|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 IMPACT OF MENTAL HEALTH DIAGNOSIS AND TREATMENT ON ASTHMA RELATED SERVICES IN MEDICAID YOUTH WITH ASTHMA By ERIC WAYNE JAMOOM 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
2 2010 Eric Wayne J amoom
3 To ewe, s pam man football fan and th at guy
4 ACKNOWLEDGMENTS I am thankful to have much support during my doctoral study. I must thank my mother for instilling within me the character of persistence and value of higher education. I am most grateful to my family for their generous support understanding, and unwavering encouragement during my education I would like to acknowledge my Uncle Si for his influence and foresight, calling me the Professor since I was about 2 years of old I would also like to express my sincere and utmost gratitude and admiration to my supervisory committee chair, Dr. Jeffrey S. Harman, for his invaluable advice, steady encouragement, unprecedented availability, and abundant patience throughout the entire process. I would like to thank the Florida Agency for Health Care Administration for access to their data, and t he Florida Center for Medicaid and the Uninsured for facilitating this process, specifically Heather, Lorna, and Jianyi. I must extend my deepest appreciation to the following mentors. I thank Dr. Christine Chase for introducing me to the world of researc h in her amazing lab as an undergraduate student. I appreciate Dr. Elena Andresen for teaching me the art of research, collaboration, and mentorship. I am fond of Dr. Allyson Hall for her candor availability, advice and counsel, as well as her consistent unwavering support I am indebted to Dr. Paul Duncan for his reliable stream of graduate funding and advice, and his contribution to my academic and professional development as a health services researcher. I thank Dr. David Janicke for without his sug gest ion, this important work in asthma might not have been conducted. I appreciate the expert collaboration of Dr. Marylin Dumont Driscoll, and I appreciate her flexibility, mentorship, and participation in
5 my graduate experience and contributing to this research the practical insight of the pediatrician The support received from friends and classmates was impressive and deeply appreciated. I must acknowledge Britta, Keva, Cameron, Alex, Jingbo, and others for their help. I must especially thank Michael Morris for his gracious support invaluable insight, and lending me his Little SAS book.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...................................................................................................... 4 TABLE OF CONTENTS ...................................................................................................... 6 LIST OF TABLES ................................................................................................................ 8 LIST OF FIGURES ............................................................................................................ 10 LIST OF ABBREVIATIONS .............................................................................................. 11 ABSTRACT ........................................................................................................................ 13 CHAPTER 1 INTRODUCTION ........................................................................................................ 15 Overview ..................................................................................................................... 15 Study Objectives ......................................................................................................... 16 Specific Sub Aims ................................................................................................ 16 Healthy People 2010 Objectives ......................................................................... 17 2 LITERATURE REVIEW .............................................................................................. 18 Pediatric Asthma ......................................................................................................... 18 Mental Health and Asthma ......................................................................................... 20 Asthma and Mental Health Pathophysiology ...................................................... 22 Asthma, Mental Health and Health Service Use ................................................. 25 ADHD association with health services. ....................................................... 26 Anxiety and depression association with health services. ........................... 26 Asthma Pharmacotherapy and Effects on Mood Disorders ............................... 29 Mental Health and Adherence to Treatments ..................................................... 32 Summary of the Literature. ......................................................................................... 34 Features of this Study and Conceptual Overview ..................................................... 35 Study Features ..................................................................................................... 35 Conceptual Framework ........................................................................................ 36 3 METHODOLOGY ....................................................................................................... 41 Research Questions and Hypotheses ....................................................................... 41 Data Source ................................................................................................................ 44 State of Florida Claims Data ................................................................................ 45 Incl usion Criteria .................................................................................................. 46 Description of Outcome, Explanatory, and Control Variables ................................... 48
7 Asthma-Related Expenditures ............................................................................. 49 Volume Data ......................................................................................................... 49 Mental Health Time E ffect ................................................................................... 51 Mental Health Treatment ..................................................................................... 51 Control Variables .................................................................................................. 52 Study Design and Analytic Plan ................................................................................. 53 Fixed Effect Model ............................................................................................... 54 FE model of mental health as modifier of asthma services. ........................ 55 FE Model with mental health treatment as a mediator of mental health in asthma servi ces. .................................................................................... 58 Fixed Effects Model Assumptions ....................................................................... 60 GEE Modeling ...................................................................................................... 61 GEE Model Assumptions ..................................................................................... 62 Institutional Review Board Approval. ......................................................................... 63 Software ...................................................................................................................... 63 4 RESULTS .................................................................................................................... 67 Overview ..................................................................................................................... 67 Sample Characteristics ............................................................................................... 67 Average Monthly Expenditure ............................................................................. 68 Average Monthly Utilization ................................................................................. 68 Negative Binomial Regression ................................................................................... 69 Summary of Utilization Analyses ................................................................................ 80 Generalized Estimating Equation Models and Expenditure Estimates .................... 82 Summary of Expenditure Analyses ............................................................................ 93 5 DISCUSSION ............................................................................................................ 129 Overview ................................................................................................................... 129 Summary and Interpretation of Findings .................................................................. 129 ADHD and Asthma -Related Services ............................................................... 130 Anxiety and Asthma-Related Services .............................................................. 134 Depression and Asthma-Related Services ....................................................... 136 Asthma-Related Pharmacy Claims and Expenditures ...................................... 139 Policy and Research Implications ............................................................................ 141 Future Research ....................................................................................................... 144 Limitations of Study .................................................................................................. 145 Conclusion ................................................................................................................ 152 LIST OF REFERENCES ................................................................................................. 154 BIOGRAPHICAL SKETCH .............................................................................................. 163
8 LIST OF TABLES Table page 3 -1 ICD -9 codes used to select asthma, depression, anxiety, and ADHD ................. 64 3 -2 Summary of monthly outcome measures .............................................................. 65 3 -3 Pharmacy measures defined by Therapeutic Class or National Drug Code ....... 66 4 -1 Sample chara cteristics of youth with asthma (n=8,241) ....................................... 94 4 -2 Average monthly expenditure by mental health condition .................................... 95 4 -3 Average monthly use by mental health condition. ................................................ 96 4 -4 Incident rate ratios for monthly asthma related inpatient admissions .................. 97 4 -5 Predicted mean estimates f or monthly asthma -related inpatient admissions ...... 98 4 -6 Incident rate ratios for monthly asthma related inpatient length of stay (if admissions>0) ........................................................................................................ 99 4 -7 Predicted mean estimates for monthly asthma-related length of stay ............... 100 4 -8 Incident rate ratios for monthly asthma-related medical claims ......................... 101 4 -9 Predicted mean estimates for monthly asthma-related medical claims ............. 102 4 -10 Incident rate ratios on monthly asthma-related outpatient visits ........................ 103 4 -11 Predicted mean estimates for monthly asthma-related outpatient visits ............ 104 4 -12 Incident Rate Ratios for asthma-related urgent visits ......................................... 105 4 -13 Predicted average estimates of monthly asthma-related urgent visits ............... 106 4 -14 Incident Rate Ratios for monthly asthma-related total pharmacy claims ........... 107 4 -15 Predicted average estimates of monthly asthma-related total pharmacy claims .................................................................................................................... 108 4 -16 Incident Rate Ratios for monthly direct asthma pharmacy claims ..................... 109 4 -17 Predicted average estimates of monthly direct asthma pharmacy claims. ........ 110 4 -18 Incident Rate Ratios for monthly indirect asthma pharmacy claims ................... 111 4 -19 Predicted average estimates of monthly indirect asthma pharmacy claims. ..... 112
9 4 -20 Incident Rate Ratios for asthma controller pharmacy claims. ............................ 113 4.21 Predicted average monthly estimates of asthma controller claims. ................... 114 4 -22 Incident Rate Ratios for asthma rescuer pharmacy claims. ............................... 115 4.23 Predicted average monthly estimates of asthma rescuer claims. ...................... 116 4 -24 Predictors of monthly tot al and direct asthma-related expenditure .................... 117 4 -25 Predictors of monthly asthma related medical and pharmacy expenditure. ...... 118 4 -26 Predicted average estimates of total asthma expenditure. ................................. 119 4 -27 Predicted average estimates of direct asthma total expenditure. ...................... 120 4 -28 Predicted mean estimates of monthly asthma-related medical expenditure. .... 121 4 -29 Predicted mean estimates of monthly asthma-related pharmacy expenditure .. 122 4 -30 Predictors of expenditure vs. no expenditure for inpatient and outpatient monthly asthma related services. ........................................................................ 123 4 -31 Predictors of monthly inpatient and outpatient expenditure for asthma related services (assuming expenditure > $0). ................................................................ 124 4 -32 Models 1 and 2 two part model predicted mean estimates of asthma related inpatient expenditure ($). ..................................................................................... 125 4 -33 Models 3 and 4 two part model predicted mean estimates of asthma related inpatient expenditure ($). ..................................................................................... 126 4 -34 Models 1 and 2 two part model predicted mean estimates of asthma related outpatient expenditure ($). ................................................................................... 127 4 -35 Models 3 and 4 two part model predicted mean estimates of asthma related outpatient expenditure ($). ................................................................................... 128
10 LIST OF FIGURES Figure page 2 -1 Mental health utilizati on model for youth with asthma. ........................................ 40
11 LIST OF ABBREVIATION S AH C A Agency for Health Care Administration ADHD Attention deficit hyperactivity disorder CPT Current Procedural Terminology DSM Diagnostic and Statistical Manual of Mental Disorders DSM IV Diagnostic and Statistical Manual of Mental Disorders 4th edition ED Emergency department ER Emergency room FE Fixed effect FFS Fee for Service HLM Hierarchical Linear Model HEDIS Healthcare Effectiveness Data and Information Set ICD -9 International Classifications of Diseases, ninth vers ionICS I nhaled corticosteroids IRR Incidence Rate Ratio LABA Long acting betaadrenergic agonist NAEPP National Asthma Education and Prevention Program NCQA National Committee for Quality Assurance NDC National Drug Code OCS Oral corticosteroids OCD Obsessive -Compu l sive Disorder ODD Oppositional Defiant Disorder OR Odds Ratio PDD Pervasive developmental disorders SABA Short acting beta adrenergic agonist
12 SE Standard error TCC Therapeutic Class Code
13 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 IMPACT OF MENTAL H EALTH DIAGNOSIS AND TREATMENT ON ASTHMARELATED SERVICES IN MEDICAID YOUTH WITH ASTHMA By Eric Wayne Jamoom D e c e m b e r 2010 Chair: Jeffrey Harman Major: Health Services Research The most common chronic condition among children and adolescents in the United States is asthma. An association between asthma and mental health ha s been reported by a multitude of studies. However, characterizing, specifying and quantifying the relationships between mental health comorbidities with the delivery of asthma related services ha ve not been accomplished. This study examines asthma related health s ervices in youth with asthma including assessment of specific factors which may impact asthma -related utilization. Specifically, this study investigates the impact of depression, anxiety and attention deficit hyperactivity disorder ( ADHD ) and the treatment of t hese conditions on asthma-related expenditure s and health service use using a statewide sample of 8,241 youth with asthma between age 6 years and 16 yea rs continuously enrolled in Florida Medicaid over 36 months from January 2002 through December 2004 Th e current research methodology improves upon prior studies by using ind ividual mental health diagnoses a larger sample size, and longitudinal data F our models are used to address the delivery of health services. The first model seeks to explain the effect of ADHD, anxiety, and depression on asthma related use and expenditures The
14 second model examines the length of mental health diagnosis on asthma-related use and expenditure s The third and fourth model s assess the effect of mental health treatment on asthma-related use and expenditures General findings suggest asthma care for those with anxiety is cos tly with high utilization across all asthma -related outcomes indicative of poor coordination. Comorbid depression was associated with g reater inpat ient service use, but generally less asthma medication fills suggesting poor asthma adherence and coordination Comorbid ADHD was associated with lower inpatient use and expenditures, and higher asthma related medication fills and medical claims suggestin g better coordination and adherence to medication regimen and coordination of asthma management Generally, mental health t reatment tended to mediate the inpatient utilization for those with mental health conditions towards similar utilization to those wit hout the condition. Depression treatment suggested better adherence, whereas anxiety treatment suggested more inpatient care. Better coordination of asthma management is indicated for those with comorbid anxiety and depression, as pediatricians tend to ref er these pa tients to psychiatric services.
15 CHAPTER 1 INTRODUCTION Overview Asthma is the most common chronic condition affecting the U S population under age 18 years. While prevalence of pediatric asthma has more than doubled during the last 30 years, the costs associated with pediatric asthma treatment are substantial. Asthma in youth has been associated with significant morbidity and functional impairment leading to youth with asthma having greater health care utilization and costs than children without asthma (Wang, Zhong, & Wheeler, 2005; Blackman & Gurka, 2007; Ortega, Huertas, Canino, Ramirez, & Rubio-Stipec 2002; Sapra, Nielsen, & Martin, 2005). Additionally, a number of studies have suggested that pediatric asthma is associated with having mental health comorbidities ( Katon et al., 2007; Sapra et al. 2005; Ortega et al.,2002; Craske, Poulton, Tsao, & Plotkin, 2001; Goodwin, Fergus son, & Horwood, 2004; Feldman, Ortega, McQuaid, & Canino 2006; Bussing, Burket, & Kelleher, 1996; Blackman & Gurka, 2007; McQuaid, Kopel, & Nassau, 2001; Vila, Nollet -Clemencon, de Blic, Mouren-Simeoni, & Scheinmann et al., 2000). Recent studies are beginning to emphasize the impact of mental health comorbidities on health service use in youth with asthma ( Richardson, Russo, Lozano, McCauley, & Katon 2008; Kewalramani, Bollinger, & Postolache, 2008). Given the high cost and increas ing prevalence of asthma in children and the overwhelming evidence of co occurring mental health conditions like depression and anxiety, there remains a need to further understand how specific mental health diagnose s (e.g., attention deficit hyperactivity disorder ( ADHD), depression anxiety ) impact the delivery of asthma related services. Additionally, there is the question of h ow treatment of each mental
16 health condition impact s the service delivery of asthma-related medical care. This study has implications for mental health and the role of mental health treatment in chronic disease. Therefore, understanding how mental health and its treatment is associated with asthma -related health service use and expenditures is critical to inform develop, and facilit ate appropriate cost containment strategies and highlight areas for policy intervention to improve the quality of health care Study Objectives The main purpose of this study is to further understand the relationship between being diagnosed with asthma and high asthma related service utilization and expenditures. S pecifically this study is interested in how individual mental health diagnosis of ADHD, anxiety, and depression modify utilization and expenditures of asthma-related services. This study expands o n an early study that reported an assoc iation between mental illness and health services use in pediatric asthma (Jamoom, 2010) The specific objectives of this study include Objective 1. To measure the impact of mental health diagnosis on asthma-related utilization and expenditure s Objective 2. To understand time -related effects of having a mental health diagnosis on asthma-related utilization and expenditure s Objective 3. To measure the impact mental health treatment (e.g., pharmacological intervention and counseling) has on the relationship between mental health diagnosis and asthma -related utilization and expenditure s Specific Sub Aims The sub aims include.
17 Sub Aim 1 To understand the relationship of having comorbid mental health diagnoses on asthma -related utilization and expenditures in youth with pediatric asthma. The specific mental health diagnoses that will be assessed include ADHD, anxiet y, and depression. Sub Aim 2 To understand the effect of mental health treatment on asthma-re lated u tilization and expenditures among children with asthma with a comorbid ADHD, anxiety, or depression diagnosis. Healthy People 2010 Objectives T he study objectives are aligned with the goals described in the national plan Healthy People 2010 (USDHHS, 2010) These objectives include: 1 ) T o identify and ultimately increase the proportion of children with mental health problems who receiv e treatment (Objective 18-07) 2 ) T o identify and ultimately reduce hospitalizations and hospital emergency department visits for asth ma (Objective s 24 02 & 2403) 3 ) T o increase the proportion of persons with asthma who receive appropriate asthma care acc ording to the NAEPP Guidelines (2007) ( Focus on written asthma management plans from health care providers (247a) and patient education to recognize and respond to early signs and symptoms (247c) ) (Objective 24 -07). Ultimately successful understanding of the major objectives of this study will further other Healthy People 2010 objectives to increase formal patient education (2406) and track asthma outcomes and access to medical care and asthma management (2408) (USDHHS, 2010)
18 CHAPTER 2 LITERATURE REVIEW Pediatric Asthma Asthma, an illness characterized by airway inflammation, wheezing, and chest tightness, impacts app roximately 9% of the U S population under 18 years and represents one of the most common chronic diseases in children (Moorman et al., 2007; Blair, Breit, & Berkow 2007; Vila, Nollet Clemencon, de Blic, Mouren -Simeoni, & Scheinmann et al., 1998). Pediatr ic asthma is the leading cause of missed school days and hospitalizations in children, and rates of disability due to asthma are increasing (Bousquet, Bousquet, Godard, & Daures, 2005; Blair et al. 2007). Moreover, the prevalence of pediatric asthma has d oubled from 1980 t o 2000 (Sapra et al. 2005 ; Blair et al., 2007 ). This observed increase in prevalence has been thought to be associated with rural and urban distribution of the population, increasing as more communities adopting modern lifestyles and be coming more urbanized ( Bousquet, Ndiaye, Ait -Khaled, Annesi Maesano, & Vignola, 2003 ; Gold & Wright, 2005). In addition to urbanization, changes in diet ( Devereux, 2006 ) and the hygiene hypothesis (Becker, 2007) have been implicated as contributors to the relatively recent rise in asthma prevalence. The hygiene hypothesis expands on the urbanization theme by suggesting that cleaner environmental conditions are implicated in the rise of asthma prevalence due to ha ving less exposure to allergens during immune system development, and are at least in part responsible for an increase in asthma prevalence in developed count ries (Becker, 2007) Supporting this hypothesis, being in a farming environment being exposed to cats, going to child care, having a large family during the immune system developing
19 years, as well as birth order have been associated with lower asthma prevalence (Becker, 2007). The costs associated with the treatment of pediatric asthma are substantial In 2002, health care expenditures for asthma totaled $14 billion ( Blair et al. 2007). Most recently Kamble and Bharmal (2009) have estimated direct expenditures for the treatment of chi ldren with asthma in the U S at $6.39 billion. The same study deter mined that the annual direct medical expenditure attributable to asthma treatment (both adults and children) is estimated at approximately $37.2 billion in 2007 U.S. dollars, representing a significant portion of healthcare resource use in the U.S. (Kamble & Bharmal, 2009). Kamble and Bharmals estimate is relatively higher than the previously reported estimates of cost in asthma which range from $3.6 billion to $30.8 billion ( Smith, Malone, Lawson, Okamoto, & Battista, 199 7; Weiss, Gergen, & Hodgson, 1992; Weiss, Sullivan, & Lyttle, 2000; Druss et al., 2001; Wang et al. 2005; Lozano, Sullivan, Smith, & Weiss 1999; Yelin et al., 2002). Children with asthma use approximately double the level of health services (i.e., hospitalizations, outpati ent visits, emergency department visits) relative to children without asthma ( Sun, Kao, Lu, Chou, & Lue, 2007; Kamble & Bharmal, 2009). A study of the North Carolina Medicaid system reported the average annual health care cost for patients with asthma exceeded that of patients without asthma with similar demographic characteristics by over $1000 (Sapra et al. 2005). Moreover, approximately 20% of patients with asthma generated nearly 80% of direct expenditures (Sapra et al., 2005); suggesting a subset of p atients with particular characteristics are at increased risk for using more health services. Given the high cost and increasing prevalence of asthma in
20 children and adolescents, enhancing our understanding of factors associated with health service utiliza tion in this population is critical to facilitate appropriate cost containment strategies and improve the quality of care. Mental Health and Asthma Emotional causes have frequently been believed to be associated with asthma exacerbations (Bloomberg & Chen 2005; Lehrer, 1998). However correlation does not imply causality, and the causal relationship between asthma and mental health has not been fully appreciated nor understood. Many studies fail to address mental health and asthma exclusively in children, o ften including a mixed age sample or exclusively adult populations. Therefore, studies using mixed samples or adult populations may be used to fill in any gaps in the literature. ADHD and asthma. A 1996 literature review on the relationship between childhood asthma and ADHD failed to find any association between asthma and ADHD (Daly et al., 1996). However, the sample size of most studies contained within the literature review article seemed to not address asthma or ADHD directly or exclusively and had fair ly small sample s of youth with asthma. Hence, the literature has not been able to establish a clear association between asthma and ADHD and other externalizing psychiatr ic symptoms (e.g., OCD, ODD). Externalizing psy chiatric d isorders are manifested in chi ldren's outward behavior rather than (or in addition to) their internal thought s and feelings. Internalizing psychiatric disorders are manifested in a childrens inward behavior, and include depression and various types of anxiety, such as separation anxiety disorder and generalized anxiety disorder. In a longitudinal study examining the effects of internalizing and externalizing behavior problems in adolescence, there was no association noted between asthma prevalence and
21 externalizing symptoms; however, c hildren diagnosed with asthma by age 5 years were at increased risk during adolescence for internalizing behavior problems (Alati et al., 2005). A nxiety and depression association with asthma. Studies have found asthma patients to have higher rates of depressive and anxiety s ymptoms than healthy controls. In 2000, a literature review of asthma found that of 8 studies reviewed, all had indicated that depressive symptoms were more common in children and adults with asthma than in the general population (Z ielinski et al., 2000). Up to this point, no consistency was obtained on the prevalence of formal depressive disorder s and diagnoses in patients with asthma. In a study of 743 adults with asthma, Eisner Katz, Lactao, and Iribarren, (2005) found 18% of all those adults with asthma had depressive symptoms based on the Center for Epidemiologic Studies Depression Scale. Children and adolescents with asthma have higher rates of depressive and anxiety disorders Goodwin et al (2004) studied asthma in adolescence and young adulthood and found that asthma was also associated with an increased likelihood of major depression, panic attacks, and any anxiety disorder. Furthermore, the Youth Risk Behavior Survey has shown that high school students with asthma report h igher rates of depressive symptoms such as hopelessness or feeling sad (45.3% vs 29.3%) than non asthma peers (Bender, 2007). In a prospective study examining the relationship between mental health problems in childhood at age 8 years and physical disorders at follow -up during early adult hood at ages 18 to 23 years, Goodwin and colleagues (2009) found that h aving early onset asthma (at age 8 years) was associated with moderate to severe conduct problems,
22 and moderate to severe depressive symptoms at start of the study. However, having incident onset of asthma at follow -up as a young adult was predicted by moderate and severe depressive problems at age 8 years (Goodwin et al., 2009) This association suggests childhood depression may be involved in the development of asthma yet correlation does not necessarily imply causation. Katon et al (2007) reported youth with asthma demonstrated twofold higher prevalence of comorbid DSM -IV anxiety and depressive disorders compared with nonasthma controls. Furthe rmore, a meta analysis of childhood asthma studies revealed these children exhibited poorer behavioral and emotional functioning and had more internalizing and externalizing disorders than either comparison group of healthy peers or a normative sample (McQ uaid et al. 2001). Glazebrook and colleagues (2006) reported children with asthma scored higher on measures of emotional disturbance than nonasthma controls. Likewise, Vila et al (2000) determined in their study of French children with asthma between ag es 8 years to 15 years that 42% had one or more DSM diagnose s. Asthma and M ental H ealth Pathophysiology While asthma is described as chronic inflammation involving airway hyperresponsiveness and bronchial obstruction leading to symptoms of coughing, wheez ing, chest tightness, and dyspnea, the underlying pathophysiology of asthma can be present without overt symptoms (Kewalramani et al. 2008). Regulated by T lymphocytes, immune responses consist of T helper type 1 (Th1) or T helper type 2 (Th2) cells. Th1 cells are primarily involved in response to infection, whereas Th2 cells are primarily involved in the allergic response. Technically, asthma is a disorder of the
23 conducting airways characterized by Th2 cell mediated inflammation and increased me diator release (Kewalramani et al. 2008, p p. 4 5). Asthma is often brought on by viral infection or allergen exposure. Antigen presenting cells (APCs) recognize the allergens and present them to T lymphocytes. If one is predisposed to the allergy phenotype, the T helper cells develop into Th2 cells inducing B lymphocytes to undergo a class switch from immunoglobulin M (IgM) to immunoglobulin E (IgE). IgE levels are typically increased in atopic asthma, as IgE binds to receptors on effector cells (e.g., mast cells basophils and eosinophils) found in the respiratory mucosa leading to sensitization. Thereafter, when individuals are exposed to these allergens again the allergen is cross -linked with IgE molecules and induce the e ffector cells to degranulate. Thereby, the effector cells release a host of cell mediators including histamine, tryptase, cytokines, leukotrienes, and prostaglandins inducing asthma symptoms (Lily, 2005). Asthma and other atopic disorders are thought to be exacerbated by psychological stress (Slattery 2005). Some studies suggest that psychological stress shift cytokine balance from Th1 towards Th2, causing an immune dysregulation with more hyperresponsive immune response during periods of high stress. Higher periods of stress activate the sym pathetic nervous system and hypothalamic -pituitary adrenocortical axis leading to an increase in cortisol and catecholamine secretion, which suppress Th1 cytokines (e.g., IL-12, IFN gamma) shifting immune response towards the Th2 phenotype leading to asthm a pathogenesis (Chrousos, 2000). Asthma is also associated with nocturnal symptoms and a decrease in lung func tion (Kewalramani et al., 2008). Those who have asthma have an increased
24 likelihood to have rhinitis (Bousquet et al., 2008). Both allergic rhinit is and nonallergic rhinitis are risk factors for sleep apnea ( Kavut, & Elili, 2009 ; Bousquet et al., 2008). Impaired sleep leads to daytime fatigue, difficulty concentrating, reduced productivity, deteriorating mood, overall lower quality of life and can exacerbate depression in individuals (McEwen, 2006). Additionally, impaired sleep due to sleep disordered breathing can be misdiagnosed as ADHD due to the substantial overlap between impairments of sleep disordered breathing and diagnostic criteria for ADHD (Owens, 2009 ). The extent that treating the sleep disorder has been suggested to normalize the ADHD scores and even find that 50% of children no longer met ADHD criteria (Owens, 2009). Generally, factors associated with impaired sleep are also considered depressogenic, and anxiety has been associated with the uncertainty of asthma and its attacks (Levenson, 2005). Other studies have suggested that increased anxiety can result from hypercapnia (or condition of elevated carbon dioxide in the blood associated) by directly impacting an activity change in the locus coeruleus. The locus coeruleus is the area of the br ain associated with the noradrenergic system that supplies norepinephrine throughout the central nervous system, and ultimately manages ones physiological response to stress and panic (Carr, 1998; Zaubler & Katon, 1998). In the review by Daly and colleagu es (1996), some medications have been associated with ADHD like symptoms in some youth with asthma. Specifically, the asthma medications with ADHD symptomology include theophylline ( Furukawa et al., 1988; Bender & Milgrom, 1992; Rachelefsky et al. 1986) a nd corticosteroids ( Milgrom & Bender, 1993 ). Compared to children with asthma taking theophylline, youth with
25 asthma who discontinued the use of theophylline had a significant improvement on concentration scores from the Stroop Test I (Furukawa et al., 198 8). Daly and colleagues review suggests no evidence exists on asthma medications causing minimum symptoms for ADHD diagnostic criteria However, their literature review article assessing the association between asthma medication and ADHD symptomology is l imited by the small sample size ranges of 6 to 13 subjects in their compiled studies Asthma, Mental Health and H ealt h Service U se Studies suggest the presence of psychiatric diagnoses or childhood behavioral, social, or emotional problems are associated w ith greater pediatric health service utilization (Janicke, Finney, & Riley, 2001). This finding is especially important for those with pediatric asthma, as youth with asthma have been associated with mental health issues, most notably comorbid internalizin g psychiatric diagnoses, such as depressive or anxiety dis orders ( Bussing et al., 1996 ; Vila et al., 2000; Craske et al. 2001 ; McQuaid et al. 2001; Ortega et al., 2002; Goodwin et al., 2004; Sapra et al. 2005; Feldman et al., 2006 ; Blackman & Gurka 2007; Katon et al., 2007). The association between asthma and ADHD, however, has not been fully appreciated (Daly et al., 1996). This association between asthma and behavioral symptoms can result in greater asthma burden (Richardson et al., 2006), additi ve functional impairment (McQuaid et al. 2001; McCauley, Katon, Russo, Richardson, & Lozano, 2007), and increased asthma medications, emergency room visits ( Nouwen, Freeston, Labbe, & Boulet 1999), and hospitalizations ( Dirks, Kinsman, Horton, Fross and Jones 1978). Moreover, the association between asthma and psychiatric symptoms has been suspected to be associated with adverse effects of controller and rescue medications, difficulties with managing a life -threatening illness, family stress and low soci oeconomic status, all of
26 which negatively impact both asthma symptomatology and psychiatric illne ss (Vila et al., 1998; McQuaid et al. 2001; Goodwin, Messineo, Bregante, Hoven, & Kairam 2005; Mrazek, 1992). ADHD association with health services. The ec onomic impact of ADHD on asthma is incomplete, as many studies fail to separate mental health services from other health care (Pelham, Foster, & Robb, 2006). Independently ADHD has been discussed as contributing to increased general utilization and health costs ( Guevara, Lozano, Wickizer, Mell, & Gephart, 2001). Specifically, in their study addressing ADHD utilization and cost compared to youth without, those with ADHD had incurred significantly greater total costs ($1465 vs $690), and had 3.4 times more pharmacy fills (11.25/year vs 3.30/year), and 1.6 times more primary care visits (3.84/ year vs 2.36/year) While suggestive of increased use and higher costs, an association between asthma prevalence and ADHD has not been fully understood (Daly et al. 1996). I t has been suggested, however, that ADHD may be actually misdiagnosed for what is an undiagnosed sleep disorder, like obstructive sleep apnea (Owens, 2009). I n a national Medicaid youth sample, externalizing conditions like ADHD have been associated with increasing costs and health services utilization (Chan, Zhan, & Homer, 2002). Overall costs for either ADHD or asthma have been shown generally to be similar. However, ADHD -related prescriptions costs, total out of -pocket expenses, outpatient visits and prescriptions claims were higher for ADHD than asthma. Anxiety and depression association with health services. Few studies however have examined factors associated with health service use in children with asthma, much less the additive effect of ps ychiatric diagnoses.
