Signal Detection Analysis of Factors Associated With Diabetes Among Semirural Mexican American Adults

Abstract

Signal detection analysis was used to evaluate a combination of sociodemographic, acculturation, mental health, health care, and chronic disease risk factors potentially associated with diabetes in a sample of 4,505 semirural Mexican American adults. Overall, 8.9% of adults had been diagnosed with diabetes. The analysis resulted in 12 mutually exclusive groups, with diabetes prevalence ranging from 1.8% to 44.1%. Three groups were at the highest risk (34.5%-44.1% diabetic) and accounted for almost half of those with diabetes. Each of these groups was distinguished by their middle to older ages and presence of one or more chronic conditions (high blood pressure, high cholesterol, obesity, and/or poor mental health) in addition to diabetes. The differing composition of the groups identified by the signal detection analysis has important implications for the design and implementation of public health interventions and health care treatment programs for Mexican Americans with diabetes.

Keywords

diabetes Mexican American comorbidities signal detection analysis

Type II diabetes mellitus is a major cause of morbidity and mortality in the United States (Harris, 1991) and its prevalence continues to increase among U.S. adults (Mokdad et al., 2001). Compared with their European American counterparts, Latinos in the United States experience higher poverty rates and poorer health outcomes, including obesity and diabetes (CDC, 2011; Cowie et al., 2006). The risk of diagnosed diabetes is 66% higher among Latinos compared to non-Latino Whites. Among Latinos the risk of diagnosed diabetes is 87% higher for Mexican Americans (CDC, 2011). It has been predicted that two out of five Latino youth and one in two Latina youth born in 2000 will have diabetes by 2050 (Narayan, Boyle, Thompson, Sorensen, & Williamson, 2003).

Latinos are a very heterogeneous group in the United States, with distinct modes of incorporation into the cultural fabric and associated diversity in health outcomes, access, and health care needs. For example, acculturation among Mexican Americans has been shown to significantly correlate with a higher prevalence of obesity but is also associated with a lower likelihood of diabetes (Hubert, Snider, & Winkleby, 2005; Pérez-Escamilla & Putnik, 2007; Sundquist & Winkleby, 1999). This complexity applies when considering psychiatric and other comorbid conditions among Latinos. The prevalence of psychiatric disorders for Latinos is similar to non-Latino Whites, but increases with time in the United States (Cabassa, Zayas, & Hansen, 2006). However Latinos are less likely to use mental health services in part due to self-reliant attitudes toward healthcare (Berdahl & Torres Stone, 2009). For Mexican Americans, anxiety has been associated with a higher prevalence of diabetes and the association of psychiatric and comorbidities have not been fully explained by acculturation factors (Ortega, Feldman, Canino, Steinman, & Alegria, 2006). Since many studies of Latinos do not include mental health assessments, more research will help elucidate the pathways and resulting prevalence of psychiatric-comorbid conditions in various Latino subgroups. In particular, treatment and support opportunities may present with development of patient-centered medical homes, for example with coplacement of mental health and primary care.

Studying the factors associated with diabetes is important for developing and implementing effective public health interventions. Studies of diabetes in Latinos have generally assessed the relative importance of one or several factors associated with the onset of diabetes. However, due to collinearity among many of these factors and the complexity of interpreting higher order interaction terms from regression models, there are relatively few studies that have assessed a combination of sociodemographic, acculturation, mental health, health care, and chronic disease risk factors potentially associated with diabetes. Signal detection analysis (Kraemer, 1992) provides an opportunity to explore the interactions of these many variables. We applied this analytic methodology to a relatively culturally cohesive Mexican American population to provide insight into the interplay of acculturation and comorbidities, including mental health, and thus help guide further program refinement for diabetes management programs for Mexican Americans.

Method

Population Studied

Salinas, California, with 144,278 residents in 2009 (U.S. Census Bureau, 2011), is a semirural medically underserved community. Latinos make up 64% of the population, among whom 88% are of Mexican descent and 15% are agricultural workers (U.S. Census Bureau, 2011). Surveys of community and agricultural labor camp residents, conducted in 1990 and 2000 showed that this community may represent a relatively culturally cohesive population as most residents were young, had low education levels, spoke Spanish at home, and had lived in the United States 10 years or more. Furthermore, most were at continuing or increased risk for chronic diseases (Winkleby et al., 2006). These surveys found that over the 10-year period obesity increased substantially and diabetes was 3 to 6 times more common, high blood pressure was 2.5 to 3 times more common, and high cholesterol was 1.5 to 3 times more common in the heaviest compared to the leanest weight groups. Acculturation differences were seen in relation to obesity, with women and men who were obese being 1.6 times more likely than the leanest group to be second-generation, U.S. born.

