<%BANNER%>

Modifiable Contributors to Rural Disparities in Type 2 Diabetes and Cardiovascular Disease

Permanent Link: http://ufdc.ufl.edu/UFE0024502/00001

Material Information

Title: Modifiable Contributors to Rural Disparities in Type 2 Diabetes and Cardiovascular Disease
Physical Description: 1 online resource (51 p.)
Language: english
Creator: Ewigman, Nathan
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: cardiovascular, diabetes, disparity, lifestyle, modifiable, rural
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The prevalence of both type 2 diabetes and cardiovascular disease (CVD) are greater in rural than urban areas. Obesity, smoking, and physical inactivity are known modifiable contributors to both diseases and are also more prevalent in rural areas. The current study utilized a nationally representative database, the Medical Expenditures Panel Survey (MEPS), to test the hypothesis that modifiable lifestyle factors contribute significantly to the association between (1) rurality and diabetes and (2) rurality and CVD. After controlling for nonmodifiable contributors (e.g. demographics, access to health care), rurality and diabetes were not statistically related (p = .082). However, when modifiable contributors were controlled for, the odds ratio decreased (from OR = 1.23 to 1.14) at a significant level (p = .007). For CVD, the association with rurality lost significance only after modifiable factors were added to the model (p = .049 to p = .278). Adding modifiable contributors to the model significantly decreased (p = .01) the odds ratio of having CVD among rural vs. urban populations by 44%. These results support the hypothesis that the association between rurality and both diseases were partially predicted by modifiable contributors beyond nonmodifiable factors. The higher rates of obesity, smoking and physical inactivity seen in rural areas may be contributing to the higher rates of these diabetes and CVD in rural areas. Effective interventions targeting these factors in rural areas may help ameliorate the rural/urban disparities in type 2 diabetes and CVD.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Nathan Ewigman.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Perri, Michael G.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-11-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024502:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024502/00001

Material Information

Title: Modifiable Contributors to Rural Disparities in Type 2 Diabetes and Cardiovascular Disease
Physical Description: 1 online resource (51 p.)
Language: english
Creator: Ewigman, Nathan
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: cardiovascular, diabetes, disparity, lifestyle, modifiable, rural
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The prevalence of both type 2 diabetes and cardiovascular disease (CVD) are greater in rural than urban areas. Obesity, smoking, and physical inactivity are known modifiable contributors to both diseases and are also more prevalent in rural areas. The current study utilized a nationally representative database, the Medical Expenditures Panel Survey (MEPS), to test the hypothesis that modifiable lifestyle factors contribute significantly to the association between (1) rurality and diabetes and (2) rurality and CVD. After controlling for nonmodifiable contributors (e.g. demographics, access to health care), rurality and diabetes were not statistically related (p = .082). However, when modifiable contributors were controlled for, the odds ratio decreased (from OR = 1.23 to 1.14) at a significant level (p = .007). For CVD, the association with rurality lost significance only after modifiable factors were added to the model (p = .049 to p = .278). Adding modifiable contributors to the model significantly decreased (p = .01) the odds ratio of having CVD among rural vs. urban populations by 44%. These results support the hypothesis that the association between rurality and both diseases were partially predicted by modifiable contributors beyond nonmodifiable factors. The higher rates of obesity, smoking and physical inactivity seen in rural areas may be contributing to the higher rates of these diabetes and CVD in rural areas. Effective interventions targeting these factors in rural areas may help ameliorate the rural/urban disparities in type 2 diabetes and CVD.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Nathan Ewigman.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Perri, Michael G.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-11-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024502:00001


This item has the following downloads:


Full Text

PAGE 1

1 MODIFIABLE CONTRIBUTORS TO RURAL DI SPARITIES IN TYPE 2 DIABETES AND CARDIOVASCULAR DISEASE By NATE L. EWIGMAN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Nate L. Ewigman

PAGE 3

3 To my parents, Coleen Kivlahan and Bernar d Ewigman; to my best friends; and to my grandfather, LB Ewigman

PAGE 4

4 ACKNOWLEDGMENTS I would first like to sincerely thank my m entors, Michael G. Perri and Jeffrey S. Harman, for their generous support and guidance on this ma sters thesis. They have been generous with their time and encouragement. I would also like to thank my parents who instilled a passion for contribution in me. My best friends have kept me light and ridi culous which has been equally important. My life is very rich, and the support I have been shown thr ough this process has yet again proven this to me.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7ABSTRACT ...................................................................................................................... ...............8 CHAP TER 1 INTRODUCTION .................................................................................................................. 10Overview of the Problem ........................................................................................................10Background .................................................................................................................... .........10Type 2 Diabetes ......................................................................................................................11Cardiovascular Disease ...........................................................................................................12Rural/Urban Differences in Prevalence of Diabetes and Cardiovascular Disease ................. 13Type 2 Diabetes in Rural Populations .................................................................................... 13Nonmodifiable Contributors to Di abetes in Rural Populations ................................... 14Socioeconomic and demographic .............................................................................14Race and ethnicity .................................................................................................... 15Access to care ........................................................................................................... 15Obesogenic environment ..........................................................................................16Modifiable Contributors to Di abetes in Rural Populations ..........................................16Cardiovascular Disease in Rural Populations .........................................................................17Nonmodifiable Contributors to Cardiova scular Disease in Rural Populations ........... 18Socioeconomic and demographic .............................................................................18Race and ethnicity .................................................................................................... 18Access to care ........................................................................................................... 19Modifiable Contributors to Cardiovascular Disease in Rural Populations .................. 19Summary ....................................................................................................................... ..........21Current Study ..........................................................................................................................212 DATA AND METHODS ....................................................................................................... 22Data Source .............................................................................................................................22Variables ..................................................................................................................... ............23Dependent Variables ....................................................................................................... 23Independent Variables ..................................................................................................... 24Mediator Variables ..........................................................................................................24Obesity ..................................................................................................................... 24Physical Activity ......................................................................................................24Smoking Status ......................................................................................................... 24Control Variables .............................................................................................................25Socioeconomic status ...............................................................................................25Age ........................................................................................................................... 25

PAGE 6

6 Sex ............................................................................................................................25Marital status ............................................................................................................ 25Region of the country ...............................................................................................26Race and ethnicity .................................................................................................... 26Physical limitations .................................................................................................. 26Access to care ........................................................................................................... 27Statistical Analyses .......................................................................................................... .......273 RESULTS ....................................................................................................................... ........30Diabetes ...................................................................................................................... ............30Participant Characteristics ............................................................................................... 30Association with Modifiable Contributors ......................................................................30Association with Rurality ................................................................................................ 30Change in Odds Ratios ....................................................................................................31Cardiovascular Disease ...........................................................................................................31Participant Characteristics ............................................................................................... 31Association with Modifiable Contributors ......................................................................31Association with Rurality ................................................................................................ 32Change in Odds Ratios ....................................................................................................324 DISCUSSION .................................................................................................................... .....36Limitations ................................................................................................................... ...........38Implications .................................................................................................................. ..........39Future Research ......................................................................................................................44LIST OF REFERENCES ...............................................................................................................45BIOGRAPHICAL SKETCH .........................................................................................................51

PAGE 7

7 LIST OF TABLES Table page 3-1 Diabetes, CVD, obesity, smoker status, physical activity by urban versus rural .............. 333-2 Changes in odds ratios after controlling for nonmodifiab le and modifiable contributors to the association be tween rurality and diabetes ...........................................333-3 Changes in odds ratios after controlling for nonmodifiab le and modifiable contributors to the associati on between rural ity and CVD ................................................ 333-4 Odds ratios of all nonmodifiable and m odifiable variables predicting diabetes ................343-5 Odds ratios of all nonmodifiable and modifiable variables predicting CVD. ................... 35

PAGE 8

8 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science MODIFIABLE CONTRIBUTORS TO RURAL DI SPARITIES IN TYPE 2 DIABETES AND CARDIOVASCULAR DISEASE By Nate L. Ewigman May 2009 Chair: Michael G. Perri Major: Psychology The prevalence of both type 2 diabetes and ca rdiovascular disease (CVD) are greater in rural than urban areas. Obesity, smoking, a nd physical inactivity are known modifiable contributors to both diseases and are also more prevalent in rural areas. The current study utilized a nationally representative database, the Medical Expenditures Panel Survey (MEPS), to test the hypothesis that modifiable lifestyl e factors contribute si gnificantly to the a ssociation between (1) rurality and diabetes and (2) ru rality and CVD. After controlling for nonmodifiable contributors (e.g. demographics, access to health care), rurality and diabetes were not statistically related (p = .082). However, when modifiable contributors we re controlled for, the odds ratio decreased (from OR = 1.23 to 1.14) at a significant level (p = .007). For CVD, the associa tion with rurality lost significance only after modifiable factors were added to the model (p = .049 to p = .278). Adding modifiable contributors to the model si gnificantly decreased (p = .01) the odds ratio of having CVD among rural vs. urban populations by 44%. These re sults support the hypothesis that the association between rurality and both diseases were pa rtially predicted by modifiable contributors beyond nonmodifiable factors. The higher rates of obesity, smoking and physical inactivity seen in rural areas may be contributin g to the higher rates of these diabetes and CVD

PAGE 9

9 in rural areas. Effective interventions targeting these factors in rural areas may help ameliorate the rural/urban disparities in type 2 diabetes and CVD.