27 Mohammed and colleagues (2006) reported 40% of pediatric patients with asthma had two or more visits to the emergency department over a oneyear retrospective period. Additionally, repeat visits were associated with younger patient ag e, presence of maternal asthma, and exposure to environmental triggers. However, psychiatric symptoms were not investigated as a predictor of emergency department visits. In a study of pediatric and adult asthma patients, Sapra, Nielsen, and Martin (2005) detected increased healthcare costs for those patients presenting with a diagnosis of depression. Likewise, Goodwin and colleagues (2005) reported that symptoms of anxiety and depression were quite common among 5 to 11 year old inner city patients with a sthma, and linked these psychiatric symptoms with an increase in health care utilization. In a set of unadjusted analyses, Richardson et al (2008) found that compared to youth with asthma alone, those with comorbid anxiety/depressive disorders had more p rimary care visits, emergency department visits, outpatient mental health specialty visits, other outpatient visits and pharmacy claims. Specifically after controlling for asthma severity and covariates, total health care costs were approximately 51% higher for youth with asthma that had depression with or without an anxiety disorder but not for youth with asthma that had an anxiety disorder alone. The authors discussed that most of the increase in health care costs was attributable to nonasthma and nonme ntal health related increases in primary care and laboratory/radiology expenditures. Preliminary study of nonpsychiatric utilization and expenditure by psychiatric diagnosis group in Medicaid youth with asthma. In a Florida Medicaid study of youth with as t hma, Jamoom and colleagues (2010) found that youth between
28 age 5 and 15 years with a comorbid externalizing psychiatric diagnosis, such as ADHD had fewer nonpsychiatric related ER visits and inpatient costs compared to those without the externalizing psyc hiatric diagnosis. Furthermore, Jamoom and colleagues contributed to the consistency in the literature with their cross-sectional findings that comorbid internalizing psychiatric diagnoses such as depression and anxiety were associated w ith a comprehensive increase in annual nonpsychiatric related expenditures as well as the amount of health ca re resources used. While their research finding assessed general trends associated with groups of psychiatric diagnoses and nonpsychiatric related expenditures and ut ilization in youth with asthma, they had several limitations which could be improved upon. L imitations from that study include both a liberal definition of asthma and mental health For mental health, general psychiatric diagnostic groups (e.g., internaliz ing or externalizing) were used rather than individual psychiatric diagnosis type (e.g., ADHD, anxiety). Therefore, they suggest assessing individual mental health conditions rather than at the aggregate grouping by symptomology (Jamoom et al., 2010). Addi tionally, they did not require more than one ICD -9 diagnosis for asthma, thus increasing the likelihood for false positives. Specifically, their research suggests further assessment of asthma-specific services rather than the more general trends associated with assessing nonpsychiatric -related services Improvement in stringent criteria for asthma diagnosis, as sessing mental health treatment and asthma sev erity, assessing the effects of time with mental illness o n the patterns of health service delivery, and using a longitudinal approach were targeted goals for future research.
29 Kewalramani, Bollinger, and Postolache (2008) indicate in their review of asthma and mood disorders that there exi st overwhelming evidence on the high rate of co occurrence of depre ssion/anxiety and asthma for children, adolescents and adults, and further research is necessary to establish the link between these conditions and specifically whether concurrent treatment of depression and anxiety in asthma patients improves asthma symptoms. Additionally the review implies the need to pay attention to different triggers, both environmental (e.g., spring pollen) and genetic (e.g., immunologic/cellular mediators of allergic response) that lead to increased periods of vulnerability. Asthm a Pharmacotherapy and Effects on Mood Disorders Asthma Pharmacotherapy is divided into controller medic ations and rescue medications. Controller medications are taken on a daily basis, whereas rescue medications are used for acute asthma episodes to reliev e episodes of crisis. Many of these medications can result in adverse psychological events, such as depression or anxiety symptomology ( Kewalramani et al. 2008). Controller medications include inhaled corticosteroids (ICS), leukotrine modifiers, cromones long acting beta adrenergic agonists (LABA ), immunomodulators, and methylxanthines. Rescuer medications include anticholinergics, systemic corticosteroids, and short acting beta androgenic agonists (SABA ) (NAEPP, 2007). Corticosteroids inhibit cytokine, prostaglandin, and leukotrienes production through preventing inflammatory cell activation and migration, and decreasing microvascular leakage ( Kewalramani et al. 2008). Corticosteroids are considered the most potent and effective longterm anti inflammat ory medication for asthma ( Kewalramani et al. 2008).
30 For those with acute asthma exacerbation, oral corticosteroid (OCS) bursts of prednisone and prednisolone are frequently prescribed. Uses of systemic corticosteroids have been associated with depression, mania, and psychosis. In a study of adults evaluated for receiving a minimum of 40mg of prednisone for asthma exacerbation during at least a seven day period, Brown, Suppes Khan, and Carmody (2002) found that after assessing patients before during and after systemic corticosteroid therapy with the Hamilton Rating Scale for Depression, the Young Mania Rating Scale, the Brie f Psychiatric Rating Scale and the Internal State Scale, there was a significant increase in manic symptoms. For patients with depre ssion, their study found a reduction in symptoms compared to those without depression. However these changes resolved with discontinuation of the steroids. However other studies have suggested that chronic steroid use seems to be associated with an increa se in depressive symptoms (Craig, Teets, Lehman, Chinchilli, & Zwillich, 1998). Brown, Vera, Frol, Woolston, and Johnson (2007) found in a follow up study, that those with asthma chronically using prednisone was associated with higher scores on psychiatric measures than those not on chronic steroids. For patients with moderate to severe asthma, a complex adjunct therapy in addition to ICS may include leukotrienes modifiers, LABAs, chromones, theophylline, and omalizumab, which th e impact of these medications on comorbid depressive and anxiety disorders in patients with asthma have not been extensively studied ( Kewalramani et al. 2008). Other studies have found that depressive symptoms are associated with some of the steroids used in treating asthma. Morrison, Goli, Van Wagoner, Brown, and Khan (2002) studied 46 patients age 6 to 17 years who presented to a low income family
31 asthma clinic. The clinic found 86% of children were on medium to high doses of inhaled corticosteroids for their asthma and 41% had mild to moderate airway obstruction on inhaled steroids. Furthermore, 30% of these patients also had Carroll Depression Scale -Revised scores consistent with likely, very likely or almost certain depressive disorder. Leukotrine modifiers have been shown to decrease use of rescue medications, decrease night awakenings and improve lung function (NAEPP, 2007). According to Singulars package insert, Montelukast, a leukotrine receptor antagonist, has uncommonly been associated with dream abnormalities, drowsiness, insomnia, depression, and suicidal thoughts (Kewalramani et al. 2008) No effects on depression or anxiety have been reported for chromones, bronchodialtors, and immunomodulators like Omalizumab ( Kewalramani et al. 2008). How ever, side effects of theophylline, a phosphodiesterase inhibitor, include seizures, insomnia, anxiety, and tac hyarrhythmia. Theophyllines have decreased in use for children. Studies on the pharmacological side effects of asthma medications remain something that continues to be researched due to the large numbers of people impacted with asthma in the U S. Biological consequences of asthma include biological predispositions to mood and mental health conditions ( Kewalramani et al. 2008). The complex interac tion between asthma and the treatment of asthma have included increased depression, anxi ety, and an inability to sleep. Insomnia seems to be a common side effect of asthma medicati ons, but may also relate to allergic or nonallergic rhinitis. Asthma has a s trong association with allergic rhinitis, which has been implicated in obstructive sleep apnea and sleep disordered breathing ( Kavut, & Elili, 2009; Owens, 2009).
32 Having sleep disordered breathing results in inability to concentrate and sympt omology associated with depressive disorders and ADHD (Owens, 2009). Sleep disordered breathing has been reported as being often misdiagnosed as ADHD (Owens, 2009). The treatment of mental health and asthma together seems quite complex. In one study by Pre torius (2004), it was suggested th at some individuals receiving asthma treatments like corticostero ids that lower serotonin levels might present with symptoms of depression, ADHD, oppositional defiant disorder and even conduct disorder; and that treating the mental health comorbidity with selective serotonin reuptake inhibitors (SSRIs ) and psycho -stimulants may result in the upregulati ng of serotonin levels which in turn trigger s asthma (Pretorius, 2004) Concerns surrounding the side effects of asthma me dications and diagnosis of ADHD have been suggested but not supported by the literature (Daly et al. 1996). Furthermore, the same study did not show any effects of ADHD medications and the impacts on asthma. In fairness, the literature revi ew reported sma ll scale studies with fairly small sample sizes that did not support a significant association of ADHD on asthma or asthma medications causing ADHD. Mental Health and A dherence to T reatments Better consideration of mood disorders like depression and anxiet y among i ndividuals with asthma has been gaining momentum due to the potential for poor adherence to asthma medication s (Kewalramani et al. 2008). Poor adherence has been observed in those suffering with specific chronic conditions like asthma. Among thos e with comorbid mood disorders, major depression was associated with poor adherence in diabetes care (Lin et al., 2004). Specifically, r esearchers suggest that m ajor depression was associated with less physical activity, unhealthy diet, and lower adherence to oral hypoglycemic, antihypertensive, and lipid -lowering medications (Lin et
33 al., 2004) Generally, compared to those without depression patients with depression are more likely to h ave other chronic comorbidities and are up to 3 times more likely to have adherence problems to medications prescribed for their medical comorbidities (Cramer & Rosenheck, 1998; DiMatteo, Lepper, & Croghan, 2000). While mood disorders like depression has led to poor adherence, some studies have shown antidepressant drug adherence improves comorbid disease medication adherence and reduced total medical costs for chronic disease (Katon, Cantrell, Sokol, Chino, & Gdovin, 2005). A review article from the L ancet addressing medication adherence in youth and young adults with cancer, reports that up to 63% of patients do not adhere to their treatment regimens with adherence factors which include patient emotional functioning ( e.g., depression and self esteem), patient health beliefs ( e.g., perceived illne ss severity and vulnerability), and family environment ( e.g., parental support and parent -child concordance) (Kondryn, Edmonson, Hill, & Eden, 2010) In a study of hypertension, 89 patients with mental health comorbidities, specifically psychotic disorders were matched with randomly selected age comparable subjects that were assessed for similar antihypertensive medication adherence (Dolder, Furtek, Lacro, & Jeste 2005). Researchers found that those with psychotic disorder were significantly less likely to have controlled blood pressure during the first year study period, suggesting that mental health comorbidities adds a concern to medical providers monitoring of chronic disease treatment adherence (Dolder et al., 2005). Some studies suggest poor adherence is more an issue of education and health beliefs. In a study assessing adherence in youth with asthma in Australia, poor
34 adherence was f ound to be associated with health beliefs held about specific asthma medications (Naimi et al., 2009). Therefore, ones health beliefs about asthma medications may not just impact adherence but ultimately their asthma-related utilization and expenditure. Current ly, the literature on asthma treatment adherence and mental health is quite limited (Opolski &Wilson, 2005). Ast hma adherence has a negative association with age (McQuaid, Kopel, Klein, & Fritz, 2003). As children age, they may be at particular risk for poor adherence and clinicians need to be aware and adherence studies are needed in youth with asthma (McQuaid et al., 2003). Summary of the Literature Asthma is a costly, prevalent condition impacting youth with asthma. Additionally, asthma has been associated with mental health conditions at relatively high rates compared to those without asthma. Large scale studies that assess the impact of mental health on the use and expenditure for youth with asthma have not been done despite the high asthma prevalence and cost. M ental health comorbidities may impact asthma medication adherence in youth with asthma. Additionally, having asthma symptoms (e.g., airway constriction) have been suggested to influence mental health condition symptomology and vice versa. U nderstanding the relationship of individual mental health conditions and mental health treatment impact on the medical care of chronic conditions like asthma r epresents a novel and important rationale behind conducting this study Moreover, findings from this study have implications on the management and quality of care for such patients, such that findings may lead to b ett er management and coordination of care. Thus, t he underlying objective for this very important research to investigate the relationship between ADHD, anxiety, and
35 depression as a modifier of the relationship between youth with asthma and their asthma re lated service use. The effects of time on this relationship also have not been addressed in the literature. Time with a mental health condition should in theory impact adherence to asthma medications, continuity of care, and utilization patterns an d expend iture for types of asthma -care (e.g., inpatient, outpatient, or medical care). While the effects of mental health treatment have not fully been discussed in the literature, an argument can be made that the treatment of mental health will improv e adherence to asthma-related therapy ( Katon et al., 2005; Kewalramani et al. 2008). Additionally assessing the exposure to the medical care system (e.g., mental health treatment) over time may have implications on disease management and coordination care. Therefore there is a need to conduct a longitudinal study to answer whether the experiences for youth with asthma that have mental health conditions and receive treatment for their mental health condition have improved asthma disease management, measured a s utilization differences in asthma related care and asthma related pharmacological therapy Features of this Study and Conceptual O verview Study Features This study will fill some of the existing literature gap by using several techniques to improve on th e limitations of previous studies. In this study, health care utilization and exp enditures are focused to asthma-specific use and expenditure, as distinct from prior studies that were using a general cost or nonasthma specific dependent variables. Second, this study improves on other definitions of asthma in the claims data by limiting the definition of asthma to those youth with at least 2 asthma claims in the data within the first year of the study an d then following the cohort for an entire 36 month pe riod to
36 understand their mental health experience. Additionally, this study follows a vulnerable population, Florida Medicaid youth with asthma. The sample was adjusted to remove all those enrollees with over a 2 month gap in coverage at one time during th e 36 month period (note that multiple gaps in eligibility was allowed, as long as those gaps in eligibility were not longer than a 62 day period) These eligibility gaps result from the constant enrollment and disenrollment that occurs in the Medicaid population. The asthma literature displays consistency in the increased association between mental health conditions like depression and anxiety with asthma. However, the relationship has not been fully elucidated with respect to quantifying the effect of how mental health conditions impact asthma management and costs in youth with asthma (Kewalramani et al., 2008; Jamoom et al., 2010 ). Since a number of areas exist where the literature fails to illuminate the impact of common cooccurring individual mental health diagnosis (i.e., depression, anxiety, and ADHD) on asthma -related costs and service use, this study will address how co occurring depression, anxiety, and ADHD will affect asthma -related utilization and expenditure. Conceptual Framework The literature implies that ADHD, anxiety, and depression would impact asthma outcomes in different ways. Therefore, in the proposed Mental Health Utilization Model for youth with asthma (Figure 21), mental health acts as a modifier of the asthma utilization relationship for those children with an asthma diagnosis. The underlying premise is that youth with asthma are going to use asthma related health services based on their age, sex, race/ethnicity, and SSI status. Since mental health comorbidities like ADHD, anxiety, and depression often accompany chronic conditions like asthma, as described in the literature, this conceptual model seeks to quantify,
37 explain, and understand the degree to which ADHD, anxiety and depression impact asthma-related outcomes. The second par t of this conceptual model reflects the time-related association by which the number of months one has ADHD, anxiety, and depression impacts asthmarelated expenditure and utilization. It would be expected that different types of asthmarelated services wo uld have different time -related effects associated with length of time one has their mental health condition. This is particularly true if management of the specific mental health condition impacts adherence to asthma treatment or has a very different coor dination for the care of mental health and asthma related services. For example, for youth with asthma that have a new diagnosis of depression, they may initially use more asthma services. However, adherence to medications represents a potential problem, a s suggested by the literature. In theory the longer one has depression, the more likely they are to not manage their asthma and have higher inpatient care, or become less likely to adhere to asthma medication. If mental health services are handled outside of the medical home (i.e., psychiatrist referral), a psychiatrist may be unaware of asthma-related issues and the pediatrician may be unaware of mental health related issues. Hence, poor coordination of asthma and mental health care can result in different utilization trends. Understanding the length of time one has a mental health condition may help in further understanding the trends associated asthma -related utilization. The third part of the model assesses the mediating effects of mental health treatmen t on the overall relationship between mental health and asthma -related services. The literature suggests poor adherence for those with comorbid depression to
38 pharmacological treatment for chronic conditions, like asthma. Having a treatment for the mental h ealth condition like depression, should improve adherence to asthma treatment (e.g., direct asthma fills). This conceptual framework allows for the ability to assess the overall effects of ADHD, anxiety, and depression treatment on asthma related utilizati on. The literature suggests that mental health treatment for depression improves adherence to other chronic disease medications (Katon et al., 2005). The continuity of mental health treatment, whether you regularly fill and use your medication, happens to be another piece of the adherence (Katon et al., 2005). In other words just presence or absence of treatment does not indicate regular use. Therefore, the length of mental health treatment can offer insight as a proxy measure for adherence. This framework helps to address whether the presence or absence of treatment or a more continuous measure of mental health treatment has a larger effect on the propensity for youth with comorbid ADHD, anxiety, or depression to use asthma related services. This utilization model provides a conceptual framework to generally analyze asthma-related health service use and expenditures among youth with asthma and the impact of mental health and associated mental health treatments for this population using claims data. This model intentionally omits other traditional behavioral components impacting utilization described frequently in the literature (e.g., parental factors, smoking, etc). There are benefits to consider this model within a larger behavioral utilization model context. For example, Dr. Andersen posits using psychological characteristics as part of his predisposing characteristics in his Behavioral Model of Health Service Use (Andersen, 1995). One can see how a modified Andersen would potentially work in this
39 study, a s well. However, for simplicity, this study is also addressing mental health treatment, which may for this population of continuously enrolled youth with asthma be thought of as implied utilization (i.e., one must use services to have treatment) but also t hought of as an enabling resource (i.e., prior use increases future use), but also may be an indicator of need (i.e., as continuous mental health treatment may indicate regular adherence) for those with high number of months of continuous treatment. In thi s sample everyone is continuously enrolled, hence treatment has a variety of different roles in a traditional Andersen Behavioral approach. So the model proposed in Figure 31, allows for each component of this claims -based analysis to be accounted for in a simplified utilization model that allows for both mental health pathology as a modifier of asthma service use and mental health treatment as a mediator of the underlying mental health pathology. The Andersen Model can be called upon in discussing some of the findings with respect to some limitations of utilization analyses with claims data.
40 Figure 21. Mental health utilization model for y outh with a sthma. Overall conceptualization of mental health treatment mediating the underlying mental health conditions impact on modifying the delivery of asthma -related health services Depression, anxiety, and ADHD modify the relationship between asthma and asthma related services. Additionally, the conceptualization assumes the number of months with a mental health diagnosis may have an effect on asthma related services. Lastly, mental health treatment (either presence or absence of the respective treatment or as continuous months of treatment) may mediate the relationship between mental health and asthma rela ted services.
41 CHAPTER 3 METHODOLOGY Research Questions and Hypotheses This study addresses three main questions 1 How do es the presence of ADHD, anxiety, or depression among youth with asthma influence their asthma management, as measured as asthma -related health service use and expenditures ? 2 Is there a relationship between onset of a diagnosis of ADHD, anxiety or depression among youth with asthma and their asthma management, as measured by asthma -related service use and expendit ures ? 3 If a relationship exists between having a diagnosis of ADHD, anxiety or depression among youth with asthma and their asthma, as measured by asthmarelated health service use and expenditure does treatment (either pharmacological and/or pharmacologi cal or physician counseli ng) affect asthma management, as measured by asthma-rel ated health service use and expenditures ? Hypothesis1 a Th e literature is incomplete regarding the association between asthma and ADHD. Generally, children with ADHD use significantly more medical resources and incur significantly higher costs than children without ADHD ( Guevara et al. 2001 ). However, the associated increase in costs may not be the case in youth with asthma with respect to ADHD. P rior work by Jamoom et al (2010) suggest that given at lea st some health care expenditure, yout h with externalizing disorders or conditions where the psychiatric disorder is expressed externally ( e.g., ADHD) had significantly lower annual nonpsychiatric -related inpatient, outpat ient, and total expenditures compared to those without a comorbid externalizing psychological diagnosis. Additionally, having a comorbid externalizing psychiatric diagnosis was associated with significantly fewer nonpsychiatric -related annual inpatient adm issions, shorter inpatient lengths of stay, fewer inpatient ED Visits, and fewer outpatient visits compared to those without exter nalizing psychiatric diagnosis. Therefore, with the exception of the
42 preliminary s t udy by Jamoom et al., the literature on ADH D and asthma suggest no association between youth with asthma and ADHD. However, this study hypothesizes that those youth with asthma with a comorbid ADHD diagnosis should use fewer asthma-related services and expenditure s compared to those without ADHD. H ypothesis 1 b &1c. The literature has suggested that youth with asthma that have a cooccurring depression and or anxiety have greater health service expenditure s. Jamoom and colleagues (20 10) have also supported the findings in the literature that those with comorbid internalizing psychiatric diagnoses (e.g., anxiety, depression) had incurred significantly more nonpsychiatric -related yearly inpatient admissions, longer length of stay, more inpatient ED visits, more outpatient visits, more outpatient ED visits, more medical/physician claims, and more pharmacy claims relative to those without internalizing diagnosis. The same study found that children with an internalizing psychiatric diagnosis were significantly more likely to have greater averag e yearly nonpsychiatric -related inpatient expenditures (p<.001), outpatient expenditure s (p<.001), and medical/physician expenditures (p<.01) compared to those without an internalizing diagnosis. Therefore, based on prior literature and a prior study asses sing this similar population, youth with asthma that have comorbid anxiety or depression should use more asthma-related services and incur greater expenditure s compared to those without anxiety (1b) or depression (1c). Since adherence may be an issue for y outh with comorbid depressive disorders, using less asthma-related pharmacy claims is also hypothesized. Hypothesis 2 The impact of the number of months with a mental health condition may impact the utilization of asthma-related services While this temp oral relationship is
43 unclear this study seeks to furth er understand the impact of time with a mental health condition on the as sociation with asthma care Ultimately, asthma related services may have independently different e ffects for the length one has depression, anxiety, or ADHD over time depending on the type of the health service used (e.g., pharmacy use vs. inpatient use) and mental health condition experience over time. From the literature, it is unclear whether mental health conditions trigger ast hma or asthma triggers mental health issues or neither Airway constriction and asthma symptomology has been implicated in changing brain function (Carr, 1998; Zaubler & Katon,1998; Kewalramani et al., 2 008). While the evidence of changing brain function d ue to asthma symptomology has been implicated in the are a of the brain associated with norepinephrine fear and panic, mental health symptomology triggering asthma attacks has also been duly noted (Blair et al. 2008; Lehrer, 1998). The asthma utilization differences associated with length of time with chronic ADHD, anxiety, or depression may help to explain these different trends. However, a significant increase or decrease in asthma utilization over time could suggest a feedback loop of requiring more or less utilization due to changing medical needs, which has relatively low mutability (Andersen, 1995). Therefore, needs related to the treatment of asthma may be complex for the number of months that youth with asthma have different mental health comorbidi ties. However, considering the length of time with ADHD, anxiety, and depression over time, stable utilization trends may be present For example, the number of months one has anxiety or depression may be associated with more asthma -related inpatient care, suggesting poorly managed asthma or may be associated with decreased asthma pharmacy fills if asthma adherence is poor.
44 Ultimately, time effects for asthma medication use would be expected to be greater in those with better adherence. In theory, depressi on has been implicated in poor adherence in a variety of chronic conditions. Poor adherence has been associated with needing more emergency care (Katon et al., 2005). Therefore, the months one has a depressive disorder is hypothesized to be associated with a d ecrease in asthma medication over time, but associated with more asthma i npatient use over time; whereas better asthma treatment adherence would be associated with more asthma medi c ation expenditure and use over time. ADHD and anxiety may have better u tilization patterns over time, as described. Hypothesis 3 From the literature, t he impact of successful treatment of mental health comorbidities in youth with asthma remains unclear. However, it has been suggested by a number of studies that the regular t reatment of mental health conditions help to improve chronic conditions, like asthma related outcomes through better adherence (Katon et al., 2005; Kewalramani et al., 2008). Therefore, I hypothesize that successful treatment for each mental health conditi on will mediate the asthma -related service use to appear more similar to those without mental illness (i.e., observe diminished effects in the presence of treatment variables for ADHD anxiety, and depression). Direct treatment effects for all conditions s hould include more use and expenditure for pharmacological management of asthma. Data Source State of Florida Medicaid data will be used for analysis in this study. This section focuses on the design and scope of the Florida Medicaid Claims data, and limit ations of the database used for this study.
45 State of Florida Claims Data State of Florida Medicaid Claims Data represents the database for Medicaid claims from the State of Florida, which managed by the Agency for Health Care Administration (AHCA). Data a nalyzed for this study were extracted from the State of Florida Medicaid database. Data collected by AHCA resulted from the transaction of Medicaid patients generating a claim. Therefore, there was no complex sampling for these data. All collected data is stored and maintained by AHCA. All specific health care visit data were extracted from all children within the specified age range. This process allows for comparison of health care use patterns across all children with and without the specified mental health conditions (e.g., anxiety, depression, ADHD). This existing administrative data comes from past stored Medicaid claims records. The variables extracted for analysis include gender, race/ethnicity, date of birth, diagnosis/treatment codes and dates, medications prescribed and dates prescribed, dollar amount paid by Medicaid for the medical visits/treatments, insurance provider, scrambled Medicaid ID number which is de identified The extracted files were sent from AHCA to the Florida Center for Medicai d and the Uninsured. AHCA generated an eligibility file, inpatient file, outpatient file, pharmacy file, and medical physician file. De -identified data. For the purposes of this study, the data set provided by AHCA contained no patient identifiers, as these data have been de-identified with only a unique, scrambled Medicaid ID number t hat is used for purposes to match all five claims files.
