Stroke in women between 2006 and 2018: Demographic,socioeconomic,and age disparities

Abstract

Background:

Black Americans have a higher prevalence of stroke and stroke-related deaths than any other racial group. Racial disparities in stroke outcomes are even wider among women than men. Conventional studies have cited differences in lifestyle (i.e. smoking, alcohol consumption, etc.) and vascular risk factors between races as the source of these disparities. However, these studies fail to account for the higher prevalence of minoritized populations at the lower end of the socioeconomic distribution.

Objectives:

This study explores differences in stroke risk factors across age and socioeconomic cohorts to determine whether comorbidities can sufficiently explain disparities at all ages and income levels.

Design:

Using the 2006–2018 National Health Interview Survey data, statistical analysis evaluated differences in risk factors among a full sample cohort (aged 18–85 years; n = 131,091) and a “young” subsample cohort (aged 18–59 years; n = 6183) of women.

Methods:

Logistics and unconditional quantile regression models assessed the relationship between stroke and comorbid, demographic, and behavioral characteristics across socioeconomic classes.

Results:

Results suggest that Black women had a 1.415-fold (confidence interval = 1.259, 1.591) higher likelihood of stroke compared with White women after controlling for age, behavior, and comorbidities. Racial disparities were not statistically significant at the higher income ranges for either the full (odds ratio = 1.404, p = 0.3114) or young samples (odds ratio = 1.576, p = 0.7718). However, Blacks had significantly higher odds of stroke in the lower quartiles (lower odds ratio: 1.329, p = 0.0242; lower middle odds ratio: 1.233, p = 0.0486; and upper middle odds ratio: 1.994, p = 0.0005). Disparities were larger among young women (odds ratio = 1.449, confidence interval = 1.211, 1.734).

Conclusion:

While comorbidities were highly associated with stroke prevalence in all socioeconomic cohorts, Blacks only had higher relative odds in the lower income classes. Lack of biological or behavioral explanations for these findings suggests that unobserved or uncontrolled factors such as systemic racism, prejudicial institutions, or differential treatment may contribute to this.

Keywords

health equity healthcare utilization income racial disparities stroke woman

Introduction

Evidence shows that low income, minority race or ethnicity, and female sex increase an individual’s risk of stroke^1
–3—the third leading cause of death for women.⁴ While the longer life span in women is generally used to explain sex disparities in stroke prevalence, this justification fails to rationalize the higher stroke rates among young women and those from historically oppressed populations in the United States.^5
–7 While cardiovascular risk factors are considered to be the major risk factors of stroke, recent evidence suggests that the prevalence of stroke in young women is increasing faster than in men despite lower prevalence of vascular risk factors.^8
–10 To date, the causal predictor of stroke disparities among young Black and Hispanic women has been largely inclusive, leaving those factors that underlie this alarming trend still unknown.

Some evidence suggests that greater neighborhood disadvantage could indirectly contribute to the rise in stroke incidence among young women,^11
–13 but studies show that neighborhood socioeconomic status (SES) was not able to explain the variation in stroke risk within different racial groups in the United States.¹ Further complicating the unexplained race and sex variations in stroke prevalence, these disparities are not consistent over time.¹⁴ Age-adjusted prevalence of first ischemic stroke is more than twice as high in Black as in White individuals and the Black/White prevalence rate ratio is the highest at midlife and decreases thereafter.¹⁵ Race and sex-stratified analyses suggest that these racial disparities are more pronounced and significant in women.¹⁶ However, age-related stroke research has shown inconsistent results. Smaller cohort studies indicate that a strong and persistent disparity in stroke risk exists across the age spectrum for Black women, but not for White women.¹⁷ The primary nonclinical, nonvascular explanation for this phenomenon is SES, which has shown to be a predictor of stroke severity, mortality, and impact in multinational studies.^18,19 There is evidence among high-income and low-income countries showing higher prevalence of stroke and greater stroke mortality among individuals from low compared with high socioeconomic backgrounds.^18,20,21 More specifically, individuals from low socioeconomic backgrounds are more likely to exhibit severe post-stroke deficits potentially due to lack of receipt of evidence-based care known to improve stroke outcomes or unobserved differences that emerge during the post-acute community living period.^21
–24

Recent findings have shown that between the ages of 45 and 65 years,²⁵ only half of the racial disparity in stroke risk is attributable to traditional risk (primarily high blood pressure and high cholesterol) and socioeconomic factors, suggesting a critical need to understand and explore more nontraditional risk factors that could contribute to the differential.² Given that half of the excess stroke risk in Blacks is not attributable to traditional risk or socioeconomic factors, researchers should explore structural, social, and other nontraditional risk factors to explain this disparate prevalence among young, historically oppressed, low-income women.³ This study builds upon previous findings by incorporating social, demographic, clinical, lifestyle, and socioeconomic factors into a large, female-focused study evaluating the association with stroke incidence among socioeconomic and age cohorts. Using the income-to-poverty ratio (IPR) to define socioeconomic classes, inter-quantile regression examines factors that contribute to racial differences in stroke prevalence.²⁶

Materials and methods

Data

Data used in this study were obtained from the 2006 to 2018 National Health Interview Survey (NHIS)—an annual, nationally representative, cross-sectional interview survey of the noninstitutionalized civilian population. Information is collected from a geographically clustered sample of US households concerning the prevalence of illness and disability and services rendered for these conditions.²⁷ NHIS is conducted by the US Census Bureau and the National Center for Health Statistics to monitor the health of the US population by collecting data on a broad range of health topics, demographic information, and socioeconomic characteristics.²⁸ The NHIS employs a complex sampling strategy that partitions dwelling units in the target universe into strata and clusters. Therefore, all analyses of these data utilize parameters that account for sample clustering and stratification as well as weights that estimates are unbiased and standard errors are consistent.^27,28 Additional information can be found at https://www.cdc.gov/nchs/nhis/methods.htm. NHIS content is updated every 15–20 years, most recently in 2019. Content and structural changes made to NHIS in 2019 created incomparability between survey annum, therefore only data from 2006 to 2018 are included in this study.

Sample

The NHIS sample includes women responding (“yes” or “no”) to the question “Have you EVER been told by a doctor or other health professional that you had . . . A stroke?” In total, 211,309 women were identified and 7105 reported being told by a doctor or health professional that they had a stroke, while 204,204 did not report having a stroke. Copies of the NHIS questionnaires used in each year are available on their website.²⁹ Since the NHIS does not include information on respondent age at the time of stroke diagnosis, respondent’s age at the time of survey response was used to identify “young” (age < 60) or “old” (age ⩾ 60). Stroke events were assumed to have occurred prior to the time of survey response. Individuals who did not respond to this question or who stated that they did not know were excluded from the study. Binary indicators for being married, having health insurance, and smoking regularly were created. Age, household size, mean hours of sleep, number of days respondents reported consuming alcohol in the past year, and BMI were included as discrete variables. In addition, respondents self-identified their race as White only, Black/African American only, Aleutian Islander/Alaskan native (AIAN) only, Asian only, or multiracial. Due to sample size constraints, AIAN, Asian, and multiracial groups were combined into a single category. Respondents separately classified their ethnicity as either Hispanic origin or ancestry (Hispanic) or non-Hispanic, which was included as a dichotomous variable as was residence in the southern United States.

