<%BANNER%>

Racial Disparities in the Diagnosis and Treatment of Depressive Disorders in Medicaid Primary Care

Permanent Link: http://ufdc.ufl.edu/UFE0042092/00001

Material Information

Title: Racial Disparities in the Diagnosis and Treatment of Depressive Disorders in Medicaid Primary Care
Physical Description: 1 online resource (126 p.)
Language: english
Creator: Swaine, Zoe
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: african, depression, disparities, ethnicity, florida, health, medicaid, mental, primary, race, racial
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Racial Disparities in the Diagnosis and Treatment of Depressive Disorders in Medicaid Primary Care Racial disparities in health care are a major concern for the country. Evidence shows that African Americans in a physicians care are less likely to be diagnosed and treated for depression when compared Caucasians. There is a small but significant amount of evidence that shows that African Americans with depression are more likely to present with somatic symptoms than Caucasians which may influence diagnosis. It is speculated that these somatic symptoms of depression are frequently misattributed as symptoms of physical illness and mask the underlying etiology. This study examines the prevalence, diagnosis, and treatment of depression in a Medicaid primary care population, and the presence of racial disparities in diagnosis, treatment, and expenditures and the role of somatic symptoms in diagnosis and treatment in primary care among depressed Medicaid enrollees. A random sample of 2,106 Florida Medipass enrollees participated in a telephone survey assessing depressive symptoms, while their Medicaid claims data were collected for the years 2003 to 2005. Information on their somatic symptoms, physician diagnosis of depression, treatment history, race, and other demographic characteristics were gathered. A depression screening tool, the PHQ-2, was used to identify those enrollees with a likely depressive disorder. Of the initial 2,106, one third of these enrollees met the screening criteria and were included in the depressed sample. The analyses showed that African Americans were not less likely to be physician diagnosed as depressed, but were significantly less likely to receive treatment for depression. Caucasians had approximately four times the odds of African Americans of obtaining treatment for depression. The lower likelihood of obtaining treatment also led to a significantly lower likelihood of having any mental health expenditure. The role of somatic symptoms was not found to be significant in any of the analyses, although three analyses did approach significance. Those who endorsed somatic symptoms had approximately twice the odds of being misdiagnosed, however, once diagnosed those with somatic symptoms had 1.5 times the odd of receiving treatment, and there was an unexpected trend that indicated Caucasians had almost twice the odds of endorsing somatic symptoms. The final analysis did not support the theory that increased somatic symptoms among African Americans cause lower rates of diagnosis and treatment. This study is an important step to understanding the role of somatic symptoms in racial disparities. Larger studies are needed to fully evaluate the relationship due to the very low numbers of African Americans being diagnosed in Medicaid primary care.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Zoe Swaine.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Frank, Robert G.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042092:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042092/00001

Material Information

Title: Racial Disparities in the Diagnosis and Treatment of Depressive Disorders in Medicaid Primary Care
Physical Description: 1 online resource (126 p.)
Language: english
Creator: Swaine, Zoe
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: african, depression, disparities, ethnicity, florida, health, medicaid, mental, primary, race, racial
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Racial Disparities in the Diagnosis and Treatment of Depressive Disorders in Medicaid Primary Care Racial disparities in health care are a major concern for the country. Evidence shows that African Americans in a physicians care are less likely to be diagnosed and treated for depression when compared Caucasians. There is a small but significant amount of evidence that shows that African Americans with depression are more likely to present with somatic symptoms than Caucasians which may influence diagnosis. It is speculated that these somatic symptoms of depression are frequently misattributed as symptoms of physical illness and mask the underlying etiology. This study examines the prevalence, diagnosis, and treatment of depression in a Medicaid primary care population, and the presence of racial disparities in diagnosis, treatment, and expenditures and the role of somatic symptoms in diagnosis and treatment in primary care among depressed Medicaid enrollees. A random sample of 2,106 Florida Medipass enrollees participated in a telephone survey assessing depressive symptoms, while their Medicaid claims data were collected for the years 2003 to 2005. Information on their somatic symptoms, physician diagnosis of depression, treatment history, race, and other demographic characteristics were gathered. A depression screening tool, the PHQ-2, was used to identify those enrollees with a likely depressive disorder. Of the initial 2,106, one third of these enrollees met the screening criteria and were included in the depressed sample. The analyses showed that African Americans were not less likely to be physician diagnosed as depressed, but were significantly less likely to receive treatment for depression. Caucasians had approximately four times the odds of African Americans of obtaining treatment for depression. The lower likelihood of obtaining treatment also led to a significantly lower likelihood of having any mental health expenditure. The role of somatic symptoms was not found to be significant in any of the analyses, although three analyses did approach significance. Those who endorsed somatic symptoms had approximately twice the odds of being misdiagnosed, however, once diagnosed those with somatic symptoms had 1.5 times the odd of receiving treatment, and there was an unexpected trend that indicated Caucasians had almost twice the odds of endorsing somatic symptoms. The final analysis did not support the theory that increased somatic symptoms among African Americans cause lower rates of diagnosis and treatment. This study is an important step to understanding the role of somatic symptoms in racial disparities. Larger studies are needed to fully evaluate the relationship due to the very low numbers of African Americans being diagnosed in Medicaid primary care.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Zoe Swaine.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Frank, Robert G.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042092:00001


This item has the following downloads:


Full Text





RACIAL DISPARITIES IN THE DIAGNOSIS AND TREATMENT OF DEPRESSIVE
DISORDERS IN MEDICAID PRIMARY CARE















By

ZOE N. SWAINE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2010



























2010 Zoe N. Swaine



























To the family and friends who have helped me throughout this journey









ACKNOWLEDGMENTS

I would like to thank the Florida Center for Medicaid and the Uninsured for their

support of me throughout my graduate training and the Florida Agency for Healthcare

Administration for their generosity in allowing me to utilize their Medicaid data. I would

also like to acknowledge Robert G. Frank, Ph.D., who, even from across the country,

continued to support my work, and to the other members of my dissertation committee,

Jeffrey S. Harman, Ph.D., David M. Janicke, Ph.D., and Brenda A. Wiens, Ph.D.









TABLE OF CONTENTS

paae

A C KNO W LEDG M ENTS ............................. ......... ..... ...................... .................. 4

LIST OF TABLES ........... ..... .............................. 7

LIST OF FIGURES................................................... .......... 9

ABSTRACT .............. ....... .. ................ ........... 10

CHAPTER

1 R EV IEW O F LITERATU R E ......... ........................................... ........................... 13

Prevalence and Cost of Depression ................. ............................ .............. 13
Racial Disparities in Mental Health Services................. .......................... 13
The Prevalence of Depression among African Americans................................... 15
Disparities in the Diagnosis of Depression................................................ 18
Disparities in the Treatment of Depression ....... ...... ...................................... 20
Potential Causes of Racial Disparities in Diagnosis.................. ...... ........... 24
Symptom Presentation in Racial Disparities........................ ....................... 24
Depression and African Americans in the Medicaid Population........................... 27
Depression and African Americans in Primary Care......... ..... ..................... 28

2 PURPOSE AND SIGNIFICANCE OF THE STUDY.............. ..................... 29

The Purpose of the Study ................... ..... ... ... ...... ... ............... ........... .... 29
The Significance of the Study .............................................. ......................... 29
Specific A im s and Hypotheses ........................................ ........................... 30

3 CO NCEPTUAL M O DEL................................... ..... ....................... 32

Race, Somatic Symptoms of Depression, Depression Diagnosis, and Treatment.. 32

4 DATA AND M ETHO DS................. ....................... ....................... 34

Sample and Data ........... ................................... 34
Sam ple Identification ....................... ......... ............ ........ .. 34
Depression Case Identification...................... .......................... 35
D a ta s e t ....................................................... .......... ....... 3 5
M medicaid Claim s Data ....................... ...... ........... ..... .......... 36
Survey Data.................................... ......... ........... 37
Variable Construction and Definition........................... .......................... 37
Dependent Variables.................. ..... .............................. ..................... 37
Independent Variables and Covariates ................................ ... .......... ..... 40
Statistical Methods............... ................. .............. ... ................. 43









5 R E S U LT S .................................................... 55

Sam ple Characteristics....................... .. ................. 55
Final Survey Sam ple ...................... ...... ............... ............ ... ... 55
PHQ-2 Identified Depressed Sam ple ................................... ....... ........... 56
Research Goals......................... .. ... .. ......................... 57
Aim 1. To examine racial disparities in the diagnosis of depression and
treatment of depression the Medicaid primary care population..................... 57
H y pothe sis 1 a ....................................... ......... .................. 57
Hypothesis 1 b ........... .... ............................... .... ................. 58
H y pothe sis 1 c ..................... .......................... .................. 58
Hypothesis 1 d ................................... .. ........ ................. 59
Aim 2: To examine the role of somatic symptoms in the diagnosis and
treatm ent of depression ............... ........................................................ 59
H y pothe sis 2 a ....................................... ......... .................. 59
Hypothesis 2b ............... ............... .. ................ 60
H y pothe sis 2 c ....................................... ......... .................. 6 1
H y pothe sis 2 d ....................................... ......... .................. 6 1
Hypothesis 2e .................................................... ................. 62

6 D IS C U S S IO N ................................................... 80

Depression Diagnosis and Treatment in Medicaid Primary Care......................... 80
Racial Disparities in Diagnosis and Treatm ent .............................. ... ........ ....... 82
Racial Disparities in Diagnosis ........................................... .......................... 82
Racial D isparities in Treatm ent........................................... .......................... 82
Racial Disparities in Mental Health Expenditures ................ ... .............. 83
The Role of Somatic Symptoms in Racial Disparities................ ................... 84
Implications and Recommendations............................................. 87
Lim stations ............ .................. ........................... 89
Future research ....... ....... ................................ 91

APPENDIX BMS DEPRESSION SURVEY ............................................... ............. 93

LIST O F R EFER ENC ES ......... .............. .................... ........................ .. ................. 120

BIO G RA PH ICA L S KETC H .......... .............. ................ ................ ................. 125









LIST OF TABLES


Table page

4-1 Sample characteristics of the full random sample and of the stratified survey
sam ple .............. ....................... ................. 51

4-2 Probability of major depression or any depressive disorder at each score of
the PHQ-2. ........................................ 51

4-3 ICD-9 diagnosis codes for all depression diagnoses.................... .......... 52

4-4 A list of all included Antidepressant Medications..................... ............... 52

4-5 ICD-9 Diagnosis codes for common cancers ....................................... ..... 53

4-6 Categories of Psychiatric Comorbidities: ICD-9-CM diagnosic codes................ 54

5-1 Sample Characteristics of the final survey sample ........................................... 63

5-2 Sample Characteristics of the PHQ-2 identified depressed sample ................... 64

5-3 Results of the Logit Analysis for Hypothesis la...................... .................. 65

5-4 Power analysis for hypothesis a ......................................................... 66

5-5 Results of the Logit Analysis for Hypothesis 1b predicting treatment with
antidepressants ......... ......... ... ..... ......... .................................. 67

5-6 Results of the Logit Analysis for Hypothesis 1b predicting mental health
v is its .................................................. ......... ........... ....... 6 8

5-7 Results of the Logit Analysis for Hypothesis 1b predicting any type of
treatment. ..... .. ................................. ................... 69

5-8 Results of the Logit Analysis for Hypothesis 1 c predicting any mental health
expenditure....................................................... 70

5-9 Results of the Gamma Model for Hypothesis 1d predicting mental health
expenditure (assuming expenditures > $0).................. ......... ................ 71

5-10 Means of mental health expenditure by race (assuming expenditures > $0)...... 71

5-11 Results of the Logit Analysis for Hypothesis 2a predicting depression
diagnosis. .... ... .................................. .................. 72

5-12 Pow er analysis for hypothesis 2a ........................................ ....................... 73









5-13 Results of the Logit Analysis for Hypothesis 2a predicting depression
diagnosis, with PCS removed................... ...................... 74

5-14 Results of the Logit Analysis for Hypothesis 2b predicting any depression
treatment. ..... .. ................................. ................... 75

5-15 Results of the Logit Analysis for Hypothesis 2b predicting any depression
treatment, with PCS removed................ .............................. 76

5-16 Results of the Logit Analysis for Hypothesis 2c predicting significant somatic
s y m p to m s .................................................. ....................... 7 7

5-17 Results of the Logit Analysis for Hypothesis 2c predicting significant somatic
symptoms, with PCS removed. ................ ............................ ............... 78

5-18 Results of the Mediation Analysis for Hypothesis 2d .................................. 79

5-19 Results of the Mediation Analysis for Hypothesis 2e .................................. 79

5-20 Results of the Mediation Analysis for Hypothesis 2d with PCS removed. ......... 79

5-21 Results of the Mediation Analysis for Hypothesis 2e with PCS removed. ......... 79









LIST OF FIGURES

Figure page

3-1 Conceptual model of the relationship between race, somatic symptoms,
diagnosis and treatment and expenditures...................... .................. 32









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

RACIAL DISPARITIES IN THE DIAGNOSIS AND TREATMENT OF DEPRESSIVE
DISORDERS IN MEDICAID PRIMARY CARE

By

Zoe N. Swaine

August 2010

Chair: Robert G. Frank
Major: Psychology

Depressive disorders are common, chronic, and costly. In primary care settings,

the point prevalence of major depression ranges from 5% to 10%, with close to three

times as many individuals experiencing "sub-threshold" depressive symptoms. In 2001

the Surgeon General released a report that highlighted the prevalence of racial

disparities in mental health care, including depression, and proposed addressing these

disparities as a top priority for the nation.

African Americans in particular are vulnerable to health disparities. There is

growing evidence that even when African Americans do access and utilize healthcare

services, they are less likely to be diagnosed and less likely to be treated for depression

when compared with Caucasians. There is also some evidence that African Americans

have a lower likelihood of having any mental health expenditures, and lower total mental

health expenditures when compared with Caucasians.

Given the evidence for disparities in diagnosis and treatment, it is important to

look for potential factors that might contribute to these disparities. A 2003 Institute of

Medicine report on racial disparities states, "the most significant gap in [disparities]

research is the failure to identify mechanisms by which these disparities occur". One of









the least researched and most speculated-upon areas of research is the role of somatic

symptoms in the diagnosis and treatment of depression. There is a small but significant

amount of evidence that shows that African Americans with depression are more likely

to present with somatic symptoms than Caucasians. It is speculated that these somatic

symptoms of depression are frequently misattributed as symptoms of physical illness

and mask the underlying etiology.

This study examines the prevalence, diagnosis, and treatment of depression in a

Medicaid primary care population. It then examines the presence of racial disparities in

diagnosis, treatment, and expenditures. It also examines the effects of somatic

symptoms of depression and their role in diagnosis and treatment in primary care

among depressed Medicaid enrollees.

A random sample of 2,106 Medipass enrollees participated in a telephone

survey, assessing depressive symptoms, while their Medicaid claims data were

collected for the years 2003, 2004, and 2005. Information on their somatic symptoms,

physician diagnosis of depression, treatment history, race, and other demographic

characteristics were gathered. From this sample the a depression screening tool, the

PHQ-2, was used to identify those enrollees with a likely depressive disorder.

Of the initial 2,106, one third of these enrollees met the screening criteria and were

included in the depressed sample.

The analyses showed that African Americans were not less likely to be diagnosed

as depressed, however, this analysis was underpowered due to the very low numbers

of African Americans actually diagnosed with depression. African Americans were less

likely to receive treatment for depression. Caucasians had approximately four times the









odds of obtaining treatment for depression. The lower likelihood of obtaining treatment

also led to a significantly lower likelihood of having any mental health expenditure, and

a trend showing that African Americans' mental health expenditures were approximately

half that of Caucasians. The role of somatic symptoms was not found to be significant

in any of the analyses, although two analyses did approach significance. Those who

endorsed somatic symptoms had approximately twice the odds of not being diagnoses,

and there was an unexpected trend that indicated Caucasians had almost twice the

odds of endorsing somatic symptoms. The final analysis did not support the theory that

increased somatic symptoms among African Americans cause lower rates of diagnosis

and treatment. This study is an important step to understanding the role of somatic

symptoms in racial disparities. Lager studies are needed to fully evaluate the

relationship due to the very low numbers of African Americans being diagnosed in

Medicaid primary care.









CHAPTER 1
REVIEW OF LITERATURE

Prevalence and Cost of Depression

Depressive disorders are common, chronic, and costly. Lifetime prevalence rates

obtained from community-based surveys indicate the range to be between 5% to 17%

(Kessler, Andrews, Mroczek, Ustun, & Wittchen, 1998). One of the most recent large-

scale epidemiological surveys completed in 2005 found 12 month prevalence rates of

5% and a lifetime prevalence of 13%. In primary care settings, the point prevalence of

major depression ranges from 5% to 10%, with close to three times as many individuals

experiencing "sub-threshold" depressive symptoms (Katon and Schulberg, 1992).

Second to hypertension, depression is the most commonly presenting problem in

primary care (Kanton and Schulberg, 1992).

By 2020, depressive illness is projected to be the second leading cause of

disability worldwide (Murray and Lopez, 1997). Depression has substantial public health

implications and great economic significance. In 1990 it was estimated that depression

cost the United Sates $43 billion annually, of which $17 billion represents lost work days

(Greenberg, Stiglin, Finklestein & Berndt, 1993). In an update to this research

Greenberg et al. (2003) found that these costs had increased to $84 billion.

Racial Disparities in Mental Health Services

In 2001 the Surgeon General released a report that highlighted the prevalence of

racial disparities in mental health care, and proposed addressing these disparities as a

top priority for the nation (DHHS, 2001). The Surgeon General's report found that

minorities have less access to, and lower availability of, mental health services, that

they are less likely to receive needed mental health services, that when in treatment









they often receive a poorer quality of care, and that minorities are underrepresented in

mental health research (DHHS, 2001). The report also found evidence that minorities

experience a greater disability burden from mental illness than do Caucasians,

stemming from both receiving less care and a poorer quality of care, rather than from

increased prevalence or increased severity of illness.

African Americans in particular are vulnerable to disparities in health for several

reasons. African Americans have been shown to have similar rates of mental illness as

Caucasians, however, they are over-represented among vulnerable populations in

which the rates of mental illness are higher. African Americans are also

disproportionately served by the "safety net providers," such as Medicaid providers or

county health departments, who have been coming under increasing pressure over the

past two decades from healthcare reforms. In addition, African Americans are more

likely to live in poor areas, where there are also shortages of healthcare providers.

Finally, African Americans have been shown to have lower utilization rates of health

services than Caucasians (DHHS, 2001).

Consensus on the definition of disparity has been slow to form. The Institute of

Medicine (IOM) released a report in 2003, defining a disparity in healthcare as "racial or

ethnic differences in the quality of health care that are not due to access-related factors

or clinical needs, preferences, and appropriateness of intervention" (IOM, 2003, pg. 32).

However, some of those who use the IOM definition to model disparities have chosen to

disregard the IOM's inclusion of "not due to access-related factors" in their

conceptualization of disparities (McGuire, Alegria, & Cook, 2006). The main contention

with the IOM definition is that it only addressed disparities arising from within the clinical









encounter, not from factors arising prior to the encounter, including health care system

factors such as health insurance. The debate continues with regards to defining

disparities with one extreme arguing for models that do not adjust for any

socioeconomic factors, thereby allowing them to mediate the race-healthcare

relationship, to the other extreme of regression-based models that carefully control for

all variables.

The Prevalence of Depression among African Americans

In order to demonstrate equal need among the African American population, the

prevalence of depression among African Americans must be examined. Several large

scale epidemiological studies of the prevalence of mental disorders have produced

mixed results with regards to rates of depression in African Americans. The

Epidemiologic Catchment Area study (ECA) sampled residents of Baltimore, St. Louis,

Durham-Piedmont, Los Angeles, and New Haven to examine the prevalence of major

depression among racially and ethnically diverse adults (Sommervell, Leif, and

Weismann & Bruce, 1989). In total, it sampled 4,638 African Americans, 12,944

Caucasians, and 1,600 Hispanics who were surveyed between 1980-1983. Diagnoses

of depression were identified through the use of Diagnostic and Statistical Manual, Third

Edition criteria. Age, sex, and site-adjusted analyses did not show any significant

differences in lifetime prevalence or six-month prevalence between Caucasians and

African Americans. There were some differences among the 18-24 years age group in

6 month prevalence: African American women showed a trend for higher six-month

prevalence than Caucasian women, and Caucasian men showed a trend for higher six-

month prevalence than African American men. Three sites showed significantly lower









rates of depression among African Americans when compared to Caucasians, after

controlling for income and age.

Dunlop, Song, Lyons, Manheim and Chang (2003) estimated depression

prevalence rates among African American, Hispanic, and Caucasian adults from a

population-based national sample (1996 HRS). In order to allow for sufficient numbers

for comparison their sample included an oversampling of African Americans and

Hispanics relative to Caucasians. The study used the short form of the World Health

Organization's Composite International Diagnostic Interview (CIDI-SF) to identify likely

depression cases. Results showed that African Americans (88.5 per 1000) and

Hispanics (107.8 per 1000) exhibited elevated rates of major depression relative to

Caucasians (77.5 per 1000). However, after controlling for socio-demographic, health,

and economic factors, Hispanics and Caucasians exhibited similar rates, and African

Americans exhibited significantly lower rates than Caucasians. This indicates that

African Americans in the community suffer from higher rates of mental illness than

Caucasians, but that the difference may be explained by demographic factors and

socioeconomic factors. In other words, factors associated with depression were more

frequent among members of minority groups than among Caucasians. Elevated

depression rates among minority individuals are heavily associated with serious chronic

illness, functional limitations, a lack of health insurance, and health behaviors such as

smoking and not exercising.

Williams et al. (2007) used the National Study of American Life (NSAL), a national

household probability sample assessing the mental health of African Americans, to

survey a total of 3,570 African Americans, 1,621 Caribbean-blacks, and 891 non-









Hispanic white adults. The survey included an adaptation by the World Health

Organization of the Composite International Diagnostic Interview (CIDI). Lifetime Major

Depressive Disorder prevalence estimates were highest for Caucasians (17.9%),

followed by Caribbean-blacks (12.9%) and African Americans (10.4%). However, 12-

month prevalence of Major Depressive Disorder was similar across groups (African

American 5.9%, Caribbean-black 7.2%, and Caucasian 6.9%).

The same study examined the chronicity of Major Depressive Disorder. This was

assessed by comparing the ratio of individuals with 12-month Major Depressive

Disorder to lifetime Major Depressive Disorder cases, which showed that chronicity was

significantly greater for both minority groups (56.5% for African Americans and 56.0%

for Caribbean-blacks) than for whites (38.6%). Depression severity and impairment was

assessed through the Sheen Disability Scale and the Quick Inventory of Depressive

Symptomatology Self-Report (QIDS-SR). These instruments showed that, relative to

Caucasians, both minority groups were more likely to rate their depression as severe

and disabling. This study indicates that while lifetime risk for depression may be lower

among African Americans and Caribbean blacks, their 12-month risk is similar to

Caucasians due to the greater chronicity of depression among these two groups.

While findings are mixed regarding racial differences in the prevalence of

depression, the results from these studies are consistent in that they indicate the rates

of major depression among African Americans are similar to those of Caucasians.

However, the relationship between depression and race is complex, and there is an

overrepresentation of African Americans in high-need populations which are harder to

access when surveying (for example, those living in inner cities, poor rural areas, and









prisons are populations). If members of these groups were included, then higher rates

of depression among African Americans might be detected.

Disparities in the Diagnosis of Depression

Given that findings indicate that prevalence rates for depression are similar

across racial groups, the question remains where do racial disparities in healthcare

come from? There is growing evidence that even when African Americans do access

and utilize healthcare services, they are less likely to be diagnosed with depression

when compared to Caucasians.

In one of the more recent study of disparities in depression diagnosis in primary

care, Stockdale et al.(2008) used the National Ambulatory Medical Care Survey

(NAMCS) for 1995-2005 to examine disparities in diagnosis and treatment in primary

care. The survey gathers information about ambulatory office visits to primary and

specialty care. Minorities were underrepresented in the primary care sample (Caucasian

80.68%, African American 10.6%, Hispanic 8.72%), perhaps reflecting their lower rates

of healthcare utilization. Primary care visits by African Americans had significantly

lower odds of resulting in a depression or anxiety diagnosis when compared to

Caucasians (odds ratios ranged from 0.56-0.65). A rate by time interaction analysis,

also indicated that these disparities in primary care did not change during the 10 years

of the study.

Borowsky et al.(2000) conducted a large scale cross-sectional survey of 19,309

patients and 349 internists and family physicians in Boston, Chicago, and Los Angeles.

As part of the Medical Outcomes Study, participants completed a self-administered

screening survey that included a brief depression screening questionnaire while

physicians completed questionnaires regarding the diagnosis and treatment of the









participants. The study showed that both Hispanics (OR=0.94, p< 0.05) and African

Americans (OR = 0.63, p= 0.05) were less frequently diagnosed with any mental health

problem than Caucasians, despite having the same level of mental health functioning.

Lower diagnosis rates were also found by Skaer, Sclar and Robinson (2000) in

an analysis of data from the US National Ambulatory Medical Care Survey, between

1992 and 1997. This is a nationwide probabilistic survey of physician office visits

completed by the National Center for Health Statistics, which yielded a total of 36,875

patient records. This data showed evidence that rates of diagnosis of depression

among Caucasians (11.3%) were significantly higher than among African Americans

(5.5%) and Hispanics (8.3%).

In contrast to findings of disparities in diagnoses, Minsky et al. (2003) studied

new admissions to the New Jersey behavioral health system between January 2000

and August 2001. They collected data on mental health functioning on 19,219 patients

and found that Latinos were more likely to be diagnosed than Caucasians (OR = 1.74),

while no differences were found between African Americans and Caucasians (OR =

0.99). In another study, Crystal et al. (2003) used the Medicare Current Beneficiary

Survey's (MCBS) cost and use files, between 1992 and 1998, to obtain estimates of

depression diagnoses, and the rates of treatment of those with depression. Their

sample consisted of 20,966 elderly individuals from which they derived 51,058 person-

years. Diagnoses recorded in Medicare claims were used to identify individuals who

received a diagnosis of depression from a healthcare provider and pharmacy and

claims data were used to identify receipt of antidepressants and/or psychotherapy. This









data showed no racial disparities in the likelihood of being diagnosed with depression

among this elderly population.

Other researchers have examined how varying patient attributes influences

physician diagnosis. McKinley et al. (2002) presented videotapes of staged patient-

physician encounters for depression and polymyalgia rheumatica (PMR) in order to

examine the effects of patient attributes on the diagnosis of depression. They examined

age (65 years or 80 years), sex, race (African American or Caucasian), and occupation

(blue or white collar) in various combinations to assess their impact on physician

diagnosis. The study found no significant influence of any of the patient attributes on

the physicians' "most likely diagnosis" of either depression or PMR. However,

characteristics of the physicians (e.g., medical specialty, race, and age) did impact the

decision. Obviously physician behavior in such an experimental environment may not

be a valid representation of behavior in practice in the community.

Overall, these studies show that the relationship between physician diagnosis

and race is clearly a complex one. While there is a large volume of evidence that lends

weight to the existence of disparities in diagnosis, the reasons behind these disparities

are not always well understood.

Disparities in the Treatment of Depression

Given that some studies show evidence for disparities in diagnosis, with African

Americans being less likely to be diagnosed than Caucasians, this might logically lead

to a decreased likelihood of African Americans being treated for depression. Williams et

al. (2007), who examined the prevalence of depression among African Americans,

Caribbean blacks, and non-Hispanic whites in the NSAL study, showed that fewer than

half of the African Americans (45.0%) and fewer than a quarter (24.3%) of the









Caribbean blacks who met the criteria for a major depressive disorder received any

form of therapy for their depression. While these numbers seem very low, they had no

data on treatment of non-Hispanic Caucasians and so comparisons could not be made.

Claims data were used in one study of Medicaid recipients to examine racial

disparities in depression treatment. Melfi, Croghan and Hanna (1999) examined racial

differences among Medicaid recipients in treatment for depression. Their sample was

46% African American and showed that African Americans were less likely to receive

antidepressant medication on the first diagnosis of depression and once treatment was

initiated, they were more likely to receive older try-cyclic antidepressants (TCAs).

African Americans were also more likely to prematurely discontinue treatment and less

likely to receive a second medication. While this study appears to offer compelling

evidence, it must be noted that they did not control for demographic variables,

depression severity, or the presence of comorbid illness.

The National Center for Health Statistics (NCHS) annually samples a nationally

representative sample of office visits to physicians in Ambulatory practice, including

primary care and all specialties. Using this data Harman et al. (2001) analyzed visits to

primary care physicians and psychiatrists where a depression diagnosis was recorded,

during two time periods: 1993-1994 and 1996-1997. In the 1993-1994 time period

African Americans had significantly lower odds of receiving any depression treatment

when compared with Caucasians (OR =0.89). However, in the 1996-1997 time period

this difference had disappeared. Harmon argued that this provides evidence that racial

disparities in the treatment of depression may be shrinking. However, these results

should be interpreted with caution as their sample only included those who had been









diagnosed with depression by their physician, and there is clear evidence that

depression is generally under-diagnosed, especially in primary care.

Harman, Fortney, and Edlund (2004) did a further study, this time using the 2000

Medical Expenditure Panel Survey (MEPS). MEPS is an annual, nationally

representative survey that gathers data on healthcare use, expenditures, health status,

and demographic variables. Their analyses showed that among those with self-

reported depression, African Americans were significantly less likely to fill an anti-

depressant prescription (OR=.47) when compared with Caucasians. Once treatment

was initiated, there were no racial differences in the likelihood of receiving an adequate

course of antidepressants. There were also no racial differences in the likelihood of

receiving psychotherapy; however, of those who received psychotherapy, African

Americans were significantly more likely to receive an adequate course (OR=2.47). The

odds of receiving any treatment for depression were significantly lower for African

Americans when compared with Caucasians (OR=.44). These studies suggest that

disparities in treatment may stem from gaining the initial access to treatment rather than

in maintaining an adequate course of treatment. Once treatment was initiated, African

Americans were more likely to receive an adequate course of counseling and

psychotherapy and equally as likely to receive an adequate course of antidepressant

treatment. What this study was unable to reveal were the rates of under-treatment

among those who had not been diagnosed as depressed or those who did not report a

diagnosis.

In addition to examining diagnosis disparities, Stockdale, Lagomasino, Siddique,

McGuire, and Miranda (2008) used the NAMCS to evaluate racial disparities in









treatment. They compared rates of antidepressant prescriptions, counseling or referral

for counseling, and any depression treatment (a dichotomous variable that indicated

either form of treatment). The data showed that African Americans had significantly

lower odds of receiving counseling (or a counseling referral) in the 1997-1999 time

period (OR = 0.59) and significantly lower odds of receiving an antidepressant

prescription in the 1997-2005 time period (OR range = 0.53 to 0.67). Time by race

interactions proved non-significant indicating that disparities in treatment did not change

over time. Finally their analysis of the odds of receiving any treatment showed that

African Americans had significantly lower odds of receiving any type of treatment when

compared to Caucasians across all time periods (OR range = 0.59 to 0.74). These

differences persisted even after controlling for the effect of receiving a diagnosis.

Again, the time by race interaction proved non-significant, indicating no change in racial

disparities in treatment across the period of the study from 1995 to 2005.

Further evidence that racial disparities in mental health treatment still exist comes

from a 2007 MEPS study. Cook, Maguire, & Miranda (2007) found a worsening of racial

disparities in mental health treatment from 2003/2004 to 2006/2007. This study and the

Stockdale study indicate the continued existence of racial disparities in the treatment of

depression in primary care, contrary to the work by Harman et al. (2001). While

Stockdale's goal was to examine disparities without eliminating the influence of SES on

this population, Harman et al. (2001) controlled carefully for demographic variables,

insurance variables, and physician specialty (primary care versus psychiatry) among

other influential variables, which may have removed most of the effect of race. This









difference in methodology stems from the ongoing debate surrounding the definition of

disparities.

Potential Causes of Racial Disparities in Diagnosis

Given the evidence for disparities in diagnosis and treatment, it is important to

look for potential factors that might contribute to these disparities. The IOM report on

racial disparities states, "the most significant gap in [disparities] research is the failure to

identify mechanisms by which these disparities occur" (IOM, 2003). Several potential

causes that contribute to the existence of racial disparities in mental health care have

been proposed and can be classified as systemic factors, patient factors, and physician

factors. The 2001 Surgeon General's report on racial and ethnic disparities in mental

health provides a detailed review of these (DHHS, 2001). Patient factors include:

coping style, mistrust of physicians, treatment seeking behavior, stigma, immigration

and acculturation, and health status. Physician factors include: communication

difficulties, clinician bias and stereotyping, and poor recognition in primary care. Finally,

systemic factors include: service setting and limitations in access, financing of the

system and health insurance, paucity of evidence-based treatment in the community,

cultural competence, effectiveness of medication among minorities, poverty, community

violence, and marginal neighborhoods. One of the most speculated-over causes, but

one of the least studied is the patient's symptom presentation.

Symptom Presentation in Racial Disparities

Mental illness is a worldwide phenomena, but the way patients express or

present their symptoms to clinicians is affected by culture. There is a small but

significant amount of evidence that shows that African Americans with depression are

more likely to present with somatic symptoms than Caucasians. It is speculated that









these somatic symptoms of depression are frequently misattributed as symptoms of

physical illness and mask the underlying etiology.

The misattribution of somatic symptoms to physical illness has some support. In

one Italian study examining the influence of symptom presentation, PCPs were asked to

record the reason for the patient visit, either psychological/family problems, physical

illness, or pain symptoms. Depressed patients were then identified by psychiatric

assessment. Among depressed patients those who presented with physical symptoms

were at 2.3 times the risk for non-recognition and those presenting with pain had four

times the risk of not being recognized as depressed Menchetti, Belvederi-Murri,

Bertakis, Bortolotti & Berardi). Another study of primary care found that somatization

reduced physician recognition of depression from 77% to 22% (Kirmayer, Robbins,

Dworkind & Yaffe, 1993).

There is also some evidence that African Americans may present with more

somatic symptoms of depression than Caucasians. In a study with 665 African

American and Caucasian psychiatric inpatients, differences in diagnosis were examined

using the DSM-III-R Symptom Checklist (Hudziak, Helzer and Wetzel et al., 1993).

Results showed that while disparities still existed in the diagnosis of Schizophrenia and

Bipolar disorder, there were no differences in depression diagnosis between the two

groups. However, they did find evidence for differences in symptom attribution, by race

of the patient, especially in schizophrenia. However, these results may not generalize

to outpatient care or primary care.

Wohl, Lesser and Smith (1997) also used a structured interview to compare the

nature and severity of depressive symptoms in depressed, medically healthy African









Americans and Caucasians. Twenty matched subjects were assessed using a

structured interview and with the Hamilton Depression Rating Scale (HAM-D). HAM-D

items were then grouped into 7 factors: diurnal, sleep, weight, reality, mood, anxiety and

somatic. When these were compared, overall severity of depression was comparable

between groups however, Caucasians showed significantly more mood and anxiety

symptoms, whereas African Americans had significantly more diurnal variation to their

depression. There were no differences on other neurovegetative symptoms. While the

sample size of this study was small, the results do suggest that there are differences in

symptom presentation between racial groups.

In a study that aimed to retrospectively determine the impact of race on treatment

adherence and outcomes among patients being treated for major depression in urban

primary care settings, Brown Schulberg and Madonia (1996) analyzed their data to

compare psychiatric history and clinical presentation between 119 African American and

153 Caucasians. While the two groups showed no significant differences in their

treatment histories, the severity of their depression, the severity of medical illness, or in

their level of psychosocial functioning, there were significant racial differences in other

areas. African Americans were more likely to have psychiatric and medical

comorbidities, they showed more severe sleep disturbance, and they had more severe

somatic symptoms, greater limitations in their self-reported physical functioning, higher

life stress, and more negative health beliefs.

While these studies are valuable in advancing the understanding of racial

disparities, none of these studies has systematically evaluated the role of somatic

symptoms of depression in diagnosis.









Depression and African Americans in the Medicaid Population

The Medicaid population is a unique and valuable population to study. African

Americans are disproportionately covered by Medicaid, 21% versus 8% of Caucasians

among non-elderly adults (Brown, Ojeda, Wyn, and Levan, 2000), and in 1997 36.7% of

African American children were covered by Medicaid compared to 17.1% of Caucasian

children (US Bureau of the Census, 1998). Nationally, 23.1% of the Medicaid population

is African American, which is approximately 13 million residents (CMS, 2007).

Disadvantaged populations, such as those in Medicaid, have been shown to be at

increased risk for poor mental health. Poverty itself has been shown to be a risk factor

for depression. Factors contributing to chronic stress, such as substandard housing,

high crime neighborhoods, and poor nutrition are associated with an increased risk for

psychological dysfunction (Bennet, 1987). Studies of low-income groups and Medicaid

recipients have shown higher rates of depression among these populations. In addition

to being at greater risk for depression, studies have shown that Medicaid enrollees are

particularly vulnerable to under-treatment for psychological disorders (Harman et al.,

2001, Harman, Fortney, and Edlund, 2004; Melfi, Croghan, and Hanna, 1999).

Depressed Medicaid recipients are significantly less likely to receive any treatment for

depression (OR = .81) than those with private insurance (Harman et al, 2001) and less

likely to receive SSRI's, psychotherapy, and an adequate length of therapy than

privately insured individuals (Melfi, Croghan, and Hanna, 1999). With the dual factors of

increased risk for depression and under-treatment, and the high proportions of African

Americans who rely on Medicaid, the Medicaid population is important to study in racial

disparities research.









Depression and African Americans in Primary Care

There is evidence that shows that a disproportionate number of African

Americans seek psychiatric care in primary care settings rather than with specialists

(DHHS, 2001). However, studies have shown that usual care by primary care

physicians fails to recognize 30% to 50% of depressed patients (Valenstein, Vijan,

Zeber, Boehm, and Buttar, 2001), and of those who are recognized, treatment rates can

be as low as 27% (Tylee, 2006). Klinkman, Coyne, Gallo, and Schwenk (1998) explored

physician recognition and diagnostic sensitivity to the disorder by comparing physician

ratings against the "gold standard" of the SCID completed by a Licensed Clinical Social

Worker. While the SCID found a 13% prevalence of depression, sensitivity of the

physician diagnosis was low, only 0.34. This means that physicians identified only 34%

of depressed individuals. However, specificity was high at .93, while positive predictive

value was .45 which means less than half of the patients identified as depressed were

actually depressed.

Given that African Americans frequently seek mental health treatment from their

primary care physician, and given the evidence for under-diagnosis and under-

treatment of African Americans combined with the under treatment among the Medicaid

population, individuals that belong to all three of these groups (African Americans,

primary care, Medicaid) are in a uniquely disadvantaged position.