For this analysis, we used data from the Centers for Disease Control and Prevention (CDC) augmented Behavioral Risk Factor Surveillance System (BRFSS) survey conducted annually in Salinas, California, from 2004 through 2008. BRFSS is an annual state-based system of surveys that uses random-digit dial Computer Assisted Telephone technology and specially trained interviewers to collect information on health risk behaviors, preventive health practices, and health care access primarily related to chronic disease and injury in adults 18 years and older (CDC, National Center for Chronic Disease Prevention and Health Promotion, 2011). Additional questions were added relating to acculturation (generation status, age of immigration, mother’s and father’s country of birth), as a large proportion of Salinas residents were born in Mexico. Completion rates for each year of the survey ranged from 53.1% to 58.6%, within the median range of response rates for the BRFSS. Criteria for inclusion in this analysis were as follows: (1) self-identification as Mexican American, (2) 18 years of age or older, (3) reported being diagnosed by a doctor as having diabetes, and (4) not currently pregnant.

Signal detection analysis was used to identify the variables most highly associated with being a diabetic adult. Signal detection is a nonparametric, distribution free modeling methodology (Kraemer, 1992). Unlike linear models, signal detection overcomes the problem of collinearity among predictor variables and is relatively sensitive to interactive effects of predictors. Signal detection identifies groups that are characterized by multiple factors and higher order interactions. The results are particularly useful for planning and implementing intervention programs, since groups that have distinct characteristics and levels of risk can be identified. We examined each predictor variable, with the analyses identifying variables that resulted in the greatest discrimination in relation to the outcome of having diabetes, based on maximizing both sensitivity and specificity.

The binary outcome variable for the signal detection analysis was “Have you ever been told by a doctor that you have diabetes?” (yes/no, excluding gestational diabetics). The most important predictor variable was identified first and split into optimal subgroups, and then each subgroup was explored in relation to all predictor variables to determine the next best predictor variable and its optimal split. The analyses continued until there were no more significant variables detected in a newly divided group at a level of p < 0.01 or there were not enough subjects in a subgroup (n<25).

Predictor variables as correlates of diabetes in Latinos were selected based on a literature review (Harris, 1991, Ortega et al., 2006, Winkleby et al., 2006, Burroughs et al., 2008):

Acculturation-related factors

Generation status: born in Mexico, born in United States with >1 parent born in Mexico, born in United States with neither parent born in Mexico;

Primary language(s) spoken at home: English only, Spanish only, both languages;

Years lived in the United States: <5, 5 to 9, 10 to 19, >20.

Sociodemographic factors

Gender: woman, man;

Age (years): 18 to 24, 25 to 44, 45 to 64, >65;

Education, highest grade or year completed: <12 years, 12 years, 13 to 15 years, >16 years;

Annual household income: <US$25,000, US$25,000 to US$49,999, >US$50,000;

Employment status: not in the labor force, employed in a blue-collar occupation, employed in a white-collar occupation;

Marital status: never married, previously married, currently married.

Health care-related

Health insurance status: had health insurance in last 12 months (yes/no);

Health care use: doctor seen in last 12 months (yes/no);

Affordability of doctor: could not afford medical care when needed in last 12 months (yes/no);

Affordability of prescription drugs: could not afford prescription drugs when needed in last 12 months (yes/no).

Mental Health

Number of days in past 30 days mental health not good;

Comorbidities/chronic disease risk factors

Hypertension, “Have you ever been told by a doctor, nurse, or other health professional that you have high blood pressure?” (yes/no);

Hypercholesterolemia, “Have you ever been told by a doctor, nurse, or other health professional that your cholesterol is high?” (yes/no);

Body mass index (BMI; kg/m²): calculated from self-reported height and weight and used as a bivariate variable of BMI<30 and BMI>30.