NHIS collects household income at the time of the survey. To account for differences in price level across time and geographic regions, the income to poverty ratio (IPR)—a calculation based on the published poverty thresholds in the year prior to interview—was used in this analysis. IPR does not require inflationary or price adjustments; reflects the relative social and economic climate in each year and the ability of households to spend within that climate; and serves as an additional indicator of relative economic well-being contextualizing deeper economic conditions.³⁰ Including IPR in a regression model captures the relationship between the likelihood of stroke and the unconditional mean of IPR. However, estimation at the unconditional mean fails to capture heterogeneity in the likelihood of stroke that may exist across IPR distribution. To account for this potential heterogeneity, we must estimate the relationship at different points along the IPR distribution. There is currently no consensus for the appropriate or meaningful segmentation of the income distribution into classes or groups.³¹ Studies have divided income or socioeconomic classes by relative income, median income, and specific percentiles of the income ladder. After an extensive review of various income divisions and segmentation methods, Stephen Rose³² concluded that over time the portion of the population at various points along the income distribution and the share of income that they controlled were continually shifting. Therefore, any definition too narrow has limited use because it ignores the large inequality between the high and low end of the distribution³³ as well as within individual economic classes.^34,35 Therefore, this study follows Rose’s recommendation as well as the structure proposed by Reeves³⁶ to utilize a four-group segmentation at uniform intervals along the distribution. These four groups—referred to as low, lower middle, upper middle, and upper based on their position along the distribution—partition IPR at 5, 9, and 14. Analysis of these four IPR segments uses the technique developed³⁷ to estimate the unconditional impact of covariates on stroke prevalence along the unconditional IPR distribution,³⁸ allowing us to draw conclusions about whether the difference in the relationships between the covariates and stroke is different at various locations along the distribution.^39,40

Finally, primary and secondary comorbidities associated with stroke were identified from an extensive review of the current literature.^41,42 Those identified conditions that were available in NHIS included hypertension, diabetes, asthma, chronic obstructive pulmonary disease (COPD), heart attack, other heart condition/heart disease, chronic bronchitis, kidney disease, and liver disease. Additional comorbidities, such as atrial fibrillation and arrhythmias, were not available in the NHIS. A count of each respondent’s comorbid conditions was created. While many of these included comorbidities that extend beyond those traditionally associated with stroke, this study utilized a more comprehensive list of primary and secondary conditions due to increasing scientific evidence that multimorbidity was associated with stroke risk and stroke outcomes, particularly among socially disadvantaged groups.^43,44

To ensure that these data could provide robust results, the incidence of stroke among young and old women, 1% and 3%, respectively, was used to conduct a post hoc power analyses. Using an alpha of 5% and 80% power, a sample of at least 1536 would be needed to detect a statistically significant difference if one existed. Given that the sample used herein consists of 211,309 individuals, this analysis was assumed to have sufficient power.

STROBE guidelines were followed in the preparation of this article.

Statistical analyses

Empirical specification

To evaluate the relationships between stroke and the covariates outlined above, a logistic regression model was first run on the age inclusive sample, then repeated in the young cohort (age < 60). This model used a binary indicator of stroke diagnosis as the dependent variable and included age, race/ethnicity, education, IPR, BMI, hours of sleep, smoking, alcohol consumption, having health insurance, residence in the South, marital status, household size, and the total number of comorbid conditions as independent variables. Regression models included population weights to reflect the national noninstitutionalized civilian population. SAS software (version 9.4) was used for all analyses. The use of SAS survey procedures accounted for the complex sampling design of the NHIS and allowed for the accurate calculation of variance estimates using defined clustering and stratification parameters. Odds ratios and 95% confidence intervals (CI) were also calculated for each covariate. First, we estimate this logistic specification to determine the relationship between stroke diagnosis and the unconditional population mean IPR. Second, we estimate the relationship between stroke diagnosis at the four quantiles of the income distribution.

Sensitivity analysis

Since comorbidities are associated with other behavioral and lifestyle characteristics included in the regression model, only the total count of comorbid conditions were included in the regression specification rather than each individual condition. Correlation between independent variables in a regression model causes issues of multicollinearity, which reduces the precision of the estimated coefficients, thereby weakening the statistical power of the regression model and leading to unreliable estimates. However, to allow the differential association between individual comorbidities and stroke prevalence to vary, a sensitivity analysis was conducted using each comorbid condition as an individual term in the regression model, in addition to the other demographic and behavioral covariates.

Power analysis

Since the data were previously collected, a post hoc power analysis was conducted to ensure that a sufficient sample existed in the data to detect a statistically significant difference should one exist. Given a 29% prevalence of young Hispanic women and 25% prevalence of young Black women in the sample cohort, a statistically significant difference between these groups could be detected between these groups with an alpha level of 0.05 and a power of 0.8 if 3866 women were in the sample. Therefore, the study sample was deemed to have sufficient power.

Results

Table 1 provides the demographic, lifestyle, and comorbidity sample means for the full sample (211,309) of women (all ages) as well as the young female subsample (142,965). Most women in the full sample cohort reported living in households of two to three people (standard deviation (SD) = 1.47). A total of 40% of women were married and 37% resided in the southern United States; 15% of the sample was Black and 16% reported being of Hispanic descent. Only 15% reported being regular or occasional smokers and sleeping roughly 7 hours per night (SD = 1.44) and consuming alcohol (SD = 25.30) roughly 5 times per year. The mean BMI of the sample was 27.65 (SD = 6.82), which would be considered overweight by BMI standards⁴⁵ and the mean number of comorbidities is one (SD = 1.37). Only 3% of the full sample (SD = 0.18) and 1% of the young female sample (SD = 0.12) were diagnosed with stroke.

Table 1.

Demographic, lifestyle, and health-related variable mean values by IPR classification.

	All women				Young women
	N	Mean/PCT	SD	Range	N	Mean/PCT	SD
All women
Age	211,309	49.62	18.6	18–85	142,965	39.11	11.78
Household size	211,309	2.41	1.47	1–18	142,965	2.77	1.53
Comorbidities	211,308	1.02	1.37	0–11	142,964	0.66	1.08
Sleep	206,244	7.14	1.44	1–24	139,879	7.05	1.37
Alcohol	151,942	5.01	25.3	0–365	106,818	4.63	21.69
BMI	200,421	27.65	6.82	10.6–94.1	135,833	27.68	7.04
Stroke	211,309	3%			142,965	1%
Married	211,309	41%			142,965	44%
Black	211,309	16%			142,965	17%
Hispanic	211,309	16%			142,965	20%
South	211,309	37%			142,965	37%
Smoker	211,309	16%			142,965	19%
Insurance	208,160	89%			140,868	86%
Lower IPR
Age	51,575	46.52	19.92	18–85	36,654	35.94	11.99
Household size	51,575	2.45	1.69	1–17	36,654	2.86	1.75
Comorbidities	51,575	1.2	1.55	0–11	36,654	0.81	1.26
Sleep	50,409	7.09	1.67	1–24	35,890	7.04	1.62
Alcohol	32,616	3.87	20.23	0–365	24,478	3.98	18.96
BMI	49,661	28.69	7.56	10.6–94.1	35,269	28.77	7.75
Stroke	51,575	5%			36,654	3%
Married	51,575	21%			36,654	24%
Black	51,575	24%			36,654	25%
Hispanic	51,575	25%			36,654	29%
South	51,575	41%			36,654	41%
Smoker	51,575	23%			36,654	27%
Insurance	50,908	85%			36,163	82%
Lower middle IPR
Age	55,685	49.53	18.94	18–85	37,109	38.33	11.44
Household size	55,685	2.43	1.49	1–18	37,109	2.83	1.56
Comorbidities	55,685	1.05	1.38	0–11	37,109	0.64	1.05
Sleep	54,753	7.13	1.44	1–24	36,517	7.02	1.34
Alcohol	40,658	4.72	23.89	0–365	28,086	4.45	20.57
BMI	53,760	28.08	6.89	12.0–85.8	35,842	28.12	7.12
Stroke	55,685	3%			37,109	1%
Married	55,685	40%			37,109	43%
Black	55,685	15%			37,109	17%
Hispanic	55,685	17%			37,109	21%
South	55,685	37%			37,109	38%
Smoker	55,685	17%			37,109	20%
Insurance	55,111	88%			36,725	85%
Upper middle IPR
Age	67,007	48.86	15.93	18–85	48,968	41.46	11.15
Household size	67,007	2.39	1.24	1–18	48,968	2.61	1.3
Comorbidities	67,007	0.85	1.19	0–11	48,968	0.6	0.97
Sleep	65,906	7.14	1.25	1–24	48,173	7.07	1.21
Alcohol	54,895	5.98	28.46	0–365	40,759	5.3	24.14
BMI	64,237	26.93	6.28	11.4–88.6	46,976	26.83	6.4
Stroke	67,007	2%			48,968	1%
Married	67,007	57%			48,968	59%
Black	67,007	10%			48,968	11%
Hispanic	67,007	10%			48,968	11%
South	67,007	34%			48,968	34%
Smoker	67,007	11%			48,968	13%
Insurance	66,242	92%			48,395	91%
Upper IPR
Age	13,424	54.35	18.57	18–85	7757	41.21	11.83
Household size	13,424	2.47	1.41	1–18	7757	2.94	1.48
Comorbidities	13,424	1.07	1.4	0–11	7757	0.62	1.03
Sleep	12,912	7.19	1.39	1–24	7457	7.08	1.27
Alcohol	9398	5.52	28.82	0–365	5611	4.8	23.5
BMI	12,243	26.84	6.34	11.7–94.0	7037	26.84	6.66
Stroke	13,424	3%	0.18		7757	1%	0.1
Married	13,424	49%	0.5		7757	53%	0.5
Black	13,424	13%	0.34		7757	15%	0.35
Hispanic	13,424	14%	0.34		7757	19%	0.39
South	13,424	35%	0.48		7757	34%	0.47
Smoker	13,424	11%	0.31		7757	14%	0.34
Insurance	13,146	90%	0.29		7579	87%	0.33

IPR: income to poverty ratio; PCT: percent; SD: standard deviation; BMI: body mass index; COPD: chronic obstructive pulmonary disease; NHIS: National Health Interview Survey.