CHAPTER 2
PURPOSE AND SIGNIFICANCE OF THE STUDY

The Purpose of the Study

The main purpose of the proposed study is to examine factors that contribute to

the racial disparities seen in the diagnosis and treatment of depression in a Medicaid

primary care population. The study will firstly examine the prevalence, diagnosis, and

treatment of depressive symptomatology in a Medicaid primary care population. It will

then assess for the presence of racial disparities in diagnosis, treatment, and mental

health expenditure among the depressed population. The study will then go on to

examine the mediating effects of somatic symptoms of depression between race and

diagnosis of depression and also between race and treatment among primary care

Medicaid enrollees.

The Significance of the Study

In order to reduce racial disparities in mental health care, we must better

understand the mechanisms that cause it. If evidence can be found for the role of

somatic symptoms contributing to racial disparities, screening and treatment efforts may

be better targeted to reduce the rates of untreated depression, and minimize the

extensive costs associated with this. To do this be it is important to clarify the

relationship between race, symptom presentation and diagnosis and treatment. While

many studies have examined the role of race in physicians' decision to diagnose, none

have examined the mediating role of somatic symptoms of depression, and none have

examined these factors among the vulnerable Medicaid population.

A major limiting factor in many of the existing studies is the difficulty in identifying

depressed patients. Those studies that look at treatment differences all rely on









physician diagnosis as a valid indicator of depression in the study population. However,

as demonstrated earlier, this method can miss a high proportion of the truly depressed

population. This study will use a widely validated tool (PHQ-2) to assess depression in

order to identify our depressed sample, independently of physician diagnosis. This

method of identifying depressed cases, combined with access to physician diagnoses,

treatment, and expense data make this study unique. Another limitation of previous

studies is that while they have the ability to identify differences in symptom presentation

between races, these studies have not been able to identify if this impacts diagnosis.

The Patient Health Questionnaire PHQ-9 will be used to identify exactly which

symptoms of depression each individual is experiencing in order to investigate the role

of symptom presentation in disparities in diagnosis. Taken all together, this study will

allow a thorough and comprehensive examination of racial disparities in depression

care in the Medicaid population, as well as one potential cause of disparities.

Specific Aims and Hypotheses

* AIM 1. To examine racial disparities in the diagnosis of depression, treatment of
depression, and healthcare expenditures of a depressed Medicaid primary care
population.

* HYPOTHESIS 1A. Depressed patients who are African American will be less likely to
be diagnosed with a depressive disorder than Caucasians.

* HYPOTHESIS 1B. Depressed patients who are African American will be less likely to
receive treatment for depression than Caucasians.

* HYPOTHESIS 1 c. Depressed patients who are African American will be less likely to
have any mental health care expenditures compared to Caucasians.

* HYPOTHESIS 1 D. Depressed patients who are African American will have
significantly lower mental healthcare expenditures than Caucasians.

* AIM 2. To examine the role of somatic symptoms in the diagnosis and treatment of
depression









* HYPOTHESIS 2A. Depressed patients who endorse somatic symptoms will be less
likely to be diagnosed with a depressive disorder than those who do not.

* HYPOTHESIS 2B. Depressed patients who endorse somatic symptoms will be less
likely to be receive treatment than those who do not.

* HYPOTHESIS 2c. Depressed patients who are African Americans will be more likely
to endorse somatic symptoms of depression than Caucasians.

* HYPOTHESIS 2D. Of those who are depressed, the number of somatic symptoms of
depression will mediate the relationship between race and the diagnosis of
depression.

* HYPOTHESIS 2E. Of those who are depressed, the number of somatic symptoms of
depression will mediate the relationship between race and the treatment of
depression.









CHAPTER 3
CONCEPTUAL MODEL

Race, Somatic Symptoms of Depression, Depression Diagnosis, and Treatment

The proposed study will be based on a conceptual model that relates race,

somatic symptoms of depression, diagnosis, treatment, and healthcare expenditures.

This conceptual framework proposes that race will influences the presentation of

somatic symptoms of depression which will then, in turn impact the likelihood of a

depression diagnosis. The diagnosis, along with racial factors, then influences the

likelihood of obtaining treatment, which will in turn increase mental health expenditures.

Other patient, physician and systemic factor will also impact the likelihood of a physician

visit and of diagnosis and treatment. The model for the proposed study is depicted as

follows:





Somatic
Symptoms of
Depression


Physician
Race Physician Depression
Visit Diagnosis




Other Patient, Depression Mental
Physician, & Treatment Health
Systemic Factors Expenditures




Figure 3-1. Conceptual model of the relationship between race, somatic symptoms,
diagnosis and treatment and expenditures











The proposed study includes only those individuals who have had at least one

physician visit and will examine the relationship between race and diagnosis, race and

treatment, and race and expenditure, while controlling for covariates, in Aim 1. In Aim 2

the study will examine the relationship between race and somatic symptoms of

depression, the relationship between somatic symptoms of depression and diagnosis

and treatment, and the relationship between race and diagnosis and race and

treatment, mediated by the somatic symptoms of depression, while controlling for

covariates known to impact depression diagnosis and treatment.









CHAPTER 4
DATA AND METHODS

Sample and Data

Sample Identification

A random sample of 31,775 Medipass enrollees were selected from the entire

population of Florida Medicaid enrollees, from which a total of 2,106 enrollees

participated in a telephone survey and are included in the final sample. Medicaid

Provider Access System (MediPass) is a primary care case management program for

Medicaid beneficiaries. Primary care providers receive a $2.00 monthly case

management fee for each of their enrolled patients, in addition to fee-for-service for all

services rendered. The sample was stratified by sex and was obtained through the use

of Computer Assisted Telephone Interviewing (CATI). The full sample of 2,106 was

used for initial descriptive analyses of the data, and to produce an estimate of the

prevalence of depression and of treatment rates among the Medicaid population in

Florida as a whole. Sub-samples of those identified as depressed will then used in the

main analyses of the study. To examine possible issues of sampling bias table 4-1

shows a comparison of demographics variables (sex, age, and race/ethnicity) obtained

from Medicaid claims data between the non-surveyed sample and the surveyed sample.

Medicaid data was used as this was available for both the surveyed and non-surveyed

groups. The comparison indicates that there has been an oversampling of female

enrolees (deliberate) and an over sampling of "black" and Hispanic enrollees, while

"white" enrollees are underrepresented.









Depression Case Identification

The primary aims of the study are to examine diagnosis and treatment among

those who are depressed, therefore the study must identify those who are depressed.

Case identification will be accomplished using the PHQ-2. The PHQ-2 is a 2-item

depression-screening scale that inquires about the frequency of depressed mood and

anhedonia over the past 2 weeks, scoring each as 0 ("not at all") to 3 ("nearly every

day"). Any participant that scored four or above on the PHQ-2 was included in the

depressed sample. Table 4-2 shows the probabilities of actually having any depressive

disorder (81.2%) or major depressive disorder (45.5%) at this and all other cut offs of

the PHQ-2 (Kroenke, Spitzer, & William, 2003). The PHQ-2 is widely used in

epidemiological research and has been validated among many populations, including

the elderly and poor, and has been shown to have excellent validity and reliability

among minority populations (Kroenke, Spitzer, & William, 2003). While the PHQ-2 can

indicate the presence of a depressive episode, it does not specify whether a depressive

episode occurs in the course of a major depressive disorder or whether it occurs in the

course of another psychological disorder (for example bipolar disorder or

schizophrenia).

Dataset

The dataset consists of two major elements: 1) healthcare claims data obtained

from the state of Florida's Medicaid administrator, the Agency for Healthcare

Administration (AHCA) for the years 2002, 2003, and 2004, and 2) healthcare survey

data obtained by the Bureau of Economic and Business Research (BEBR) in the

employ of the Florida Center for Medicaid and the Uninsured, as part of an evaluation

for AHCA, obtained between August 2004 and March 2005. The data contains









Protected Health Information so full IRB and HIPAA approval was been obtained for this

study and for use of this data.

Medicaid Claims Data

The AHCA supplies Florida Medicaid archival claims data as part of a contractual

agreement with the Center for Medicaid and the Uninsured (FCMU), at the University of

Florida. The healthcare claims data provide information on recipient eligibility, facility

claims (outpatient and inpatient), medical claims (physician claims), and pharmacy

claims.

Medicaid claims were compiled for the years 2003, 2004 and 2005 and included

in the study if the claim fell within two years of the survey date of the enrollee.

Florida Medicaid eligibility files include recipient information including, date of

birth, gender, race/ethnicity, Medicaid assistance category, Medicaid plan, length of

enrollment, and eligibility start and end dates.

Pharmacy, facility, and medical claims files provide event-level information.

Pharmacy claims are submitted if a recipient filled one or more prescriptions

during the period of the study. These claims include the date the medication is filled,

therapeutic drug class, National Drug Code number, and a refill indicator. Facility

claims, which include claims from both outpatient and inpatient facilities, provide ICD-9-

CM diagnoses (up to 11), procedure codes, data on provider characteristics (including

type and specialty), and a flag if the event was an emergency department visit. Medical

claims are claims billed by physicians and include ICD-9-CM diagnoses (up to 2),

procedure codes, and provider characteristics. All claims files have a unique identifier

for each Medicaid recipient to enable merging of the 9 claims datasets and the survey

dataset. Medicaid primary care claims were identified by use of the provider specialty









code. The following specialties will be included as primary care providers: general

practice (11), family practice (09), internal medicine (18), pediatrics (35), adolescent

medicine (1), geriatrics (13), geriatric nurse practitioner (82), and adult primary care

nurse practitioner (75).

Survey Data

The survey data utilized for this study is part of a broader AHCA study assessing

a disease management program targeting depression, implemented by Florida's

Medicaid program. The survey contained questions on mental health, including

depression, health-related quality of life, and general demographic questions. Partial

funding for this study came from the pharmaceutical company Bristol-Myers Squibb.

The sample used in this study is the control group that received no intervention. The

survey was completed using the Computer Assisted Telephone Interviewing system

(CATI) with a random sample of adult Florida Medicaid enrollees between August 2004

and March 2005. Data collection was completed by trained staff at the UF survey

Research Center (UFSRC) at the Bureau of Economic and Business Research (BEBR).

The full survey can be seen in Appendix 1.

Variable Construction and Definition

The variables to be used in the study will retain their numerical values. However,

some of the variables are categorical in nature and are dummy coded to enable their

use in regression analyses.

Dependent Variables

Physician diagnosis (depphys). In hypotheses la, 2a, and 2d, physician

diagnosis of depression is the dependent variable. Each of these aims examines

patient characteristics thought to influence physicians' decisions to diagnose









depression. For this reason, the depression diagnosis is treated as a dichotomous

variable. No diagnosis will be coded as "0" and a diagnosis of depression is coded as

"1". Physician-diagnosed depression is identified by the International Classification of

Diseases Clinical Modification (ICD-9-CM) codes of unipolar depression appearing on

one or more health claims during the study period. ICD-9-CM diagnostic codes

(including all primary and secondary diagnoses) are used to identify any type of unipolar

mood disorder (single episode, recurrent or unspecified). Any eligible diagnosis in any

of the diagnosis fields in either the facility claims file or the medical claims files indicate

a diagnosis. See Table 4-3 for ICD-9-CM codes corresponding to unipolar mood

disorder diagnoses used in this study.

Treatment (mhvisit). Mental health outpatient Current Procedural Terminology

(CPT) codes and antidepressant therapeutic drug class codes were extracted from the

claims data to indicate the receipt of treatment for depressive disorders. Mental health

outpatient visits are identified by CPT codes used for billing psychological services.

Using this definition a dichotomous variable was created (mhvisit) where "1" designates

1 or more mental health visits during the 24-month period and "0" indicates no mental

health care visit.

Treatment (medany). A dummy-coded variable was also created to examine

treatment with antidepressant medication. See Table 4-4 for the list of medications and

their FDA indications. The variable (medany) indicates whether any antidepressant

prescription was filled in the 24-month period of the study. Because pharmacy claims

cannot be linked to the specialty of the prescribing physician, all antidepressant

medications listed in Table 4-2 will be counted.









Treatment (txany). A final variable was constructed to indicate the presence or

absence of any type of treatment. It is a dichotomous variable with "0" indicating no

medication or mental health visits and "1" indicating treatment with either medication or

a mental health visit, or both. Treatment is the dependent variable in hypothesis lb

and 2d to examine the effects of race on treatment.

Somatic Symptoms of Depression (somat). The presence of somatic symptoms

of depression is identified through the use of the Patient Health Questionnaire (PHQ-9),

which is included in the survey. The Patient Health Questionnaire is a brief, 9-item, self-

report depression assessment specifically developed for use in primary care. It utilizes

diagnostic criteria from the Diagnostic and Statistical Manual of Mental disorders, Fourth

Edition (DSM-IV) (APA, 1994). Several studies support its validity, feasibility, and its

capacity to detect changes of depressive symptoms over time (Kroenke, Spitzer, Janet

and Williams, 2001). A dichotomous variable (somat) was created that indicates the

presence of significant somatic symptoms (a total score of 5 or more on the 4 somatic

diagnostic criteria of depression: weight change, sleep difficulties, feeling

agitated/slowed, and fatigue).

Mental Health Expenditures. Expenditures was identified through claims data

and is measured in two ways. A dichotomous variable (mhexpany) will be created to

identify the presence of any mental health expenditures with "1" indicating expenditures

and "0" indicating no expenditures. A continuous variable will be based on a dollar figure

of total mental health expenditures (mhexptot). Mental health expenditures was

identified through similar means as treatment, through use of CPT codes to identify

mental health visits and the corresponding paid dollar amount, and through the









identification of all antidepressant medications and their paid dollar amount. The cost of

all mental health expenditures was then totaled to produce one figure for each

individual.

Independent Variables and Covariates

Race (race). The variable race was obtained from the survey data. While the

Medicaid eligibility files include data on racial/ethnic group classification, the data do not

identify race and ethnicity separately. Coding of the race variable is "1" for Caucasian,

"2" for African American, and "3" for all other races. In the analyses these are treated

as three separate dummy variables with Caucasian being omitted and treated as the

reference group..

Ethnicity (ethnic). Ethnicity was also obtained from self-reported survey data.

The dummy variable was created with Non-Hispanic coded as "0" and Hispanic coded

as "1". In 1997 the Office of Budget Management (OBM) released federal guidelines for

the assessment of race and ethnicity (of Hispanic origin or not), based on the logic that

people of Hispanic origin can be any race. Ethnicity has been shown to affect access,

utilization of health services and rates of disability. Ethnicity is used as a covariate in all

analyses to control for the effects of ethnicity separately from the effects of race in the

analyses.

Sex (Sex). The sex of the enrollees was extracted from the eligibility file and is

dummy coded (1 =male, 2=female). Sex serves as a covariate in all hypotheses, where

it will be used to control for the effects of sex on the outcome variable. Sex has been

shown to affect depression and health care utilization.

Age (age). Claims data was used to extract age at the date the survey interview

was completed (a unique date for each recipient). Coding of the age variable is "1" for









18-35, "2" 36-64, and "3" for 65 and older. In all hypotheses age is used as a covariate.

Age has been shown to be related to diagnosis and treatment of depression, and many

other health behaviors. In the analyses these are treated as three individual dummy

variables with 18-35 being omitted and treated as the reference group.

Medical Comorbidities (comortot). Five medical comorbidities were identified

from the ICD-9 codes from the claims files: asthma and other chronic lung diseases

(ICD code 490 through 496.9), diabetes (ICD code 250 through 250.9), heart disease

(ICD codes 392.0, 393.0 through 398.99, 410 through 414.9, 415.0 through 416.9, and

420 through 429.9), hypertension (ICD code 401.0 through 405.99), and cancer (see

Table 4-5 for a list of included ICD-9 cancer diagnoses). A comorbidity is indicted if any

of these diagnoses are present in any of the ICD-9 code fields. These conditions have

been selected for their prevalence in primary care and Medicaid in general, and their

comorbidity with depressive disorders. In addition, by specifying large categories of

chronic disease it will eliminate the effects of counting all comorbidities, even if they are

minor and/or acute.

A count variable was created that identified the number of these comorbidities that

each individual has (comortot). Medical comorbidities will be used as a covariate or risk

adjustment variable in all other analyses. Wang and colleagues (2000) suggested this

parsimonious use of diagnosis counts performs comparatively to more complicated

comorbidity indexes.

Psychiatric Comorbidities (psycoany). Psychiatric comorbidities was identified

from the ICD-9 diagnostic fields in the claims files (see Table 4-6 for a list of all included

diagnoses). Four dichotomous dummy variables were created to indicate the presence









of specific categories of psychiatric comorbidities. These are important covariates as

some diagnoses, such as schizophrenia, are likely to be associated with much higher

expenditures than other psychiatric comorbidities. Dummy variables for anxiety

disorders (anxdiag), bipolar disorders (bipodiag), schizophrenia (schizdiag), and

substance abuse disorders (sadiag) were created. The presence of a diagnosis in each

of these categories is coded as "1" and the absence of a diagnosis will be coded as a

"O". The presence of a psychiatric comorbidity has been shown to increase individuals

chances of obtaining a depression diagnosis. It is hypothesized that the presence of a

psychiatric comorbidity acts as a cue to physicians to consider mental health diagnoses.

Education level (educat). Higher levels of education has been shown to be

associated with better health outcomes. This variable was obtained from the survey

data which asks about the highest level of education the respondent completed.

Responses choose one of 6 ordinal categories (1) 8th grade or less, 2) some high

school, 3) high school graduate or GED, 4) some college or two-year degree, 5) 4- year

degree, and 6) more than 4 year degree). This variable will be included as a covariate in

all analyses to control for the effects of years of education. In the analyses these are

treated a 6 individual dummy variables, with 8th grade or less being omitted and treated

as the reference group.

Self-reported physical and mental health (PCS & MCS). Self-reported physical and

self reported mental health will be obtained from the survey data. The survey contains

the Medical Outcomes Study Short Form 12 (SF-12) (Ware, Kosinski, and Keller, 1996).

The SF-12 is a 12 item measure of General Health Function. It was designed to assess

an individuals self-reported physical and mental quality of life over the previous 4









weeks. The SF-12 was constructed to reproduce the SF-36 physical and mental health

summary measures with at least 90% accuracy and allows for calculation of the

Physical Component Summary (PCS) and the Mental Component Summary (MCS)

scores. The SF-12 is a valid and reliable measure of overall health status (Ware,

Kosinski, and Keller, 1996). The SF-12 is used as a covariate in all analyses to control

for physical and mental health status.

Number of Primary Care Visits (pcvisit). The number of visits to a primary care

provider was obtained from the Medicaid outpatient claims data. The higher the number

of visits to primary care, the greater the opportunity there is to receive a depression

diagnosis, so it is important to control for this. Any claim encounter that included a

provider code from those identified earlier as primary care providers were totaled to

create a count variable indicating the number of visits to those providers. This will be

used as a covariate in all analyses.

Psychiatrist Visits (psyvisit). A dichotomous variable indicating the presence of

any visits with a psychiatrist was obtained from the outpatient claims data. Individuals

with any claim with a psychiatrist (specialty code: 42) or child psychiatrist (specialty

code: 43) was coded as "1" and those who have had no contact with a psychiatrist were

coded as "O". It is important to control for psychiatrist visits as these individuals are

more likely to have obtained treatment than those who have not seen a psychiatrist.

Statistical Methods

Hypothesis la: Depressed patients who are African American will be less likely to

be diagnosed with a depressive disorder than Caucasians A Logit was used in

hypothesis la as the outcome variable, depression diagnosis, is dichotomous. Ordinary

Least Squares (OLS) could not be used with dichotomous data as the error term for this









data is not normally distributed and is heteroskedastic, which both violate the

assumptions of OLS. This would result in OLS estimates that are not efficient and it is

theoretically possible to obtain predictions that are below 0 or above 1 (not possible in

binary/dichotomous data). Instead, the logit model was used as it does not require the

dependant variable to be normally distributed, it does not require linearity between

independent and dependent variables, it is robust against heteroscedasticity (the non-

homogeneity of variance), and it does not require normally distributed error terms. The

Logit was used to examine the odds of being diagnosed with depression by race, while

controlling for all other variables. The equation for the model is depicted below.

logit(p) = 3o + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) + ps(anxdiag) +

p6(bipodiag) + y7(schizdiag) + p8(sadiag) + p9(educatl) + o10(educat2) + p11(educat3) +

P12(educat4) + p13(educat5) + i14(pcs) + p15(mcs) + p16(pcvisit) + y17(agel) +

P18(age2) + P19(RaceAA) + p20(Raceother) + s

Where p is the probability of diagnosis and 1-p is the odds of no diagnosis.

Hypothesis 1 b: Depressed patients who are African American will be less likely to

receive treatment for depression than Caucasians. Three separate Logit analyses were

used in hypothesis 1 b as all three of the outcome variables were dichotomous (any

antidepressant, any mental health visit, or any treatment). These regressions examined

the odds of having any treatment by race, while controlling for all other variables. The

equation for the model is depicted below.

logit(p) = 3o + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) + ps(anxdiag) +

s6(bipodiag) + y7(schizdiag) + 1B(sadiag) + BP(educatl) + o10(educat2) + P11(educat3) +









Pi2(educat4) + P13(educat5) + p14(pcs) + p15(mcs) + P16(pcvisit) + p17(agel) +

P18(age2) + P19(RaceAA) + p20(Raceother) + E

Where p is the probability of treatment (any antidepressant, any MH visit, any

treatment) and 1-p is the odds of no treatment.

Hypothesis 1c: Depressed patients who are African American will be less likely to

have any mental health care expenditures compared to Caucasians. A Logit was used

in hypothesis 1c as the outcome variable, any mental health expenditure, is

dichotomous. This regression examined the odds of having any mental health

expenditures by race, while controlling for all other variables. The equation for the

model is depicted below:

logit(p) = 3o + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) + s5(anxdiag) +

p6(bipodiag) + y7(schizdiag) + p8(sadiag) + p9(educatl) + p0o(educat2) + p11(educat3) +

P12(educat4) + p13(educat5) + i14(pcs) + p15(mcs) + p16(pcvisit) + y17(agel) +

P18(age2) + P19(RaceAA) + p20(Raceother) + s

Where p is the probability of any mental health expenditure and 1-p is the odds of

no mental health expenditure. The logit transformation is defined as in hypothesis la.

Hypothesis id: Depressed patients who are African American will have

significantly lower mental healthcare expenditures than Caucasians. The Two-Part

model was used to predict mental health expenditures due to expenditure data being

censored at 0, and the high likelihood of significant skewness. This model first predicts

the probability of having any expenditures using the logit equation in hypothesis 1c, and

then in the second part predicts expenditures given that expenditures are greater than

zero, using a gamma model. While Tobit models are an estimation procedure that









accounts for censored values (here, anything below zero) of the dependent variable,

which is seen in expenditures data, it was not used due to it's sensitivity to non-normally

distributed data and homoskedasticity. These are characteristics of Medicaid

expenditure data. These two regressions estimate the difference in mental health

expenditures by race, while controlling for all other variables. The first step of the model

is as follows:

Equation 1: logit(p) = po + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) +

p5(anxdiag) + p6(bipodiag) + y7(schizdiag) + p8(sadiag) + p9(educatl) + p0o(educat2) +

311(educat3) + P12(educat4) + P13(educat5) + p14(pcs) + p15(mcs) + P16(pcvisit) +

P17(agel) + p18(age2) + P19(RaceAA) + p20(Raceother) + s

Where p is the probability of any mental health expenditure and 1-p is the odds of no

mental health expenditure. The logit transformation is defined as in hypothesis 1 a

In step 2, given expenditures are >0 (identified by the logit), a gamma model was

used to predict expenditure. The benefit of this model is that allows for skewness in the

dependent variable, it is not sensitive to heteroskedasticity, and it does not need to be

retransformed once the analysis has been run. The gamma model is as follows:

f(mhexptot) = [(mhexptot /b)c-1] [(exp(-mhexptot /b)/bF(c)]

Hypothesis 2a: Depressed patients who endorse somatic symptoms will be less

likely to be diagnosed with a depressive disorder than those who do not. A Logit was

used in hypothesis 2a as the outcome variable, depression diagnosis, is dichotomous.

This regression examined the odds of being diagnosed with depression by somatic

symptoms, while controlling for all other variables. The equation for the model is

depicted below.








logit(p) = 3o + P3(psyvisit) + P2(sex) + p3(ethnic) + p4(comortot) + 35(anxdiag) +

p6(bipodiag) + y7(schizdiag) + 38(sadiag) + 39(educatl) + p1o(educat2) + 311(educat3) +

P12(educat4) + 313(educat5) + 314(pcs) + p15(mcs) + 316(pcvisit) + p17(agel) +

Pis(age2) + 3s9(RaceAA) + p20(Raceother) + p21(somat) + E

Where p is the probability of diagnosis of depression and 1-p is the odds of no

depression diagnosis.

Hypothesis 2b: Depressed patients who endorse somatic symptoms will be less

likely to be receive treatment than those who do not. A Logit was used in hypothesis 2b

as the outcome variable, any treatment, is dichotomous. This regression examined the

odds of receiving any treatment by somatic symptoms, while controlling for all other

variables. The equation for the model is depicted below.

logit(p) = 3o + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) + s5(anxdiag) +

p6(bipodiag) + y7(schizdiag) + p8(sadiag) + p9(educatl) + o10(educat2) + p11(educat3) +

i12(educat4) + p13(educat5) + i14(pcs) + p15(mcs) + p16(pcvisit) + y17(agel) +

P18(age2) + p19(RaceAA) + p20(Raceother) + p21(somat) + E

Where p is the probability of having any treatment and 1-p is the odds of no

treatment.

Hypothesis 2c: Depressed patients who are African Americans will be more likely

to endorse somatic symptoms of depression than Caucasians. A Logit was used in

hypothesis 2c as the outcome variable, significant somatic symptoms, is dichotomous.

This regression examined the odds of experiencing somatic symptoms by race, while

controlling for all other variables. The equation for the model is depicted below:









logit(p) = 3o + P3(psyvisit) + P2(sex) + p3(ethnic) + p4(comortot) + s5(anxdiag) +

p6(bipodiag) + y7(schizdiag) + 38(sadiag) + 39(educatl) + p1o(educat2) + 311(educat3) +

P12(educat4) + 313(educat5) + 314(pcs) + p15(mcs) + 316(pcvisit) + p17(agel) +

P18(age2) + 319(RaceAA) + p20(Raceother) + E

Where p is the probability of somatic symptoms and 1-p is the odds of no somatic

symptoms.

Hypothesis 2d: Of those who are depressed, the number of somatic symptoms of

depression will mediate the relationship between race and the diagnosis of depression.

A Logit was used in hypothesis 2d as the outcome variable, depression diagnosis, is

dichotomous. In order to measure the mediating effects of somatic symptoms between

race and diagnosis, the change in the race odds ratio from the base model in

hypothesis la was compared with the more fully specified model which includes the

somatic symptoms variable. This is a common method of measuring the mediation of

variables (Baron & Kenny, 1986). If race effects disappear (change from significant to

non-significant) when somatic symptoms are controlled for, this indicates that somatic

symptoms do in fact play a significant role in mediating the relationship between race

and diagnosis. However, some critics argue that a coefficient changing from below the

arbitrary 0.05 threshold to above is not the most conservative approach. For this

reason, to ensure that the change in the race coefficient is a meaningful change the

difference between the two coefficients was tested for statistical significance using the

Hausman Test. The equations for the models are depicted below:

Equation 1: logit(p) = po + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) +

ps(anxdiag) + p6(bipodiag) + y7(schizdiag) + ps(sadiag) + p9(educatl) + o10(educat2) +








P11(educat3) + P12(educat4) + P13(educat5) + p14(pcs) + p15(mcs) + P16(pcvisit) +

P17(agel) + p18(age2) + P19(RaceAA) + p20(Raceother) + E


Equation 2: logit(p) = po + Pi(psyvisit) + P2(sex) + p3(ethnic) + p4(comortot) +

p5(anxdiag) + 36(bipodiag) + y7(schizdiag) + 38(sadiag) + 39(educatl) + p1o(educat2) +

311(educat3) + 312(educat4) + 313(educat5) + 1p4(pcs) + s15(mcs) + 316(pcvisit) +

P17(age1) + s18(age2) + p19(RaceAA) + p20(Raceother) + p21(somat) + E

Where p is the probability of diagnosis of depression and 1-p is the odds of no

depression diagnosis.

Hypothesis 2e: Of those who are depressed, the number of somatic symptoms of

depression will mediate the relationship between race and the treatment of depression.

As before a Logit was used in hypothesis 2e as the outcome variable depression

treatment is dichotomous. The same technique was used as in hypothesis 2d to test for

the mediating effects of somatic symptoms between race and treatment. The equations

for the models are depicted below:

Equation 1: logit(p) = po + PB(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) +

p5(anxdiag) + p6(bipodiag) + y7(schizdiag) + ps(sadiag) + p9(educatl) + o10(educat2) +

311(educat3) + p12(educat4) + p13(educat5) + i14(pcs) + p15(mcs) + p16(pcvisit) +

3i17(age1) + p18(age2) + P19(RaceAA) + p20(Raceother) + E


Equation 2: logit(p) = po + Pi(psyvisit) + p2(sex) + p3(ethnic) + p4(comortot) +

B5(anxdiag) + s6(bipodiag) + y7(schizdiag) + s8(sadiag) + B9(educatl) + o10(educat2) +








P11(educat3) + p12(educat4) + p13(educat5) + p14(pcs) + p15(mcs) + p16(pcvisit) +

P17(agel) + p18(age2) + pi9(RaceAA) + p20(Raceother) + p21(somat) + E

Where p is the probability of treatment (antidepressant, psychotherapy, and any

treatment) and 1-p is the odds of no treatment.









Table 4-1 Sample characteristics of the full random sample and of the stratified survey
sample.
Total Random Sample S y
Survey Sample
(non-surveyed)
N =29,669 N=2,106
Sex
Male 56% 50%
Age 46.7 years 44.7 years
Race (Medicaid Data)
White 43.0 % 36.9%
Black 24.4% 30.4%
American Indian .1% .0%
Oriental .5% .3%
Hispanic 12.4% 16.5%
Other 19.6% 15.9%


Table 4-2 Probability of major depression or any depressive disorder at each score of
the PHQ-2.
PHQ-2 score Probability of major Probability of any
depressive disorder (%) depressive disorder (%)
1 15.4 36.9
2 21.1 48.3
3 38.4 75.0
4 45.5 81.2


56.4


84.6
92.9









Table 4-3 ICD-9 diagnosis codes for all depression diagnoses
Unipolar Mood Disorder ICD-9-CM
Depression NOS 311.00
Major Depressive Disorder-Single Episode 296.20-296.26
Major Depressive Disorder-Recurrent Episode 296.30-296.36
Dysthymia 300.40


Table 4-4 A list of all included Antidepressant Medications
Drug Class Generic Name Brand Name
Antidepressant Imipramine Tofanil
Desipramine Norpramin
Amitriptyline Elavil
Nortriptyline Aventyl, Pamelor
Protriptyline Vivacil
Trimipramine Surmontil
Doxepin Sinequan, Adapin
Maprotiline Ludiomil
Amoxapine Asendin
Trazodone Desyrel
Fluoxetine Prozac
Bupropion Wellbutrin
Sertraline Zoloft
Paroxetine Paxil
Venlafaxine Effexor
Nefazodone Serzone
Fluvoxamine Luvox
Phenelzine Nardil
Tranylcypromine Parnate











Table 4-5 ICD-9 Diagnosis codes
Cancer Group
Bones/soft tissue
Brain
Breast
Colon
Endocrine
Gyn
Head and neck
Lung
Lymph node spread
Melanoma


Non-colon GI
Non-melanomatous
Non-specific site
Pleura/mediastinum
Prostate
Secondary cancer
Testes/Male GU
Urinary Tract


skin cancer


for common cancers
ICD-9 code ranges
170, 171,238.1,238.2
190-192.9, 237.5, 237.6, 239.6
174, 175, 239.3
153, 154, 235.2
193, 194, 237.0,237.4, 239.7
180, 182, 183, 184, 236.1,236.2


140-149.9, 160, 1
162, 235.9, 239.1
196
172
150-152.9, 155-1
173,238.2


61,162, 195.0


59.9, 235, 239.0


199, 238.8, 238.9, 239.8, 239.9
164
236.5
197
187.3, 187.4, 187.9, 236.4, 236.6
189, 236.7, 236.91, 239.4, 239.5










Table 4-6 Categories of Psychiatric Comorbidities:
Psychiatric Comorbidity Categories
Anxiety Disorders
Anxiety State-Unspecified
Panic Disorder
Generalized Anxiety Disorder
Other Anxiety State
Phobia-Specified
Agoraphobia with Panic Attacks
Agoraphobia without Panic Attacks
Social Phobia
Other Phobias
Obsessive-Compulsive Disorder
Acute Reaction to Stress
Adjustment Disorder with Anxious Mood
Prolonged Post Traumatic Stress Disorder


Schizophrenia


ICD-9-CM diagnosic codes
ICD-9-CM Codes


300.00
300.01
300.02
300.09
300.20
300.21
300.22
300.23
300.29
300.30
308.30
309.24
309.81

295.xx


Bipolar Disorders
Bipolar- Single Episode
Bipolar I
Bipolar II
Bipolar NOS

Substance Abuse Disorders
Alcohol-induced Disorders
Drug-induced Disorders


296.00-296.06
296.40-296.60
296.89
296.80


291.xx
292.xx









CHAPTER 5
RESULTS

Sample Characteristics

Final Survey Sample

Initially, the data collection obtained 2,411 completed surveys. Using the study

criteria, 2,106 Medipass Medicaid recipients were included in the final sample, this was

lower than the 2,411 sampled as several Medicaid recipients did not have any primary

care claims within 2 years of their survey date. Table 5-1 shows sample characteristics

of this sample. Almost 53% of the full survey sample was male Given that sampling

was stratified by sex to include equal proportions of male and females, this is an

overrepresentation of men in the final survey sample, presumably due to the fact that

men had more primary care claims and so were more likely to end up in the sample.

African American and Hispanic enrollees also appeared to be overrepresented (33.5%

and 32.7% respectively in the survey sample, as opposed to 24.4% and 12.4% in the

non survey sample), however, it is impossible to compare these numbers in this way

due to the different way that race and ethnicity is collected in Medicaid claims and the in

the survey sample. The age of the survey group was only slightly lower than the non-

surveyed group (44.7 versus 45.8 years).

Of the overall survey sample, 6.9% obtained a physician diagnosis of a depressive

disorder, and 23.8% received some kind of mental health treatment. One third (33.7%)

of the sample met criteria for a depressive disorder as measured by the PHQ-2 (N =

709). Of the 68 patients diagnosed with depression by their physician, 82% received at

least one prescription for antidepressants. Thirty six percent of physician-diagnosed

cases received at least one mental health visit, as assessed by the CPT codes.









PHQ-2 Identified Depressed Sample

Table 5-2 shows sample characteristics of the PHQ-2 identified depressed

sample. The majority (59.8%) of the sample was Caucasian (N = 422) and 29.2% were

African American (N= 206), the remainder, classified as "other", accounted for 11% of

recipients' (N = 78) race group membership. Almost 35% of the sample indicated that

they were of Hispanic ethnicity (N = 256). This race/ethnicity distribution corresponds to

a slight over-representation of Caucasians as compared to African Americans and

Hispanics when compared to the overall survey sample.

The mean age of recipients in the study sample was 48.16 years (SD = 12.5), with

a range fro 18 years to 102 years. Most of the sample were high school graduates,

however only 7.6% of the sample had a college degree. Their PCS and MCS scores

were both very low compared to national norms, indicating poorer physical and mental

functioning. A large percent of the sample also endorsed significant somatic symptoms

of depression (84%). This level of physical complaints does appear to be reflected in

the average number of primary care visits during the period of the study, which was 39

visits.

Of the 709 cases identified as depressed by the PHQ-2. The rate of physician

diagnosis among this group was 11.1%. The rate of antidepressant treatment was

35.7% and 8.4% of survey-identified cases received at least one mental health visit,

while 37% received any type of mental health treatment. Eighteen percent had at least

one psychiatrist visit. The average mental health expenditure of the sample was

$385.11, with a large number having $0 expenditures and heavy skewness and

kurtosis.









Research Goals

Aim 1. To examine racial disparities in the diagnosis of depression and treatment
of depression the Medicaid primary care population.

Hypothesis la

Hypothesis la states that patients who are African American will be less likely to

be diagnosed with a depressive disorder than Caucasians. The results of the Logit

indicates that this analysis does not support this hypothesis, see table 5-3 for the results

of the Logit analysis. While controlling for the covariates of age, sex, education, number

of physical comorbidities, number of primary care visits, psychiatrist visit, physical and

mental wellbeing, and being diagnosed with an anxiety disorder, bipolar disorder, or a

psychotic disorder; being African American did not impact the likelihood of being

diagnosed with depression in Medicaid Primary Care (OR = .973, ps.96). The variable

substance abuse diagnosis was dropped due to the lack of this diagnosis in all of the

patients. This analysis highlighted the effect of psychiatric caseness, whereby patients

that already have seen a psychiatrists are significantly more likely to obtain a diagnosis

in primary care when compare with those who did not have a visit with a psychiatrist.

Here we see very large effect sizes with those who had seen a psychiatrist having

almost 40 times the odds of being diagnosed with depression (OR = 39.765, p< .001).

Of note is the very small numbers involved in the analysis, crosstabs showed that only 9

African Americans in the sample obtained a diagnosis of depression from their

physician, indicating the lack of power in any analyses examining race and diagnosis.

See Table 5-4 for the results of the power analysis.









Hypothesis lb

Hypothesis 1b states that patients who are African Americans will be less likely to

receive treatment (antidepressants, a mental health visit, or any treatment) for

depression than Caucasians. The results of the 3 separate Logits support this

hypothesis. See table 5-5, 5-6, and 5-7 for the results of the three respective Logit

analyses. While controlling for all covariates, Caucasians had almost 4 times the odds

receiving antidepressants when compared to African Americans (OR = .258, p .001),

race does not predict mental health visits (OR = .203, ps .461), but it predicted receiving

any type of mental health treatment in Medicaid Primary Care, with Caucasians having

approximately 4 times the odds of receiving any mental health treatment when

compared with African Americans (OR = .239, ps .001). Being female, having more

PCP visits, having a psychiatrist visit, and being above the age of 65 all increased the

odds of receiving any depression treatment. The lack of effect in predicting mental

health visits appears to be due to a lack of power due to small sample size.

Hypothesis 1c

Hypothesis 1 c states that patients who are African American will be less likely to

have any mental health expenditures than Caucasians. The results of the Logit analysis

supports this hypothesis, see table 5-8 for the results of the Logit analysis. While

controlling for the covariates of age, sex, education, number of physical comorbidities,

number of primary care visits, psychiatrist visit, physical and mental wellbeing, and

being diagnosed with an anxiety disorder, bipolar disorder, or a psychotic disorder;

Caucasian had 4 times the odds of African Americans of having any mental health

expenditures (OR = .239, p5 .001). Being female, having more PCP visits, having a

psychiatrist visit, being above the age of 65, all increased the likelihood of having any









mental health expenditures. Table 5-9 shows the average expenditure of each racial

group.

