Impact of Stroke Risk Factors on Ethnic Stroke Disparities among Midlife Mexican Americans and Non-Hispanic Whites

Abstract

Background and Purpose

We examined the contribution of stroke risk factors to midlife (age 45–59) Mexican American (MA) and non-Hispanic White (NHW) ischemic stroke (IS) rate disparities from 2000–2010.

Methods

Incident IS cases (N = 707) and risk factors were identified from the BASIC Project, Nueces County, Texas (2000–2010). US Census data (2000–2010) was used to estimate the population at-risk for IS, and the Behavioral Risk Factor Surveillance System (2000–2010) was used to estimate risk factor prevalence in the stroke free population. Poisson regression models combined IS counts (numerator) and population at-risk counts (denominator) classified by ethnicity and risk factor status to estimate unadjusted and risk factor adjusted associations between ethnicity and IS rates. Separate models were run for each risk factor and extended to include an interaction term between ethnicity and risk factor.

Results

The crude rate ratio (RR) for ethnicity (MA vs. NHW) was 2.01 (95% CI: 1.71, 2.36), and was attenuated in models that adjusted for diabetes (RR: 1.50, 95% confidence interval (CI): 1.26, 1.78) and hypertension (RR:1.84, 95% CI: 1.50, 2.26). Additionally, diabetes had a stronger association with IS rates among MAs (RR: 6.42, 95% CI: 5.31, 7.76) compared with NHWs (RR: 4.07, 95% CI: 3.68,4.51).

Conclusions

The higher prevalence of diabetes and hypertension and stronger association of diabetes with IS among midlife MAs likely contribute to persistent midlife ethnic stroke disparities.

INTRODUCTION

Overall stroke incidence has decreased in the United States in the past decade, but ischemic stroke incidence has remained stable or increased among midlife adults (ages 45–59), and racial-ethnic stroke risk disparities remain greatest in this age range. Stroke risk disparities among midlife Mexican Americans (MAs), as compared to midlife non-Hispanic Whites (NHWs), have been especially high and persistent¹. Several hypotheses might explain the ethnic disparity in this age group, including a higher prevalence of risk factors, greater influence of stroke risk factors, and challenges accessing primary stroke prevention among midlife MAs^2–4. We examined the contribution of traditional stroke risk factors to ethnic differences in stroke rates among midlife MAs and NHWs. Additionally, we examined if stroke risk factors had differential associations with ischemic stroke rates in midlife MAs as compared to midlife NHWs.

METHODS

Data sources

Brain Attack Surveillance in Corpus Christi (BASIC)

BASIC is an ongoing stroke surveillance study in Nueces County, Texas. A combination of active and passive stroke surveillance, as well as geographic isolation from other major cities, provides an opportunity to capture all cerebrovascular events in the county⁵. Only incident ischemic strokes among MAs and NHWs aged 45–59 residing in Nueces County, Texas identified from 2000–2010 were included in the current study.

Stroke risk factor data (diabetes, hypertension, high cholesterol, coronary heart disease, smoking status, less than high school education, health care coverage) and demographics for stroke cases were extracted from medical records or patient interviews. Data on body mass index was only available beginning in 2005, and was categorized as >30 or ≤30 for the subgroup of cases with this data. Because education level was only available for the subset of cases who were interviewed in the BASIC project, missing values were imputed (please see http://stroke.ahajournals.org). Study was approved by the University of Michigan Institutional Review Board, and the Institutional Review Board of local hospital systems.

Behavioral Risk Factor Surveillance System (BRFSS)

BRFSS is a cross-sectional telephone survey that collects self-reported data on health related risk behaviors. BRFSS data from Public Health Region 11 was used to estimate stroke risk factor prevalence for the stroke-free population in Nueces County. Data from the region was pooled from 2000–2010, and limited to self-identified Hispanics and NHWs aged 45–59. Stroke risk factors included the same risk factors collected in the BASIC study. Multiple imputation was used to account for missing data (please see http://stroke.ahajournals.org).

Statistical Analysis

Poisson regression models were run combining yearly stroke counts (numerator) and yearly populations at-risk counts (denominator) to estimate unadjusted and risk factor adjusted (separate models for each risk factor) associations between ethnicity (MA versus NHW) and rate of ischemic stroke. For each imputed dataset, separate regressions were run to obtain the rate ratio (RR) and 95% confidence intervals (CI) for ethnicity adjusted for each risk factor and time modeled as years since 2000. Each risk factor model was extended to include an interaction term between ethnicity and the risk factor. Risk factor adjusted ethnic RR’s and ethnic-specific interaction effect estimates from each imputed dataset were then combined to obtain pooled RR’s and 95% CI estimates.

RESULTS

A total of 707 incident ischemic strokes among those 45–59 years were identified, 493 among MAs and 214 among NHWs. Stroke risk factor prevalence among stroke cases and the stroke-free population is reported in Supplemental Table 1 (please see http://stroke.ahajournals.org).

The crude RR comparing ischemic stroke rates in MAs and NHWs was 2.01 (95% CI: 1.71, 2.36). Models that included diabetes (RR: 1.50; 95% CI: 1.26, 1.78) and hypertension (RR: 1.84; 95% CI: 1.50, 2.26) attenuated the RR for ethnicity compared to the crude model (Table 1). Models including current smoking status and no health insurance increased the RR for ethnicity compared to the crude model.