Descriptive summary statistics were generated using average for poor mental health days and proportions for ordinal, categorical, or dichotomous variables based on the subgroups identified by the signal detection analysis. All descriptive analyses were conducted using the Statistical Analysis System (SAS; SAS Institute Inc., 2010). The signal detection analyses were conducted using software developed by the Sierra Pacific Mental Illness Research Education and Clinical Center funded by the U.S. Department of Veterans Affairs. The stopping rule of p < 0.01 for the signal detection analysis was not a test of a statistical hypothesis, but was used to reduce the probability of false-positive outcomes that can be associated with exploratory data analysis. Responses to questions related to other lifestyle factors (diet and nutrition) and doctor’s advice during last health care visit were also assessed for the groups identified from the signal detection analysis. Relevant outcomes are reported by subgroup.

Results

The majority of the sample was women (60.5%), and adults who were younger than 45 years of age (69.8%) and married (70.6%; Table 1). Educational attainment and household income were low, with 58.0% of the sample having less than 12 years of education and 57.3% reporting an annual household income of <US$25,000. Although 73.9% were born in Mexico, 80.9% had lived in the United States for 10 years or longer. While the majority reported having some health care coverage and seeing a doctor in the last 12 months, approximately 25.0% had not been able to afford medical care or prescription medications in the last 12 months. One in four respondents reported having hypertension and/or hypercholesterolemia, one in three was obese, and respondents had an average of 4.4 poor mental health days in the past month.

Table 1.

Sociodemographic and Health Profile, Mexican Americans, Aged 18 to 88 Years, Behavioral Risk Factor Surveillance System (BRFSS) Survey, Salinas, California, 2004-2008.

Variable	Sample size^a	Percent
Sociodemographics
Gender
Women	2,726	60.5
Men	1,779	39.5
Age
18-24 years	509	11.3
25-44 years	2,637	58.5
45-64 years	1,127	25.0
≥ 65 years	232	5.2
Education
< 12 years	2,556	58.0
12 years	944	21.4
13-15 years	553	12.6
≥ 16 years	352	8.0
Annual household income
< US$25,000	2,369	57.3
US$25,000-US$49,999	1,111	26.9
≥ US$50,000	651	15.8
Marital status
Never married	635	14.4
Previously married	662	15.0
Currently married	3,116	70.6
Primary language spoken at home
English only	906	20.5
Spanish only	3,041	68.7
Both languages	477	10.8
Years lived in U.S.
< 5	319	7.1
5-9	536	12.0
10-19	1,230	27.6
≥ 20	2,378	53.3
Generation status
Born in Mexico	3,291	73.9
Born in U.S., ≥ 1 parent born in Mexico	441	9.9
Born in U.S., neither parent born in Mexico	724	16.2
Employment status
White collar	1,671	38.5
Blue collar	1,849	42.6
Not in labor force	820	18.9
Health care-related
Had health care coverage last 12 months	2,808	62.4
Doctor seen in last 12 months	3,370	75.0
Could not afford medical care last 12 months	1,128	25.0
Could not afford prescription medicine last 12 months	1,047	23.3
Health conditions
Hypertension, ever told by a doctor	939	20.9
Hypercholesterolemia, ever told by a doctor	864	19.5
Obesity (BMI ≥30 kg/m²)	1,320	32.4
Mental health, number of days in past 30 days poor mental health (mean)	4.4	7.9

Sample sizes may not add up to 4,505 because of missing data. BMI = body mass index.

The overall prevalence of diabetes was 8.9%; 400 cases among the sample of 4,505 individuals. The signal detection analysis resulted in 12 mutually exclusive groups with diabetes prevalence ranging from a low of 1.8% to a high of 44.1% (Figure 1). Three groups (Groups 8, 11, and 12) were at the highest risk for diabetes (34.5%-44.1% diabetic), and accounted for almost half of those with diabetes. Each of these three groups was distinguished by their middle to older ages and presence of one or more other chronic diseases in addition to diabetes (high cholesterol plus obesity for Group 8, poor mental health plus diabetes for Group 11, and high blood pressure plus diabetes for Group 12). The largest of these three high-risk groups was Group 12 (36.8% of all diabetics) that was composed of hypertensive individuals who were 55 to 88 years of age among whom 43.1% were diabetic.

Figure 1.