PAGE 10

10 CHAPTER 1 INTRODUCTION Overview of the Problem The prevalence of type 2 diabet es and cardiov ascular disease (C VD) is greater in rural than urban areas. Contributors such as obesity, sm oking and physical inactivity are known lifestyle contributors to both diseases and are also more prevalent in rural areas. The purpose of the current study is to examine the role of modifiable lifestyle contributors in predicting the prevalence of self-reported type 2 diabetes and CVD in rural and urban populations. The proposed study attempts to quantify the unique cont ribution of lifestyle fact ors to these diseases in rural and urban populations on a national level. Understanding this contribution is a first step in determining the role of lifestyle interventions in reducing rural disparities in diabetes and CVD. Background W e will review the (a) disease burden of di abetes and CVD, (b) national rural/urban differences in the prevalence of these diseases and (c ) their contributors in nonmodifiable and modifiable terms. Sociodemographic risk factors for diabetes and CVD, such as age, region of the country and educational status can be categ orized as nonmodifiable because of the poor understanding of their association with chronic diseases and the re lative difficulty of altering the negative health consequences of these factors. Similarly, acc ess to care is a relatively nonmodifiable factor. In comparison, lifestyle f actors are modifiable in that there is a canon of efficacy and, to a lesser extent, effectiven ess literature on the be nefit of intervening upon lifestyle factors to im prove health outcomes. The comparatively higher prevalence of diabetes and CVD in rural populations appears to be caused by a combination of nonmodifiable a nd modifiable factors (Gamm et al., 2003). As

PAGE 11

11 will be discussed, rural areas are characterized by both sociodemographic and lifestyle contributors to diabetes. Some (Lee et al., 2007) have argued that higher rates of obesity and chronic diseases are due to disease-promoting en vironments while others contend that lifestyle factors play a unique role (Patterson et al., 2004). Gamm et al. (2003) has conjectured that as rural areas adopt healthier lifestyl es, the higher rates of these diseas es will be primarily explained by sociodemographic factors. Currently, however, the unique contribution of these factors has not been examined empirically on a national le vel. Currently, it has not been shown that modifiable, lifestyle factors contribute to th e higher rates of diab etes and CVD in rural populations. This is an important step towards justifying interventions ta rgeting the contributing lifestyle factors in rural areas. Type 2 Diabetes Type 2 diabetes differs from type 1 diabetes in that it is usuall y adult onset, generally treated with oral medications and is largely driven by lifestyle factors (ADA, 2008). Type 2 diabetes is a debilitating chronic disease charact erized by the inability to break down and utilize glucose (ADA, 2008). High glucose levels can even tually result in a host of secondary medical complications such as CVD, renal disease and retinopathy. Approximately 16 million Americans have type 2 diabetes (Mainous et al., 2004) and diabetes accounts for over 300,000 deaths in the United States annually (ADA, 1998). The health care cost of diabetes was $100 billion in 1997 (Mokdad et al., 2001). Moreover, strong evidence exists that both the prevalence and incidence have been increasing rapidly (Mokdad et al ., 1999; Mokdad et al., 2000; Geiss et al., 2006). National estimates for the prevalence of dia gnosed diabetes in recent years are between five to eight percent of the population (Mokdad et al., 2001; Sta gnitti & Pancholi, 2004; Narayan et al., 2006). However, when the known underreporting bias and lack of dete ction are taken into account, the prevalence is closer to ten percen t (Mokdad et al., 2003; Engelgau et al., 2004;

PAGE 12

12 Harris et al., 1998). Narayan et al (2006) predict that the prevalence of diabetes will double from 2005 to 2050. Changes in the incidence of diabetes ar e even more alarming. From 1997 to 2003, the incidence, or new cases, of di abetes increased by 41% (4.9 to 6.9 per 1,000; Geiss et al., 2006). Furthermore, evidence suggests th at the increasing inci dence of diabetes is being driven by lifestyle factors such as obesity and lack of exercise. Geiss et al. (2006) found that the new cases in 2002-2003 were significantly more likely to be obese than new cases in 1997-1998. This finding underscores the connection be tween diabetes and lifestyle c ontributors such as obesity and lack of exercise (Sullivan et al., 2005). Cardiovascular Disease CVD refers to a clu ster of diseases includi ng coronary heart disease, congestive heart failure and related symptoms of angina, hypert ension, stroke and myocar dial infarction (AHA, 2008). An important determinant of CVD is lifestyle factors such as ciga rette smoking, lack of exercise and obesity (Khot et al., 2003; Alexander et al., 2003). A majority of patients with coronary CVD have these risk factors (Khot et al., 2003). Although CVD can be prevented, treated and in some cases even reversed by lif estyle changes (Ornish et al., 1990), CVD accounts for 900,000 deaths annually and is still the lead ing cause of death (Cooper et al., 2000; NCHS, 2008). In 2005, the national financial burden of CVD was estimated at nearly $400 billion (CDC, 2005). Approximately one of every three Americans has one or more types of CVD (AHA, 2008). Despite the high prevalence of most types of CVD, its prevalence has been decreasing since the 1960s (with the exception of conge stive cardiovascular failure) and continues to decrease (Cooper et al., 2000). The incidence of most types of CVD has remained stable overall, with increases in certain subgroups a nd rural populations. Findings in the incidence of CVD must be

PAGE 13

13 considered in the context of in creasing rates of adverse health conditions such as obesity (Cooper et al., 2000). Yusuf et al. (2004) found that over 90% of the population attributable risk for CVD can be accounted for by unhealthy lifestyle alone in 95% of people. Similar to diabetes, CVD ameliorates in response to lifes tyle intervention. In the Lifest yle Heart Trial, Ornish and colleagues (1990, 1998) found that a five year intensive lifestyle intervention including smoking cessation, improvement in diet and physical activity led to an overall reversal of CVD, including a 91% reduction in anginal events and a 5 year sustained weight loss of 12.8 lbs compared to minimal change in the control group (Ornish et al., 1990; 1998). Lifestyle factors play a crucial role in the onset of CVD. Rural/Urban Differences in Prevalence of Diabetes and Cardiovascular Disease Defining rurality. Rural areas are defined by thei r low population density and are typically ch aracterized by high rates of poverty and lower acce ss to services and commodities (Census Bureau, 2007). An urban area refers to a central city and surrounding area with a combined population of 50,000 or more and at least 1,000 inhabitants per square mile (Census Bureau, 2007). One of the most commonly used clas sification system and the one used in this paper for urban is metropolitan statistical ar ea (MSA) versus non-metropo litan statistical area (non-MSA) for rural as define d by the federal Office of Management and Budget standards to Census 1990 data. These standards generally define MSAs as an urban core with at least 50,000 people and a total population (including the surrounding area) of 100,000. Approximately 20% of the US population is rural or non-MSA (Larson et al., 2003). Type 2 Diabetes in Rural Populations The prevalence of type 2 diabetes is dispropor tionately higher in ru ral areas, as com pared to urban ones. According to the 1995 National Hea lth Interview Survey (NHIS), the self-reported

PAGE 14

14 prevalence of diabetes in ur ban populations is 3.2% compared to 3.6% for rural populations (Cooper et al., 2000; Pearson et al., 1998; Barnett et al. 2000; Gamm et al., 2003). Another NHIS estimate suggests that the self-reported prevalen ce of diabetes in rural areas was 17 percent higher than in central cities and 11.7 percent higher than in all othe r classifications of urban areas (Gamm et al., 2003). Although recent estimates of diabetes incidence in rural vs. urban populations appear to be unavailable, the overall incidence of diabetes is increasing rapi dly among certain subgroups. An analysis of people with previously diagnosed diabetes, the National Health and Nutrition Examination Survey III indicated that the prevalence of diabetes is highest among rural African Americans (9.5%), as compared with urban Afri can Americans (6.0%), rural whites (6.5%) and urban whites (4.5%; Mainous et al ., 2004). From these prevalence ra tes, an approximation of the national prevalence of previously diagnosed diabet es in rural areas is 8% compared to 5.25% in the national urban population (Main ous et al., 2004). Another indica tion that the rural disparity in diabetes is still extant is th at new cases of diabetes are charact erized by risk factors associated with rurality. Specifically, th e incident diabetic cases fr om 1997-2003 in the NHIS were characterized by older age and obesi ty status (Geiss et al., 2006). Nonmodifiable Contributors to D iabetes in Rural Populations Higher rates of diabetes in ru ral areas are partially caused by nonmodifiable factors. Below we review the contribution of these factors to the prevalence of diabet es in rural populations. Socioeconomic and demographic Im portant sociodemographic risk factors help explain higher rates of diabetes in rural areas. Lower income and educational status, fo r example, has an inverse trend with the prevalence and incidence of diabetes (G eiss et al., 2006; Mokdad et al., 2001; Mokdad et al., 2003). Although the mechanism by which lower socioeconomic status is not specifically known,

PAGE 15

15 chronic stress, potentially manife sting as the disturbance of th e Hypothalamic-Pituitary-Adrenal axis, has been proposed as an independent mediat or of the SES diabetes association (Rosmond, 2003). Aging has a similar trend with diabetes, ho wever, it only partially accounts for the rising prevalence (Dabney & Gosschalk, 2003). Socioec onomic variables such as low income and educational status as well as hi gher age are positively associated with rural status (Gamm et al., 2003). Race and ethnicity Race is ano ther risk factor for diabetes. Over all, African Americans have the highest rates of diabetes (Mokdad et al., 2001; Mainous et al., 2004) Compared to white men, for example, black men are 100% likelier to have or devel op diabetes (Bracanti et al., 2000). Although rural areas have lower proportions of minorities, minority and rural status may have an additive effect on risk for diabetes (Mainous et al., 2004). Data from the third NHANES show rural blacks as having the highest prevalence of diabetes comp ared to urban blacks, rural whites and urban whites (9.5% vs. 6.0%, 6.5%, and 4.5% re spectively; Mainous et al., 2004). Access to care Having a usual source of care is seen as an entry point for getting p reventive services, which is particularly important for the manageme nt of chronic diseases (Larson et al., 2003). Rurality is also associated with a lower likelihood of having health insurance, getting prompt and even routine care (Bolin & Gamm et al., 2003). Rura l residents have fewer outpatient visits per year (Larson et al., 2003; Gamm et al., 2003) as well as fewer physic ians and hospitals per capita (280 per 100,000 vs. 156 per 100,000; Merwin et al., 2006) Thus rural residents are more likely to live farther away from a usual source of care as compared to urban residents (Larson et al., 2003; Gamm et al., 2003). Almost thirteen percent of rural households have no source of regular care and report fewer ambulatory visits than urban (Pearson et al., 1998 ; Larson et al., 2003).