46 Inclusion Crit eria The Florida Medicaid claims that were received from AHCA include all eligibility and claims files from February 1, 2001 thru January 31, 2005 for children and adolescents aged 5 to 15 years as of February 1, 2001 that were enrolled in the Florida Medi caid Program. These claims data were then processed based on the following inclusion criteria. Continuous enrollment criteria: Enrolled in Florida Medicaid during the 36 month period with no more than 62 days of eligibility per lapse in coverage during the study period of January 1, 2002 to December 31, 2004. Multiple eligibility gaps of less than 62 days is permitted Asthma criteria: Inpatient, outpatient, and medical physician claims file contained at least 2 proxy asthma diagnosis claims via ICD -9 code 493.x in the year 2002. Pervasive developmental disorder criteria: Inpatient, outpatient, and medical physician claims file contained no more than 1 diagnosis of pervasive developmental disorders (ICD -9 code:299.x) over a 3 year period were not included in this study. One diagnosis of PDD is allowed over a 3 year period to reduce excluding those with a false positive diagnosis for PDD. Continuous coverage criteria. Continuous coverage conveys important protections and is strongly associated with utilization and access to health care services (Honberg, McPherson, Strickland, Gage, & Newacheck 2005). F or sample criteria of continuous coverage to be met, no enrollee must have more than a 62 day enrollment gap between enrollment periods. Initial enrollees pulled from the claims if they met the age criteria (5 to 15 years), and were enrolled in Florida Medicaid between February 1, 2001 and January 31, 2005. The analytical eligible file was all those eligible Medicaid enrollees who had a start date from 2001 to 2004, and an eligible end date from 2002 to 2005. The total n umber of enrollees that met criteria was 7 3,914 enrollees Enrollees were then assessed for gaps in coverage from January 1, 2002 to December 31, 2004 First eligibility gaps were
47 calculated from Jan uary 1, 2002 eligibility. Intermittent enrollment gaps were calculated by taking the difference between the last enrollment end date (plus one) from the next enrollment start date with a difference of 0 respecting no gap in coverage. If any eligibility gap was larger than 62 days, that enrollee was flagged as not being continuously enrolled. This allows for multiple lapses in coverage under 62 days at a time. Additionally exclusion flags were creat ed for those who started their first eligible begin date after March 3, 2002. Another exclusion flag was created for those with last eligibility end date before October 31, 2004 representing 62 day earliest end latest dates for eligibility. Based on the 62 day gap rule and the two exclusion beginning and end dates, 38,607 enrollees had at least one 62 day gap in eligibility and were excluded. Therefore, 35, 307 enrollees met the definition of continuous coverage. Asthma criteria. This study is analyzing Medi caid youth with asthma. In order to truly analyze the effect of mental health in youth with asthma, a one year cohort of youth with asthma had to be constructed. The inclusion would have to be enrollees that have at lea st two ICD9 diagn oses of asthma (Tab le 3.1) within the first year of the study (January 1, 2002 thr o u gh December 31, 2002) These asthma diagnoses would be counted for the first year in the outpatient, inpatient, and medical physician claims database. Of the remaining 3 5,307 continuous cover ed enrollees those without at least 2 ICD 9 diagnosis for asthma during the first year of the study were excluded from the sample (n= 26973 ). Therefore, 8, 334 enrollees had at least 2 ICD 9 diagnoses of asthma and were continuously covered on Medicaid. Pervasive developmental disorder To be included in the final sample, the 8, 334 r emaining enrollees that met both the inclusion definition of asthma and continuous
48 enrollment definition were evaluated for pervasive developmental delay as these enrollees require special services that would make the interpretability of mental health diagnosis difficult If they had more than 1 ICD -9 diagnosi s code for pervasive developmental disorders (e.g., autism ), they were excluded from the study (n=93) The rationale to not exclude all pervasive developmental disorders claims could be due to codes relating to testing or false positives. There were only 38 members of the sample left in the final eligibility group that had one ICD -9 diagnosis code of 99.xx (Table 3.1) ov er a 3 year p e riod. Therefore the total sample is now 8,241 youth with asthma that are continuously covered over the 36. Final eligible sample. The final sample contains 8,2 4 1 enrollees that had at least 2 diagnos es of asthma in the claims data that were c ontinuously enro lled over the three year period without more than 1 diagnosis of pervasive developmental disorder during the 3 years pe riod. This sample represents the eligible analytic sample Analytical working sample. For those 8, 241 enrollees that met c riteria, they were expanded into 36 personmonth observations, and explanatory variables and dependent variables were created from the each of the claims file. Since all subjects have been enrolled in Medicaid all 3 years, there is not a need to control for length of enrollment in the Medicaid Program. Description of Outcome, Explanatory, and Control Variables Since this study aims to assess the impact of mental health diagnosis and related treatment on asthma related health services, outcome measures are limited to claims for all ast hma -related services and expenditures. All dependent variables of utilization and expenditure were created using 3 6 person months of Medicaid c laims data and therefore, total expenditure and volume
49 data is accrue d over one month for 36 observation months from January 1, 2002 until December 31, 2004. For purposes of longitudinal analysis, monthly assessment is then ranked 1 to 36 corresponding with the 36 months personmonths between January 1, 2002 and Decem ber 31, 2004. Asthma-R elated Expenditures Total monthly asthma healt h care expenditure represents an approach to quantify all asthma health care utilization and total asthma -related expense of care. Expenditures for Florida Medicaid claims are paid by the State of Florida. Expenditures or expenses in the claims database are direct payments to Medicaid pr oviders by Florida Medicaid. E xpenditure variables were constructed from the event files for inpatient, outpatient, medical/physician, and pharmacy Total accrued person month expenditure was created by summing all individual asthm a specific expenditures for each individual unique scrambled Me dicaid ID by month (Table 3.2 ). Total asthma expenditure. Accrued monthly health care expenditure s were calculated from the claims data to create 36 personmonth entries for total monthly asthma-related expenditure variables from outpatient, inpatient, medical/physician, and pharmacy expenditures related to asthma. Two variables for asthma expenditure were created bas ed on different pharmacy claims for asthma di rect and asthma indirect expenditure s (Table 3.2) Volume Data Asthma-related outpatient visits, urgent care visits, pharmacy claims for total, direct, indirect, rescuer, and controller asthma medications inpat ient admissions, inpatient length of stay and average length of stay that occurred over the defined monthly periods were calculated from the claims data (Table 3 2)
50 Outpatient visits Outpatient visits relate to all claims that were accrued for each mont h from the outpatient file linked to specifically to the ICD -9 code for asthma. Urgent care v isits. Urgent care visits were defined from the outpatient claims file as claims with an E visit flag. T he E Flag represents urgent care on an outpatient basis Pharmacy c laims Pharmacy c laims relate to all pharmacy related claims accrued for each time period from the pharmacy file. Specific asthma and m ental health related claims were defined thro ugh specific National Drug Code ( NDC) and therapeutic class codes defined in Table 3 3 Pharmacy claims were defined as asthma -related total, direct, indirect, controller, and rescuer pharmacy claims. Direct asthma medications are considered all rescuer and controller medications. Re scuer and controller medication classifications help to understand utilization of mild asthma (e.g., reliance of rescuer medications) and moderate to severe asthma (e.g., reliance on controller medications) Indirect asthma pharmaceutical claims represent a proxy outcome measure of asthm a exacerbations for out of control asthma, cough and colds, allergic rhinitis, and inflammation. Assessing the effect of indirect asthma -related claims helps to understand the association of pharmacy claims related to treating allergic and nonallergic rhi nitis, a condition that is common among yout h with asthma. The National Committee on Quality Assurance provided NDC codes for all measures of asthma related medications (NCQA, 2010) and was used to create the different pharmacy claim variables (see Table 3 3) Monthly l ength of s tay and average monthly length of stay The monthly length of stay variable was created using a span of days using the beginning and end
51 date of servi ce values for length of inpatient care accrued each person-month this is same as monthly accrued inpatient file coverage days The average monthly length of stay represents total length of stay days in a month divided by the number of admissions. Admissions. To understand inpatient asthma -related volume of services, a ccrued monthly a sthma -related admissions were counted. Mental Health T ime Effect Trigger for mental health diagnosis. Over the 36 personmonths flags for ADHD, anxiety, and depression w ere created for the month that a diagnosis code for each condition was entered in the inpatient, outpatient, medical physician, or pharmacy claim files (Table 3-1) For each condition these diagnosis flags triggered subsequent months to be classified with that specific mental health condition (e.g., depression). In otherwords, t he first m onth that ADHD, anxiety, or depression claims is found over the 36 month period, or at the first incidence of AD HD, anxiety, and depression, that individual is flagged at that moment as having the respective condition for the remainder of the months. Hence the first diagnosis of depression, anxiety, or ADHD acts as a trigger in the matrix and will be used as the primary explanatory variables to model to understand the association between time effects and the ment al health condition with asthma-related serv ices There was no minimum number of months required to be flagged with an ADHD, anxiety, or depression diagnosis. Mental Health Treatment Flag for mental health pharmacological treatment. Over the 36 person months of data mental health treatment flags fo r A DHD, anxiety, and depression were created for any person month where utilization attributed to each mental health c ondition was
52 occurred, as described in Table 3-3. A depression, anxiety, and ADHD flag was created for treatment indicating treatment was given in that month. Ch r onicity of mental health treatment. Over the 36 month period it is possible for many to not adhere to their mental health treatment, and a meas ure of treatment chronicity is needed. The definition of treatment chronicity for this study is a count of continuous months of mental health treatment that would restart at 0 if the child goes 3 months goes without a prescription fill for the specific mental health condition. Mental Health and Pharmacological Counseling. Defined by ICD 9 codes of V66.3 and V67.3, and CPT codes of 90862, 90816, 90818, 90821, 90846, 90847, 90849, 96152, 96153, and 96154 to account for patients receiving some extra mental health or pharmacological cons ult during their medical experience This variabl e captures counseling through either physician support or physician drug maintenance with their mental health condition. Control Variables Observed differences in the change of asthma related utilization and expenditure as the presence of mental health com orbidities could be associated with other individual characteristics (e.g., age ). C ontrol variables, therefore, assist in adjusting for r ange of factors known to affect health care utilization such as age, gender, race/ethnicity, and supplemental security income (SSI) status. While other control factors implicated in asthma, such as urban/rural characteristics or air quality index (AQI) may impact asthma related utilization and expenditure through increased reports of environmental induced asthma (e.g., h igh pollen). However, this study seeks to understand the impact of mental health conditions and its treatment on asthma -related utilization and expenditure. Therefore, omission of
53 environmental triggers may be a source of omitted variable bias in these analyses yet should not impact the general relationship of co occurring mental health and treatment on asthma-related utilization and expenditure. The expectation is that environmental causes of asthma should neither impact mental health nor mental health treatment, which is the purpose for this study. Future analyses may look to assess the impact of the environment on the magnitude of the association. Demographic characteristics. Since Medicaid beneficiaries will be between age 5 years and 15 years, a discrete variable controlling for age at the beginning of each person month is used. A measure of supplemental security income ( SSI ) status controls for significant disability, morbidity, and complex care, and will be controlled for as a dichotomous variable of SSI and non-SSI status. To account for differences in health utilization and expenditure by sex, a dichotomous variable will be used to categorize male and female. Race and ethnicity has also been used to control for different health service utilization an d expenditure patterns among groups, and race/ethnicity will be categorized into Caucasian, African American, Hispanic, and Other. Study Design and Analytic Plan Three years of claims data from January 1, 2002 until December 31, 2004 will be used to asses s the objectives of this study. According to Shadish, Cook, and Campbell (2002), an interrupted -short time series design with control groups improves th e strength of interpretability. Longitudinal data represents the best way to assess the impact of mental health comorbidity and treatment on asthma-related use and expenditure s over time. The primary analysis will use panel data methods to determine causation between mental health and asthma. Specifically, fixed and random effect model s for utilization
54 will be used to estimate impact of mental health and treatment effects from the panel data (Wooldridge, 2006). For utilization data, the majority of analyses will used a fixed effect negative binomial regression to assess asthma specific utilization. However e xpenditure s will be addressed through a variety of generalized estimating equations (GEE) with an unstructured correlation to allow for autocorrelation that tends to be common with panel data. This sort of analysis on predicting expenditure data is not a f ixed effect model, as it considers population averages over the panel to model differences in expenditure. Below is an example of deriving and fitting a fixed effect model to demonstrate the analytical approach to model utilization and expenditures While not the only method used to predict the impact of asthma -related outcomes from control and explanatory variables in this study, represents one main analytical approach to model utilization. Fixed Effect Model The fixedeffects model is = + + + (3 -1) F rom which it follows that = + + + (3 -2) W here are with averages of within i Subt racting ( 3 -2) from ( 3 1), to obtain = + (3 -3) From ( 3 -3) remains unestimated in this formula. = + + + (3 -4) W here are the grand averages of
55 For instance, = Summing ( 3 -3) and ( 3 4) to obtain, + = + + + + + (3 -5) Fixed effect models estimate the constraint v = 0, such that model estimates, + = + + + However, this FE model will estimate fixed parameters. Fixedeffects regression produces consistent coefficient estimates and standard errors as ordinary regression when indicator (dummy) variables are included for each of the groups. Since the fixedeffects model is ( 3 -1) and vi are fixed parameters to be estimated, this is the same as = + 1+ 2+ (3 -6) FE model of mental health as modifier of asthma services Using equation ( 3 6) to parameterize the equation such that, = + + (3 -7) Consider a model with a dependent variable of asthma-related utilization and expenditure, of individual i, in panel month j = + + + + + + + + + + + (3 -8) This example model shows the basic parameterized fixed effect model ( 3 -8) desc ribing each mental health diagnosis (i.e., ADHD anxiety, and depression) as a modifier in the delivery of asthma related services. F o r each mental health (i.e., ADHD,
56 anxiety, and depression) impact on asthma related services, interpretation is read through each mental health coefficient. However, in order to obtain a causal association, adding condition*month interaction terms, allows for the temporal association between each condition and the person month to be assessed. This condition*month interaction represents the number of months of having the mental health condition. = + + + + + + + + + + + + + (3 -9) Using such a model ( 3 -9), allows specific beta coefficients to determine the magnitude and fixed effects of ADHD, anxiety, and depression on asthma-related use and expenditure Th e time related effects or the effect of the number of months having the specific mental health condition, will be captured by the interaction terms between ADHD*Month, Anxiety*Month, and Depression*Month. The fixed effects for the control variables are represented through gamma coefficients age, sex, race/ethnicity, and SSI status, which are controlled in this model. To answer the first hypothesis, model ( 3 8) or model ( 3 -9) helps in understanding the men tal health role as a modifier of asthma service However, with model ( 3 9), analysis for each condition must take into account the number of months one has the specific mental health condition The overall impact for each condition is addressed by jointly examining the significance for that condition as well as the respective coefficient on the interaction with time to fully assess th e utilization effect of having the mental health condition (i.e., is time an effect modifier) In depression, for example, to assess
57 depression significance and magnitude, one would have to use the model without the time interaction (3 8) However, the model containing the time interaction (3 -9) for depression, for example, the overall effect of depression would need to be asses sed jointly ( and ) as the effect of depression on utilization is split into both and and cannot be assessed independently for overall depression effect To answer the second hypothesis as to the temporal relationship between asthma service use and each mental health condition, the significance for the temporal interaction, specifically the coefficients associated with the mental health condition as well as the interaction of month and mental health condition must be considered. The question is essentially does the number of months with a mental health condition m odify the effect of depression? Well to determine the answer, first the significance and magnitude of both and would need to be assessed. First would represent the numbe r of months of depression. The significance of the magnitude and directionality of this effect would describe depression trends over the 36 person month period Whereas, does not assess the complete effect of time rather the general effect of depression on utilization not attributed to time. Hence the need to assess both and to consider the total effect of depression on specific asthma service utilization for both the time effect of depression as well as the general e ffect of depression utilization and requires caution when talking about one without the other Generally, the overall impact of depression (depression both with and without time effects) are contained and assessed in model 3-8. Model 3-9 allows for underst anding the effects of time on asthma related utilization attributed to depression. For example, if the depression*month coefficient ( ) has an incidence rate ratio over 1 with a pvalue <0.05, this would
58 suggest significant positive utilization for those with depression over time. Then its important to assess the magnitude and significance of the relationship for to fully understand the effect of the time. If the significance and directionality change, then time is an effect modifier and must be taken into account to fully understand the relationship of depression. I n estimating t he impact of time on asthma related use in comorbid anxiety this would be done by assessing and For ADHD, the impact of time on asthma related use for those with comorbid ADHD would be done by assessing and Again this approach only attributes utilization to the mental health condition after the mental health diagnos is occurs in the claims data Therefore, before a mental health diagnosis is obtained in the c laims, this does not count toward the effect. This provides better evidence for a causal effect. Thus, causality on asthma services and each mental health condition can be determined. However, caution is needed on causal inference due to limitations. Such limitations include false negatives in claims data for mental health condition identification. For example, those with ADHD, anxiety, and depression may have had claims prior to January 1, 2002 which would not appear in the claims data until after treated by the doctor and coded in the claims file Also those with ADHD, anxiety or depression tend to be underreported in claims data (Spettell et al., 2003) FE M odel with mental health t reatment as a mediator of mental health in asthma services Using equation (3 8 ) further parameterization of the equation is required to allow for new dummy variables were added for months of any anxiety, depression, and ADHD treatment.
59 Consider a fixed effects model with a dependent variable of asthma-related utilization and ex penditure, of individual i in panel month j, dummy variables for any pharmacological specific month treatment for ADHD, anxiety, and depression was added to the equation ( 3 10 ), as well as any pharmacological consult/ psychosocial counseling (MHC) = + + + + + + + + + + + + + + (3 -10) For the next model, however, an additional variable per mental health treatment was created. This variable assesses the impact of consecutive mental health treatment. Mental health treatment laps of 3 months or more will reset to 0 Hence this variable repr esents the number of months of consecuti ve mental health treatment by attempting to control for mental health chronicity. Therefore, model 3 -11 represents these 3 added variables. = + + + + + + + + + + + + + + + + + + (3 -11) T o answer the third hypothesis, we must analyze each mental health condition from model ( 3 -8 ) and ( 3 11) independently for the treatment effect on asthma-related utilization and expenditure and assess the respective coefficient on the interaction with time to fully understand the expenditure/use association of having treatment on the mental health condition. In depression treatment, for example, examining and
60 will give an understanding of how co morbid depression treatment impacts asthmarelated us e and expenditur e. In u nderstanding the impact of anxiety treatmen t, this would be done by examining and For ADHD treatment, examining and would allow for the assessment of the impact on asthma related services While the assumption of having consistent treatment is to imply controlled mental health, it may provide a good understanding of how regular mental health treatment impacts asthmarel ated services The counseling variable allows for the assessment of the presence of mental health counseling and mental health pharmacological management impacts utilization. Fixed Effects Model Assumptions Using a FEM, certain assumptions must be stated. First for each i, the model is, = + + + where each beta coefficient represent parameters to estimate and is the unobserved effect. Next, the assumption must be made that the data represents a random sample from the cross section. Third, each explanatory variable changes over time (for at least some i), and no perfect linear relationship exist among the explanatory variables. Fourth, for each time period, the expected value of the idiosyncratic error given the explanatory variables in all time periods and unobserved effect is zero ( i.e., exogeneity assumption). These four assumptions make the fixed effect model estimator unbiased. Also, the Var ( | ) = Var( ) = for all t =1,,T (Woodridge, 2006) While fixed effects was used as an example to illustrate modeling the impact of mental health and mental health t reatment on asthma -related utilization and expenditure random effect models may also be used in the event that a fixed effect
61 model will not converge. Random effect assumes the above four assumptions plus additional requirement that the unobserved heterogeneity, is independent of all explanatory variables at all time periods. If is correlated, then fixed effects is preferred. Random effects does not allow for arbitrary correlation between the unobserved het erogeneity, and key explanatory variables, Fixed effects tend to be preferable as it allows for estimating ceteris paribus effects, or effects that stay the same. However both have consistent estimates (Woodridge, 2006). GEE Modeling Using general estimating equation method allows for the fitting of populationaveraged panel -data models, which allows a great deal of flexibility in fitting models. In STATA the option is xtgee, which fits general linear models and allows for the specification of within group correlation structure for the panels. The xtgee command in STATA allow for the specifying a distribution and link function and within -group correlation. For the volume analyses, for example, a negative binomial distribution and link functio n with unstructured correlation was specified. This would represent a population averaged model, which is not the same as what would be observed in fixed effect models. For expenditure data, specifying a gamma distribution and log link function with unstructured correlation. For predicting expenditure using a logistic regression, the xtgee model specifying a binomial distribution with a link function of logit to obtain the natural log of the odds ln ( y /(1 -y )). For each of these models, predicted estimates a nd standard errors were obtained. The autocorrelation/autoregressive nature of panel data may lead to some problems with certain within-group correlation structures. Hence, autocorrelation is problematic with time series data when the error is
62 correlated o ver time, and testing for serial correlation is needed AR(q). Adjustment through correlation specification would improve autoregressive nature of time series data (Wooldridge, 2006). There are a variety of different specifications of correlation available with xtgee in STATA. For example, use of the exchangeable, or compound symmetry, as the working correlation matrix with measured data and identity link function is equivalent to using a random effects model with a random intercept per cluster (Horton & Lipsitz, 1999). For small observations per cluster with a balanced and complete design, an unstructured matrix works quite well, yet for those with clustered observations, a lack of logical ordering for observations within a cluster may suggest an exchangeabl e correlation matrix (Horton & Lipsitz, 1999). Additionally specifying autoregressive (AR) correlation by the order number is allowed. The nature of 36 person months suggests 36 observations per cluster, may suggest selection of either an exchangeable, AR( #) or unstructured correlation matrix depending on the specific dependent variable. GEE Model Assumptions Unlike FEM for utilization/count data, GEE Models were used in constructing all logistic regression (if necessary) and gamma loglink regressions for all expenditure data Two part models were used for expenditure data with a high number of zero dollar expenditure Hence, two part models were constructed for inpatient and outpatient expenditures Asthma expenditure modeling would be adjusted based on amount of kurtosis. Manning and Mullahy (2001 ) suggests that kurtosis score over 4 would use a log -linear approach to estimating asthma expenditure While a score under 4 would indicate using
63 a gamma distribution approach. All expenditure variables had a kurt osis score of 4 or less to use a gamma famil y distribution with log link using STATA (StataCorp, 2007) Institutional Review Board Approval. This study received University of Florida Institutional Review Board (IRB) approval on July 8, 2009 (#236-2009). Software All data programming was done using SAS 9 ( SAS Institute Inc., Cary, NC), and all model estimates were generated using Stata 10.0 (StataCorp., 2007 ).
64 Table 3 1. I CD-9 codes used to select asthma, depression, anxiety and ADHD Heath Condition ICD 9 Codes Asthma 493.xx Allergic Rhinitis 477.xx Depression 296.2, 296.20, 296.21, 296.22, 296.23, 296.24, 296.25, 296.26, 296.3, 296.30, 296.31, 296.32, 296.33, 296.34, 296.35, 296.36, 296.9, 296.90, 296.99, 298.0, 293.83, 309.0, 309.1, 300.4, 311, 313.1 Anxiety 300.0, 300.00, 300.02, 300.09, 300.1, 300.10, 300.11, 300.12, 300.13, 300.14, 300.15, 300.16, 300.19, 300.2, 300.20, 300.21, 300.22, 300.23, 300.29, 300.3, 300.5, 293.84, 313.0 ADHD 314.0, 314.00, 314.01 PDD 299.xx ICD -9 codes used in defining the different health conditions used in this study.
65 Table 3 2 Summary of monthly outcome measures Asthma measures Expenditure Total expenditure 1 using indirect asthma medications Total expenditure 2 using direct asthma medications Total asthma related inpatient expenditure s Total asthma related outpatient expenditure s Total asthma related medical physician expenditure s Total asthma related pharmacy expendi ture s Volume Number of asthma related admissions Number of asthma related length of stay Number of asthma related outpatient visits Number of asthma related urgent care visits Number of asthma related medical claims Number of asthma related total pharmacy claims Number of asthma related direct pharmacy claims Number of asthma related indirect pharmacy claims Number of asthma related rescuer medications Number of asthma related controller medications Dependent variables list u sed in this study.