Signal detection analysis on diabetes in Mexican Americans, aged 18 to 88, Behavioral Risk Factor Surveillance System (BRFSS) Survey, Salinas, California, 2004–2008. Outcome variable: Were you ever told by a doctor that you had diabetes (yes or no)? For full list of predictor variables, see Table 1. The total number of diabetics in the final 12 subgroups is slightly lower than the 400 total cases because of missing values for some predictor variables identified by the signal detection analysis.

Age was identified as the optimal variable to predict diabetes status and split the sample into two groups ( $χ_{1, N = 4, 505}^{2}$ = 505.5, p<0.0001). Diabetes prevalences were six times higher in those 55 years of age and older (33.4%) as compared to those less than 55 years old (5.2%). The optimal variables, apart from age that identified further subgroups were factors representing comorbid conditions, mental health, ability to pay for medications, and for many groups, a combination of two or more of these factors. After the initial split on age, the analysis split on those with normal and high cholesterol, followed by age again, and then high blood pressure, affordability of prescription drug use, and obesity. The highest burden of diabetes (44.1%) was carried by those in Group 8 whose members were only 47 to 54 years of age, had high cholesterol, and were obese. Their diabetes prevalence was more than twice that of those in this age group with high cholesterol who were not obese ( $χ_{1, N = 177}^{2}$ = 13.0, p = 0.0003). In those who were 18 to 46 years of age, the model split again on obesity. Those in Group 6 who were 18 to 46 and obese were also more than twice as likely to have diabetes as those who were not obese (20.9% as compared to 7.9%).

A significantly higher diabetes burden (43.1%) was carried by group 12, whose members were 55 and older and hypertensive ( $χ_{1, N = 589}^{2}$ = 34.5, p<0.0001). They were more than twice as likely to have diabetes as those who were not hypertensive (19.9%). Even among individuals who were younger (aged 18-49), those who were hypertensive (Group 2) were more than three times as likely to have diabetes as those who were not hypertensive ( $χ_{1, N = 3, 019}^{2}$ = 31.9, p<0.0001).

Poor mental health also distinguished some groups. Those 55 and older who had normal blood pressure but who had four or more poor mental health days in the past thirty days had a significantly higher prevalence of diabetes than those with better mental health ( $χ_{1, N = 246}^{2}$ = 9.4, p = 0.002). This group (Group 11) had the third highest diabetes prevalence at 34.5%.

There were group-specific characteristics and levels of risk of the 12 signal detection groups.¹ While groups had varying prevalences for diabetes and most comorbidities, all groups had a median BMI >29 indicating that one half of the sample were overweight. Looking at the first age group split (Groups 1-8 vs. 9-12), it appears that comorbidities and mental health days play an additive role in relation to diabetes prevalence. In the 18 to 54 aged groups, Group 1 had the lowest diabetes prevalence (1.8%), the lowest number of poor mental health days (3.7 days), and only one comorbidity (obesity, 26.4%) while the highest diabetes prevalence group (Group 8 at 44.1%) had the third highest number of days of poor mental health (8.3 days), and high prevalences for three comorbidities (obesity, 100%; and high blood pressure, 47.1%; and high cholesterol, 100%). In this age range, the group with the second highest diabetes prevalence (Group 4 at 29.1%) had the second highest number of poor mental health days (8.8 days) of all 12 groups and moderate prevalences for two comorbidities (obesity, 33.3%; and hypertension, 48.3%). It was also the one group with a complex interaction of acculturation, sociodemographic, and health care access factors. Group 4 had higher percentages of individuals who had less than a high school education (80.4%), fewer years in the United States (13.8 % < 10 years), were primarily Spanish speaking (80.4%), blue-collar workers (58.9%), had no health insurance (56.9%), had not seen a doctor in the past year (27.6%), and could not afford a doctor (77.6%) or medications (100%) over the past year. Among those in the groups representing ages 47 to 88, a similar pattern was seen as in several groups in the lower age ranges, with only minor variability in acculturation, sociodemographic, and health care access variables. However, among groups with the highest diabetes prevalences (Groups 11 and 12, 42% of the diabetics, diabetes prevalences of 34.5% and 43.1%, respectively) had higher prevalences for comorbidities. Group 11 experienced the highest average number of poor mental health days over the past month (16.5 days). These two groups also had higher prevalences for obesity (60.0% and 43.9%) and high cholesterol (60.4% and 57.6%) when compared to groups with a substantially lower prevalence of diabetes (Group 9, with a diabetes prevalence of 9.2% and 17.2% obese and 37.8% with high cholesterol).