PAGE 16

16 The inability to get routine and preventive primary care is problema tic because it can lead to late diagnosis and improper management (Bolin & Ga mm et al., 2003). This c ould contribute to the higher prevalence of diabetes through fewer pr eventive services, late diagnosis and improper management. Obesogenic environment Another im portant structural contributor to diab etes in rural populations is what Lee et al. (2007) refers to as an obesogenic envir onmentcharacterized by poor access to physical activity and healthy foods at the same time r eady access to unhealthy foods. Given the strong association between diabetes and obesity, obe sogenic environmenta l contributors could partially explain the rural/urban di sparity in diabetes (Lee et al., 2007). In California, the ratio of fast food restaurants and conveni ence stores versus supermarkets and produce venders (Retail Food Environment Index, or RFEI) significantly pr edicts the prevalence of diabetes even after controlling for demographic characteristics (Des igned for Disease, 2008). Rural areas are less likely to have supermarkets than urban areas, so a similar pattern with diabetes could be inferred in rural areas (Kaufman, 1999). A lthough these structural factors are potential determinants of diabetes, it is crucial to focus on the modifiable contributions to diabet es in rural settings. Modifiable Contributors to Diabetes in Rural Populations The etiology of type 2 diabetes is also partially explained by modifiable lifestyle behaviors. O besity and obesity-related lifestyle c ontributors such as not adhering to physical activity recommendations and a healt hy diet are crucial predictors of diabetes (Sullivan et al., 2005; Hu et al., 2001). Obesity appears to be the str ongest lifestyle predictor of diabetes. First, obesity and diabetes prevalence are highly corre lated. In 2001, national data sugge sted that the prevalence of diabetes in normal weight individuals is 4.1%. The prevalence increased according to increases

PAGE 17

17 in weight classification (i.e. Overweight = 7.3%, Obese Class 2 = 14.9%, Obese Class 3, 25.6%; Mokdad et al., 2003). Second, there is strong evid ence that obesity lead s to higher rates of incident diabetes. Geiss et al (2006) showed that from 1997 to 2003, the incidence of diabetes increased by 41%. In a follow-up study, Narayan et al. (2006) found that obesity was the main contributor to the increase in incidence. Third, th ere is a direct correlation between weight gain and risk of diabetes. Mokdad (2001) estimated from 1990-1998 using the BRFSS that every 1-kg increase in self-reported weight was associated with 9% increase in the risk of having diabetes. Given these trends between obesi ty and diabetes, it is particul arly concerning that in the 1998 National Health Interview Survey, the prevalen ce of obesity in rural populations was 20.4% compared to 17.8% in urban (Patterson et al., 2004). Lack of exercise is anothe r lifestyle predictor of diab etes (Hu et al ., 2001). Although physical inactivity is a known cause of obesity, it also predicts diabetes independently of BMI (Sullivan et al., 2005). Improvements in physical activ ity and diet in clinical trials have been shown to reduce incident diabetes and related ri sk factors (Mokdad et al., 2001; Sullivan et al., 2005). Similar to obesity, lack of exercise, as we ll as poor diet, are more prevalent among rural populations (Patterson et al ., 2004; Lee et al., 2007). Cardiovascular Disease in Rural Populations Until th e late 1970s, CVD was less prevalent in rural than urban populations (Pearson et al., 1998). Although the rates of CVD began to declin e in urban areas during the latter part of the 20th century, rural areas did not experience the sa me downward trend. Several factors potentially explain this shift: 1) the increasing mechanizat ion of traditionally phys ically strenuous rural occupations, 2) the late adoption of healthy lifestyles more prom inent in urban areas, and 3) better access to modern medical technologies to treat and prevent CVDs in urban areas (Pearson et al., 1998; Patterson et al., 2004). By the 1980s, rurality was a clear risk factor for CVD and by

PAGE 18

18 1996, CVD was 1.34 times more prevalent in rural areas (98.8 per 1,000) compared to urban (72.6 per 1,000; Pearson et al., 1998; Gamm et al., 2003). Nonmodifiable Contributors to Cardiovascular Disease in Rural Populations The association between CVD and sociodem ogra phic factors is well established (Cooper et al., 2000). The major nonmodifiable risk factors for CVD associated with rural status are age, educational status and poverty (Cooper et al., 20 00). CVD is by far the largest cause of mortality among people over 65 (NCHS, 2008). Moreover, th e highest proportions of people over the age of 65 reside in rural areas (12% in central counties vs. 15% in most rural c ounties; Eberhardt et al., 2001). Socioeconomic and demographic CVD rates are higher in populations with lo wer incom es and education (Cooper et al., 2000; Diez-Roux et al., 1997). Being poor and living in poorer neighbo rhoods is associated with CVD beyond individual-level variables such as race (Cooper et al., 2000; Diez-Roux et al., 1997). Although rates of poverty are high in inner cities, rural areas are also characterized by poverty (Eberhardt et al., 2001; Lee et al., 2007; Pearson et al., 1998). Rural populations also have a higher proportion of people with less than a high school education (Patterson et al., 2004). Additionally, CVD prevalence is strongly correlated with educ ational status within rural populations (Pearson et al., 1998). Si milar to diabetes, chronic stre ss is a proposed mediator of the association between socioeconomic status and CVD (Kaplan & Keil, 1993). Race and ethnicity On a nationa l level, race is also a notable predictor of CVD. African American men have the highest burden of CVD (National Center fo r Health Statistics; NC HS, 2008; Gamm et al., 2003). Although there are fewer minorities in rural areas, the preval ence of hypertension among

PAGE 19

19 rural African Americans (23%) is higher than (a) urban African Americans (20%), (b) rural whites (13%) and (c) urban white s (10%; Mainous et al., 2004). Access to care Lim ited access to care also contributes to the higher rates of CVD in rural populations. Rural areas have fewer physicians and health care centers per capita as noted earlier (Merwin et al., 2006). As discussed, people in ru ral areas are more likely to have to drive greater distances to access care than urban counterparts (Gamm et al., 2003). Additionally, rural persons are less likely than urban counterparts to have had their bl ood pressure checked in th e previous five years or to have taken action to lower it (Gamm et al ., 2003). All of these gaps in care experienced by rural populations potentially lead to (a) late diagnosis, (b) inad equate management of chronic CVD, (c) higher mortality rates due to CVD as a re sult of longer travel times for care (Bolin & Gamm et al., 2003). Modifiable Contributors to Cardiovascular Disease in Rural Populations The strongest and most consistent risk f actors for CVD are smoking, type 2 diabetes, hypertension and hyperlipidemia. Only 15-20% of CVD patients lack any of these risk factors (Khot et al., 2003). Smoking, diabetes, hypertension and hyperlipidemia are all more prevalent in rural populations (Eberhardt et al., 2001; Gamm et al., 2003; Mainous et al., 2004; Cooper et al., 2000). Obesity is another strong risk factor for CVD as well as hypertension, hyperlipidemia and diabetes. For example, among women in the Nurs es Health Study who developed diabetes, prediagnosis weight gain increased their future risk of coronary CVD (Cho et al., 2002). Obesity is also a direct predictor of coronary CVD and other types of CVD (Must et al., 1999). Closely related to obesity, physic al inactivity is a pr edictor of CVD. P hysical activity has been shown to reduce the risk of cardiac events, high blood choles terol and blood pressure levels

PAGE 20

20 (Ornish et al., 1998; Khot et al., 2003; Hu et al., 2001; Lee et al., 2001) Sedentary women who became physically active in middle age showed a lowe red risk of coronary events (Manson et al., 1999). As with obesity, physical inactivity is more prevalent in rural ve rsus urban populations (Patterson et al., 2004; Lee et al., 2007) Although obesity and related health behaviors are intimately linked to three of the four conventional risk factors, perhaps the strongest determinant of CVD is smoking (Khot et al., 2003). Smoking alone was responsible for 180,000 deaths related to CVD in 1990 and also appears to decrease the time of onset for corona ry CVD (Cooper et al., 2000; Jousilahti et al., 1999; Office of the Surgeon General, 2004; Khot et al., 2003). Additionally, being a smoker increases risk of CVD by 1.5 to 3 fold (Jous ilahti et al., 2000; Kannel et al., 1999). A recent meta-analysis revealed that smoking cessation lead s to a 36% risk reducti on in cardiac mortality regardless of age, sex and type of cardiac event (Critchley & Capewell, 2003). It has been suggested that higher rates of sm oking is a result of lower educa tional status and lower access to health education resources in rura l areas (Eberhardt et al., 2001). The conventional risk factors, as well as other contributors to CVD such as obesity and physical inactivity are disproportionately highe r in rural populations (Cooper et al., 2000; Eberhardt et al., 2001; Pearson et al., 1998; Gamm et al., 2003). Moreover, rural persons are less likely than urban to have their bl ood cholesterol levels checked in the last 5 years and take action to reduce their high blood pr essure, CVD (Pearson et al., 1998; DHHS, Healthy People 2010, 2000). Although rural areas have high er prevalence of lifestyle risk factors and worse preventive care, CVD can be prevented and treated through lifestyle interventions (Ornish et al., 1998). Thus, understanding whether lifesty le contributors predict the hi gher rates of CVD in rural

PAGE 21

21 populations is essential in determ ining if lifestyle in terventions can amelio rate the rural/urban disparity. Summary The causes of chronic diseases su ch as diabetes and CVD are multifaceted. Nonmodifiable and modifiable contributors to these diseases are common in rural areas. To our knowledge, there has not been a national asse ssment of the relative contributions of these factors to the higher rates of diabetes and CVD in rura l populations. Understanding the contributions of modifiable factors to diabet es and CVD may help in identifying appropriate targets for intervention in rural areas. Current Study The current study attempts to m easure the unique contribution of m odifiable factors in explaining the higher rates of diabetes and CVD in rural populations. We propose the following primary hypotheses: 1. Controlling for modifiable f actors will weaken the associ ation between rurality and diabetes even after considering nonmodifiable factors. 2. Controlling for modifiable f actors will weaken the associat ion between rurality and CVD even after considering nonmodifiable factors.