66 Table 3 3 P harmacy measures defined by Therapeutic Class or National Drug Code Pharmacy measures TCC HEDIS NDC files from NCQA Asthma Treatment Rescuer J5D, J5A, J5F, A1D, J2B SABA from asthma HEDIS Controler A1B, P5A, J5J, J5G, Z4B, J5D, Z2F, Z2L Anticholinergics HEDIS Asthma HEDIS w/o SABA Direct Asthma A1B, P5A, J5J, J5G, Z4B, J5D, Z2F, Z2L, J5A, J5F, A1D, J2B Anticholinergics HEDIS Asthma HE DIS Indirect asthma: or allergic rhinitis Q7A, Q7C, Q7E, Q7P, Q7W, B3A, B3J, B3K, B3R, B3T, B3X, B4Q, B4R, B4W, H6A, Z2A, H3A, H3E, H3F Anticholinergics HEDIS Pharyngeal antibiotics HEDIS ADHD Treatment H2A, H2V, H7Y ADD/ADHD HEDIS Anxiety Treatment H2D, H2E. H2F, H7C, H2U, H2S, H7J, J7C Depression Treatment H2U, H7B, H7N, H7D, H2S, H7C, H7E, H2W, H2X, H7Z, H7J Depression HEDIS TCC=Therapeutic class code; NDC = National Drug Code; NCQA HEDIS retrieved from http://www.ncqa.org/tabid/1091/Default.aspx
67 CHAPTER 4 RESULTS Overview The beginning of this section describes the sample characteristics and average monthly utilization and expenditure data for the sample of Medicaid youth with asthma. The second part of this section describes the model s and estimates for negative binomial r egression using fixed effects random effects, or population averag ed models. The third part of this results section analyzes expenditures using population averaged models. For both volume and expenditure analyses, four models will be used to assess the association between mental health and mental health treatment on asthma-related service use and expenditure. Spe cifically, the four models: Model 1 tests whether ADHD, anxiety, and depression independen tly act as modifier s of asthma -related use. Model 2 tests whether months with the specific mental health diagnosis modifies the impact of ADHD anxiety, and depression on asthma -related use and expenditure using the addition of monthmental health condition interaction term s added to model 1. Model 3 tests whether presence of mental h ealth treatment mediate s the association of ADHD, anx iety, and depression with asthma -related service use and expenditu re Model 4 adds to model 3 by including indicators for the number of month s of ADHD, anxiety, and depression treatment to assess whether continuous mental health treatment has an additive effect on asthma-related service use and expenditure Sample Characteristics Sample characteristics are reported in Table 4-1. The sample mean age was 11.2 years (s.d.=3.03) and consisted of more boys (55.7%) tha n girls. Also, the sample was racially and ethnically diverse, having 30.9% black, 25.7% white, 24% Hispanic, and
68 19.4% other Approximately 30% of the sample had supplemental security i ncome (SSI). Over the three year period, about 1 of 10 youth with asthma had depression (11.4%), 18.5% had ADHD, and 3.5% of youth with asthma had anxiety diagnosis. Over a third of the youth with asthma had allergic rhinitis over the 3 year period. Lastly only 5.6% had received pharmacological consult or physician counseling over the three year period. Average Monthly Expenditure The average monthly expenditure for the sample of Medicaid youth with asthma are found in Table 4 -2. Relative to those without ADHD youth with comorbid ADHD had less average monthly inpatient expenditures ($23.68 vs. $25.67), yet more pharmacy ($61.57 vs $53.96) and total asthma expenditures ($108.39 vs.103.83). Compared to those without anxiety youth with comorbid anxiety had higher observed average monthly asthma -related total expenditures ($150.86 vs. $97.73), direct expenditures ($128.85 vs. $82.28), pharmacy expenditures ($75.69 vs. $52.27), and inpatient expenditures ($48.71 vs. $21.83). Lastly average monthly expenditures for those with depression were higher for inpatient ($44.23 vs. $22.35), pharmacy ($61.43 vs. 54.35), direct ($131.22 vs. $100.46) and total asthma ($112.66 vs. $84.52) expenditures compared to those without depression. Average Monthly Utilization The obs erved average monthly asthma-related volume data for the sample can be found on Table 43. Compared to those without ADHD, youth with comorbid ADHD had fewer observed average monthly admissions (.0058 vs. .0064), fewer outpatient visits (.034 vs.038 ), fewer urgent care visits (.022 vs.024), and fewer medical claims (.28 vs. .32). However, those with comorbid ADHD had more total asth ma -related pharmacy
69 claims (.97 vs. .86 ) compared to those without ADHD Compared to those without anxiety those with comorbid anxiety more average monthly admissions (.009 vs. .006), longer monthly length of stay (4.5 vs. 3.3), longer average monthly length of stay (4.4 vs. 3.1), more medical claims (.323 vs. .316), and more pharmacy claims (1.11 vs. 0.84). Compared to those wit hout depression, those with comorbid depression was associated with more admissions (.009 vs. .006 ), longer length of stay (4.5 vs.3.3), longer average length of stay (4.3 vs. 3.1) and more total pharmacy claims (.98 vs. .86 ). However, depression was assoc iated with fewer urgent visits (.23 vs .24) and fewer medical claims (.27 vs. .32 ) compared to those without. T he observed average pharmacy utilization is found in Table 4 3. For those with ADHD, there seemed to be more monthly direct (.61 vs. .57) and in direct (.42 vs. .36) pharmacy claims, as well as more controller claims (.35 vs. .30) relative to those without ADHD. For tho se with anxiety, there were more average monthly direct (.70 vs. .56) and indirect asthma ( .49 vs. .35) related pharmacy claims, as well as more rescuer (.30 vs. .23) and controller (.37 vs. .30) medication use relative to those without an xiety. Depression had more average monthly direct (.60 vs. .58) indirect (.45 vs. .35) and controller (.34 vs. .30) medication use compared to tho se without depression. Negative Binomial Regression Prediction of u tilization among youth with asthma was modeled through negative binomial regression. N egative binomial regression was fit for inpatient admissions, length of stay, outpatient and urgent ca re visits, and medical and pharmacy claims for yo uth with asthma on Medicaid. All 4 models described above were used to predict utilization controlling for age, sex, race/ethnicity, and SSI status. The majority of these models used a fixed effects analysis with the exception of inpatient admissions and
70 length of stay which used random effects Presentation of the results will be through two tables. The first table presents incident rate ratios (IRR) obtained through the fitted model. The second table presents the predicted mean estimates from the model. Inpatient Utilization A fixed effects negative binomial model with random effects was used to construct IRRs (Table 44) and predict average monthly asthma related admissions (Table 45). With respect to mo nthly asthma-related admissions the first model shows a statistically significant 33% (p<0.01) and 43% (p<0.01) increase in admissions for those with anxiety and depression, respectively compared to those without. However, those without ADHD had 27 % more admissions compared to those with ADHD. Therefore, these results support the first hypothesis suggesting mental health diagnosis is associated with the utilization of asthma services. Specifically, these results support the hypot hesis of an increase in admissions use for anxiety and depression, yet a decrease in service use for those with ADHD. There were no significant mental health conditionmonth interactions observed in model 2, suggesting the number of months a person has any of the mental healt h diagnose s does not modify the association with a sthma related inpatient admissions Model s 3 and 4 suggest mental health treatment diminishes the effect of all mental health conditions on asthma related hospital admissions. Specifically for overall impac t of treatment m odel 3 finds that youth with asthma receiving anxiety treatment had 47% (p<0.01) more inpatie nt admissions than those not receiving treatment Where no main ADHD treatment effect was observed model 4 finds t hat continuous treatment of ADH D was associated with asthma related admissions Specifically the rate of asthma -related admissions
71 decreased for every additional month of continuous ADHD treatment (IRR=.973, p<0.01) A mi xed effect (both fixed and random effect) negative binomial model examining monthly length of stay for those with inpatient admissions was used to construct IRR s (Table 46) and predict average monthly length of stay (Table 4-7). Addressing length of stay for those with inpatient admissions, the first model shows a significant 29% longer monthly length of stay (LOS) for those with anxiety compared to those without (3.95 vs. 3.05 (this notation represents those predicted estimates for average monthly asthmarelated length of stay for anxiety being 3.95 days versus 3.0 5 days for those without, and this notation will be used throughout the results section ); p<0.0 1) There were no significant mental health conditionmonth interactions observed in model 2, suggesting the number of months a person has a mental health diagno sis does not have an association with asthma related monthly length of stay. Models 3 and 4 suggests that mental health treatment did not diminish the effect of anxiety for asthma -related LOS as anxiety maintained an elevated association with length of st ay ( Model 3: IRR=1.28, p<0.01; Model 4: IRR=1.26, p<0.01) compared to those without anxiety N o main treatment effects were associated with length of stay. Medical claims Results for the fixed effects negative binomial f or the medical claims are found on T ables 48 (IRR) & 4 -9 (predicted estimates) M odel 1 illustrates youth with asthma that have ADHD (.26 vs. .23 (this notation represents predicted medical claims estimates for ADHD vs. those without ADHD and will be used throughout the results section) o r anxiety (.26 vs. .23) had an 11% and 13% significant increase in medical claims respectively, compared to those without anxiety
72 Depression, however, had fewer incident asthma-related medical claims (IRR = .900, pvalue<0.001). Unlike anxiety, utilizati on for those with depression and ADHD were different than hypothesized. Model 2 suggests that the number of months with comorbid ADHD or depression diagnosis was associated with incident asthma related medical claims Namely, the rate of medical claims inc reased by 0.3% (p<0.05) for every additional month of having a diagnosis of ADHD However, the number of months with depression was associated with fewer incident claims (IRR=.991) While the ADHD*month and depression*month term s were significant, the individual mental health disorders lost significance implying the number of months with the mental health condition adjusted away the main overall effect observed in model 1 Model 3 was assessed for the effects of mental health treatment. Youth with asthma with depression had fewer medical claims (IRR=.842, p<0.01) compared to those without depression, whereas those with anxiety had 8% (p<0.01) more medical claims compared to those without anxiety Overall, those youth with asthma on ADHD anxiety, and depre ssion treatment had 23% (p<0.01) 10% (p<0.05), and 17% (p<0.01) more incident medical claims, respectively, compared to those not taking those medications. Overall diminished effects observed with treatment models (model 3 and model 4) suggest that those receiving treatment for anxiety and ADHD use medical claims similarly to those without the condition whereas those receiving treatment for depression had resulted in fewer incident medical claims than those without depression Fo r continuous treatment eff ects from model 4, the rate of medical claims increased by 1.0% (p<0.05) for every additional month of continuous depression treatment. The main effect for d epression
73 treatment (continuous or dichotomous ) was associated with more asthma-related medical claim s Outpatient visits. Fixed effects negative binomial regression f or outpatient visits can be found on Tables 4 10 (IRR) and 4-11 (predicted estimates) T he first model suggests that those with anxiety had 23% more outpatient visits (.63 vs. .5 2 ) compare d to those without an anxiety. Thereby, suggesting that those with anxiet y have a propensity to have more outpatient experiences compared to those without anxiety. There were no significant mental health conditionmonth interactions observed in model 2, su ggesting the number of months a person has a m ental health diagnosis does not modify the association with asthma related outpatient visits Model 3 showed diminished effects for all 3 mental health conditions. Specific overall treatment effects for depression and anxiety treatment was observed. Youth receiving anxiety treatment had 49% more incident asthma-related outpatient visits (p<0.01) compared to those without. Those youth with asthma receiving depression treatment, however, had fewer asthma related outpatient visits (IRR=.814, p<0.05) compared to those without. Overall mental health treatment diminishes the effect for all 3 mental health condit ions association with asthma-related outpatient visits, such that those who receive d mental health treatment use asthma outpatient services similar to those without the condition. Model 4 did not show any continuous mental health treatment effects Urgent care visits. F ixed effects negative binomial regression models were constructed for urgent care utili z ation, results presented in Tables 412 (IRR) and 413 (predicted estimates) The first model suggests youth with anxiety had 33% more asthma-related urg ent care visits (1.6 vs. 1.2, p<0.01) t o those without anxiety
74 Interaction terms for m onths of having any of the 3 mental health di agnoses were not significant in model 2, although some general associations with depression and anxiety were observed ; that is, c ompared to those without anxiety those with anxiety had 51% more urgent care visits (IRR=1.51; 1.5 vs.1.3 p<0.01) when controlling for months with anxiety ; whereas controlling for months of depression, those with depression had incurred significantly fewer urgent care visits (IRR=0.700; .86 vs. 1.3; p<0.05) compared to those without depression. Model 3 suggested that overall depression treatment was associated with fewer urgent care visits (IRR=.654, p<0.01), where as anxiety treatment w as associated with 77% more asthma -related urgent care visits (p<0.01). Model 4 suggested no appreciable continuous mental health treatment effects. While no significant relationship was observed for depression, anxiety, and ADHD, treatment diminishes the effect (e.g., IRR magnitude) of the mental health conditions association with asthma -related urgent care to be similar to those without the respective mental health condition Total monthly a sthma pharmacy claims. Total monthly a sthma -related pharmaceutic a l claims predicted estimates and incident rate ratios from the fixed effects negative binomial are found in Table 4 14 (IRR) and Table 4 -15 (predicted estimates) Model 1 shows that youth with ADHD and anxiety have 19% ( 1.6 vs. 1.4 p<0.01) and 37% ( 1.9 vs 1.4 p<0.01) more monthly total asthma-related fills than those without the respective conditions whereas those with depression had significantly fewer total asthma-rel ated fills (IRR=0.930; 1.3 vs. 1.4, p<0.01) compared to those without depression. Ass essing the number of months with each mental health condition and the association with total asthma fills suggest that there wa s a significant interaction for
75 anxiety*month (p<0.01) as well as depression*month (p<0.01). The rate for total asthma fills increased by 0.4% (p<0.01) for every additional month of having an anxiety diagnosis. The rate for total asthma fills decreased for every additional month of having a depression diagnosis (IRR=.993, p<0.01) Controlling for number of months with depression, those with depression had 7 .4 % (p<0.01) more total asthma fills than those without depression Controlling for number of months with ADHD and anxiety, those with ADHD and those with anxiety had 17% and 27% increase in tot al asthma fills, respectively, compared to those without ADHD and anxiety. Model 3 shows treatment diminishes the effect of ADHD and anxiety on asthma pharmacy claims, as ADHD (IRR=.990 p<0.01) and anxiety (IRR=1.24, p<0.01) suggest that those with ADHD and anxiety receiving treatment had more similar use to those without ADHD and anxiety However those with depression had fewer total fill s, suggesting those with depression used even fewer total asthma fills compared to those without depression (IRR=.828, p<0.01). Those who had received ADHD, anxiety, and depression treatment had 49%, 26%, and 24% more monthly asthma-related total fills compared to those untreated for the respective mental health medication. Additionally having pharmacological counseling w as associated with 24% greater asthma related prescription fills (1.8 vs. 1.4 p<0.01). Model 4 assess continuous treatment effects. Th e rate of total asthma fills increased by 2% (p<0.01) for every additional month o f continuous ADHD treatment. The rate o f total asthma fills increased by 2% (p<0.01) for every additional month of continuous anxiety treatment. Direct asthma pharmacy claims. Fixed effect negative binomial regression models for monthly direct asthma pharmacy claims are presented in Table 4 16 (IRR)
76 and 4 -17 (predicted estimates) Model 1 shows youth with ADHD (IRR=1.15; 2.2 vs 1.9, p<0.01) and anxiety (IRR=1.29, 2.4 vs. 1.9, p<0.01) have significantly more monthly direct asthma prescription fills compared to those without. Whereas, those with depression had signific antly fewer direct asthma fills (IRR=.907; 1.80 vs 1.98 p<0.01) compared to those without. Similar to total asthma pharmacy claims, there was a significant effect associated with months with a mental health condition (AD HD*month (p<0.01), anxiety*month (p<0.01), and depression*month (p<0.01)) and direct asthma pharmaceutical claims The interpretation for ADHD and anxie ty suggest th e number of months wi th ADHD (p<0.01) or anxiety (p<0.01) was associated with more direct a sthma -related prescriptions. The number of months with depression was associated with fewer direct asthma fills (IRR=.991, P<0.01) yet the controlling for months of depression modified the main depression effect to having 9% more direct asthma pharmacy cl aims (IRR 1.09, p<0.01) compared to those without depression. Model 3 s uggests that mental h ealth treatment diminishes the effect for ADHD and anxiety on direct pharmacy claims such that those receiving treatment for ADHD and anxiety fill direct asthma pr escriptions similarly to those without ADHD or anxiety However for depression, the effect of treatment resulted in even lower direct asthma -related prescr iption fills (IRR=.796, p<0.01) compared to those without depression. Those with physician counseling had 16% more direct asthma fills than those who did not receive physician counseling. Model 3 also show that overall t reatment for all mental health conditions were associated with increased direct asthma-related prescripti on fills. Specifically the mean effect of ADHD, anxiety and depression treatment had used 44% (p<0.01) 18% (p<0.01) and 26% (p<0.01) more direct asthma fills, respectively,
77 compared to those not receiving the particular mental health treatment. In model 4, continuous treatment effects were observed. The rate of direct asthma pharmacy claims increased 0.9% (p<0.01) for every additional month of continuous ADHD treatment. The rate of direct asthma pharmacy claims increased 2.0% (p<0.01) for every addition al month of continuous anxiety treatment Indirect asthma pharmacy claims. A fixed effect model negative binomial regression was used for indirect asthma pharmacy fills on Tables 4 -18 (IRR) and 419 (predicted estimates). M odel 1 shows that ADHD and anxiet y were associated with a 17% ( 5.4 vs. 4.6, p<0.01) and 15% ( 5.3 vs. 4.6 p<0.01) increase respectively, in indirect asthma prescription fills compared to those without ADHD or anxiety Model 2 shows that the rate of indirect asthma claims decreased for ev ery month with a depression diagnosis (IRR=.997, p<0.05). Controlling for months with the specific mental health condition, ADHD (IRR=1.17, p<0.01), anxiety (IRR=1.13, p<0.01), and depression (IRR=1.08, p<0.05) had significantly more indirect asthma claims compared to those without ADHD, anxiety, and depression. Model 3 assess the impact of mental health treatment as a mediator of ADHD, anxiety, and depression on indirect asthma prescription fills. Mental health treatment diminishes the effects for all ment al health conditions. Specifically, mental health treatment resulted in fewer indirect asthma prescription fills for those with depression compared to those without depression (IRR=.925; 4.4 vs. 4.8 p<0.01), yet resulted in 8% more indirect fills for those with anxiety compared to those without anxiety (IRR=1.08; 4.9 vs. 4.5, p<0.01). Treatment effects are present, as those receiving ADHD, anxiety and depression treatment w ere associated with 38% (p<0.01), 24% (p<0.01), and 18% (p<0.01) more indire ct asthma
78 prescription fills respectively, compared to those not receiving the particular treatment. Also those having physician counseling had 67% (p<0.01) more indirect asthma-related pharmacy claims compared to those who did not. Model 4 shows continuous treatment effects were observed for ADHD and anxiety treatment. The rate of indirect asthma fills increased by 0.8% (p<0.01) for every additional month of continuous ADHD treatment. The rate of indirect asthma fills increased by 0.9% (p<0.01) for every additional month of continuous anxiety treatment Controller Medications. Tables 4 -20 (IRR) and 421 (predicted estimates) show the GEE models with negative binomial for controller medications. Model 1 suggests that those with ADHD or anxiety had 11% and 10% more controller medication fills respectively, compared to th ose without ADHD (IRR=1.11; .32 vs. .29 p<0.01) or anxiety (IRR=1.10 ; .3 3 vs. .29 p <0.01). Assessing the number of months with a mental health diagnosis from model 2, the rate of controller fills decreased for every additional month of having a depression diagnosis (IRR=.995, p<0.01), while the rate of controller fills increased by 2% (p<0.05) for every additional month with an ADHD diagnosis. Controlling for number of months with an ADHD an d depression diagnosis, those with and ADHD or depression had 6% (p<0.05) and 7% (p<0.05) more controller fills compared to those without ADHD or depression. Model 3 assess es the impact of mental health treatment and contr oller medication use. Mental healt h treatment diminishes the ef fect for those with ADHD and anxiety. Depression had fewer monthly controller fills compared to those without depression, suggesting a further reduction in controller fills (IRR=.884; .256 vs.300; p<0.01). ADHD, anxiety, and de pression treatment had 24% (p<0.01), 12% (p<0.01), and 21% (p<0.01) more asthma controller
79 fills, respectively, compared to those without the particular mental health treatment. Also, those receiving physician pharmacological counseling had a fewer control ler medication fills than those who did not have such counseling (IRR=.905, p<0.05). For model 4, continuous ADHD treatment effects were observed. The rate of controller fills increased by 0.4% (p<0.01) for every additional month of continuous ADHD treatme nt. Rescuer Medications. Rescuer medication modeling usi n g GEE can be found on Tables 4 -22 (IRR) and 4-2 3 (predicted estimates) Model 1 shows the anxiety and depression as modifiers of monthly rescuer medication utilization. Specifically those with anxiety had 21% ( .29 vs. .2 3 p<0.01) more rescuer medication fills compared to those without. Those with depression had fewer mont hly rescuer fills compared to thos e without depression (IRR = 869; .20 vs. .24, p<0.01). Model 2 shows the number of months with depression or anxiety was significantly associated with rescuer fills. Specifically, the number of months with depression was associated with fewer rescuer fills (IRR=.993, p<0.01), where rate of rescuer fills increased by 5% (p<0.01) for e very additional month of anxiety diagnosis. Controlling for number of months with anxiety, those with anxiety ha d 11% more rescuer fills (IRR=1.11, p<0.01) compared to those without anxiety Model 3 shows that those with ADHD and anxiety who receive treatment use asthma services similar to those without ADHD and anxiety Specifically, ADHD and anxiety treatment diminishes the effects fo r ADHD and anxiety on asthmarelated rescuer fills. Those with depression who received treatment use asthma services differe ntly to those without depression, as those with depression had filled fewer rescuer medications compared to those without depression (i.e., difference between model 1 (IRR=.869, p<0.01) and model 3 ( IRR= .812, p<0.01) ). Those
80 receiving ADHD, anxiety, and de pression treatment had 24% (p<0.01) 15% (p<0.01), and 13% (p<0.01) more rescuer medication fills, respectively, compared to those without the particular mental health treatment. M odel 4 showed continuous treatment ef fects for depression and anxiety Speci fically the rate of rescuer medication claims decreased for every additional month of continuous depression treatment (IRR=.993, p<0.05), whereas the rate of rescuer medication claims increased by 2% (p<0.01) for every additional month of continuous anxie ty treatment Summary of Utilization Analyses ADHD and utilization. ADHD modified asthma related utilization. Specifically, ADHD was associated with having fewer admissions yet more medical claims, and total, direct, indirect, and controller asthma-related medication fills. This utilization pattern suggests those with ADH D used more preventive and pharmacy services for their asthma, and had less emergency type care (e.g., admissions) than those without ADHD. Additionally the number of months with ADHD was associated with more medical claims and phar macy claims. ADHD treat ment results in having more medical claims and more pharmacy fills across all categories (total, direct, indirect, rescuer, and controller medications), whereas the rate of admissions decreases for every additional month of continuous ADHD tre atment. Treat ment diminishes the eff ect of ADHD for all asthma utilization categories as hypothesized Anxiety and utilization. Anxiety modified asthma related utilization. Specifically, anxiety was associated with having longer length s of stay, more admissions, urgent care visits, medical claims, outpatient visits, and total, direct, indirect, controller and rescuer asthma -related medication fills. This utilization pattern suggests those with anxiety used more preventive and pharmacy services for their asthma, but als o had
81 more emergency type (e.g., admissions, LOS, urgent care visits) care than those without anxiety. Additionally the number of months with anxiety was associated with more utilization for total, direct, and rescuer pharmacy claims The treatment of anx iety has main effect s associated with more admissions, medical claims, outpatient visits, urgent care visits, and more pharmacy fills across all categories (total, direct, indirect, rescu er, and controller medications) Findings suggest that the treatment of anxiety may have implications for asthma management. Treatment diminishes the effect of anxiety for all asthma utilization categories, as hypothesized. Depression and utilization. Depression modified asthma related utilization. Specifically, depression was associated with having more admissions yet fewer medical claims, and total, direct, indirect, controller and rescuer asthma -related medication fills. This utilization pattern suggests those with depression u sed fewer preventive and pharmacy services for their asthma, and had m ore emergency type care than those without depression Additionally, controll ing for months with depression revealed a n increase in utilization of medical claims and all pharmacy claims categories for those with depression relat ive to those without depression. However, the number of months with a depression diagnosis (i.e., the longer one has depression) was associated with a decrease in medical claims and all pharmacy claims categories. The treatment of depression was associated with more medical claims and more pharmacy fills across all categories (total, direct, indirect, rescuer, and controller medications), yet fewer outpatient visits urgent care visits compared t o those not receiving depression treatment This finding sugges t s some beneficial aspects to depression treatment on asthma-related treatment Treatment of depression did not diminish the effect of
82 depression for all asthma related categories. Specifically, treatment of depression diminished the effect of depression f or asthma related admissions, outpatient visits, urgent care visits, and indirect pharmacy claims. Generalized Estimating Equation Models and Expenditure Estimates Four GEE models with gamma distribution and log link function with v arying correlation struc tures were conducted for each asthma-related expenditure dependent variable to assess the three objectives of this study. Model 1 uses mental health triggers and control variables to assess the modifying impact of the mental health triggers on asthma-relat ed expenditure. Model 2 uses mental health triggers, control variables and number of months with mental health diagnosis (e.g., condition*month interaction ) to assess temporal effects on asthma -related utilization and expenditure. Model s 3 and 4 use mental health triggers, control variables, and mental health treatment variables to assess treatment effects (Model 3 captures overall treatment effects, whereas 4 adds additional continuous m ental health treatment ). Total and direct asthma expenditures Models predicting total and direct asthma expenditures are presented in Table 4 2 4 and predicted estimates in Table 4 2 6 (total expenditures ) and Table 427 (direct expenditures ) For both total and direct asthma expenditures were fit to GEE with gamma distribution and log link function with an unstructured correlation. Total monthly asthma -related expenditures. Model 1 illustrates that compared to those without anxiety, those with a nxiety had incurred 29 .1 % greater m onthly total asthma related expenditures (B =.291; predicted estimates for those with anxiety was $126.85 vs. predicted estimates for those without anxiety was $ 94.80, p<0.01) (Results will be presented from this point on wi th beta estimate, predicted expenditures for those
83 with vs. those without the mental health diagnosis and the specific pvalue) The number of months for each mental health diagnosis was not significantly associated with total asthma expenditures. However cont rolling for the number of months with anxiety and depression, in model 2, revealed a significant 22% and 21% increase in total asthma -related expenditures for those with anxiety ($125.77 vs. $94.81, p<0.05) and depression ($108.18 vs. $98.25, p<0.05), resp ectively, compared to those without anxiety and depression. Model 3 suggests that anxiety and depression treatment had no significant effects on total asthma expenditure s for those with anxiety and depression. However, ADHD treatment resulted in those with ADHD having 10.4 % less total asthma expenditures (B = -.104; $91.37 vs. $101 .37, p<0.05) than those without ADHD. Mainly, mental health treatment diminished the effects of total asthma -related expenditures for all conditions to look more like thos e without the relative mental health condition. T hose who received treatment for ADHD ($114.76 vs. $97.57, p<0.01) anxiety ($130.31 vs. $96.73, p<0.01) and depression ($132.67 vs. $96.74, p<0.01) had incurred 16.2%, 29.8%, and 31.6% greater total asthma expenditures, respectively, compared to those not receiving the particular mental health treatment Model 4 did have a significant continuous anxiety treatment effect, such that the rate of total asthma related expenditures increased by 2.2% (p<0.01).for each additional month of continuous anxiety t reatment Direct monthly asthma -related expenditures. A similar pattern was seen in direct monthly asthma -related expenditures Model 1 found those with anxiety had incurred 30 .4 % greater direct monthly asthma re lated expenditures compared to those without anxiety ($108.52 vs. $80.02, p<0.01). Again, the number of months with a
84 mental health diagnosis was not significantly associated with direct asthma-related expenditure s However, controlling for the number of m onths with an anxiety and depression diagnosis revealed that those with anxiety incurred 24.2% greater monthly direct asthma expenditures compared to those without anxiety (B=.242; $107.54 vs. $80.04, p<0.05), and those with depression incurred 23.2% great er monthly direct asthma-related expenditures than those without depression (B=.232; $93.70 vs. $82.74, p<0.05) Model 3 illustrated that mental health treatment diminished the effects of all mental health conditions for direct asthma expenditures. Treatment effects for ADHD suggest that those with ADHD incurred 11.6%less monthly direct asthma expenditures compared to those without ADHD ($76.66 vs. $86.12, p<0.05) Compared to those without treatment for each of the mental health conditions, treatment for ADHD (B=.130, $94.43 vs. $82.96, p<0.01), anxiety (B=. 3 09 $11.45 vs. $81.48, p<0.01), and depression (B=.328; $113.62 vs. $81.85, p<0.01) was associated with incurring more direct monthly asthma expenditure s compared to those without treatment Model 4 finds th at the rate for monthly direct asthma -related expenditure increased by 2.3% (p<0.01) for each additional month of con tinuous anxiety treatment. Medical and pharmacy monthly asthma -related expenditures Medical monthly asthma-related expenditure s a nd pharmacy expenditures were modeled using GEE with gamma distribution, log link function, and correlation matrix specified with AR -6. The fit GEE model predictors can be found in Table 4-25 with predicted es timates for average monthly asthma -related m edical expenditures in Table 4 2 8 and asthma -related pharmacy expenditures in Table 429.