Hypothesis ld

Hypothesis 1d states that patients who are African American will have lower

mental health expenditures than Caucasians. The results of the Gamma regression

show a trend to supporting this hypotheses, table 5-10 shows the results of the Gamma

analysis. Assuming expenditures were greater than zero, African Americans had 43%

lower mental health expenditures (3 =-.433; p 5.087), relative to Caucasians, however

this did not reach significance.

Aim 2: To examine the role of somatic symptoms in the diagnosis and treatment
of depression

Hypothesis 2a

Hypothesis 2a states that patients who endorse greater somatic symptoms will be

less likely to be diagnosed with a depressive disorder than those who do not. For this

model, significant somatic symptoms (the dichotomous variable) proved to have the

strongest relationship with diagnosis and so was used in place of a count of somatic

symptoms. The results of the Logit do not support this hypothesis, see table 5-11 for

the results of the Logit analysis. While controlling for all covariates (including the

number of physical comorbidities and the score on the SF-12 PCS) the presence of

significant somatic complaints does not impact the likelihood of being diagnosed with

depression in Medicaid primary care (OR = .433, ps .137). However, this analysis

almost met significance, indicating that there may be some influence of somatic

symptoms on diagnosis and this lack of effect is likely due to a lack of power (1 -r err

prob = .31), see Table 5-12 for the power analysis.









However, there may be shared variance between somatic symptoms and the 2

other measure of physical wellbeing: number of physical comorbidities and the Physical

Component Score of the SF12 (PCS). It is possible that the inclusion of both these in

the analyses may be masking the effect of somatic symptoms. While the number of

physical comorbidities does not appear to significantly impact any of the analyses, PCS

does appear to influence several of the analyses. To test this theory the analysis was

re-run without the PCS variable, and the results are shown in Table 5-13. The removal

of the PCS variable weakened the relationship between somatic symptoms and

diagnosis (OR = .50, ps .21).

Hypothesis 2b

Hypothesis 2b states that patients who endorse greater somatic symptoms will be

less likely to receive any mental health treatment than those who do not. The results of

the Logit do not support this hypothesis, see table 5-14 for the results of the Logit

analysis. While controlling for all covariates (including the number of physical

comorbidities and the score on the SF-12 PCS) the presence of significant somatic

complaints does not impact the likelihood of receiving any mental health treatment in

Medicaid primary care (OR = 1.091, p< .795).

As with the previous analysis, this analysis was re-run with the PCS variable

removed and the results are shown in Table 5-15. In this case, the removal of PCS did

appear to influence the results, although the results did not become significant,

removing PCS did strengthen the relationship between somatic symptoms and receiving

any treatment (OR = 1.56, p .157). However, this shows the reverse of the expected

direction, that when depressed individuals have significant somatic symptoms of









depression their likelihood of obtaining treatment increases, although, this did not reach

significance.

Hypothesis 2c

Hypothesis 2c states that patients who are African Americans will be more likely to

endorse somatic symptoms of depression than Caucasians. The results of the Logit do

not support this hypothesis, see table 5-16 for the results of the Logit analysis. While

controlling for all covariates being African American does not impact the likelihood of

endorsing somatic symptoms in Medicaid Primary Care (OR = .900, p .733).

As with the previous two analyses, this analysis was re-run with the PCS variable

removed and the results are shown in Table 5-17. The removal of PCS from the

analyses did impact the relationship between somatic symptoms of depression and

race. With the removal of PCS from the analyses this relationship became very close to

significant (OR = .588, ps .062), however this was also in the reverse direction to what

was predicted, that Caucasians have approximately 1.7 the odds of endorsing somatic

symptoms of depression, when compared with African Americans.

Hypothesis 2d

Hypothesis 2d states that the number of somatic symptoms of depression will

mediate the relationship between race and the diagnosis of depression. This was

achieved through examining the change in the odds ratio for Race when somatic

symptoms is added to the model predicting diagnosis. The addition of somatic

symptoms to this equation did not significantly change the value of the odds ratio of

Race (8.4% change, ps.480). This indicates that relationship between race and

diagnosis of depression is not mediated by somatic symptoms. This change in the odds

ratio can be seen in Table 5-18.









Once more, due to the possible shared variance between somatic symptoms and

the PCS score, the analysis was re-run with PCS removed from the model. The results

are shown in Table 5-19. The removal of PCS from the model predicting diagnosis did

not significantly change the value of the odds ratio of Race (9.8% change, ps.73).

This again indicates that the relationship between race and diagnosis of depression is

not mediated by somatic symptoms.

Hypothesis 2e

Hypothesis 2e states that the number of somatic symptoms of depression will

mediate the relationship between race and the treatment of depression. This was

achieved through examining the change in the odds ratio for Race when somatic

symptoms is added to the model predicting any treatment. The addition of somatic

symptoms to this equation did not significantly change the value of the odds ratio of

Race (.8% change, ps.899). This indicates that relationship between race and

treatment is not mediated by somatic symptoms. This change in the odds ratio can be

seen in table 5-20.

The analysis was re-run with PCS removed from the model and the results are

shown in Table 5-21. The removal of PCS from the model predicting treatment did not

significantly change the value of the odds ratio of Race (1.7% change, ps.90). This

once more indicates that the relationship between race and treatment of depression is

not mediated by somatic symptoms.









Table 5-1 Sample Characteristics of the final survey sample
Final Survey Sample
N = 2,106
Sex (Male) 52.4%
Age mean 45.86 years
Race
Caucasian 54.6%
African American 33.5%
Other 11.9%
Ethnicity (Hispanic) 32.7%
Education
8th Grade or less 13.3%
Some High School 21.9%
High School Grad 34.7%
Some College 22.3%
4 Year College 5.5%
Post Graduate 2.4%
Physician Diagnosis Depression 6.9%
Any Treatment 23.8%
PHQ-2 Criteria of Depression 33.7%









Table 5-2 Sample Characteristics of the PHQ-2 identified depressed sample
PHQ-2 Depressed Sample N = 709
Sex (Male) 49.6%
Age mean (range) 48.16 years (18-102)
Race


Caucasian
African American
Other
Ethnicity (Hispanic)
Education
8th Grade or less
Some High School
High School Grad
Some College
4 Year College
Post Graduate
PCS mean (SD)
MCS mean (SD)
Significant Somatic Symptoms
Number of Physical Comorbidities
0
1
2
3
4
5
Number of PCP Visits Mean (SD)
Psychiatrist visit
Physician Diagnosed
Depressive Disorder
Anxiety Disorder
Schizophrenia
Bipolar Disorder
Any Antidepressant
Any Mental Health Visit
Any Mental Health Treatment
Total Mental Health Expenditure
Mean (SD)
Range
Skewness, Kurtosis


35.77
39.94


59.8%
29.2%
11.0%
34.8%

15.7%
23.8%
34.4%
18.5%
5.3%
2.3%
(12.83)
(12.20)
83.6%


81.1%
7.8%
6.9%
2.7%
1.1%
.4%
38.73 (85.98)
18.2%

11.1%
2.1%
0.4%
0.1%
35.7%
8.4%
37.1%

$385.11 ($993.20)
$0 -$11,532.62
4.53, 30.74









Table 5-3 Results of the Logit Analysis for Hypothesis 1 a
95% C.I. for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant .079 1.319 .004 .952 1.082 1.082 .079
Race (Caucasian) .842 .656


African American
Other
Ethnicity
Sex
Number of
Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia Diag

Bipolar Diagnosis

Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


.342 .559


22.414


-1.798
-.825
-.673
.450
-1.281

.001


40192.970


1.138

2.061
.485


.737 1.757


10.097
7.411


.000 1.000

10.069 .073


.632 8.078 .004


2.365
1.270
.269
.891


.426 .514


60.190
1.167
13.688


39.765
.981
.936


1.683 .431


2.710
5.546
1.655
2.856


.572
1.254
1.645
8.609
3.969

1.004


15.683
.947
.904


100.830
1.016
.969


This table shows the results from the logit analysis predicting depression
race is the variable of interest here.


.225
.336
diagnosis,


1.999
6.222









Table 5-4 Power analysis for Hypothesis 1 a
Tails 1
Odds Ratio .973
a err prob .050
Total Sample Size 528.000
R2 .480
Critical z -1.640
Power .072









Table 5-5 Results of the Logit Analysis for Hypothesis 1 b predicting treatment with
antidepressants
95% C.l.for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 2.174 .870 6.247 .012 8.791


Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia Diag
Education
Some High School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Saw a psychiatrist
SF12 PCS
SF12 MCS
Age Categories (18-
35)
Age 36-64
Age 65+


-1.355
-1.289
-.017
.428
.036

1.512
19.254

.281
.612
.679
-.007
-.298
.004

1.681
-.043
-.041


.310
.731


This table shows the results from
treatment with antidepressants.


.304
.438
.294
.244
.142

1.092
21768.816

.393
.386
.428
.567
.811
.002


26.783
19.872
8.683
.003
3.076
.065

1.917
.000
5.161
.513
2.516
2.519
.000
.135
3.676


4.536
2.301 E8

1.325
1.844
1.972
.993
.742
1.004


.362 21.600 .000
.011 14.459 .000
.011 15.366 .000
3.244 .197

.370 .703 .402
.419 3.048 .081
the logit analysis examining race


.142
.117
.552
.951
.785

.533
.000

.614
.866
.853
.327
.151
1.000

2.644
.936
.940


.660
.914
icting


.468
.649
1.749
2.473
1.369

38.564


2.861
3.929
4.559
3.014
3.638
1.007

10.920
.979
.980


2.816
4.719


5.373
.958
.960


1.364
2.077
in pred









Table 5-6 Results of the Logit Analysis for Hypothesis 1 b predicting mental health visits
95% C.l.for
B S.E. Wald Sig. Exp(B) EXP(B)
EXP(B)
Lower Upper
Constant .853 1.314 .422 .516 2.347


Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia Diag

Bipolar Diagnosis

Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


-.775
-1.040
-.562
-.832
.147

-1.735
2.052

18.216

-.998

-1.389
-1.818
-1.588
-.566
.000

3.220
-.028
-.035


-.789
-.640


.609
.760
.420
.401
.221

1.226
1.679
40192.970


3.216
1.618
1.873
1.787
4.309
.443

2.001
1.494
.000


.157
.222
1.000


.461
.354
.570
.435
1.158

.176
7.786
.000


9.583
.566 3.106


6.204
6.891
1.665
.261
.005

50.942
2.127
3.954
2.002

2.000
.642


1.520
1.567
1.299
.955
1.785

1.952
209.078


.121 1.118


25.031
.973
.966


.454
.528


10.338
.937
.933


.152
.110


.744
.631
2.281
4.977
1.003

60.604
1.010
1.000


1.356
2.520


This table shows the results from the logit analysis examining race in predicting
treatment with a mental health visit.









Table 5-7 Results of the Logit Analysis for Hypothesis 1 b predicting any type of
treatment
95% C.l.for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 1.354 .744 3.306 .069 3.872


Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia Diag

Bipolar Diagnosis

Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


1.920 1.087
19.805 21940.476
40192.970
21.364


32.474
27.433
6.671
.294
3.839
3.313

3.120
.000
.000


.077
.999
1.000


.408
.785
1.873
2.253
1.635

57.404


6.820
3.990E8
.000


2.592
.351 .414


.298
.181
.604
.342
5.381

33.362
20.515
12.760
14.174

.687
11.383


.254
1.237


This table shows the results from
type of treatment for depression.


.630 2.493


5.077
.956
.968


1.289
3.445


2.925
.938
.951


.707
1.679


2.312
2.402
1.834
2.531
1.007

8.811
.975
.986


2.351
7.067


the logit analysis examining race in predicting any









Table 5-8 Results of the Logit Analysis for Hypothesis 1c predicting any mental health
expenditure
95% C.I. for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 1.354 .744 3.306 .069 3.872


Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia Diag

Bipolar Diagnosis

Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


1.920 1.087
19.805 21940.476
40192.970
21.364


32.474
27.433
6.671
.294
3.839
3.313

3.120
.000
.000


.077
.999
1.000


.408
.785
1.873
2.253
1.635

57.404


6.820
3.990E8
.000


2.592
.351 .414


.298
.181
.604
.342
5.381

33.362
20.515
12.760
14.174

.687
11.383


.254
1.237


.630 2.493


5.077
.956
.968


1.289
3.445


2.925
.938
.951


.707
1.679


2.312
2.402
1.834
2.531
1.007

8.811
.975
.986


2.351
7.067


This table shows the results from the logit analysis examining race in predicting any
mental health expenditure.










Table 5-9 Results of the Gamma Model for Hypothesis 1d predicting mental health
expenditure (assuming expenditures > $0)
B S.E. 95% Wald Confidence Interval Hypothesis Testing
Wald
Lower Upper C uare Sig.
Chi-Square
Intercept 7.041 .0959 6.853 7.229 5389.784 .000
Race (Caucasian)
African American -.433 .2525 -.928 .062 2.934 .087
Other -.537 .3248 -1.173 .100 2.728 .087


Table 5-10 Means of mental


health expenditure by race (assuming expenditures > $0).


Mean


Race
Caucasian $1,142.82
African American $741.48
Other $668.29











Table 5-11 Results of
diagnosis


the Logit Analysis for Hypothesis 2a predicting depression


95% C.I. for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 40193.012 .000 1.000 .000
21.890Constant
21.890


Significant Somatic
Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia Diag
Bipolar Diagnosis
Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


-.838


-.779
.685
22.990

-1.842

-.853
-.722
.456
-1.186
.001

3.664
-.025
-.073


-.422
.315


This table shows the results from
predicting diagnosis.


.564 2.207
.534
.539 .045
.605 .444
.418 .810
.369 .576
.224 .312


.806 .936
1.387 .244
40193.012 .000
10.349
.639 8.315


2.475
1.433
.275
.774
.434

59.872
1.767
15.288
1.611

.570
.180


the logit analysis examining somatic symptoms in


.333
.621
1.000
.066
.004

.116
.231
.600
.379
.510


1.307

2.566
4.903
3.309
2.729
1.760

2.225
30.081


.554

1.233
1.585
8.655
4.291
1.004

98.705
1.012
.964


1.960
5.849


.459
1.984
9.649E9

.159

.426
.486
1.577
.305
1.001

39.018
.976
.930


.656
1.370


15.424
.941
.897


.219
.321









Table 5-12 Power analysis for hypothesis 2a
Tails 1
Odds Ratio .433
a err prob .050
Total Sample Size 521.000
R2 .490
Critical z -1.644
Power .31









Table 5-13 Results of the Logit Analysis for Hypothesis 2a predicting depression
diagnosis, with PCS removed
95% C.l.for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 40193.049 .000 1.000 .000
23.724Constant
23.724


Significant Somatic
Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of
Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia
Diag
Bipolar Diagnosis
Education
Some High
School
High School
Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


-.680

-.182
.513
.275
.285
.211


-.724
.712


.544 1.566
.979
.537 .115
.589 .758
.410 .451
.368 .599
.215 .967


23.372 40193.049

-1.902 .637

-.887 .540

-.760 .600
.490 .860
-1.261 1.324
.001 .001


3.672
-.068


.000
11.205
8.929


2.700 .100


1.601
.324
.907
.521

59.084
14.195
1.364

.072
.716


.485
2.037

1.413E10

.149

.412

.468
1.632
.283
1.001


39.326
.934


.871
1.822


.175 1.470


2.387
5.295
2.941
2.735
1.881


2.410
30.775


.520

1.186

1.517
8.807
3.795
1.004


15.419
.902


.319
.454


100.299
.968


2.383
7.319









Table 5-14 Results of the Logit Analysis for Hypothesis 2b predicting any depression
treatment
95% C.I. for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 1.796 45787.106 .000 1.000 6.028


Significant Somatic
Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis

Schizophrenia Diag

Bipolar Diagnosis
Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


.333


-1.429
-.997
-.135
.409
.235

-1.926

19.801
21.292


.254
1.235


This table shows the results from
predicting depression treatment.


1.087
21931.378


.068
32.320
27.363
6.605
.291
3.884
3.256

3.138
.000


40193.026 .000
2.637
.352 .440


.317
.191
.598
.340
5.427

33.397
18.828
11.475
14.139

.686
11.355


1.091


.076
.999

1.000
.756
.507

.573
.662
.439
.560
.020


.146
.000

1.765E9

1.263

1.208
1.174
.673
.675
1.004


5.077
.957
.969


1.289
3.440


.567 2.096


.409
.789
1.428
2.261
1.632

1.227


.000

.633 2.518


2.926
.938
.951


.707
1.677


2.329
2.415
1.838
2.533
1.007

8.810
.976
.987


2.351
7.057


the logit analysis examining somatic symptoms in









Table 5-15 Results of the Logit Analysis for
treatment, with PCS removed


Hypothesis 2b predicting any depression


95% C.I. for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant -1.040 45780.128 .000 1.000 .353


Significant Somatic
Race (Caucasian)
African American
Other
Ethnicity
Sex
Number of Physical
Comorbidities
Anxiety Diagnosis

Schizophrenia Diag

Bipolar Diagnosis
Education
Some High
School
High School Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


.442

-1.569
-.877
-.334
.426
.288

-2.212

19.703
21.715


.313


1.096
21916.579


2.003
38.170
34.574
5.479
1.906
4.381
4.932

4.070
.000


40193.062 .000
3.221
.345 .779


.549
.116
.594
.252
6.795

30.186
5.516
22.103

7.569
22.063


.843 2.871


.044
.999

1.000
.666
.377

.459
.733
.441
.615
.009


.109
.000

2.695E9

1.356

1.273
1.130
.683
.713
1.004


4.472
.980


2.181
5.223


.351
.867
1.151
2.280
1.719

.939


.000

.689 2.668


2.621
.963


1.251
2.620


2.413
2.278
1.803
2.673
1.008

7.630
.997


3.800
10.410









Table 5-16 Results of
symptoms


the Logit Analysis for Hypothesis 2c predicting significant somatic


95% C.I.for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 9.467 1.236 58.680 .000 12925.902


Race (Caucasian)
African
American
Other
Ethnicity
Sex
Number of
Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia
Diag

Bipolar Diagnosis

Education
Some High
School
High School
Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 PCS
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+
This table shows the
symptoms.


-.105

-.434
-.145
.416
.241


-.631
18.181


23.437


1.004
.307 .117


1.134
21372.373

40192.970


.900 .493 1.644


.969
.191
2.583
1.268


.309
.000


1.537
1.655
2.515
1.936


4.914


.532
7.868E7


.000 1.000


.000 .000


4.786
.498 3.009


.474 3.436 .064


1.300
.073
.000
.214


.156 .394 .156 .692
-.081 .012 43.961 .000
-.087 .012 52.559 .000
1.430 .489

-.366 .330 1.224 .269
-.047 .433 .012 .914
results from the logit analysis examining rac


1.169
.922
.917


.694
.954
e in pred


.159 1.119

.164 1.052


1.519
3.150
13.146
1.002

2.532
.944
.938


.363 1.326
.408 2.231
icting somatic









Table 5-17 Results of the Logit
symptoms, with PCS


Analysis for
removed


Hypothesis 2c predicting significant somatic


95% C.I. for
B S.E. Wald Sig. Exp(B) EXP(B)
Lower Upper
Constant 3.997 .805 24.660 .000 54.448


Race (Caucasian)
African
American
Other
Ethnicity
Sex
Number of
Physical
Comorbidities
Anxiety Diagnosis
Schizophrenia
Diag

Bipolar Diagnosis

Education
Some High
School
High School
Grad
Some College
4 Year College
Post Graduate
Number of PCP
Visits
Psychiatrist visit
SF12 MCS
Age Categories
(18-35)
Age 36-64
Age 65+


-.531

-.269
.282
.340
.455


-.184
17.900


24.451


3.609
.284 3.492


1.108
23095.521

40192.970


.588 .337 1.026


.412
.847
1.980
4.643


.027
.000


1.738
2.420
2.256
2.385


7.307


.832
5.940E7


.000 1.000


3.365
.472 1.759


.000 .000


.535 .212 1.349


.447 2.960 .085


.193 1.113


1.609
.215
.019
.077


.365 .004
.010 38.861
6.152


4.559
4.170


1.397
2.631
7.906
1.003

2.000
.956


3.202
4.782


1.835
2.222









Table 5-18 Results of the Mediation Analysis for Hypothesis 2d
Sig of Change
% Change in from Previous
Race (AA) and Diagnosis Exp(B) Sig. OR from Previous
Model
Unadjusted .973 .959
With Somatic Symptoms .892 .833 8.4% .480
This table shows the change in the odds ratio for the race variable when the somatic
symptoms variable is entered into the model predicting diagnosis.

Table 5-19 Results of the Mediation Analysis for Hypothesis 2e
% han n Sig of Change
% Change in from Previous
Race (AA) and Treatment Exp(B) Sig. OR from Previous
Model
Unadjusted .239 .000
With Somatic Symptoms .240 .000 0.8% .899
This table shows the change in the odds ratio for the race variable when the somatic
symptoms variable is entered into the model predicting treatment.


Table 5-20 Results of the Mediation Analysis for Hypothesis 2d with PCS removed
% Cha n Sig of Change
% Change in from Previous
Race (AA) and Diagnosis Exp(B) Sig. OR from Previous
Model
Unadjusted 1.851 .292
With Somatic Symptoms 1.670 .384 9.8% .731
This table shows the change in the odds ratio for the race variable when the somatic
symptoms variable is entered into the model predicting diagnosis.

Table 5-21 Results of the Mediation Analysis for Hypothesis 2e with PCS removed
Sig of Change
% Change in from Previous
Race (AA) and Treatment Exp(B) Sig. OR from Previous
Model
Unadjusted .409 .017 -
With Somatic Symptoms .416 .019 1.7% .902
This table shows the change in the odds ratio for the race variable when the somatic
symptoms variable is entered into the model predicting treatment.









CHAPTER 6
DISCUSSION

The Surgeon Generals report in 2001 highlighted the several areas of shortcoming

in providing mental health services to minorities within the United States. The report

discussed such concerns as lack of access to mental health services, and even when

access is gained, minorities have less chance of receiving needed mental health

services (DHHS, 2001). This study supports the existence of such disparities, however,

provides only partial evidence into one of the hypothesized reasons for this. The study

also supports the previously published evidence that the Medicaid population as a

whole is vulnerable to under-diagnosis and under-treatment, as evidenced by the

numbers of undiagnosed and untreated Medicaid enrollees in this study.

Depression Diagnosis and Treatment in Medicaid Primary Care

This study supports that depression in Medicaid primary care is markedly under-

diagnosed. While one third of the full survey sample met PHQ-2 criteria for a

depressive disorder at the time of the telephone survey, only 6.9% of the full survey

sample was diagnosed as depressed by their physician during the period of the study.

This supports research that shows that detection in the primary care settings is

suboptimal (Valenstein, Vijan, Zeber, Boehm, and Buttar, 2001) and research that finds

lower rates of depression identification in Medicaid (Melfi, Croghan, and Hanna, 1999).

Among the 706 PHQ-2 identified depression group, their physician diagnosis rates

were also low, 11.1% obtaining a diagnosis during the study period, despite showing

possible signs of depression at the time of the survey. These results indicate that as

much as 90% of depressed Medicaid enrollees may be undiagnosed. Given what we

know about the personal and societal burden of depression, this is a major public health









concern. However, what this study does not address is the number of enrollees who are

being successfully treated, and therefore their symptoms are well managed. This would

mean that they would not be identified using the PHQ-2 as a selection tool, and would

not be included in the final sample, therefore reducing the observed rates of diagnosis

and treatment among the depressed.

While diagnosis rates are low among both the full survey sample and the PHQ-2

identified group, treatment rates are higher. Previous research has shown that of those

who are recognized as depressed by primary care physicians, treatment rates can be

as low as 27% (Tylee, 2006). In this study's full sample, among those who were

diagnosed with depression by their physician (N = 145), 82% received at least one

prescription for antidepressants. A similar pattern is seen among the PHQ-2 identified

depression group, while 11.1% received a diagnosis, 37.1% received treatment during

the same time period. Of the entire survey sample approximately one quarter received

some form of treatment for depression during the course of the study. An interesting

aspect of depression management in this setting are the much higher rates of treatment

than of diagnosis, indicating that treatment frequently occurs in the absence of a

depression diagnosis.

This study is consistent with studies have shown that Medicaid enrollees are also

particularly vulnerable to under-treatment when compared with other populations

(Harman, 2004; Melfi, 1999). However, due to lack of a non-Medicaid comparison

sample in this study it is not possible make this claim.









Racial Disparities in Diagnosis and Treatment

Racial Disparities in Diagnosis

This study did not provide support for racial disparities in the diagnosis of

depression. After controlling for a variety of covariates known to influence diagnosis,

race showed no relationship with the likelihood of being diagnosed with depression by a

physician in primary care. However, this result may be due to the very low numbers of

depressed African Americans who actually received a depression diagnosis in our

sample (N=9). Due to this lack of power, the analysis results may be unreliable In

addition there is also some debate as to the appropriateness of using a tool such as the

PHQ-2 among different cultural groups due to it's focus on more

psychological/emotional questions about depression, which in some studies have

shown a lack sensitivity to different expressions of depression across cultures (Kerr &

Kerr, 2001).

Racial Disparities in Treatment

This study provides clear support for the existence of racial disparities in the

treatment of depression in Medicaid primary care. In the PHQ-2 identified depressed

sample, after controlling for a variety of covariates known to influence treatment, race

was a significant factor in whether an individual obtained treatment or not. The data

indicated that Caucasians had almost 4 times the odds of African Americans of

receiving antidepressants and of receiving any type of treatment for depression. The

lack of effect of mental health visits may again be the result of low numbers as only four

African Americans from our depressed sample received any type of mental health visit

from their primary care physician, while not significant, the odds ratio indicated that

Caucasians have approximately 5 times the odds of African Americans of receiving a









mental health visit. This supports research that consistently shows that African

Americans less frequently obtain treatment for depression (Melfi, Croghan & Hanna,

1999, Harman et al. 2001, Harman, Fortney, & Edlund 2004, Stockdale, Lagomasino,

Siddique, McGuire, & Miranda, 2008).

The fact that the data shows a disparity in treatment but not in initial diagnosis

(perhaps unreliably so due to power issues) may indicate that it is not the initial

identification of need that leads to such disparities, but the initiation of treatment. This

is consistent with research that shows stigma may play a greater role among the

treatment of African Americans, than the treatment of Caucasians. Research has shown

that treatment with antidepressants and individual counseling is less acceptable to

African Americans than Caucasians (Gonzalez, Croghan & West, 2008; Givens, Katz,

Bellamy, & Holmes, 2007). This difference also existed within a primary care setting

(Cooper, Gonzales & Gallo et al, 2003). African Americans also associate greater

stigma with depression treatment than Caucasians in primary care settings (Menke &

Flynn, 2009). The stigma associated with depression treatment may lead to treatment

being less frequently offered to the patient, or to increased rates of refusal of treatment

by the patient. However, this pattern of results may still be consistent with the role of

somatic symptoms in diagnosis and treatment. While the hypothesized relationship

between race and diagnosis is not supported here, increased somatic symptoms could

still impact treatment initiation.

Racial Disparities in Mental Health Expenditures

In line with lower treatment rates for African Americans, this study also found that

African Americans were less likely to have any mental health expenditures, and that

when they did, their expenditures were approximately half that of Caucasians, although









this analysis did not quite reach significance. Based on our previous analyses, this

disparity in expenditures does not stem from a lack of recognition of depression, but

from lower rates of treatment once depression is identified. While saving money for

Medicaid may be viewed as a good thing, the resulting costs of untreated depression

may be far greater. However, it does raise the question about the ability of the system

to support treatment for the large numbers of unidentified and untreated depressed

individuals should they become identified, and obtain treatment.

The Role of Somatic Symptoms in Racial Disparities

Despite the existence of racial disparities in treatment, this study showed limited

support for somatic symptoms masking depression diagnosis. After controlling for all

other covariates, the presence of somatic symptoms of depression was not a significant

predictor in either of the models predicting depression or treatment, although there did

appear to be a trend in one analysis suggesting that with those without somatic

symptoms had approximately twice the odds of diagnosis than those with somatic

symptoms, but this did not reach significance. When the covariate PCS score was

removed from the models it did not lead to different findings. This analysis was also

underpowered due to the low numbers who are diagnoses in this sample.

The lack of a significant relationship between somatic symptoms and diagnosis

could possibly be a result of the overall ill health of this sample and of the Medicaid

population in general. The Medicaid population has generally higher rates of many

disabling physical conditions and poorer physical functioning than the general

population. This is supported in this study by the average score of 36 on the Physical

Component Summary of the SF-12 (PCS), which is lower than the national norm for 75

years old, despite having an average age of 48. This dramatic depiction of this









population indicates that this sample reflects a group with generally very poor physical

wellbeing. In addition 83% of the population endorsed a significant level of somatic

symptoms associated with depression. At such high levels of physical symptoms it is

possible that the physical health problems are so great that they may mask any

variation in the expression somatic symptoms of depression and prevent addressing of

anything other than the wealth of physical problems within patient consultations in

primary care. This is especially true given the average time spent in consultation with

patients in non-academic primary care is only about 10-13 minutes (Tai-Seale, McGuire

& Zhang, 2007). This picture is contrary to the picture painted by the ICD-9 diagnosis

codes in the claims data, with only 20% of the sample having at least one physical

comorbidity. This interesting contrast highlights some of the likely problems with

Medicaid claims datasets. This problem was also seen with depression diagnosis. The

lack of a diagnosis code in the diagnosis fields may not mean a lack of recognition of a

disorder, but just a lack of recording each diagnosis the person carries, as evidenced by

the stark contrast of the PCS score and the comorbidity diagnoses and by patients

being treated for depression in the absence of a diagnosis.

The initial analysis between somatic symptoms and treatment proved non-

significant, however, when PCS was removed from the model the results moved

towards significance (p< .157), but the odds ratio indicated that the effect was in the

reverse direction than hypothesized, with those who endorsed somatic symptoms

having about 1.5 the odds of receiving treatment than those who did not. This is also

particularly interesting in relations to the effects of race and somatic symptoms.









The relationship between somatic symptoms and race initially showed no

significant effect, however, once PCS was removed from the model, the results

changed to becoming very near significance (p < 062). However, this was in the

opposite direction of the hypothesized effect, indicating that Caucasians have

approximately 1.7 the odds of African Americans of endorsing somatic symptoms of

depression. This reversal of effect could be a result of using an unvalidated measure of

somatic symptoms in the study. It is possible that given that the four somatic questions

appear as part of nine questions about depression, anhedonia, suicide, and feeling bad

about oneself, that this may cause a response bias, influenced by the stigma among

African Americans towards depression, leading African American to endorse these

items less frequently.

This pattern of results is interesting, while somatic symptoms may mask diagnosis,

once someone is identified as depressed, somatic symptoms perhaps suggest greater

severity and a greater need for treatment. When this is coupled with Caucasians

increased rates of somatic symptom reporting and their greater likelihood of obtaining

treatment, the presence of somatic symptoms may actually act as a catalyst for the

initiation of treatment.

The final analysis modeled the mediating effect of somatic symptoms between

race and diagnosis and race and treatment also showed a lack of significant effect,

where the addition of somatic symptoms to the model did not change the relationship

between race and diagnosis and race and treatment. The removal of PCS from the

models predicting diagnosis and treatment did not significantly change the value of the

odds ratio of Race, indicating that while there may be some significant findings between









race, somatic symptoms and depression diagnosis and treatment, somatic symptoms

does not play a mediating role in the relationship between race and diagnosis and race

and treatment. Taken together, these analyses and their patterns of results are not easy

to interpret as a whole, partially due to the lack of power in several analyses, and

perhaps partly due to the use of an unvaildated tool to measure somatic symptoms.

However there seems to be some evidence for the masking effects of somatic

symptoms in diagnosis, and a facilitating effect of somatic symptoms in initiating

treatment once it is recognized.

Implications and Recommendations

This study is the first to systematically evaluated the widely accepted somatic

symptoms of depression and their role racial disparities. This study provides further

evidence that African Americans continue to experience racial disparities in the

treatment of depression in Medicaid primary care. The combination of high rates of

depression in this population, 30% in this sample, with the low rate of diagnosis and

treatment results in a very large and costly group of individuals living with undiagnosed

and untreated depression. What this study suggests is that more needs to be done to

improve the identification and treatment of depression in primary care, in low-income

groups, and among African Americans.

Patient education programs should be expanded to cover the symptoms of

depression, especially in clinics that serve low-income communities, such as health

departments or clinics that serve Medicaid recipients. This would aid patients in being

able to recognize their own symptoms of depression, and would also go someway to

challenge the stigma that many still associate with mental illness. To tackle disparities

directly there should be culturally targeted patient education efforts that should highlight









the role of physical symptoms of depression in addition to emotional and psychological

symptoms.

When evaluating patients for depression, physicians should be sure to look for

somatic and neurovegetative symptoms rather than just mood or cognitive symptoms of

depression, regardless of race as there is evidence that these may mask depression.

Automatic screening of patients for depression could be included in the initial

screenings performed by nurses. So while patients are being weighed, and having their

blood pressure checked the nurse could administer one of the brief depression

screeners, such as the PHQ-2 and then notate the chart as with any other health

measure.

Ensuring that primary care practices have best practice guidelines to guide the

treatment either within the primary care setting or to know when to refer for specialized

treatment. This ensures that the physician has the knowledge to be able to follow

through when patients are identified.

All these methods have been used in the past to try to improve the management of

depression in primary care, and their inclusion into a coordinated approach to

depression management is vital. Research suggests that models of integrated care

within primary care clinics are most successful in identifying and treating those with

depression (U.S. Preventive Services Task Force, 2002). Integrative care models on the

whole incorporate mental and behavioral health professions within the primary care

clinic. Levels of collaborations vary within such models, however the most effective

have shown to be those that are highly collaborative and structured in their delivery of

care (U.S. Preventive Services Task Force, 2002). These models of care also improve









the likelihood that the patient will obtain an adequate course of treatment and result in

better overall health outcomes. However currently, few studies have looked at whether

this model of integrative care is effective in reducing racial disparities.

Limitations

This study has several limitations. The first is common among studies that use

administrative datasets. These datasets can be incomplete and may not reflect what is

actually occurring in the course of health service. For example the relatively low rate of

physician diagnosis among this sample compared to the much higher rate identified by

the survey may not be the result of a lack of recognition, it may just be the lack of

documentation. Using a billing diagnosis as a means to track diagnosis therefore leads

to an underestimation of the numbers that are identified by the physicians. Frequently

physicians may just enter one diagnostic code for the purposes of billing, despite

providing treatment for several conditions. This would be consistent with this study's

finding that far more recipients received treatment than received a diagnosis. Records

of treatment with antidepressants within this dataset may be more reliable as these

claims come from claims paid to the pharmacy who must document all medications

provided to enable them to be reimbursed, whereas physicians are not required to

include every diagnosis to be reimbursed.

Another limitation linked to what is actually occurring in the primary care

consultation is that endorsing items on the PHQ-2 and the somatic items on the PHQ-9

over the telephone during a survey, does not mean that the enrollees reported these

symptoms while consulting with their primary care physician. This means that while this

study relies on the survey endorsed symptoms being reported to the physician, which is

then assumed to influence the physicians to diagnose or treat, there is no way to know









what type of information is shared with the physician. In order to be sure, future studies

must somehow gain access to this information, for example, by record review or by

some other study design that accesses consultation specific information.

While the PHQ-2 is a valuable tool for use in primary care to improve efforts to

identify depression, its use by this study may have resulted in an over-inclusive sample,

or the inclusion of a number false positives. This means that the study sample may

include individuals who do not have a depressive disorder, and also result in an over-

estimate the prevalence rate among the Medicaid primary care population, and an

overestimate of under diagnosis and under treatment.

Another common limitation among studies concerned with specific populations is a

lack of generalizable. Given the very specific sample of individuals, Medicaid enrollees

who are were seen in primary care, is unlikely that these findings could be taken as

representative of other populations, for example inpatients or privately insured

individuals, or as representative of the population as a whole, due to the unique

characteristics of Medicaid recipients and of service in primary care settings.

Another limitation of this study is the lack of power in several of the analyses.

While the sample was a good size, approximately 700 Medicaid enrollees, the low rates

of physician diagnosis among African Americans and the low rates of mental health

visits among African Americans weakened the power of several analyses. When

studies examine an event that is very rare the sample size must be large enough in

order to obtain sufficient power, however this was not possible in this study given that

the dataset was part of a completed larger study.









Finally, as mentioned previously, one of the studies primary variables, somatic

symptoms of depression as obtained from the PHQ-9, has never been validated for use

in this way, and has never been used in this way in other studies. It is possible that this

is not the best means to assess for the somatic symptoms of depression, and that other

important aspects of depression were not included, for example gastric problems, and

pain.

Despite these limitations this study tells a powerful story about the quality of life of

this population, and about the existence of racial disparities in this already vulnerable

population.

Future research

This study suggests that several areas of research need to be developed. While

this study attempted to use large-scale health services research techniques to examine

the problem of racial disparities, and despite using a large administrative data set, the

study still encountered problems of lack of power and low numbers. While the odds

ratios in several cases were very large, indicating a strong effect, the lack of numbers

could not provide the study with statistical significance. Future research must include

even larger sample sizes to enable enough power to support these hypotheses.

This study also encountered problems with the validity of the Somatic Symptoms

measure. Future studies should include the DSM-IV criteria used by the PHQ-9, but

should also include wider measures of somatic symptoms such as headache,

migraines, sexual dysfunction, menstrual-related symptoms, chronic pain, digestive

problems (e.g., diarrhea, constipation), and fatigue (Kerr & Kerr, 2001).

There is a also need for a systematic validation of existing depression measures

across racial and ethnic groups and further exploration and development of tools









specifically designed to assess depression among minorities, that may include broader

concepts such as worry, feeling pressured, feeling empty, or feeling cut-off.

More generally, there is clearly a need to examine collaborative models of care in

primary care, and their effectiveness in reducing racial disparities by improving

diagnosis and treatment rates across a wider array of populations and settings,

including settings that provide care to low-income populations. Such models of service

are also valuable when serving remote communities and show promise in dealing with

the myriad of behavioral health issues such as heart disease, diabetes, and high blood

pressure, which are frequently comorbid with depression.









APPENDIX
BMS DEPRESSION SURVEY

Florida Center for Medicaid & the Uninsured
Survey to be conducted with adult enrollees in the Florida Medicaid program in the Spring of
2004.

Interviewer notes/administration instructions in italics.
Instrument names & notes for researcher's use only in
Field names in ALL CAPS to the left of each item.
For all items: -8 = Don't Know, -9 = Refused. Unless noted, the skip sequences for these options are the
same as for "No" or "Disagree."

Programmer note: please use the field names and response category values indicated
on this hard copy!