Health insurance rates were lower than the rates for visiting a doctor for all but Group 10. For all groups with higher prevalences of diabetes and comorbidities, rates of receiving doctors’ advice for nutrition and physical activity ranged from 62.3% to 82.1%. Irrespective of diabetes prevalence or age, all groups had poor adherence to recommended nutrition and physical activity guidelines and 39.4% to 63.4% had consumed more than one soda on the previous day.

Discussion

This study is one of the first to demonstrate the complex interplay between socioeconomic and comorbid conditions (hypertension, hypercholesterolemia, obesity, poor mental health) in relation to diabetes prevalence in a community-based sample of Mexican Americans. This Mexican American community had an overall diabetes population prevalence of 8.9%, similar to that seen nationally for Latinos (CDC, 2011). However, the twelve mutually exclusive subgroups identified by signal detection analysis had considerable variability in their diabetes prevalences, ranging from a low of 1.8% to a high of 44.1%. Several groups with high rates of co-occurring morbidities had diabetes prevalences four or more times the national average, underscoring the poor health burden experienced by diabetic Mexican Americans in this community.

The differing composition of the groups identified by the signal detection analysis has important implications for the design and implementation of public health interventions and health care treatment programs for Mexican Americans with diabetes. The two groups with 100% prevalence of obesity (Group 8) and 100% prevalence of high blood pressure (Group 12) and 44.1% and 43.1% prevalences for diabetes are at particular risk for serious health complications. These two groups had moderate access to care, providing opportunities for the development of health care-associated intervention programs, such as targeted diabetes self-management behavioral programs. Such programs should consider the groups’ varying levels of acculturation, as culturally mediated factors can exert a more pervasive influence on diabetes and obesity in Mexican Americans than socioeconomically mediated factors (Hazuda, Haffner, Stern, & Eifler, 1988). Self-management behavioral activities can be effective for Mexican Americans (Martinez & Bader, 2007), but need modification to be as effective as in non-Hispanic Whites (Kurian & Borders, 2006). The chronic care model framework applied to chronic disease management programs (Wagner, 1998) may be useful for addressing such complexities of care required by diabetic patients with multiple comorbidities. Given the co-occurring mental health needs for several subgroups with high diabetes prevalences identified in this analysis, the opportunity to provide mental health services along with primary care, potentially through colocation of services, is an important part of such a framework. It will be critical for Affordable Care Act expansion of secondary prevention programs for diabetes to incorporate a framework informed by studies such as this one if we are to be effective in meeting health care reform challenges and escalating diabetes prevalence among all racial/ethnic groups.

Groups 1, 2, 3 and 5 had diabetes prevalences less than the national average. However, the median BMI for these groups was >25, and Group 2 had high blood pressure and Group 5 had high cholesterol as well. In addition, though two thirds to three quarters of the respondents in these groups were trying to lose weight through exercise, they generally had poor nutrition habits. Thus individuals in these subgroups display a panorama of factors that are associated with diabetes (CDC, 2011), possibly indicating that those of the group who do not currently have diabetes are on a pathway to developing diabetes. In addition, individuals in these groups were young (<54 years of age) and higher proportions of these groups had spent less time in the United States, and though they had moderate levels of education and income compared to the other groups, they were still uninsured at rates less than the national average for Mexican Americans (27%; Rhoades & Vistnes, 2004). A third of those with high blood pressure had not received medical advice in the past year to eat more fruits or vegetables or to exercise more, even though this factor is highly correlated with developing serious health conditions such as diabetes and cardiovascular disease (Chobanian et al, 2003).

Group 4 stood out in that they were less acculturated, less educated, had less income, had the second highest number of poor mental health days in the past month, and had much lower access to care in the past year compared to the other subgroups. This group may represent an important subgroup of diabetics within the greater Mexican American population, who might benefit from more culturally oriented health interventions which consider their socioeconomic situation, Spanish language preference, possible low literacy skills, and mental health support needs. Programs using a peer educator model are often presented as a low cost approach to self-management education (Rhodes, Foley, Zometa, & Bloom, 2007) and particularly effective when tailored to accommodate the cultural, economic, and learning needs of low-income adults with low literacy skills (Howard-Pitney, Winkleby, Albright, Bruce, & Fortmann, 1997). Peer-led diabetes education programs have demonstrated improvement in glucose and metabolic control for low-income, low education, diabetic Mexican Americans when compared with standard approaches (Philis-Tsimikas, Fortmann, Lleva-Ocana, Walker, & Gallow, 2011) and other needs, such as for mental health, can be incorporated into the peer training (Reinschmidt & Chong, 2007). Surveys with this population indicate using peer-training programs would be culturally supported (Hanni, Mendoza, Snider, & Winkleby, 2007; Hanni, Garcia, Ellemberg, & Winkleby, 2009). Such programs must also consider low-cost options for those who cannot afford prescription medications like in Group 4, especially if other factors such as legal status (not asked in our survey) would preclude qualification for reimbursement for medication costs through government programs.