8 5 Medical monthly asthma-related expenditures. In model 1, youth with anxiety had incurred 26.7% greater asthma -related medical expenditure (B=.267; $17.13 vs. $13.12, p<0.05) compared to those without anxiety M odel 1 suggest s youth with comorbid depres sion incurred less medical expenditure compared to those without depression but the association is not s ignificant (p<0.12) Temporal effects were not observed from Model 2 which suggest the number of months with a comorbid mental health condition was not associat ed with asthma -related medical expenditure s Treatment effects observed in Model 3 suggest treatment of ADHD and anxiety diminishe s the effects for the association between ADHD and anxiety with asthma related medical expenditures. Those receiving ADHD treatment (B=.164 ; $15.72 vs. $13.34; p<0.01) and anxiety treatment (B=.163; $15.65 vs. $13.24; p<0.05) had incurred 16.4% and 16.3% greater monthl y asthma-related medical expenditure s respectively, compared to those without the particular mental health treatment. Counseling with a physician or medication management consult was associated with incurring 39.6% less medical expenditures than those wit hout such a consultation (B= .396; $9.31 vs. $13.83 p<0.05). C ontinuous treatment effects for asthma-related medical expenditures were not observed in model 4. Pharmacy expenditures. In m odel1, those with ADHD ( B=.072; $59.19 vs. $55.06, p<0.01) and anxie ty ( B=.199; $66.22 vs. $54.25, p<0.01) incurred 7.2% and 19.9% greater monthly asthma-related pharmacy expenditures respectively compared to those without the mental health condition. However, those with depression had incurred 10.3% less asthma -related pharmacy expenditures ( B= -.103; $51.27 vs. $56.81, p<0.01) compared to those without depression Model 2 has a significant
86 depressionmonth interaction (B= -.010, p<0.01), indicating the rate of pharmacy monthly asthma -related expenditure decreased by 1.0% for every additional month with depression diagnosis. While controlling for the number of months with a depression diagnosis was associated with a main trend of decreasing pharmacy expenditures for every month with a depression diagnosis, the relative depr ession effect observed in model one lost significance (i.e., was adjusted away by the depression*month interaction). Model 3 finds ADHD, anxiety, and depression treatment diminishes the effect of ADHD, anxiety, and depression on monthly asthma-related phar macy expenditures supporting hypothesis 3. For those receiving ADHD ($70.33 vs. $54.18; p<0.01), anxiety ($66.24 vs. $54.93; p<0.01) and depression ($70.16 vs. $54.78; p<0.01) treatment incurred 26.1%, 18.7%, and 24.8% greater asthma -related pharmacy expe nditures respectively, compared to those not receiving the particular mental health treatment. Model 4 did show some significant effects when controlling for continuous mental health treatment. For every additional month of continuous anxiety treatment, t he rate of monthly asthma-related pharmacy expenditures increased by 1.3% (p<0.01). Two part model for inpatient and outpatient expenditures Given that the number of observations without any expenditure was high, a two part model was constructed for inpatient and outpatient expenditures First part of th e model is to first predict expenditures using logistic regression (Table 4-3 0 ). The odds of inpatient and outpatient expenditures are modeled using a binomial distribution and a logit link functi on with independent correlation. The second part is to model expenditure given at least some expenditure (Table 4 31) GEE models using exchangeable correlation matrix with the gamma distribution and log link function was used to predict inpatient
87 and outp atient expenditures given expenditures greater than $0. The final constructed predicted mean estimates of inpatient and outpatient expenditures are constructed by multiplying predicted probability estimates of expenditure by predicted estimates of expenditu re given incurring some expenditure. Result s are shown in Table 432 and Table 4 33 for prediction estimates of inpatient expenditure and Table 434 and Table 435 for prediction estimates of outpatient expenditure. Inpatient expenditures. Model 1 reporting logistic regression for inpatient expenditures, shows the increased odds of having inpatient expenditures f or anxiety (OR=1.34, p<0.01) and depression (OR=1.36, p<0.01), yet decreased for ADHD (OR=.768, p<0.01). The gamma regression for inpatient expendi tures in Model 1, illustrates those with anxiety incurred 25.3% greater inpatient expenditures than those without anxiety (B=.253, $4694.57 vs. $3645.54, p<0.01). The combined predicted estimates are based on multiplying the probability of expenditure and the estimated inpatient expenditure given some expenditure. For those with anxiety and depression, combined estimates showed that both anxiety ($545.22 vs. $328.83) and depression ($496.12 vs. $338.63) were associated with greater predicted monthly inpatient expenditure compared to those without anxiety or depression However, those with ADHD ($300.61 vs. $391.97) had less predicted monthly inpatient expenditures compared to those without ADHD. The logistic regression for m odel 2 assessed the effect of the number of months with a mental health condition on the odds of having inpatient expenditure. The results from m odel 2 suggest that for every additional month with either an ADHD (OR=.991, p<0.01) or d epression (OR=.995, p< 0.05) diagnosis, the odds of having monthly
88 asthma-related inpatient expenditure decreases Controlling for the number of months with depression, the odds of having inpatient expenditures significantly increased for those with depression compared to those without depression (OR=1. 54, p<0.01). The number of months with anxiety was not associated with increased odds of having inpatient expenditure. However, controlling for the number of months with anxiety revealed an increase in odds of having inpatient expenditure for those with anxiety compared to those without anxiety (OR=1.34, p<0.01). T he gamma regression for model 2 found no significant association between the numbers of months with a mental health diagnosis predicting inpatient expenditure given some expenditure. The combined predicted average monthly inpatient expenditures were lower for those with ADHD compared to those without ADHD ($304.58 vs. $391.88). The combined predicted average monthly inpatient expenditur es were higher for anxiety and depress ion compared to those without anxiety ($553.45 vs. $326.45) or depression ($497.52 vs. $339.67). Logistic regression for model 3 and model 4 shows mental health treatment slightly diminishes the effects for the odds of having inpatient expenditures for all mental health conditions thus mental health treatment mediates the mental health condition Receiving treatment was associated with decreased odds of inpatient expenditure for those with ADHD compared to those without ADHD (OR=.814, p<0.01). However, rec eiving treatment for anxiety and depression was associated with increased odds of inpatient expenditure for those with anxiety and depression compared to those without anxiety (OR=1.28, p<0.01) and depression (OR=1.35, p<0.01). ADHD treatment was associat ed with significant decrease in odds of inpatient expenditure compared to those
89 without ADHD treatment (OR=.823 p<0.01), whereas anxiety treatment was associated with an increase in the odds of inpatient expenditure (OR=1.13, p<0.01). Counseling was associ ated with a 1.23 times increase in the odds of having inpatient expenditure compared to those who did not receive counseling (OR=1.23, p<0.01). ADHD and anxiety treatment diminishes the odds of inpatient expenditure for those with ADHD and anxiety. Model 4 shows that the odds of inpatient expenditure decreased for every additional month of continuous ADHD treatment (OR=.983, p<0.01). The gamma regression for model 3 predicted inpatient expenditures given some expenditure. Treatment of anxiety and depression had diminished effects for those with anxiety and depression on asthma-related inpatient expenditures Controlling for mental health treatment those with anxiety had incurred 21.9% greater asthma -related inpatient expenditures compared to those without anxiety (p<0.05) No other treatment effects were observed nor were there continuous treatment effects observed in model 4. For model 3 and model 4 considering treatment effects, the combined predicted average monthly inpatient expenditures were lower for t hose with ADHD compared to those without ADHD (Model 3: $303.42 vs. 389.88; Model 4: $312.39 vs. $386.50). The combined predicted average monthly inpatient expenditures were higher for anxiety or depression compared to those without anxiety (Model 3: $514. 19 vs. $333.20; Model 4: $490.32 vs. $337.08) or depression (Model 3: $489.30 vs. $339.33; Model 4: $500.83 vs. $337.68). In model 3 predicted estimates, those receiving ADHD or depression treatment had lower inpatient expenditure (ADHD: $335.38 vs. $371.94; depression: $357 vs. $369.68), respectively, compared to those not receiving treatment. Treatment of anxiety, however, was associated with greater monthly inpatient expenditure
90 ($425.93 vs. $353.14). For model 4 considering continuous mental health treatment, those receiving ADHD or anxiety treatment had more inpatient expenditure than those not receiving ADHD treatment ($378.88 vs. $367.60) or anxiety treatment (383.32 vs. $367.18). Those receiving depression treatment have less average inpatient expend itures compared to those not receiving depression treatment ($367.28 vs. 368.76). Outpatient expenditures. Model 1 uses logistic regression for predicting odds of monthly asthma-related outpatient expenditures Those with ADHD or anxiety were associated with decreased odds for monthly asthma -related outpatient expenditures compared to those without ADHD (OR=.887, p<0.01) or anxiety (OR=.946, p<0.01). However, those with depression were associated with increased odds for monthly asthma-related outpatient expenditures compared to those without depression (OR=1.18, p<0.01). Gamma regression for model 1 found mental health conditions were not significant predictors of outpatient expenditures. The combined predicted outpatient expenditures were higher for anxiety or depression compared to those without anxiety ($128.32 vs. $125.69) or depression ($130.98 vs. $125.36). However, the combined predicted monthly asthma-related outpatient expenditures were lower for ADHD co mpared to those without ADHD ($116.42 vs. $128.34). The logistic regression for model 2 assessed the effect of the number of months with a mental health condition on the odds of having outpatient expenditure. The results suggest that the odds of monthly as thma -related outpatient expenditures decreases for every additional month with ADHD (OR=.996, p<0.01) and decreases for every additional month with depression (OR=.996, p<0.05) Controlling the number of months
91 of having a mental health diagnosis, the odds of having outpatient expenditures increased for those with depression compared to those without depression ( OR=1.28 p<0.01) Gamma regression for m odel 2 illustrated that for monthly asthma-related outp atient expenditures increased by 0.6% (p<0.01) for ev ery additional month of having an ADHD diagnosis Controlling for the number of months with a mental health diagnosis, those with ADHD had incurred 14% (B= -.140; $119.37 vs. $127.43, p<0.01) less monthly asthma -related outpatient expenditures compared to t hose without ADHD. The combined predicted outpatient expenditures were higher for anxiety or depression compared to those without anxiety ($128.95 vs. $125.52) or depression ($131.25 vs. $125.29). However, the combined predicted monthly asthma -related outp atient expenditures were lower for ADHD compared to those without ADHD ($119.37 vs. $127.43). Logistic regression for Model 3 assesses the treatment effects for the odds of having monthly asthma -related outpatient expenditure, and illustrates that mental h ealth treatment diminishes effects for all 3 mental health conditions on asthma -related outpatient expenditures Receiving treatment was associated with decreased odds of outpatient expenditure for those with ADHD compared to those without ADHD (OR=.925, p <0.01). However, receiving treatment for anxiety was associated with increased odds of outpatient expenditure for those with anxiety compared to those without anxiety (OR=1.08, p<0.01). T reatment effects included the decrease in the odds of outpatient expenditure for those receiving ADHD or anxiety t reatment compared to those with untreated ADHD (OR=.907, p<0.01) or untreated anxiety (OR=.641, p<0.01) D epression treatment was associated with an increase in the odds of having
92 outpatient expenditure compared to those untreated for depression (OR=1.29, p<0.01) Additionally, physician counseling was associated with decreased odds of having outpatient expenditure compared to those without physician counseling (OR=.941, p<0.01). Model 4 shows a decrease in odds of monthly outpatient expenditure for every additional month of continuous ADHD (OR=.990, p<0.01) and continuous anxiety treatment ( OR=.974, p<0.01). The odds of monthly outpatient expenditures increase by 1% (p<0.01) for every additional month of c ontinuous depression treatment Gamma regression showed that treatment did not mediate the effect of outpatient expenditure However, trends for all three ADHD, anxiety, and depression, seemed to exhibit some diminished effects (B closer to 0). Model 4 did have a significant effect for continuous treatment of anxiety, such that monthly asthma-related outpatient expenditures increased by 2.2% (p<0.01) for every additional month of continuous anxiety treatment. For model 3 and model 4 considering treatment effects, the combined predicted monthly asthma-related outpatient expenditures were lower for those with ADHD compared to t hose without ADHD (Model 3: $120.66 vs. 127.24; Model 4: $ 121.59 vs. $127.13). The combi ned predicted average monthly asthm a -related out patient expenditures were higher for anxiety or depression compared to those without anxiety (Model 3: $133.37 vs. $124.93; Model 4: $132.10 vs. $125.1 8) or depression (Model 3: $129.87 vs. $125.48; Model 4: $130.74 vs. $125.41). In model 3 pr edicted estimates, those receivin g ADHD or anxiety treatment had lower monthly asthma-related out patient expenditure respectively, compared to those with untreated ADHD (Model 3: $114.34 vs. $127.07; Model 4: $118.64 vs. $126.75) or untreated anxiety (Mod el 3: $106.79 vs. $126.94; Model 4: $112.52 vs. $126.89) Those receiving treatment for depression had
93 greater predicted monthly asthma-related outpatient expenditure compared to those untreated for depression (Model 3: $132.65 vs. $125.70; Model 4: $134.04 vs. $125.74) Summary of Expenditure Analyses ADHD and expenditure. Generally, those with ADHD had less asthma-related total, direct, inpatient, and outpatient expenditures yet had greater expenditures for monthly asthma -related pharmacy expenditures T his is similar to patterns observed in the negative binomial models for predicting utilization. Treatment of ADHD was associated with greater average monthly total, direct, medical and pharmacy asthmarelated expenditures yet had less monthly asthma-related inpatient expenditures Anxiety and expenditure Anxiety was associated with incurring 29% greater total asthma -related expenditures, 30.4% greater direct asthma expenditures, 26.7% greater asthma -related medical expenditures, 19.9% greater asthmarelate d pharmacy expenditure, and 25.3% greater asthma-related inpatient expenditures compared to those without anxiety. Additionally, those with anxiety had incurred more predicted inpatient expenditures compared to those without. Treatment of anxiety was assoc iated with greater total, direct, inpatient, medical, and pharmacy expenditures. Depression and expenditure Depression was generally associated with incurring greater total, direct, and inpatient asthma-related monthly expenditures, yet less medical and p harmacy monthly expenditures. Treatment of d epression had diminished effects for those with depression. S pecifically those with depression treatment had incurred greater total, direct, and pharmacy asthma-related monthly expenditures but also had less predicted inpatient expenditure.
94 Table 4 1. Sample characteristics of youth with asthma (n=8,241) Sample Characteristics % or Mean (s.d.) Mean Age (s.d.) 11.23 (3.03) Sex % Male 57.7% Race/Ethnicity % White 25.7% % Black 30.9% % Hispanic 24.0% % Other 19.4% % Supplemental Security Income (ssi) 30.6% % ADHD (over 3 yr, n=1,531) 18.5% % Anxiety (over 3 yr, n=285) 3.5% % Depression (over 3 yr, n=938) 11.4% % Allergic Rhinitis (over 3yr) 35.5% % w/ drug management consult/ counseling 5.6% s.d. =standard deviation; Characteristics of continuous enrolled youth with asthma in Florida Medicaid (n=8,241).
95 Table 4 2. Average monthly expenditure by mental health condition Inpatient Outpatient Medical Pharmacy Direct Asthma Total Asthma Mental Health condition Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) ADHD 23.68 (2.00) 9.39 (.31) 13.75 (.53) 61.57 (.60) 89.45 (2.22) 108.39 (2.27) No ADHD 25.67 (.93) 10.96 (.16) 13.23 (.18) 53.96 (.28) 88.09 (1.03) 103.83 (1.05) Anxiety 48.71 (3.61) 10.48 (.40) 15.98 (.80) 75.69 (.88) 128.85 (3.86) 150.86 (3.90) No Anxiety 21.83 (.81) 10.71 (.15) 12.92 (.16) 52.27 (.26) 82.28 (.91) 97.73 (.93) Depression 44.23 (3.46) 10.22 (.38) 15.34 (.77) 61.43 (.70) 112.66 (3.67) 131.22 (3.69) No Depression 22.35 (.82) 10.75 (.15) 13.01 (.17) 54.35 (.27) 84.52 (.92) 100.46 (.94) Total 25.32 (.85) 10.68 (.14) 13.32 (.18) 55.31 (.25) 88.34 (.94) 104.63 (.95) Average expenditure is for n=296,676 person months of 8,241 youth with asthma. Monthly expenditure s are measured in unadjusted U.S. dollars. Based on the entire population, inpatient and outpatient information is diluted over 3 years among all youth with asthma.
96 Table 4 3. Average monthly use by mental health condition Inpatient Admissions LOS* Average LOS* Outpatient Visits Urgent Visits Medical Claims Mental Health condition Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) ADHD .0058(.00035) 3.7 (.19) 3.5 (.17) .0339 (.0009) .0217 (.0006) .281 (.005) No ADHD .0064 (.00017) 3.5 (.08) 3.3 (.08) .0384 (.0004) .0244 (.0003) .324 (.003) Anxiety .0092 (.00051) 4.5 (.19) 4.4 (.18) .0377 (.0011) .0232 (.0008) .323 (.008). No Anxiety .0058 (.00015) 3.3 (.08) 3.1 (.07) .0376 (.0004) .0240 (.0003) .316 (.003) Depression .0088 (.00016) 4.5 (.21) 4.3 (.21) .0377 (.0011) .0232 (.0007) .273 (.006) No Depression .0059 (.00043) 3.3 (.08) 3.1 (.07) .0376 (.0004) .0240 (.0003) .324 (.003) Total .0063 (.00015) 3.5 (.07) 3.4 (.07) .0377 (.0004) .0239 (.0003) .317 (.003) Total Pharmacy Claims Direct Pharmacy Claims Indirect Pharmacy Claims Controller Pharmacy Claims Rescuer Pharmacy Claims Mental Health condition Mean (SE) Mean (SE) Mean (SE) Mean (SE) Mean (SE) ADHD .968 (.007) .614 (.005) .421 (.003) .345 (.003) .238 (.002) No ADHD .860 (.003) .572 (.002) .355 (.002) .295 (.001) .243 (.001) Anxiety 1.111 (.008) .700 (.006) 493 (.004) .368 (.004) .304 (.003) No Anxiety .844 (.003) .561 (.002) .348 (.001) .295 (.001) .233 (.001) Depression .979 (.008) .601 (.006) .447 (.004) .335 (.003) .243 (.002) No Depression .863 (.003) .576 (.002) .354 (.001) .299 (.001) .242 (.001) Total .879 (.003) .579 (.002) .367 (.001) .304 (.001) .242 (.001) SE= standard error; LOS=length of stay. *The length of stay variable was assessed conditional admission>0 LOS is the total length of stay for the month, whereas average LOS was equal to total monthly length of stay divided by number of admissions in that month.
97 Table 4 4. Incident rate ratios for monthly asthma related inpatient admissions Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD .792 (.659,.951)** .938 (.701,1.25) .802 (.647,.995)* .867 (.695,1.08) Anxiety 1.33 (1.05,1.68)** 1.35 (.909,2.01) 1.09 (.827,1.43) 1.10 (.837,1.45) Depression 1.43 (1.13,1.82)** 1.53 (1.02,2.30)* 1.28 (.977,1.68) 1.32 (1.01,1.74)* Month .968 (.963,.972)** .970 (.964,.975)** .969 (.964,.974)** .970 (.965,.975)** ADHD*month .990 (.977,1.004) Anxiety*month .999 (.982,1.02) Depression*month .997 (979,1.02) Counseling 1.24 (.931,1.65) 1.25 (.937,1.66) ADHD treatment .926 (.719, 1.19) 1.04 (.792,1.36) Anxiety treatment 1.47 (1.04, 2.07)* 1.42 (.977,2.05) Depression treatment 1.25 (.887, 1.77) 1.39 (.956,2.02) Continuous ADHD treatment .973 (.952,.995)** Continuous anxiety treatment 1.00 (.978,1.03) Continuous depression treatment .978 (.948,1.01) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthmaspecific tied to asthma ICD 9 codes; Negative binomial models controlled for age, gender, race, supplemental security income using random effects
98 Table 4 5. Predicted mean estimates for monthly asthma -related inpatient admissions Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE ) Mean (SE ) Mean (SE ) Mean (SE ) All use .00627 (.359) .00627 (.357) .00628 (.360) .00628 (.360) ADHD .00520 (.368) .00531 (.371) .00526 (.373) .00560 (.374) No ADHD .00656 (.357) .00657 (.355) .00656 (.360) .00646 (.360) Anxiety .00789 (.372) .00794 (.379) .00671 (.381) .00679 (.381) No anxiety .00595 (.359) .00595 (.358) .00618 (.361) .00616 (.361) Depression .00842 (.373) .00857 (.380) .00768 (.380) .00789 (.380) No depression .00588 (.359) .00587 (.358) .00600 (.362) .00597 (.362) ADHD treatment .00586 (.381) .00651 (.384) No ADHD treatment .00632 (.360) .00626 (.361) Anxiety treatment .00889 (.397) .00859 (.402) No anxiety treatment .00605 (.361) .00607 (.362) Depression treatment .00771 (.399) .00847 (.404) No depression treatment .00614 (.361) .00610 (.362) Counseling .00764 (.384) .00769 (.384) No Counseling .00617 (.359) .00617 (.359) SE=standard error; All predictions came from models in table 44.
99 Table 4 6. Incident rate ratios for monthly asthma related inpatient length of stay (if admissions>0) Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD .950 (.849,1.06) .943 (.782,1.14) .923 (.806, 1.06) .937 (.815,1.08) Anxiety 1.29 (1.14,1.47)** 1.35 (1.07,1.69)** 1.28 (1.08, 1.50)** 1.26 (1.06,1.48)** Depression 1.11 (.968,1.27) 1.09 (.859,1.39) 1.05 (.894, 1.24) 1.07 (.903,1.26) Month .996 (.992,.999)* .996 (.992,1.000)* .995 (.992,1.000)* .996 (.992,.999)* ADHD*month 1.00 (.978,1.01) Anxiety*month .998 (.986,1.01) Depression*month 1.00 (.900,1.01) Counseling 1.12 (.960,1.30) 1.11 (.957,1.29) ADHD treatment 1.04 (.873,1.23) 1.08 (.892,1.34) Anxiety treatment 1.002 (.818,1.23) .974 (.786,1.21) Depression treatment 1.08 (.872,1.33) 1.10 (.881,1.38) Continuous ADHD treatment .992 (.977,1.01) Continuous anxiety treatment 1.01 (.991,1.02) Continuous depression treatment .995 (.976,1.01) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthmaspecific tied to asthma ICD 9 codes; Negative binomial models controlled for age, gender, race, supplemental security income using random effects
100 Table 4 7. Pr edicted mean estimates for monthly asthma -related length of stay Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE ) Mean (SE ) Mean (SE ) Mean (SE ) All use 3.18 (.146) 3.18 (.149) 3.17 (.149) 3.16 (.152) ADHD 3.05 (.153) 3.05 (.160) 2.97 (.159) 2.99 (.163) No ADHD 3.21 (.145) 3.21 (.147) 3.22 (.149) 3.20 (.153) Anxiety 3.95 (.156) 3.94 (.166) 3.89 (.166) 3.83 (.170) No anxiety 3.05 (.146) 3.06 (.149) 3.05 (.150) 3.05 (.153) Depression 3.46 (.160) 3.46 (.170) 3.30 (.169) 3.33 (.173) No depression 3.12 (.146) 3.12 (.149) 3.14 (.150) 3.13 (.153) ADHD t reatment 3.26 (.173) 3.39 (.181) No ADHD treatment 3.16 (.149) 3.14 (.153) Anxiety t reatment 3.17 (.178) 3.08 (.185) No anxiety t reatment 3.17 (.150) 3.16 (.153) Depression t reatment 3.39 (.180) 3.46 (.187) No depression t reatment 3.15 (.150) 3.14 (.153) C ounseling 3.51 (.165) 3.49 (.168) No Counseling 3.14 (.148) 3.14 (.152) SE=standard error; All predictions came from models in table 46.
101 Table 4 8 Incident rate ratios for monthly asthma-related medical claims Model 1 Model 2 Model 3 Model 4 Mental Health I RR(CI) IRR (CI) IRR (CI) IRR (CI) ADHD 1.11 (1.06,1.16)** 1.05 (.981,1.13) 1.01 (.955, 1.06) 1.01 (.955,1.06) Anxiety 1.13 (1.06,1.20)** 1.05 (.948,1.16) 1.08 (1.006, 1.15)* 1.08 (1.01,1.15)* Depression .900 (.845,.959)** 1.07 (.958,1.19) .842 (.785, .903)** .834 (.777,.895)** Month .968 (.967,.970)** .969 (.967,.970)** .969 (.968, .970)** .969 (.968,.970)** ADHD*month 1.003(1.000,1.006)* Anxiety*month 1.004(.9996,1.008) Depression*month .991 (.987,.996)** Counseling 1.05 (.959, 1.15) 1.04 (.955,1.14) ADHD treatment 1.23 (1.17,1.30)** 1.24 (1.18,1.32)** Anxiety treatment 1.10 (1.01, 1.20)* 1.12 (1.02,1.23)* Depression treatment 1.17 (1.07, 1.28)** 1.12 (1.02,1.24)** Continuous ADHD tr eatment .998 (.994,1.002) Continuous anxiety treatment .996 (.989,1.004) Continuous depression t reatment 1.01(1.002,1.018)* IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthmaspecific tied to asthma ICD 9 codes; Negative binomial models controlled for age, gender, race, supplemental security income using fixed effe cts
102 Table 4 9. Predicted mean estimates for monthly asthma -related medical claims Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All use .234 (.030) .234 (.031) .235 (.032) .235 (.033) ADHD .255 (.035) .253 (.038) .236 (039) .236 (.041) No ADHD .230 (.029) .230 (.030) .234 (.032) .234 (.033) Anxiety .261 (.041) .257 (.046) .250 (.045) .250 (.046) No anxiety .231 (.030) .231 (.031) .232 (.033) .232 (.034) Depression .214 (.041) .223 (.047) .202 (.046) .201 (.047) No depression .238 (.030) .237 (.031) .240 (.033) .241 (.034) ADHD treatment .283 (.041) .286 (.043) No ADHD treatment .230 (.032) .230 (.033) Anxiety treatment .257 (.055) .261 (.058) No anxiety treatment .233 (.033) .233 (.034) Depression treatment .272 (.055) .262 (.059) No depression treatment .233 (.032) .233 (.034) Counseling .245 (.054) .244 (.054) No Counseling .234 (.031) .234 (.032) SE=standard error; All predictions came from models in table 48.
103 Table 4 10 Incident rate ratios on monthly asthma-related outpatient visits Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD .995 (.875,1.13) 1.02 (.861,1.21) 1.03 (.895, 1.18) 1.04 (.906,1.19) Anxiety 1.23 (1.05,1.43)** 1.25 (.972,1.61) 1.11 (.948, 1.31) 1.11 (.946,1.31) Depression .917 (.786,1.07) .802 (.627,1.03) .941 (.800, 1.11) .941 (.800,1.11) Month .974 (.971,.977)** .974 (.970,.977)** .973 (.971, .977)** .974 (.971,.977)** ADHD*month .998 (.993,1.008) Anxiety*month .999 (.989,1.01) Depression*month 1.01 (.998,1.02) Counseling 1.07 (.717, 1.61) 1.06 (.705,1.58) ADHD treatment .931 (.828, 1.05) .951 (.844,1.07) Anxiety treatment 1.49 (1.22,1.83)** 1.46 (1.18,1.80)** Depression treatment .814 (.669, .990)* .811 (.661,.994)* Continuous ADHD tr eatment .991 (.982,1.001) Continuous anxiety treatment 1.01 (.988,1.03) Continuous depression t reatment 1.00 (.984,1.02) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthmaspecific tied to asthma ICD 9 codes; Negative binomial models controlled for age, gender, race, supplemental security income using fixed effe cts
104 Table 4 11. Predicted mean estimates for monthly asthma-related outpatient visits Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All use .531 (.134) .531 (.135) .534 (.141) .535 (.143) ADHD .529 (.143) .530 (.147) .545 (.153) .552 (.154) No ADHD .531 (.132) .532 (.133) .532 (.140) .531 (.143) Anxiety .634 (.151) .634 (.159) .586 (.160) .586 (.162) No anxiety .517 (.134) .518 (.135) .536 (.142) .527 (.144) Depression .492 (.150) .479 (.157) .507 (.159) .507 (.161) No depression .537 (.134) .538 (.136) .538 (.142) .539 (.144) ADHD treatment .501 (.152) .511 (.155) No ADHD treatment .537 (.141) .537 (.143) Anxiety treatment .780 (.174) .764 (.178) No anxiety treatment .522 (.142) .523 (.144) Depression treatment .440 (.171) .511 (.176) No depression treatment .541 (.142) .537 (.144) Counseling .571 (.237) .562 (.238) No Counseling .532 (.136) .533 (.139) SE= standard error; All predictions came from models in table 410.
105 Table 4 12. Incident Rate Ratios for asthma-related urgent visits Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD 1.07 (.900,1.27) 1.07 (.854,1.33) 1.09 (.912, 1.30) 1.10 (.918,1.31) Anxiety 1.33 (1.10,1.61)** 1.51 (1.10,2.07)** 1.18 (.969, 1.44) 1.18 (.962,1.44) Depression .852 (.704,1.03) .700 (.515,.952)* .917 (.752, 1.12) .919 (.753,1.12) Month .977 (.972,.983)** .977 (.971,.982)** .977 (.972, .983)** .977 (.972,.983)** ADHD*month 1.000 (.993, 1.007) Anxiety*month .994 (.983,1.01) Depression*month 1.01 (.998,1.02) Counseling .862 (.304, 2.45) .848 (.303,2.38) ADHD Treatment .960 (.832, 1.11) .976 (.844,1.13) Anxiety Treatment 1.77 (1.38, 2.29)** 1.72 (1.33,2.24)** Depression Treatment .654 (.510, .840)** .655 (.506,.848)** Continuous ADHD tr eatment .992 (.980,1.005) Continuous anxiety treatment 1.01 (.986,1.04) Continuous depression t reatment .999 (.974,1.03) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthmaspecific tied to asthma ICD 9 codes or asthma-specific pharmacy claims (Table 3 3); Negative binomial models controlled for age, gender, race, supplemental security income using fixed effects
106 Table 4 13. Predicted average estimates of monthly asthma-related urgent visits Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All use 1.25 (.314) 1.26 (.315) 1.28 (.334) 1.28 (.335) ADHD 1.32 (.322) 1.32 (.325) 1.37 (.343) 1.38 (.344) No ADHD 1.24 (.312) 1.24 (.314) 1.26 (.333) 1.26 (.334) Anxiety 1.61 (.326) 1.66 (.333) 1.47 (.347) 1.47 (.348) No anxiety 1.21 (.314) 1.21 (.315) 1.25 (.334) 1.25 (.335) Depression 1.09 (.325) 1.05 (.331) 1.19 (.345) 1.19 (.347) No depression 1.28 (.314) 1.29 (.316) 1.30 (.334) 1.30 (.335) ADHD treatment 1.23 (.341) 1.26 (.342) No ADHD treatment 1.28 (.333) 1.28 (.335) Anxiety treatment 2.19 (.358) 2.13 (.361) No anxiety treatment 1.24 (.334) 1.23 (.335) Depression treatment .858 (.356) .862 (.359) No depression treatment 1.31 (.334) 1.31 (.335) Counseling 1.11 (.595) 1.10 (.590) No Counseling 1.29 (.319) 1.29 (.321) SE=standard error; All predictions came from models in table 412.
107 Table 4 14. Incident Rate Ratios for monthly asthma-related total pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD 1.19 (1.15,1.22)** 1.17 (1.12,1.22)** .990 (.958, 1.02) .971 (.934,1.004) Anxiety 1.37 (1.32,1.42)** 1.27 (1.21,1.35)** 1.24 (1.19,1.29)** 1.22 (1.17,1.27)** Depression .930 (.896,.965)** 1.07 (1.01,1.14)** .828 (.795,.863)** .814 (.781,.849)** Month .988 (.987,.989)** .988 (.987,.989)** .989 (.988,.989)** .988 (.987,.989)** ADHD*month 1.001 (.999,1.002) Anxiety*month 1.004 (1.00 2 ,1.006)** Depression*month .993 (.991,.995)** Counseling 1.24 (1.13,1.36)** 1.20 (1.10,1.32)** ADHD Treatment 1.49 (1.45,1.54)** 1.45 (1.41,1.50)** Anxiety Treatment 1.26 (1.21,1.32)** 1.20 (1.15,1.25)** Depression Treatment 1.24 (1.18,1.29)** 1.24 (1.19,1.30)** Continuous ADHD tr eatment 1.02 (1.01,1.02)** Continuous anxiety treatment 1.02 (1.01,1.02)** Continuous depression t reatment .998 (.995,1.002) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthmaspecific tied to asthma ICD 9 codes or asthma-specific pharmacy claims (Table 3 3); Negative binomial models controlled for age, gender, race, supplemental security income using fixed effects
108 Table 4 15. Predicted average estimates of monthly asthma-related total pharmacy cl aims Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All use 1.42 (.026) 1.42 (.026) 1.43 (.028) 1.43 (.028) ADHD 1.62 (.028) 1.63 (.029) 1.42 (.031) 1.40 (.031) No ADHD 1.37 (.025) 1.37 (.026) 1.44 (.027) 1.45 (.027) Anxiety 1.85 (.030) 1.83 (.032) 1.71 (.033) 1.68 (.033) No anxiety 1.35 (.026) 1.35 (.026) 1.38 (.028) 1.38 (.028) Depression 1.34 (.031) 1.38 (.033) 1.23 (.033) 1.22 (.034) No depression 1.44 (.026) 1.44 (.026) 1.49 (.028) 1.50 (.028) ADHD treatment 2.03 (.030) 1.98 (.031) No ADHD treatment 1.36 (.028) 1.36 (.028) Anxiety treatment 1.77 (.035) 1.69 (.036) No anxiety treatment 1.40 (.028) 1.41 (.028) Depression treatment 1.74 (.035) 1.75 (.036) No depression treatment 1.41 (.028) 1.41 (.028) Counseling 1.75 (.052) 1.70 (.052) No Counseling 1.41 (.026) 1.41 (.027) SE=standard error; All predictions came from models in table 414.