Comorbidities included hypertension, diabetes, asthma, COPD, hyperlipidemia, heart attack, other heart condition/heart disease, chronic bronchitis, kidney disease, and liver disease. Data were weighted to reflect nationally representative population. Analysis accounted stratification and clustering of the NHIS.

Baseline comparisons using chi-square tests (Table 2) indicated significant differences in the stroke diagnosis between IPR quantiles (χ² = 998.66, p < 0.0001). Figure 1 illustrates stroke prevalence for the full sample as well as the young and aged >60 samples by income group. In each age group, stroke prevalence was higher in the lower IPR classifications. The lower IPR classifications also showed a disproportionate number of individuals living in the southern United States (χ² = 711.28, p < 0.0001), smoking regularly (χ² = 3463.72, p < 0.0001), and with higher numbers of comorbidities (F = 721.45, p < 0.0001). Comparisons using ANOVA indicated that stroke survivors from lower IPR subgroups had higher mean BMI compared with the higher quantile (F = 784.2, p < 0.0001). The lower IPR groups also have a higher frequency of alcohol consumption (F = 51.50, p < 0.0001).

Table 2.

Distribution across income groups.

	Lower IPR	Lower middle IPR	Upper middle IPR	Upper IPR	Chi-square/F-stat	Prob
Stroke
All ages	42.36%	30.82%	20.25%	6.57%	998.66	<0.0001
Age < 60	49.53%	23.92%	22.43%	4.13%	485.11	<0.0001
Age ⩾ 60	39.19%	33.88%	19.29%	7.64%	550.43	<0.0001
Demographic characteristics
Age (mean)	46.5	49.4	48.7	54.3	745.28	<0.0001
Household size (mean)	2.5	2.4	2.4	2.5	27.25	<0.0001
Comorbidities (mean)	1.2	1.0	0.8	1.1	721.45	<0.0001
Sleep (mean)	7.1	7.1	7.1	7.2	17.47	<0.0001
Alcohol (mean)	3.8	4.6	5.9	5.5	57.80	<0.0001
BMI (mean)	28.7	28.0	26.9	26.8	784.20	<0.0001
Married	14.2%	28.7%	49.3%	7.8%	17100.86	<0.0001
Black	41.8%	29.4%	23.3%	5.6%	4513.39	<0.0001
Hispanic	43.0%	30.3%	21.2%	5.6%	5834.49	<0.0001
South	30.6%	30.3%	32.9%	6.2%	711.28	<0.0001
Smoker	38.9%	31.6%	25.2%	4.3%	3463.72	<0.0001

IPR: income to poverty ratio; BMI: body mass index.

Figure 1.

Stroke prevalence among women age ≤60 and >60 by income group.

The robustness of racial and income subgroups was tested among a cohort of “young” women (age < 60 years) as well as the full cohort allowing the relationships between comorbidities, lifestyle habits, and race to vary along the income distribution using both basic logistic and quantile regression.^26,46,47 Logistic regression models assessed the contribution of these covariates simultaneously using the full sample of women and a subset of young women aged <60 years. Regression results in Table 3 indicate that higher age (OR = 1.053, CI = 1.049, 1.056), larger household size (OR = 1.048, CI = 1.004, 1.094), residing in the south (OR = 1.208, CI = 1.103, 1.322), and having health insurance (OR = 1.388, CI = 1.097, 1.755) are significantly related to the likelihood of stroke in women. Among young women, IPR is negatively related to stroke diagnosis, suggesting a lower prevalence of stroke at higher IPR (OR = 0.941, CI = 0.93, 0.951). While quite small, the coefficient for year is negative, indicating that between 2006 and 2018 there was a slight decrease in the prevalence of stroke diagnosis among women (OR = 0.961, CI = 0.949, 0.974).

Table 3.

Multinomial logistic regression: prevalence of stroke in women, 2006–2018.

	All women					Young women
	Estimate	p value	Odds ratio	95% confidence limits		Estimate	p value	Odds ratio	95% confidence limits
Intercept	71.719	<0.0001				59.027	0.015
Age	0.052	<0.0001	1.053	1.049	1.056	0.055	<0.0001	1.057	1.049	1.065
HH size	0.047	0.028	1.048	1.004	1.094	0.012	0.830	1.012	0.95	1.078
Married	−0.029	0.227	0.944	0.847	1.053	−0.013	0.812	0.973	0.805	1.177
Comorbidities	0.421	<0.0001	1.523	1.481	1.565	0.496	<0.0001	1.642	1.567	1.721
Year	−0.039	<0.0001	0.961	0.949	0.974	−0.033	0.005	0.968	0.947	0.989
Black	0.228	<0.0001	1.415	1.259	1.59	0.230	0.002	1.449	1.211	1.734
Other race	−0.110	0.162	1.009	0.834	1.221	−0.089	0.482	1.053	0.797	1.392
Hispanic	0.020	0.924	1.041	0.886	1.223	0.009	0.929	1.019	0.805	1.29
South	0.094	<0.0001	1.208	1.103	1.322	0.097	0.011	1.215	1.046	1.412
Income/poverty	−0.061	<0.0001	0.941	0.93	0.951	−0.081	<0.0001	0.922	0.905	0.939
Smoker	0.386	<0.0001	1.471	1.314	1.647	0.286	0.000	1.331	1.134	1.561
Sleep	0.056	0.000	1.058	1.026	1.091	−0.010	0.477	0.99	0.932	1.051
Alcohol	−0.001	0.375	0.999	0.997	1.001	−0.002	0.369	0.998	0.995	1.002
BMI	0.000	0.971	1	0.993	1.007	0.001	0.826	1.001	0.991	1.011
Insurance	0.328	0.003	1.388	1.097	1.755	0.235	0.090	1.264	0.937	1.707
N	131091					94,438
AIC	224,737,695					97,148,318
Likelihood ratio	378.58					127.49

HH: household; BMI: body mass index; AIC: Akaike information criterion.

Bold font indicates significance at the 95% confidence level. Dependent variable: Have you ever been diagnosed with stroke? (1 = Yes, 2 = No).

The comparatively large odds ratios for smoking (OR = 1.471, CI = 1.314, 1.647), comorbidities (OR = 1.523, CI = 1.481, 1.565), and race (Black) (OR = 1.415, CI = 1.259, 1.59) suggest that stroke prevalence is 1.47 times higher among smokers compared with nonsmokers, and 1.41 times higher in Blacks compared with Whites. Similarly, each additional comorbidity increases the relative stroke prevalence by 52.3%, relative to the reference group. For women aged <60 years, relative prevalence of stroke diagnosis is higher among Blacks than Whites (OR = 1.449, CI = 1.211, 1.734), smokers than nonsmokers (OR = 1.331, CI = 1.134, 1.561), and individuals with more comorbidities (OR = 1.642, CI = 1.567, 1.721). Other covariates are similar in both magnitude and direction for both the young subsample and full sample.

IPR-specific regression

To test for differential effects along the income distribution, quantile regression allowed the relationships between covariates and stroke prevalence to vary between high and low subgroups along the income (IPR) distribution. The results are listed in Table 4. Age, race, residence, smoking behavior, and comorbidities continue to be significantly associated with stroke among women in the lower, lower middle, and upper middle IPR quantiles. However, in the upper IPR quantile, racial differences are no longer significant. Age (OR = 1.081, CI = 1.032, 1.132) and comorbidities (OR = 1.511, CI = 1.234, 1.85)—logical biological attenuates to stroke—are the only significant associations among women aged <60 years. In the full sample cohort, southern residence (OR = 1.452, CI = 1.026, 2.053) and IPR (OR = 0.664, CI = 0.514, 0.859) are also positively and negatively associated, respectively.