It takes a support network of the individual, friends and family, and health care providers to reduce the occurrence of diabetes complications (CDC, 2011). While we identified health factors that were the most optimal predictors of diabetes in this population, these and other factors may also play a significant role in the ability of Mexican American diabetics to manage their disease. For example, some studies have shown that racial/ethnic discrimination is associated with lower rates of diabetes management when controlling for a variety of health and demographic factors (Ryan, Gee, & Griffith, 2008). Future studies should assess barriers in this population to optimum diabetes self-management.

This analysis was limited to a sample of semirural Mexican Americans in Monterey County, California, and one should be cautious in extrapolating the results to Mexican Americans in general. The cross-sectional analysis limits the ability to make inferences about causality. There is also variability in self-reported health measures from health surveys, ranging from underestimates of weight by Mexican Americans by up to 20% (Gillum & Sempos, 2005), to varying sensitivity and specificity for various groupings for high cholesterol (Natarajan, Lipsitz, & Nietert, 2002), to overall high reliability for diabetes (Espelt, Goday, Franch, & Borrell, 2011). While Type II diabetes most likely accounts for 90 to 95% of the cases of diabetes in this sample, our survey did not differentiate between Type I or Type II diabetes nor did it ask about enrollment in diabetes or other health outcome-related interventions. A final limitation of the study is the relatively low response rates of approximately 50% for each survey.

It can be difficult to capture the complexity of acculturation using proxy measurements such as generation status, language usually spoken at home, and years lived in the United States (Perez-Escamilla & Putnik, 2007; Wallace, Pomery, Latimer, Martinez, & Salovey, 2010). However, studies have indicated that the addition of social context and psychosocial measures to analyses, which include individual acculturation measures serve to contextualize results, especially in regards to various health outcomes such as diabetes, nutrition, healthy lifestyles, and self-reported health among Latinos (Perez-Escamilla & Putnick 2007; Wallace et al. 2010; Johnson, Carroll, Fulda, Cardarelli, & Cardarelli, 2010).

This exploratory analysis indicates that providers need to consider the interplay of various socioeconomic, comorbid, acculturation, and mental health factors for different age groups of Mexican Americans with diabetes when considering appropriate intervention and treatment strategies. While we do not know the reasons for the doctor’s visits in the prior year, the adults in this study received only moderate amounts of appropriate nutrition and exercise messaging at their last doctor’s visit despite potentially life-threatening health conditions. This suggests a need for more routine and effective behavior change messaging and culturally appropriate diabetes interventions for Mexican American communities. Tailoring diabetes interventions to specific needs of particular subgroups of Mexican Americans will be important if we are to reduce medical expenditures among people diagnosed with diabetes—currently 2.3 times higher than what expenditures would be in the absence of diabetes (ADA, 2008).

Footnotes

Acknowledgements

Special thanks to colleagues involved with the Salinas Steps program.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for the Steps to a Healthier Salinas Program was provided by the U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, as part of its Steps to a Healthier U.S. program.

Notes

Author Biographies

K. D. Hanni has a doctorate from the University of California at Davis in epidemiology and currently works for the Monterey County Health Department as Planning, Evaluation, and Policy Manager. Her current work focuses on access to care and health inequities, including health in all policies, and measuring the impact and outcome of public health programs.

D. A. Ahn has a doctorate in applied statistics from the Stanford University School of Education. He is a biostatistician/senior computer analyst at the Stanford Prevention Research Center where he has applied a wide range of statistical methodologies in the areas of cancer epidemiology, cardiovascular risk factors, nutrition studies, healthy aging studies, and neighborhood influences on health.