HELLO Hello. My name is _, and I am calling from the Survey Research Center
at the University of Florida.

ADULTA May I speak to (target name)?
this!>

If yes, reintroduce yourself if necessary

If no, Is there another time I could call back to talk to him/her? Schedule
a call back if necessary and thank the respondent for his/her time.

INTRO I am calling you today because I'm doing a scientific research study on the health of
people in Florida's Medicaid program. If you're willing to complete my survey, I will
send you a $5 Wal-Mart gift certificate in the mail. My survey usually takes about 10
minutes and it has a lot of questions about whether you have had different physical and
emotional problems recently.

to only a very small group of people. Most people will opt-out of around 75% of these
questions. In our trials, the average administration time was 10 minutes.>

ADVISE Ijust need to tell you a couple of things before we get started. You do not have to answer
any question you don't want to. You don't have to participate in the survey at all. No
one not even Medicaid will know if you participated or not, and your name will not be
reported to anyone else. The results from our study will be reported back to Florida
Medicaid and may be published in a scientific journal. Ok, let's get started.

1. Yes (go to Survey)
2. No (go to SORRY)


SORRY Ok. Thank you very much for your time. (End of Interview)












BMS1 My first group of 9 questions is about some problems you may or may not have
had over that past 2 weeks. For each question. I'm going to ask you how often
you've had that problem.

Over the last 2 weeks, how often have you been bothered by: having little
interest or pleasure in doing things. Have you been bothered: not at all, several
days in past 2 weeks, more than half the days in the past 2 weeks, or nearly
every day in the past 2 weeks?

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS2 feeling down, depressed, or hopeless? (Interviewer Remind as necessary that
the time period is the last 2 weeks. Read response categories as necessary.)

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS3 trouble falling or staying asleep, or sleeping too much? (Interviewer Remind as
necessary that the time period is the last 2 weeks. Read response categories as
necessary.)

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS4. feeling tired or having little energy? (Interviewer: Remind as necessary that the
time period is the last 2 weeks. Read response categories as necessary.)

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS5. poor appetite or overeating (Interviewer: Remind as necessary that the time
period is the last 2 weeks. Read response categories as necessary.)









1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS6. feeling bad about yourself or that you are a failure or have let yourself or your
family down? (Interviewer Remind as necessary that the time period is the last
2 weeks. Read response categories as necessary.)

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused


BMS7. Trouble concentrating on things, such as reading the newspaper or watching
television? (Interviewer: Remind as necessary that the time period is the last 2
weeks. Read response categories as necessary.)


1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS8. Moving or speaking slowly that other people could have noticed? (Interviewer:
Remind as necessary that the time period is the last 2 weeks. Read response
categories as necessary.)

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused

BMS9. Thoughts that you would be better off dead or of hurting yourself in some way?
(Interviewer: Remind as necessary that the time period is the last 2 weeks.
Read response categories as necessary.)

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day
-8. Don't know
-9. Refused











BMS 10 Programmer: This question should be asked of everyone unless they answered
1, -8, or-9 to all nine questions above. You indicated that you had a problem
with . How
difficult problems made it for you to do your work, take
care of things at home, or get along with other people?

1. Not difficult at all
2. Somewhat difficult
3. Very difficult
4. Extremely difficult
-8. Don't Know
-9. Refused





NOTE: There was a mistake in the CATI program that I did not catch when I proofed the
program. One of the SF-36 questions was left out. Luckily, however, the question that
was left out was not one of the SF-12 items. So we have to roll up to the SF-12 level.

BMS11 Now I'm going to ask you some questions about your general health. This
information will help keep track of how you feel and how well you are able to do
your usual activities.

In general, would you say your health is excellent, very good, good, fair, or poor?

1. Excellent
2. Very good
3. Good
4. Fair
5. Poor
-8/-9 Don't Know/Refused

BMS12 Compared to one year ago, how would you rate your health in general now? Is it
much better now that one year ago, somewhat better now than one year ago,
about the same as one year ago, somewhat worse now than one year ago, or
much worse now than one year ago?

1. Much better
2. Somewhat better
3. About the same
4. Somewhat worse
5. Much worse

BMS13 The following questions are about activities you might do during a typical day.
I'm going to ask you if your health limits you in these activities, and if so, how
much?

Does your health limit you in vigorous activities, such as running, lifting heavy
objects, or participating in a strenuous sport?










1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Does your health limit you in moderate activities, such as moving a table,
pushing a vacuum cleaner, bowling, or playing golf?

1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Does your health limit you in Lifting or carrying groceries?


1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Does your health limit you in Cl

1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Does your health limit you in Be

1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Does your health limit you in W

1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Walking several hundred yards

1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Walking one hundred yards

1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

Bathing or dressing yourself


imbing several flights of stairs?





!nding, kneeling or stooping





walking more than a mile


BMS14






BMS15


BMS16





BMS17





BMS18





BMS19





BMS20





BMS21









1. Yes, limited a lot
2. Yes, limited a little
3. No, not limited at all

BMS22 Now I'm going to ask you about some problems you may have had with your
work or daily activities as a result of your physical health.

In the last 4 weeks, how much of the time have you cut down the amount of time
you spent on work or other activities as a result of your physical health?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS23 In the last 4 weeks, how much of the time have you accomplished less than you
would like as a result of your physical health?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS24 In the last 4 weeks, how much of the time were you limited in the kind of work or
other
activities you did as a result of your physical health?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS25 In the last 4 weeks, how much of the time have you had difficulty performing work
or other
activities (for example it took extra effort) as a result of your physical health

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS26 Now I'm going to ask you some questions problems you may have had with your
work or daily activities as a result of any emotional problems such as feeling
depressed or anxious.

In the last 4 weeks, how much of the time have you cut down the amount of time
you spent on work or other activities as a result of any emotional problems?









1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS27 In the last 4 weeks, how much of the time have you accomplished less than you
would like as a result of any emotional problems?
1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time


BMS28 In the last 4 weeks, how much of the time did you do work or other activities less
carefully than usual as a result of any emotional problems?
1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS29 During the past 4 weeks, to what extent has your physical health or emotional
problems interfered with your normal social activities with family, friends,
neighbors, or groups?

1. Not at all
2. Slightly
3. Moderately
4. Quite a bit
5. Extremely


BMS30 How much bodily pain have you had during the past 4 weeks?

1. None
2. Very mild
3. Mild
4. Moderate
5. Severe
6. Very Severe

BMS31 During the past 4 weeks, how much did pain interfere with your normal working
(including both work outside the home and housework):

1. Not at all
2. Slightly
3. Moderately
4. Quite a bit
5. Extremely












BMS32 These questions are about how you feel and how things have been with you
during the past 4 weeks. For each question, please give the one answer that
comes closest to the way you have been feeling.


How much of the time during the past 4 weeks did you feel full of life?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS33 How much of the time during the past 4 weeks have you been very nervous?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS34 How much of the time during the past 4 weeks have you felt so down in the
dumps that
nothing could cheer your up?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS35 How much of the time during the past 4 weeks Have you felt calm and peaceful ?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS36 How much of the time during the past 4 weeks Did you have a lot of energy?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS37 How much of the time during the past 4 weeks Have you felt downhearted and
depressed?











1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS38 How much of the time during the past 4 weeks Did you feel worn out?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS39 How much of the time during the past 4 weeks Have you been happy?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS40 How much of the time during the past 4 weeks Did you feel tired?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS41 During the past 4 weeks, how much of the time has your physical health or
emotional problems interfered with your social activities (like visiting friends,
relatives, etc.)?

1. All of the time
2. Most of the time
3. Some of the time
4. A little of the time
5. None of the time

BMS42 Now, I am going to read some statements to you. I would like to know how
TRUE or FALSE each statement is for you. Your choices will be definitely true,
mostly true, don't know, mostly false, and definitely false.

How true or false is this statement: I seem to get sick a little easier than other
people
1. Definitely true
2. Mostly true
3. don't know
4. mostly false


101









5. definitely false

How true or false is this statement: I am as healthy as anybody I know

1. Definitely true
2. Mostly true
3. don't know
4. mostly false
5. definitely false


true or false is this
definitely true
mostly true
don't know
mostly false
definitely false

true or false is this
definitely true
mostly true
don't know
mostly false
definitely false


statement: I expect my health to get worse






statement: my health is excellent





CIDIINTR Now I'm going to ask you some questions about your emotional health.




During The Past 12 months, was there ever a time when you felt sad, blue, or
depressed for two weeks or more in a row?


No (Go To CIDI-A9)
(If Volunteered) I was on Medication/Anti-Depressants


CIDI-A1A For the next few questions, please think of the two-week period during the past
12 months when
these feelings were worst. During that time did the feelings of being sad, blue, or
depressed usually
last all day long, most of the day, about half the day, or less than half the day?


All Day Long
Most
About Half (Go To CIDI-A9)
Less Than Half (Go To CIDI-A9)


CIDI-A1B During those two weeks, did you feel this way every day, almost every day, or I
ess often?


BMS43.


BMS44






BMS45


CIDIA1











1. Every Day
2. Almost Every Day
3. Less Often (Go To CIDI-A9)

CIDI-A1C During those two weeks did you lose interest in most things like hobbies, work, or
activities that
usually give you pleasure?

1. Yes
5. No

CIDI-A1D Thinking about those same two weeks, did you feel more tired out or low on
energy than is usual
for you?

1. Yes
5. No

CIDI-A2 Did you gain or lose weight without trying, or did you stay about the same?
Interviewer: If R asks: "Are we still talking about the same two weeks?" Answer:
"Yes."

1. Gained
2. Lost
3. (If Volunteered) Both gained and lost weight
4. Stay about the same (Go to CIDI-A3)
5. (If Volunteered) R was on a diet (Go to CIDI-A3)

CIDI-A2A About How Much Did (You Gain/You Lose/Your Weight Change)?

SPOUNDS Interviewer: Accept a range response

Cl DI-A2 B Interviewer: Did R's weight change by 10 pounds or more?
the interviewer does not have to stop and do this. >

1. Yes
5. No

CIDI-A3 Did you have more trouble falling asleep than you usually do during those two
weeks?
1. Yes
5. No (Go To CIDI-A4)

CIDI-A3A Did that happen every night, nearly every night, or less often during those two
weeks?

1. Every night
2. Nearly every night
3. Less often











CIDI-A4 During those two weeks, did you have a lot more trouble concentrating than
usual?
Interviewer If R Asks: "Are we still talking about the same two weeks?" Answer
"Yes."

1. Yes
5. No

CIDI-A5 People sometimes feel down on themselves, no good, or worthless. during that
two week period, did you feel this way?
Interviewer If R Asks: "Are we still talking about the same two weeks?" Answer:
"Yes."

1. Yes
5. No

CIDI-A6 Did you think a lot about death either your own, someone else's, or death in
general during those two
weeks?
Interviewer: If R Asks: "Are we still talking about the same two weeks?" Answer:
"Yes."

1. Yes
5. No


CIDI-A7 Interviewer Checkpoint (Count Qualifying Responses In CIDI-A 1c
Throughcidi-A6. Qualifying Responses Are As Follows: CIDI-A 1D=1, CIDI-
A1D=1, CIDI-A2B=1, CIDI-A3A=1 Or2, CIDI-A4=1, CIDI-A5=1, And CIDI-A6=1)

Programmer- Please construct an algorithm to do this so that the
interviewer doesn't have to stop and do it.

1. Zero Qualifying Responses (Go To CIDI-B1)
2. One Or More Qualifying Responses


CIDI-A8 To review, you had two weeks in a row during the past 12 months when you
were sad, blue, or depressed and also had some other feelings or problems like
(Read Up To The First Three "Yes" Responses To CIDI-A 1C through CIDI-A6).

About how many weeks altogether did you feel this way during the past 12
months?

# Of Wks

52 (If Volunteered) Entire year (Go To C DI-A8B)

CIDI-A8a Think about this most recent time when you had two weeks in a row when you
felt this way. How










long ago was that?


Months in the past

CIDI-A8B Did you tell a doctor about these problems? (By "Doctor" I mean either a medical
doctor or
osteopath, or a student in training to be either a medical doctor or osteopath.)

1. Yes
5. No

CIDI-A8c Did you tell any other professional (such as a psychologist, social worker,
counselor, nurse,
clergy, or other helping professional)?

1. Yes
5. No

CIDI-A8D Did you take medication or use drugs or alcohol more than once for these
problems?

1. Yes
5. No

CIDI-A8E How much did these problems interfere with your life or activities a lot, some,
a little,
or not at all

1. A Lot (Go To CIDI-B1)
2. Some (Go To CIDI-B1)
3. A Little (Go To CIDI-B1)
4. Not At All (Go To CIDI-B 1)

CIDI-A9 During the past 12 months, was there ever a time lasting two weeks or more
when you lost interest in
most things like hobbies, work, or activities that usually give you pleasure?

1. Yes
5. No (Go To CIDI-B1)
6. (If Volunteered) I Was On Medication/Anti-Depressants

CIDI-A9A For the next few questions, please think of the two-week period during the past
12 months when
you had the most complete loss of interest in things. During that two-week
period, did the loss of
interest usually last all day long, most of the day, about half the day, or less than
half the day?

1. All Day Long
2. Most
3. About Half (Go to CIDI-B1)










4. Less Than Half (Go To CIDI-B1)

CIDI-A9B Did you feel this way every day, almost every day, or less often during the two
weeks?


Every Day
Almost Every Day
Less Often (Go to CIDI-B1)


CIDI-A9C During those two weeks, did you feel more tired out or low on energy than is
usual for you?


Did you gain or lose weight without trying, or did you stay about the same?
Interviewer If R asks: "Are we still talking about the same two weeks?" Answer
"Yes."


1. Gain
2. Lose
3. (If Volunteered) Both Gained and Lost Weight
4. Stay About the Same (Go to CIDI-A 11)
5. (If Volunteered) R Was on a diet (Go to CIDI-A 11)

About how much did (you gain/you lose/your weight change)?


Pounds
Interviewer: Accept a range response.


Interviewer: Did R's weight change by 10 pounds or more?
Programmer- Please construct an algorithm to do this so that the
interviewer doesn't have to stop and do it.


Did you have more trouble falling asleep than you usually do during those two
weeks?


Yes
No (Go To CIDI-A12)


Did that happen every night, nearly every night, or less often during those two
weeks?


Every night
Nearly every night
Less often


CIDI-A10


CIDI-A10A


CI DI-A10B


CIDI-Al1


CIDI-A11A










CIDI-A12 During those two weeks, did you have a lot more trouble concentrating than
usual?
Interviewer If R Asks: "Are we still talking about the same two weeks?" Answer
"Yes."

1. Yes
5. No

CIDI-A13 People sometimes feel down on themselves, no good, or worthless. did you feel
this way during that two week period?
Interviewer If R Asks: "Are we still talking about the same two weeks?" Answer:
"Yes."

1. Yes
5. No

CIDI-A14 Did you think a lot about death either your own, someone else's, or death in
general during
those two weeks?
Interviewer: If R Asks: "Are we still talking about the same two weeks?" Answer:
"Yes."

1. Yes
5. No

CIDI-A 15 Interviewer Checkpoint (Count "Yes" Responses In CIDI-A9C through CIDI-
A14)

1. Zero "Yes" Responses On CIDI-A9C, CIDI-A 12, CIDI-A 13, CIDI-A 14, and
(EitherCIDI-A 10=4-5 or CIDI-A 10A is less than 10 pounds) and (Either
CIDI-A 11=5 or CIDI-A 11A =3)

Go To CIDI-B1

2. All Others (Go to CIDI-A16)

Programmer- Please construct an algorithm to do this so that the
interviewer doesn't have to stop and do it.


CIDI-A16 To review, you had two weeks in a row during the past 12 months when you lost I
interest in most things
and also had some other things like (Read up to the first 3 "Yes" responses to
CIDI-A9c through CIDI-A14). About how many weeks did you feel this way
during the past 12 months?

# Of Wks

52. (If Volunteered) Entire Year (Go To CIDI-A 16B)










Think about this most recent time when you had two weeks in a row when you
felt this way. How
long ago was that?


Months In The Past


Did you tell a doctor about these problems? (By "Doctor" I mean either a medical
doctor or
osteopath, or a student in training to be either a medical doctor or osteopath.)


1. Yes
5. No


Did you tell any other professional (such as a psychologist, social worker,
counselor, nurse,
clergy, or other helping professional)?


Did you take medication or use drugs or alcohol more than once for these
problems?


1. Yes
5. No


How much did these problems interfere with your life or activities a lot, some,
a little,
or not at all?
1. A Lot
2. Some
3. A Little
4. Not At All




CIDI-B1 During the past 12 months, did you ever have a period lasting one month or
longer when most of the time you felt worried, tense, or anxious?


Yes (Go to CIDI-B2)
No


CIDI-B1A People differ a lot in how much
the past
12 months when you worried a
situation?


they worry about things. did you have a time in

lot more than most people would in your


Yes
No (Go To CIDI-E1)


CIDI-A16A


CIDI-A16B


CIDI-A16C


CIDI-A16D


CIDI-A16E










CIDI-B2. Has that period ended or is it still going on?

1. Ended
2. Still going on (Go To CIDI-B2B)

CIDI-B2A How many months or years did it go on before it ended?

# of months or go to CIDI-B3

# of years go to CIDI-B3
89 (If Volunteered) "All My Life" or "As Long As I Can Remember" (Go to
CIDI-B3)

CIDI-B2B How many months or years has it been going on?

# of months or

# of years
89 (If Volunteered) "All My Life" or "As Long As I Can Remember"

CIDI-B3 INTERVIEWER CHECKPOINT

1. CIDI-B2A/CIDI-B2P is six months or longer, or (If Volunteered) "All My
Life" Or "As Long As I Can Remember"
2. CIDI-B2A/CIDI-B2P is less than six months (Go to CIDI-E1)

Programmer- Please construct an algorithm to do this so that the
interviewer doesn't have to stop and do it.


CIDI-B4 (During that period, was your/is your) worry stronger than in other people?

1. Yes
5. No

CIDI-B5 (Did/Do) you worry most days?

1. Yes
5. No

CIDI-B6 (Did/Do) you usually worry about one particular thing, such as your job security
or the failing health of a loved one, or more than one thing?

1. One Thing
2. More than one thing

CIDI-B7 (Did/Do) you find it difficult to stop worrying?

1. Yes
5. No










CIDI-B8


CIDI-B12


CIDI- B12A


CIDI- B12P


When you (were/are) worried or anxious, (were/are) you also...
1. Yes
5. No

Restless?


1. Yes
5. No


(Were/Are) you keyed up or on edge?


CIDI- B12c. (Were/Are) you easily tired?

1. Yes
5. No

CIDI- B12d. (Did/Do) you have difficulty keeping your mind on what you (were/are) doing?


(Did/Do) you ever have different worries on your mind at the same time?

1. Yes
5. No

How often (was/is) your worry so strong that you (couldn't/can't) put it out of your
mind no matter how hard you (tried/try) often, sometimes, rarely, or never?

1. Often
2. Sometimes
3. Rarely
4. Never

How often (did/do) you find it difficult to control your worry often, sometimes,
rarely, or never?

1. Often
2. Sometimes
3. Rarely
4. Never

What sort of things (did/do) you mainly worry about? (Probe: Any other main
worries?)


CIDI-B9


CIDI-B10


CIDI-B11










CIDI- B12e. (Were/Are) you more irritable than usual?


CIDI- B12f. (Did/Do) you have tense, sore, or aching muscles?


CIDI- B12g. (Did/Do) You Often Have Trouble Falling Or Staying Asleep?

1. Yes
5. No

CIDI-B13. CHECKPOINT


CIDI-B14


CIDI-B15


CIDI-B16


CIDI-B17
lot, some, a


1. 0-1 Yes responses in the CIDI-B12 series (Go to CIDI-E1)
2. All others

Programmer- Please construct an algorithm to do this so that the
interviewer doesn't have to stop and do it.


Did you tell a doctor about your worry or about the problems it was causing? (by
"Doctor" I mean either
a medical doctor or osteopath, or a student in training to be either a medical
doctor or osteopath.)

1. Yes
5. No

Did you tell any other professional (such as a psychologist, social worker,
counselor, nurse, clergy, or
other helping professional)?

1. Yes
5. No

Did you take medication or use drugs or alcohol more than once for the worry or
the problems it was
causing?

1. Yes
5. No

How much (did/does) the worry or anxiety interfere with your life or activities a
little, or
not at all?
1. A Lot
2. Some









A Little
Not At All





CIDI-E1


The next questions are about things that make some people so afraid that they
avoid them, even when there is no real danger.


Do you have an unreasonably strong fear or avoid any of the following things...

CIDI-E1A First, being in a crowd or standing in line? (Do you have an unreasonably strong
fear or avoid either of these situations?)


CIDI-E1B (How about) being away from home alone? (Do you have an unreasonably
strong fear or avoid this situation?)


CIDI-E1C (How about) traveling alone? (Do you have an unreasonably strong fear or avoid
this situation?)


CIDI-E1D (How About) traveling in a bus, train, or car? (Do you have an unreasonably
strong fear or avoid any of these situations?)

1. Yes
5. No

CIDI-E1E (How about) being in a public place like a department store? (Do you have an
unreasonably strong fear or avoid this type of situation?)


CI DI-E2


CIDI-E3


INTERVIEWER CHECKPOINT -See CIDI-E1A through CIDI-E1E

1. One or more "Yes" responses in CIDI-E1A through CIDI-E1E
2. All others (Go to CIDI-F1)

Thinking only of the situations) that we just reviewed that causes) you
unreasonably strong fears, do
you get very upset every time you are in (this/these) situationss, most of the
time, only some of the









time, or never?


Every Time
Most Of The Time
Some Of The Time (Go to CIDI-F1)
Never (Go to CIDI-F1)
(If Volunteered) Only one or two times ever (Go to CIDI-F1)


CIDI-E4 How long have you had (this/these) fear(s) less than 1 year, between 1 and 5
years, or more than 5
years?


Less Than 1 Year
Between 1 And 5 Years (Go to CIDI-E5)
More Than 5 Years (Go to CIDI-E5)


CIDI-E4A About how many months?

Number Of Months

CIDI-E5 When you are in (this/these) situationss, are you afraid that you might faint, lose
control, or embarrass
yourself in other ways?

1. Yes
5. No


CIDI-E6 When you are
without any way to
escape?


CIDI-E7 When you are
available if you needed
it?


CIDI-E8


in (this/these) situationss, do you worry that you might be trapped


in (this/these) situationss, do you worry that help might not be


During the past 12 months, how much did (this/these) fear(s) interfere with your
life or activities a lot, some, a little, or not at all?


A Lot
Some
A Little
Not At All














CIDI-F1 During the past 12 months, did you ever have a spell or an attack when all of a
sudden you felt
frightened, anxious, or very uneasy?

1 Yes
5 No (Go to PAININTR)

CIDI-F1A Did any of these attacks occur when you were in a life-threatening situation?

1. Yes
5. No (Go to CIDI-F2)
8. (If Volunteered) Don't Know (Go to CIDI-F2) Programmer- For this
question only, please use 8 and not-8 for the Don't Know category.

CIDI-F1B Did any of these attacks occur when you were not in a life-threatening situation?

1. Yes
5. No (Go to PAININTR)

CIDI-F2 About how many attacks did you have in the past 12 months?

Number

CIDI-F3 How long ago did you have the most recent (one/attack)?

Months In The Past

CIDI-F4 Did (this attack/all of these attacks) happen in a situation when you were not in
danger or not the
center of attention?

1. Yes
5. No (Go To PAININTR)

CIDI-F5 A moment ago, we discussed situations that cause unreasonably strong fears.
when you have
attacks of the sort you just described, do they usually occur in situations that
cause you unreasonably
strong fear?

1. Yes
5. No (Go To CIDI-F6)

CIDI-F5A Did you ever have an attack in the past 12 months when you were not in a
situation that usually causes you to have unreasonably strong fears?

1. Yes
5. No (Go To PAININTR)

CIDI-F6 When you have attacks, ...











CIDI-F6A ...Does your heart pound or race?

1. Yes
5. No


CIDI-F6B ...Do

1.
5.

CIDI-F6C ...Do

1.
5.

CIDI-F6D ...Do

1.
5.

CIDI-F6E ...Do

1.
5.


CIDI-F6F


you have tightness, pain, or discomfort in your chest or stomach?

Yes
No

you sweat?

Yes
No

you tremble or shake?

Yes
No

you have hot flashes or chills?

Yes
No

you, or things around you, seem unreal?

Yes
No




PAININTR Now I have some questions for you about bodily pain that you might have had.

PAIN1. Have you experienced pain lasting for more than two weeks?

1. Yes
2. No = 2 (Go to DEMOINTR)

PAIN2. How much did the pain bother you? (Did it bother you extremely, quite a bit,
moderately, very little, or not at all?)
1. Extremely
2. Quite a bit
3. Moderately
4. Very little (Go to DEMOINTR)
5. Not at all (Go to DEMOINTR)
-8/-9 Don't Know/Refused (Go to DEMOINTR)

PAIN3 Have you sought treatment for your pain?











(Go to PAIN5)


I'm going to read a list of possible reasons why you
Please tell me which of the following apply to you.

Didn't think I needed treatment
Received treatment for this in the past
Couldn't afford treatment
Didn't know where to find treatment
Couldn't get an appointment when I could go
I was refused treatment when I could get it
I don't feel comfortable speaking English and
couldn't find treatment where they spoke
my language
Some other reason (specify )


did not seek treatment.


PAIN4


PAIN4A
PAIN4B
PAIN4C
PAIN4D
PAIN4E
PAIN4F
PAIN4G


PAIN4H


Go to PAIN12

How soon did you get medical help for this problem after it started?


field)


(Specify number of days, weeks or months, each in a separate


How long did the pain last?

1. 2 To 4 Weeks
2. 1 To 6 Months
3. More Than 6 Months

Do you know what caused your pain?

1. Yes
2. No (Go toPAIN9)


What Was It That Caused Your Pain?

(Specify- e.g., Headache, Backache, GI
Problems, Etc.)


Has your provider offered you any of the following
your pain?


medications or therapies for


Elavil (Amitriptiline), Desipramine, Nortryptiline, or Another Tricyclic
Antidepressant Drug
1. Yes 2. No


1. Yes 2. No
1. Yes 2. No


PAIN5


PAIN6


PAIN7


PAIN8


PAIN9




PAIN9A










PAIN9B Aspirin, Tylenol, Motrin, or other Aspirin-like drugs
1. Yes 2. No
PAIN9C Mexiletine (Mexitil)
1. Yes 2. No
PAIN9D Capsaicin (Capsin, A Lotion Made From Chili Peppers)
1. Yes 2. No
PAIN9E Mild Opiate drugs Like Codeine, Vicodin, Percodin, or Percoset
1. Yes 2. No
PAIN9F Strong Opiate Drugs like Morphine, Dilaudid, or Demerol
1. Yes 2. No
PAIN9G Alternative therapies like acupuncture, acupressure, massage, visualization,
or herbal remedies
1. Yes 2. No
PAIN9H Another medication
1. Yes 2. No
PAIN91 Was not offered medication
1. Yes (Go to PAIN12) 2. No


PAIN10 What best describes how well the medications or therapies controlled your pain
most of the time? Would you say the medications or therapies:

1. Didn't Help
2. Slightly Relieved Your Pain
3. Significantly Relieved Your Pain
4. Completely Relieved Your Pain


PAIN11 How satisfied were you with your provider's efforts to control your pain? Would
you say:

1. Completely Satisfied
2. Mostly Satisfied
3. Somewhat Satisfied
4. A Little Satisfied
5. Or Not At All Satisfied

PAIN12 During the past 4 weeks, how many days did pain cause you to stay in bed for a
half day or more?

# of days:

DEMOINTR Ok. We're almost finished now. I just have a few last questions.

HEALTH In general, how would you rate your overall health now?

1. Excellent
2. Very Good
3. Good
4. Fair
5. Poor











DOB What is your date of birth? Programmer- This can be entered as a single field,
or as a separate field for month, day, and year. Use whichever format is easiest
for SAS programming.

SEX Are you male or female?

1. Male
2. Female

EDUCAT What is the highest grade or level of school that you have completed?
1. 8th grade or less
2. Some high school, but did not graduate
3. High school graduate or GED
4. Some college or 2-year degree
5. 4-year college graduate
6. More than 4-year college degree

ETHNIC Are you of Hispanic or Latino origin or descent?
1. Hispanic or Latino
2. Not Hispanic or Latino

RACE I am going to read you a list of race categories, and I'd like you to tell me which one or
ones you think describe you best. Just so you know, the reason I'm asking you about this
is because the researchers want to make sure they have gathered the opinions of enough
people from all different races and ethnicities.

Here is the list. Tell me yes or no for each category. Are you:

RACEWH White 1. Yes 2. No
RACEBL Black or African American 1. Yes 2. No
RACEAS Asian 1. Yes 2. No
RACEHAW Native Hawaiian or other Pacific Islander 1. Yes 2. No
RACEIND American Indian or Alaska Native 1. Yes 2. No

LANGUAGE What language do you mainly speak at home?
1. English
2. Spanish
3. Some other language (Specfy

INCENT1 Ok. You've answered all my questions now. Thank you very much. Now I just
need to get your name and address so I can send you your $5 WalMart card.

Programmer Please create separate fields for first name, last name, address
line 1,address line 2, apartment #, city, state, and zip.

INCENT2 Let me read all of that back to you so that we can be sure it's right.









Programmer set up a screen that will display the name and address info again
so the interviewer can confirm it.

INCENT3 Ok. Your gift card will arrive within 8 weeks. Just one more thing. We have
been talking about some very sensitive subjects today. Some people might feel
upset after talking about these things. If you do, there is a number you can call
to talk about these feelings. Would you like me to give you that number?

1. Yes
2. No (go to OUT)

NUMBER The number is 1-888-784-2433. You can call 24 hours a day, 7 days a week.

OUT Ok. Thank you very much for your time.









LIST OF REFERENCES


Andrews, G. & Peters, L. (1998). The psychometric properties of the Composite
International Diagnostic Interview. Social Psychiatry and Psychiatric
Epidemiology, 33, 80-88.

APA (1994). Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,
(DSM-IV).

Bennett, M. B. (1987). Afro-American women, poverty and mental health: A social
essay. Women and Health, 12, 213-228.

Baron, R. M. & Kenny, D. A. (1986). The moderator-mediator variable distinction in
social psychological research: conceptual, strategic, and statistical
considerations. Journal of Personality and Social Psychology, 51, 1173- 1182.

Borowsky, S., Rubenstein, L. V., Meredith, L. S., Camp, P., Jackson-Triche, M., &
Wells, K.B. (2000). Who is at risk for non-detection of mental health problems in
primary care? Journal of Global Information Management, 15,381-388.

Brown, C., Schulberg, H. C., & Madonia, M. J. (1996). Clinical presentations of major
depression by African Americans and whites in a primary medical care practice.
Journal of Affective Disorders, 41, 181-191.

Brown, E. R., Ojeda, V. D., Wyn, R., & Levan, R. (2000). Racial and ethnic disparities in
access to health insurance and health care. UCLA Center for Health Policy
Research and Kaiser Family Foundation, April 2000. Retrieved July 1, 2010, from
www.kff.org.

Centers for Medicaid & Medicare Services (2007). Medicaid Statistical Information
System Tables FY 2004. Retrieved July 1, 2010, from
http://www.cms.hhs.gov/MedicaidDataSourcesGenlnfo/02_MSISData.asp.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ: Lawrence Earlbaum Associates.

Cook, B. L., McGuire, T., & Miranda. J. (2007) Measuring trends in mental health care
disparities, 2000-2004. Psychiatrc Services, 58, 1533-1540.

Cooper, L. A., Gonzales, J. J., Gallo, J.J., Rost, K. M., Meredith, L. S., Rubenstein, L.
V., Wang, N. Y., & Ford, D.E. (2003). The acceptability of treatment for
depression among African-American, Hispanic, and white primary care patients.
Medical Care, 41, 479-448.

Crystal, S., Sambamoorthi, U., & Walkup, J. T. (2003) Diagnosis and treatment of
depression in the elderly Medicare population: predictors, disparities, and trends.
Journal of the Amercan Geriatrics Society, 51, 1718-1728.










Dunlop, D. D., Song, J., Lyons, J. S., Manheim, L. M., & Chang, R. W. (2003)
Racial/ethnic differences in rates of depression among preretirement adults.
American Journal of Public Health, 93, 1945-52.

Faul, F., Erdfelder, E., Lang, A., & Buchner, A. (2007). G*Power 3: A flexible statistical
power analysis program for the social, behavioral, and biomedical sciences.
Behavior Research Methods, 39, 175-191.

Givens, J. L., Katz, I. R., Bellamy, S., Holmes, W. C. (2007). Stigma and the
acceptability of depression treatments among african americans and whites.
Journal of General Internal Medcine. 22,1292-1297.

Gonzalez, H. M., Croghan, T., West, B., Williams, D., Nesse, R., Tarraf, W., Taylor, R.,
Hinton, L., Neighbors, H., Jackson, J. (2008) Antidepressant Use in Black and
White Populations in the United States. Psychiatric Services, 59, 1131-1138.

Greenberg, P.E., Kessler, R.C., Birnbaum, H.G., Leong, S.A., Lowe, S.W., Berglund,
P.A., & Corey-Lisle, P.K. (2003). The economic burden of depression in the
United States: how did it change between 1990 and 2000? Journal Clinical
Psychiatry, 64, 1465-1475.

Greenberg, P.E., Stiglin, L.E., Finkelstein, S.N., & Berndt, E.R. (1993) The economic
burden of depression in 1990. Joumal Clinical Psychiatry, 54, 405-418.

Harman, J.S., Mulsant, B.H., Kelleher, K.J., Schulberg, H.C., Kupfer, D.J., & Reynolds,
C.F. (2001). Narrowing the gap in treatment of depression. International Journal
of Psychiatry in Medicine, 31, 239-253.

Harman, J. S., Fortney, J., & Edlund, M. (2004) Disparities in the adequacy of
depression care in the United States. Psychiatric Services, 55,1379-1385.

Hudziak, J. J., Helzer, J. E., Wetzel, M. W., Kessel, K. B., McGee, B., Janca, A., &
Przybeck, T. (1993). The use of the DSM-III-R Checklist for initial diagnostic
assessments. Comprehensive Psychiatry, 34, 375-83.

Institute of Medicine (2003). Unequal treatment: Confronting racial and ethnic disparities
in health care, ed. B.D. Smedley, A.Y. Stith, and A.R. Nelson. Washington, DC:
The National Academies Press.

Katon, W., & Schulberg, H. (1992). Epidemiology of depression in primary care. General
Hospital Psychiatry, 14, 237-247.

Kirmayer, I. J., Robbins, J. L., Dworkind, M., & Yaffe, M.J. (1993). Somatization and the
recognition of depression and anxiety in primary care, American Journal of
Psychiatry, 150, 734-741.










Kroenke K, Spitzer RL, Williams JB.(2003). The Patient Health Questionnaire-2: validity
of a two-item depression screener. Medical Care, 41, 1284-1292.

McKinlay, J. B., Lin, T., Freund, K., & Moskowitz, M. (2002). The unexpected influence
of physician attributes on clinical decisions: results of an experiment. Journal
Health and Social Behavior, 43, 92-106.

Kerr, L. K., Kerr, L. D. (2001). Screening tools for depression in primary care: the effects
of culture, gender, and somatic symptoms on the detection of depression.
Western Journal of Medicine, 175, 349-352.

Kessler, R. C., Andrews, G., Mroczek, D., Ustun, T. B., & Wittchen, H. U. (1998). The
World Health Organization composite international Diagnostic Interview Short
Form (CIDI-SF). International Journal of Methods in Psychiatric Research, 7,
171-185.

Klinkman, M. S., Coyne, J. C., Gallo, S., & Schwenk, T. L. (1998). False positives, false
negatives, and the validity of the diagnosis of major depression in primary care.
Archives of Family Medicine, 7, 451-461.

McGuire, T. G., Alegria, M, & Cook, B. L.(2006). Implementing the Institute of Medicine
definition of disparities: an application to mental health care. Health Services
Research, 41,1979-2005.

Melfi, C. A., Croghan, T. W., & Hanna, M. P. (1999). Access to treatment for depression
in a Medicaid population. Journal of Health Care for the Poor and Underserved,
10, 201-215.

Menchetti, M., Belvederi-Murri, M., Bertakis, K., Bortolotti, B., Berardi, D. (2009).
Recognition and treatment of depression in primary care: effect of patients'
presentation and frequency of consultation. Journal of Psychosomatic Research.
66, 335-341.

Menke, R. & Flynn, H. (2009). Relationships between stigma, depression, and treatment
in white and African American primary are patients. The Journal of Nervous and
Mental Disease. 197, 407-411.

Murray, C. J., Lopez, A. D. (1997) Global mortality, disability, and the contribution of risk
factors: Global burden of disease study. Lancet. 349,1436-1442.

Office of Management and Budget (1997). Revisions to the standards for the
classification of Federal data on race and ethnicity. Federal Register 62FR58782-
58790 (58790): Retrieved July 1, 2010, from
http://www.whitehouse.gov/omb/fedreg/ombdirl 5.html.









Ortho Biotech Products, L.P. (2007) Quick Reference to 2008 ICD-9-CM Diagnosis
Codes for Common Cancers. Retrieved July 1, 2010, from
www.procritline.com/pubs/icd9/icd9.pcl.onc.pdf

Skaer, T.L., Sclar, D.A., Robison, L.M., & Galin, R.S. (2000). Trends in the rate of
depressive illness and use of antidepressant pharmacotherapy by ethnicity/race:
an assessment of office-based visits in the United States, 1992-1997. Clinical
Therapeutics, 22,1575-1589.

Somervell, P. D., Leaf, P., Weissman, M. M., Glazer, D., & Bruce, M. L. (1989). The
Prevalence of Major Depression In Black and White Adults in Five United States
Communities. American Journal of Epidemiology. 130, 725-735.

Stockdale, S. E., Lagomasino, I T., Siddique, J., McGuire, T., & Miranda, J. (2008).
Racial and Ethnic Disparities in Detection and Treatment of Depression and
Anxiety Among Psychiatric and Primary Health Care Visits, 1995-2005. Medical
Care, 46, 668-677.

U.S. Department of Health and Human Services. (2001). Mental Health: Culture, Race,
and Ethnicity-A Supplement to Mental Health: A Report of the Surgeon
General. Rockville, MD: U.S. Department of Health and Human Services,
Substance Abuse and Mental Health Services Administration, Center for Mental
Health Services.

U.S. Preventive Services Task Force. (2002). Screening for depression:
recommendations and rationale. Annals of Internal Medicine, 21, 760-764.

Tacchini, G., Coppola, M. T., Musazzi, A., Altamura, A. C., & Invernizzi, G. (1994).
Multinational validation of the Composite International Diagnostic Interview
(CIDI). Minerva Psichiatr, 35, 63-80.