Table 1.

Unadjusted and Adjusted Incidence Rate Ratios Among Midlife (45–59) Mexican Americans and Non-Hispanic Whites, Nueces County, Texas, 2000–2010

	Point Estimate	Lower Confidence Limit	Upper Confidence Limit	Percent Change
Crude*	2.01	1.71	2.36	---
Risk Factor†
Diabetes	1.50	1.26	1.78	−25.45
Hypertension	1.84	1.50	2.26	−8.31
Coronary Artery Disease	2.04	1.70	2.44	1.63
High Cholesterol	2.07	1.73	2.47	3.14
Body Mass Index > 30	2.10	1.63	2.71	4.71
No Insurance	2.19	1.83	2.61	8.91
Current Smoker	2.13	1.79	2.52	5.91
Less than High School	1.93	1.62	2.30	−3.80

^*Unadjusted Model: Stroke Risk = Intercept + years since 2000 +β₁*Ethnicity

^†Adjusted Model: Stroke Risk = Intercept + years since 2000+ β₁*Ethnicity + β₂*Risk Factor

Stroke risk factors with significant interaction (P < 0.20) with ethnicity included diabetes, no insurance, and less than high school education (Table 2). The stroke RR comparing MAs (RR: 6.42; 95% CI: 5.31, 7.76) with and without diabetes was greater than the RR among NHWs (RR: 4.07; 95% CI: 3.68, 4.51), while the stroke RR comparing MAs (RR: 0.59; 95% CI: 0.48, 0.71) with and without health insurance was in the opposite direction than the RR among NHWs (RR: 2.60; 95% CI: 1.89, 3.58). Stroke RRs comparing NHWs with and without education less than high school were greater than those in MAs (Table 2).

Table 2.

Midlife (45–59) Stroke Rate Ratios Comparing Those With and Without Risk Factor by Ethnicity, Nueces County, Texas, 2000–2010

	Mexican American			non-Hispanic White
Risk Factors	Rate Ratio	LCL	UCL	Rate Ratio	LCL	UCL
Diabetes*	6.42	5.31	7.76	4.07	3.68	4.51
Hypertension	4.11	3.15	5.36	3.58	2.31	5.53
Coronary Artery Disease	4.88	3.22	7.41	4.59	2.39	8.83
High Cholesterol	2.21	1.68	2.90	2.00	1.36	2.93
No Insurance*	0.59	0.48	0.71	2.60	1.89	3.58
Body Mass Index > 30	1.33	1.02	1.75	2.00	1.30	3.09
Current Smoker	2.76	2.26	3.36	3.04	2.26	4.07
Less than High School*	0.92	0.77	1.10	4.12	2.83	5.99

Abbreviations: LCL, Lower Confidence Limit; UCL, Upper Confidence Limit.

^*P <0.20

DISCUSSION

We found that midlife MAs have double the ischemic stroke risk compared to NHWs. The higher prevalence of diabetes and, to a lesser extent, hypertension in MAs contributed to their greater midlife stroke rates. Additionally, the influence of diabetes on ischemic stroke rates was significantly greater in midlife MAs compared with midlife NHWs.

While causes of race-ethnic disparities in diabetes and hypertension have been elucidated and interventions aimed at preventing and treating these risk factors have been developed, disparities remain largely unabated ⁶. Social epidemiologists argue that race-ethnic inequalities in cardiovascular risk factors and the downstream clinical outcomes, such as stroke, will persist unless upstream causes of disparities, including poor social conditions and low socioeconomic status that are prevalent among minority populations, are addressed ⁷. Although health insurance did not explain midlife ischemic stroke rate disparities in our analysis, there were large ethnic differences in health insurance status with nearly half of stroke-free MAs being uninsured. Therefore, policies that improve healthcare access may be particularly important for primary stroke prevention in midlife MAs. In particular, policies that curtail disparities in risk factors in early life (e.g., disparities in childhood obesity) may be more successful in the long-term by reducing disparities in major risk factors in midlife ⁸.

Reasons for the stronger association of diabetes with ischemic stroke among MAs compared to NHWs are not known but may be due race-ethnic variation in the pathogenesis of diabetes or more likely to greater rates of uncontrolled diabetes among midlife MAs ⁹. Additionally, it is unclear why lack of health insurance was protective for MAs, but this may be due to low income healthy MAs opting out of health insurance programs, while MAs with overt disease may be choose to purchase health insurance due to high costs of medications.

This study is subject to several limitations. First, due to the ecologic nature of the study, definitive statements about the causal role of stroke risk factors in midlife ischemic stroke disparities cannot be made. Second, data from BRFSS is self-reported which is subject to recall bias, and this bias may differ between cases and controls. Lastly, the analysis was focused on only on the presence or absence of risk factors and did not take into consideration if a patient was currently receiving treatment and/or if the risk factor was being controlled. Future studies should consider use of continuous risk factor data such as blood pressure, hemoglobin A1c, or low-density lipoprotein levels to examine the influence of treatment and control of risk factors on ethnic stroke disparities in midlife.

SUMMARY

Our findings suggest that diabetes and to a lesser extent hypertension contribute to the persistent ischemic stroke disparities among midlife MAs. Social policies may be particularly relevant to reducing the consequences of stroke in midlife MAs, including greater years of living with disability, greater lifetime lost earnings, and greater healthcare expenditures from stroke.