PAGE 22

22 CHAPTER 2 DATA AND METHODS Data Source The Medical Expenditure Panel Survey (MEPS) is jointly sponsored by the Agency for Health Care Policy and Research an d the National Center for Health Statistics. The MEPS consists of a set of large-scale, nationally repres entative surveys which document utilization, cost and insurance information among the civilia n non-institutionalized U.S. population. The household component (MEPS-HC) provides deta iled information on demographics, health conditions/status, medical care utilization, acce ss to care and income. The MEPS-HC utilizes a sampling frame, or a set of units from which the sample was drawn, from respondents to the National Health Interview Survey (NHIS) as we ll as an overlapping panel design of sample households which entails an initia l contact and five interviews fo r data collection over a 2 year period (AHRQ, 2003; Sullivan et al., 2005). Data collection continues in the subsequent year with a new sample of househol ds, creating overlapping panels of survey data (Cohen et al., 1999). Combining these data with other panels allows for conti nuous and current estimates of health care expenditures (Cohen et al., 1999). Each household inte rview consists of computerassisted personal interviewing tec hnology as well as utilization and cost information on medical care for 2 calendar years (Cohen et al., 1999). Sampling from the National Health Interview Survey ensures a nationally representative sample of the US civilian non-institutionalized population with oversampling of Hispanics and African Americans (Cohen et al., 1999; Sullivan et al., 2005). As a result of the disproportionate sampling of minorities and its complex sampling procedure, MEPS data are weighted (Cohen et al., 1999). The weights are derived from the prev ious years NHIS weights and are based on demographic probabilities to correct for complete or partial non-respons e, differences between

PAGE 23

23 NHIS and MEPS eligibility, and corrections to better match the Current Population Survey ranging from the regional to person level (Cohen et al., 1999). The MEPS-HC utilizes stratification, clustering and multiple stages of selection to further adjust for complex sampling (Cohen et al., 1999). Participants from the MEPS-HC provide names of their medical providers and employers during the survey. The Medica l Provider Component of MEPS (MEPS-MPC) validates medical care and conditi on information at the person level based on this information. Diagnoses of medical conditions in the MEPSMPC are based on ICD-9 clinical modification codes (Sullivan et al., 2005). The current study used data from the 2005 MEPS, which samples from the 2004-2005 National Health Interview Survey, to establish th e unique contribution of li festyle contributors to the higher rates of diabetes and CVD in rural populations. Variables Dependent Variables The prim ary outcomes of the present study were the presence of type 2 diabetes and CVD (including myocardial infarcti on, congestive cardiovascular failure, angina, cardiovascular disease, hypertension, stroke and other cardiovascular conditions) in urban and rural populations as measured by ICD-9 codes in the MEPS-MPC. Although CVD can refer to a broad cluster of disorders of the heart and arteries, the opera tionalization of CVD in the current study is consistent with the American Heart Associati on definition (AHA, 2008). As part of the MEPSHC, respondents were asked if they had ever been diagnosed as having type 2 diabetes or various forms of CVD. From this self-reported info rmation, medical providers and facilities are contacted for corroboration of thes e self-reported diseases and in formation is collected in the MEPS-MPC. ICD-9 codes are generated in the MPC data files by disease and were assimilated

PAGE 24

24 into variables that flagged the respective conditions. Presence of diabetes and CVD are based off of these flag vari ables (AHRQ, 2003). Independent Variables Metropo litan Statistical Area (MSA) status, the proxy for rurality, wa s the predictor of interest. MSA status is assigne d according to the OMB standard s of the Census 1990 data based on the respondents address. Counties placed alo ng the urban-rural continuum are categorized as MSA/urban (includes metro and ne ar-metro, see Appendix A: 1-6) and non-MSA/rural (includes near-rural and rural, see Appendix A: 7-9; AHRQ, 2003). Mediator Variables The prim ary aim of this study is to determin e whether the lifestyl e factors of obesity, smoking and exercise predict rura l/urban differences in diabetes and CVD while controlling for nonmodifiable factors (i.e. demographics, access to care, etc.). Obesity For adults over age 18, body m ass index (BMI) [w eight in kg/height in m] was calculated using self-reported height and weight (which are not included for public use because of confidentiality concerns). Obesity constitu tes a BMI greater than or equal to 30. Physical Activity Physical activity was measured by asking res pondents if they engaged in moderate or vigorous physical activity for 30 minutes three or more times per week. Smoking Status Sm oking status was measured by asking respond ents if they currently smoke cigarettes (Sullivan et al., 2005).

PAGE 25

25 Control Variables To exam ine the lifestyle contributors to the higher rates of diabetes and CVD, sociodemographic (income, race, gender, age, ed ucation, geographic region) and access to care (usual source of care, distance to usual source of care) variables were used to control for de facto differences in morbidity of diabetes a nd CVD not due to lifestyle factors. Socioeconomic status Lower socioeconom ic status is generally measured by income/poverty level and educational status and is a we ll-established risk factor for diabetes and CVD (Braveman & Tarimo, 2002). Given that rural areas are ch aracterized by lower socioeconomic status, socioeconomic status is a de facto contributor to thes e diseases in rural areas (Lee et al., 2007). In MEPS, poverty status is measured as family income as a percent of the federal poverty line (poor, near poor, low income, middle income and high income). Education is measured as the years of education when first entering MEPS ( no school/kindergarten only, grades 1-8, grades 911, grade 12, 1 yr of college, et c. up to 5+ years of college). Age Age is stron gly associated with disease mo rbidity and rural status (Gamm et al., 2003; Iezzoni, 2003). Age is measured by difference in years from self-reported date of birth and 12/31/05. Sex Sex is related to differences in disease m orb idity and is thus important to control for statistically. Sex is m easured by self report. Marital status Marital status is associated with a poorer prognosis of heart disease in wom en (OrthGomer et al., 2000) whereas, among men, being married is generally protective of health (Lillard

PAGE 26

26 & Panis, 1996). Thus, marital status by gender interactions may represent de facto differences in heart disease and potentially diabetes. Marital st atus is measured by self report as of 12/31/05 (married, widowed, divorced, separated, never married). Region of the country Regional dif ferences in morbidity of both C VD and diabetes exist, thus region of the country is also a predictor to disease morbidity (Cooper et al ., 2000). Region is determined by census region criteria (North east, Midwest, South, West). Race and ethnicity Racial and ethnic difference in diabetes a nd CVD are stark (Cooper et al., 2000; Iezzoni, 2003). For exam ple, African American men have the highest rates of CVD and rural African Americans have higher rates than their urban counterparts (Mainous et al., 2004). African Americans have two times the risk of diabetes -related deaths than wh ite counterparts (Clark, 1998). Race is self reported and consists of white (no other race reported) black (no other race reported), American Indian/Alaska native (no other race reported), Asian (no other race reported), native Hawaiian/Pacific islander (no ot her race reported) and multiple races reported. Compared to non-Hispanic white people, Hisp anics have a worse CVD risk profile but paradoxically lower CVD mortality (Swenson et al., 2002). In terms of diabetes, Hispanics (particularly Mexican Americans and Puerto Ricans) have roughly double the prevalence of diabetes than non-Hispanic whites. Hispanic ethnicity was measured as Hispanic vs. nonHispanic (Flegal et al., 1991). Physical limitations Another individual predictor of disease m orbidity that is also more pr evalent in rural areas is physical limitation (Mainous & Kohrs, 1995). Physical limitati on is measured by self report and operationalized as any limitation in walking.

PAGE 27

27 Access to care Access to care is another im portant assessment of extant differences in morbidity of CVD and diabetes. An accurate assessment of access to care in rural populations is the time it takes to get to a usual source of care (Larson et al., 2003 ). In MEPS, this is measured by self-reported minutes it takes to drive to th e reported usual source of care (less than 15 minutes, 15 to 30 minutes, 31 to 60 minutes, 61 to 90 minutes, 91 to 120 minutes, more than 120 minutes). Statistical Analyses To identify the self-reported pr ev alence of diabetes and CVD, we estimated the number of rural and urban persons flagged by the diab etes and CVD dummy variables. We then dichotomized variables based on disease pres ence and conducted a proportion analysis and a simple Chi Square analysis to determ ine if the difference was significant. Given the dichotomous nature of diabetes and CVD and our research question of estimating the relative contribution of our independent variables, logistic regression was deemed an appropriate statistical analysis Logistic regression transforms the dependent variable, in this case diabetes and CVD, into log it variables which expr ess the natural log odds of the dependent variable occurring or not occurring. The odds ratio (OR) then, represents th e ratio of the odds of the dependent variable occurring in one group of a particular independent, e.g. urban status, versus the odds of it occurring in another group, e.g. rural status. An odds ratio of one indicates that the dependent variable is equally likely under both conditions of the independent variable. We tested the goodness of fit of our model for both diabetes and CVD using the HosmerLemeshow goodness-of-fit test. Based on the Hosme r-Lemeshow test our models fit the data overall although tended to slightly overpredict for those with the greatest probability of having CVD.