Tai-Seale, M., McGuire, T. G. & Zhang W. (2007). Time allocation in primary care office.
visits. Health Services Research, 42, 1871-1894.

Tylee, A. (2006). Identifying and managing depression in primary care in the United
Kingdom. Journal Clinical Psychiatry, 67 Suppl 6, 41-45.

U.S. Bureau of the Census (1998) Current Population Reports, Series P60-202.

Valenstein, M., Vijan, S., Zeber, J., Boehm, K., & Buttar, A. (2001). The cost-utility of
screening for depression in primary care. Annals of Internal Medicine, 134, 345-
360.

Wang, P.S., Walker, A., Tsuang, M., et al. (2000). Strategies for improving comorbidity
measures based on Medicare and Medicaid claims data. Journal Clinical
Epidemiology, 53, 571-578.










Williams, D. R., Gonzalez, H. M., Neighbors, H. W., Nesse, R., Abelson, J. M.,
Sweetman, J. & Jackson, J. S. (2007). Prevalence and Distribution of Major
Depressive Disorder in African Americans, Caribbean Blacks, and Non-Hispanic
Whites: Results From the National Survey of American Life. Archives of General
Psychiatry, 64, 305-315.









BIOGRAPHICAL SKETCH

Zoe is originally from Aberdeen in Scotland, and completed her Bachelor of

Science in psychology at the University of Glasgow.

After moving to the United States in 2001, she began working as a day-treatment

counselor and clinical case manager with a non-profit mental health organization

providing services to Medicaid enrollees. During this time, Zoe developed a strong

interest in working with underserved populations and in health policy. She spent one

year as a study coordinator at the H. Lee Moffitt Cancer Center at the University of

South Florida before returning to graduate school.

While at the University of Florida, she sought out placements aimed at improving

access to mental health services for underserved populations, including the uninsured,

rural communities, and those living in poverty. These experiences have led her to a

special interest in the integration of mental health services into traditional healthcare

settings, such as primary care. During her final year at the University of Florida, she

was involved in the pioneering of an integrated mental health service within a student

run "Equal Access Clinic" in order to increase mental health services for the uninsured

and also to provide training for clinical psychology students with this unique and

important population.

Zoe completed her pre-doctoral Fellowship at Yale University School of Medicine

and obtained her Ph.D. in Clinical and Health Psychology at the University of Florida in

2010.

Zoe's goals for the future are to continue to work to improve access to mental

health services for underserved populations and to use these experiences to guide her

research in tackling the complex mental health needs of the underserved.





























































126





PAGE 1

1 RACIAL DISPARITIES IN THE DIAGNOSIS AND TREATMENT OF DEPRESSIVE DISORDERS IN MEDICAID PRIMARY CARE By ZO N. SWAINE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Zo N. Swaine

PAGE 3

3 To the family and friends who have helped me throughout this journey

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank the Florida Center for Medicaid and the Uninsured for their support of me throughout my graduate training and the Florida Agency for Healthcare Administration for their generosity in allowing me to utilize their Medicaid data. I would also like to acknowledge Robert G. Frank, Ph.D., who, even from across the country, continued to support my work, and to the other members of my dissertation committee, Jeffrey S. Harman, Ph.D., David M. Janicke, Ph.D., and Brenda A. Wiens, Ph.D.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS..................................................................................................4 LIST OF TABLES............................................................................................................7 LIST OF FIGURES..........................................................................................................9 ABSTRACT ...................................................................................................................10 CHAPTER 1 REVIEW OF LITERATURE....................................................................................13 Prevalence and Cost of Depression.......................................................................13 Racial Disparities in Ment al Health Services...........................................................13 The Prevalence of Depression among African Americans......................................15 Disparities in the Diagnosis of Depression..............................................................18 Disparities in the Treatment of Depression.............................................................20 Potential Causes of Racial Disparities in Diagnosis................................................24 Symptom Presentation in Racial Disparities.....................................................24 Depression and African Americans in the Medicaid Population..............................27 Depression and African Amer icans in Primary Care...............................................28 2 PURPOSE AND SIGNIFIC ANCE OF THE STUDY................................................29 The Purpose of the Study.......................................................................................29 The Significance of the Study.................................................................................29 Specific Aims and Hypotheses...............................................................................30 3 CONCEPTUAL MODEL..........................................................................................32 Race, Somatic Symptoms of Depression, Depression Diagnosis, and Treatment..32 4 DATA AND METHODS...........................................................................................34 Sample and Data....................................................................................................34 Sample Identification........................................................................................34 Depression Case Identification.........................................................................35 Dataset...................................................................................................................35 Medicaid Claims Data......................................................................................36 Survey Data......................................................................................................37 Variable Constr uction and Definition.......................................................................37 Dependent Vari ables........................................................................................37 Independent Variables and Covariates............................................................40 Statistical Methods..................................................................................................43

PAGE 6

6 5 RESULTS...............................................................................................................55 Sample Characteristics...........................................................................................55 Final Survey Sample........................................................................................55 PHQ-2 Identified D epressed Sample...............................................................56 Research Goals......................................................................................................57 Aim 1. To examine racial disparities in the diagnosis of depression and treatment of depression the Medicaid primary care population.....................57 Hypothesis 1a............................................................................................57 Hypothesis 1b............................................................................................58 Hypothesis 1c............................................................................................58 Hypothesis 1d............................................................................................59 Aim 2: To examine the role of somatic symptoms in the diagnosis and treatment of depression................................................................................59 Hypothesis 2a............................................................................................59 Hypothesis 2b............................................................................................60 Hypothesis 2c............................................................................................61 Hypothesis 2d............................................................................................61 Hypothesis 2e............................................................................................62 6 DISCUSSION.........................................................................................................80 Depression Diagnosis and Treatment in Medicaid Primary Care............................80 Racial Disparities in Diagnosis and Treatment.......................................................82 Racial Disparities in Diagnosis.........................................................................82 Racial Disparities in Treatment.........................................................................82 Racial Disparities in Mental Health Expenditures.............................................83 The Role of Somatic Symptoms in Racial Disparities.............................................84 Implications and Recommendations.......................................................................87 Limitations...............................................................................................................89 Future research......................................................................................................91 APPENDIX BMS DEPRE SSION SU RVEY ...................................................................93 LIST OF REFERENCES.............................................................................................120 BIOGRAPHICAL SKETCH..........................................................................................125

PAGE 7

7 LIST OF TABLES Table page 4-1 Sample characteristics of the full random sample and of the stratified survey sample................................................................................................................51 4-2 Probability of major depression or any depressive disorder at each score of the PHQ-2..........................................................................................................51 4-3 ICD-9 diagnosis codes for all depression diagnoses..........................................52 4-4 A list of all included Antidepressant Medications................................................52 4-5 ICD-9 Diagnosis codes for common c ancers.....................................................53 4-6 Categories of Psychiatric Comorbidities: ICD-9-CM diagnosic codes................54 5-1 Sample Characteristics of the final survey sample.............................................63 5-2 Sample Characteristics of the PHQ-2 identified depressed sample...................64 5-3 Results of the Logit Analysis for Hypothesis 1a..................................................65 5-4 Power analysis fo r hypothesis 1a.......................................................................66 5-5 Results of the Logit Analysis for Hypothesis 1b predicting treatment with antidepressants..................................................................................................67 5-6 Results of the Logit Analysis for Hypothesis 1b predicting mental health visits...................................................................................................................68 5-7 Results of the Logit Analysis for Hypothesis 1b predicting any type of treatment ............................................................................................................69 5-8 Results of the Logit Analysis for Hypothesis 1c predicting any mental health expenditure.........................................................................................................70 5-9 Results of the Gamma Model for Hypothesis 1d predicting mental health expenditure (assuming expenditures > $0).........................................................71 5-10 Means of mental health expenditure by race (assuming expenditures > $0)......71 5-11 Results of the Logit Analysis for Hypothesis 2a predicting depression diagnosis............................................................................................................72 5-12 Power analysis fo r hypothesis 2a.......................................................................73

PAGE 8

8 5-13 Results of the Logit Analysis for Hypothesis 2a predicting depression diagnosis, with PCS removed.............................................................................74 5-14 Results of the Logit Analysis for Hypothesis 2b predicting any depression treatment ............................................................................................................75 5-15 Results of the Logit Analysis for Hypothesis 2b predicting any depression treatment, with PCS removed.............................................................................76 5-16 Results of the Logit Analysis for Hypothesis 2c predicting significant somatic symptoms...........................................................................................................77 5-17 Results of the Logit Analysis for Hypothesis 2c predicting significant somatic symptoms, with PCS removed...........................................................................78 5-18 Results of the Mediation Analysis for Hypothesis 2d..........................................79 5-19 Results of the Mediation Analysis for Hypothesis 2e..........................................79 5-20 Results of the Mediation Analysis for Hypothesis 2d with PCS removed...........79 5-21 Results of the Mediation Analysis for Hypothesis 2e with PCS removed...........79

PAGE 9

9 LIST OF FIGURES Figure page 3-1 Conceptual model of the relationship between race, somatic symptoms, diagnosis and treatment and expenditu res.........................................................32

PAGE 10

10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy RACIAL DISPARITIES IN THE DIAGNOSIS AND TREATMENT OF DEPRESSIVE DISORDERS IN MEDICAID PRIMARY CARE By Zo N. Swaine August 2010 Chair: Robert G. Frank Major: Psychology Depressive disorders are common, chroni c, and costly. In primary care settings, the point prevalence of major depression ranges from 5% to 10%, with close to three times as many individuals experiencing “sub-threshold” depressive symptoms. In 2001 the Surgeon General released a report that highlighted the prevalence of racial disparities in mental health care, including depression, and proposed addressing these disparities as a top priority for the nation. African Americans in particular are vulnerable to health disparities. There is growing evidence that even when African Am ericans do access and utilize healthcare services, they are less likely to be diagnosed and less likely to be treated for depression when compared with Caucasians. There is al so some evidence that African Americans have a lower likelihood of having any mental health expenditures, and lower total mental health expenditures when compared with Caucasians. Given the evidence for disparities in diagnosis and treatment, it is important to look for potential factors that might contribut e to these disparities. A 2003 Institute of Medicine report on racial disparities states, “the most significant gap in [disparities] research is the failure to identify mechanisms by which these disparities occur”. One of

PAGE 11

11 the least researched and most speculated-upon areas of research is the role of somatic symptoms in the diagnosis and treatment of d epression. There is a small but significant amount of evidence that shows that African Americans with depression are more likely to present with somatic symptoms than Caucasians. It is speculated that these somatic symptoms of depression are frequently misattributed as symptoms of physical illness and mask the underlying etiology. This study examines the prevalence, diagnosis, and treatment of depression in a Medicaid primary care population. It then examines the presence of racial disparities in diagnosis, treatment, and expenditures. It also examines the effects of somatic symptoms of depression and their role in diagnosis and treatment in primary care among depressed Medicaid enrollees. A random sample of 2,106 Medipass enrollees participated in a telephone survey, assessing depressive symptoms, while their Medicaid claims data were collected for the years 2003, 2004, and 2005. Information on their somatic symptoms, physician diagnosis of depression, treatment history, race, and other demographic characteristics were gathered. From this sample the a depression screening tool, the PHQ-2, was used to identify those enrollees with a likely depressive disorder. Of the initial 2,106, one third of these enrollees met the screening criteria and were included in the depressed sample. The analyses showed that African Americans were not less likely to be diagnosed as depressed, however, this analysis was underpowered due to the very low numbers of African Americans actually diagnosed with depression. African Americans were less likely to receive treatment for depression. Caucasians had approximately four times the

PAGE 12

12 odds of obtaining treatment for depression. The lower likelihood of obtaining treatment also led to a significantly lower likelihood of having any mental health expenditure, and a trend showing that African Americans’ mental health expenditures were approximately half that of Caucasians. The role of somatic symptoms was not found to be significant in any of the analyses, although two analyses did approach significnce. Those who endorsed somatic symptoms had approximately twice the odds of not being diagnoses, and there was an unexpected trend that indicated Caucasians had almost twice the odds of endorsing somatic symptoms. The final analysis did not support the theory that increased somatic symptoms among African Am ericans cause lower rates of diagnosis and treatment. This study is an important step to understanding the role of somatic symptoms in racial disparities. Lager studies are needed to fully evaluate the relationship due to the very low numbers of African Americans being diagnosed in Medicaid primary care.

PAGE 13

13 CHAPTER 1 REVIEW OF LITERATURE Prevalence and Cost of Depression Depressive disorders are common, chronic, and costly. Lifetime prevalence rates obtained from community-based surveys indicate the range to be between 5% to 17% (Kessler, Andrews, Mroczek, Ustun, & Wittchen, 1998). One of the most recent largescale epidemiological surveys completed in 2005 found 12 month prevalence rates of 5% and a lifetime prevalence of 13%. In primary care settings, the point prevalence of major depression ranges from 5% to 10%, wit h close to three times as many individuals experiencing “sub-threshold” depressive symptoms (Katon and Schulberg, 1992). Second to hypertension, depression is the most commonly presenting problem in primary care (Kanton and Schulberg, 1992). By 2020, depressive illness is projected to be the second leading cause of disability worldwide (Murray and Lopez, 1997) Depression has substantial public health implications and great economic significance. In 1990 it was estimated that depression cost the United Sates $43 billion annually, of which $17 billion represents lost work days (Greenberg, Stiglin, Finklestein & Berndt, 1993). In an update to this research Greenberg et al. (2003) found that thes e costs had increas ed to $84 b illion. Racial Disparities in Mental Health Services In 2001 the Surgeon General released a report that highlighted the prevalence of racial disparities in mental health care, and proposed addressing these disparities as a top priority for the nation (DHHS, 2001). The Surgeon General’s report found that minorities have less access to, and lower av ailability of, mental health services, that they are less likely to receive needed mental health services, that when in treatment

PAGE 14

14 they often receive a poorer quality of care, and that minorities are underrepresented in mental health research (DHHS, 2001). The report also found evidenc e that minorities experience a greater disability burden fr om mental illness than do Caucasians, stemming from both receiving less care and a poorer quality of care, rather than from increased prevalence or increased severity of illness. African Americans in particular are vulnerable to disparities in health for several reasons. African Americans have been shown to have similar rates of mental illness as Caucasians, however, they are over-repres ented among vulnerable populations in which the rates of mental illness are higher. African Americans are also disproportionately served by the “safety net providers,” such as Medicaid providers or county health departments, who have been coming under increasing pressure over the past two decades from healthcare reforms. In addition, African Americans are more likely to live in poor areas, where there are also shortages of healthcare providers. Finally, African Americans have been shown to have lower utilization rates of health services than Caucasians (DHHS, 2001). Consensus on the definition of disparity has been slow to form. The Institute of Medicine (IOM) released a report in 2003, defining a disparity in healthcare as “racial or ethnic differences in the quality of health ca re that are not due to access-related factors or clinical needs, preferences, and appropriat eness of intervention” (IOM, 2003, pg. 32). However, some of those who use the IOM definition to model disparities have chosen to disregard the IOM’s inclusion of “not due to access-related factors” in their conceptualization of disparities (McGuire, Aleg ria, & Cook, 2006). The main contention with the IOM definition is that it only addressed disparities arising from within the clinical

PAGE 15

15 encounter, not from factors arising prior to the encounter, including health care system factors such as health insurance. The debate continues with regards to defining disparities with one extreme arguing for models that do not adjust for any socioeconomic factors, thereby allowing them to mediate the race-healthcare relationship, to the other extreme of regression-based models that carefully control for all variables. The Prevalence of Depression among African Americans In order to demonstrate equal need among the African American population, the prevalence of depression among African Americans must be examined. Several large scale epidemiological studies of the prevalence of mental disorders have produced mixed results with regards to rates of depression in African Americans. The Epidemiologic Catchment Area study (ECA) sampled residents of Baltimore, St. Louis, Durham-Piedmont, Los Angeles, and New Haven to examine the prevalence of major depression among racially and ethnically diverse adults (Sommervell, Leif, and Weismann & Bruce, 1989). In total, it sampled 4,638 African Americans, 12,944 Caucasians, and 1,600 Hispanics who we re surveyed between 1980-1983. Diagnoses of depression were identified through the use of Diagnostic and Statistical Manual, Third Edition criteria. Age, sex, and site-adjusted analyses did not show any significant differences in lifetime prevalence or si x-month prevalence between Caucasians and African Americans. There were some differences among the 18-24 years age group in 6 month prevalence: African American women showed a trend for higher six-month prevalence than Caucasian women, and Cauc asian men showed a trend for higher sixmonth prevalence than African American men. Three sites showed significantly lower

PAGE 16

16 rates of depression among African Americans when compared to Caucasians, after controlling for income and age. Dunlop, Song, Lyons, Manheim and Chang (2003) estimated depression prevalence rates among African American, Hispanic, and Caucasian adults from a population-based national sample (1996 HRS). In order to allow for sufficient numbers for comparison their sample included an oversampling of African Americans and Hispanics relative to Caucasians. The study used the short form of the World Health Organization’s Composite International Diagnostic Interview (CIDI-SF) to identify likely depression cases. Results showed that African Americans (88.5 per 1000) and Hispanics (107.8 per 1000) exhibited elevated rates of major depression relative to Caucasians (77.5 per 1000). However, after controlling for socio-demographic, health, and economic factors, Hispanics and Caucasians exhibited similar rates, and African Americans exhibited significantly lower rates than Caucasians. This indicates that African Americans in the community suffer from higher rates of mental illness than Caucasians, but that the difference may be explained by demographic factors and socioeconomic factors. In other words, fa ctors associated with depression were more frequent among members of minority groups than among Caucasians. Elevated depression rates among minority individuals are heavily associated with serious chronic illness, functional limitations, a lack of health insurance, and health behaviors such as smoking and not exercising. Williams et al. (2007) used the National St udy of American Life (NSAL), a national household probability sample assessing the m ental health of African Americans, to survey a total of 3,570 African Americans, 1,621 Caribbean-blacks, and 891 non-

PAGE 17

17 Hispanic white adults. The survey included an adaptation by the World Health Organization of the Composite International Diagnostic Interview (CIDI). Lifetime Major Depressive Disorder prevalence estimates were highest for Caucasians (17.9%), followed by Caribbean-blacks (12.9%) and African Americans (10.4%). However, 12month prevalence of Major Depressive Disorder was similar across groups (African American 5.9%, Caribbean-black 7.2%, and Caucasian 6.9%). The same study examined the chronicity of Major Depressive Disorder. This was assessed by comparing the ratio of individuals with 12-month Major Depressive Disorder to lifetime Major Depressive Disorder cases, which showed that chronicity was significantly greater for both minority groups (56.5% for African Americans and 56.0% for Caribbean-blacks) than for whites (38.6%). Depression severity and impairment was assessed through the Sheen Disability Scale and the Quick Inventory of Depressive Symptomatology Self-Report (QIDS-SR). T hese instruments showed that, relative to Caucasians, both minority groups were more likely to rate their depression as severe and disabling. This study indicates that while lifetime risk for depression may be lower among African Americans and Caribbean blacks, their 12-month risk is similar to Caucasians due to the greater chronicity of depression among these two groups. While findings are mixed regarding racial differences in the prevalence of depression, the results from these studies are consistent in that they indicate the rates of major depression among African Americans are similar to those of Caucasians. However, the relationship between depression and race is complex, and there is an overrepresentation of African Americans in high-need populations which are harder to access when surveying (for example, those liv ing in inner cities, poor rural areas, and

PAGE 18

18 prisons are populations). If members of these groups were included, then higher rates of depression among African Americans might be detected. Disparities in the Diagnosis of Depression Given that findings indicate that prevalence rates for depression are similar across racial groups, the question remains where do racial disparities in healthcare come from? There is growing evidence that even when African Americans do access and utilize healthcare services, they are less likely to be diagnosed with depression when compared to Caucasians. In one of the more recent study of disparities in depression diagnosis in primary care, Stockdale et al.(2008) used the National Ambulatory Medical Care Survey (NAMCS) for 1995-2005 to examine disparities in diagnosis and treatment in primary care. The survey gathers information about ambulatory office visits to primary and specialty care. Minorities were underrepresent ed in the primary care sample (Caucasian 80.68%, African American 10.6%, Hispanic 8.72%), perhaps reflecting their lower rates of healthcare utilization. Primary care vi sits by African Americans had significantly lower odds of resulting in a depression or anxiety diagnosis when compared to Caucasians (odds ratios ranged from 0.56-0.65). A rate by time interaction analysis, also indicated that these disparities in primary care did not change during the 10 years of the study. Borowsky et al.(2000) conducted a large sca le cross-sectional survey of 19,309 patients and 349 internists and family physicians in Boston, Chicago, and Los Angeles. As part of the Medical Outcomes Study, participants completed a self-administered screening survey that included a brief depression screening questionnaire while physicians completed questionnaires regarding the diagnosis and treatment of the

PAGE 19

19 participants. The study showed that both Hispanics (OR=0.94, p< 0.05) and African Americans (OR = 0.63, p= 0.05) were less frequently diagnosed with any mental health problem than Caucasians, despite having the same level of mental health functioning. Lower diagnosis rates were also found by Skaer, Sclar and Robinson (2000) in an analysis of data from the US National Ambulatory Medical Care Survey, between 1992 and 1997. This is a nationwide probabilistic survey of physician office visits completed by the National Center for Health Statistics, which yielded a total of 36,875 patient records. This data showed evidence that rates of diagnosis of depression among Caucasians (11.3%) were significantly higher than among African Americans (5.5%) and Hispanics (8.3%). In contrast to findings of disparitie s in diagnoses, Minsky et al. (2003) studied new admissions to the New Jersey behav ioral health system between January 2000 and August 2001. They collected data on mental health functioning on 19,219 patients and found that Latinos were more likely to be diagnosed than Caucasians (OR = 1.74), while no differences were found between African Americans and Caucasians (OR = 0.99). In another study, Crystal et al. (2003) used the Medicare Current Beneficiary Survey’s (MCBS) cost and use files, between 1992 and 1998, to obtain estimates of depression diagnoses, and the rates of treatment of those with depression. Their sample consisted of 20,966 elderly individuals from which they derived 51,058 personyears. Diagnoses recorded in Medicare claims were used to identify individuals who received a diagnosis of depression from a healthcare provider and pharmacy and claims data were used to identify receipt of antidepressants and/or psychotherapy. This

PAGE 20

20 data showed no racial disparities in the likelihood of being diagnosed with depression among this elderly population. Other researchers have examined how varying patient attributes influences physician diagnosis. McKinley et al. (2002) presented videotapes of staged patientphysician encounters for depression and polymyalgia rheumatica (PMR) in order to examine the effects of patient attributes on the diagnosis of depression. They examined age (65 years or 80 years), sex, race (African American or Caucasian), and occupation (blue or white collar) in various combinations to assess their impact on physician diagnosis. The study found no significant influence of any of the patient attributes on the physicians’ “most likely diagnosis” of either depression or PMR. However, characteristics of the physicians (e.g., medical specialty, race, and age) did impact the decision. Obviously physician behavior in such an experimental environment may not be a valid representation of behavior in practice in the community. Overall, these studies show that the relationship between physician diagnosis and race is clearly a complex one. While there is a large volume of evidence that lends weight to the existence of disparities in diagnosis, the reasons behind these disparities are not always well understood. Disparities in the Treatment of Depression Given that some studies show evidence fo r disparities in diagnosis, with African Americans being less likely to be diagnosed than Caucasians, this might logically lead to a decreased likelihood of African American s being treated for depression. Williams et al. (2007), who examined the prevalence of depression among African Americans, Caribbean blacks, and non-Hispanic whites in the NSAL study, showed that fewer than half of the African Americans (45.0%) and fewer than a quarter (24.3%) of the

PAGE 21

21 Caribbean blacks who met the criteria for a major depressive disorder received any form of therapy for their depression. While these numbers seem very low, they had no data on treatment of non-Hispanic Caucasians and so comparisons could not be made. Claims data were used in one study of Medicaid recipients to examine racial disparities in depression treatment. Melf i, Croghan and Hanna (1999) examined racial differences among Medicaid recipients in treatment for depression. Their sample was 46% African American and showed that African Americans were less likely to receive antidepressant medication on the first diagnosis of depression and once treatment was initiated, they were more likely to receive older try-cyclic antidepressants (TCAs). African Americans were also more likely to prematurely discontinue treatment and less likely to receive a second medication. While this study appears to offer compelling evidence, it must be noted that they did not control for demographic variables, depression severity, or the presence of comorbid illness. The National Center for Health Stat istics (NCHS) annually samples a nationally representative sample of offi ce visits to physicians in Ambulatory practice, including primary care and all specialties. Using this data Harman et al. (2001) analyzed visits to primary care physicians and psychiatrists where a depression diagnosis was recorded, during two time periods: 1993-1994 and 1996-1997. In the 1993-1994 time period African Americans had significantly lower odds of receiving any depression treatment when compared with Caucasians (OR =0.89). However, in the 1996-1997 time period this difference had disappeared. Harmon argued that this provides evidence that racial disparities in the treatment of depression may be shrinking. However, these results should be interpreted with caution as their sample only included those who had been

PAGE 22

22 diagnosed with depression by their physician, and there is clear evidence that depression is generally under-diagnosed, especially in primary care. Harman, Fortney, and Edlund (2004) did a further study, this time using the 2000 Medical Expenditure Panel Survey (M EPS). MEPS is an annual, nationally representative survey that gathers data on healthcare use, expenditures, health status, and demographic variables. Their analyses showed that among those with selfreported depression, African Americans were significantly less likely to fill an antidepressant prescription (OR=.47) when compared with Caucasians. Once treatment was initiated, there were no racial differences in the likelihood of receiving an adequate course of antidepressants. There were also no racial differences in the likelihood of receiving psychotherapy; however, of thos e who received psychotherapy, African Americans were significantly more likely to receive an adequate course (OR=2.47). The odds of receiving any treatment for depression were significantly lower for African Americans when compared with Caucasians (OR=.44). These studies suggest that disparities in treatment may stem from gaining the initial access to treatment rather than in maintaining an adequate course of treatment. Once treatment was initiated, African Americans were more likely to receive an adequate course of counseling and psychotherapy and equally as likely to receive an adequate course of antidepressant treatment. What this study was unable to reveal were the rates of under-treatment among those who had not been diagnosed as depressed or those who did not report a diagnosis. In addition to examining diagnosis disparities, Stockdale, Lagomasino, Siddique, McGuire, and Miranda (2008) used the NAMCS to evaluate racial disparities in

PAGE 23

23 treatment. They compared rates of antidepressant prescriptions, counseling or referral for counseling, and any depression treatment (a dichotomous variable that indicated either form of treatment). The data showed that African Americans had significantly lower odds of receiving counseling (or a counseling referral) in the 1997-1999 time period (OR = 0.59) and significantly lower odds of receiving an antidepressant prescription in the 1997-2005 time period (OR range = 0.53 to 0.67). Time by race interactions proved non-significant indicating that disparities in treatment did not change over time. Finally their analysis of the odds of receiving any treatment showed that African Americans had significantly lower odds of receiving any type of treatment when compared to Caucasians across all time periods (OR range = 0.59 to 0.74). These differences persisted even after controlling for the effect of receiving a diagnosis. Again, the time by race interaction proved non-significant, indicating no change in racial disparities in treatment across the period of the study from 1995 to 2005. Further evidence that racial disparities in mental health treatment still exist comes from a 2007 M EPS study. Cook, Maguire, & Mir anda (2007) found a worsening of racial disparities in mental health treatment from 2003/2004 to 2006/2007. This study and the Stockdale study indicate the continued existenc e of racial disparities in the treatment of depression in primary care, contrary to the work by Harman et al. (2001). While Stockdale’s goal was to examine disparities without eliminating the influence of SES on this population, Harman et al. (2001) controlled carefully for demographic variables, insurance variables, and physician specialty (p rimary care versus psychiatry) among other influential variables, which may have removed most of the effect of race. This

PAGE 24

24 difference in methodology stems from the ongoing debate surrounding the definition of disparities. Potential Causes of Racial Disparities in Diagnosis Given the evidence for disparities in diagnosis and treatment, it is important to look for potential factors that might contri bute to these disparities. The IOM report on racial disparities states, “the most significant gap in [disparities] research is the failure to identify mechanisms by which these disparities occur” (IOM, 2003). Several potential causes that contribute to the existence of ra cial disparities in mental health care have been proposed and can be classified as systemic factors, patient factors, and physician factors. The 2001 Surgeon General’s report on racial and ethnic disparities in mental health provides a detailed re view of these (DHHS, 2001). Patient factors include: coping style, mistrust of physicians, treatm ent seeking behavior, stigma, immigration and acculturation, and health status. Physician factors include: communication difficulties, clinician bias and stereotyping, and poor recognition in primary care. Finally, systemic factors include: service setting and limitations in access, financing of the system and health insurance, paucity of evidence-based treatment in the community, cultural competence, effectiveness of medication among minorities, poverty, community violence, and marginal neighborhoods. One of the most speculated-over causes, but one of the least studied is the patient’s symptom presentation. Symptom Presentation in Racial Disparities Mental illness is a worldwide phenomena, but the way patients express or present their symptoms to clinicians is affected by culture. There is a small but significant amount of evidence that shows that African Americans with depression are more likely to present with somatic symptoms than Caucasians. It is speculated that

PAGE 25

25 these somatic symptoms of depression are frequently misattributed as symptoms of physical illness and mask the underlying etiology. The misattribution of somatic symptoms to physical illness has some support. In one Italian study examining the influence of symptom presentation, PCPs were asked to record the reason for the patient visit, either psychological/family problems, physical illness, or pain symptoms. Depressed pati ents were then identified by psychiatric assessment. Among depressed patients t hose who presented with physical symptoms were at 2.3 times the risk for non-recognition and those presenting with pain had four times the risk of not being recognized as depressed Menchetti, Belvederi-Murri, Bertakis, Bortolotti & Berardi). Another study of primary care found that somatization reduced physician recognition of depression from 77% to 22% (Kirmayer, Robbins, Dworkind & Yaffe, 1993). There is also some evidence that African Americans may present with more somatic symptoms of depression than Caucasians. In a study with 665 African American and Caucasian psychiatric inpatients, differences in diagnosis were examined using the DSM-III-R Symptom Checklist (Hudzi ak, Helzer and Wetzel et al., 1993). Results showed that while disparities still ex isted in the diagnosis of Schizophrenia and Bipolar disorder, there were no differences in depression diagnosis between the two groups. However, they did find evidence for di fferences in symptom attribution, by race of the patient, especially in schizophrenia. However, these results may not generalize to outpatient care or primary care. Wohl, Lesser and Smith (1997) also used a structured interview to compare the nature and severity of depressive symptoms in depressed, medically healthy African

PAGE 26

26 Americans and Caucasians. Twenty matc hed subjects were assessed using a structured interview and with the Hamilton Depression Rating Scale (HAM-D). HAM-D items were then grouped into 7 factors: diurnal, sleep, weight, reality, mood, anxiety and somatic. When these were compared, overall severity of depression was comparable between groups however, Caucasians showed significantly more mood and anxiety symptoms, whereas African Americans had significantly more diurnal variation to their depression. There were no differences on other neurovegetative symptoms. While the sample size of this study was small, the results do suggest that there are differences in symptom presentation between racial groups. In a study that aimed to retrospectively determine the impact of race on treatment adherence and outcomes among patients being treated for major depression in urban primary care settings, Brown Schulberg and Madonia (1996) analyzed their data to compare psychiatric history and clinical pr esentation between 119 African American and 153 Caucasians. While the two groups showed no significant differences in their treatment histories, the severity of their depression, the severity of medical illness, or in their level of psychosocial functioning, there were significant racial differences in other areas. African Americans were more likely to have psychiatric and medical comorbidities, they showed more severe sleep disturbance, and they had more severe somatic symptoms, greater limitations in their self-reported physical functioning, higher life stress, and more negative health beliefs. While these studies are valuable in advancing the understanding of racial disparities, none of these studies has syst ematically evaluated the role of somatic symptoms of depression in diagnosis.

PAGE 27

27 Depression and African Americans in the Medicaid Population The Medicaid population is a unique and valuable population to study. African Americans are disproportionately covered by Medicaid, 21% versus 8% of Caucasians among non-elderly adults (Brown, Ojeda, Wyn, and Levan, 2000), and in 1997 36.7% of African American children were covered by Medicaid compared to 17.1% of Caucasian children (US Bureau of the Census, 1998). Nationally, 23.1% of the Medicaid population is African American, which is approxima tely 13 million residents (CMS, 2007). Disadvantaged populations, such as those in Medicaid, have been shown to be at increased risk for poor mental health. Poverty itself has been shown to be a risk factor for depression. Factors contributing to chronic stress, such as substandard housing, high crime neighborhoods, and poor nutrition are associated with an increased risk for psychological dysfunction (Bennet, 1987). Studies of low-income groups and Medicaid recipients have shown higher rates of depression among these populations. In addition to being at greater risk for depression, studies have shown that Medicaid enrollees are particularly vulnerable to under-treatment for psychological disorders (Harman et al., 2001, Harman, Fortney, and Edlund, 2004; Melfi, Croghan, and Hanna, 1999). Depressed Medicaid recipients are significantly less likely to receive any treatment for depression (OR = .81) than those with private insurance (Harman et al, 2001) and less likely to receive SSRI’s, psychotherapy, and an adequate length of therapy than privately insured individuals (Melfi, Croghan, and Hanna, 1999). With the dual factors of increased risk for depression and under-treatment, and the high proportions of African Americans who rely on Medicaid, the Medicaid population is important to study in racial disparities research.

PAGE 28

28 Depression and African Americans in Primary Care There is evidence that shows that a disproportionate number of African Americans seek psychiatric care in primary care settings rather than with specialists (DHHS, 2001). However, studies have shown that usual care by primary care physicians fails to recognize 30% to 50% of depressed patients (Valenstein, Vijan, Zeber, Boehm, and Buttar, 2001), and of those who are recognized, treatment rates can be as low as 27% (Tylee, 2006). Klinkman, Coyne, Gallo, and Schwenk (1998) explored physician recognition and diagnostic sensitivity to the disorder by comparing physician ratings against the “gold standard” of the SC ID completed by a Licensed Clinical Social Worker. While the SCID found a 13% prevalence of depression, sensitivity of the physician diagnosis was low, only 0.34. This means that physicians identified only 34% of depressed individuals. However, specificity was high at .93, while positive predictive value was .45 which means less than half of the patients identified as depressed were actually depressed. Given that African Americans frequently seek mental health treatment from their primary care physician, and given the evidence for under-diagnosis and undertreatment of African Americans combined with the under treatment among the Medicaid population, individuals that belong to all three of these groups (African Americans, primary care, Medicaid) are in a uniquely disadvantaged position.

PAGE 29

29 CHAPTER 2 PURPOSE AND SIGNIFICANCE OF THE STUDY The Purpose of the Study The main purpose of the proposed study is to examine factors that contribute to the racial disparities seen in the diagnosis and treatment of depression in a Medicaid primary care population. The study will fi rstly examine the pr evalence, diagnosis, and treatment of depressive symptomatology in a Medicaid primary care population. It will then assess for the presence of racial disparities in diagnosis, treatment, and mental health expenditure among the depressed po pulation. The study will then go on to examine the mediating effects of somatic symptoms of depression between race and diagnosis of depression and also between race and treatment among primary care Medicaid enrollees. The Significance of the Study In order to reduce racial disparities in mental health care, we must better understand the mechanisms that cause it. If evidence can be found for the role of somatic symptoms contributing to racial disp arities, screening and treatment efforts may be better targeted to reduce the rates of untreated depression, and minimize the extensive costs associated with this. To do this be it is important to clarify the relationship between race, symptom presentation and diagnosis and treatment. While many studies have examined the role of race in physicians’ decision to diagnose, none have examined the mediating role of somatic symptoms of depression, and none have examined these factors among the vulnerable Medicaid population. A major limiting factor in many of the existing studies is the difficulty in identifying depressed patients. Those studies that look at treatment differences all rely on

PAGE 30

30 physician diagnosis as a valid indicator of depression in the study population. However, as demonstrated earlier, this method can miss a high proportion of the truly depressed population. This study will use a widely validat ed tool (PHQ-2) to assess depression in order to identify our depressed sample, independently of physician diagnosis. This method of identifying depressed cases, combined with access to physician diagnoses, treatment, and expense data make this study unique. Another limitation of previous studies is that while they have the ability to identify differences in symptom presentation between races, these studies have not been able to identify if this impacts diagnosis. The Patient Health Questionnaire PHQ9 will be used to identify exactly which symptoms of depression each individual is experiencing in order to investigate the role of symptom presentation in disparities in diagnosis. Taken all together, this study will allow a thorough and comprehensive examination of racial disparities in depression care in the Medicaid population, as well as one potential cause of disparities. Specific Aims and Hypotheses • AIM 1. To examine racial disparities in the diagnosis of depression, treatment of depression, and healthcare expenditures of a depressed Medicaid primary care population. • HYPOTHESIS 1A. Depressed patients who are African American will be less likely to be diagnosed with a depressive disorder than Caucasians. • HYPOTHESIS 1B. Depressed patients who are African American will be less likely to receive treatment for depression than Caucasians. • HYPOTHESIS 1C. Depressed patients who are African American will be less likely to have any mental health care expenditures compared to Caucasians. • HYPOTHESIS 1D. Depressed patients who are African American will have significantly lower mental healthcare expenditures than Caucasians. • AIM 2. To examine the role of somatic symptoms in the diagnosis and treatment of depression

PAGE 31

31 • HYPOTHESIS 2A. Depressed patients who endorse somatic symptoms will be less likely to be diagnosed with a depressive disorder than those who do not. • HYPOTHESIS 2B. Depressed patients who endorse somatic symptoms will be less likely to be receive treatment than those who do not. • HYPOTHESIS 2C. Depressed patients who are African Americans will be more likely to endorse somatic symptoms of depression than Caucasians. • HYPOTHESIS 2D. Of those who are depressed, the number of somatic symptoms of depression will mediate the relationshi p between race and the diagnosis of depression. • HYPOTHESIS 2E. Of those who are depressed, the number of somatic symptoms of depression will mediate the relationshi p between race and the treatment of depression.

PAGE 32

32 CHAPTER 3 CONCEPTUAL MODEL Race, Somatic Symptoms of Depression, Depression Diagnosis, and Treatment The proposed study will be based on a conceptual model that relates race, somatic symptoms of depression, diagnosis, treatment, and healthcare expenditures. This conceptual framework proposes that race will influences the presentation of somatic symptoms of depression which will then, in turn impact the likelihood of a depression diagnosis. The diagnosis, along with racial factors, then influences the likelihood of obtaining treatment, which will in turn increase mental health expenditures. Other patient, physician and systemic factor will also impact the liklihood of a physician visit and of diagnosis and treatment. The model for the proposed study is depicted as follows: Figure 3-1. Conceptual model of the relationship between race, somatic symptoms, diagnosis and treatment and expenditures Other Patient, Physician, & S y stemic Factors Race Physician Depression Diagnosis Depression Treatment Somatic Symptoms of Depression Mental Health Expenditures Physician Visit

PAGE 33

33 The proposed study includes only those individuals who have had at least one physician visit and will examine the relationship between race and diagnosis, race and treatment, and race and expenditure, while controlling for covariates, in Aim 1. In Aim 2 the study will examine the relationship between race and somatic symptoms of depression, the relationship between somatic symptoms of depression and diagnosis and treatment, and the relationship between race and diagnosis and race and treatment, mediated by the somatic symptoms of depression, while controlling for covariates known to impact depression diagnosis and treatment.