109 Table 4 16. Incident Rate Ratios for monthly direct asthma pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD 1.15 (1.11,1.20)** 1.11 (1.06,1.17)** .989 (.950,1.03) .974 (.936,1.01) Anxiety 1.29 (1.23,1.34)** 1.16 (1.09,1.24)** 1.20 (1.14,1.25)** 1.18 (1.13,1.23)** Depression .907 (.868,.948)** 1.09 (1.02,1.17)** .806 (.769,.847)** .796 (.758,.835)** Month .988 (.987,.989)** .988 (.987,.989)** .989 (.988,.990)** .988 (.987,.989)** ADHD*month 1.002(1.001,1.004)** Anxiety*month 1.005(1.003,1.007)** Depression*month .991 (.989,.994)** Counseling 1.16 (1.01,1.32)* 1.11 (.973,1.27) ADHD treatment 1.44 (1.39,1.48)** 1.40 (1.36,1.45)** Anxiety treatment 1.18 (1.12,1.24)** 1.13 (1.07,1.19)** Depression treatment 1.26 (1.19,1.32)** 1.27 (1.20,1.34)** Continuous ADHD tr eatment 1.009 (1.007,1.011)** Continuous anxiety treatment 1.02 (1.01,1.02)** Continuous depression t reatment .997 (.992,1.001) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Fixed effect models controlled for age, gender, race, supplemental security income. Pharmacy claims are defined in Table 3 -3.
110 Table 4 17. Predicted average e stimates of monthly direct asthma pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE ) Mean (SE ) Mean (SE ) Mean (SE ) All use 1.95 (.037) 1.95 (.037) 1.94 (.039) 1.93 (.039) ADHD 2.17 (.040) 2.17 (.040) 1.92 (.043) 1.89 (.043) No ADHD 1.88 (.037) 1.89 (.037) 1.95 (.039) 1.94 (.039) Anxiety 2.40 (.041) 2.36 (.043) 2.25 (.044) 2.21 (.045) No anxiety 1.87 (.037) 1.87 (.037) 1.88 (.039) 1.88 (.040) Depression 1.80 (.042) 1.87 (.044) 1.63 (.045) 1.61 (.046) No depression 1.98 (.037) 1.98 (.037) 2.02 (.039) 2.02 (.039) ADHD treatment 2.66 (.042) 2.60 (.042) No ADHD treatment 1.85 (.039) 1.85 (.039) Anxiety treatment 2.26 (.046) 2.16 (.047) No anxiety treatment 1.91 (.039) 1.91 (.040) Depression treatment 2.39 (.047) 2.40 (.047) No depression treatment 1.90 (.039) 1.89 (.040) Counseling 2.21 (.076) 2.12 (.075) No Counseling 1.92 (.037) 1.92 (.037) SE= standard error; All predictions came from models in T able 4 -16.
111 Table 4 18. Incident Rate Ratios for monthly indirect asthma pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD 1.15 (1.10,1.20)** 1.17 (1.11,1.24)** 1.005 (.961,1.05) .995 (.950,1.04) Anxiety 1.17 (1.12,1.23)** 1.13 (1.06,1.22)** 1.08 (1.03,1.14)** 1.07 (1.02,1.13)** Depression 1.02 (.967,1.07) 1.08 (1.00,1.16)* .925 (.878,.976)** .911 (.863,.960)** Month .987 (.986,.989)** .987 (.986,.989)** .988 (.987,.989)** .987 (.986,.988)** ADHD*month .999 (.998,1.001) Anxiety*month 1.002 (.999,1.004) Depression*month .997 (.995,.999)* Counseling 1.67 (1.18,2.38)** 1.57 (1.12,2.20)** ADHD treatment 1.38 (1.33,1.42)** 1.35 (1.30,1.40)** Anxiety treatment 1.24 (1.18,1.31)** 1.22 (1.15,1.29)** Depression treatment 1.18 (1.12,1.25)** 1.18 (1.11,1.24)** Continuous ADHD tr eatment 1.008(1.006,1.011)** Continuous anxiety treatment 1.009(1.005,1.013)** Continuous depression t reatment 1.001 (.997,1.005) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Pharmacy claims are defined in Table 3 3. Fixed effect models controlled for age, gender, race, supplemental security income
112 Table 4 19. Predicted average estimates of monthly indirect asthma pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All use 4.76 (.083) 4.77 (.084) 4.69 (.088) 4.60 (.086) ADHD 5.29 (.085) 5.31 (.085) 4.70 (.090) 4.58 (.089) No ADHD 4.60 (.083) 4.60 (.083) 4.68 (.087) 4.61 (.086) Anxiety 5.44 (.086) 5.42 (.087) 4.99 (.091) 4.86 (.090) No anxiety 4.63 (.084) 4.64 (.084) 4.62 (.088) 4.54 (.087) Depression 4.82 (.087) 4.89 (.088) 4.41 (.091) 4.27 (.090) No depression 4.74 (.083) 4.75 (.084) 4.76 (.087) 4.69 (.086) ADHD treatment 6.16 (.089) 5.94 (.088) No ADHD treatment 4.48 (.087) 4.40 (.087) Anxiety treatment 5.72 (.092) 5.50 (.091) No anxiety treatment 4.60 (.088) 4.52 (.087) Depression treatment 5.45 (.092) 5.34 (.091) No depression treatment 4.62 (.086) 4.53 (.087) Counseling 7.43 (.193) 6.90 (.186) No Counseling 4.44 (.082) 4.39 (.081) SE=standard error; All predictions came from models in T able 4 18.
113 Table 4 20. Incident Rate Ratios for asthma controller pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health IRR(CI) IRR(CI) IRR(CI) IRR(CI) ADHD 1.10 (1.06,1.13)** 1.06 (1.01,1.11)* .975 (.937,1.02) .968 (.930,1.01) Anxiety 1.11 (1.06,1.16)** 1.09 (1.02,1.16)* 1.03 (.983,1.08) 1.03 (.978,1.08) Depression .987 (.943,1.03) 1.07 (.999,1.15)* .884 (.839,.931)** .878 (.833,.925)** Month .992 (.991,.993)** .993 (.991,.994)** .993 (.992,.994)** .993 (.992,.994)** ADHD*month 1.002(1.000,1.004)* Anxiety*month 1.001 (.998,1.004) Depression*month .995 (.992,.998)** Counseling .905 (.822,.983)* .899 (.828,.976) ADHD treatment 1.24 (1.21,1.27)** 1.23 (1.20,1.26)** Anxiety treatment 1.12 (1.07,1.17)** 1.11 (1.06,1.17)** Depression treatment 1.21 (1.15,1.26)** 1.19 (1.14,1.25)** Continuous ADHD tr eatment 1.004(1.001,1.006)** Continuous anxiety treatment 1.002 (.997,1.006) Continuous depression t reatment 1.004 (.999,1.008) IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Pharmacy claims are defined in Table 3 3. Generalized estimating equations with negative binomial family and link with unstructured correlation was used to control for age, gender, race, supplemental security income.
114 Table 4.21. Predicted average monthly estimates of asthma controller claims Model 1 Model 2 Model 3 Model 4 Mean (SE ) Mean (SE ) Mean (SE ) Mean (SE ) All average use .292 (.029) .292 (.030) .292 (.030) .292 (.031) ADHD .321 (.032) .323 (.033) .285 (.034) .283 (.035) No ADHD .285 (.029) .285 (.029) .294 (.030) .295 (.030) Anxiety .327 (.035) .327 (.038) .303 (.038) .302 (.038) No anxiety .286 (.029) .286 (.030) .290 (.030) .291 (.031) Depression .288 (.035) .289 (.039) .256 (.039) .254 (.039) No depression .293 (.029) .294 (.030) .300 (.030) .301 (.031) ADHD treatment .379 (.032) .375 (.033) No ADHD treatment .284 (.030) .284 (.031) Anxiety treatment .335 (.037) .334 (.038) No anxiety treatment .289 (.030) .289 (.031) Depression treatment .370 (.037) .365 (.038) No depression treatment .288 (.030) .288 (.031) Counseling .259 (.050) .257 (.050) No counseling .295 (.029) .295 (.030) SE=standard error; All predictions came fr om negative binomial models on T able 420.
115 Table 4 22. Incident Rate Ratios for asthma rescuer pharmacy claims Model 1 Model 2 Model 3 Model 4 Mental Health IRR (CI) IRR ( CI) IRR ( CI) IRR (CI) ADHD .989 (.950,1.03) .972 (.922,1.03) .892 (.852,.935)** .897 (.856,.940)** Anxiety 1.21 (1.15,1.27)** 1.11 (1.03,1.19)** 1.12 (1.06,1.19)** 1.11 (1.05,1.17)** Depression .869 (.825,.917)** .975 (.902,1.05) .812 (.765,.862)** .816 (.769,.867)** Month .989 (.988,.990)** .989 (.988,.990)** .990 (.989,.991)** .990 (.989,.991)** ADHD*month 1.001 (.999,1.003) Anxiety*month 1.005(1.002,1.009)** Depression*month .993 (.990,.997)** Counseling .926 (.854,1.004) .925 (.853,1.002) ADHD treatment 1.24 (1.19,1.28)** 1.24 (1.20,1.29)** Anxiety treatment 1.15 (1.09,1.22)** 1.10 (1.04,1.17)** Depression treatment 1.13 (1.07,1.20)** 1.15 (1.09,1.22)** Continuous ADHD tr eatment .997 (.994,1.001) Continuous anxiety treatment 1.02 (1.01,1.02)** Continuous depression t reatment .993 (.988,.999)* IRR=Incident Rate Ratios, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Pharmacy claims defined in Table 3 3. Generalized estimating equations with negative binomial family and link with unstructured correlation was used to control for age, gender, race, supplemental security income.
116 Table 4.23. Predicted average monthly estimates of asthma rescuer claims Model 1 Model 2 Model 3 Model 4 Mean (SE) Mean (SE) Mean (SE) Mean (SE) All average use .235 (.028) .235 (.028) .235 (.029) .235 (.030) ADHD .232 (.032) .232 (.034) .210 (.035) .210 (.036) No ADHD .236 (.027) .235 (.028) .242 (.029 .241 (.030) Anxiety .289 (.035) .289 (.039) .267 (.039) .262 (.040) No anxiety .228 (.028) .227 (.029) .231 (.030) .230 (.030) Depression .203 (.037) .204 (.041) .190 (.040) .190 (.041) No depression .241 (.028) .242 (.029) .245 (.029) .244 (.030) ADHD treatment .302 (.040) .303 (.035) No ADHD treatment .230 (.030) .230 (.030) Anxiety treatment .278 (.040) .263 (.041) No anxiety treatment .233 (.030) .233 (.030) Depression treatment .272 (.040) .278 (.042) No depression treatment .233 (.030) .233 (.030) Counseling .215 (.049) .215 (.049) No counseling .237 (.028) .236 (.029) SE= standard error; All predictions came fr om negative binomial models on T able 422.
117 Table 4 24. Predictors of monthly total and direct asthma related expenditure Total expenditure D irect expenditure Beta (CI) Beta (CI) Model 1 ADHD .019 ( .104, .067) .049 ( .144, .046) Anxiety .291 ( .174, .408)** .304 ( .174, .436)** Depression .082 ( .035, .198) .110 ( .022, .241) Model 2 ADHD .085 ( .207, .038) .127 ( .266, .013) Anxiety .224 ( .033, .415)* .242 ( .023, .460)* Depression .212 ( .021, .402)* .232 ( .014, .450)* Month -.016 ( -.018, -.013)** -.017 ( -.019, -.014)** ADHD*month .004 ( -.001, .010) .005 ( -.002, .011) Anxiety*month .003 ( .005, .012) .003 ( .006, .012) Depression*month .007 ( .015, .001) .006 ( .016, .003) Model 3 ADHD -.104 ( -.199, -.009)* -.116 ( .223, -.010)* Anxiety .111 ( -.015, .238) .116 ( .027, .258) Depression .052 ( .178,.073) .022 ( .163, .119) ADHD treatment .162 ( .073, .251)** .130 ( .026, .233)** Anxiety treatment .298 ( .142, .454)** .309 ( .129, .489)** Depression treatment .316 ( .164, .467)** .328 ( .153, .503)** Counseling .039 ( .115, .194) .021 ( .146, .188) Model 4 ADHD .109 ( .205, .012)* .115 ( .224, .007) Anxiety .074 ( -.054, .201) .076 ( -.067, .219) Depression .039 ( .165, .088) .003 ( .145, .139) ADHD treatment .152 ( .060, .244)** .125 ( .018, .232)* Anxiety treatment .249 ( .084, .414)** .253 ( .062, .444)** Depression treatment .352 ( .193, .513)** .374 ( .188, .599)** Continuous ADHD tr eatment .003 ( -.005, .011) .001 ( -.008, .010) Continuous anxiety treatment .022 ( .008, .037)** .023 ( .007, .039)** Continuous depression t reatment .014 ( .029, .001) .016 ( .033, .001) Counseling .037 ( .118, .191) .020 ( .147, .187) CI= 95% Confidence Intervals; pvalue: <0.05; ** < 0.01; Dependent variables are asthma-specific tied to asthma ICD -9 codes; Generalized estimating equations w/ gamma family and log link were used to control for age, gender, race, supplemental security income. Total and direct expenditure s modeled w/ unstr uctured correlation.
118 Table 4 25. Predictors of monthly asthma related medical and pharmacy expenditure Medical expenditure Pharmacy expenditure Beta (CI) Beta (CI) Model 1 ADHD .022 ( .122, .167) .072 ( .013, .131)** Anxiety .267 ( .087, .447)** .199 ( .123, .276)** Depression .144 ( .324, .035) .103 ( .179, .026)** Model 2 ADHD .069 ( .284, .148) .106 ( .013, .199)* Anxiety .290 ( .018, .598) .102 ( .033, .238) Depression .021 ( .283, .325) .078 ( .056, .212) Month .024 ( .029, .019)** .000 ( .002, .002) ADHD*month .006 ( .004, .016) .002 ( .006, .003) Anxiety*month .000 ( .014, .014) .005 ( .001, .011) Depression*month .010 ( .024, .004) .010 ( .015, .004)** Model 3 ADHD .032 ( .181, .118) .060 ( .122,.002) Anxiety .168 ( .020, .355) .084 ( .003, .164)* Depression -.165 ( .351, .021) -.233 ( .312, -.153)** ADHD treatment .164 ( .088, .241)** .261 ( .222, .300)** Anxiety treatment .163 ( .019, .307)* .187 ( .115, .260)** Depression treatment .029 ( .110, .168) .248 ( .178, .318)** Counseling .396 ( .773, .020)* .059 ( .187, .069) Model 4 ADHD .029 ( .179, .122) .068 ( .132, .004)* Anxiety .159 ( .030, .347) .061 ( .021, .142) Depression .171 ( .357, .014) .232 ( .313, .151)** ADHD treatment .174 ( .097, .250)** .260 ( .220, .300)** Anxiety treatment .160 ( .014, .306)* .164 ( .089, .238)** Depression treatment .022 ( -.119, .163) .256 ( .184, .328)** Continuous ADHD tr eatment .009 ( .021, .003) .002 ( .003, .008) Continuous anxiety treatment .006 ( .015, .027) .013 ( .004, .022)** Continuous depression t reatment .004 ( .017, .025) .002 ( .012, .007) Counseling .379 ( .751, .007)* .068 ( .197, .060) CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01; Dependent variables are asthma-specific tied to asthma ICD -9 codes or asthma -specific pharmacy claims (Table 3 -3); Generalized estimating equations w/ gamma family and log link were used to control for age, gender, race, supplemental security income. Medical and pharmacy expenditure s were modeled w/ AR -6 correlation.
119 Table 4 26. Predicted average estimates of total asthma expenditure Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All average use 99.47 (.053) 99.50 (.055) 99.17 (.057) 99.37 (.059) ADHD 98.01 (.063) 97.96 (.068) 91.37 (.069) 91.21 (.072) No ADHD 99.85 (.052) 99.70 (.053) 101.37 (.056) 101.71 (.058) Anxiety 126.85 (.075) 125.77 (.085) 108.70 (.081) 105.58 (.084) No anxiety 94.80 (.053) 94.81 (.055) 97.26 (.057) 98.04 (.059) Depression 106.42 (.075) 108.18 (.085) 95.01 (.081) 96.26 (.083) No depression 98.09 (.053) 98.25 (.056) 100.10 (.057) 100.06 (.059) ADHD treatment 114.76 (.071) 113.94 (.074) No ADHD treatment 97.57 (.056) 97.85 (.059) Anxiety treatment 130.31 (.096) 124.55 (.102) No anxiety treatment 96.73 (.057) 97.14 (.059) Depression treatment 132.67 (.095) 137.60 (.100) No depression treatment 96.74 (.057) 96.67 (.060) Counseling 102.85 (.093) 102.79 (.095) No counseling 98.88 (.055) 99.10 (.057) SE=s tandard error; All predictions came from gamma regression models on T able 4 24.
120 Table 4 27. Predicted average estimates of direct asthma total expenditure Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All average use 84.21 (.058) 84.23 (.060) 84.03 (.062) 84.19 (.064) ADHD 80.98 (.069) 80.86 (.076) 76.66 (.077) 76.86 (.800) No ADHD 85.06 (.056) 84.92 (.058) 86.12 (.061) 86.27 (.064) Anxiety 108.52 (.083) 107.54 (.095) 92.41 (.090) 89.57 (.093) No anxiety 80.02 (.058) 80.04 (.061) 82.32 (.062) 83.02 (.065) Depression 92.22 (.083) 93.70 (.095) 82.53 (.090) 83.98 (.093) No depression 82.63 (.058) 82.74 (.061) 84.37 (.063) 84.24 (.065) ADHD treatment 94.43 (.080) 94.22 (.083) No ADHD treatment 82.96 (.062) 83.15 (.065) Anxiety treatment 111.45 (.109) 105.90 (.116) No anxiety treatment 81.84 (.062) 82.22 (.065) Depression treatment 113.62 (.107) 118.77 (.114) No depression treatment 81.85 (.063) 81.74 (.065) Counseling 85.70 (.101) 85.74 (.104) No counseling 83.90 (.060) 84.07 (.063) SE=standard error; All predictions came f rom gamma regression models on T able 4 24.
121 Table 4 28 Predicted mean estimates of monthly asthma-related medical expenditure Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All average use 13.60 (.122) 13.64 (.125) 13.53 (.126) 13.46 (.127) ADHD 13.85 (.135) 13.91 (.143) 13.19 (.140) 13.15 (.142) No ADHD 13.55 (.121) 13.55 (.123) 13.61 (.124) 13.53 (.126) Anxiety 17.13 (.147) 17.55 (.161) 15.64 (.152) 15.44 (.154) No anxiety 13.12 (.123) 13.12 (.125) 13.24 (.126) 13.15 (.128) Depression 12.02 (.146) 12.23 (.161) 11.74 (.152) 11.61 (.153) No depression 13.89 (.123) 13.99 (.126) 13.85 (.126) 13.78 (.128) ADHD treatment 15.72 (.131) 15.78 (.133) No ADHD treatment 13.34 (.126) 13.26 (.128) Anxiety treatment 15.65 (.145) 15.64 (.147) No anxiety treatment 13.24 (.126) 13.33 (.128) Depression treatment 13.91 (.144) 13.75 (.146) No depression treatment 13.51 (.126) 13.44 (.128) Counseling 9.31 (.219) 9.41 (.218) No counseling 13.83 (.121) 13.74 (.123) SE=standard error; All predictions came f rom gamma regression models on T able 4 25.
122 Table 4 29 Predicted mean estimates of monthly asthma-related pharmacy expenditure Model 1 Model 2 Model 3 Model 4 Mental Health Mean (SE) Mean (SE) Mean (SE) Mean (SE) All average use 55.88 (.042) 55.97 (.044) 55.61 (.044) 55.66 (.046) ADHD 59.19 (.048) 59.52 (.052) 53.04 (.051) 52.75 (.053) No ADHD 55.06 (.042) 55.13 (.043) 56.32 (.044) 56.48 (.045) Anxiety 66.22 (.055) 66.14 (.062) 59.71 (.058) 58.58 (.060) No anxiety 54.25 (.042) 54.24 (.044) 54.91 (.045) 55.14 (.046) Depression 51.27 (.054) 52.04 (.062) 45.76 (.057) 45.84 (.059) No depression 56.81 (.042) 57.09 (.044) 57.74 (.045) 57.82 (.046) ADHD treatment 70.33 (.048) 70.32 (.050) No ADHD treatment 54.18 (.044) 54.21 (.046) Anxiety treatment 66.24 (.057) 64.79 (.059) No anxiety treatment 54.93 (.045) 55.01 (.046) Depression treatment 70.16 (.057) 70.74 (.059) No depression treatment 54.78 (.045) 54.77 (.046) Counseling 52.64 (.075) 52.21 (.077) No counseling 55.82 (.043) 55.91 (.045) SE= standard error; All predictions came f rom gamma regression models on T able 4 25.
123 Table 4 30 Predictors of expenditure vs. no expenditure for inpatient and outpatient monthly asthma related services Inpatient Outpatient OR (CI) OR (CI) Model 1 ADHD .768 (.741,796)** .887 (.868,.907)** Anxiety 1.34 (1.28,1.40)** .946 (.917,.976)** Depression 1.36 (1.30,1.42)** 1.18 (1.14,1.21)** Model 2 ADHD .940 (.861,1.03) .971 (.923,1.02) Anxiety 1.34 (1.19,1.51)** .962 (.890, 1.04) Depression 1.54 (1.36,1.74)** 1.28 (1.19, 1.39)** Month 1.04 (1.04,1.04)** 1.05 (1.05,1.05)** ADHD *month .991 (.988,.995)** .996 (.994,.998)** Anxiety*month 1.000 (.995,1.01) .999 (.996,1.002) Depression* month .995 (.990,.9996)* .996 (.993,.999)* Model 3 ADHD .814 (.780,.850)** .925 (.901,.951)** Anxiety 1.28 (1.22,1.35)** 1.08 (1.04,1.12)** Depression 1.35 (1.28,1.42)** 1.11 (1.07,1.15) ADHD treatment .823 (.775,.875)** .907 (.875,.941)** Anxiety treatment 1.13 (1.04,1.23)** .641 (.603,.681)** Depression treatment .961 (.882,1.05) 1.29 (1.21,1.37)** Counseling 1.23 (1.17,1.29)** .941 (.908,.974)** Model 4 ADHD 841 (.805,.878)** .945 (.919,.972)** Anxiety 1.27 (1.20,1.34)** 1.11 (1.07,1.15)** Depression 1.35 (1.28,1.42)** 1.11 (1.07,1.15)** ADHD treatment .965 (.898,1.04) .989 (.948,1.03) Anxiety treatment 1.07 (.960,.1.19) .831 (.772,.897)** Depression treatment .927 (.832,1.03) 1.17 (1.08,1.25)** Continuous ADHD tr eatment .983 (.979,.987)** .990 (.914,.980)** Continuous anxiety treatment 1.00 (.998,1.01) .974 (.969,.978)** Continuous depression t reatment 1.006 (1.000, 1.013)* 1.01 (1.002,1.012)** C ounseling 1.23 (1.17,1.30)** .947 (.914,.980)** OR=Odds Ratio, CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01 Generalized estimating equations with binomial family and logit link with independent correlation controlled for age, gender, race, supplemental security income. Dependent variables are asthma specific (T able 3-1).
124 Table 4 31 Predictors of monthly inpatient and outpatient expenditure for asthma related services (assuming expenditure > $0) Inpatient Outpatient Predictors of Expenditure Beta (CI) Beta (CI) Model 1 ADHD -.035 ( -.177, .107) -.030 (-.088, .028) Anxiety .253 ( .082, .423)** .052 ( .031, .135) Depression .118 ( .057, .292) .045 ( .128, .038) Model 2 ADHD .070 ( .304, .164) .140 ( .238, .041)** Anxiety .170 ( -.140, .481) .039 ( -.118, .196) Depression .156 ( .160, .472) .089 ( .244, .066) Month .000 ( .005, .005) .003 ( .005, .001)** ADHD *month .002 ( .010, .013) .006 ( .002, .011)** Anxiety*month .005 ( .010, .020) .001 ( .006, .008) Depression* month .002 ( .017, .013) .002 ( .005, .009) Model 3 ADHD -.071 ( .243, .101) -.009 ( .080, .061) Anxiety .219 ( .008, .431)* .024 ( .070, .118) Depression .107 ( .100, .314) -.024 ( .117,.070) ADHD treatment .067 ( .149, .282) .051 ( .138, .037) Anxiety treatment .055 ( .219, .329) .094 ( .045, .233) Depression treatment .001 ( .279, .282) .081 ( .218, .056) Counseling .048 ( .148, .244) .025 ( .005, .080) Model 4 ADHD -.062 ( -.238, .113) -.013 ( -.085, .059) Anxiety .168 ( .049, .385) .003 ( .098, .091) Depression .135 ( .074, .345) .015 ( .109, .079) ADHD treatment .061 ( -.174, .295) -.060 ( -.155, .035) Anxiety treatment .015 ( .305, .275) .014 ( .169, .141) Depression treatment .062 ( .239, .363) .020 ( .173, .134) Continuous ADHD tr eatment .000 ( .020, .019) .002 ( .005, .090) Continuous anxiety treatment .019 ( .003, .042) .022 ( .010, .035)** Continuous depression t reatment .014 ( .040, .013) .011 ( .024, .001) Counseling .039 ( .157, .235) .021 ( .070, .113) CI= 95% Confidence Intervals; pvalue: <0.05; ** <0.01 Generalized estimating equations with gamma family and log link with exchangeable correlation controlled for age, gender, race, supplemental security income. Dependent variables are asthma specific (T able 3-1).
125 Table 4 32 Models 1 and 2 two part model predicted mean estimates of asthma -related inpatient expenditure ($) Model 1 Model 1 Model 1 Model 2 Model 2 Model 2 Mental Health Any Use Mean Expenditure Predicted Mean Any Use Mean Expenditure Predicted Mean All average use .090 3796.57 369.33 .090 3794.03 369.55 ADHD .075 3691.54 300.61 .076 3686.21 304.58 No ADHD .095 3821.53 391.97 .095 3816.25 391.88 Anxiety .111 4694.57 545.22 .111 4699.00 553.45 No anxiety .086 3645.54 328.83 .086 3631.80 326.45 Depression .112 4188.25 496.12 .114 4182.57 497.52 No depression .086 3722.53 338.63 .086 3725.15 339.67 This two part model uses predicted probability estimates of inpatient use from lo gistic regression (T able 4 30) and the predicted estimated inpatient monthly mean expenditure given expenditure>$0 f rom gamma regression (T able 4-31 ) to calculate predicted mean expenditure.
126 Table 4 33 Models 3 and 4 two part model predicted mean estimates of asthma-related inpatient expenditure ($) Model 3 Model 3 Model 3 Model 4 Model 4 Model 4 Mental Health Any Use Mean Expenditure Predicted Mean Any U se Mean Expenditure Predicted Mean All average use .090 3791.22 368.60 .090 3784.69 368.77 ADHD .078 3578.92 303.42 .080 3598.70 312.39 No ADHD .094 3843.33 389.88 .093 3829.87 386.50 Anxiety .108 4557.76 514.19 .107 4354.49 490.32 No anxiety .087 3659.85 333.20 .087 3681.45 337.08 Depression .112 4144.85 489.30 .112 4237.82 500.83 No depression .086 3724.47 339.33 .086 3701.70 337.68 ADHD treatment .077 4029.60 335.38 .088 4000.48 378.88 No ADHD treatment .092 3770.00 371.94 .091 3765.48 367.60 Anxiety treatment .100 3990.54 425.93 .096 3732.52 383.32 No anxiety treatment .090 3777.94 363.14 .090 3788.61 367.18 Depression treatment .088 3796.02 357.72 .085 4010.42 367.28 No depression treatment .091 3790.90 369.68 .091 3770.41 368.76 Counseling .107 3965.55 452.64 .107 3925.18 449.94 No counseling .089 3779.25 361.63 .089 3775.04 362.04 This two part model uses predicted probability estimates of inpatient use from logistic regression (T able 4 30) and the predicted estimated inpatient monthly mean expenditure given expenditure >$0 f rom gamma regression (T able 4 -31 ) to calculate predicted mean expenditure.
127 Table 4 34 Models 1 and 2 two part model predicted mean estimates of asthma -related outpatient expenditure ($) Model 1 Model 1 Model 1 Model 2 Model 2 Model 2 Mental Health Any Use Mean Expenditure Predicted Mean Any Use Mean Expenditure Predicted Mean All average use .394 319.09 126.12 .394 319.01 126.09 ADHD .373 311.33 116.42 .374 312.82 119.37 No ADHD .399 320.74 128.34 .399 319.45 127.43 Anxiety .383 333.90 128.32 .384 334.19 128.95 No anxiety .396 316.99 125.69 .396 316.71 125.52 Depression .425 306.96 130.98 .428 304.33 131.25 No depression .389 321.00 125.36 .389 321.05 125.29 The two part models use predicted probability estimates of outpatient use from logistic regression (T able 4 -31) and the predicted estimates of average outpatient monthly expenditure given some expendi ture>$0 from gamma regression (T able 432) to calculate predicted mean expenditure.