PAGE 28

28 We used separate logistic regressions to determine the odds ratio between (1) MSA (rural) status and diabetes and (2) MSA status a nd CVD controlling for nonmodifiable variables (sociodemographic and access to care). To assure nationally representative estimates and to adjust for the complex sample design of MEPS, person-level, sample and variance adjustment weights were used (Sullivan et al., 2005). To determine the unadjusted association of ru ral status and diabetes and CVD, we first determined the OR for the regressions (1) MSA (rural) status and diabetes and (2) MSA status and CVD without any covariates In order to understand the contribution of nonmodifiable (sociodemographic and access to ca re) factors to the rural dis ease association, we recomputed the model controlling for nonmodifiable factors and re-estimated the OR. Fina lly, as a test of our primary hypotheses, we recomputed the m odel controlling for both nonmodifiable and modifiable (obesity, smoking and physical activ ity for CVD; obesity and physical activity for diabetes) factors and re-estimated the OR. By comparing the model with nonmodifiable factors alone to the model with both nonmod ifiable and modifiable factors, we were able to determine whether controlling for modifiable factors woul d significantly weaken the association between rurality and each disease, thereby implying that modifiable factors help explain the rural disparity in these diseases. In other words, we set up a mediational test of modifiable factors on the association between rurality and each disease. Due to collinearity, several variables were either combined (i.e. Midwest + West = Western; never married + separated + widowed = not married) or not used as covariates in the analysis (African American, married, Northeast) Additionally, several va riables were modified to ameliorate the fit of our model. For diabetes, age was divided into 5 categories by years (1825, 26-35, 36-45, 46-54 and 56-65). To adjust for th e over-prediction of CVD, we used the

PAGE 29

29 square of age and obesity, and we divided educ ation into less than hi gh school, high school and more than high school. To test our mediational hypothesis, we determ ined the significance of the difference of the odds ratio between the two models (the original model with only nonmodif iable factors vs. the model with both nonmodifiable and modifiable f actors) we used an Adjusted Wald Test. Although the Wald Test is generally used to test the significance of individual regression coefficients, it can be adjusted to test the significance of the difference between any two dichotomous variables. We creat ed two arbitrary dichotomized variables that were coded opposite of each other (i.e., D = 0 when G = 1, vice versa ) and created intera ction variables for both nonmodifiable and modifiable variables. To test for the significance of adding modifiable variables to the model, we coded all of the nonm odifiable interaction vari ables as D & G and the modifiable variables as G only. We then used an Adjusted Wald Test to test thef significance of the difference betw een G and D (G minus D). All of the above analyses used the survey procedures of Stata 10 statistical software (StataCorp, 2002).

PAGE 30

30 CHAPTER 3 RESULTS Diabetes Participant Characteristics The sam ple with diabetes used to test th e association with rura lity consisted of 2,007 respondents. Among people with diabetes, 85% we re either overweight or obese, 37.5% were physically active and 21% were ru ral (compared to the overall sample proportion of 18%). The crude proportion of people with diabetes am ong urban was 5.5% compared to 7.2% among rural populations. For the proportion of diabetes, CVD, obesity, physical inactivity and smoking by urban vs. rural, please see Table 3-1. Association with Modifiable Contributors Obesity was associated with self-reported diabetes ( p < .001); the association of physical activity and diabetes was marginally significant ( p = .06; see Table 3-4). Association with Rurality A logistic regression analysis determined that the unadjusted associated b etween rurality and diabetes was significant (OR = 1.37, p = .003). When controlling for nonmodifiable contributors, the associati on lost significance (OR = 1.23, p = .082). Although rurality and diabetes were not statistically re lated when nonmodifiable contribu tors were accounted for, there was an additive effect of modifiable contri butors on the associati on between rurality and diabetes. When modifiable contributors were added to the model (already controlling for nonmodifiable contributors), th e association further weakened (from OR = 1.23, p = .082 to OR = 1.14, p = .265). Thus, both nonmodifiable and modifiab le factors appear to contribute to the rural/urban differences in the prevalence of diabetes.

PAGE 31

31 Change in Odds Ratios A logistic regression analysis showed that the unadjusted odd s ratio of diabetes was 38% higher am ong the rural population th an for the urban one (OR = 1.37, p = .003). When controlling for nonmodifiable contributors, the odds ratio dropped to 18% higher for rural compared to the urban (OR = 1.23, p = .082). As previously stated, adding modifiable contributors weakened the odds ra tio of having diabetes and bei ng rural to approximately 14%, representing a 38.5% decrease in the odds ratio (OR = 1.14, p = .265). The Adjusted Wald Test showed that this decrease in odds ratio was si gnificant (p = .007). See Table 3-2 for changes in the odds ratio in the association between rura lity and diabetes when controlling for (a) nonmodifiable and both (b) nonmodifiab le and modifiable contributors. Cardiovascular Disease Participant Characteristics The sam ple with CVD used to test the re lationship with rurali ty consisted of 5,577 respondents. In the sample, 77.7% of those with CVD were overwei ght/obese, 47.8% were physically active, 15.9% were smokers and 19.8% were rural. The overall prevalence of CVD among rural urban was 20.8% compared to 16.6% among urban populations. For the proportion of diabetes, CVD, obesity, physical inactivity an d smoking in urban vs. rural, please see Table 31. Association with Modifiable Contributors Obesity and physical activity were independe ntly associated with self-reported CVD (both ps < .001); however, the association between being a current smoker and CVD was not significant ( p > .05; see Table 3-5).

PAGE 32

32 Association with Rurality In a logistic regression analysis, the unadjus ted relationship between rurality and CVD was significant (OR = 1.27; p = .001). The association was weakened but sti ll significant after controlling f or nonmodifiable contributors (OR = 1.16; p = .049). However, addition of the modifiable contributors to the model rendere d the odds ratio non-significant (OR = 1.09; p = .278). Change in Odds Ratios Logistic regression showed the unadjuste d odds ratio of having CVD a mong the rural population was approximately 29% higher than for the urban one (OR = 1.29, p < .001). When controlling for nonmodifiable contributors, the od ds ratio dropped to approximately 16% higher for rural compared to th e urban population (OR = 1.16, p = .049). Adding modifiable contributors further weakened th e odds ratio of having CVD and being rural to approximately 9%, representing a 44% decrease in the odds ratio (OR = 1.09, p = .278). The Adjusted Wald Test revealed that this decrease in the odds ratio was significant (p = .01). See Table 3-3 for changes in the odds ratio in the association betw een rurality and CVD when controlling for (a) nonmodifiable and both (b) nonmodifiab le and modifiable contributors.

PAGE 33

33 Table 3-1. Diabetes, CVD, obesity, smoker st atus, physical activity by urban versus rural Table 3-2. Changes in odds ratios after c ontrolling for nonmodifi able and modifiable contributors to the association between rurality and diabetes Rurality Diabetes OR P % Change in OR p of Change from Previous Model Unadjusted Association 1.37 0.003-With NonModifiable 1.23 0.08238%0.015 With NonModifiable and Modifiable 1.14 0.27639%0.007 Table 3-3. Changes in odds ratios after c ontrolling for nonmodifi able and modifiable contributors to the associati on between rura lity and CVD RuralityCVD OR p % Change in Odds p of Change from Previous Model Unadjusted Association 1.27 0.001-With NonModifiable 1.16 0.04944%0.012 With NonModifiable and Modifiable 1.09 0.27844%0.010 Rural Urban Percentage Confidence Intervals Percentage Confidence Intervals Self Reported Diabetes 7.15%.062 .0815.49% .051 .059 Self Reported CVD 20.76%.191 .22416.65% .159 .174 Overweight/Obese 74.41%.731 .75871.20% .704 .721 Current Smoker Status 25.64%.235 .27819.39% .184 .204 Mod/Vig Physical Exercise (3x/wk) 57.40%.561 .58757.85% .540 .617

PAGE 34

34 Table 3-4. Odds ratios of all nonmodifiable a nd modifiable variable s predicting diabetes Diabetes OR p Rural 1.14 0.27 Western Region 1.37 0.04 South 1.34 0.08 Young Adult 0.08 0.00 Mid Young Adult 0.12 0.00 Mid Adult 0.30 0.00 Mid Old Adult 0.59 0.00 Sex 1.14 0.19 White 0.61 0.00 Hispanic 1.57 0.00 Divorced 0.89 0.36 Other Marital 0.88 0.34 Years of Education 0.93 0.00 Poverty Category 0.88 0.00 Walking Limitation 1.77 0.00 Time Takes to get to Usual Source of Care 1.07 0.20 BMI 1.09 0.00 Physical Activity (x3/wk, mod to vig) 0.85 0.06 Current Smoker 1.04 0.73

PAGE 35

35 Table 3-5. Odds ratios of all nonmodifiable and modifiable variab les predicting CVD. CVD OR p Rural 1.09 0.28 Western Region 0.97 0.71 South 1.33 0.00 Age 1.15 0.00 Age Squared 1.00 0.07 Sex 1.29 0.00 White 0.62 0.00 Hispanic 1.06 0.50 Divorced 1.10 0.28 Other Marital 1.05 0.53 Less than High School Education 1.23 0.02 High School Education 1.26 0.00 Poverty Category 1.01 0.64 Walking Limitation 1.39 0.00 Time Takes to get to Usual Source of Care 1.12 0.00 BMI 1.10 0.00 BMI Squared 1.15 0.19 Physical Activity (x3/wk, mod to vig) 0.82 0.00 Current Smoker 0.92 0.33

PAGE 36

36 CHAPTER 4 DISCUSSION The current study exam ined the unique cont ributions of various nonmodifiable and modifiable factors to the rural/ urban disparities in type 2 diabetes and cardiovascular disease (CVD). Although it is known that bot h types of contributors to th ese diseases are more common in rural areas, to our knowledge, no studies ha ve evaluated the relativ e contribution of these factors using a national sample. Assessing the un ique contribution of modifiable factors is important because of the implicati on that special efforts targeting modifiable factors in the rural population might reduce the rural/urban disp arity in diabetes and heart disease. The present analysis examined the unique c ontribution of three sp ecific modifiable lifestyle factors (obesity, curre nt smoking, physical activity) to the association between (1) rurality and diabetes and (2) rura lity and heart disease. Consistent with our original hypotheses, modifiable lifestyle factors cont ributed significantly to the vari ance of the asso ciation of both diseases with rurality. However, the specific patter n of contributions appears to vary by disease. For diabetes, nonmodifiable factors (sociodem ographic and access to care) fully mediated the association with rurality. However, when mo difiable factors were added to the ruralitydiabetes model, the p value decreased (p = .134 to = .276). Correspondingly, when modifiable factors were added to the model already accoun ting for nonmodifiable factors, there was a statistically significant reduction in the odds ratio. Thus, although the high observed rates of diabetes in rural areas is largely due to nonmodifiable factors, modi fiable factors are also at play. Given the connection between many nonmodifiable factors associated with rurality (e.g. poverty, age, low access to care etc.) and diabetes, it follows that these are significant contributors to the rural/urban disparity in diabetes However, factors such as obesity and physical inactivity are also known contributors to diabetes and are associated with rural ity. In the present study, these