PAGE 34

34 CHAPTER 4 DATA AND METHODS Sample and Data Sample Identification A random sample of 31,775 Medipass enrollees were selected from the entire population of Florida Medicaid enrollees, from which a total of 2,106 enrollees participated in a telephone survey and are included in the final sample. Medicaid Provider Access System (MediPass) is a pr imary care case management program for Medicaid beneficiaries. Primary care providers receive a $2.00 monthly case management fee for each of their enrolled patients, in addition to fee-for-service for all services rendered. The sample was stratified by sex and was obtained through the use of Computer Assisted Telephone Interviewing (CATI). The full sample of 2,106 was used for initial descriptive analyses of the data, and to produce an estimate of the prevalence of depression and of treatment rates among the Medicaid population in Florida as a whole. Sub-samples of thos e identified as depre ssed will then used in the main analyses of the study. To examine possible issues of sampling bias table 4-1 shows a comparison of demographics variables (sex, age, and race/ethnicity) obtained from Medicaid claims data between the non-surveyed sample and the surveyed sample. Medicaid data was used as this was av ailable for both the surveyed and non-surveyed groups. The comparison indicates that there has been an oversampling of female enrolees (deliberate) and an over sampling of “black” and Hispanic enrollees, while “white” enrollees are underrepresented.

PAGE 35

35 Depression Case Identification The primary aims of the study are to examine diagnosis and treatment among those who are depressed, therefore the study must identify those who are depressed. Case identification will be accomplished using the PHQ-2. The PHQ-2 is a 2-item depression-screening scale that inquire s about the frequency of depressed mood and anhedonia over the past 2 weeks, scoring each as 0 ("not at all") to 3 ("nearly every day"). Any participant that scored four or above on the PHQ-2 was included in the depressed sample. Table 4-2 shows the prob abilities of actually having any depressive disorder (81.2%) or major depressive disorder (45.5%) at this and all other cut offs of the PHQ-2 (Kroenke, Spitzer, & William, 2003). The PHQ-2 is widely used in epidemiological research and has been validated among many populations, including the elderly and poor, and has been shown to have excellent validity and reliability among minority populations (Kroenke, Spitzer, & William, 2003). While the PHQ-2 can indicate the presence of a depressive episode, it does not specify whether a depressive episode occurs in the course of a major depressive disorder or whether it occurs in the course of another psychological disorder (for example bipolar disorder or schizophrenia). Dataset The dataset consists of two major elemen ts: 1) healthcare claims data obtained from the state of Florida’s Medicaid administrator, the Agency for Healthcare Administration (AHCA) for the years 2002, 2003, and 2004, and 2) healthcare survey data obtained by the Bureau of Economic and Business Research ( BEBR) in the employ of the Florida Center for Medicaid and the Uninsured, as part of an evaluation for AHCA, obtained between August 2004 and March 2005. The data contains

PAGE 36

36 Protected Health Information so full IRB and HIPAA approval was been obtained for this study and for use of this data. Medicaid Claims Data The AHCA supplies Florida Medicaid archival claims data as part of a contractual agreement with the Center for Medicaid and the Uninsured (FCMU), at the University of Florida. The healthcare claims data provid e information on recipient eligibility, facility claims (outpatient and inpatient), medical claims (physician claims), and pharmacy claims. Medicaid claims were compiled for the years 2003, 2004 and 2005 and included in the study if the claim fell within two years of the survey date of the enrollee. Florida Medicaid eligibility files include recipient inform ation including, date of birth, gender, race/ethnicity, Medicaid assistance category, Medicaid plan, length of enrollment, and eligibility start and end dates. Pharmacy, facility, and medical claims f iles provide event-level information. Pharmacy claims are submitted if a recipient filled one or more prescriptions during the period of the study. These claims include the date the medication is filled, therapeutic drug class, National Drug C ode number, and a refill indicator. Facility claims, which include claims from both outpatient and inpatient facilities, provide ICD-9CM diagnoses (up to 11), procedure codes, da ta on provider characteristics (including type and specialty), and a flag if the event was an emergency department visit. Medical claims are claims billed by physicians and include ICD-9-CM diagnoses (up to 2), procedure codes, and provider characteristics. All claims files have a unique identifier for each Medicaid recipient to enable merging of the 9 claims datasets and the survey dataset. Medicaid primary care claims were identified by use of the provider specialty

PAGE 37

37 code. The following specialties will be included as primary care providers: general practice (11), family practice (09), inter nal medicine (18), pediatrics (35), adolescent medicine (1), geriatrics (13), geriatric nurse practitioner (82), and adult primary care nurse practitioner (75). Survey Data The survey data utilized for this study is part of a broader AHCA study assessing a disease management program targeting depression, implemented by Florida’s Medicaid program. The survey contained questions on mental health, including depression, health-related quality of life, and general demographic questions. Partial funding for this study came from the pharmaceutical company Bristol-Myers Squibb. The sample used in this study is the control group that received no intervention. The survey was completed using the Computer Assisted Telephone Interviewing system (CATI) with a random sample of adult Florida Medicaid enrollees between August 2004 and March 2005. Data collection was completed by trained staff at the UF survey Research Center (UFSRC) at the Bureau of Economic and Business Research ( BEBR). The full survey can be seen in Appendix 1. Variable Construction and Definition The variables to be used in the study will retain their numerical values. However, some of the variables are categorical in nature and are dummy coded to enable their use in regression analyses. Dependent Variables Physician diagnosis (depphys) In hypotheses 1a, 2a, and 2d, physician diagnosis of depression is the dependent variabl e. Each of these aims examines patient characteristics thought to influence physicians’ decisions to diagnose

PAGE 38

38 depression. For this reason, the depressi on diagnosis is treated as a dichotomous variable. No diagnosis will be coded as “0” an d a diagnosis of depression is coded as “1”. Physician-diagnosed depression is identified by the International Classification of Diseases Clinical Modification (ICD-9-CM) co des of unipolar depression appearing on one or more health claims during the study period. ICD-9-CM diagnostic codes (including all primary and secondary diagnoses ) are used to identify any type of unipolar mood disorder (single episode, recurrent or un specified). Any eligible diagnosis in any of the diagnosis fields in either the facility claims file or the medical claims files indicate a diagnosis. See Table 4-3 for ICD-9-CM codes corresponding to unipolar mood disorder diagnoses used in this study. Treatment (mhvisit) Mental health outpatient Current Procedural Terminology (CPT) codes and antidepressant therapeutic drug class codes were extracted from the claims data to indicate the receipt of treatment for depressive disorders. Mental health outpatient visits are identified by CPT c odes used for billing psychological services. Using this definition a dichotomous variable was created (mhvisit) where “1” designates 1 or more mental health visits during the 24-month period and “0” indicates no mental health care visit. Treatment (medany) A dummy-coded variable was also created to examine treatment with antidepressant medication. See Table 4-4 for the list of medications and their FDA indications. The variable (medany) indicates whether any antidepressant prescription was filled in the 24-month period of the study. Because pharmacy claims cannot be linked to the specialty of the prescribing physician, all antidepressant medications listed in Table 4-2 will be counted.

PAGE 39

39 Treatment (txany) A final variable was constructed to indicate the presence or absence of any type of treatment. It is a dichotomous variable with "0" indicating no medication or mental health visits and "1" indicating treatment with either medication or a mental health visit, or both. Treatment is the dependent variable in hypothesis 1b and 2d to examine the effects of race on treatment. Somatic Symptoms of Depression (somat) The presence of somatic symptoms of depression is identified through the use of the Patient Health Questionnaire (PHQ-9), which is included in the survey. The Patient Health Questionnaire is a brief, 9-item, selfreport depression assessment specifically developed for use in primary care. It utilizes diagnostic criteria from the Diagnostic and Statis tical Manual of Mental disorders, Fourth Edition (DSM-IV) (APA, 1994). Several studies support its validity, feasibility, and its capacity to detect changes of depressive symptoms over time (Kroenke, Spitzer, Janet and Williams, 2001). A dichotomous variable (somat) was created that indicates the presence of significant somatic symptoms (a total score of 5 or more on the 4 somatic diagnostic criteria of depression: weight change, sleep difficulties, feeling agitated/slowed, and fatigue). Mental Health Expenditures Expenditures was identified through claims data and is measured in two ways. A dichotom ous variable (mhexpany ) will be created to identify the presence of any mental health expenditures with “1” indicating expenditures and “0” indicating no ex penditures. A continuous variable will be based on a dollar figure of total mental health expenditures (mhexptot). Mental health expenditures was identified through similar means as treatment through use of CPT codes to identify mental health visits and the corresponding paid dollar amount, and through the

PAGE 40

40 identification of all antidepressant medications and their paid dollar amount. The cost of all mental health expenditures was then totaled to produce one figure for each individual. Independent Variables and Covariates Race (race) The variable race was obtained from the survey data. While the Medicaid eligibility files include data on racial/ethnic gr oup classification, the data do not identify race and ethnicity separately. Coding of the race variable is “1” for Caucasian, “2” for African American, and “3” for all other races. In the analyses these are treated as three separate dummy variables with Caucasian being omitted and treated as the reference group.. Ethnicity (ethnic) Ethnicity was also obtained from self-reported survey data. The dummy variable was created with Non-Hispanic coded as “0” and Hispanic coded as “1”. In 1997 the Office of Budget Management (OBM) released federal guidelines for the assessment of race and ethnicity (of Hispanic origin or not), based on the logic that people of Hispanic origin can be any race. Ethnicity has been shown to affect access, utilization of health services and rates of disab ility. Ethnicity is used as a covariate in all analyses to control for the effects of ethnicity separately from the effects of race in the analyses. Sex (Sex) The sex of the enrollees was extracted from the eligibility file and is dummy coded (1=male, 2=female). Sex serves as a covariate in all hypotheses, where it will be used to control for the effects of sex on the outcome variable. Sex has been shown to affect depression and health care utilization. Age (age) Claims data was used to extract age at the date the survey interview was completed (a unique date for each recipient). Coding of the age variable is “1” for

PAGE 41

41 18-35, “2” 36-64, and “3” for 65 and older. In a ll hypotheses age is used as a covariate. Age has been shown to be related to diagnosis and treatment of depression, and many other health behaviors. In the analyses these are treated as three individual dummy variables with 18-35 being omitted and treated as the reference group. Medical Comorbidities (comortot) Five medical comorbidities were identified from the ICD-9 codes from the claims files: asthma and other chronic lung diseases (ICD code 490 through 496.9 ), diabetes (I CD code 250 through 250.9), heart disease (ICD codes 392.0, 393.0 through 398.99, 410 through 414.9, 415.0 through 416.9, and 420 through 429.9), hypertension (ICD code 401.0 through 405.99), and cancer (see Table 4-5 for a list of included ICD-9 cancer diagnoses). A comorbidity is indicted if any of these diagnoses are present in any of the ICD-9 code fields. These conditions have been selected for their prevalence in primary care and Medicaid in general, and their comorbidity with depressive disorders. In addition, by specifying large categories of chronic disease it will eliminate the effects of counting all comorbidities, even if they are minor and/or acute. A count variable was created that identified the number of these comorbidities that each individual has ( comortot ). Medical comorbidities will be used as a covariate or risk adjustment variable in all other analyses. Wang and colleagues (2000) suggested this parsimonious use of diagnosis counts performs comparatively to more complicated comorbidity indexes. Psychiatric Comorbidities ( psycoany) Psychiatric comorbid ities was identified from the ICD-9 diagnostic fields in the claims files (see Table 4-6 for a list of all included diagnoses). Four dichotomous dummy variabl es were created to indicate the presence

PAGE 42

42 of specific categories of psychiatric comorbid ities. These are important covariates as some diagnoses, such as schizophrenia, are likely to be associated with much higher expenditures than other psychiatric comorbidities. Dummy variables for anxiety disorders ( anxdiag ), bipolar disorders ( bipodiag ), schizophrenia ( schizdiag ), and substance abuse disorders ( sadiag ) were created. The presence of a diagnosis in each of these categories is coded as “1” and the absence of a diagnos is will be coded as a “0”. The presence of a psychiatric comorb idity has been shown to increase individuals chances of obtaining a depression diagnosis. It is hypothesized that the presence of a psychiatric comorbidity acts as a cue to physi cians to consider ment al health diagnoses. Education level (educat) Higher levels of education has been shown to be associated with better health outcomes. This variable was obtained from the survey data which asks about the highest level of education the respondent completed. Responses choose one of 6 ordinal categories (1) 8th grade or less, 2) some high school, 3) high school graduate or GED, 4) some college or two-year degree, 5) 4year degree, and 6) more than 4 year degree). This variable will be included as a covariate in all analyses to control for the effects of years of education. In the analyses these are treated a 6 individual dummy variables, with 8th grade or less being omitted and treated as the reference group. Self-reported physical and mental health (PCS & MCS). Self-reported physical and self reported mental health will be obtained from the survey data. The survey contains the Medical Outcomes Study Short Form 12 (SF-12) (Ware, Kosinski, and Keller, 1996). The SF-12 is a 12 item measure of General Health Function. It was designed to assess an individuals self-reported physical and mental quality of life over the previous 4

PAGE 43

43 weeks. The SF-12 was constructed to reproduce the SF-36 physical and mental health summary measures with at least 90% accuracy and allows for calculation of the Physical Component Summary (PCS) and the Mental Component Summary (MCS) scores. The SF-12 is a valid and reliable measure of overall health status (Ware, Kosinski, and Keller, 1996). The SF-12 is used as a covariate in all analyses to control for physical and mental health status. Number of Primary Care Visits (pcvisit) The number of visits to a primary care provider was obtained from the Medicaid outpa tient claims data. The higher the number of visits to primary care, the greater the opportunity there is to receive a depression diagnosis, so it is importent to control for this. Any claim encounter that included a provider code from those identified earlier as primary care providers were totaled to create a count variable indicating the number of visits to those providers. This will be used as a covariate in all analyses. Psychiatrist Visits (psyvisit) A dichotomous variable indicating the presence of any visits with a psychiatrist was obtained fr om the outpatient clai ms data. Individuals with any claim with a psychiatrist (specialty code: 42) or child psychiatrist (specialty code: 43) was coded as “1” and those who have had no contact with a psychiatrist were coded as “0”. It is important to control for psychiatrist visits as these individuals are more likely to have obtained treatment than those who have not seen a psychiatrist. Statistical Methods Hypothesis 1a: Depressed patients who are African American will be less likely to be diagnosed with a depressive disorder than Caucasians A Logit was used in hypothesis 1a as the outcome variable, depr ession diagnosis, is dichotomous. Ordinary Least Squares (OLS) could not be used with dic hotomous data as the error term for this

PAGE 44

44 data is not normally distributed and is heteroskedastic, which both violate the assumptions of OLS. This would result in OLS estimates that are not efficient and it is theoretically possible to obtain predictions that are below 0 or above 1 (not possible in binary/dichotomous data). Instead, the logit model was used as it does not require the dependant variable to be normally distributed, it does not require linearity between independent and dependent variables, it is robust against heteroscedasticity (the nonhomogeneity of variance), and it does not r equire normally distributed error terms. The Logit was used to examine the odds of being diagnosed with depression by race, while controlling for all other variables. The equation for the model is depicted below. logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Where p is the probability of diagnosis a nd 1-p is the odds of no diagnosis. Hypothesis 1b: Depressed patients who are African American will be less likely to receive treatment for depression than Caucasians. Three separate Logit analyses were used in hypothesis 1b as all three of the outcome variables were dichotomous (any antidepressant, any mental health visit, or any treatment). These regressions examined the odds of having any treatment by race, while controlling for all other variables. The equation for the model is depicted below. logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) +

PAGE 45

45 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Where p is the probability of treatment (any antidepressant, any MH visit, any treatment) and 1-p is the odds of no treatment. Hypothesis 1c: Depressed patients who are African American will be less likely to have any mental health care expenditures co mpared to Caucasians. A Logit was used in hypothesis 1c as the outcome variable, any mental health expenditure, is dichotomous. This regression examined the odds of having any mental health expenditures by race, while controlling for all other variables. The equation for the model is depicted below: logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Where p is the probability of any mental he alth expenditure and 1-p is the odds of no mental health expenditure. The logit transformation is defined as in hypothesis 1a. Hypothesis 1d: Depressed patients who are African American will have significantly lower mental healthcare expenditures than Caucasians. The Two-Part model was used to predict mental health expenditures due to expenditure data being censored at 0, and the high likelihood of significant skewness. This model first predicts the probability of having any expenditures usi ng the logit equation in hypothesis 1c, and then in the second part predicts expenditures given that expenditures are greater than zero, using a gamma model. While Tobit models are an estimation procedure that

PAGE 46

46 accounts for censored values (here, anyth ing below zero) of the dependent variable, which is seen in expenditures data, it was not used due to it’s sensitivity to non-normally distributed data and homoskedasticity. These are characteristics of Medicaid expenditure data. These two regressions estimate the difference in mental health expenditures by race, while controlling for all other variables. The first step of the model is as follows: Equation 1: logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Where p is the probability of any mental heal th expenditure and 1-p is the odds of no mental health expenditure. The logit trans formation is defined as in hypothesis 1a In step 2, given expenditures are >0 (identified by the logit), a gamma model was used to predict expenditure. The benefit of this model is that allows for skewness in the dependent variable, it is not sensitive to heteroskedasticity, and it does not need to be retransformed once the analysis has been run. The gamma model is as follows: f(mhexptot) = [(mhexptot /b)c-1] [(exp(-mhexptot /b)/b (c)] Hypothesis 2a: Depressed patients who endorse somatic symptoms will be less likely to be diagnosed with a depressive disorder than those who do not. A Logit was used in hypothesis 2a as the outcome variable, depression diagnosis, is dichotomous. This regression examined the odds of being diagnosed with depression by somatic symptoms, while controlling for all other variables. The equation for the model is depicted below.

PAGE 47

47 logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + 21(somat) + Where p is the probability of diagnosis of depression and 1-p is the odds of no depression diagnosis. Hypothesis 2b: Depressed patients who endorse somatic symptoms will be less likely to be receive treatment than those who do not. A Logit was used in hypothesis 2b as the outcome variable, any treatment, is dichotomous. This regression examined the odds of receiving any treatment by somatic symptoms, while controlling for all other variables. The equation for the model is depicted below. logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + 21(somat) + Where p is the probability of having any tr eatment and 1-p is the odds of no treatment. Hypothesis 2c: Depressed patients who are African Americans will be more likely to endorse somatic symptoms of depression than Caucasians. A Logit was used in hypothesis 2c as the outcome variable, signi ficant somatic symptoms, is dichotomous. This regression examined the odds of experiencing somatic symptoms by race, while controlling for all other variables. The equation for the model is depicted below:

PAGE 48

48 logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Where p is the probability of somatic sympto ms and 1-p is the odds of no somatic symptoms. Hypothesis 2d: Of those who are depressed, the number of somatic symptoms of depression will mediate the relationship between race and the diagnosis of depression. A Logit was used in hypothesis 2d as the outcome variable, depression diagnosis, is dichotomous. In order to measure the m ediating effects of somatic symptoms between race and diagnosis, the change in the race odds ratio from the base model in hypothesis 1a was compared with the more fully specified model which includes the somatic symptoms variable. This is a common method of measuring the mediation of variables (Baron & Kenny, 1986). If race effects disappear (change from significant to non-significant) when somatic symptoms are controlled for, this indicates that somatic symptoms do in fact play a significant role in mediating the relationship between race and diagnosis. However, some critics argue that a coefficient changing from below the arbitrary 0.05 threshold to above is not the most conservative approach. For this reason, to ensure that the change in the race coefficient is a meaningful change the difference between the two coefficients was tested for statistical significance using the Hausman Test. The equations for the models are depicted below: Equation 1: logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) +

PAGE 49

49 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Equation 2: logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + 21(somat) + Where p is the probability of diagnosis of depression and 1-p is the odds of no depression diagnosis. Hypothesis 2e: Of those who are depressed, the number of somatic symptoms of depression will mediate the relationship between race and the treatment of depression. As before a Logit was used in hypothesis 2e as the outcome variable depression treatment is dichotomous. The same technique was used as in hypothesis 2d to test for the mediating effects of somatic symptoms between race and treatment. The equations for the models are depicted below: Equation 1: logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) + 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + Equation 2: logit(p) = o + 1(psyvisit) + 2(sex) + 3(ethnic) + 4(comortot) + 5(anxdiag) + 6(bipodiag) + 7(schizdiag) + 8(sadiag) + 9(educat1) + 10(educat2) +

PAGE 50

50 11(educat3) + 12(educat4) + 13(educat5) + 14(pcs) + 15(mcs) + 16(pcvisit) + 17(age1) + 18(age2) + 19(RaceAA) + 20(Raceother) + 21(somat) + Where p is the probability of treatment (antidepressant, psychotherapy, and any treatment) and 1-p is the odds of no treatment.

PAGE 51

51 Table 4-1 Sample characteristics of the full random sample and of the stratified survey sample. Total Random Sample (non-surveyed) Survey Sample N = 29,669 N = 2,106 Sex Male 56% 50% Age 46.7 years 44.7 years Race (Medicaid Data) White 43.0 % 36.9% Black 24.4% 30.4% American Indian .1% .0% Oriental .5% .3% Hispanic 12.4% 16.5% Other 19.6% 15.9% Table 4-2 Probability of major depression or any depressive disorder at each score of the PHQ-2. PHQ-2 score Probability of major depressive disorder (%) Probability of any depressive disorder (%) 1 15.4 36.9 2 21.1 48.3 3 38.4 75.0 4 45.5 81.2 5 56.4 84.6 6 78.6 92.9

PAGE 52

52 Table 4-3 ICD-9 diagnosis codes for all depression diagnoses Unipolar Mood Disorder ICD-9-CM Depression NOS 311.00 Major Depressive Disorder-Single Episode 296.20-296.26 Major Depressive Disorder-Recurrent Episode 296.30-296.36 Dysthymia 300.40 Table 4-4 A list of all included Antidepressant Medications Drug Class Generic Name Brand Name Antidepressant Imipramine Tofanil Desipramine Norpramin Amitriptyline Elavil Nortriptyline Aventyl, Pamelor Protriptyline Vivacil Trimipramine Surmontil Doxepin Sinequan, Adapin Maprotiline Ludiomil Amoxapine Asendin Trazodone Desyrel Fluoxetine Prozac Bupropion Wellbutrin Sertraline Zoloft Paroxetine Paxil Venlafaxine Effexor Nefazodone Serzone Fluvoxamine Luvox Phenelzine Nardil Tranylcypromine Parnate

PAGE 53

53 Table 4-5 ICD-9 Diagnosis codes for common cancers Cancer Group ICD-9 code ranges Bones/soft tissue 170, 171, 238.1, 238.2 Brain 190–192.9, 237.5, 237.6, 239.6 Breast 174, 175, 239.3 Colon 153, 154, 235.2 Endocrine 193, 194, 237.0, 237.4, 239.7 Gyn 180, 182, 183, 184, 236.1, 236.2 Head and neck 140–149.9, 160, 161, 162, 195.0 Lung 162, 235.9, 239.1 Lymph node spread 196 Melanoma 172 Non-colon GI 150–152.9, 155–159.9, 235, 239.0 Non-melanomatous skin cancer 173, 238.2 Non-specific site 195, 199, 238.8, 238.9, 239.8, 239.9 Pleura/mediastinum 163, 164 Prostate 185, 236.5 Secondary cancer 196, 197 Testes/Male GU 186, 187.3, 187.4, 187.9, 236.4, 236.6 Urinary Tract 188, 189, 236.7, 236.91, 239.4, 239.5

PAGE 54

54 Table 4-6 Categories of Psychiatric Comorbidities: ICD-9-CM diagnosic codes Psychiatric Comorbidity Ca tegories ICD-9-CM Codes Anxiety Disorders Anxiety State-Unspecified 300.00 Panic Disorder 300.01 Generalized Anxiety Disorder 300.02 Other Anxiety State 300.09 Phobia-Specified 300.20 Agoraphobia with Panic Attacks 300.21 Agoraphobia without Panic Attacks 300.22 Social Phobia 300.23 Other Phobias 300.29 Obsessive-Compulsive Disorder 300.30 Acute Reaction to Stress 308.30 Adjustment Disorder with Anxious Mood 309.24 Prolonged Post Traumatic Stress Disorder 309.81 Schizophrenia 295.xx Bipolar Disorders Bipolar – Single Episode 296.00-296.06 Bipolar I 296.40-296.60 Bipolar II 296.89 Bipolar NOS 296.80 Substance Abuse Disorders Alcohol-induced Disorders 291.xx Drug-induced Disorders 292.xx

PAGE 55

55 CHAPTER 5 RESULTS Sample Characteristics Final Survey Sample Initially, the data collection obtained 2,411 completed surveys. Using the study criteria, 2,106 Medipass Medicaid recipients we re included in the final sample, this was lower than the 2,411 sampled as several Medicaid recipients did not have any primary care claims within 2 years of their survey date. Table 5-1 shows sample characteristics of this sample. Almost 53% of the full survey sample was male Given that sampling was stratified by sex to include equal proportions of male and females, this is an overrepresentation of men in the final survey sample, presumably due to the fact that men had more primary care claims and so were more likely to end up in the sample. African American and Hispanic enrollees also appeared to be overrepresented (33.5% and 32.7% respectively in the survey sample, as opposed to 24.4% and 12.4% in the non survey sample), however, it is impossible to compare these numbers in this way due to the different way that race and ethnicity is collected in Medicaid claims and the in the survey sample. The age of the survey group was only slightly lower than the nonsurveyed group (44.7 versus 45.8 years). Of the overall survey sample, 6.9% obtained a physician diagnosis of a depressive disorder, and 23.8% received some kind of mental health treatment. One third (33.7%) of the sample met criteria for a depressive disorder as measured by the PHQ-2 (N = 709). Of the 68 patients diagnosed with depression by their physician, 82% received at least one prescription for antidepressants. Thirty six percent of physician-diagnosed cases received at least one mental health visit, as assessed by the CPT codes.

PAGE 56

56 PHQ-2 Identified Depressed Sample Table 5-2 shows sample characteristics of the PHQ-2 identified depressed sample. The majority (59.8%) of the sample was Caucasian (N = 422) and 29.2% were African American (N= 206), the remainder, classified as “other”, accounted for 11% of recipients’ (N = 78) race group membership. Almost 35% of the sample indicated that they were of Hispanic ethnicity (N = 256). This race/ethnicity distribution corresponds to a slight over-representation of Caucasians as compared to African Americans and Hispanics when compared to the overall survey sample. The mean age of recipients in the study sample was 48.16 years (SD = 12.5), with a range fro 18 years to 102 years. Most of the sample were high school graduates, however only 7.6% of the sample had a college degree. Their PCS and MCS scores were both very low compared to national norms, indicating poorer physical and mental functioning. A large percent of the sample also endorsed significant somatic symptoms of depression (84%). This level of physical complaints does appear to be reflected in the average number of primary care visits during the period of the study, which was 39 visits. Of the 709 cases identified as depressed by the PHQ-2. The rate of physician diagnosis among this group was 11.1%. The rate of antidepressant treatment was 35.7% and 8.4% of survey-identified cases received at least one mental health visit, while 37% received any type of mental health treatment. Eighteen percent had at least one psychiatrist visit. The average mental health expenditure of the sample was $385.11, with a large number having $0 expenditures and heavy skewness and kurtosis.

PAGE 57

57 Research Goals Aim 1. To examine racial disparities in the diagnosis of depression and treatment of depression the Medicaid primary care population. Hypothesis 1a Hypothesis 1a states that patients who are African American will be less likely to be diagnosed with a depressive disorder than Caucasians. The results of the Logit indicates that this analysis does not support this hypothesis, see table 5-3 for the results of the Logit analysis. While controlling for the covariates of age, sex, education, number of physical comorbidities, number of primar y care visits, psychiatrist visit, physical and mental wellbeing, and being diagnosed with an anxiety disorder, bipolar disorder, or a psychotic disorder; being African American did not impact the likelihood of being diagnosed with depression in Medicaid Primary Care (OR = .973, p .96). The variable substance abuse diagnosis was dropped due to the lack of this diagnosis in all of the patients. This analysis highlighted the effect of psychiatric caseness, whereby patients that already have seen a psychiatrists are sign ificantly more likely to obtain a diagnosis in primary care when compare with those who did not have a visit with a psychiatrist. Here we see very large effect sizes with those who had seen a psychiatrist having almost 40 times the odds of being diagnosed with depression (OR = 39.765, p .001). Of note is the very small numbers involved in the analysis, crosstabs showed that only 9 African Americans in the sample obtained a diagnosis of depression from their physician, indicating the lack of power in any analyses examining race and diagnosis. See Table 5-4 for the results of the power analysis.

PAGE 58

58 Hypothesis 1b Hypothesis 1b states that patients who are African Americans will be less likely to receive treatment (antidepressants, a mental health visit, or any treatment) for depression than Caucasians. The results of the 3 separate Logits support this hypothesis. See table 5-5, 5-6, and 5-7 for the results of the three respective Logit analyses. While controlling for all covariates, Caucasians had almost 4 times the odds receiving antidepressants when compared to African Americans (OR = .258, p .001), race does not predict mental health visits (OR = .203, p .461), but it predicted receiving any type of mental health treatment in Medi caid Primary Care, with Caucasians having approximately 4 times the odds of receiving any mental health treatment when compared with African Americans (OR = .239, p .001). Being female, having more PCP visits, having a psychiatr ist visit, and being above the age of 65 all increased the odds of receiving any depression treatment. The lack of effect in predicting mental health visits appears to be due to a lack of power due to small sample size. Hypothesis 1c Hypothesis 1c states that patients who are African American will be less likely to have any mental health expenditures than Caucasians. The results of the Logit analysis supports this hypothesis, see table 5-8 for the results of the Logit analysis. While controlling for the covariates of age, sex, education, number of physical comorbidities, number of primary care visits, psychiatris t visit, physical and m ental wellbeing, and being diagnosed with an anxiety disorder, bipolar disorder, or a psychotic disorder; Caucasian had 4 times the odds of Africa n Americans of having any mental health expenditures (OR = .239, p .001). Being female, having more PCP visits, having a psychiatrist visit, being above the age of 65, all increased the likelihood of having any

PAGE 59

59 mental health expenditures. Table 5-9 shows the average expenditure of each racial group. Hypothesis 1d Hypothesis 1d states that patients who are African American will have lower mental health expenditures than Caucasians. The results of the Gamma regression show a trend to supporting this hypotheses, table 5-10 shows the results of the Gamma analysis. Assuming expenditures were greater than zero, African Americans had 43% lower mental health expenditures ( =-.433; p .087), relative to Caucasians, however this did not reach significance. Aim 2: To examine the role of somatic symptoms in the diagnosis and treatment of depression Hypothesis 2a Hypothesis 2a states that patients who endorse greater somatic symptoms will be less likely to be diagnosed with a depressive disorder than those who do not. For this model, significant somatic symptoms (the di chotomous variable) proved to have the strongest relationship with diagnosis and so was used in place of a count of somatic symptoms. The results of the Logit do not support this hypothesis, see table 5-11 for the results of the Logit analysis. While controlling for all covariates (including the number of physical comorbidities and the score on the SF-12 PCS) the presence of significant somatic complaints does not impact the likelihood of being diagnosed with depression in Medicaid primary care (OR = .433, p .137). However, this analysis almost met significance, indicating that there may be some influence of somatic symptoms on diagnosis and this lack of effect is likely due to a lack of power (1- err prob = .31), see Table 5-12 for the power analysis.

PAGE 60

60 However, there may be shared variance between somatic symptoms and the 2 other measure of physical wellbeing: number of physical comorbidities and the Physical Component Score of the SF12 (PCS). It is possible that the inclusion of both these in the analyses may be masking the effect of somatic symptoms. While the number of physical comorbidities does not appear to significantly impact any of the analyses, PCS does appear to influence several of the analyses. To test this theory the analysis was re-run without the PCS variable, and the results are shown in Table 5-13. The removal of the PCS variable weakened the relationship between somatic symptoms and diagnosis (OR = .50, p .21). Hypothesis 2b Hypothesis 2b states that patients who endorse greater somatic symptoms will be less likely to receive any mental health treatment than those who do not. The results of the Logit do not support this hypothesis, see table 5-14 for the results of the Logit analysis. While controlling for all covariat es (including the number of physical comorbidities and the score on the SF-12 PCS) the presence of significant somatic complaints does not impact the likelihood of receiving any mental health treatment in Medicaid primary care (OR = 1.091, p .795). As with the previous analysis, this analysis was re-run with the PCS variable removed and the results are shown in Table 5-15. In this case, the removal of PCS did appear to influence the results, although the results did not become significant, removing PCS did strengthen the relationship between somatic symptoms and receiving any treatment (OR = 1.56, p .157). However, this shows the reverse of the expected direction, that when depressed individuals have significant somatic symptoms of

PAGE 61

61 depression their likelihood of obtaining treatment increases, although, this did not reach significance. Hypothesis 2c Hypothesis 2c states that patients who are African Americans will be more likely to endorse somatic symptoms of depression than Caucasians. The results of the Logit do not support this hypothesis, see table 5-16 for the results of the Logit analysis. While controlling for all covariates being African American does not impact the likelihood of endorsing somatic symptoms in Medicaid Primary Care (OR = .900, p .733). As with the previous two analyses, this analysis was re-run with the PCS variable removed and the results are shown in Table 5-17. The removal of PCS from the analyses did impact the relationship between somatic symptoms of depression and race. With the removal of PCS from the analyses this relationship became very close to significant (OR = .588, p .062), however this was also in the reverse direction to what was predicted, that Caucasians have approxim ately 1.7 the odds of endorsing somatic symptoms of depression, when compared with African Americans. Hypothesis 2d Hypothesis 2d states that the number of somatic symptoms of depression will mediate the relationship between race and the diagnosis of depression. This was achieved through examining the change in the odds ratio for Race when somatic symptoms is added to the model predicting diagnosis. The addition of somatic symptoms to this equation did not significantly change the value of the odds ratio of Race (8.4% change, p .480). This indicates that relationship between race and diagnosis of depression is not mediated by somatic symptoms. This change in the odds ratio can be seen in Table 5-18.

PAGE 62

62 Once more, due to the possible shared variance between somatic symptoms and the PCS score, the analysis was re-run with PCS removed from the model. The results are shown in Table 5-19. The removal of PCS from the model predicting diagnosis did not significantly change the value of the odds ratio of Race (9.8% change, p .73). This again indicates that the relationship between race and diagnosis of depression is not mediated by somatic symptoms. Hypothesis 2e Hypothesis 2e states that the number of somatic symptoms of depression will mediate the relationship between race and the treatment of depression. This was achieved through examining the change in the odds ratio for Race when somatic symptoms is added to the model predicting any treatment. The addition of somatic symptoms to this equation did not significantly change the value of the odds ratio of Race (.8% change, p .899). This indicates that relationship between race and treatment is not mediated by somatic symptoms. This change in the odds ratio can be seen in table 5-20. The analysis was re-run with PCS removed from the model and the results are shown in Table 5-21. The removal of PCS from the model predicting treatment did not significantly change the value of the odds ratio of Race (1.7% change, p .90). This once more indicates that the relationship between race and treatment of depression is not mediated by somatic symptoms.

PAGE 63

63 Table 5-1 Sample Characteristics of the final survey sample Final Survey Sample N = 2,106 Sex (Male) 52.4% Age mean 45.86 years Race Caucasian 54.6% African American 33.5% Other 11.9% Ethnicity (Hispanic) 32.7% Education 8th Grade or less 13.3% Some High School 21.9% High School Grad 34.7% Some College 22.3% 4 Year College 5.5% Post Graduate 2.4% Physician Diagnosis Depression 6.9% Any Treatment 23.8% PHQ-2 Criteria of Depression 33.7%

PAGE 64

64 Table 5-2 Sample Characteristics of the PHQ-2 identified depressed sample PHQ-2 Depressed Sample N = 709 Sex (Male) 49.6% Age mean (range) 48.16 years (18-102) Race Caucasian 59.8% African American 29.2% Other 11.0% Ethnicity (Hispanic) 34.8% Education 8th Grade or less 15.7% Some High School 23.8% High School Grad 34.4% Some College 18.5% 4 Year College 5.3% Post Graduate 2.3% PCS mean (SD) 35.77 (12.83) MCS mean (SD) 39.94 (12.20) Significant Somatic Symptoms 83.6% Number of Physical Comorbidities 0 81.1% 1 7.8% 2 6.9% 3 2.7% 4 1.1% 5 .4% Number of PCP Visits Mean (SD) 38.73 (85.98) Psychiatrist visit 18.2% Physician Diagnosed Depressive Disorder 11.1% Anxiety Disorder 2.1% Schizophrenia 0.4% Bipolar Disorder 0.1% Any Antidepressant 35.7% Any Mental Health Visit 8.4% Any Mental Health Treatment 37.1% Total Mental Health Expenditure Mean (SD) $385.11 ($993.20) Range $0 $11,532.62 Skewness, Kurtosis 4.53, 30.74

PAGE 65

65 Table 5-3 Results of the Logit Analysis for Hypothesis 1a B S.E. Wald Sig. Exp(B) 95% C.I. for EXP(B) Lower Upper Constant .079 1.319 .004 .952 1.082 1.082 .079 Race (Caucasian) .842 .656 African American -.027 .522 .003 .959 .973 .350 2.710 Other .540 .598 .816 .366 1.717 .531 5.546 Ethnicity -.307 .414 .551 .458 .735 .327 1.655 Sex .330 .367 .806 .369 1.390 .677 2.856 Number of Physical Comorbidities .130 .221 .342 .559 1.138 .737 1.757 Anxiety Diagnosis .723 .811 .795 .372 2.061 .421 10.097 Schizophrenia Diag -.724 1.392 .271 .603 .485 .032 7.411 Bipolar Diagnosis 22.414 40192.970 .000 1.000 .000 .000 Education 10.069 .073 Some High School -1.798 .632 8.078 .004 .166 .048 .572 High School Grad -.825 .537 2.365 .124 .438 .153 1.254 Some College -.673 .597 1.270 .260 .510 .158 1.645 4 Year College .450 .869 .269 .604 1.569 .286 8.609 Post Graduate -1.281 1.357 .891 .345 .278 .019 3.969 Number of PCP Visits .001 .001 .426 .514 1.001 .998 1.004 Psychiatrist visit 3.683 .475 60.190 .000 39.765 15.683 100.830 SF12 PCS -.019 .018 1.167 .280 .981 .947 1.016 SF12 MCS -.066 .018 13.688 .000 .936 .904 .969 Age Categories (18-35) 1.683 .431 Age 36-64 -.399 .557 .513 .474 .671 .225 1.999 Age 65+ .369 .744 .246 .620 1.447 .336 6.222 This table shows the results from the logit analysis predicting depression diagnosis, race is the variable of interest here.