Table 4.

Multinomial logistic regression: prevalence of stroke in women by IPR classification, 2006–2018.

	All women					Young women
	Estimate	p value	Odds ratio	95% confidence limits		Estimate	p value	Odds ratio	95% confidence limits
Lower IPR
Intercept	54.428	0.006				54.824	0.076
Age	0.045	<0.0001	1.046	1.041	1.051	0.055	0.061	1.056	1.045	1.067
HH size	−0.012	0.708	0.988	0.934	1.045	−0.033	0.363	0.967	0.901	1.038
Married	−0.003	0.883	0.994	0.805	1.228	−0.001	0.850	0.998	0.746	1.333
Comorbidities	0.395	<0.0001	1.484	1.423	1.548	0.469	<0.0001	1.599	1.494	1.711
Year	−0.030	0.002	0.97	0.95	0.991	−0.030	0.047	0.97	0.939	1.002
Black	0.154	0.022	1.329	1.128	1.565	0.171	0.072	1.376	1.086	1.743
Other race	−0.024	0.882	1.112	0.821	1.508	−0.022	0.943	1.135	0.749	1.719
Hispanic	−0.094	0.058	0.829	0.66	1.04	−0.091	0.237	0.834	0.588	1.184
South	0.096	0.014	1.211	1.047	1.402	0.140	0.015	1.322	1.059	1.651
Income/poverty	−0.110	0.000	0.896	0.848	0.946	−0.136	0.001	0.873	0.806	0.946
Smoker	0.224	0.005	1.252	1.065	1.47	0.190	0.062	1.209	0.968	1.511
Sleep	0.027	0.293	1.027	0.986	1.071	−0.027	0.258	0.974	0.908	1.044
Alcohol	−0.003	0.212	0.997	0.991	1.002	−0.003	0.411	0.997	0.991	1.004
BMI	0.000	0.936	1	0.991	1.01	0.003	0.649	1.003	0.989	1.016
Insurance	0.331	0.060	1.392	0.952	2.036	0.156	0.372	1.168	0.728	1.875
N	31,180					23,448
AIC	62,672,406					31,706,119
Likelihood ratio	139					65
Lower middle IPR
Intercept	75.458	0.001				31.950	0.679
Age	0.052	<0.0001	1.053	1.047	1.06	0.063	<0.0001	1.065	1.051	1.079
HH Size	0.039	0.392	1.039	0.959	1.127	0.001	0.852	1.001	0.887	1.131
Married	0.024	0.674	1.049	0.872	1.261	0.083	0.338	1.18	0.834	1.668
Comorbidities	0.424	<0.0001	1.527	1.455	1.603	0.474	<0.0001	1.607	1.469	1.758
Year	−0.041	0.000	0.96	0.938	0.982	−0.020	0.541	0.981	0.941	1.022
Black	0.185	0.043	1.233	0.99	1.535	0.180	0.024	1.148	0.796	1.655
Other race	−0.160	0.293	0.874	0.603	1.266	−0.222	0.392	0.768	0.404	1.46
Hispanic	0.069	0.535	1.147	0.843	1.56	0.033	0.909	1.069	0.692	1.651
South	0.098	0.014	1.216	1.037	1.427	0.113	0.110	1.254	0.943	1.668
Income/poverty	−0.021	0.399	0.979	0.925	1.037	−0.008	0.695	0.992	0.894	1.1
Smoker	0.475	<0.0001	1.608	1.309	1.975	0.314	0.044	1.369	1.009	1.859
Sleep	0.040	0.203	1.041	0.981	1.104	−0.058	0.322	0.944	0.828	1.076
Alcohol	0.000	0.764	1	0.996	1.003	0.001	0.798	1.001	0.996	1.006
BMI	−0.008	0.248	0.992	0.98	1.004	−0.011	0.209	0.989	0.97	1.008
Insurance	0.354	0.082	1.425	0.946	2.146	0.334	0.181	1.397	0.819	2.383
Number of observations used	39,026					26,977
AIC	73,127,119					28,014,901
Likelihood ratio	125					34
Upper middle IPR
Intercept	90.844	0.000				102.400	0.016
Age	0.059	<0.0001	1.06	1.052	1.068	0.046	0.072	1.047	1.032	1.063
HH size	0.114	0.017	1.12	1.016	1.235	0.067	0.418	1.069	0.918	1.245
Married	−0.034	0.461	0.934	0.77	1.132	−0.032	0.885	0.938	0.677	1.3
Comorbidities	0.447	<0.0001	1.563	1.469	1.662	0.556	<0.0001	1.743	1.577	1.927
Year	−0.050	<0.0001	0.952	0.928	0.976	−0.055	0.009	0.947	0.91	0.985
Black	0.375	0.001	1.694	1.291	2.222	0.384	0.011	1.831	1.233	2.72
Other race	−0.224	0.141	0.93	0.642	1.349	−0.164	0.445	1.058	0.634	1.766
Hispanic	0.039	0.702	1.082	0.792	1.476	0.000	0.982	0.999	0.646	1.546
South	0.060	0.118	1.127	0.941	1.351	0.005	0.943	1.01	0.754	1.352
Income/poverty	−0.023	0.336	0.977	0.909	1.051	−0.060	0.260	0.941	0.832	1.065
Smoker	0.404	0.001	1.497	1.166	1.923	0.288	0.108	1.334	0.944	1.885
Sleep	0.102	0.000	1.107	1.04	1.18	0.043	0.711	1.044	0.904	1.206
Alcohol	−0.001	0.590	0.999	0.996	1.002	−0.002	0.617	0.998	0.991	1.006
BMI	0.003	0.625	1.003	0.989	1.018	0.014	0.177	1.014	0.993	1.036
Insurance	0.322	0.259	1.38	0.83	2.292	0.302	0.458	1.352	0.696	2.629
N	52,432					38,965
AIC	67,223,183					29,954,591
Likelihood ratio	83					23
Upper IPR
Intercept	113.900	0.057				65.926	0.566
Age	0.053	<0.0001	1.054	1.04	1.068	0.078	0.001	1.081	1.032	1.132
HH size	0.026	0.768	1.026	0.865	1.217	0.042	0.773	1.043	0.785	1.385
Married	−0.039	0.709	0.926	0.618	1.387	−0.176	0.461	0.703	0.276	1.793
Comorbidities	0.361	<0.0001	1.435	1.3	1.585	0.413	<0.0001	1.511	1.234	1.85
Year	−0.058	0.049	0.944	0.891	1	−0.034	0.541	0.966	0.865	1.079
Black	0.203	0.312	1.404	0.852	2.314	0.090	0.772	1.576	0.67	3.703
Other race	−0.066	0.793	1.073	0.521	2.21	0.274	0.435	1.894	0.686	5.23
Hispanic	0.108	0.526	1.241	0.637	2.416	0.320	0.235	1.896	0.66	5.445
South	0.186	0.035	1.452	1.026	2.053	0.207	0.296	1.512	0.696	3.285
Income/poverty	−0.409	0.002	0.664	0.514	0.859	−0.348	0.232	0.706	0.399	1.249
Smoker	0.303	0.236	1.353	0.821	2.232	0.157	0.714	1.17	0.505	2.715
Sleep	0.125	0.081	1.133	0.985	1.303	0.207	0.136	1.23	0.937	1.615
Alcohol	0.001	0.629	1.001	0.996	1.006	−0.049	0.333	0.952	0.863	1.051
BMI	0.008	0.667	1.008	0.97	1.048	−0.024	0.394	0.977	0.925	1.031
Insurance	0.629	0.138	1.876	0.818	4.306	0.433	0.478	1.543	0.466	5.108
N	8453					5048
AIC	17,404,924					5,437,989
Likelihood ratio	23					5

IPR: income to poverty ratio; HH: household; BMI: body mass index; AIC: Akaike information criterion.

Bold font indicates significance at the 95% confidence level. Dependent variable: Have you ever been diagnosed with stroke? (1 = Yes, 2 = No).