PAGE 37

37 modifiable factors also appear to play a role in the rural dispar ity in diabetes. Traditionally, a mediator is defined as an expl anatory factor between a significantly associated predictor and outcome variable (Baron & Kenny, 1986). In this case, rurality and diabetes were no longer associated after controlling for nonmodifiable contributors. Although modifiable contributors would not be considered a mediator by th e traditional definition (Baron & Kenny, 1986), modifiable factors significantly reduced the association between diabetes and rurality even after controlling for nonmodifiable contri butors. Thus modifiable contri butors help explain the higher rates of diabetes in rural vs. urba n populations on a national level. The results for heart disease are consistent wi th a substantial contri bution of modifiable lifestyle factors. The associati on between rurality and heart dis ease remained significant after accounting for all of the nonmodifi able factors. This is a surp rising finding given the strong association of nonmodifiable factors (e.g. race, age) to heart disease that are also associated with rural status. Additional factors associated with rurality predic ted heart disease beyond nonmodifiable factors. Indeed, adding modifiable factors to th e model rendered the association of rurality with heart disease non-significant. Moreover, the reduction in the odds ratio was also significant. Thus, it appears that modifiable fact ors potentially explain th e rural/urban disparity in heart disease above and be yond nonmodifiable contributors. An example of a longitudinal cohort study that supports the results of the current study is the Nurses Health Study. Among 84,129 women, t hose who with healthy lifestyles (not smokers, not overweight, consumed a healthy diet, exercised moderately or vigorously for half an hour a day and consumed moderate amounts of alcohol) had an incidence of coronary events 80% lower than in the rest of the population (Stampfer et al., 2000). These results independently predicted lower coronary event incidence beyond nonmodifiable fa ctors such as age and other

PAGE 38

38 medical risk (Stampfer et al., 2000). Findings from this prosp ective cohort study suggest that coronary events can be prevente d through targeting modifiable life style factors such as obesity, smoking and physical inactivity (Sta mpfer et al., 2000). Taken togeth er with the results of the current study, providing interven tions that specifically target healthy lifestyles in the rural population has the potential to reduce the incidenc e of coronary events in rural residents and thereby decrease the observed rural/urban differences in CVD. Limitations Several lim itations of the presen t study should be considered. First, all of the variables of interest were based on self report, and thus the potential for a soci al desirability bias exists. For example, obese people tend to underreport their weight (Stevens et al., 1998), potentially underrepresenting the role of obesity in predicting self-reported diabetes and CVD. Additionally, given the self report of conditions clinical verification of diabet es and CVD were not available, thus introducing the possibility of recall bias. However, systematic recall bias has not been implicated as a function of rural vs. urban areas (Larson et al., 2003). Sec ond, the availability of modifiable variables in the MEPS was limited. For example, no variables measure (1) smoking history, (2) diet quality or (3 ) sedentary behavior. Although th ese variables would provide a wider range of modifiable contri butors, obesity and current smoker status are both major causes of diabetes and CVD. Another po tential limitation of the study is the use of non-discrete disease outcomes such as hypertension, angina, and stroke (versus the aggregate cl assification of CVD). This limits the generalizability to the plethora of studies that focus on discrete cardiovascular diseases. In the overall sample of adults for 2005, 7,276 people reported ev er having a diagnosis of hypertension, angina and stroke (combined). This is most likely an overestimate of the current sample because we used data from people who reported a CVD diagnosis this year. Additionally, given the high co-occurrence of thes e conditions and the card iovascular diseases,

PAGE 39

39 our results may not be different even if excluding hypertension, angina and stroke. Lastly, our operationalization of CVD is consistent with the definition used by the American Heart Association (AHA, 2008). Another limitation is the operationalization of nonmodifiable and modifiable variables. Although lifestyle are traditionally considered modifiable, nonlifestyle predictors of morbidity, such as better access to health care is also potentially modifiable (Gamm et al., 2003). Moreover, it is likely that a combination of nonmodifiable and modifiable variables, are contributing to hi gher rates of diabetes and CVD in rural areas. Finally, the current analysis was cross-sectional and consequen tly can not address the causal relationship between modifiable f actors and the higher rates of disease in rural areas. However, given the sufficient evidence that modifiable factor s lead to both diabetes and CVD, our results imply that these factors contribu te to the rural/urban disparity in these diseases on a national level. Implications The higher rates of unhealthy lifes tyle behaviors in rural area s appear to be contributing significantly to the higher rates of diabetes and C VD in rural areas. These results sugg est that the rural/urban disparity in the prev alence of diabetes and CVD w ould be reduced if modifiable lifestyle factors were equivalent between th e rural and urban populat ion. Indeed, if these modifiable lifestyle factors were equivalent between rural and ur ban areas, it would theoretically result in approximately 200,000 fewer cases of di abetes and 550,000 fewer cases of CVD in rural areas (when extrapolated to the US rural populati on of 50 million; Eberhardt et al., 2001). Given the high costs of these diseases, targeting modifiable lifestyle behaviors in rural areas could decrease the economic impact of diabetes and CVD in rural areas and in th e country as a whole. Finally, given the strong associ ation of obesity and both diab etes and CVD (see Table 3-4 and Table 3-5), focusing on obesity may be an appr opriate target for interventions and policies.

PAGE 40

40 Thus, as we will discuss later, weight loss programs such as Treatment of Obesity in Underserved Rural Settings (TOURS; Perri et al ., 2008) represents an ex ample of the type of intervention that could be used to target the hi gher rates of diabetes and CVD in rural areas. The demonstration of modifiable factors inde pendently contributing to national disparities implies that, in order to reduce the rural/urban disparity in these diseas es, policy and research should focus on contributors to unhealthy li festyle behaviors in rural areas. To understand how to reduce the incidence of diabetes and CVD in rural areas, it is necessary to understand contributor s to unhealthy behaviors in rural areas; identify potential interventions to improve adherence to a healthful lifestyle in rural areas; execute clinical translational research to determine effective inte rventions for rural areas; and develop a national model to create partnerships with rural counties to assist in the implementation of these interventions. The first step in targeting the rural/urban di sparity in diabetes and CVD is to understand contributors to the higher rates of unhealthy behaviors in rural areas One potential contributor to higher rates of unhealthy behavior s is lack of education. As noted, rural residents have lower education (Gamm et al., 2003) and fewer outpati ent health visits (L arson et al., 2003). Additionally, rural heal th providers are burdened by high patient volume and low access to continuing medical education (Pearson et al., 19 98). Consequently rural residents have less exposure to health education and promotion. Add itionally, there are cultur al factors associated with being rural that influen ce unhealthy behaviors. For example, it is customary in rural America for meals to be highly caloric and nutritionally unbalanced (Flora et al., 2004). Traditionally, high caloric intake was sustainable because of the physically demanding nature of rural labor (Pearson et al., 1998). However, the increasing mechanization of farming has

PAGE 41

41 disrupted the caloric equation among many rural resi dents, leading to high er rates of obesity (Pearson et al., 1998). In addition to cultural co ntributors, being among people who engage in unhealthy behaviors increases an individuals risk of adopting these behaviors. For example, over a 32 year period in the Framingham Hear t Study, individuals who had a friend become obese had a 57% greater chance of becoming obese than those who did not (Christakis & Fowler, 2007). There are also environmenta l contributors to these unhealthy behaviors in rural areas. For example, rural areas have limited access to s upermarkets (Kaufman, 1999) or environments conducive to physical activity (Eyler, 2003). Moreove r, rural populations have been described as slow adopters of healthy behaviors (Pearson et al., 1998). Through the Framingham Heart Study and other research, innovations about the ca re and prevention of chronic diseases are disseminated first to urban areas for the reasons discussed above (Pearso n et al., 1998). Another potential contributor to unhealthy behaviors in rural areas is unt reated depression. For example, one known predictor of unhealthy behaviors is untreated depression (S trine et al., 2008). Although rural populations have similar rates of depression as urban, the treatment rates are lower in rural areas (Hauenstein et al., 2006). Thus, the higher prev alence of untreated depression may be an underlying predictor of unhealthy behavior s in rural settings. Hartley (2004) posed the challenge of identifying cont ributors to unhealthy behaviors in rural populations as a question: "Why does rural residence (culture, community, and environment) reinforce negative health behaviors?" By understanding contri butors to the higher rates of unhealthy behaviors in rural populations, interven tions aimed at improving lifestyle behaviors can be designed to be more effective in rural communities. The next step in improving the chronic dis ease burden in rural areas is identifying efficacious interventions to improve lifestyle beha viors. Chronic diseases are largely driven by