PAGE 66

66 Table 5-4 Power analysis for Hypothesis 1a Tails 1 Odds Ratio .973 err prob .050 Total Sample Size 528.000 R2 .480 Critical z -1.640 Power .072

PAGE 67

67 Table 5-5 Results of the Logit Analysis for Hypothesis 1b predicting treatment with antidepressants B S.E. Wald Sig. Exp(B) 95% C.I.for EXP(B) Lower Upper Constant 2.174 .870 6.247 .012 8.791 Race (Caucasian) 26.783 .000 African American -1.355 .304 19.872 .000 .258 .142 .468 Other -1.289 .438 8.683 .003 .275 .117 .649 Ethnicity -.017 .294 .003 .954 .983 .552 1.749 Sex .428 .244 3.076 .079 1.534 .951 2.473 Number of Physical Comorbidities .036 .142 .065 .799 1.037 .785 1.369 Anxiety Diagnosis 1.512 1.092 1.917 .166 4.536 .533 38.564 Schizophrenia Diag 19.254 21768.816 .000 .999 2.301E8 .000 Education 5.161 .397 Some High School .281 .393 .513 .474 1.325 .614 2.861 High School Grad .612 .386 2.516 .113 1.844 .866 3.929 Some College .679 .428 2.519 .112 1.972 .853 4.559 4 Year College -.007 .567 .000 .990 .993 .327 3.014 Post Graduate -.298 .811 .135 .713 .742 .151 3.638 Number of PCP Visits .004 .002 3.676 .055 1.004 1.000 1.007 Saw a psychiatrist 1.681 .362 21.600 .000 5.373 2.644 10.920 SF12 PCS -.043 .011 14.459 .000 .958 .936 .979 SF12 MCS -.041 .011 15.366 .000 .960 .940 .980 Age Categories (1835) 3.244 .197 Age 36-64 .310 .370 .703 .402 1.364 .660 2.816 Age 65+ .731 .419 3.048 .081 2.077 .914 4.719 This table shows the results from the logit analysis examining race in predicting treatment with antidepressants.

PAGE 68

68 Table 5-6 Results of the Logit Analysis for Hypothesis 1b predicting mental health visits B S.E. Wald Sig. Exp(B) 95% C.I.for EXP(B) Lower Upper Constant .853 1.314 .422 .516 2.347 Race (Caucasian) 3.216 .200 African American -.775 .609 1.618 .203 .461 .140 1.520 Other -1.040 .760 1.873 .171 .354 .080 1.567 Ethnicity -.562 .420 1.787 .181 .570 .250 1.299 Sex -.832 .401 4.309 .038 .435 .198 .955 Number of Physical Comorbidities .147 .221 .443 .505 1.158 .752 1.785 Anxiety Diagnosis -1.735 1.226 2.001 .157 .176 .016 1.952 Schizophrenia Diag 2.052 1.679 1.494 .222 7.786 .290 209.078 Bipolar Diagnosis 18.216 40192.970 .000 1.000 .000 .000 Education 9.583 .088 Some High School -.998 .566 3.106 .078 .369 .121 1.118 High School Grad -1.389 .558 6.204 .013 .249 .084 .744 Some College -1.818 .693 6.891 .009 .162 .042 .631 4 Year College -1.588 1.231 1.665 .197 .204 .018 2.281 Post Graduate -.566 1.108 .261 .609 .568 .065 4.977 Number of PCP Visits .000 .001 .005 .941 1.000 .997 1.003 Psychiatrist visit 3.220 .451 50.942 .000 25.031 10.338 60.604 SF12 PCS -.028 .019 2.127 .145 .973 .937 1.010 SF12 MCS -.035 .018 3.954 .047 .966 .933 1.000 Age Categories (18-35) 2.002 .368 Age 36-64 -.789 .558 2.000 .157 .454 .152 1.356 Age 65+ -.640 .798 .642 .423 .528 .110 2.520 This table shows the results from the logit analysis examining race in predicting treatment with a mental health visit.

PAGE 69

69 Table 5-7 Results of the Logit Analysis for Hypothesis 1b predicting any type of treatment B S.E. Wald Sig. Exp(B) 95% C.I.for EXP(B) Lower Upper Constant 1.354 .744 3.306 .069 3.872 Race (Caucasian) 32.474 .000 African American -1.431 .273 27.433 .000 .239 .140 .408 Other -1.002 .388 6.671 .010 .367 .172 .785 Ethnicity .136 .251 .294 .588 1.146 .701 1.873 Sex .406 .207 3.839 .050 1.501 1.000 2.253 Number of Physical Comorbidities .237 .130 3.313 .069 1.267 .982 1.635 Anxiety Diagnosis 1.920 1.087 3.120 .077 6.820 .810 57.404 Schizophrenia Diag 19.805 21940.476 .000 .999 3.990E8 .000 Bipolar Diagnosis 21.364 40192.970 .000 1.000 .000 .000 Education 2.592 .763 Some High School .226 .351 .414 .520 1.253 .630 2.493 High School Grad .183 .334 .298 .585 1.200 .623 2.312 Some College .156 .367 .181 .671 1.169 .569 2.402 4 Year College -.398 .513 .604 .437 .672 .246 1.834 Post Graduate -.394 .675 .342 .559 .674 .180 2.531 Number of PCP Visits .004 .002 5.381 .020 1.004 1.001 1.007 Psychiatrist visit 1.625 .281 33.362 .000 5.077 2.925 8.811 SF12 PCS -.044 .010 20.515 .000 .956 .938 .975 SF12 MCS -.032 .009 12.760 .000 .968 .951 .986 Age Categories (18-35) 14.174 .001 Age 36-64 .254 .306 .687 .407 1.289 .707 2.351 Age 65+ 1.237 .367 11.383 .001 3.445 1.679 7.067 This table shows the results from the logit analysis examining race in predicting any type of treatment for depression.

PAGE 70

70 Table 5-8 Results of the Logit Analysis for Hypothesis 1c predicting any mental health expenditure B S.E. Wald Sig. Exp(B) 95% C.I. for EXP(B) Lower Upper Constant 1.354 .744 3.306 .069 3.872 Race (Caucasian) 32.474 .000 African American -1.431 .273 27.433 .000 .239 .140 .408 Other -1.002 .388 6.671 .010 .367 .172 .785 Ethnicity .136 .251 .294 .588 1.146 .701 1.873 Sex .406 .207 3.839 .050 1.501 1.000 2.253 Number of Physical Comorbidities .237 .130 3.313 .069 1.267 .982 1.635 Anxiety Diagnosis 1.920 1.087 3.120 .077 6.820 .810 57.404 Schizophrenia Diag 19.805 21940.476 .000 .999 3.990E8 .000 Bipolar Diagnosis 21.364 40192.970 .000 1.000 .000 .000 Education 2.592 .763 Some High School .226 .351 .414 .520 1.253 .630 2.493 High School Grad .183 .334 .298 .585 1.200 .623 2.312 Some College .156 .367 .181 .671 1.169 .569 2.402 4 Year College -.398 .513 .604 .437 .672 .246 1.834 Post Graduate -.394 .675 .342 .559 .674 .180 2.531 Number of PCP Visits .004 .002 5.381 .020 1.004 1.001 1.007 Psychiatrist visit 1.625 .281 33.362 .000 5.077 2.925 8.811 SF12 PCS -.044 .010 20.515 .000 .956 .938 .975 SF12 MCS -.032 .009 12.760 .000 .968 .951 .986 Age Categories (18-35) 14.174 .001 Age 36-64 .254 .306 .687 .407 1.289 .707 2.351 Age 65+ 1.237 .367 11.383 .001 3.445 1.679 7.067 This table shows the results from the logit analysis examining race in predicting any mental health expenditure.

PAGE 71

71 Table 5-9 Results of the Gamma Model for Hypothesis 1d predicting mental health expenditure (assuming expenditures > $0) B S.E. 95% Wald Confidence Interval Hypothesis Testing Lower Upper Wald Chi-Square Sig. Intercept 7.041 .0959 6.853 7.229 5389.784 .000 Race (Caucasian) African American -.433 .2525 -.928 .062 2.934 .087 Other -.537 .3248 -1.173 .100 2.728 .087 Table 5-10 Means of mental health expenditure by race (assuming expenditures > $0). Mean Race Caucasian $1,142.82 African American $741.48 Other $668.29

PAGE 72

72 Table 5-11 Results of the Logit Analysis for Hypothesis 2a predicting depression diagnosis B S.E. Wald Sig. Exp(B) 95% C.I. for EXP(B) Lower Upper Constant 21.890 40193.012 .000 1.000 .000 Significant Somatic -.838 .564 2.207 .137 .433 .143 1.307 Race (Caucasian) .534 .766 African American -.114 .539 .045 .833 .892 .310 2.566 Other .403 .605 .444 .505 1.497 .457 4.903 Ethnicity .377 .418 .810 .368 1.457 .642 3.309 Sex .280 .369 .576 .448 1.324 .642 2.729 Number of Physical Comorbidities .125 .224 .312 .576 1.134 .730 1.760 Anxiety Diagnosis -.779 .806 .936 .333 .459 .095 2.225 Schizophrenia Diag .685 1.387 .244 .621 1.984 .131 30.081 Bipolar Diagnosis 22.990 40193.012 .000 1.000 9.649E9 .000 Education 10.349 .066 Some High School -1.842 .639 8.315 .004 .159 .045 .554 High School Grad -.853 .542 2.475 .116 .426 .147 1.233 Some College -.722 .603 1.433 .231 .486 .149 1.585 4 Year College .456 .869 .275 .600 1.577 .287 8.655 Post Graduate -1.186 1.348 .774 .379 .305 .022 4.291 Number of PCP Visits .001 .001 .434 .510 1.001 .998 1.004 Psychiatrist visit 3.664 .474 59.872 .000 39.018 15.424 98.705 SF12 PCS -.025 .018 1.767 .184 .976 .941 1.012 SF12 MCS -.073 .019 15.288 .000 .930 .897 .964 Age Categories (18-35) 1.611 .447 Age 36-64 -.422 .559 .570 .450 .656 .219 1.960 Age 65+ .315 .741 .180 .671 1.370 .321 5.849 This table shows the results from the logit analysis examining somatic symptoms in predicting diagnosis.

PAGE 73

73 Table 5-12 Power analysis for hypothesis 2a Tails 1 Odds Ratio .433 err prob .050 Total Sample Size 521.000 R2 .490 Critical z -1.644 Power .31

PAGE 74

74 Table 5-13 Results of the Logit Analysis for Hypothesis 2a predicting depression diagnosis, with PCS removed B S.E. Wald Sig. Exp(B) 95% C.I.for EXP(B) Lower Upper Constant 23.724 40193.049 .000 1.000 .000 Significant Somatic -.680 .544 1.566 .211 .506 .175 1.470 Race (Caucasian) .979 .613 African American -.182 .537 .115 .735 .834 .291 2.387 Other .513 .589 .758 .384 1.670 .526 5.295 Ethnicity .275 .410 .451 .502 1.317 .590 2.941 Sex .285 .368 .599 .439 1.329 .646 2.735 Number of Physical Comorbidities .211 .215 .967 .326 1.235 .811 1.881 Anxiety Diagnosis -.724 .818 .782 .376 .485 .098 2.410 Schizophrenia Diag .712 1.385 .264 .608 2.037 .135 30.775 Bipolar Diagnosis 23.372 40193.049 .000 1.000 1.413E10 .000 Education 11.205 .047 Some High School -1.902 .637 8.929 .003 .149 .043 .520 High School Grad -.887 .540 2.700 .100 .412 .143 1.186 Some College -.760 .600 1.601 .206 .468 .144 1.517 4 Year College .490 .860 .324 .569 1.632 .302 8.807 Post Graduate -1.261 1.324 .907 .341 .283 .021 3.795 Number of PCP Visits .001 .001 .521 .470 1.001 .998 1.004 Psychiatrist visit 3.672 .478 59.084 .000 39.326 15.419 100.299 SF12 MCS -.068 .018 14.195 .000 .934 .902 .968 Age Categories (18-35) 1.364 .506 Age 36-64 -.138 .513 .072 .788 .871 .319 2.383 Age 65+ .600 .709 .716 .398 1.822 .454 7.319

PAGE 75

75 Table 5-14 Results of the Logit Analysis for Hypothesis 2b predicting any depression treatment B S.E. Wald Sig. Exp(B) 95% C.I. for EXP(B) Lower Upper Constant 1.796 45787.106 .000 1.000 6.028 Significant Somatic .087 .333 .068 .795 1.091 .567 2.096 Race (Caucasian) 32.320 .000 African American -1.429 .273 27.363 .000 .240 .140 .409 Other -.997 .388 6.605 .010 .369 .172 .789 Ethnicity -.135 .251 .291 .590 .873 .534 1.428 Sex .409 .208 3.884 .049 1.505 1.002 2.261 Number of Physical Comorbidities .235 .130 3.256 .071 1.265 .980 1.632 Anxiety Diagnosis -1.926 1.087 3.138 .076 .146 .017 1.227 Schizophrenia Diag 19.801 21931.378 .000 .999 .000 .000 Bipolar Diagnosis 21.292 40193.026 .000 1.000 1.765E9 .000 Education 2.637 .756 Some High School .234 .352 .440 .507 1.263 .633 2.518 High School Grad .189 .335 .317 .573 1.208 .626 2.329 Some College .161 .368 .191 .662 1.174 .571 2.415 4 Year College -.396 .513 .598 .439 .673 .246 1.838 Post Graduate -.394 .675 .340 .560 .675 .180 2.533 Number of PCP Visits .004 .002 5.427 .020 1.004 1.001 1.007 Psychiatrist visit 1.625 .281 33.397 .000 5.077 2.926 8.810 SF12 PCS -.044 .010 18.828 .000 .957 .938 .976 SF12 MCS -.032 .009 11.475 .001 .969 .951 .987 Age Categories (18-35) 14.139 .001 Age 36-64 .254 .307 .686 .408 1.289 .707 2.351 Age 65+ 1.235 .367 11.355 .001 3.440 1.677 7.057 This table shows the results from the logit analysis examining somatic symptoms in predicting depression treatment.

PAGE 76

76 Table 5-15 Results of the Logit Analysis for Hypothesis 2b predicting any depression treatment, with PCS removed B S.E. Wald Sig. Exp(B) 95% C.I. for EXP(B) Lower Upper Constant -1.040 45780.128 .000 1.000 .353 Significant Somatic .442 .313 2.003 .157 1.556 .843 2.871 Race (Caucasian) 38.170 .000 African American -1.569 .267 34.574 .000 .208 .123 .351 Other -.877 .375 5.479 .019 .416 .200 .867 Ethnicity -.334 .242 1.906 .167 .716 .446 1.151 Sex .426 .203 4.381 .036 1.531 1.027 2.280 Number of Physical Comorbidities .288 .130 4.932 .026 1.333 1.034 1.719 Anxiety Diagnosis -2.212 1.096 4.070 .044 .109 .013 .939 Schizophrenia Diag 19.703 21916.579 .000 .999 .000 .000 Bipolar Diagnosis 21.715 40193.062 .000 1.000 2.695E9 .000 Education 3.221 .666 Some High School .305 .345 .779 .377 1.356 .689 2.668 High School Grad .242 .326 .549 .459 1.273 .672 2.413 Some College .122 .358 .116 .733 1.130 .560 2.278 4 Year College -.382 .495 .594 .441 .683 .258 1.803 Post Graduate -.339 .675 .252 .615 .713 .190 2.673 Number of PCP Visits .004 .002 6.795 .009 1.004 1.001 1.008 Psychiatrist visit 1.498 .273 30.186 .000 4.472 2.621 7.630 SF12 MCS -.021 .009 5.516 .019 .980 .963 .997 Age Categories (18-35) 22.103 .000 Age 36-64 .780 .283 7.569 .006 2.181 1.251 3.800 Age 65+ 1.653 .352 22.063 .000 5.223 2.620 10.410

PAGE 77

77 Table 5-16 Results of the Logit Analysis for Hypothesis 2c predicting significant somatic symptoms B S.E. Wald Sig. Exp(B) 95% C.I.for EXP(B) Lower Upper Constant 9.467 1.236 58.680 .000 12925.902 Race (Caucasian) 1.004 .605 African American -.105 .307 .117 .733 .900 .493 1.644 Other -.434 .441 .969 .325 .648 .273 1.537 Ethnicity -.145 .331 .191 .662 .865 .453 1.655 Sex .416 .259 2.583 .108 1.515 .913 2.515 Number of Physical Comorbidities .241 .214 1.268 .260 1.273 .837 1.936 Anxiety Diagnosis -.631 1.134 .309 .578 .532 .058 4.914 Schizophrenia Diag 18.181 21372.373 .000 .999 7.868E7 .000 Bipolar Diagnosis 23.437 40192.970 .000 1.000 .000 .000 Education 4.786 .443 Some High School -.864 .498 3.009 .083 .422 .159 1.119 High School Grad -.879 .474 3.436 .064 .415 .164 1.052 Some College -.581 .510 1.300 .254 .559 .206 1.519 4 Year College -.183 .679 .073 .787 .833 .220 3.150 Post Graduate -.024 1.327 .000 .986 .976 .073 13.146 Number of PCP Visits .000 .001 .214 .644 .999 .997 1.002 Psychiatrist visit .156 .394 .156 .692 1.169 .539 2.532 SF12 PCS -.081 .012 43.961 .000 .922 .900 .944 SF12 MCS -.087 .012 52.559 .000 .917 .895 .938 Age Categories (18-35) 1.430 .489 Age 36-64 -.366 .330 1.224 .269 .694 .363 1.326 Age 65+ -.047 .433 .012 .914 .954 .408 2.231 This table shows the results from the logit analysis examining race in predicting somatic symptoms.

PAGE 78

78 Table 5-17 Results of the Logit Analysis for Hypothesis 2c predicting significant somatic symptoms, with PCS removed B S.E. Wald Sig. Exp(B) 95% C.I. for EXP(B) Lower Upper Constant 3.997 .805 24.660 .000 54.448 Race (Caucasian) 3.609 .165 African American -.531 .284 3.492 .062 .588 .337 1.026 Other -.269 .419 .412 .521 .764 .336 1.738 Ethnicity .282 .307 .847 .357 1.326 .727 2.420 Sex .340 .242 1.980 .159 1.405 .875 2.256 Number of Physical Comorbidities .455 .211 4.643 .031 1.576 1.042 2.385 Anxiety Diagnosis -.184 1.108 .027 .868 .832 .095 7.307 Schizophrenia Diag 17.900 23095.521 .000 .999 5.940E7 .000 Bipolar Diagnosis 24.451 40192.970 .000 1.000 .000 .000 Education 3.365 .644 Some High School -.626 .472 1.759 .185 .535 .212 1.349 High School Grad -.769 .447 2.960 .085 .463 .193 1.113 Some College -.613 .483 1.609 .205 .542 .210 1.397 4 Year College -.300 .647 .215 .643 .741 .209 2.631 Post Graduate -.158 1.135 .019 .890 .854 .092 7.906 Number of PCP Visits .000 .001 .077 .782 1.000 .998 1.003 Psychiatrist visit -.022 .365 .004 .952 .978 .479 2.000 SF12 MCS -.065 .010 38.861 .000 .937 .918 .956 Age Categories (18-35) 6.152 .046 Age 36-64 .607 .284 4.559 .033 1.835 1.051 3.202 Age 65+ .798 .391 4.170 .041 2.222 1.033 4.782

PAGE 79

79 Table 5-18 Results of the Mediation Analysis for Hypothesis 2d This table shows the change in the odds ratio for the race variable when the somatic symptoms variable is entered into the model predicting diagnosis. Table 5-19 Results of the Mediation Analysis for Hypothesis 2e This table shows the change in the odds ratio for the race variable when the somatic symptoms variable is entered into the model predicting treatment. Table 5-20 Results of the Mediation Analysis for Hypothesis 2d with PCS removed This table shows the change in the odds ratio for the race variable when the somatic symptoms variable is entered into the model predicting diagnosis. Table 5-21 Results of the Mediation Analysis for Hypothesis 2e with PCS removed This table shows the change in the odds ratio for the race variable when the somatic symptoms variable is entered into the model predicting treatment. Race (AA) and Diagnosis Exp(B) Sig. % Change in OR Sig of Change from Previous Model Unadjusted .973 .959 With Somatic Symptoms .892 .833 8.4% .480 Race (AA) and Treatment Exp(B) Sig. % Change in OR Sig of Change from Previous Model Unadjusted .239 .000 With Somatic Symptoms .240 .000 0.8% .899 Race (AA) and Diagnosis Exp(B) Sig. % Change in OR Sig of Change from Previous Model Unadjusted 1.851 .292 With Somatic Symptoms 1.670 .384 9.8% .731 Race (AA) and Treatment Exp(B) Sig. % Change in OR Sig of Change from Previous Model Unadjusted .409 .017 With Somatic Symptoms .416 .019 1.7% .902

PAGE 80

80 CHAPTER 6 DISCUSSION The Surgeon Generals report in 2001 highlighted the several areas of shortcoming in providing mental health services to minorities within the United States. The report discussed such concerns as lack of access to mental health services, and even when access is gained, minorities have less chance of receiving needed mental health services (DHHS, 2001). This study supports the existence of such disparities, however, provides only partial evidence into one of the hypothesized reasons for this. The study also supports the previously published evidence that the Medicaid population as a whole is vulnerable to under-diagnosis and under-treatment, as evidenced by the numbers of undiagnosed and untreated Medicaid enrollees in this study. Depression Diagnosis and Treatment in Medicaid Primary Care This study supports that depression in Medicaid primary care is markedly underdiagnosed. While one third of the full survey sample met PHQ-2 criteria for a depressive disorder at the time of the telephone survey, only 6.9% of the full survey sample was diagnosed as depressed by their physician during the period of the study. This supports research that shows that detection in the primary care settings is suboptimal (Valenstein, Vijan, Zeber, Boehm, and Buttar, 2001) and research that finds lower rates of depression identification in Medicaid (Melfi, Croghan, and Hanna, 1999). Among the 706 PHQ-2 identified depression group, their physician diagnosis rates were also low, 11.1% obtaining a diagnosis during the study period, despite showing possible signs of depression at the time of the survey. These results indicate that as much as 90% of depressed Medicaid enrollees may be undiagnosed. Given what we know about the personal and societal burden of depression, this is a major public health

PAGE 81

81 concern. However, what this study does not address is the number of enrollees who are being successfully treated, and therefore thei r symptoms are well managed. This would mean that they would not be identified using the PHQ-2 as a selection tool, and would not be included in the final sample, therefore reducing the observed rates of diagnosis and treatment among the depressed. While diagnosis rates are low among both the full survey sample and the PHQ-2 identified group, treatment rates are higher. Previous research has shown that of those who are recognized as depressed by primar y care physicians, treatment rates can be as low as 27% (Tylee, 2006). In this study’s full sample, among those who were diagnosed with depression by their physician (N = 145), 82% received at least one prescription for antidepressants. A similar pattern is seen among the PHQ-2 identified depression group, while 11.1% received a diagnosis, 37.1% received treatment during the same time period. Of the entire survey sample approximately one quarter received some form of treatment for depression during the course of the study. An interesting aspect of depression management in this setting are the much higher rates of treatment than of diagnosis, indicating that treatment frequently occurs in the absence of a depression diagnosis. This study is consistent with studies have shown that Medicaid enrollees are also particularly vulnerable to under-treatment when compared with other populations (Harman, 2004; Melfi, 1999). However, due to lack of a non-Medicaid comparison sample in this study it is no t possible make this claim.

PAGE 82

82 Racial Disparities in Diagnosis and Treatment Racial Disparities in Diagnosis This study did not provide support for racial disparities in the diagnosis of depression. After controlling for a variety of covariates known to influence diagnosis, race showed no relationship with the likelihood of being diagnosed with depression by a physician in primary care. However, this result may be due to the very low numbers of depressed African Americans who actually received a depression diagnosis in our sample (N=9). Due to this lack of power the analysis results may be unreliable In addition there is also some debate as to the appropriateness of using a tool such as the PHQ-2 among different cultural groups due to it’s focus on more psychological/emotional questions about depression, which in some studies have shown a lack sensitivity to different expressions of depression across cultures (Kerr & Kerr, 2001). Racial Disparities in Treatment This study provides clear support for the existence of racial disparities in the treatment of depression in Medicaid primary care. In the PHQ-2 identified depressed sample, after controlling for a variety of covariates known to influence treatment, race was a significant factor in whether an individual obtained treatment or not. The data indicated that Caucasians had almost 4 times the odds of African Americans of receiving antidepressants and of receiving any type of treatment for depression. The lack of effect of mental health visits may again be the result of low numbers as only four African Americans from our depressed sample received any type of mental health visit from their primary care physician, while not significant, the odds ratio indicated that Caucasians have approximately 5 times the o dds of African Americans of receiving a

PAGE 83

83 mental health visit. This supports research that consistently shows that African Americans less frequently obtain treatment for depression (Melfi, Croghan & Hanna, 1999, Harman et al. 2001, Harman, Fortney, & Edlund 2004, Stockdale, Lagomasino, Siddique, McGuire, & Miranda, 2008). The fact that the data shows a disparity in treatment but not in initial diagnosis (perhaps unreliably so due to power issues) may indicate that it is not the initial identification of need that leads to such disparities, but the initiation of treatment. This is consistent with research that shows stigma may play a greater role among the treatment of African Americans, than the tr eatment of Caucasians. Research has shown that treatment with antidepressants and individual counseling is less acceptable to African Americans than Cauc asians (Gonzalez, Croghan & West, 2008; Givens, Katz, Bellamy, & Holmes, 2007). This difference also existed within a primary care setting (Cooper, Gonzales & Gallo et al, 2003). African Americans also associate greater stigma with depression treatment than Cauc asians in primary care settings (Menke & Flynn, 2009). The stigma associated with depression treatment may lead to treatment being less frequently offered to the patient, or to increased rates of refusal of treatment by the patient. However, this pattern of results may still be consistent with the role of somatic symptoms in diagnosis and treatment. While the hypothesized relationship between race and diagnosis is not supported here, increased somatic symptoms could still impact treatment initiation. Racial Disparities in Mental Health Expenditures In line with lower treatment rates for African Americans, this study also found that African Americans were less likely to have any mental health expenditures, and that when they did, their expenditures were appr oximately half that of Caucasians, although

PAGE 84

84 this analysis did not quite reach significance. Based on our previous analyses, this disparity in expenditures does not stem from a lack of recognition of depression, but from lower rates of treatment once depression is identified. While saving money for Medicaid may be viewed as a good thing, the resulting costs of untreated depression may be far greater. However, it does raise the question about the ability of the system to support treatment for the large num bers of unidentified and untreated depressed individuals should they become identified, and obtain treatment. The Role of Somatic Symptoms in Racial Disparities Despite the existence of racial disparities in treatment, this study showed limited support for somatic symptoms masking depre ssion diagnosis. After controlling for all other covariates, the presence of somatic symptoms of depression was not a significant predictor in either of the models predicting depression or treatment, although there did appear to be a trend in one analysis suggesting that with those without somatic symptoms had approximately twice the odds of diagnosis than those with somatic symptoms, but this did not reach significance. When the covariate PCS score was removed from the models it did not lead to different findings. This analysis was also underpowered due to the low numbers who are diagnoses in this sample. The lack of a significant relationship between somatic symptoms and diagnosis could possibly be a result of the overall ill health of this sample and of the Medicaid population in general. The Medicaid population has generally higher rates of many disabling physical conditions and poorer physical functioning than the general population. This is supported in this study by the average score of 36 on the Physical Component Summary of the SF-12 (PCS), which is lower than the national norm for 75 years old, despite having an average age of 48. This dramatic depiction of this

PAGE 85

85 population indicates that this sample reflects a group with generally very poor physical wellbeing. In addition 83% of the population endorsed a significant level of somatic symptoms associated with depression. At such high levels of physical symptoms it is possible that the physical health problems are so great that they may mask any variation in the expression somatic symptoms of depression and prevent addressing of anything other than the wealth of physical problems within patient consultations in primary care. This is especially true given the average time spent in consultation with patients in non-academic primary care is only about 10-13 minutes (Tai-Seale, McGuire & Zhang, 2007). This picture is contrary to the picture painted by the ICD-9 diagnosis codes in the claims data, with only 20% of the sample having at least one physical comorbidity. This interesting contrast highlights some of the likely problems with Medicaid claims datasets. This problem was also seen with depression diagnosis. The lack of a diagnosis code in the diagnosis fields may not mean a lack of recognition of a disorder, but just a lack of recording each diagnosis the person ca rries, as evidenced by the stark contrast of the PCS score and the comorbidity diagnoses and by patients being treated for depression in the absence of a diagnosis. The initial analysis between somatic symptoms and treatment proved nonsignificant, however, when PCS was removed from the model the results moved towards significance (p .157), but the odds ratio indicated that the effect was in the reverse direction than hypothesized, wit h those who endorsed somatic symptoms having about 1.5 the odds of receiving treatment than those who did not. This is also particularly interesting in relations to the effects of race and somatic symptoms.

PAGE 86

86 The relationship between somatic symptoms and race initially showed no significant effect, however, once PCS was removed from the model, the results changed to becoming very near significance (p 062). However, this was in the opposite direction of the hypothesized effect, indicating that Caucasians have approximately 1.7 the odds of African Americans of endorsing somatic symptoms of depression. This reversal of effect could be a result of using an unvalidated measure of somatic symptoms in the study. It is possible that given that the four somatic questions appear as part of nine questions about depression, anhedonia, suicide, and feeling bad about oneself, that this may cause a resp onse bias, influenced by the stigma among African Americans towards depression, leading African American to endorse these items less frequently. This pattern of results is interesting, while somatic symptoms may mask diagnosis, once someone is identified as depressed, somatic symptoms perhaps suggest greater severity and a greater need for treatment. When this is coupled with Caucasians increased rates of somatic symptom reporting and their greater likelihood of obtaining treatment, the presence of somatic symptoms may actually act as a catalyst for the initiation of treatment. The final analysis modeled the mediating effect of somatic symptoms between race and diagnosis and race and treatment also showed a lack of significant effect, where the addition of somatic symptoms to the model did not change the relationship between race and diagnosis and race and treatment. The removal of PCS from the models predicting diagnosis and treatment did not significantly change the value of the odds ratio of Race, indicating that while there may be some significant findings between

PAGE 87

87 race, somatic symptoms and depression di agnosis and treatment, somatic symptoms does not play a mediating role in the relationship between race and diagnosis and race and treatment. Taken together, these analyses and their patterns of results are not easy to interpret as a whole, partially due to the lack of power in several analyses, and perhaps partly due to the use of an unvaildated tool to measure somatic symptoms. However there seems to be some evidence for the masking effects of somatic symptoms in diagnosis, and a facilitating effect of somatic symptoms in initiating treatment once it is recognized. Implications and Recommendations This study is the first to systematica lly evaluated the wi dely accepted somatic symptoms of depression and their role racial disparities. This study provides further evidence that African Americans continue to experience racial disparities in the treatment of depression in Medicaid primary care. The combination of high rates of depression in this population, 30% in this sample, with the low rate of diagnosis and treatment results in a very large and costly group of individuals living with undiagnosed and untreated depression. What this study suggests is that more needs to be done to improve the identification and treatment of depression in primary care, in low-income groups, and among African Americans. Patient education programs should be expanded to cover the symptoms of depression, especially in clinics that serve low-income communities, such as health departments or clinics that serve Medicaid recipients. This would aid patients in being able to recognize their own symptoms of depression, and would also go someway to challenge the stigma that many still associate with mental illness. To tackle disparities directly there should be culturally targeted patient education efforts that should highlight

PAGE 88

88 the role of physical symptoms of depressi on in addition to emoti onal and psychological symptoms. When evaluating patients for depression, physicians should be sure to look for somatic and neurovegetative symptoms rather than just mood or cognitive symptoms of depression, regardless of race as there is evidence that these may mask depression. Automatic screening of patients for depression could be included in the initial screenings performed by nurses. So while patients are being weighed, and having their blood pressure checked the nurse could administer one of the brief depression screeners, such as the PHQ-2 and then notate the chart as with any other health measure. Ensuring that primary care practices have best practice guidelines to guide the treatment either within the primary care setting or to know when to refer for specialized treatment. This ensures that the physician has the knowledge to be able to follow through when patients are identified. All these methods have been used in the past to try to improve the management of depression in primary care, and their inclusion into a coordinated approach to depression management is vital. Research suggests that models of integrated care within primary care clinics are most successful in identifying and treating those with depression (U.S. Preventive Services Task Force, 2002). Integrative care models on the whole incorporate mental and behavioral health professions within the primary care clinic. Levels of collaborations vary within such models, however the most effective have shown to be those that are highly collaborative and structured in their delivery of care (U.S. Preventive Services Task Force, 2002). These models of care also improve

PAGE 89

89 the likelihood that the patient will obtain an adequate course of treatment and result in better overall health outcomes. However currently, few studies have looked at whether this model of integrative care is effective in reducing racial disparities. Limitations This study has several limitations. The first is common among studies that use administrative datasets. These datasets can be incomplete and may not reflect what is actually occurring in the course of health service. For example the relatively low rate of physician diagnosis among this sample compared to the much higher rate identified by the survey may not be the result of a lack of recognition, it may just be the lack of documentation. Using a billing diagnosis as a means to track diagnosis therefore leads to an underestimation of the numbers that are identified by the physicians. Frequently physicians may just enter one diagnostic code for the pur poses of billing, despite providing treatment for several conditions. This would be consistent with this study’s finding that far more recipients received treatment than received a diagnosis. Records of treatment with antidepressants within this dataset may be more reliable as these claims come from claims paid to the pharmacy who must document all medications provided to enable them to be reimbursed, whereas physicians are not required to include every diagnosis to be reimbursed. Another limitation linked to what is actually occurring in the primary care consultation is that endorsing items on the PHQ-2 and the somatic items on the PHQ-9 over the telephone during a survey, does not mean that the enrollees reported these symptoms while consulting with their primary care physician. This means that while this study relies on the survey endorsed symptoms being reported to the physician, which is then assumed to influence the physicians to diagnose or treat, there is no way to know

PAGE 90

90 what type of information is shared with the physician. In order to be sure, future studies must somehow gain access to this information, for example, by record review or by some other study design that accesses consultation specific information. While the PHQ-2 is a valuable tool for use in primary care to improve efforts to identify depression, its use by this study may have resulted in an over-inclusive sample, or the inclusion of a number false positives. This means that the study sample may include individuals who do not have a depressive disorder, and also result in an overestimate the prevalence rate among the Medicaid primary care population, and an overestimate of under diagnosis and under treatment. Another common limitation among studies concerned with specific populations is a lack of generalizable. Given the very specific sample of individuals, Medicaid enrollees who are were seen in primary care, is unlikely that these findings could be taken as representative of other populations, for example inpatients or privately insured individuals, or as representative of the population as a whole, due to the unique characteristics of Medicaid recipients and of service in primary care settings. Another limitation of this study is the lack of power in several of the analyses. While the sample was a good size, approximately 700 Medicaid enrollees, the low rates of physician diagnosis among African Americans and the low rates of mental health visits among African Americans weakened the power of several analyses. When studies examine an event that is very rare the sample size must be large enough in order to obtain sufficient power, however this was not possible in this study given that the dataset was part of a completed larger study.

PAGE 91

91 Finally, as mentioned previously, one of the studies primary variables, somatic symptoms of depression as obtained from the PHQ-9, has never been validated for use in this way, and has never been used in this way in other studies. It is possible that this is not the best means to assess for the somatic symptoms of depression, and that other important aspects of depression were not included, for example gastric problems, and pain. Despite these limitations this study tells a powerful story about the quality of life of this population, and about the existence of racial disparities in this already vulnerable population. Future research This study suggests that several areas of research need to be developed. While this study attempted to use large-scale health services research techniques to examine the problem of racial disparities, and despite using a large administrative data set, the study still encountered problems of lack of power and low numbers. While the odds ratios in several cases were very large, indicating a strong effect, the lack of numbers could not provide the study with statistical significance. Future research must include even larger sample sizes to enable enough power to support these hypotheses. This study also encountered problems with the validity of the Somatic Symptoms measure. Future studies should include the DSM-IV criteria used by the PHQ-9, but should also include wider measures of somatic symptoms such as headache, migraines, sexual dysfunction, menstrual-related symptoms, chronic pain, digestive problems (e.g., diarrhea, constipation), and fatigue (Kerr & Kerr, 2001). There is a also need for a systematic va lidation of existing depression measures across racial and ethnic groups and further exploration and development of tools

PAGE 92

92 specifically designed to assess depression among minorities, that may include broader concepts such as worry, feeling pressur ed, feeling empty, or feeling cut-off. More generally, there is clearly a need to examine collaborative models of care in primary care, and their effectiveness in r educing racial disparities by improving diagnosis and treatment rates across a wider array of populations and settings, including settings that provide care to low-income populations. Such models of service are also valuable when serving remote communities and show promise in dealing with the myriad of behavioral health issues such as heart disease, diabetes, and high blood pressure, which are frequently comorbid with depression.

PAGE 93

93 APPENDIX BMS DEPRESSION SURVEY Florida Center for Medicaid & the Uninsured Survey to be conducted with adult enrollees in the Florida Medicaid program in the Spring of 2004. Interviewer notes/administration instructions in italics. Instrument names & notes for researcher’s use only in Field names in ALL CAPS to the left of each item. For all items: -8 = Don’t Know, -9 = Refused. Unless noted, the skip sequences for these options are the same as for “No” or “Disagree.” Programmer note: please use the field names and response category values indicated on this hard copy! HELLO Hello. My name is _______, and I am calling from the Survey Research Center at the University of Florida. ADULTA May I speak to (target name) ? < Interviewer – Be certain that you get the right person on the phone. Confirm this!> If yes, reintroduce yourself if necessary If no Is there another time I could call back to talk to him/her? Schedule a call back if necessary and thank the respondent for his/her time. INTRO I am calling you today because I’m doing a scientific research study on the health of people in Florida’s Medicaid program. If you’re willing to complete my survey, I will send you a $5 Wal-Mart gift certificate in the mail. My survey usually takes about 10 minutes and it has a lot of questions about whether you have had different physical and emotional problems recently. ADVISE I just need to tell you a couple of things before we get started. You do not have to answer any question you don’t want to. You don’t have to participate in the survey at all. No one not even Medicaid will know if you participated or not, and your name will not be reported to anyone else. The results from our study will be reported back to Florida Medicaid and may be published in a scientific journal. Ok, let’s get started. 1. Yes (go to Survey ) 2. No (go to SORRY) SORRY Ok. Thank you very much for your time. (E nd of Interview)

PAGE 94

94 BMS1 My first group of 9 questions is about some problems you may or may not have had over that past 2 weeks. For each question, I’m going to ask you how often you’ve had that problem. Over the last 2 weeks, how often have you been bothered by: having little interest or pleasure in doing things. Have you been bothered: not at all, several days in past 2 weeks, more than half the days in the past 2 weeks, or nearly every day in the past 2 weeks? 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS2 feeling down, depressed, or hopeless? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS3 trouble falling or staying asleep, or sleeping too much? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS4. feeling tired or having little energy? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS5. poor appetite or overeating (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.)