128 Table 4 35 Models 3 and 4 two part model predicted mean estimates of asthma -related outpatient expenditure ($) Model 3 Model 3 Model 3 Model 4 Model 4 Model 4 Mental Health Any Use Mean Expenditure Predicted Mean Any Use Mean Expenditure Predicted Mean All average use .394 319.10 126.06 .394 319.65 126.13 ADHD .380 316.55 120.66 .384 316.27 121.59 No ADHD .397 319.53 127.24 .396 320.36 127.13 Anxiety .408 325.72 133.37 .414 318.74 132.10 No anxiety .392 318.03 124.93 .391 319.79 125.18 Depression .414 312.56 129.87 .414 315.56 130.74 No depression .391 320.01 125.48 .391 320.28 125.41 ADHD treatment .374 304.49 114.34 .392 302.40 118.64 No ADHD treatment .396 320.25 127.07 .394 321.15 126.75 Anxiety treatment .305 348.96 106.79 .356 315.36 112.52 No anxiety treatment .399 317.66 126.94 .396 319.88 126.89 Depression treatment .448 295.29 132.65 .427 313.76 134.04 No depression treatment .391 320.20 125.70 .392 319.93 125.74 Counseling .381 326.69 124.94 .383 326.18 125.00 No counseling .395 318.56 126.12 .395 319.26 126.19 The two part model s use predicted probability estimates of outpatient use from l ogistic regression (T able 4 -31) and the predicted estimates of average outpatient monthly expenditure given some expendi ture>$0 from gamma regression (T able 432) to calculate predicted mean expendit ure.
129 CHAPTER 5 DISCUSSION Overview This study uses negative binomi al and gamma regressions to predict asthmarelated utilization and expenditures for three unique mental health conditions (e.g., ADHD, anxiety, and depression) and their treatment on Medicaid youth with asthma. The findings of this study are expanded for each specific mental heal th condition and the associations with each asthma-related outcome. These findings generally suggest that comorbid ADHD, anxiety, and d epression and the respective mental health pharmacological treatment have a wide range of associations and implications for the delivery of services for youth wit h asthma. The findings for each of the associations between ADHD, anxiety and depression with asthma-related services (e.g., inpatient, outpatient, medical, and pharmacological services) have strong policy implications with respect to the management and coordination of care among youth with asthma on Florida Medicaid. Future research, strengths, an d limitations of this study are addressed. Summary and Interpretation of Findings This study addressed gaps in the asthma and mental health literature by using several techniques to improve on the limitations of previous studies. In this study, health care utilization and expenditures was limited to asthma -specific use and expenditure. Additionally, asthma was defined conservatively as having least 2 IC D -9 asthma diagnoses within the first year of the study This study uses a large sample (n=8,241) of continuously enrolled Medica id youth with asthma from the state of Florida. This cohort was analyzed over a 36 month period through a longitudinal design. The main feature of
130 this study was to bring specific attention to individual mental health diagnosis of ADHD, anxiety, and depression and the impact on asthma-related utilization and expenditures. Additionally, the study addressed the need to look at the p harmacological impact of mental health treatment on the underlying relationship between mental health and asthma-related care. This study implies that treatment for psychiatric conditions may have an impact on asthma related care. Generally, t his study sought to achieve 3 objectives which include: 1 To understand the impact of 3 unique mental health conditions as a modifier of asthma-specific utilization and expenditures 2 To understand the association of the number of months with a mental health condition on the use and expenditure of asthma-specific services 3 To understand the impact of mental health treatment for 3 unique mental health conditions as a mediator of asthma-specific utilization and expenditure s The below discussion sections addres s these three objectives by mental health diagnosis and asthma -related services ADHD and Asthma-Related Services While other studies have not found any association between childhood asthma and ADHD (Daly et al., 1996), this study reports a large sample of youth with asthma that have a high prevalence of ADHD (i.e., 18.5% of this sample has ADHD). The prevalence of ADHD in Florida is about half that observed in this study at 9.2% (CDC, 2005), suggesting a higher prevalence of ADHD in this Medicaid population of continuously enrolled youth with asthma. Since youth with asthma have an increased likelihood for obstructive sleep apnea, assessing each youth with asthma for sleep disordered breathing has been suggested ( Bousquet et al., 2008; et al., 2009; Owens, 2009).The issue of sleep disorders being misdiagnosed as ADHD
131 becomes an important consideration to explain the underlying higher ADHD prevalence observed in this study. Findings from this study suggest asthma -related services are used differ ently among those with ADHD compared to those without ADHD In this study, the hypothesized effect for youth with comorbid ADHD was to observe a decrease in asthma-specific use and expenditure. However, this was not universally observed across all asthma related services. Generally, having comorbid ADHD was associated with significant decrease in inpatient monthly admissions and expenditure s. This was consistent with earlier studies by Jamoom and colleagues (2010). However, ADHD was associated with a signifi cant increase in asthma -related medical claims, and asthma related pharmacy claims, including total, direct, controller, and indirect pharmacy claims suggesting better adherence to asthma med ications and coordination of asthma care The associated asthma -related prescription expenditure s were higher for those with ADHD compared to those without ADHD, and this may be indicative of youth with ADHD being high users of pharmacy claims as previous ly described (Chen et al., 2002). However this can also relate to better adherence and better coordination of care (Katon et al., 2005). Additionally, differences in ADHD adherence has been observed (Gau et al., 2008), and this may ultimately impact asthma treatment adherence based on different factors associated within the Medicaid population. Poor adherence in childhood with ADHD has been associated with older age, later onset of ADHD, family history of ADHD, higher paternal education level, and multi -dos e administration (Gau et al., 2008) This study did not address many of those characteristics due to the nature of claims data and very different sample characteristics (e.g., Taiwanese population)
132 Another rationale for the improved adherence and asthma management relates to the continuity of care within the medical home for those with ADHD. Pediatricians are more likely to treat ADHD then refer them outside of their medical home (Stein et al., 2008) Specifically in a recent sample of 659 pediatricians 70% agreed that pediatricians should be responsible for treating and managing ADHD. Therefore, pediatricians are not only more involved in the treatment of asthma but their patients overall ADHD treatment as well. This may be a very different experience f or other mental health conditions that physicians are more likely to refer to a specialty mental health clinic (e.g., depression, anxiety) Those with ADHD and asthma are, in theory receiving an overall higher quality of care for the management of their a sthma and ADHD by not necessarily having to leave their medical home A temporal effect was observed for those with ADHD such that the number of months one has an ADHD diagnosis is associated with an increase in medical claims, controller medications, and inpatient and outpatient expenditures The general trend of the longer one has ADHD, the more asthmarelated medical, controller, and direct pharmacy claims. This increase in utilization was described in treating ADHD (Guevara et al. 2001 ), and potential ly may extend to ward beneficial asthma treatment (Kewalramani et al. 2008 ). Also, there was a decreased propensity to use inpatient services. These findings suggest t hat those with comorbid ADHD have improved coordination of care, better management and ad herence to their asthma regimen. T he positive effects for ADHD generally may be a combination of more exposure to the medical environment more pediatrician ownership for the treatment and management of the ADHD and asthma rather than referring ADHD to psy chiatric services where the
133 asthma may not be addressed with the mental health condition treated within a medical home, and better pharmacological management o f ADHD and asthma medications (Guevara et al., 2001; Kewalramani eet al., 2008; Stein et al., 2008 ). ADHD was associated with significant reduction in total expenditure when treatment variables were included. Treatment of ADHD was associated with an overall effect of having more medical claims and more pharmacy fills across all categories. The continuous use of ADHD medications was associated with fewer inpatient admissions. ADHD treatment diminishes the effects of ADHD on monthly asthma related total expenditure s and all asthma related pharmacy expenditure For example, without ADHD treatment, ADHD w as associated with an increase in asthma related pharmacy expenditure s compared to those without ADHD. However controlling for ADHD treatment resulted in a s ignificant reduction in total asthma-related pharmacy expenditure effectively mediating asthma specific pharmacy expenditure for ADHD. Other treatment effects that require attention include potential protective effects of ADHD treatment. Further understanding into the medications used in ADHD and the impact on a sthma control can help explain some of the improved care observed in this study ADHD treatment was associated with an increase in medical claims and all pharmacy claims, and the increase pharmacy utilization while not hypothesized, makes logical sense due to increased access to pharmacy based on better continuity of care (i.e., ADHD and asthma controlling medication) and physician coordination within the medical home. This also is consistent with the ADHD literature of increased pharmacy utilization (Chen et al., 2002; Guevara et al. 2001 ).
134 A nxiety and AsthmaR elated Services Findings from this study suggest that youth with anxiety use asthma-related services differently than those without anxiety. As hypothesized, anxiety was associated with greater asthma-related utilization and expenditures across all asthma -related service categories. The rationale for the increase in all asthma -related use and expenditure categories is multifaceted. Having used more asthma -related inpatient and urgent care is indicative of poorly coordinated and controlled asthma However, high asthma-related pharmacy fills and expenditures imply asthma medication adherence is not necessarily an issue. Hence, the larger problem may relate to coordination of care. C oordination of asthma and anxiety management relates directl y to the continuity of care within and outside the medical home for those with anxiety and asthma. Pediatricians are less likely to treat anxiety and refer youth with anxiety to psychiatrists as they do not believe it is part of their responsibility to tre at anxiety (Stein et al., 2008). The result is psychiatric care that is done outside of the medical home. Specifically in a recent sample of 659 pediatricians, 29 % agreed that pediatricians should be responsible for treating and managing anxiety disorders Therefore, it can be assumed that pediatricians are only involved in the treatment of asthma not the anxiety. The coordination of management from providers outside of the medical home becomes even more important in the context of the role anxiety plays fo r the management of asthma Hence, coordination and continuity of care for those with anxiety and asthma requires further insight and context into how those conditions are m anaged Th is study finds that th ose with anxiety and asthma are, in theory, receivi ng care (e.g., higher general use and expenditure) H owever the quality of care for the management of their asthma and anxiety may not necessarily be the best as they are likely to be leaving their medical
135 home for anxiety care resulting in asthma coordination and management problems This study did not control for where anxiety and asthma health services were processed. Hence, this study did not address whether asthma was being assessed in the psychiatric facility or in the pediatricians office and is an area for future research. The role of anxiety in comorbid medical illness has been suggested as something not to ignore (Roy -Byrne et al., 2008). This study agrees with other emerging data suggesting anxiety disorders rival depression in terms of risk, comorbidity and outcome. Using more services and having greater expenditure s suggests that this population should be monitored closely by physicians with better coordination of care and routine follow up The increase in services and expenditure s for youth with comorbid anxiety may be indicative of greater severity of disease pathology. Since the study did not control for asthma severity influencing utilization, it is recommended that future studies include a measure of severity to further understand why this population of youth with asthma that are using significantly more services and expenditures. There were significant t ime effects for those with anxiety on monthly total, direct, and rescuer asthma-related pharmacy utilization These effects suggested inc reased pharmacy utilization is due, in part, to the months of having an anxiety diagnosis The longer one has anxiety t he greater the general utilization and expenditure. The implications for length of time with anxiety on asthma care suggest physicians and patients alike may benefit from a review of anxiety and asthma protocols as they relate to improved long term pharmacological management and coordination to reduce negative asthma outcomes
136 For example, anxiety treatment was a ssociated with more asthma-related inpatient care. This association may relate to anxiety treatment exacerbating asthma, disease severity, potentially harmful pharmacological interactions, or poorly managed, coordinated asthma care (Roy Byrne et al., 2008; Stein et al., 2008). However, there may be drug interactions, or potential side effects that are unable to be addressed by this study. Specifically, such pharmacological effects may occur over time to influence the biochemical pathways by which these drugs operate ( Kewalramani et al. 2008; Pretorius, 2004). For example, taking SSRIs or psycho-stimulants used to treat depression, anxiety, and ADHD and may impact up-regulation of serotonin levels and possibly trigger asthma attacks (Pretorius, 2004). The relationship between asthma and anxiety disorders suggest that psychopharmacological might improve asthma control. Anxiety treatment resulted in d iminished effect for anxiety across all asthma-related utilization and expenditure signifying anxiety tre atment helps to mediate asthma utilization and expenditure to an extent for those with anxiety The difference may be statistically significant but of little practical or clinical significance. These findings suggest that despite more interaction with th e medical environment their anxiety contributes to high asthma-related service use and expenditures. Strong increased expenditures and utilization patterns in this population require more careful attention to their chronic disease management A focus shoul d, therefore, be on decreasing asthma-related inpatient care. Depression and Asthma -R elated Services F indings from this study suggest that those youth with depression use asthma related services differently from youth without depression. However t he original
137 hypotheses were met with mixed findings. As hypothesized, depression was associated with significantly greater monthly inpatient admissions, which is consistent with other studies (Mohammed et al., 2006; Richardson et al, 2008). However, those with comorbid depression used significantly less asthma maintenance services (e.g., medical claims, pharmacy claims). Youth with comorbid depression incurred significantly less asthma related monthly medical claims and used significantly less total, direct, and res cuer pharmacy claims compared to those without depression. The combination of having significantly greater inpatient admissions, while having lower total monthly asthma -related pharmacy expenditure s compared to those without depression and lower asthma -rel ated pharmacy prescriptions fill ed may be due to poor adherence, poor coordination of depression and asthma management, or differences in asthma severity. Therefore, this study suggests that those youth with depression like other instances in the literature of comorbid depression in chronic disease (e.g., hypertension, diabetes) may be experiencing poor adherence to asthma treatment regimens, evidenced by higher monthly inpatient admissions compared to those without depression. Other possibilities include that youth with asthma that have comorbid depression are sicker. Since this study does not control for asthma severity or depression severity this currently cannot be assessed and is discussed further as a limitation. The number of months with depression diagnosis was assessed for each of the asthma-related outcomes The number of months with depression was associated with fewer asthma-related medical and pharmacy claims with every additional month of
138 having a depression diagnosis. Therefore the longer one ha s depression, the more likely the youth with depression is not filling his/her asthma medications resulting in potentially poor adherence Perhaps there is an initial founder effect where the doctor diagnoses the mental health condition during the yout h with asthmas well -check visit. Since most pediatricians do not feel it is their responsibility to manage and treat depression (Stein et al., 2008), they refer the treatment of the youth with asthmas depression to a psychiatric clinic outside the medica l home. Therefore, if the asthma is addressed through routine well -check visit within the medic al home, and the psychiatrist only focuses on the depression, then coordination of care and ultimately the management of the asthma may suffer over time. The coordination and management of both the depression and asthma are poor due to being outside of the medical home and having poorly coordinated mental health and asthma services Another possibility was that the depression was either more acute in nature to result in a decrease in utilization over time However, coupled with the first hypothesis it may reflect poor adherence and a trend towards less interaction with the medical environment (e.g., less routine care) such that when such an adverse event o ccurs it yields initial costlier care (e.g., increased incident inpatient admissions). In general, receiving treatment mediated the asthma -related use and expenditure for those with depression. Specifically, depression treatment diminishes the effect of de pression on i npatient admissions as hypothesized. D epression treatment (both presence of and continu ous treatment) was associated with increased medical claims and more asthma related pharmacy expenditure and utilization Therefore suggesting depression t reatment was associated with better asthma management and potentially
139 better adherence Depression treatment was also associated with reduced urgent care visits (IRR=.655, p<0.01) suggesting that depression treatment has some protective aspects for reducing the need for urgent care De pression treatment was a ssociated with increased asthma-related total, direct, rescuer and controller pharmacy claims suggesting increase in utilization for depression treatment improves adherence or at least filling of asthma related medication. However, coordination of depression and asthma management may also be a concern. Pediatricians are less likely to treat depression and refer youth with anxiety to psychiatrists as they do not believe it is part of their responsibility to treat anxiety (Stein et al., 2008). The result is psychiatric care is done outside of the medical home. Specifically in a recent sample of 659 pediatricians, 25% agreed that pediatricians should be responsible for treating and managing child/adolescent depression (Stein et al., 2008). Similarly experienced with anxiety, pediatricians are only involved in the treatment of asthma not the underlying mental health condition (e.g., depression). The coordination of management from providers outside of the medi cal home becomes even more important for context of the role depression plays for the management of asthma. Again, this was a very different experience for those with ADHD who are treated for their ADHD and asthma within the same medical home and have apparently better asthma management. Hence, better disease management requires better coordination and continuity of care for those with depression and asthma. Asthma-R elated Pharmacy Claims and Expenditure s This study assesses large scale impact of depressio n, anxiety, and ADHD impact on asthma-related pharmacy claims. Generally, anxiety and ADHD have higher asthma related monthly pharmacy utilization. However, there may be multiple effects which
140 relate to the success of asthma management between the two ment al health groups. Asthma medications may have adverse impacts for those with anxiety and depression (Pretorius, 2004), whereas those with ADHD may benefit from stimulants (Daly et al., 1996). The implication of depression being associated with less asthma-related pharmacy utilization may ultimately lead to higher cost inpatient care due to poor medication adherence (Katon et al., 2005; DiMatteo et al., 2000; Cramer et al., 1998) This finding may require further understanding as many other factors may be i n play with respect to adherence to asthma related treatment for youth with asthma, from specific drug interactions to parental behavior with respect to access. However the literature suggests depression is associated with poor adherence to asthma treatment (Katon et al., 2005; DiMatteo et al., 2000; Cramer et al., 1998). The other interesting finding is that treatment for mental health was associated with increase in asthma medication use suggesting youth with asthma receiving mental health treatment are more likely to fill asthma medications. Additionally, youth with comorbid anxiety have higher asthma -related pharmacy claims leading to greater reliance on rescuer and controller medications, suggesting an issue of coordination of care. Use of rescuer me dication and controller medications were assessed in this study. Rescuer medication use is meant for temporary relief for mild to moderate asthma. Most youth with asthma that have more moderate to severe asthma rely on controller medications (Rodrigo & Rodrigo, 2002; Blair et al., 2008). There is significantly more controller use for youth with comorbid ADHD. The use of controller medications may
141 help in stabilizing asthma and reduce the chance for an asthmatic episode needing medical intervention. The aspe cts of ADHD that may be potentially interesting to explore include what specific aspects of regular use of ADHD medications contribute to the protective effect of reducing the odds of having inpatient use (OR=.983). Policy and Research Implications The im plications for this study involve understanding a new role that mental health and mental health treatment play in chronic disease management for youth with asthma. ADHD was associated with better adherence and this may be a result of better coordination o f care within the medical home. ADHD may also be a misdiagnosis for obstructive sleep apnea or sleep disordered breathing, which may result in many similar diagnostic criteria symptomology observed in ADHD (Owens, 2009). It is important to consider why thi s population had twice the prevalence of ADHD compared to the rest of the state of Florida. Additionally, 50% of the original sample didnt meet continuously enrolled criteria, and that may represent a large set of youth that have a variety of different c haracteristics from this study sample of continuously enrolled Medicaid youth with asthma. Having gapes in coverage may result in less coordination and continuity of care for these youth. Further study that accounts for discontinuity in Medicaid coverage m ay be warranted Coordination of care seems to be a really important implication from this study. As we uncover the relationship in utilization for mental health and chronic disease we can start to understand the importance of coordination of mental health treatment on the overall management of asthma. C oordination of care is better in theory for those with
142 ADHD, however when doctors are referring ou t mental health conditions like depression and anxiety the problem becomes a silo or possible lack of coordi nation between the two providers; where the psychiatrist is responsible for that patient s mental health treatment, and the pediatrician is responsible for the asthma management. The problem is the mental health and asthma may be viewed as mutually exclusi ve problems. However, they must be treated in context (Kewalramani et al., 2008). I f both the pediatrician and the psychiatrist are not coordinating the asthma and mental health management, the care suffers. Some have suggested that many pediatric asthma h ospitalizations might be prevented if parents and children were better educated about the childs condition, medications, and the need for follow -up care and avoiding known disease triggers (Flores, Abreu, Tomany -Korman, & Meurer, 2005; Coffman, Cabana, Ha lpin & Yelin, 2008). A key issue is how many parents are aware of potential interactions between the medications for asthma and treatment for mental health conditions. ADHD, anxiety, and depression symptomology have not received much attention in the literature since smaller studies assessing medication in ADHD and asthma were completed by Daly and colleagues (1996). This study encourages more research and attention to larger studies like this one to assess the impact of asthma specific medications on both mental health and asthma related outcomes. Treatment tends to be associated with improved asthma-related care as use tends to exhibit diminished effects However the specific medication interaction and other multiple factors involving education on the tri ggers that exacerbate ones specific
143 asthma may help to reduce costly care (Flores et al., 2005). Some side effects of some asthma medications, and the potential interactions associated with SSRIs may trigger potential asthma attacks, and this specific asp ect could use more attention by the medical and pharmacological communities (Pretorius, 2004) This study supports at least further research to elucidate the actual pharmacological medication interactions for those with asthma and those that have depression, anxiety, and ADHD. Study strengths. This study has fulfilled the call to address the impact of mental health and mental health treatment on asthma-related utilization and expenditure (Blair et al., 2008; Kewalramani et al., 2008; Roy -Byrne et al., 2008). The strengths of this study include being one of the first large longitudinal studies to address mental health and mental health treatment in a large chronic condition, childhood asthma. This studys findings are generalizable to y outh with asthma that are continuously enrolled in Florida Medicaid and these findings extend to most national Medicaid populations. Additionally, the 36 month cohort of more than 8,000 youth with asthma was continuously enrolled in Medicaid suggesting th at all had similar access to asthma -related services. This study also addres s ed specific mental health associations w/ the delivery of asthma services, and improves on other methods through using a longitudinal approach. This study also improved upon earl ier studies by including robust methods and cohort definitions for asthma Claims data represent a good measure of service use and expenditure s to analyze the objectives of this study and obtain good estimates of expenditure and utilization for this popula tion. Furthermore all expenditure and utilizatio n were attributed to asthma through ICD 9 coding and national standards from
144 the National Center for Quality Assurance to define mental health as well as mental health and asthma treatment variables. The defi nition of asthma for this cohort was having 2 ICD 9 codes for asthma within the first 12 months of the study. This strict definition allows for better identification of youth with asthma to avoid false positives in this sample The definition of mental health was based on the concept of a trigger allowing for mental health diagnosis at any month in the 36 month observation period. This method provides a 3 year opportunity to assess mental health as it is diagnosed in a cohort of youth with asthma. This stu dy also brings more attention to policy relating to mental health and asthma. Specifically, this study reported a strong association between ADHD and asthma that has not been observed in the literature Additionally, poor adherence to asthma treatment was suggested for youth with asthma that have depression, and this study adds to the consistency of the literature (Katon et al., 2005). Coordination and management of care issues were also suggested by the general findings for youth with asthma that have anxi ety. Future Research Future research is needed to address the main findings of the study. First, there is a need to understand how differently ADHD, anxiety, and depression are coordinated with the pediatrician who manages the asthma care. Understanding th e difference in coordination and referral practices helps to understand the effect of a medical home on asthma-related care. Second, future research needs to address the cause of asthma medication adherence issues among those with depression. Strategies f or improving adherence in asthmatic youth with depression are also indicated. This may involve understanding the parental role in compliance and ensuring adherence to treatment
145 regimens. Third, a sthma and mental health severity need to be addressed with re spect to each asthma-related outcome. Understanding the relationship between asthma and mental health severity would eventually help to quantify the effect of severity on asthmarelated services. Moreover, c omplex m edication interactions between asthma and drugs to treat mental health (e.g., SSRIs, ADHD medications) need to be assessed through wide scale studies. Additionally, influence of behavioral, environment al and parental factors on asthma -related utilization and needing to be addressed in future studies. Lastly, s pecific types of asthma need to be assessed for these observed findings as different etiologies of asth m a (i.e., exercise-induced, intrinsic, allergic asthma ) may have very different associations wi th mental hea lth and mental health treatment (Blair et al., 2008) Limitations of S tudy While strengths of this study are remarkable, there are several noteworthy limitations that should be addressed. Generaliz ability. The generalizeability of this studys findings are specific to a continuously enrolled sample of Florida Medicaid youth with asthma. While most likely able to be expanded to other continuously enrolled Medicaid youth with asthma, these findings are not generalizable to those discontinuously enrolled Medicaid youth. Over half of the original sample (n=38,607) was excluded because they did not meet the definition of continuous enrollment, and that may represent a potential problem to external validity. A large number of youth with asthma were not included in the sample based on not meeting criteria for continuous eligibility or meeting the string definition of asthma within the first 12 month period of the study. Addressing limitations of large intermittent
146 periods between enroll ment, or enrollment gaps, represent a challenge in working with claims data. Using the inclusion criteria of no more than 62 days of gap at a time ensured that a basic definition of continuous enrollment was met. Using at least two asthma diagnoses within the first year allowed for more confidence in reducing the number of false positives. However, this increases the chance of removing those youth with mild asthma and forming a more severe cohort. Results may not be generalizable to people with gaps in coverage who may have different help seeking behavior. Additionally, a sthma -specific triggers may be different with in each state with environmental fluctuations in air quality indicators associated with triggering asthma (e.g., ozone) (Blair et al., 2008). E nv ironmental triggers are unlikely to be correlated with mental health diagnosis as mental health diagnosis tends to be equally distributed. For this study, mental health is consistent across all counties regardless of environmental exposures Therefore, controlling for environment was not necessary. Asthma severity. Th e impa ct of asthma severity on asthma-related use and expenditure was not fully controlled for in this study. While the supplemental security income was represented as a measure of comorbidity and disability, it does not capture asthma severity. A n accepted algorithm to calculate asthma severity was not attempted with this specific claims data. However, studies using a measure of self -perceived asthma severity had significant associations with depressive symptoms than studies that used an objective measure (Opolski & Wilson, 2005). Hence, that may have risk adjusted some of the effect associated with depression. Using a proxy measurement of asthma severity has been noted in the literature ( Richa rdson et al., 2006; Birnbaum et al., 2009), using controller and rescuer medications to partly obtain an objective
147 measure of asthma severity Specifically assessing the issue of asthma severity, anxiety and depression were associated with increased asthma symptom burden (Richardson et al., 2006). Hence using anxiety and depression as a proxy for asthma severity would not have been realistic. This study assumes that asthma severity is equally distributed across all mental health conditions. However, the obv ious limitation is that the assumption is not correct, as asthma s ympt om burden has been associated with anxiety and depression. Since the majority of utilization models used a fixed effect model, we were able to account for individual variation rather tha n variation between individual s which reduces some of the impact of not controlling for asthma severity. However, random effects models were used to calculate utilization for inpatient admissions and length of stay. The inpatient outcomes thereby accounted for variation between individuals as well, which led to potentially more efficient estimates. Both random and fixed effects allow for consistent estimates of utilization for each mental health condition and its treatment. Claims data. Using claims data w hile a powerful approach for assessing utilization and expenditure has often a high number of false positives and negatives associated i n the coding process. The diagnostic and treatment variables were constructed through pr oxy ICD 9, NDC, and TCC coding which is not necessarily an ideal characteristic of these data. Using a fixed approach reduced some omitted variable bias as these models do not measure differences between individuals, but measure within individual variation holding all variables fixed w ithin samples, among demographic and other control variables (e.g., personal medical histories ) For random effect analyses the need for
148 more control variables helps to reduce the chance for correlation between t he explanatory variables and any unobserved heterogeneity. This provides for more efficient estimates when there is no correlation between the two. However some omitted variables can include parental use, behavioral trends (e.g ., smoking), and asthma severity Mental health classification The focu s of this study was on the relationship between individual mental health diagnosis and the impact of mental health treatment on Medicaid youth with asthma. The severity for each mental health condition assessed in this study was not quantified or controlled for creating some limitation. Additionally, flags for mental health condition were triggers from the moment a diagnosis occurred. Generally, measuring mental health remains a challenge, as mental health tends toward being underreported in claims data, representing another limitation of this study. However, this study did not adjust or control for those who had an acute mental health episode, effectively threatening the internal validity for the trigger for each mental health condition. Essentially, acute mental health episodes were coded as having that condition and an early enough episode of ADHD, anxiety, and depression would dilute the association if the acute mental health condition resolved itself. Therefore, the internal validity of each mental healt h variable may be threatened in some capacity. Effects of this sort of bias would dilute the effects for those youth with asthma that have chronic ADHD, anxiety, and depression. The mental health trigger for depression, anxiety and ADHD represent one appr oach to assess the impact of causality on the impact of mental health on asthma use and expenditure s. The limitation with this method involves the omission of a
149 preexisting condition prior to the first diagnosis of depression, anxiety, and ADHD for this st udy. Hence, depression, anxiety, and ADHD may be underreported. Pharmacological classification and treatment variables. HEDIS supported NDC codes obtained from NCQA were compiled for pharmacological treatment for depression, ADHD, and asthma medications ( e.g., anticholinergics, rescuer and controller medications) were used to create asthma-related pharmacy and mental health treatment flags. However all treatments for depression, for example, may not have been identified through claims data. Additionally fo r mental health treatment, claims data underreports mental health counseling. Maturation effects may be diluted due to mental health treatment visit and diagnosis incongruity, however using a depression, anxiety, and ADHD trigger attempted to correct for t his possibility Asthma utilization is complex, multi-factorial Asthma expenditure and use in this study are based on the claims coding process. All diagnostic codes in the claims for one event may not truly reflect the full extent of use and expenditure attributed to the specific medical condition coded in the claims. Hence, coding for the diagnosis of asthma may have just been briefly discussed relating to asthma during the visit and disguise the true nature of visit (e.g., ADHD management). For example if asthma was a primary, secondary, or tertiary diagnosis code in the claims, the expenditure and use associated with the visit may not be unique to asthma, other codes listed for that specific event in the claims data. Additionally the order of the diagnosis may not reflect order of importance based on different physician attention to coding details. This study does not discriminate utilization or expenditure by primary, secondary or tertiary order diagnosis coding for asthma or other conditions.