PAGE 42

42 lifestyle behaviors and are amenable to interven tion. Ornish et al.s (1990, 1998) results suggest that 90% of patients with heart disease can benefit significantly from lifestyle interventions (without additional medical treatment). An example of this type of an e fficacious intervention is the Diabetes Prevention Program (DPP). The DPP demonstrated that intensive lifestyle changes (particularly weight loss and phys ical activity) decreas ed overall incidence of diabetes by 58% (compared to 31% for Metformin, or a contro l group) over an approximately 3 year period (Diabetes Prevention Program Research Group, 2002) This intervention included a large number of elderly, lower educated and low income people making it more applicab le to rural settings. Another example of an efficacious intervention is the Coronary Health Improvement Program (CHIP; Aldana et al., 2005). This program was deve loped from the Ornish et al. (1990) Lifestyle Heart Trial and consists of groups of people a nd printed material on lifestyle improvement. A randomized clinical trial of this intervention showed improvement in diet and physical activity in the experimental group compared to a control group (Aldana et al., 2005). Compared to many interventions, the CHIP is rela tively cost effective (Aldana et al., 2005) and may be appropriate given the limited resour ces in rural counties. Rural areas are characterized by low income, low educati on (Gamm et al., 2003) and the host of cultural and environmental obstacles to healthy behavior previously discussed. Most efficacy clinical trials are done in urban area s with higher income/educated, highly motivated participants. Thus the generalizability of interventi ons such as the Lifestyle Heart Trial is low for rural persons. The prevalence of poor lifestyle behavi ors in the rural popul ation highlights the challenge of effecting sustained changes in life style necessary to reduce incidence of diabetes and CVD in rural areas. Additi onally, rural communities have limited infrastructure, training and

PAGE 43

43 funds to support many interventions. Taken together, there is a need for clinical translation of efficacious to effectiveness interventions that can work in rural areas. TOURS was a weight loss study designed sp ecifically for rural settings and provided extended care follow-up to promote the maintenan ce of healthy behaviors (Perri et al., 2008). Extending care appears to be of particular import because particip ants regained one-third to onehalf of lost weight within a year (Perri et al., 2008). TOURS demonstrated an average weight loss of 10.0 kg in the initial 6 month intervention. After this initial peri od, participants were randomized to 26 biweekly extended-care sessions of face-to-face, telephone or weight-control information (control group of either). Participan ts in the face-to-face and telephone conditions regained significantly less weight than thos e in the control group. Additionally, the telephone condition was more cost effective compared to the face-to-face extended care group ($2554 vs. $2125; Perri et al., 2008). Our findings support the dissemination of interventions such as TOURS in rural settings on a national level. Give n our results of the important contribution of obesity to both diseases in rural areas, TOURS stands out as a model fo r lifestyle interventions that might lead to the greatest impact on reduc ing diabetes and CVD in the rural population. In particular, implementing cost-effective programs fo r sustaining weight loss, such as the TOURS telephone intervention might produ ce significant return on invest ment with respect to the incidence of diabetes and CVD. However, given the financial cons traints of rural counties, more research on the financial viability of di sseminating such interventions is needed. Lastly, developing a national model to create pa rtnerships with rural counties to assist in the implementation of these interventions is necessary. Hartley (2004) has suggested that Wagner et al.s (2001) Chronic Care Model ( CCM) is appropriate for managing the population health of rural areas. A key element of this mode l is that successful interventions involve (a)

PAGE 44

44 activated patients, (b) prepared pr actitioners and (c) community res ources with respect to local or regional idiosyncrasies (Hartley, 2004). Disseminating effective (the rapeutically and financially) interventions in rural areas will certainly requ ire these elements. Effectively disseminating the necessary interventions to rural populations to al leviate the burden of chronic disease will also necessitate partnerships and support from a nati onal organization such as the Office of Rural Health Policy, the US Department of Agricu ltures Office of Rural Development and other funding sources such as State Rural Devel opment Councils, the National Rural Health Association, state offices of rural hea lth and state/local health departments. Finally, given the strong associ ation of obesity and both diab etes and CVD (see Table 3-4 and Table 3-5), focusing on obesity rather than current smoking or physical activity may be an appropriate target for in terventions and policies. Future Research Future direc tions for this lin e of research include further examining the unique contribution that obesity in particular appears to be playing in the rural/urba n disparity. Better specifying this contribution will enable national policies to pr ioritize funding for eff ective interventions. Additionally, a replicatio n of this study that includes sm oker history and diet quality would further explicate the contribution of modifiable factors to thes e diseases in rural populations. Finally, an important line of res earch is to identify factors that predict the higher rate of poor lifestyle factors in rural areas. As discusse d, there are many causes of the higher rates of unhealthy behaviors among rural populations. Regard less, there are no studies that attempt to quantify the relative contributions of such factors to the highe r rates of unhealthy behaviors. Identifying these contributors would provide a framework by which to design effective interventions in rural areas.

PAGE 45

45 LIST OF REFERENCES Agency for Healthcare Research an d Quality (AHRQ; 2003) Computing standard errors for MEPS estimates. Available from www. meps.ahrq.gov. Accessed September 12, 2008 Aldana SG, Greenlaw RL, Diehl HA, Salberg A, Merrill RM, Ohmine S, Thomas C (2005) Effects of an intensive diet a nd physical activity modi fication program on the health risks of adults. J Amer Dietetic Assoc105: 371-381 Alexander CM, Landsman PB, Teutsch SM, Haffner SM (2003) NCEP-defined metabolic syndrome, diabetes, and prevalence of coronary heart disease among NHANES III participants age 50 years a nd older. Diabetes 52: 1210-1214 American Diabetes Association (ADA; 1998) Ec onomic consequences of diabetes mellitus in the US in 1997. Diabetes Care 21: 296-309 American Diabetes Associati on (ADA; 2008), Available from http://www.diabetes.org/type2-diabetes.jsp Accessed Septem ber 12, 2008 American Heart Association (AHA; 2008) H eart disease and stroke statistics 2008 Update. Dallas, Texas: American Heart Association. Available from http://www. americanheart.org/ downloadable/ heart/1200082005246HS_Stats%202008.final.pdf. Accessed September 12, 2008 Barnett E, Anderson T, Blosnich, J, Halvers on J, Novak J (2005) Promoting cardiovascular health: from individual goals to social envi ronmental change. Amer J Prev Med 29: 107-112 Baron R, Kenny D (1986) The moderator-mediator variable distin ction in social psychological research: Conceptual strategic, and stat istical considerations J Pers Soc Psych 51: 1173-1182 Bolin J, Gamm L (2003) Access to quality hea lth services in rural areasinsurance: a literature review. Rural Healthy People 2010: A companion document to Healthy People 2010. Volume 2. College Station, TX: The Texas A&M University System Health Science Center, School of Rural Public Health, Southwest Rural Health Research Center Bracanti FL, Linda Kao WH, Folsom AR, Wa tson RL, Szklo M (2000) Incident type 2 diabetes mellitus in African American and white adults. JAMA 283: 2253-2259 Braveman PA, Tarimo E (2002) Social inequalities in health within countries: not only an issue for affluent nations. Soc Sci Med 54: 1621 Centers for Disease Control (CDC; 2005) Prev enting chronic disease: investing wisely Census Bureau, Economic Research (2007). Available from http://www.census.gov/population/ www/estimates/metrodef.html Accessed Septem ber 12, 2008

PAGE 46

46 Cho E, Manson JE, Stampfer MJ, Solomon CG, Colditz GA, Speizer FE, Willett WC, Hu FB (2002) A prospective study of obesity and risk of coronary heart disease among diabetic women. Diabetes Care 25: 1142-1148 Christakis NA, Fowler JH (2007) The spread of obesity in a large social network over 32 years. N Engl J Med 357: 370-379 Clark C (1998) How should we respond to the wo rldwide diabetes epidemic? Diabetes Care 21: 475-476 Cohen SB, DiGaetano R, Goksel H (1999) Estimation procedures in the 1996 Medical Expenditure Panel Survey Household Component Rockville (MD):Agency for Health Care Policy and Research. MEPS Methodol ogy Report No.5. AHCPR Pub.No.99-0027 Cooper R, Cutler J, Desvigne-Nickens P, Fortmann PS, et al. (2000) Trends and disparities in coronary heart disease, stroke, and other cardiovascular diseases in the United States: findings of the national conference on cardi ovascular disease prevention. Circulation 102: 3137-3147 Critchley JA, Capewell S (2003) Mortality risk reduction associated with smoking cessation in patients with coronary heart diseas e: A Systematic Review. JAMA 290: 86-97 Dabney B, Gosschalk A (2003) Di abetes in rural areas: a lite rature review. Rural healthy people 2010: a companion document to Healt hy People 2010. Volume 2. College Station, TX: The Texas A&M University System Health Science Center, School of Rural Public Health, Southwest Rural H ealth Research Center Designed for Disease: The link between local food environments and obesity and diabetes. (2008). California Center for Public Health Advo cacy, PolicyLink, and the UCLA Center for Health Policy Research Diabetes Prevention Program Research Group (2 002) Reduction in the incidence of type 2 diabetes with lifestyle intervention or Metformin. N Engl J Med 346: 393-403 Diez-Roux AV, Nieto FJ, Muntaner C, Tyrole r HA, Comstock GW, Shahar E, Cooper LS, Watson RL, Szklo M (1997) Neighborhood enviro nments and coronary heart disease: a multilevel analysis. Am J Epidemiol 146: 48-63 Eberhardt MS, Ingram DD, Makuc DM, Pam uk ER, Freid VM, Harper SB, Schoenborn CA, Xia H (2001) Urban and rural heal th chartbook. health, United St ates. Hyattsville, Maryland: National Center for Health Statistics Engelgau MM, Geiss LS, Tierney EF, Rios-Burro ws N, Mokdad AH, Ford ES, Imperatore G, Narayan KM (2004) The evolving diabetes burde n in the United States. Ann Intern Med 140: 945-950 Eyler, A (2003) Personal, social, and environm ental correlates of physical activity in rural Midwestern white women. Amer J Prev Med 25: 86-92