Sensitivity analysis

Results from the full and young cohorts are provided in Appendix 1 Table 5. These results list each comorbidity individually rather than using the total number of comorbidities. The results for both the full and young samples are similar to those discussed above for race, age, residence in the South, and insurance status. As expected, diabetes (OR = 1.386, CI = 1.147, 1.675), high cholesterol (OR = 1.264, CI = 1.036, 1.542), and myocardial infarction (OR = 1.836, CI = 1.417, 2.38) are highly associated with stroke prevalence in both samples. The results for the full and young cohorts disaggregated by IPR are provided in Appendix 1 Table 6. The results mirror the previously discussed findings. Significant racial differences appear in the lower and lower middle IPR cohorts, but not in the upper and upper middle groups.

Discussion

This multiyear, nationally representative analysis of women with stroke yielded two primary findings. First, stroke prevalence and disease comorbidity burden were the greatest among women from the lowest IPR quantile. Second, disaggregation of the sample into IPR quantiles shows a strong racial difference in stroke prevalence among the lower IPR quantiles, but not the upper quantiles. While the factors related to the lowest IPR disparities are outside the scope of this analysis, they were not captured by the clinical or administrative data utilized here. They could reflect differences in lifestyle characteristics related to demographic factors, healthcare utilization patterns, quality of healthcare received, or responses to primary and secondary stroke prevention mechanisms. Similarly, findings may represent external factors such as healthcare providers, healthcare institutions, and other unobserved institutional inequities.

These results align with previous findings showing a higher association between the incidence of stroke risk factors and income,⁴⁸ particularly among Blacks and Hispanics.⁴⁹ Findings from the full sample cohort and the young sample cohort regression models showed a higher relative likelihood of stroke among women residing in the southern United States⁵⁰—an area known as the “Stroke Belt” or “Stroke Alley.”⁵¹ This region has been recognized by public health authorities for having an unusually high incidence of stroke and other forms of cardiovascular disease compared with the rest of the country.⁵²

Interestingly, a recent systematic review of SES and stroke prevalence did not find evidence of an association among studies completed in the United States, although differences were observed among studies completed in Sweden and Japan.¹ Those studies completed in the United States only examined the SES associations stratified by sex and/or age, which may have resulted in no observed differences.²¹ The findings reported here suggest studies designed to examine the role SES has in stroke-related outcomes, and those also exploring race must carefully consider race stratified by socioeconomic level. Furthermore, these findings highlight a recent call to carefully examine the complexity of factors related to SES that impact the likelihood of stroke and stroke-related outcomes.⁵³ For example, Lindmark (2021) argues that both mediating and moderating factors affect SES, impacting stroke outcomes. Thus, both types of factors must be considered to understand the heterogeneity and mechanisms that drive sex-related differences. Substantial research will be required to understand these issues further.

Admittedly, this analysis does not capture all possible racial groups or factors related to stroke prevalence. Disparities could result from other measures of healthcare access such as continuity and quality of care and proximity to healthcare providers. However, the findings strongly suggest that studies examining racial disparities must consider issues related to intersectionality, rather than continuing to utilize traditional key outcome variables (sex, race, etc.) which classify individuals into discrete, categorical contributors. Intersectionality considers individuals as a by-product of their life experiences grounded in the sociodemographic backgrounds.⁵⁴ Thus, analyses guided by intersectionality frameworks compute effects from simultaneous, multilevel sociodemographic classifications rather than simply the additive effects of each category. Intersectionality considers the impact of multiple social categories based on their lived experiences rather than simply the timepoint related to the health event; in this case experiencing a stroke.⁵⁵ Understanding these relationships is critically important to measuring disparities in stroke outcomes across multiple dimensions.^56,57 Disparities among low-income Black women are supported by recent findings, suggesting that racial differences in post-stroke disability align with differences in the home/community environment where disability accommodations and support are key to long-term outcomes.^56,58

While neither the mean frequency of alcohol consumption nor the mean duration of sleep was statistically significant in the majority of results presented herein, despite their known association with stroke,^59,60 studies have shown that these outcomes are often subject to social desirability⁶¹ and recall bias⁶² when compared with measured values.

Although the findings are interesting, this study has limitations that should be noted. First, NHIS does not provide age at stroke. Therefore, individuals were classified into old and young age cohorts using their age at the time of the interview. This classification strategy could have potentially misclassified older women who experienced a stroke below the age of 60 years. Second, NHIS neither differentiates between first stroke nor the total number of strokes, the type of stroke, times post-stroke, stroke severity, or the receipt of post-stroke rehabilitation. However, these unobserved factors could potentially confound estimates of functional limitations and other post-stroke outcomes. In addition, NHIS is cross-sectional, which precludes longitudinal assessment of stroke recovery. Third, most of the women in the sample had health insurance, thus the results are not generalizable to an uninsured population. Furthermore, insurance and income are likely not completely independent given that private insurance is often obtained through an employer and qualification for public insurance plans can be income- or employment-based. Therefore, the results should be interpreted with these potential associations in mind. Fourth, all NHIS data is self-reported. Self-reported data is subject to recall, telescoping, social desirability, and selectivity bias. Fifth, as mentioned previously, comorbidities such as atrial fibrillation and arrhythmias were not available in the NHIS, despite their documented association with stroke.^63,64 However, a limited number of studies have shown that the likelihood of atrial fibrillation is higher among older, postmenopausal,⁶⁵ White⁶⁶ women, therefore it is unlikely to have biased racial and ethnic estimates. Finally, the type, quality, and relationship to a primary care caregiver has been shown to impact post-stroke outcomes, disability status, and quality of life.⁶⁷ Unfortunately, caregiver information and type are not included in the NHIS.

Conclusion

This study showed that age, comorbidities, race, IPR, and lifestyle (i.e. smoking, alcohol consumption) have been the most significant factors associated with stroke among women in the past decade. Differences in stroke prevalence related to lifestyle, comorbidity, and age characteristics can be explained by the physiological and biological processes of stroke. However, a higher prevalence of stroke among Black women, compared with White women, after controlling for heterogeneity in comorbidities cannot be easily reconciled.⁶⁸ When these relationships vary, the results show that racial differences at the upper range of the income distribution can be explained by comorbidities and lifestyle differences, while those at the lower end cannot. The disparities among low-income women could be due to unmeasured or unobserved risk factors or result from other structural or systemic disparities between racial groups. While all studies based on survey data reflect a degree of unobserved heterogeneity, it would be naïve to believe that racial inequities do not exist at the lower end of the IPR spectrum that persist beyond behavioral, physiological, or geographical rationalization.⁶⁹

Since these inequities likely result from centuries of slavery, segregation, and contemporary White oppression, which has severely limited the access of many Blacks and people of color to adequate socioeconomic resources, additional research is needed to explore these causal factors. Only through sound, scientific investigation can these inequities be identified and addressed.⁵⁴ Since institutional factors for stroke in women transcend geographic, material, physiological, and financial boundaries, they are not easily addressed. Therefore, studies targeting specific changes in healthcare policy and the healthcare environment are essential to helping all women live healthy, stroke-free lives. Furthermore, the impact of these policies and interventions should address differences by race/ethnicity, age, education, and income to adequately engender the types of changes by public health, healthcare, and social service providers capable of decreasing the prevalence and severity of stroke for those who face a disproportionate burden.

Supplemental Material

sj-docx-1-whe-10.1177_17455057231199061 – Supplemental material for Stroke in women between 2006 and 2018: Demographic, socioeconomic, and age disparities

Supplemental material, sj-docx-1-whe-10.1177_17455057231199061 for Stroke in women between 2006 and 2018: Demographic, socioeconomic, and age disparities by Molly M Jacobs and Charles Ellis in Women’s Health

Supplemental Material

sj-docx-2-whe-10.1177_17455057231199061 – Supplemental material for Stroke in women between 2006 and 2018: Demographic, socioeconomic, and age disparities

Supplemental material, sj-docx-2-whe-10.1177_17455057231199061 for Stroke in women between 2006 and 2018: Demographic, socioeconomic, and age disparities by Molly M Jacobs and Charles Ellis in Women’s Health

Footnotes

Appendix 1

Table 6.

Multinomial logistic regression with conditon covariates: prevalence of stroke in women by socioeconomic class 2006–2018.

