Evaluation of race / ethnic disparities in patterns of diabetes personal health behaviors

Kimberly J Wilson, The University of Texas School of Public Health


It is recognized that Hispanics and non-Hispanic Blacks (NHB) have higher prevalence of diabetes and worse outcomes related to it. The literature is mixed on the extent to which these poorer outcomes are due to lower quality of care and adherence to recommended self-management practices. Studies have investigated race/ethnic disparities in individual management behaviors, with mixed results. However, no studies have sought to identify how these individual practices may be linked into patterns of diabetes care. Furthermore, a gap in knowledge exists about how predictors of intermediate diabetes outcomes are related to one another.^ The dual objectives of this dissertation were to (1) evaluate whether race/ethnic disparities exist in patterns of diabetes management behaviors, and (2) identify patterns of patient characteristics and care behaviors that predict success at meeting recommended A1c, blood pressure and LDL cholesterol targets. ^ For the first objective, I used data from the 2011 MEPS and latent class analysis to identify discrete categories of diabetes management behaviors. I then used multinomial logistic regression to determine whether race/ethnicity increased individuals' risk of belonging to one or more of the categories, before and after adjusting for need, predisposing and enabling characteristics. Because race/ethnicity and income are strongly related, I also assessed whether race/ethnicity moderated income's association with sub-optimal diabetes management. For the second objective, I utilized data from the 2007-2008 and 2011-2012 rounds of NHANES and a tree classification technique to create groups with statistically significant levels of success at meeting ADA-recommended A1c, blood pressure and LDL cholesterol targets using demographic (including race/ethnicity), health status and health behavior variables.^ For the first objective, four classes of behaviors were identified. In unadjusted models, being NHB or Hispanic increased risk relative to non-Hispanic Whites (NHW) of engaging in low-intensity management (RR=1.50, p<.05, and RR=2.03, p<.001, respectively). With the inclusion of confounders, the disparities were no longer statistically significant. Adding interaction terms for income and race/ethnicity indicated probable moderation with the effect of low income being a strong predictor for NHW of engaging in sub-par diabetes management, but not for NHB or Hispanics. ^ For the final objective, multiple clusters defined by both patient characteristic and behavior variables were identified for each outcome. Race/ethnicity played roles in clustering in the blood pressure and cholesterol models. Only 13.9% of NHB with one or more comorbidities who had higher than average daily intake of carbohydrates met the blood pressure goal, compared to 48.2% for the total sample. In addition, NHB and Hispanic women aged 65 years and over were less likely to meet the LDL cholesterol goal than their male counterparts (46.2% vs. 67.3%).^ The results of these studies indicate that being Hispanic or NHB increases one's likelihood of sub-par diabetes management with differences in age, perceived disease severity, education, income and insurance coverage explaining the disparities. In addition, minority status was independently associated with lower probability of meeting blood pressure and cholesterol goals for some individuals. These findings highlight the importance of removing barriers to care and highlight populations that would benefit from targeted education and care follow-up.^

Subject Area

Behavioral psychology|Public health|Ethnic studies|Health care management

Recommended Citation

Wilson, Kimberly J, "Evaluation of race / ethnic disparities in patterns of diabetes personal health behaviors" (2015). Texas Medical Center Dissertations (via ProQuest). AAI3732060.