PAGE 47

47 Flegal KM, Ezzati TM, Harris MI, Haynes SG, Juarez RZ, Knowler WC, Perez-Stable EJ, Stern MP (1991) Prevalence of diabetes in Mexican Americans, Cubans and Puerto Ricans from the Hispanic Health and Nutritio n Examination Survey, 1982. Diabetes Care 14: 628-638 Flora, CB, Flora JL, Spears JD, Swanson LE (1992) Rural communities: legacy and change, 2nd edn, Westview, Boulder Gamm LD, Hutchison LL, Dabney BJ, Dorsey, AM, eds. (2003). Rural healthy people 2010: a companion document to Healthy People 2010. Volume 1. College Station, Texas: The Texas A&M University System Health Scienc e Center, School of Rural Public Health, Southwest Rural Health Research Center Geiss LS, Pan L, Cadwell B, Gregg EW, Be njamin SM, Engelgau MM (2006) Changes in incidence of diabetes in U.S. adu lts, 1997-2003. Amer J Prev Med 30: 371-377 Hauenstein EJ, Petterson S, Rovnyak V, Merw in E, Heise B, Wagner D (2006) Rurality and mental health treatment. Administration and Policy in Mental Health and Mental Health Services Research 34: 255-267 Harris MI, Flegal KM, Cowie CC, Eberhardt MS Goldstein DE, Little RR, Wiedmeyer HM, Byrd-Holt DD (1998) Prevalence of diabetes, impaired fasting glucose, and impaired glucose tolerance in US adults. Diabetes Care 21: 518-524 Hartley D (2004) Rural health disparities, popu lation health, and rural culture. Am J Public Health 94: 1675-1678 Hu FB, Stampfer MJ, Solomon C, Liu S, Co lditz GA, Speizer FE, Willett WC, Manson JE (2001) Physical activity and risk for cardiovascular events in diabetic women. Ann Intern Med 134: 96-105 Iezzoni L, ed. (2003) Risk adjustment for meas uring health care outcomes, 3rd edn Health Administration Press, Ann Arbor, Michigan Jousilahti P, Vartiainen E, Korhonen HJ, Pusk a P, Tuomilehto J (1999) Is the effect of smoking on the risk for coronary heart disease even stronger than was previously thought? J Cardiovasc Risk 6: 293-298 Kannel WB, McGee DL, Catelli WP (2000) La test perspective on cigarette smoking and cardiovascular disease: the Framingham experience. J Cardiac Rehab 4: 267-277 Kaplan GA, Keil JE (1993) Socioeconomic fact ors and cardiovascular disease: a review of the literature. Circulation 88: 1973 Kaufman PK (1999) Rural poor have less access to supermarkets, large grocery stores. Rural Dev Perspect 13: 19-25

PAGE 48

48 Khot UN, Khot MB, Bajzer CT, Sapp SK, et al. (2003) Prevalence of conventional risk factors in patients with coronary heart disease. JAMA 290: 898-904 Larson SL, Machlin SR, Nixon A, Zodet M (2004) Chartbook #13: Health care in urban and rural areas, combined years 1998-2000. Agency for Healthcare Research and Quality, Rockville, MD Lee I-M, Rexrode KM, Cook NR, Manson JE, Buring JE, (2001) Physical activity and coronary heart disease in women: is no pain, no gain" passe? JAMA 285: 1447-1454 Lee RE, Greiner KA, Hall S, Born W, Kimminau KS, Allison A, Ahluwalia JS (2007) Ecologic correlates of obes ity in rural obese adults. J Am Coll Nutr 26: 424-433 Lillard LA, Panis CW (1996) Mar ital status and mortality: th e role of health. Demography 33: 313-27 Mainous AG, King DE, Garr DR, Pearson WS (2004) Race, rural residence, and control of diabetes and hypertension. Ann Fam Med 2: 563-568 Mainous A, Kohrs F (1995) A comparison of hea lth status between rural and urban adults. Journal of Community Health 20: 423-431 Manson JE, Hu FB, Rich-Edwards JW, Colditz GA, Stampfer MJ, Willett WC, Speizer FE, Hennekens CH (1999) A prospective study of walk ing as compared with vigorous exercise in the prevention of coronary heart disease in women. N Engl J Med 341: 650-658 Merwin E, Snyder A, Katz E (2006) Differe ntial access to quality rural healthcare: professional and policy challenges. Fa mily & Community Health 29: 186 Mokdad AH, Bowman BA, Ford ES, Vinicor F, Marks JS, Koplan JP (2001) The continuing epidemics of obesity and diabetes in the United States. JAMA 286: 1195-1200 Mokdad AH, Ford ES, Bowman BA, Dietz WH Vinicor F, Bales VS, Marks JS (2003) Prevalence of obesity, diabetes and obesity-related health ri sk factors, 2001. JAMA 289: 7679 Mokdad AH, Ford ES, Bowman BA, Nelson DE, Engelgau MM, Vinicor F, Marks JS (2000) Diabetes trends in the U. S.: 1990. Diabetes Care 23:1278 Mokdad AH, Serdula MK, Dietz WH, Bowman BA Marks JS, Kopla JP (1999) The spread of the obesity epidemic in the United States, 1991-1998. JAMA 282: 1519-1522 Must A, Spadano J, Coakley EH, Field AE, Co lditz G, Dietz WH (1999) The disease burden associated with overweight and obesity. JAMA 282: 1523-1529 Narayan KM, Boyle JP, Geiss LS, Saaddine JB, Thompson TJ (2006) Impact of recent increase in incidence on future diabetes burden: U.S., 2005-2050. Diabetes Care 29: 21142116

PAGE 49

49 National Center for Health Statistics (NCH S) Deaths-leading causes. Available from http://www.cdc.gov/nchs/fastats/lcod.htm Accessed Septem ber 12, 2008 Office of the Surgeon General (2004) The health consequences of smoking: a report of the Surgeon General. Available from http://www.surgeongeneral.gov/library/ sm okingconsequences/. Accessed September 12, 2008 Ornish D, Scherwitz W Billings JH, Gould KL, Merritt TA, Sparler S, Armstrong WT, Ports TA, Kirkeeide RL, Hogeboom C, Brand RJ(1998) Intensive lifestyle cha nges for reversal of coronary heart disease. JAMA 280: 2001-2007 Ornish D, Brown SE, Scherwitz LW, Billings JH, Armstrong WT, Ports TA, McLanahan SM, Kirkeeide RL, Brand RJ, Gould KL. (1990) Can lifestyle changes reverse coronary heart disease?: The Lifestyle Heart Trial. The Lancet 336: 129-133 Orth-Gom er K, Wamala SP, Horsten M, Sche nck-Gustafsson K, Schneiderman N, Mittleman MA (2000) Marital stress worsens prognosis in women with coronary heart disease: the Stockholm female coronary risk study. JAMA 284: 3008-3014 Patterson PD, Moore CG, Probst JC, Shinogle JA (2004) Obesity and physical inactivity in rural America. J Rur Health 20: 151-159 Pearson TA, Lewis, C (1998) Rural epidemio logy: insights from a rural population laboratory. Am J Epidemiol 148: 949-957 Perri MG, Limacher, MC Durning PE, Jani cke DM, Lutes LD, Bobroff LD, Dale MS, Daniels MJ, Radcliff TA, Martin AD (2008) Extended-care programs for weight management in rural communities: the Treatment of Obesity in Underserved Rural Settings (TOURS) Randomized Trial. Arch Intern Med 168: 2347-2354 Rosmond R (2003) Stress induced disturbances of the HPA axis: a pathway to type 2 diabetes? Med Sci Mont 9:RA35-RA39 StataCorp (2002) Stata statistical software: release 9.0 Sp ecial Edition. Stata Corporation, College Station, TX Stagnitti MN, Pancholi M (2004) Chronic condi tions and outpatient prescription medicines for persons 18 and older in the U.S. civilian noninstitutionalized population, 1987 and 2001 Agency for Healthcare Research and Quality, Rockville, MD Stampfer M, Hu F, Manson J, Rimm E, Wille tt W (2000) primary prevention of coronary heart disease in women th rough diet and lifestyle. N Engl J Med 343: 16-22 Stevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, Wood JL (1998) The effect of age on the association between body-mass i ndex and mortality. N Engl J Med 338: 1

PAGE 50

50 Strine TW, Mokdad AH, Dube SR, Balluz LS Gonzalez O, Berry JT, Manderscheid R, Kroenke K (2008) The association of depres sion and anxiety with obesity and unhealthy behaviors among community-dwelling US adults. Gen Hosp Psych 30: 127-137 Sullivan PW, Morrato EH, GhushchyanV, Wyatt HR, Hill JO (2005) Obesity, inactivity, and the prevalence of diabetes and diabetes-relate d cardiovascular comorbidities in the U.S., 2000-2002. Diabetes Care 28: 1599-1603 Swenson CJ, Trepka MJ, Rewers MJ, S carbro S, Hiatt WR, Hamman RF (2002) Cardiovascular disease mortality in Hispanic s and Non-Hispanic Whites. Am J Epidemiol 156: 919-928 U.S. Department of Health and Human Se rvices. Healthy People 2010. 2nd ed. With understanding and improving heal th and objectives for improving health. 2 vols. Washington, DC: U.S. Government Printing Office, November 2000 Wagner EH, Austin BT, Davis C, Hindmarsh M, Schaefer J, Bono mi A(2001) Improving chronic illness care: tran slating evidence into ac tion. Health Aff 20: 64-78 Yusuf S, Hawken S, unpuu S, Dans T, Avezum A, Lanas Y, McQueen M, Budaj A, Pais P, Varigos J (2004) Effect of poten tially modifiable risk factors associated with myocardial infarction in 52 countries (t he INTERHEART study): case-cont rol study. The Lancet 11: 937-52

PAGE 51

51 BIOGRAPHICAL SKETCH Nathan Lawrence Ewigm an graduated with a Ba chelor of Arts degree in psychology in June 2006 from Knox College in Ga lesburg, Illinois. He is curre ntly pursuing a doctorate in clinical and health psychology and a masters degree in public health at the University of Florida. He received his M.S. from the University of Fl orida in the spring of 2009. His academic interests lie in primary care psychology, underserved populati ons and health services research. He is a Leo.