	All women					Young women
	Estimate	p value	Odds ratio	95% confidence limits		Estimate	p value	Odds ratio	95% confidence limits
Lower
Intercept	−1.075	0.986				−72.863	0.562
Age	0.021	0.000	1.021	1.009	1.033	0.023	0.114	1.023	0.995	1.052
HH size	0.062	0.281	1.064	0.95	1.191	0.036	0.662	1.037	0.882	1.218
Married	0.059	0.509	1.125	0.793	1.596	0.249	0.074	1.644	0.952	2.838
Year	−0.001	0.962	0.999	0.939	1.062	0.034	0.584	1.035	0.915	1.17
Black	0.159	0.032	1.172	0.854	1.608	0.202	0.420	1.224	0.748	2.002
Other race	0.027	0.491	1.027	0.952	1.107	−0.047	0.516	0.954	0.829	1.099
Hispanic	−0.025	0.258	0.975	0.934	1.019	−0.007	0.183	0.994	0.984	1.003
South	−0.009	0.324	0.991	0.973	1.009	−0.007	0.634	0.993	0.963	1.023
Income/poverty	−0.126	0.819	0.882	0.301	2.586	−0.509	0.462	0.601	0.155	2.333
Smoker	0.181	0.131	1.022	0.76	1.374	0.432	0.042	0.966	0.578	1.614
Sleep	−0.340	0.058	0.607	0.353	1.044	−0.899	0.007	0.255	0.093	0.705
Alcohol	−0.060	0.582	0.887	0.578	1.361	−0.085	0.663	0.843	0.391	1.819
BMI	0.179	0.006	1.429	1.106	1.847	0.262	0.043	1.69	1.017	2.807
Insurance	−0.039	0.468	0.962	0.866	1.068	−0.018	0.837	0.982	0.826	1.168
Hypertension	0.358	0.152	1.431	0.876	2.335	0.318	0.509	1.375	0.534	3.542
Diabetes	0.288	0.035	1.334	1.021	1.743	0.368	0.157	1.445	0.868	2.404
High cholesterol	0.037	0.810	1.037	0.769	1.399	0.196	0.519	1.217	0.67	2.209
Coronary heart disease	0.430	0.010	1.538	1.107	2.136	0.409	0.193	1.505	0.814	2.784
Myocardial infarction	0.688	<0.0001	1.99	1.443	2.745	1.083	<0.0001	2.953	1.729	5.042
Heart condition/disease	0.159	0.278	1.172	0.879	1.563	−0.098	0.727	0.907	0.523	1.571
Asthma	0.370	0.010	1.447	1.095	1.913	0.517	0.041	1.677	1.021	2.754
Chronic bronchitis	0.403	0.017	1.496	1.076	2.08	0.429	0.143	1.535	0.865	2.724
Kidney disease	0.815	<0.0001	2.259	1.586	3.22	0.867	0.004	2.38	1.312	4.316
Liver disease	−0.315	0.239	0.73	0.432	1.234	−0.553	0.246	0.575	0.226	1.465
N	3675					1535
AIC	13,812,982					5,329,142
Likelihood ratio	11					5
Lower middle
Intercept	−140.300	0.043				−241.900	0.207
Age	0.042	<0.0001	1.043	1.027	1.058	0.046	0.025	1.047	1.006	1.09
HH size	0.141	0.073	1.152	0.987	1.343	−0.079	0.689	0.924	0.628	1.36
Married	−0.155	0.057	0.733	0.532	1.01	0.092	0.674	1.202	0.509	2.839
Year	0.067	0.054	1.069	0.999	1.143	0.118	0.216	1.125	0.933	1.357
Black	0.409	0.043	1.505	1.014	2.235	0.098	0.853	0.907	0.323	2.544
Other race	0.064	0.183	1.066	0.97	1.172	−0.226	0.313	0.798	0.515	1.237
Hispanic	−0.020	0.308	0.98	0.943	1.019	−0.021	0.469	0.98	0.926	1.036
South	0.003	0.772	1.003	0.983	1.023	−0.013	0.613	0.987	0.94	1.037
Income/poverty	−0.335	0.519	0.716	0.259	1.98	1.321	0.211	3.746	0.472	29.714
Smoker	−0.060	0.721	0.958	0.632	1.454	−0.692	0.132	0.419	0.12	1.462
Sleep	0.078	0.730	1.1	0.57	2.121	0.515	0.296	1.401	0.35	5.614
Alcohol	−0.066	0.669	0.877	0.48	1.602	−0.154	0.525	0.736	0.286	1.894
BMI	0.114	0.106	1.255	0.953	1.653	0.049	0.789	1.102	0.542	2.239
Insurance	0.003	0.958	1.003	0.899	1.119	−0.016	0.894	0.984	0.776	1.248
Hypertension	−0.283	0.199	0.753	0.489	1.161	−0.652	0.199	0.521	0.193	1.41
Diabetes	0.115	0.475	1.122	0.818	1.538	−0.005	0.991	0.995	0.446	2.219
High cholesterol	0.450	0.011	1.568	1.107	2.221	−0.058	0.890	0.944	0.413	2.155
Coronary heart disease	0.657	0.001	1.93	1.305	2.853	−0.857	0.252	0.425	0.098	1.842
Myocardial infarction	0.444	0.057	1.559	0.987	2.464	2.004	0.002	7.416	2.049	26.835
Heart condition/disease	0.271	0.095	1.311	0.954	1.802	−0.344	0.537	0.709	0.238	2.116
Asthma	0.153	0.457	1.165	0.778	1.746	−0.350	0.488	0.705	0.262	1.893
Chronic bronchitis	0.118	0.627	1.125	0.699	1.809	−0.462	0.382	0.63	0.223	1.775
Kidney disease	0.459	0.100	1.582	0.916	2.734	1.214	0.070	3.368	0.906	12.523
Liver disease	0.883	0.013	2.419	1.207	4.849	1.936	0.001	6.932	2.215	21.694
N	4456					1418
AIC	15,247,995					3,510,065
Likelihood ratio	11					4
Upper middle
Intercept	−32.053	0.741				469.100	0.065
Age	0.049	<0.0001	1.05	1.031	1.07	0.043	0.175	1.044	0.981	1.112
HH size	−0.054	0.593	0.948	0.779	1.154	−0.413	0.080	0.662	0.417	1.05
Married	0.023	0.821	1.047	0.704	1.555	0.167	0.490	1.397	0.54	3.613
Year	0.013	0.786	1.013	0.922	1.114	−0.236	0.062	0.79	0.616	1.012
Black	0.214	0.403	1.238	0.75	2.043	−0.396	0.467	0.673	0.232	1.956
Other race	0.100	0.066	1.105	0.993	1.23	−0.037	0.782	0.964	0.742	1.252
Hispanic	−0.003	0.406	0.997	0.991	1.004	−0.069	0.241	0.933	0.831	1.048
South	−0.002	0.893	0.998	0.963	1.034	0.001	0.987	1.001	0.94	1.065
Income/poverty	−0.860	0.188	0.423	0.117	1.526	−0.697	0.434	0.498	0.087	2.862
Smoker	0.577	0.008	2.338	1.339	4.081	0.808	0.048	2.264	0.867	5.911
Sleep	−0.306	0.273	0.967	0.426	2.192	−0.799	0.148	0.454	0.093	2.204
Alcohol	0.107	0.542	1.239	0.622	2.469	0.099	0.768	1.22	0.325	4.577
BMI	0.022	0.811	1.045	0.727	1.503	0.271	0.205	1.718	0.744	3.971
Insurance	−0.046	0.539	0.955	0.825	1.106	0.048	0.798	1.049	0.727	1.513
Hypertension	0.102	0.725	1.108	0.627	1.956	1.472	0.024	4.356	1.212	15.657
Diabetes	0.453	0.056	1.573	0.988	2.502	0.425	0.371	1.529	0.603	3.88
High cholesterol	0.314	0.133	1.368	0.908	2.061	0.687	0.149	1.987	0.782	5.046
Coronary heart disease	0.341	0.219	1.406	0.817	2.42	0.277	0.742	1.32	0.253	6.873
Myocardial infarction	0.994	0.002	2.701	1.427	5.112	1.853	0.021	6.377	1.32	30.807
Heart condition/disease	0.180	0.430	1.197	0.766	1.871	−0.837	0.241	0.433	0.107	1.754
Asthma	0.160	0.500	1.174	0.737	1.87	0.695	0.090	2.003	0.898	4.47
Chronic bronchitis	0.394	0.165	1.483	0.85	2.589	0.071	0.897	1.074	0.365	3.159
Kidney disease	1.057	0.001	2.879	1.515	5.472	1.359	0.035	3.892	1.104	13.716
Liver disease	−0.254	0.654	0.776	0.255	2.357	0.667	0.526	1.949	0.248	15.326
N	4767					1897
AIC	12,547,435					3,157,831
Likelihood ratio	8					3
Upper
Intercept	−230.600	0.170				−4432.500	0.291
Age	0.034	0.010	1.035	1.008	1.062	2.993	0.371	19.937	0.028	19.98
HH size	−0.114	0.480	0.892	0.649	1.225	7.549	0.363	0.986	0.853	1.141
Married	−0.141	0.421	0.754	0.379	1.501	−2.733	0.039	0.004	0.001	0.749
Year	0.114	0.176	1.12	0.95	1.322	2.243	0.297	9.42	0.137	645.56
Black	0.030	0.960	1.031	0.321	3.31	−0.897	0.802	0.408	0.002	460.22
Other race	0.071	0.460	1.074	0.888	1.299	−1.969	0.578	0.14	0.008	147.91
Hispanic	−0.001	0.795	0.999	0.989	1.009	−0.321	0.498	0.725	0.286	1.842
South	−0.024	0.357	0.977	0.928	1.027	0.233	0.348	1.263	0.775	2.056
Income/poverty	1.581	0.199	4.858	0.435	54.252	−31.064	0.359	0.893	0.854	0.933
Smoker	0.342	0.350	1.2	0.513	2.803	27.482	0.325	1.079	0.726	1.603
Sleep	−0.501	0.354	0.516	0.104	2.555	−25.986	0.316	0.934	0.815	1.07
Alcohol	−0.005	0.985	0.989	0.325	3.01	−22.735	0.336	0.99	0.976	1.004
BMI	0.272	0.093	1.722	0.912	3.25	−11.667	0.273	0.999	0.974	1.025
Insurance	−0.321	0.195	0.725	0.446	1.18	−17.533	0.404	0.711	0.263	1.918
Hypertension	0.276	0.511	1.317	0.578	3.004	8.311	0.026	1.259	0.68	2.332
Diabetes	1.041	0.002	2.833	1.477	5.434	5.070	0.025	1.449	0.974	2.156
High cholesterol	−0.186	0.608	0.83	0.407	1.694	−0.544	0.929	1.251	0.829	1.89
Coronary heart disease	0.788	0.132	2.199	0.787	6.146	−52.751	0.419	1.098	0.64	1.884
Myocardial infarction	−0.010	0.986	0.99	0.316	3.1	46.178	0.419	3.671	2.204	6.117
Heart condition/disease	0.419	0.233	1.521	0.763	3.031	11.192	0.463	0.762	0.482	1.202
Asthma	0.279	0.517	1.322	0.567	3.079	−18.116	0.377	1.392	0.951	2.037
Chronic bronchitis	−0.253	0.737	0.776	0.177	3.409	2.768	0.098	15.918	0.595	425.626
Kidney disease	−0.230	0.709	0.795	0.237	2.663	21.265	0.599	6.095	0.001	9.781
Liver disease	−0.927	0.104	0.396	0.129	1.21	−8.941	0.150	0.012	0.021	25.646
N	1013					253
AIC	3,748,356					430,988
Likelihood ratio	3					2