M. A. Winkleby has a doctorate in epidemiology from the University of California at Berkeley. She is a professor of medicine at the Stanford Prevention Research Center, and faculty director of the Office of Community Health at the Stanford School of Medicine. Her work has focused on cardiovascular disease health behaviors, risk factors and outcomes (smoking, physical inactivity, poor nutrition, obesity, diabetes, hypertension); health status of ethnic minority and low socioeconomic groups; and community-based intervention studies.

References

American Diabetes Association (2008). Economic costs of diabetes in the United States in 2007. Diabetes Care, 31, 596-615.

Berdahl

T. A.

Torres Stone

R. A.

(2009). Examining Latino differences in mental healthcare use: The role of acculturation and attitudes towards healthcare. Community Mental Health Journal, 45, 393-403.

Burroughs

V. J.

Nonas

Sweeney

C. T.

Rohay

J. M.

Harkins

A. M.

Kyle

T. K.

Burton

S. L.

(2008). Self-reported comorbidities among self-described overweight African-American and Hispanic adults in the United States: results of a national survey. Obesity, 16, 1400-1406.

Cabassa

L. J.

Zayas

L. H.

Hansen

M. C.

(2006). Latino adults’ access to mental health care: A review of epidemiological studies. Administration and Policy Mental Health and Mental Health Services Research, 33, 316-330.

Centers for Disease Control and Prevention (2011). National diabetes fact sheet: National estimates and general information on diabetes and prediabetes in the United States, 2011. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention.

Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion (2011). Behavioral risk factor surveillance system. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention.

Chobanian

A. V.

Bakris

G. L.

Black

H. R.

Green

L. A.

Izzo

J. L.

Jones Roccella

E. J.

(2003). Seventh report of the Joint National Committee on prevention, detection, evaluation, and treatment of high blood pressure. Hypertension, 42, 1206-1252.

Cowie

C. C.

Rust

K. F.

Byrd-Holt

D. D.

Eberhardt

M. S.

Flegal

K. M.

Engelgau

M. M.

Gregg

E. W.

(2006). Prevalence of diabetes and impaired fasting glucose in adults in the U.S. population, National Health and Nutrition Examination Survey 1999-2002. Diabetes Care, 29, 1263-1268.

Espelt

Goday

Franch

Borrell

(2011). Validity of self-reported diabetes in health interview surveys for measuring social inequalities in the prevalence of diabetes. Journal of Epidemiology of Community Health. doi:10.1136/jech.2010.112698

10.

Gillum

R. F.

Sempos

(2005). Ethnic variation in validity of classification of overweight and obesity using self-reported weight and height in American women and men: The third national health and nutrition examination survey. Nutrition Journal, 4, 27.

11.

Hanni

K. D.

Garcia

Ellemberg

Winkleby

M. A.

(2009). Targeting the taqueria: implementing healthy food options at Mexican-American restaurants. Health Promotion Practice, 10, 91S-99S.

12.

Hanni

K. D.

Mendoza

Snider

Winkleby

M. A.

(2007). A methodology for evaluating organizational change in community-based chronic disease interventions. Preventing Chronic Disease, 4, A105.

13.

Harris

M. I.

(1991). Epidemiological correlates of NIDDM in Hispanics, whites, and blacks in the U.S. population. Diabetes Care, 14, 639-638.

14.

Hazuda

H. P.

Haffner

S. M.

Stern

M. P.

Eifler

C. W.

(1988). Effects of acculturation and socioeconomic status on obesity and diabetes in Mexican Americans, The San Antonio heart study. American Journal of Epidemiology, 128, 1289-1301.

15.

Howard-Pitney

Winkleby

M. A.

Albright

C. L.

Bruce

Fortmann

S. P.

(1997). The Stanford nutrition action program: A dietary fat intervention for low-literacy adults. American Journal of Public Health, 87, 1971-1976.

16.

Hubert

Snider

Winkleby

M. A.

(2005). Health status, health behaviors, and acculturation factors associated with overweight and obesity in Latinos from a community and agricultural labor camp survey. Preventive Medicine, 40, 642-651.

17.

Johnson

K. L.

Carroll

J. F.

Fulda

K. G.

Cardarelli

(2010). Acculturation and self-reported health among Hispanics using a socio-behavioral model: The North Texas healthy heart study. BMC Public Health, 10, 53.

18.