150 This st udy therefore, assumes equal weight to all ICD 9 codes of asthma present in the claims files. Additionally while considering the primary explanatory variables, other variables were not included in this study to account for utilization Asthma is multifact orial and etiology of asthma often depends on interactions between genetic susceptibility and environmental factors. Susceptibility genes include those involved in immunity and smooth muscle, fibroblast proliferation, or cytokine production regulation immu nity (e.g., T -helper, cytokines, ADAM33 gene, tumor necrosis factor ). Often impacted by triggers of asthma, which may have a strong environmental cause include allergen exposure (e.g., dust mite, roach, pets), diet deficiencies (e.g., low vitamin C and v itamin E), perinatal factors (e.g., young maternal age, poor maternal nutrition, prematurity, low birthweight, and lack of breast feeding) (Blair et al., 2008). While these environmental and genetic factors are specific to asthma, this study uses claims da ta which does not control for these other causes of asthma. Asthma utilization is also multifactoral relying on population characteristics that are predisposed in the population like gene environment interactions or mental health condition or involve enabling resources which may include presence of insurance to transportation to the clinic, family structures, parental education and perceived need. The Andersen model would suggest that more such variables that are unavailable from claims data are needed to fully appreciate the predicted model of use for mental health utilization in youth with asthma (Andersen, 1995) This study did not use parental utilization or parental characteristics in adjusting for asthma related utilization. Often mental health condi tions like depression tend to run in families, parents are enablers of
151 health seeking behavior, and help their child in navigating through the medical environment Additionally, parents are charged with carrying out medication regimens. This study did not address parental involvement in medication adherence and potential influence in utilization. If children are depressed, parents may be as well. Family histories may play a role in better physician awareness. Children with depression may be less likely to c ommunicate needs to psychiatrist relating to asthma and vice versa with the pediatrician, resulting in a failure of coordination of care. Argues for the inclusion of psychiatric services within the medical home concept, as is exemplified by the treatment f or ADHD. Asthma and psychological health. Bender and colleagues (2000) suggest that mild to moderate asthma has imposed modest effects on daily life but explicitly state not the psychological health in youth with asthma. Hence, there is a literature that exists that psychological health is not altered in youth with asthma. However, multiple studies including this one, have suggested and quantified a higher association of psychiatric diagnosis and differential utilization among such a group. This study goes even as far as suggesting a temporal relationship between different mental health conditions and asthma-specific use. Lastly, claims data represents a powerful approach to understanding expenditure and utilization pattern s However, this kind of analysis with claims data does contain false positives (e.g., having a n asthma ICD -9 diagnosis code in the claims yet does not have asthma) and false negatives (e.g., having an asthma diagnosis but not coded in the claims data ) whe n used for proxy measures of diagnosis. However the false negatives and positives should be randomly distributed. Data from claims are only as
152 good as the provider and clinic that entered them into the system and are an inherent form of bi as within this st udy. Physi cians and clinics have multiple reimbursement incentives and practices that can influence the diagnostic coding process. Therefore, the false positives or those with an incorrectly coded ICD 9 diagnosis for asthma, depression, anxiety, ADHD, and false negatives or those that did not actually have the condition represent a limitation of working with claims data. This study attempted to control for this by making sure that at least 2 ICD -9 codes of a specific condition were used in identifying asthma, a main criteria for sample eligibility. While assessing utilization and expenditure through this sort of modeling is an extremely powerful and flexible method to analyze longitudinal data, assumptions are made that asthma is not t riggered by any other exposure. Asthma triggers include allergen exposure, diet, infection, exercise, inhaled irritants, emotion, aspirin, and GERD (Blair et al. 2008), yet not all trigger s were taken into account by these model s. Conclusion This study looks at the effects of individual mental health conditions and the impact of mental health treatment on asthma use and expenditure. ADHD, anxiety, and depression modified the delivery of asthma related services for Medicaid youth with asthma in Florida. This study represents the first large scale study to address the issue of mental health and mental health treatment on asthma related utilization and expenditure using a continuously enrolled cohort of Florida Medicaid youth with asthma. ADHD was associated with less inpatient care, yet more asthma-related pharmacy care and medical claims suggesting better management and coordination of care. Depression was associated with more admissions, yet less asthma-related pharmacy care suggesting poor adherence to asthma medications Anxiety was associated with
153 more inpatient, outpatient, medical, and pharmacy care, suggesting poor management and coordination of asthma-related care. Treatment of ADHD and depression was associated with improved asthma-related pharmacy care and medical claims, and less inpatient care. Treatment of anxiety resulted in more expenditure and utilization for all categories of asthma care, including inpatient ca re Policies to address include close monitoring of youth with anxiet y and depression treatment may be warranted to ensure youth with ast hma with comorbid psychiatric conditions have better asthma outcomes to manage their asthma in presence of mental health comorbidities. In addition to addressing strategies to improve adhe rence to asthma treatment for youth with depression, continuity of care through medical homes require s more attention for this population. The issues associated with anxiety and depression being treated outside of the pediatrician office may result in poor coordinated and managed care for youth with depression and anxiety. Those with comorbid ADHD had better control of their asthma than those with anxiety and depression. This study sought to address utilization and expenditure among ADHD, anxiety, and depr ession, findings suggest 3 very different experiences with asthmarelated services.
154 LIST OF REFERENCES Alati R OCallaghan, M Najman J M Wlliams G M Bor W & Lawlor D A. (2005). Asthma and internalizing b ehav ior problems in adolescence: A l ongitudinal s tudy. Psychosomatic M edicine 462470. Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care: Does it matter? Journa l of Health and Social Behavior 36 1 -10. Baron R M & Kenny D A. (1986). The moderator m ediator distinction in social psychological research: Conceptual, strategic and statis tical considerations. Journal of Personality and Social Psychology 51 11731182. Becker, K. G. (2007). Autism, asthma, inflammation, and the hygiene hypothesis. Medica l Hypotheses 69 (4), 731 -7 40 Bender B. G. (2007). Depression symptoms and substance abuse in adolescents with asthma. Annals of Allergy, Asthma and Immunology, 99 (4) 319 -3 24. Bender, B. G., Annet, R. D., Ikle, D., DuHamel, T. R., Rand, C., & Strunk R. C. (2000). Relationship between disease and psychological adaptation in children in the childhood asthma management program and their families. Archives of Pediatric and Adolescent Medicine, 154 706-713. Bender, B., & Milgrom, H. (1992). Theophylline -induced behavior change in children. An objective evaluation of parents perceptions. JAMA, 267 26212624. Biesanz, J. C., Deeb -Sossa, N., Papadakis, A. A., Bollen, K. A., & Curran, P. J. (2004). The role of coding time in estimating and interpreting gr owth curve models. Psychological Methods, 9 30 -52. Birnbaum, H. G., Ivanova, J. I., Yu, A. P., Hsieh, M., Seal, B., Emani, S., et al. (2009) Asthma severity categorization using a claims -based algorithm or pulmonary function testing. The Journal of Asthm a, 46, 6772. Blackman J A & Gurka, M J. (2007). Developmental and behavioral comorbidities of asthma in children. Journal of Developmental and B ehaviora l P ediatrics, 28, 92 99. Blair, R., Br eit, J. S., & Berkow, R. (2008). Asthma. In The Merck Manuals Online Medical Library (section 05, chapter 48). Retrieved from http://www.merck.com/mmpe/sec05/ch048/ch048a.html Bloomberg, G R & Chen, E. (2005). The relationship of psychological stress with childhood asthma. I mmunology and Allergy C linics of North America, 25(1), 83105.
155 Bousquet J Bousquet P J Godard P & Daures, J P. (2005). The public health implications of asthma. Bulletin of the World Health Organization, 83 5485 54. Bousquet J Khaltaev N Cruz A. A., Denburg, J. Fokkens, W. J. Togias, A., et al. (2008). Allergic Rhinitis and its Impact on Asthma (ARIA) 2008 update (in collaboration with the World Health Organization). Allergy 63( Suppl 86):8 160. Bousquet J., Ndiaye, M., Ait -Khaled, N., Annesi -Maesano, I., & Vignola, A. M. (2003). Management of chronic respiratory and allergic diseases in developing countries. Focus on sub Saharan Africa. Allergy,58, 265-283. Braman, S S. & Hanania N A. (2007). Asthma in older adults. Clin ics in Chest Med icine 28 (4) 685-702. Brown E. S. Suppes T Khan D A. & Carmody T J ., 3rd. (2002) Mood changes during prednisone bursts in outpatients with asthma. Journal of Clinical Psychopharmacology ,22 (1), 5561. Brown E. S. Vera E. Frol A. B. Woolston D J & Johnson B. (2007). Effects of chronic prednisone therapy on mood and memory. Journal of Affective Disorders, 99 (13) 2792 83. Bussing R Burket R C & Kelleher E. T. (1996). Prevalence of anxiety dis orders in a clinic based sample of pediatric asthma patients. Psychosomatics 37 108-115. Carr R E. (1998). Panic disorder and asthma: causes, effects and research implications. Journal of Psychosomatic Research, 44(1), 4352. Centers for Disease Control and Prevention. (2005) Prevalence of Diagnosis and Medication Treatment for Attention-Deficit/ Hyperactivity Disorder United States, 2003. Morbidity and Mortality Weekly Repor t, 54 (34), 842-847. Chan, E Zhan C & Homer C J. (2002) Health care use and costs for children with attention-deficit/hyperactivity disorder: national estimates from the medical expenditure panel survey. Archives of P e diatrics & Adolescent M edicine. 2002; 156(5):5045 11. Cheong, J MacKinnon D P., & Khoo, S. T. (2003). Onvestigation of Mediational Processes Using Parallel Process Latent Growth Curve Modeling. Structural Equation Modeling, 10( 2), 238 -262. Ch r ousos G P. (2000). Stress, chronic inflammation, and emotional and physical well being: concurrent effects and chronic sequelae. The Journal of Allergy and Clinical Immunology, 106 (5Suppl), S275S2 91. Coffman, J. M. Cabana, M. D., Halpin, H. A., & Yelin, E. H. (2008). Effects of asthma education on Childrens use of acute care services: A metaanalysis. Pediatrics, 121, 575-586.
156 Craig T J Teets S. Lehman, E. B. Chinchilli, V. M & Zwillich C. (1998). Nasal congestion secondary to allergic rhinitis as a cause of sleep disturbance and daytime fatigue and the response to topical nasal corticosteroids. The Journal of Allergy and Clinical Immunology 101(5) 63363 7. Cramer J A, Rosenheck R. (1998). Compliance with medication regimens for mental and physical disor ders. Psychiatric Services, 49 196201 Craske M G Poulton, R Tsao, J C & Plotkin D. (2001). Paths to panic disorder/agoraphobia: an exploratory analysis from age 3 to 21 in an unselected birth cohort. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 556563. Daly, J. M., Biederman, J., Bostic, J. Q., Maraganore, A. M., Lelon, E., Jellinek, M., Lapey A. (1996). The relationship between childhood asthma and attention deficit hyperactivity disorder: A review of the literature. Journal of Attention Disorders, 1 (1), 31-40. Devereux, G. (2006). The increase in the prevalence of asthma and allergy: Food for thought. Natur e Reviews. Immunology 6 (11), 869 -874. DiMatteo, M. R., Lepper, H. S., & Croghan, T. W. (2000). Depression is a risk factor for noncompliance with medical treatment: metaanalysis of effects of anxiety and depression on patient adherence. Archives of Internal Medicine, 160, 2101 -2107 Dirks J F Kinsman R A. Horton D J Fross K. H & Jones N F. (1978). Panic fear in asthma rehospitalization following intensive longterm treatment Psychosomatic M edicine ,40, 5 13. Dolder,C. R., Furtek, K., Lacro, J. P., & Jeste, D. V. (2005) .Antihypertensive medication adherence and blood pressure control in patients wit h psychotic disorders c ompared to p e rsons without psychiatric i llness Psychosomatics, 46 135141. ) Druss B. G Marcus S. C Olfson, M Tanielian T Elinson, L & Pincus H A. (2001). Comparing the national economic burden of five chr onic conditions. Health Affairs, 20 233241. Eisner M D Katz P. P Lactao G & Iribarren, C (2005). Impact of depressive s ymptoms on adult asthma outcomes Annal s of Allergy, Asthma & I mmunology 94(5), 5665 74. Feldman J M Ortega, A. N McQuaid E. L & Canino G. (2006). Comorbidity between asthma attacks and internalizing disorders among Puerto Rican children at oneyear follow up. Psychosomatics, 47, 333339.
157 Gau, S. S., Chen, S. J., Chou, W. J, Cheng, H., Tang, C. S., Chang, H. L., Tzang, R. F., Wu, Y. Y., Huang, Y. F., Chou, M. C., Liang, H. Y., Hsu, Y. C., Lu, H. H., Huang, Y. S. (2008). National survey of adherence, efficacy, and side effects of methlphenidate in children with attention-deficit/hyperactivity disorder in Taiwan. Journal of Clinical Psychiatry, 69 (1), 131 -140. Glazebrook C McPherson A. C Macdonald I A. Swift, J. A., Ramsay, C., Newbould, R., et al. (2006). Asthma as a barrier to children's physical activity: implications for body mass index and mental health. Pediatrics 118, 24432449. Gold, D. R., & Wright,R. (2005). Population disparities in asthma. Annu al Review of Public H ealth 25 89 -113. Goodwin, R D Fergusson, D M & Horwood L J. (2004). Asthma and depressive and anxiety disorders among young pe rsons in the community. Psychol ogical Med icine, 34 (8), 146514 74. Goodwin, R D Messineo K. Bregante, A. Hoven C W & Kairam R. (2005). Prevalence of probable mental disorders among pediatric asthma patients in an inner -city clinic. The J ournal of Asthma 42 643647. Goodwin, R. D. Sourander, A., Duarte, C. S., Niemela, S., Multimaki, P., Nikolakaros, G., et al. (2009) Do mental health problems in childhood predict chronic physical conditions among males in early adulthood? Evidence from a commu nity -based prospective study. Psychological Medicine, 39, 301311 Guevara, J., Lozano, P., Wickizer, T., Mell, L., & Gephart, H. (2001). Utilization and cost of health care services for children with attention-deficit/hyperactivity disorder. Pediatrics 10 8 71 78 Honberg, L., McPherson M., Strickland, B., Gage, J. C., & Newacheck P. W. (2005). Assuring adequate health insurance: Results of the National Survey of Children with Special Health Care Needs. Pediatrics 115 (5 ), 1233 1239. Horton, N. J., & Lipsitz, S. R. (1999). Review of software to fit generalized estimating equation regression models. The American Statistician, 53 160 -169. Janicke D M F inney J W & Riley A. W. (2001). Childrens health c are u se: A prospective investigation of fac tors related to care-seeking. Medical Care, 39 9901001. Jamoom E. W Simon, S. L Janicke, D L Harman J Adams C D & Dumont Driscoll M C (2010). Comorbid behavioral health diagnosis and medical service use in children with asthma and M edic aid i nsurance Unpublished manuscript. A. F., Kavut, A. B., & Elili, M. (2009). Allergic and nonallergic rhinitis: the threat for obstructive sleep apnea. Annal s of Allergy, Asthma & I mmunology 103(1), 2025.
158 Ka mble S. & Bharmal M. (2009). Incremental direct expenditure of treating a sthma in the United States. The J ournal of Asthma, 46 73 -80. Katon, W. Cantrell, C. R., Sokol, M. C., Chiao, E., & Gdovin, D. M. (2005). Impact of antidepressant drug adherence on comorbid medication use and resource utilization. Archives of Internal Medicine, 165, 24972503. Katon W Lozano P. Russo J McCauley E. Richardson L & Bush T. (2007). The prevalence of DSM -IV anxiety and depressive disorders in youth with asthma compared with cont rols. The J ournal of Adolesc ent Healt h 41 (5), 455 -4 63. Kewalramani A Bollinger M E. & Postolache T T. (2008) Asthma and mood d isorders. International Journal of C hild Health and Human Development 1 (2), 115-123. Kondryn, H. J., Edmondson, C. L., Hill, J., & Eden, T. O. (2010). Treatment non adherence in teenage and young adult patients with cancer. Lancet Oncology Retreived from doi:10.1016/S1470 -2045(10)70069-3 Lehrer, P. M. (1998). Emotionally triggered asthma: a review of research literature and some hypotheses for self -regulation therapies. Applied Psychology and Biofeedback, 23(1), 1341. Levenson, J L. (Ed.). (2005) The American Psychiatric Publishing Textbook of Psychosomatic Medicine. Arlington, VA: Am erican Psychiatr ic Publ ishing Lily C M. (2005). Diversity of asthma: evolving concepts of pathophysiology and lessons from genetics. The Journal of Allergy and Clinica l Immunology ; 115(4 Suppl), S526-S5 31. Lin E. H Katon, W Von Korff M Rutter C Simon, G E. Oliver M et al. (2004) Relationship of depression and diabetes self -care, medication adherence, and preventive care. Diabetes Care, 27 (9), 215421 60. Lozano, P. Sullivan S D Smith D H & Weiss K. B. (1999). The economic burden of asthma in US children: estimates from the National Medical Expenditure Survey. The Journal of Allergy and Clinical Immunology 104, 957963. Mackinnon, D P. Warsi G & Dwyer J H. (1995). A simulation study of mediated effect measures. Multivariate Behavioral Research, 30 41 -62. Manning, W. G., Mullahy, J. ( 2001). Estimating log models: To transform or not to transform? Journal of Health Economics 20(4), 4614 94. McCauley E. Katon, W Russo K. Richardson, L & Lozano, P. (2007). Impact of anxiety and depression on functional impairment in adolescents with asthma. Gen eral Hospital Psychiatry 29 214222.
159 McEwen B. S. (2006) Sleep deprivation as a neurobiologic and physiologic stressor: Allostasis and al lostatic load. Metabolism, 55(10 Suppl 2):S20 -S2 3. McQuaid E. L Kopel S. J Nassau J H. (2001) Behavioral adjustment in children with asthma: a meta a nalysis. Journal of Developmental and B ehavioral Pediatrics 22 430439. McQuaid, E. L., Kopel, S. J., Klein, R. B., & Fritz, G. K. (2003). Medication adherence in pediatric asthma: reasoning, responsibility, and behavior. Journal of Pediatric Psychology, 28 323333. Mehta, P.D., & West, S.G. (2000). Putting the individual back in individual growth curves. Psychological Methods, 5, 2343. Milgrom, H. & Bender, B. (1993). Psychological side effects of therapy with corticosteroids. American Review of Respiratory Disease, 147, 471473. Mohammed, F ., Bootoor S. Panday A. Ramdass, A., Reemaul, J., Sharma, A., et al. (2006). Predictors of repeat visits to the emergency room by asthmatic children in primary care. The J ournal of the American Medical Association, 98 1278 -1284. Moorman, J E. Rudd, R A Johnson C A. King, M., Minor, P., Bailey, C., et al. (2007). Na tional Surveillance for Asthma -United States, 1980 2004. Morbidity and Mortality Weekly Rev iew -Surveillance Summaries 56 (SS08), 1 14; 18 -54. Morrison K M Goli A Van Wagon er J Brown E. S. & Khan, D A. (2002) Depressive symptoms in inner -city children with asthma. Primary Care Companion to the Journal of Clinical P sychiatry, 4 (5), 174-17 7. Mrazek D A. (1992). Psychiatric complications of pediatric asthma. Ann als of Aller gy, 69, 285290. Naimi D R Freedman, T G Ginsburg K. R Bogen, D Rand, C S. & Apter A. J (2009). Adolescents and asthma: why bother with our meds? The Journal of Allergy and Clinical Immunology 123( 6), 1335 -13 41. National Asthma and E ducation and Prevention Program. Expert Panel Report 3: Guidelines for the diagnosis and management of asthma. (2007). Bethesda, MD: N ational Institutes of H ealth/Nat ional Heart Lung Blood Inst i tute Retrieved from http://www.nhlbi.nih.gov/guidelines/asthma/gip_rpt.pdf National Committee on Quality Assurance. (2010). HEDIS 2010 Final NDC Lists Retreived from http://www.ncqa.org/tabid/1091/Default.aspx Nouwen A. Freeston M H Labbe, R & Boulet L P. (1999). Psychological factors associated with emergency room vi sits among asthmatic patients. Behav ior Modif ication 23 217 -233.
160 Opolski, M., & Wilson, I. (2005). Asthma and depression: a pragmatic review of the literature and recommendations for future research. Clinical Practice and Epi demiology in Mental Health, 1 18. Retrieved from http://www.cpementalhealth.com/content/1/1/18 Ortega A. N Huertas S E. Canino G Ramirez R & Rubio -Stipec M. (2002). Childhood asthma, chronic illness, and psychiatric di sorders. The Journal of Nervous and M ental D isease, 190, 275281. Owens, J. (2009). Neurocognitive and behavioral impact of sleep disordered breathing in children. Pediatric Pulmonology, 44 417-422. Pelham W E. Foster E. M & Robb, J A. (2007) The economic impact of attentiondeficit/hyperactivity disorder in children and adolescents. J ournal of Pediatr ic Psychol ogy, 32 (6), 711727. Pr eto rius, E. (2004). Corticosteroids, depression and the role of serotonin. Review in the Neurosciences 15 109116. Rachelefsky, G. S., Wo, J., Adelson, J., Mickey M. R., Spector S. L., Katz R. M., Siegel, S. C., & Rohr, A. S. (1986). Behavior abnormalities and poor school performance due to oral theophylline use. Pediatrics, 78, 1133 -1138. Richardson, L P Lozano, P Russo J McC auley, E., Bush, T., & Katon, W. (2006). Asth ma symptom burden: relationship to asthma severity and anxiety and depression symptoms. Pediatrics 118, 1042 -1051. Richardson, L Russo J Lozano, P. McCauley E. & Katon, W. (2008) The effect of comorbid anxiety and depressive disorders on health care utilization and costs among adolescents with asthma. Ge neral Hospital Psychiatry, 30 398 406. Rodrigo, G. J., & Rodrigo, C. (2002). The role of anticholinergics in acute asthma treatment. Chest, 121 19771987. Roy Byrne, P. P., Davidson, K. W., Kessle r, R. C., Asmundson, G. J. G., Goodwin, R. D., Kubzansky, L., et al. (2008). Anxiety disorders and comorbid medical illness. General Hospital Psychiatry, 30 208225. Roy Byrne, P. P., Katon, W. (1997). Generalized anxiety disorder in primary care: the precursor/modifier pathway to increased health care utilization. Journal of Clinical Psychiatry, 58 (Supplemental 3), 33-40. Sapra S. Nielsen, K & Martin, B. C. (2005). The net cost of asthma to North Carolina Medicaid and the Influence of comorbidities that drive asthma costs. The J ournal of Asthma 42 469-477. Shadish W R Cook T D & Campbell D T (2002) Experimental and Quasi Experimental Designs Boston, MA : Houghton Mifflin Company
161 Singer J D. (1998). Using SAS PROC MIXED to Fit Multil evel Models, Heirarchical Models, and Individual Growth Models. Journal of Educational and Behavioral Statistics, 24 ( 4) 323-355. Slattery M J (2005) Psychiatric comorbidity associated with atopic disorders in children and adolescents. Immunology and a llergy clinics of North America, 25 (2), 407-4 20. Smith D H Malone, D C Lawson K. A. Okamoto, L J & Battista C. (1997). A national estimate of the economic costs of asthma. American Journal of Respiratory and Critical Care M edicine 156, 787-793. Spettell, C. M., Wall, T. C., Allison, J., Calhoun, J., Kobylinski, R., Fargason, R., & Kiefe, C. I. (2003) Identifying Physician-Recognized Depression from Administrative Data: Consequences for Quality Measurement. Health Services Research, 38 (4), 1 0811102. StataCorp. (2007). Stata Statistical Software: Release 10. College Station, TX: StataCorp LP. Stein, R. E. K., Horwitz, S. M., Storfer -Isser, A., Heneghan, A., Olson, L., & Hoagwood, K. E. (2008). Do pediatricians think they are responsible for i dentification and management of child mental health problems? Results of the AAP Periodic Survey. Ambulatory Pediatrics, 8(1), 11 -17. Sun H L Kao, Y H Lu, T H Chou, M C & Lue, K. H. (2007). Health-care utilization and costs in Taiwanese pediatric patients with asthma. Pediatr ics Int ernationa l, 49 48 -52. United States Department of Health and Human Services. (2010). Healthy People 2010. Retrieved at www.healthypeople.gov Vila G Nollet -Clemencon, C de Blic J Mouren-Simeoni M C & Scheinmann, P. (1998). Asthma severity and psychopathology in a tertiary care department for children and adolescents. European Child & Adolescent P sychiatry, 7 1 37144. Vila G Nollet -Clemencon, C de Blic J Mouren-Simeoni M C & Scheinmann P. (2000). Prevalence of DSM IV anxiety and affective disorders in a pediatric population of asthmatic children and adolescents. J ournal of Affect ive Disord ers, 58, 223231. Wang L Y Zhong, Y & Wheeler L (2005) Direct and indirect costs of asthma in school age children. Prev enting Chronic Dis ease, 2, 1. Retrieved from http://www.cdc.gov/pcd/issu es/2005/jan/04_0053.htm Weiss K. B. Gergen, P. J & Hodgson, T A (1992) An economic evaluation of asthma in the United States. The New England Journal of Medicine 326, 862 86 6.
162 Weiss K. B. Sullivan S D & Lyttle C S. (2000). Trends in the cost of illness for asthma in the United States, 1985 1994. The Journal of Allergy and Clinical Immunology 106, 493 499. Wooldridge J M. (2006). Introductory Econometrics A Modern Approach. Mason, OH: Thomson South-Western Yelin, E. Trupin, L Cisternas M Eisner M Katz P. & Blanc P. (2002). A national study of medical care expenditures for respiratory conditions. The European Respiratory Journal, 19, 414 421. Zaubler T S. & Katon, W. (1998). Panic disorder in the general medical setting. Journal of Psychosomatic Research, 44(1), 25 -42. Zielinski T A. Brown E. S. Nejtek V. A Khan, D A Moore, J J & Rush A. J (2000) Depres sion in asthma: Prevalence and clinical i mplications. Primary Care Companion to the Journal of Clinical P sychiatry, 2 (5), 15315 8
163 BIOGRAPHICAL SKETCH Eric Wayne Jamoom was born i n the spring of 1 979, in New York City, New York to Victor a nd Randee Jamoom When Eric was four months old, his father died of a massive heart attack at 32. The apparent passing of nearly every paternal relative at early ages f rom heart disease instilled a curiosity around genetics and chronic disease. Therefore, Eric received his Bachelor of Science in Mic robiology/ Cell S cience from the University of Florida in May of 2001, and his Master of Science in Medi cal Genetics and training in genetic c ounseling from the University of Minnesota in August of 2003 Not just wanting to inform individual patients of their personal and family health, Eric decided he wanted to explain the health and the delivery of health services for vulnerable populations So in 2005, Eric came to UF to pursue his Master of Public Health and doctoral degree in Health Services Research. While at the University of Florida, Eric has had a num ber of different research collaborations and teaching opportunities including working with the Research Rehabilitation and Training Center on Health and Wellness Expert Health Status Measurement Panel, completing an internship at the Florida Center for Medicaid and the Uninsured, and developing, collecting and implementing outreach studies in Medicaid clinics across Gainesville. Eric has published multiple articles and journal entries during his time at UF After receiving his Ph. D. i n 2010, Eric moved to Washington, DC, as an Associate Health Services Research Fellow at the National Center for Health Statistics He plans on applying his experiences to tackle the current challenges associated with the delivery of health services and assessing the effects o f electronic health records