HH: household; BMI: body mass index; AIC: Akaike information criterion.

Bold font indicates significance at the 95% confidence level. Dependent variable: Have you ever been diagnosed with stroke? (1 = Yes, 2 = No).

Acknowledgements

None.

Declarations

ORCID iD

Molly M Jacobs

Supplemental material

Supplemental material for this article is available online.

References

Kim

Twardzik

Judd

, et al. Neighborhood socioeconomic status and stroke incidence: a systematic review. Neurology 2021; 96(19): 897–907.

Howard

Kleindorfer

Judd

, et al. Disparities in stroke incidence contributing to disparities in stroke mortality. Ann Neurol 2011; 69(4): 619–627.

Kleindorfer

Khoury

Moomaw

, et al. Stroke incidence is decreasing in whites but not in blacks: a population-based estimate of temporal trends in stroke incidence from the Greater Cincinnati/Northern Kentucky Stroke Study. Stroke 2010; 41(7): 1326–1331.

Bushnell

CD.

Stroke and the female brain. Nat Clin Pract Neurol 2008; 4(1): 22–33.

Viticchi

Falsetti

Plutino

, et al. Sex influence in ischemic stroke severity and outcome among metabolically unhealthy overweight patients. J Neurol Sci 2020; 416: 116955.

Akhtar

Kate

Kamran

, et al. Sex-specific differences in short-term and long-term outcomes in acute stroke patients from Qatar. Eur Neurol 2020; 83(2): 154–161.

Bushnell

McCullough

Awad

, et al. Guidelines for the prevention of stroke in women: a statement for the healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2014; 45(5): 1545–1588.

Béjot

Delpont

Giroud

Rising stroke incidence in young adults: more epidemiological evidence, more questions to be answered. J Am Heart Assoc 2016; 11(5): e003661.

Ekker

Verhoeven

Vaartjes

, et al. Stroke incidence in young adults according to age, subtype, sex, and time trends. Neurology 2019; 92(21): e2444–e2454.

10.

Leppert

Burke

, et al. Young women had more strokes than young men in a large, United States claims sample. Stroke 2020; 51(11): 3352–3355.

11.

Hunt

Deot

Whitman

Stroke mortality rates vary in local communities in a metropolitan area: racial and spatial disparities and correlates. Stroke 2014; 45(7): 2059–2065.

12.

Gaglioti

Rivers

Ringel

, et al. Individual and neighborhood influences on the relationship between waist circumference and coronary heart disease in the REasons for Geographic and Racial Differences in Stroke Study. Prev Chronic Dis 2022; 19: E20.

13.

Lang

I-M

Antonakos

Judd

, et al. A longitudinal examination of objective neighborhood walkability, body mass index, and waist circumference: the REasons for Geographic And Racial Differences in Stroke study. Int J Behav Nutr Phys Act 2022; 19(1): 17.

14.

Tian

Liu

, et al. Long-term trends in stroke management and burden among low-income women in a rural area from China (1992-2019): a prospective population-based study. Front Neurol 2021; 12: 720962.

15.

Koton

Schneider

ALC

Rosamond

, et al. Stroke incidence and mortality trends in US communities, 1987 to 2011. JAMA 2014; 312(3): 259–268.

16.

Virani

Alonso

Aparicio

, et al. Heart disease and stroke statistics—2021 update: a report from the American Heart Association. Circulation 2021; 143(8): e254–e743.

17.

Kamalesh

Shen

Tierney

WM.

Stroke mortality and race: does access to care influence outcomes?

Am J Med Sci 2007; 333(6): 327–332.

18.

Addo

Ayerbe

Mohan

, et al. Socioeconomic status and stroke: an updated review. Stroke 2012; 43(4): 1186–1191.

19.

Wang

Zhai

Wei

, et al. Socioeconomic status predicts the risk of stroke death: a systematic review and meta-analysis. Prev Med Rep 2020; 19: 101124.

20.

Avan

Digaleh

Di Napoli

, et al. Socioeconomic status and stroke incidence, prevalence, mortality, and worldwide burden: an ecological analysis from the Global Burden of Disease Study 2017. BMC Med 2019; 17(1): 1–30.

21.

Avan

Digaleh

Di Napoli

, et al. Socioeconomic status and stroke incidence, prevalence, mortality and worldwide burden: an ecological analysis from the Global Burden of Disease Study 2017 (2219). Am Acad Neurol 2020; 94: 2219.

22.

Marshall

Wang

Crichton

, et al. The effects of socioeconomic status on stroke risk and outcomes. Lancet Neurol 2015; 14(12): 1206–1218.

23.

Schulz

Kannan

Dvonch

, et al. Social and physical environments and disparities in risk for cardiovascular disease: the healthy environments partnership conceptual model. Environ Health Perspect 2005; 113(12): 1817–1825.

24.

Tyson

Turner

Discharge and follow-up for people with stroke: what happens and why. Clin Rehabil 2000; 14(4): 381–392.

25.

GBD 2019 Stroke Collaborators. Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Neurol 2021; 20(10): 795–820.

26.

Gardener

Sacco

Rundek

, et al. Race and ethnic disparities in stroke incidence in the Northern Manhattan Study. Stroke 2020; 51(4): 1064–1069.

27.