Kiernan

Kraemer

H. C.

Winkleby

M. A.

King

A. C.

Taylor

C. B.

(2001). Do logistic regression and signal detection identify different subgroups at risk? Implications for the design of tailored interventions. Psychology Methods, 6, 35-48.

19.

Kraemer

H. C.

(1992). Evaluating medical tests: Objective and quantitative guidelines. Newbury Park, CA: SAGE.

20.

Kurian

A. K.

Borders

T. F.

(2006). Racial and ethnic differences in the effects of regular providers and self-management education on diabetes preventive care. Ethnicity and Disease, 16, 786-791.

21.

Martinez

N. C.

Bader

(2007). Analysis of behavioral risk factor surveillance system data to assess the health of Hispanic Americans with diabetes in El Paso County, Texas. Diabetes Education, 33, 691-699.

22.

Mokdad

A. H.

Bowman

B. A.

Ford

E. S.

Vinicor

Marks

J. S.

Koplan

J. P.

(2001). The continuing epidemics of obesity and diabetes in the United States. Journal of the American Medical Association, 286, 1195-1200.

23.

Narayan

K. M.

Boyle

J. P.

Thompson

T. J.

Sorensen

S. W.

Williamson

D. F.

(2003). Lifetime risk for diabetes mellitus in the United States. Journal of the American Medical Association, 290, 1884-1890.

24.

Natarajan

Lipsitz

S. R.

Nietert

P. J.

(2002). Self-report of high cholesterol: Determinants of validity in U.S. adults. American Journal of Preventive Medicine, 23, 13-21.

25.

Ortega

A. N.

Feldman

J. M.

Canino

Steinman

Alegria

(2006). Co-occurrence of mental and physical illness in U.S. Latinos. Social Psychiatry Psychiatric Epidemiology, 41, 927-934.

26.

Pérez-Escamilla

Putnik

(2007). The role of acculturation in nutrition, lifestyle, and incidence of type 2 diabetes among Latinos. Journal of Nutrition, 137, 860-870.

27.

Philis-Tsimikas

Fortmann

Lleva-Ocana

Walker

Gallow

L. C.

(2011). Peer-led diabetes education programs in high-risk Mexican Americans improve glycemic control compared with standard approaches: A Project Dulce promotora randomized trial. Diabetes Care, 34, 1926-1931.

28.

Reinschmidt

K. M.

Chong

(2007). SONRISA: A curriculum toolbox for promotoras to address mental health and diabetes. Preventing Chronic Disease, 4, A101.

29.

Rhoades

Vistnes

(2006). Health insurance status of Hispanic subpopulations in 2004: Estimate for the U.S. civilian noninstitionalized population under age 65. Rockville, MD: Agency for Healthcare Research and Quality.

30.

Rhodes

S. D.

Foley

K. L.

Zometa

C. S.

Bloom

F. R.

(2007). Lay health advisor interventions among Hispanics/Latinos: A qualitative systematic review. American Journal of Preventive Medicine, 33, 418-427.

31.

Ryan

A. M.

Gee

G. C.

Griffith

(2008). The effects of perceived discrimination on diabetes management. Journal of Health Care for the Poor and Underserved, 19, 149-163.

32.

SAS Institute Inc. (2010). SAS Enterprise Guide 9.2, help and documentation. Cary, NC: SAS Institute Inc.

33.

Sundquist

Winkleby

(2000). Country of birth, acculturation status and abdominal obesity in a national sample of Mexican-American women and men. International Journal of Epidemiology, 29, 470-477.

34.

U.S. Census Bureau. (2011). State and county quick facts. Retrieved from http://quickfacts.census.gov/qfd/states/06/0664224.html

35.

Wagner

E. H.

(1998). Chronic disease management: What will it take to improve care for chronic illness? Effective Clinical Practice, 1, 2-4.

36.

Wallace

P. M.

Pomery

E. A.

Latimer

A. E.

Martinez

J. L.

Salovey

(2010). A review of acculturation measures and their utility in studies promoting Latino health. Hispanic Journal of Behavioral Science, 32, 37-54.

37.

Winkleby

M. A.

Kim

Urizar

G. G.

Ahn

Jennings

M. G.

Snider

(2006). Ten-year changes in cancer-related health behaviors and screening practices among Latino women and men in California. Ethnicity and Health, 11, 1-17.