PAGE 95

95 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS6. feeling bad about yourself – or that you are a failure or have let yourself or your family down? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS7. Trouble concentrating on things, such as reading the newspaper or watching television? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS8. Moving or speaking slowly that other people could have noticed? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused BMS9. Thoughts that you would be better off dead or of hurting yourself in some way? (Interviewer: Remind as necessary that the time period is the last 2 weeks. Read response categories as necessary.) 1. Not at all 2. Several days 3. More than half the days 4. Nearly every day -8. Don’t know -9. Refused

PAGE 96

96 BMS10 Programmer: This question should be asked of everyone unless they answered 1, -8, or –9 to all nine questions above. You indicated that you had a problem with . How difficult problems made it for you to do your work, take care of things at home, or get along with other people? 1. Not difficult at all 2. Somewhat difficult 3. Very difficult 4. Extremely difficult -8. Don’t Know -9. Refused NOTE: There was a mistake in the CATI program that I did not catch when I proofed the program. One of the SF-36 questions was left out. Luckily, however, the question that was left out was not one of the SF-12 items. So we have to roll up to the SF-12 level. BMS11 Now I’m going to ask you some questions about your general health. This information will help keep track of how you feel and how well you are able to do your usual activities. In general, would you say your health is excellent, very good, good, fair, or poor? 1. Excellent 2. Very good 3. Good 4. Fair 5. Poor -8/-9 Don’t Know/Refused BMS12 Compared to one year ago how would you rate your health in general now ? Is it much better now that one year ago, somewhat better now than one year ago, about the same as one year ago, somewhat worse now than one year ago, or much worse now than one year ago? 1. Much better 2. Somewhat better 3. About the same 4. Somewhat worse 5. Much worse BMS13 The following questions are about activities you might do during a typical day. I’m going to ask you if your health limits you in these activities, and if so, how much? Does your health limit you in vigorous activities, such as running, lifting heavy objects, or participating in a strenuous sport?

PAGE 97

97 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS14 Does your health limit you in moderate activities such as moving a table, pushing a vacuum cleaner, bowling, or playing golf? 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS15 Does your health limit you in Lifting or carrying groceries? 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS16 Does your health limit you in Climbing several flights of stairs? 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS17 Does your health limit you in Bending, kneeling or stooping 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS18 Does your health limit you in Walking more than a mile 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS19 Walking several hundred yards 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS20 Walking one hundred yards 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS21 Bathing or dressing yourself

PAGE 98

98 1. Yes, limited a lot 2. Yes, limited a little 3. No, not limited at all BMS22 Now I’m going to ask you about some problems you may have had with your work or daily activities as a result of your physical health. In the last 4 weeks, how much of the time have you cut down the amount of time you spent on work or other activities as a result of your physical health? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS23 In the last 4 weeks, how much of the time have you accomplished less than you would like as a result of your physical health? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS24 In the last 4 weeks, how much of the time were you limited in the kind of work or other activities you did as a result of your physical health? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS25 In the last 4 weeks, how much of the time have you had difficulty performing work or other activities (for example it took extra effort) as a result of your physical health 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS26 Now I’m going to ask you some questions problems you may have had with your work or daily activities as a result of any emotional problems such as feeling depressed or anxious. In the last 4 weeks, how much of the time have you cut down the amount of time you spent on work or other activities as a result of any emotional problems?

PAGE 99

99 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS27 In the last 4 weeks, how much of the time have you accomplished less than you would like as a result of any emotional problems? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS28 In the last 4 weeks, how much of the time did you do work or other activities less carefully than usual as a result of any emotional problems? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS29 During the past 4 weeks to what extent has your physical health or emotional problems interfered with your normal social activities with family, friends, neighbors, or groups? 1. Not at all 2. Slightly 3. Moderately 4. Quite a bit 5. Extremely BMS30 How much bodily pain have you had during the past 4 weeks ? 1. None 2. Very mild 3. Mild 4. Moderate 5. Severe 6. Very Severe BMS31 During the past 4 weeks how much did pain interfere with your normal working (including both work outside the home and housework): 1. Not at all 2. Slightly 3. Moderately 4. Quite a bit 5. Extremely

PAGE 100

100 BMS32 These questions are about how you feel and how things have been with you during the past 4 weeks For each question, please give the one answer that comes closest to the way you have been feeling. How much of the time during the past 4 weeks did you feel full of life? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS33 How much of the time during the past 4 weeks have you been very nervous? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS34 How much of the time during the past 4 weeks have you felt so down in the dumps that nothing could cheer your up? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS35 How much of the time during the past 4 weeks Have you felt calm and peaceful ? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS36 How much of the time during the past 4 weeks Did you have a lot of energy? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS37 How much of the time during the past 4 weeks Have you felt downhearted and depressed?

PAGE 101

101 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS38 How much of the time during the past 4 weeks Did you feel worn out? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS39 How much of the time during the past 4 weeks Have you been happy? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS40 How much of the time during the past 4 weeks Did you feel tired? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS41 During the past 4 weeks how much of the time has your physical health or emotional problems interfered with your social activities (like visiting friends, relatives, etc.)? 1. All of the time 2. Most of the time 3. Some of the time 4. A little of the time 5. None of the time BMS42 Now, I am going to read some statements to you. I would like to know how TRUE or FALSE each statement is for you. Your choices will be definitely true, mostly true, don’t know, mostly false, and definitely false. How true or false is this statement: I seem to get sick a little easier than other people 1. Definitely true 2. Mostly true 3. don’t know 4. mostly false

PAGE 102

102 5. definitely false BMS43. How true or false is this statement: I am as healthy as anybody I know 1. Definitely true 2. Mostly true 3. don’t know 4. mostly false 5. definitely false BMS44 How true or false is this statement: I expect my health to get worse 1. definitely true 2. mostly true 3. don’t know 4. mostly false 5. definitely false BMS45 How true or false is this statement: my health is excellent 1. definitely true 2. mostly true 3. don’t know 4. mostly false 5. definitely false CIDIINTR Now I’m going to ask you some questions about your emotional health.
CIDIA1 During The Past 12 months, was there ever a time when you felt sad, blue, or depressed for two weeks or more in a row? 1. Yes 5. No (Go To CIDI-A9) 6. (If Volunteered) I was on Medication/Anti-Depressants CIDI-A1A For the next few questions, please think of the two-week period during the past 12 months when these feelings were worst. During that time did the feelings of being sad, blue, or depressed usually last all day long, most of the day, about half the day, or less than half the day? 1. All Day Long 2. Most 3. About Half (Go To CIDI-A9) 4. Less Than Half (Go To CIDI-A9) CIDI-A1B During those two weeks, did you feel this way every day, almost every day, or l ess often?

PAGE 103

103 1. Every Day 2. Almost Every Day 3. Less Often (Go To CIDI-A9) CIDI-A1C During those two weeks did you lose interest in most things like hobbies, work, or activities that usually give you pleasure? 1. Yes 5. No CIDI-A1D Thinking about those same two weeks, did you feel more tired out or low on energy than is usual for you? 1. Yes 5. No CIDI-A2 Did you gain or lose weight without trying, or did you stay about the same? Interviewer: If R asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Gained 2. Lost 3. (If Volunteered) Both gained and lost weight 4. Stay about the same (Go to CIDI-A3) 5. (If Volunteered) R was on a diet (Go to CIDI-A3) CIDI-A2A About How Much Did (You Gain/You Lose/Your Weight Change)? __________ POUNDS Interviewer: Accept a range response CIDI-A2B Interviewer: Did R’s weight change by 10 pounds or more? 1. Yes 5. No CIDI-A3 Did you have more trouble falling asleep than you usually do during those two weeks? 1. Yes 5. No (Go To CIDI-A4) CIDI-A3A Did that happen every night, nearly every night, or less often during those two weeks? 1. Every night 2. Nearly every night 3. Less often

PAGE 104

104 CIDI-A4 During those two weeks, did you have a lot more trouble concentrating than usual? Interviewer: If R Asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Yes 5. No CIDI-A5 People sometimes feel down on themselves, no good, or worthless. during that two week period, did you feel this way? Interviewer: If R Asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Yes 5. No CIDI-A6 Did you think a lot about death — either your own, someone else’s, or death in general during those two weeks? Interviewer: If R Asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Yes 5. No CIDI-A7 Interviewer Checkpoint — (Count Qualifying Responses In CIDI-A1c Throughcidi-A6. Qualifying Responses Are As Follows: CIDI-A1D=1,CIDIA1D=1, CIDI-A2B=1, CIDI-A3A=1 Or 2, CIDI-A4=1, CIDI-A5=1, And CIDI-A6=1 ) Programmer – Please construct an algorithm to do this so that the interviewer doesn’t have to stop and do it. 1. Zero Qualifying Responses (Go To CIDI-B1) 2. One Or More Qualifying Responses CIDI-A8 To review, you had two weeks in a row during the past 12 months when you were sad, blue, or depressed and also had some other feelings or problems like ( Read Up To The First Three "Yes" Responses To CIDI-A1C through CIDI-A6). About how many weeks altogether did you feel this way during the past 12 months? __________# Of Wks 52 ( If Volunteered ) Entire year (Go To CIDI-A8B) CIDI-A8a Think about this most recent time when you had two weeks in a row when you felt this way. How

PAGE 105

105 long ago was that? ____________________Months in the past CIDI-A8B Did you tell a doctor about these problems? (By "Doctor" I mean either a medical doctor or osteopath, or a student in training to be either a medical doctor or osteopath.) 1. Yes 5. No CIDI-A8c Did you tell any other professional (such as a psychologist, social worker, counselor, nurse, clergy, or other helping professional)? 1. Yes 5. No CIDI-A8D Did you take medication or use drugs or alcohol more than once for these problems? 1. Yes 5. No CIDI-A8E How much did these problems interfere with your life or activities — a lot, some, a little, or not at all 1. A Lot (Go To CIDI-B1) 2. Some (Go To CIDI-B1) 3. A Little (Go To CIDI-B1) 4. Not At All (Go To CIDI-B1) CIDI-A9 During the past 12 months, was there ever a time lasting two weeks or more when you lost interest in most things like hobbies, work, or activities that usually give you pleasure? 1. Yes 5. No (Go To CIDI-B1) 6. (If Volunteered) I Was On Medication/Anti-Depressants CIDI-A9A For the next few questions, please think of the two-week period during the past 12 months when you had the most complete loss of interest in things. During that two-week period, did the loss of interest usually last all day long, most of the day, about half the day, or less than half the day? 1. All Day Long 2. Most 3. About Half (Go to CIDI-B1)

PAGE 106

106 4. Less Than Half (Go To CIDI-B1) CIDI-A9B Did you feel this way every day, almost every day, or less often during the two weeks? 1. Every Day 2. Almost Every Day 3. Less Often (Go to CIDI-B1) CIDI-A9C During those two weeks, did you feel more tired out or low on energy than is usual for you? 1. Yes 5. No CIDI-A10 Did you gain or lose weight without trying, or did you stay about the same? Interviewer: If R asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Gain 2. Lose 3. (If Volunteered) Both Gained and Lost Weight 4. Stay About the Same (Go to CIDI-A11) 5. (If Volunteered) R Was on a diet (Go to CIDI-A11) CIDI-A10A About how much did (you gain/you lose/your weight change)? ______________Pounds Interviewer: Accept a range response. CIDI-A10B Interviewer: Did R’s weight change by 10 pounds or more? Programmer – Please construct an algorithm to do this so that the interviewer doesn’t have to stop and do it. 1. Yes 5. No CIDI-A11 Did you have more trouble falling asleep than you usually do during those two weeks? 1. Yes 5. No (Go To CIDI-A12) CIDI-A11A Did that happen every night, nearly every night, or less often during those two weeks? 1. Every night 2. Nearly every night 3. Less often

PAGE 107

107 CIDI-A12 During those two weeks, did you have a lot more trouble concentrating than usual? Interviewer: If R Asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Yes 5. No CIDI-A13 People sometimes feel down on themselves, no good, or worthless. did you feel this way during that two week period? Interviewer: If R Asks: "Are we st ill talking about the same two weeks?" Answer: "Yes." 1. Yes 5. No CIDI-A14 Did you think a lot about death — either your own, someone else’s, or death in general during those two weeks? Interviewer: If R Asks: "Are we st ill talking about the same two weeks?" Answer: “Yes." 1. Yes 5. No CIDI-A15 Interviewer Checkpoint — (Count "Yes" Responses In CIDI-A9C through CIDIA14) 1. Zero “Yes” Responses On CIDI-A9C, CIDI-A12, CIDI-A13, CIDI-A14, and (EitherCIDI-A10=4-5 or CIDI-A10A is less than 10 pounds) and (Either CIDI-A11=5 or CIDI-A11A=3) Go To CIDI-B1 2. All Others (Go to CIDI-A16) Programmer – Please construct an algorithm to do this so that the interviewer doesn’t have to stop and do it. CIDI-A16 To review, you had two weeks in a row during the past 12 months when you lost I nterest in most things and also had some other things like (Read up to the first 3 "Yes" responses to CIDI-A9c through CIDI-A14). About how many weeks did you feel this way during the past 12 months? ______________# Of Wks 52. (If Volunteered) Entire Year (Go To CIDI-A16B)

PAGE 108

108 CIDI-A16A Think about this most recent time when you had two weeks in a row when you felt this way. How long ago was that? ____________ Months In The Past CIDI-A16B Did you tell a doctor about these problems? (By "Doctor" I mean either a medical doctor or osteopath, or a student in training to be either a medical doctor or osteopath.) 1. Yes 5. No CIDI-A16C Did you tell any other professional (such as a psychologist, social worker, counselor, nurse, clergy, or other helping professional)? 1. Yes 5. No CIDI-A16D Did you take medication or use drugs or alcohol more than once for these problems? 1. Yes 5. No CIDI-A16E How much did these problems interfere with your life or activities — a lot, some, a little, or not at all? 1. A Lot 2. Some 3. A Little 4. Not At All
CIDI-B1 During the past 12 months, did you ever have a period lasting one month or longer when most of the time you felt worried, tense, or anxious? 1. Yes (Go to CIDI-B2) 5. No CIDI-B1A People differ a lot in how much they worry about things. did you have a time in the past 12 months when you worried a lot more than most people would in your situation? 1. Yes 5. No (Go To CIDI-E1)

PAGE 109

109 CIDI-B2. Has that period ended or is it still going on? 1. Ended 2. Still going on (G o To CIDI-B2B) CIDI-B2A How many months or years did it go on before it ended? __________ # of months or go to CIDI-B3 __________# of years go to CIDI-B3 89 (If Volunteered) “All My Life” or “As Long As I Can Remember” ( Go to CIDI-B3) CIDI-B2B How many months or years has it been going on? __________ # of months or __________# of years 89 (If Volunteered) “All My Life” or “As Long As I Can Remember” CIDI-B3 INTERVIEWER CHECKPOINT 1. CIDI-B2A/CIDI-B2 is six months or longer, or (If Volunteered) “All My Life” Or “As Long As I Can Remember” 2. CIDI-B2A/CIDI-B2 is less than six months (Go to CIDI-E1) Programmer – Please construct an algorithm to do this so that the interviewer doesn’t have to stop and do it. CIDI-B4 (During that period, was your/is your) worry stronger than in other people? 1. Yes 5. No CIDI-B5 (Did/Do) you worry most days? 1. Yes 5. No CIDI-B6 (Did/Do) you usually worry about one particular thing, such as your job security or the failing health of a loved one, or more than one thing? 1. One Thing 2. More than one thing CIDI-B7 (Did/Do) you find it difficult to stop worrying? 1. Yes 5. No

PAGE 110

110 CIDI-B8 (Did/Do) you ever have different worries on your mind at the same time? 1. Yes 5. No CIDI-B9 How often (was/is) your worry so strong that you (couldn’t/can’t) put it out of your mind no matter how hard you (tried/try) — often, sometimes, rarely, or never? 1. Often 2. Sometimes 3. Rarely 4. Never CIDI-B10 How often (did/do) you find it difficult to control your worry — often, sometimes, rarely, or never? 1. Often 2. Sometimes 3. Rarely 4. Never CIDI-B11 What sort of things (did/do) you mainly worry about? (Probe: Any other main worries?) ___________________________________________________________ CIDI-B12 When you (were/are) worried or anxious, (were/are) you also... 1. Yes 5. No CIDIB12A Restless? 1. Yes 5. No CIDIB12 (Were/Are) you keyed up or on edge? 1. Yes 5. No CIDIB12c. (Were/Are) you easily tired? 1. Yes 5. No CIDIB12d. (Did/Do) you have difficulty keeping your mind on what you (were/are) doing? 1. Yes 5. No

PAGE 111

111 CIDIB12e. (Were/Are) you more irritable than usual? 1. Yes 5. No CIDIB12f. (Did/Do) you have tense, sore, or aching muscles? 1. Yes 5. No CIDIB12g. (Did/Do) You Often Have Trouble Falling Or Staying Asleep? 1. Yes 5. No CIDI-B13. CHECKPOINT 1. 0-1 Yes responses in the CIDI-B12 series (Go to CIDI-E1 ) 2. All others Programmer – Please construct an algorithm to do this so that the interviewer doesn’t have to stop and do it. CIDI-B14 Did you tell a doctor about your worry or about the problems it was causing? (by "Doctor" I mean either a medical doctor or osteopath, or a student in training to be either a medical doctor or osteopath.) 1. Yes 5. No CIDI-B15 Did you tell any other professional (such as a psychologist, social worker, counselor, nurse, clergy, or other helping professional)? 1. Yes 5. No CIDI-B16 Did you take medication or use drugs or alcohol more than once for the worry or the problems it was causing? 1. Yes 5. No CIDI-B17 How much (did/does) the worry or anxiety interfere with your life or activities — a lot, some, a little, or not at all? 1. A Lot 2. Some

PAGE 112

112 3. A Little 4. Not At All
CIDI-E1 The next questions are about things that make some people so afraid that they avoid them, even when there is no real danger. Do you have an unreasonably strong fear or avoid any of the following things... CIDI-E1A First, being in a crowd or standing in line? (Do you have an unreasonably strong fear or avoid either of these situations?) 1. Yes 5. No CIDI-E1B (How about) being away from home alone? (Do you have an unreasonably strong fear or avoid this situation?) 1. Yes 5. No CIDI-E1C (How about) traveling alone? (Do you have an unreasonably strong fear or avoid this situation?) 1. Yes 5. No CIDI-E1D (How About) traveling in a bus, train, or car? (Do you have an unreasonably strong fear or avoid any of these situations?) 1. Yes 5. No CIDI-E1E (How about) being in a public place like a department store? (Do you have an unreasonably strong fear or avoid this type of situation?) 1. Yes 5. No CIDI-E2 INTERVIEWER CHECKPOINT —See CIDI-E1A through CIDI-E1E 1. One or more "Yes" responses in CIDI-E1A through CIDI-E1E 2. All others (Go to CIDI-F1) CIDI-E3 Thinking only of the situation(s) that we just reviewed that cause(s) you unreasonably strong fears, do you get very upset every time you are in (this/these) situation(s), most of the time, only some of the

PAGE 113

113 time, or never? 1. Every Time 2. Most Of The Time 3. Some Of The Time (Go to CIDI-F1) 4. Never (Go to CIDI-F1) 7. (If Volunteered) Only one or two times ever (Go to CIDI-F1) CIDI-E4 How long have you had (this/these) fear(s) — less than 1 year, between 1 and 5 years, or more than 5 years? 1. Less Than 1 Year 2. Between 1 And 5 Years (Go to CIDI-E5) 3. More Than 5 Years (Go to CIDI-E5) CIDI-E4A About how many months? _______________Number Of Months CIDI-E5 When you are in (this/these) situation(s), are you afraid that you might faint, lose control, or embarrass yourself in other ways? 1. Yes 5. No CIDI-E6 When you are in (this/these) situation(s), do you worry that you might be trapped without any way to escape? 1. Yes 5. No CIDI-E7 When you are in (this/these) situation(s), do you worry that help might not be available if you needed it? 1. Yes 5. No CIDI-E8 During the past 12 months, how much did (this/these) fear(s) interfere with your life or activities — a lot, some, a little, or not at all? 1. A Lot 2. Some 3. A Little 4. Not At All


PAGE 114

114 CIDI-F1 During the past 12 months, did you ever have a spell or an attack when all of a sudden you felt frightened, anxious, or very uneasy? 1 Yes 5 No (Go to PAININTR) CIDI-F1A Did any of these attacks occur when you were in a life-threatening situation? 1. Yes 5. No (Go to CIDIF2) 8. (If Volunteered ) Don’t Know (Go to CIDI-F2) Programmer – For this question only, please use 8 and not –8 for the Don’t Know category. CIDI-F1B Did any of these attacks occur when you were not in a life-threatening situation? 1. Yes 5. No (Go to PAININTR) CIDI-F2 About how many attacks did you have in the past 12 months? __________Number CIDI-F3 How long ago did you have the most recent (one/attack)? _______________Months In The Past CIDI-F4 Did (this attack/all of these attacks) happen in a situation when you were not in danger or not the center of attention? 1. Yes 5. No (Go To PAININTR) CIDI-F5 A moment ago, we discussed situations that cause unreasonably strong fears. when you have attacks of the sort you just described, do they usually occur in situations that cause you unreasonably strong fear? 1. Yes 5. No (Go To CIDI-F6) CIDI-F5A Did you ever have an attack in the past 12 months when you were not in a situation that usually causes you to have unreasonably strong fears? 1. Yes 5. No (Go To PAININTR) CIDI-F6 When you have attacks, ...

PAGE 115

115 CIDI-F6A ...Does your heart pound or race? 1. Yes 5. No CIDI-F6B …Do you have tightness, pain, or discomfort in your chest or stomach? 1. Yes 5. No CIDI-F6C …Do you sweat? 1. Yes 5. No CIDI-F6D …Do you tremble or shake? 1. Yes 5. No CIDI-F6E ...Do you have hot flashes or c hills? 1. Yes 5. No CIDI-F6F ...Do you, or things around you, seem unreal? 1. Yes 5. No PAININTR Now I have some questions for you about bodily pain that you might have had. PAIN1. Have you experienced pain lasting for more than two weeks? 1. Yes 2. No = 2 (Go to DEMOINTR) PAIN2. How much did the pain bother you? (Did it bother you extremely, quite a bit, moderately, very little, or not at all?) 1. Extremely 2. Quite a bit 3. Moderately 4. Very little (Go to DEMOINTR) 5. Not at all (Go to DEMOINTR) -8/-9 Don’t Know/Refused (Go to DEMOINTR) PAIN3 Have you sought treatment for your pain?

PAGE 116

116 1. Yes (Go to PAIN5) 2. No PAIN4 I’m going to read a list of possible reasons why you did not seek treatment. Please tell me which of the following apply to you. PAIN4A Didn't think I needed treatment 1. Yes 2. No PAIN4B Received treatment for this in the past 1. Yes 2. No PAIN4C Couldn't afford treatment 1. Yes 2. No PAIN4D Didn't know where to find treatment 1. Yes 2. No PAIN4E Couldn't get an appointment when I could go 1. Yes 2. No PAIN4F I was refused treatment when I could get it 1. Yes 2. No PAIN4G I don't feel comfortable speaking English and couldn't find treatment where they spoke my language 1. Yes 2. No PAIN4H Some other reason ( specify __________) 1. Yes 2. No Go to PAIN12 PAIN5 How soon did you get medical help for this problem after it started? ____________ (Specify number of days, weeks or months, each in a separate field) PAIN6 How long did the pain last? 1. 2 To 4 Weeks 2. 1 To 6 Months 3. More Than 6 Months PAIN7 Do you know what caused your pain? 1. Yes 2. No (Go to PAIN9) PAIN8 What Was It That Caused Your Pain? ___________________________ (Specify – e.g., Headache, Backache, GI Problems, Etc.) PAIN9 Has your provider offered you any of the following medications or therapies for your pain? PAIN9A Elavil (Amitriptiline), Desipramine, Nortryptiline, or Another Tricyclic Antidepressant Drug 1. Yes 2. No

PAGE 117

117 PAIN9B Aspirin, Tylenol, Motrin, or other Aspirin-like drugs 1. Yes 2. No PAIN9C Mexiletine (Mexitil) 1. Yes 2. No PAIN9D Capsaicin (Capsin, A Lotion Made From Chili Peppers) 1. Yes 2. No PAIN9E Mild Opiate drugs Like Codeine, Vicodin, Percodin, or Percoset 1. Yes 2. No PAIN9F Strong Opiate Drugs like Morphine, Dilaudid, or Demerol 1. Yes 2. No PAIN9G Alternative therapies like acupuncture, acupressure, massage, visualization, or herbal remedies 1. Yes 2. No PAIN9H Another medication 1. Yes 2. No PAIN9I Was not offered medication 1. Yes (Go to PAIN12) 2. No PAIN10 What best describes how well the medications or therapies controlled your pain most of the time? Would you say the medications or therapies: 1. Didn't Help 2. Slightly Relieved Your Pain 3. Significantly Relieved Your Pain 4. Completely Relieved Your Pain PAIN11 How satisfied were you with your provider's efforts to control your pain? Would you say: 1. Completely Satisfied 2. Mostly Satisfied 3. Somewhat Satisfied 4. A Little Satisfied 5. Or Not At All Satisfied PAIN12 During the past 4 weeks, how many days did pain cause you to stay in bed for a half day or more? # of days:_________ DEMOINTR Ok. We’re almost finished now. I just have a few last questions. HEALTH In general, how would you rate your overall health now? 1. Excellent 2. Very Good 3. Good 4. Fair 5. Poor

PAGE 118

118 DOB What is your date of birth? Programmer – This can be entered as a single field, or as a separate field for month, day, and year. Use whichever format is easiest for SAS programming. SEX Are you male or female? 1. Male 2. Female EDUCAT What is the highest grade or level of school that you have completed ? 1. 8th grade or less 2. Some high school, but did not graduate 3. High school graduate or GED 4. Some college or 2-year degree 5. 4-year college graduate 6. More than 4-year college degree ETHNIC Are you of Hispanic or Latino origin or descent? 1. Hispanic or Latino 2. Not Hispanic or Latino RACE I am going to read you a list of race categories, and I’d like you to tell me which one or ones you think describe you best. Just so you know, the reason I’m asking you about this is because the researchers want to make sure they have gathered the opinions of enough people from all different races and ethnicities. Here is the list. Tell me yes or no for each category. Are you: RACEWH White 1. Yes 2. No RACEBL Black or African American 1. Yes 2. No RACEAS Asian 1. Yes 2. No RACEHAW Native Hawaiian or other Pacific Islander 1. Yes 2. No RACEIND American Indian or Alaska Native 1. Yes 2. No LANGUAGE What language do you mainly speak at home? 1. English 2. Spanish 3. Some other language (Specify ___________________) INCENT1 Ok. You’ve answered all my questions now. Thank you very much. Now I just need to get your name and address so I can send you your $5 WalMart card. Programmer – Please create separate fields for first name, last name, address line 1,address line 2, apartment #, city, state, and zip. INCENT2 Let me read all of that back to you so that we can be sure it’s right.

PAGE 119

119 Programmer – set up a screen that will display the name and address info again so the interviewer can confirm it. INCENT3 Ok. Your gift card will arrive within 8 weeks. Just one more thing. We have been talking about some very sensitive subjects today. Some people might feel upset after talking about these things. If you do, there is a number you can call to talk about these feelings. Would you like me to give you that number? 1. Yes 2. No (go to OUT) NUMBER The number is 1-888-784-2433. You can call 24 hours a day, 7 days a week. OUT Ok. Thank you very much for your time.

PAGE 120

120 LIST OF REFERENCES Andrews, G. & Peters, L. (1998). The psychometric properties of the Composite International Diagnostic Interview. Social Psychiatry and Psychiatric Epidemiology, 33 80-88. APA ( 1994). Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, (DSM-IV). Bennett, M. B. (1987). Afro-American women, poverty and mental health: A social essay. Women and Health, 12 213-228. Baron, R. M. & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51 1173– 1182. Borowsky, S., Rubenstein, L. V., Meredith, L. S., Camp, P., Jackson-Triche, M., & Wells, K.B. (2000). Who is at risk for non-detection of mental health problems in primary care? Journal of Global Information Management, 15 ,381–388. Brown, C., Schulberg, H. C., & Madonia, M. J. (1996). Clinical presentations of major depression by African Americans and whites in a primary medical care practice. Journal of Affective Disorders, 41 181-191. Brown, E. R., Ojeda, V. D., Wyn, R., & Levan, R. (2000). Racial and ethnic disparities in access to health insurance and health care UCLA Center for Health Policy Research and Kaiser Family Foundation, April 2000 Retrieved July 1, 2010, from www.kff.org. Centers for Medicaid & Medicare Services (2007 ). Medicaid Statistical Information System Tables FY 2004 Retrieved July 1, 2010, from http://www.cms.hhs.gov/MedicaidDataSourcesGenInfo/02_MSISData.asp. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Earlbaum Associates. Cook, B. L., McGuire, T., & Miranda. J. (2007) Measuring trends in mental health care disparities, 2000-2004. Psychiatric Services, 58 1533-1540. Cooper, L. A., Gonzales, J. J., Gallo, J.J., Rost, K. M., Meredith, L. S., Rubenstein, L. V., Wang, N. Y., & Ford, D.E. (2003). The acceptability of treatment for depression among African-American, Hispanic, and white primary care patients. Medical Care, 41, 479-448. Crystal, S., Sambamoorthi, U., & Walkup, J. T. (2003) Diagnosis and treatment of depression in the elderly Medicare populati on: predictors, disparities, and trends. Journal of the American Geriatrics Society, 51 1718-1728.

PAGE 121

121 Dunlop, D. D., Song, J., Lyons, J. S., Manheim, L. M., & Chang, R. W. (2003) Racial/ethnic differences in rates of depression among preretirement adults. American Journal of Public Health, 93 1945-52. Faul, F., Erdfelder, E., Lang, A., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39 175-191. Givens, J. L., Katz, I. R., Bellamy, S., Holmes, W. C. (2007). Stigma and the acceptability of depression treatments among african americans and whites. Journal of General Internal Medcine 22,1292-1297. Gonzalez, H. M., Croghan, T., West, B., W illiams, D., Nesse, R., Tarraf, W., Taylor, R., Hinton, L., Neighbors, H., Jackson, J. (2008) Antidepressant Use in Black and White Populations in the United States. Psychiatric Services 59, 1131-1138. Greenberg, P.E., Kessler, R.C., Birnbaum, H.G., Leong, S.A., Lowe, S.W., Berglund, P.A., & Corey-Lisle, P.K. (2003). The economic burden of depression in the United States: how did it change between 1990 and 2000 ? Journal Clinical Psychiatry, 64 1465–1475. Greenberg, P.E., Stiglin, L.E., Finkelstein, S.N., & Berndt, E.R. (1993) The economic burden of depression in 1990. Journal Clinical Psychiatry 54 405-418. Harman, J.S., Mulsant, B.H., Kelleher, K.J., Schulberg, H.C., Kupfer, D.J., & Reynolds, C.F. (2001). Narrowing the gap in treatment of depression International Journal of Psychiatry in Medicine, 31 239-253. Harman, J. S., Fortney, J., & Edlund, M. (2004) Disparities in the adequacy of depression care in the United States. Psychiatric Services, 55 ,1379-1385. Hudziak, J. J., Helzer, J. E., Wetzel, M. W., Kessel, K. B., McGee, B., Janca, A., & Przybeck, T. (1993). The use of the DSM-III-R Checklist for initial diagnostic assessments. Comprehensive Psychiatry 34, 375-83. Institute of Medicine (2003). Unequal treatment: Confronting racial and ethnic disparities in health care, ed. B.D. Smedley, A.Y. Stith, and A.R. Nelson. Washington, DC: The National Academies Press. Katon, W., & Schulberg, H. (1992). Epidemiology of depression in primary care General Hospital Psychiatry, 14 237-247. Kirmayer, l. J., Robbins, J. L., Dworkind, M., & Yaffe, M.J. (1993). Somatization and the recognition of depression and anxiety in primary care, American Journal of Psychiatry, 150, 734–741.

PAGE 122

122 Kroenke K, Spitzer RL, Williams JB.(2003). T he Patient Health Ques tionnaire-2: validity of a two-item depression screen er. Medical Care, 41, 1284-1292. McKinlay, J. B., Lin, T., Freund, K., & Moskowitz, M. (2002). The unexpected influence of physician attributes on clinical decisions: results of an experiment. Journal Health and Social Behavior, 43 92-106. Kerr, L. K., Kerr, L. D. (2001). Screening tools for depression in primary care: the effects of culture, gender, and somatic sympt oms on the detection of depression. Western Journal of Medicine, 175, 349–352. Kessler, R. C., Andrews, G., Mroczek, D., Ustun, T. B., & Wittchen, H. U. (1998). The World Health Organization composite international Diagnostic Interview Short Form (CIDI-SF ). International Journal of Methods in Psychiatric Research, 7, 171-185. Klinkman, M. S., Coyne, J. C., Gallo, S., & Schwenk, T. L. (1998). False positives, false negatives, and the validity of the diagnosis of major depression in primary care Archives of Family Medicine, 7 451-461. McGuire, T. G., Alegria, M, & Cook, B. L.( 2006). Implementing the Institute of Medicine definition of disparities: an application to mental health care Health Services Research, 41 ,1979–2005. Melfi, C. A., Croghan, T. W., & Hanna, M. P. (1999). Access to treatment for depression in a Medicaid population Journal of Health Care for the Poor and Underserved, 10 201-215. Menchetti, M., Belvederi -Murri, M., Bertakis, K., Bortol otti, B., Berardi, D. (2009). Recognition and treatment of depression in primary care: effect of patients' presentation and frequency of consultation. Journal of Psychosomatic Research. 66, 335-341. Menke, R. & Flynn, H. (2009). Relationships between stigma, depression, and treatment in white and African American primary are patients. The Journal of Nervous and Mental Disease. 197, 407-411. Murray, C. J., Lopez, A. D. (1997) Global mortality, disability, and the contribution of risk factors: Global burden of disease study Lancet. 349 ,1436-1442. Office of Management and Budget (1997). Revisions to the standards for the classification of Federal data on race and ethnicity Federal Register 62FR5878258790 (58790): Retrieved July 1, 2010, from http://www.whitehouse.gov/omb/fedreg/ombdir15.html.

PAGE 123

123 Ortho Biotech Products, L.P. (2007 ) Quick Reference to 2008 ICD-9-CM Diagnosis Codes for Common Cancers. Retrieved July 1, 2010, from www.procritline.com/pubs/icd9/icd9.pcl.onc.pdf Skaer, T.L., Sclar, D.A., Robison, L.M., & Galin, R.S. (2000). Trends in the rate of depressive illness and use of antidepressant pharmacotherapy by ethnicity/race: an assessment of office-based visits in the United States, 1992-1997 Clinical Therapeutics, 22 ,1575-1589. Somervell, P. D., Leaf, P., Weissman, M. M ., Glazer, D., & Bruce, M. L. (1989). The Prevalence of Major Depression In Black and White Adults in Five United States Communities American Journal of Epidemiology. 130 725-735. Stockdale, S. E., Lagomasino, I T., Siddique, J., McGuire, T., & Miranda, J. (2008). Racial and Ethnic Disparities in Detection and Treatment of Depression and Anxiety Among Psychiatric and Primary Health Care Visits, 1995-2005. Medical Care, 46 668-677. U.S. Department of Health and Human Services. (2001 ). Mental Health: Culture, Race, and Ethnicity—A Supplement to Mental Health: A Report of the Surgeon General Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Servic es Administration, Center for Mental Health Services. U.S. Preventive Services Task Force. (2002). Screening for depression: recommendations and rationale. Annals of Internal Medicine 21, 760-764. Tacchini, G., Coppola, M. T., Musazzi, A., Altamura, A. C., & Invernizzi, G. (1994). Multinational validation of the Composite International Diagnostic Interview (CIDI). Minerva Psichiatr 35, 63-80. Tai-Seale, M., McGuire, T. G. & Zhang W. (2007). Time allocation in primary care office. visits. Health Services Research, 42, 1871-1894. Tylee, A. (2006). Identifying and managing depression in primary care in the United Kingdom. Journal Clinical Psychiatry, 67 Suppl 6, 41-45. U.S. Bureau of the Census (1998) Current Population Reports, Series P60-202 Valenstein, M., Vijan, S., Zeber, J., Boehm, K., & Buttar, A. (2001). The cost-utility of screening for depression in primary care. Annals of Internal Medicine, 134, 345360. Wang, P.S., Walker, A., Tsuang, M., et al. (2000). Strategies for improving comorbidity measures based on Medicare and Medicaid claims data Journal Clinical Epidemiology, 53, 571–578.

PAGE 124

124 Williams, D. R., Gonzalez, H. M., Neighbor s, H. W., Nesse, R., Abelson, J. M., Sweetman, J. & Jackson, J. S. (2007). Prevalence and Distribution of Major Depressive Disorder in African Americans, Caribbean Blacks, and Non-Hispanic Whites: Results From the National Survey of American Life Archives of General Psychiatry, 64, 305-315.

PAGE 125

125 BIOGRAPHICAL SKETCH Zo is originally from Aberdeen in Scotland, and completed her Bachelor of Science in psychology at the University of Glasgow. After moving to the United States in 2001, she began working as a day-treatment counselor and clinical case manager with a non-profit mental health organization providing services to Medicaid enrollees. During this time, Zo developed a strong interest in working with underserved populations and in health policy. She spent one year as a study coordinator at the H. Lee Moffitt Cancer Center at the University of South Florida before returning to graduate school. While at the University of Florida, she sought out placements aimed at improving access to mental health services for unde rserved populations, including the uninsured, rural communities, and those living in poverty These experiences have led her to a special interest in the integration of mental health services into traditional healthcare settings, such as primary care. During her final year at the University of Florida, she was involved in the pioneering of an integrated mental health service within a student run “Equal Access Clinic” in order to increase mental health services for the uninsured and also to provide training for clinic al psychology students with this unique and important population. Zo completed her pre-doctoral Fellowship at Yale University School of Medicine and obtained her Ph.D. in Clinical and Health Psychology at t he University of Florida in 2010. Zo’s goals for the future are to continue to work to improve access to mental health services for underserved populations a nd to use these experiences to guide her research in tackling the complex mental health needs of the underserved.

PAGE 126

126