Lynn

Rivera

Griffin

, et al. IPUMS health surveys: medical expenditure panel survey, version 1.1 (dataset). Minneapolis, MN: IPUMS, 2019. https://meps.ipums.org/

28.

L. Blewett J. Rivera

. IPUMS health surveys: National Health Interview Survey (Version 6.2), 2016, https://nhis.ipums.org/

29.

National Center for Health Statistics, National Health Interview Survey (NHIS). Public-use data files and documentation, https://www.cdc.gov/nchs/data_access/ftp_data.htm#:~:text=Public%2Duse%20data%20files%20are,format%20appropriate%20for%20their%20analyses

30.

Semega

Kollar

Shrider

, et al. Income and poverty in the United States: 2019 (P60-270). Washington, DC: US Census Bureau, 2020.

31.

Eisenhauer

An economic definition of the middle class. For Soc Econ 2008; 37: 103–113.

32.

Rose

The growing size and incomes of the upper middle class. Urban Inst 2016; 21: 10.

33.

Wenger

Zaber

MA.

Who is middle class? Santa Monica, CA: RAND Corporation, 2021.

34.

Pickett

Wilkinson

RG.

Income inequality and health: a causal review. Soc Sci Med 2015; 128: 316–326.

35.

Tóth

Wachs

Di Clemente

, et al. Inequality is rising where social network segregation interacts with urban topology. Nat Commun 2021; 12(1): 1143.

36.

Reeves

Gourtsoyannis

Basu

, et al. Financing universal health coverage—effects of alternative tax structures on public health systems: cross-national modelling in 89 low-income and middle-income countries. Lancet 2015; 386(9990): 274–280.

37.

Powell

Did the economic stimulus payments of 2008 reduce labor supply? Evidence from quantile panel data estimation. Santa Monica, CA: RAND Corporation, 2014.

38.

Lanehart

Rodriguez de Gil

Kim

, et al. Propensity score analysis and assessment of propensity score approaches using SAS procedures. In: Proceedings of the SAS global forum 2012 conference, Orlando, FL, 22–25 April 2012, pp. 22–25. Cary, NC: SAS Institute.

39.

Guo

Fraser

Chen

Propensity score analysis: recent debate and discussion. J Soc Social Work Res 2020; 11(3): 463–482.

40.

Desai

Rothman

Bateman

, et al. A propensity-score-based fine stratification approach for confounding adjustment when exposure is infrequent. Epidemiology 2017; 28(2): 249–257, https://pubmed.ncbi.nlm.nih.gov/27922533

41.

Cipolla

Liebeskind

Chan

SL.

The importance of comorbidities in ischemic stroke: impact of hypertension on the cerebral circulation. J Cereb Blood Flow Metab 2018; 38(12): 2129–2149.

42.

Benjamin

Virani

Callaway

, et al. Heart disease and stroke statistics-2018 update: a report from the American Heart Association. Circulation 2018; 137(12): e67–e492.

43.

Dugravot

Fayosse

Dumurgier

, et al. Social inequalities in multimorbidity, frailty, disability, and transitions to mortality: a 24-year follow-up of the Whitehall II cohort study. Lancet Public Health 2020; 5(1): e42–e50.

44.

Gallacher

Jani

Hanlon

, et al. Multimorbidity in Stroke. Stroke 2019; 50(7): 1919–1926.

45.

Fryar

Carroll

Ogden

CL.

Prevalence of obesity among children and adolescents: United States, trends 1963–1965 through 2009–2010. Hyattsville, MD: National Center for Health Statistics, 2012.

46.

Buie

JNJ

Zhao

Burns

, et al. Racial disparities in stroke recovery persistence in the post-acute stroke recovery phase: evidence from the health and retirement study. Ethn Dis 2020; 30(2): 339–348.

47.

Magwood

Ellis

Nichols

, et al. Barriers and facilitators of stroke recovery: perspectives from African Americans with stroke, caregivers and healthcare professionals. J Stroke Cerebrovasc Dis 2019; 28(9): 2506–2516.

48.

Nguyen

Khan

Valero-Elizondo

, et al. Association of income status with stroke in non-elderly adults in the United States, 2012-2018. Curr Probl Cardiol 2023; 48(8): 101235, https://linkinghub.elsevier.com/retrieve/pii/S0146280622001323

49.

Trimble

Morgenstern

LB.

Stroke in minorities. Neurol Clin 2008; 26(4): 1177–1190, https://linkinghub.elsevier.com/retrieve/pii/S0733861908000698

50.

Glymour

Kosheleva

Boden-Albala

Birth and adult residence in the Stroke Belt independently predict stroke mortality. Neurology 2009; 73(22): 1858–1865, https://www.neurology.org/lookup/doi/10.1212/WNL.0b013e3181c47cad

51.

Howard

VJ.

Twenty years of progress toward understanding the Stroke Belt. Stroke 2020; 51(3): 742–750, https://www.ahajournals.org/doi/10.1161/STROKEAHA.119.024155

52.

Bathala

A visit to the stroke belt of the United States. J Neurosci Rural Pract 2012; 3(3): 426–428, http://www.thieme-connect.de/DOI/DOI?10.4103/0976-3147.102653

53.

Lindmark

Neighborhood socioeconomic differences in stroke risk? Highlighting the need for further research. Neurology 2021; 96: 879–880.

54.

Hankivsky

Reid

Cormier

, et al. Exploring the promises of intersectionality for advancing women’s health research. Int J Equity Health 2010; 9: 5.

55.

Kapilashrami

Hankivsky

Intersectionality and why it matters to global health. Lancet 2018; 391(10140): 2589–2591.

56.

Wickramasinghe

Feminist research methodology: making meanings of meaning-making. London: Routledge, 2009.

57.

Beckman

LJ.

Training in feminist research methodology: doing research on the margins. Women Ther 2014; 37(1–2): 164–177.

58.

Ironstone-Catterall

McDonough

Robertson

, et al. Feminist research methodology and women’s health: a review of the literature. Toronto, ON, Canada: National Network on Environments and Women’s Health, 1998.

59.

Reynolds

Lewis

Nolen

JDL

, et al. Alcohol consumption and risk of stroke. JAMA 2003; 289(5): 579–588, https://jamanetwork.com/journals/jama/fullarticle/195880

60.

Zhang

Chang

Zhang

, et al. Correlation analysis of sleep quality and youth ischemic stroke. Behav Neurol 2014; 2014: 246841–246846, http://www.hindawi.com/journals/bn/2014/246841/

61.

Davis

Thake

Vilhena

Social desirability biases in self-reported alcohol consumption and harms. Addict Behav 2010; 35(4): 302–311, https://linkinghub.elsevier.com/retrieve/pii/S0306460309003074

62.

Lauderdale

Knutson

Yan

, et al. Self-reported and measured sleep duration. Epidemiology 2008; 19(6): 838–845, https://journals.lww.com/00001648-200811000-00015

63.

Lip

GYH

Freedman

De Caterina

, et al. Stroke prevention in atrial fibrillation: past, present and future. Thromb Haemost 2017; 117(07): 1230–1239.

64.

Healey

Connolly

Gold

, et al. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med 2012; 366(2): 120–129.

65.

Azarbal

Stefanick

Salmoirago Blotcher-E , et al. Obesity, physical activity, and their interaction in incident atrial fibrillation in postmenopausal women. J Am Heart Assoc 2014; 3(4): e001127, https://www.ahajournals.org/doi/10.1161/JAHA.114.001127

66.

Perez

Wang

Larson

, et al. Risk factors for atrial fibrillation and their population burden in postmenopausal women: the Women’s Health Initiative Observational Study. Heart 2013; 99(16): 1173–1178, http://heart.bmj.com/content/99/16/1173.abstract

67.

Pindus

Mullis

Lim

, et al. Stroke survivors’ and informal caregivers’ experiences of primary care and community healthcare services–a systematic review and meta-ethnography. PLoS ONE 2018; 13(2): e0192533.

68.

Mahesh

PKB

Gunathunga

Jayasinghe

, et al. Factors influencing pre-stroke and post-stroke quality of life among stroke survivors in a lower middle-income country. Neurol Sci 2018; 39(2): 287–295.

69.

Dunbar

Khavjou

Bakas

, et al. Projected costs of informal caregiving for cardiovascular disease: 2015 to 2035: a policy statement from the American Heart Association. Circulation 2018; 137(19): e558–e577.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

2.93 MB

0.03 MB