Stroke Disparities

Introduction

In the United States, the death rate from stroke has been declining since the early 1900s, falling from the third to fourth leading cause of death in 2008, and to the fifth leading cause of death in 2013. However, this decline has stalled or reversed in recent years. With the “graying of America” the number of people having stroke could more than double by 2050 and the total cost of stroke could increase by 129% by 2030. Unfortunately, the burden of stroke does not fall equally across the population, with a higher burden in blacks, residents of the southeastern “Stroke Belt” and rural regions nationwide, in men, and individuals with lower socioeconomic status (SES). While there has been a significant decline in overall stroke mortality, age-adjusted stroke death rates have remained higher in blacks than whites. The National Institutes of Health is required by the Minority Health and Health Disparities and Education Act (2000) to assess the impact and contributors to disparities in (1) minority health and related activities, (2) rural health and related activities, and (3) other activities related to the socioeconomically disadvantaged in the urban setting. In addition to persistent urban-rural geographic disparities in stroke, the “Stroke Belt” in the southeastern US states represents an additional geographic disparity in stroke. While the risk of stroke increases with age, stroke at younger ages is also a significant public health challenge. Of note, most reports either adjust for age or focus solely on younger or older populations. As early as 1992, documentation of higher stroke mortality with lower county-specific SES was emerging. Finally, while the age-adjusted burden of stroke mortality falls heavier on men, this sex-specific burden of stroke differs by age and race. ^,

Racial and Ethnic Disparities in Stroke

One of the first descriptions of the racial disparity in stroke mortality was a 1970 publication providing nonwhite-to-white stroke mortality ratios, where the nonwhite-to-white mortality ratio between 1949 and 1950 was estimated to be 1.63 for men and 1.92 for women. During this time, the nonwhite population was largely African American. Using age-specific stroke mortality rates for 2017 available from the Centers for Disease Control and Prevention (CDC) WONDER, and calculating age-adjusted rates standardized to the population distribution in 1950, the comparable black-to-white stroke mortality ratios for the year 2017 are 1.83 for men and 1.56 for women. Hence, there has been little change in the overall (nonwhite-to-white) disparity in stroke mortality in almost 70 years.

In the more recent period from 1999 to 2017, the stroke mortality rates by specific race-ethnicity strata can also be described using data from the CDC WONDER system ( Fig. 15.1 ). Here stroke mortality (ICD 10: I60–I69) is shown for non-Hispanic whites, blacks, and Asians and for Hispanics of all races. Between 1999 and 2013 there was a 37% (for blacks) to 45% (for Native Americans) decline in stroke mortality across the race-ethnic strata, with a 2% (for Native Americans) to 8% (for Hispanics and Asians) increase between 2013 and 2017 (left panel). Despite the considerable changes over time, with the exception of a small increase for the disparity in blacks, stroke mortality ratios contrasting each minority group with whites has remained relatively stable (right panel). Relative to whites, the excess stroke mortality falls clearly and solely on the black population, with the excess over time ranging from 36% to 48%. The data suggest that the black-to-white stroke mortality ratio increased between 1999 and 2008, decreased from 2008 to 2012, and has been increasing from 2012 to 2017. The data also suggest that the lower mortality for Hispanics relative to whites has been decreasing over time.

The black-to-white disparity in stroke mortality, however, is disproportionately distributed across age groups, with a relative risk 2.5-times greater in blacks than whites for ages 45–54 and 55–64, but then a decreasing disparity with increasing age to the point of no increased risk for blacks relative to whites above the age of 85 years ( Fig. 15.2 ). With the possible exception of Native Americans aged 45–54, there were no other substantial racial/ethnic disparities in stroke mortality.

While the racial/ethnic differences in stroke mortality data are remarkably persistent, there are two possible concerns with these data. First, the numerator data of deaths from stroke and other diseases is from the vital statistics systems (information usually collected by physicians and funeral directors) while the denominator data on population is based on self-reported race/ethnicity from the Census, and it is possible that these two sources report race/ethnicity differently. A report by Arias and colleagues concluded this is of minimal concern for whites and blacks, only modest concern for Hispanic and Asian/Pacific-Islander populations, but may be substantial for Native American populations. There could also be errors introduced by misclassification of causes of death; however, in a national cohort study, death certificates were found to have a sensitivity of 82% and a specificity of 96% relative to physician-adjudicated stroke events. Hence, misclassification could be playing a potentially different role across the palette of diseases, particularly for the Native American population.

The difference in stroke mortality between blacks and whites may be a result of either a higher incidence of stroke in blacks than whites, or a higher case fatality among blacks suffering stroke than whites suffering stroke. There is strong evidence of a higher incidence of stroke in the black population than the white population, with both the Northern Manhattan Study (NOMAS) and the Greater Cincinnati/Northern Kentucky Stroke Study (GCNKSS) showing black-to-white incidence ratios of 2.0 and greater. In addition, the REasons for Geographic And Racial Differences in Stroke (REGARDS) study has shown a pattern of black-to-white incidence ratios strikingly similar to the black-to-white mortality ratios, with risk of incident stroke three times greater in blacks than whites for ages 45–54 and virtually no difference in stroke risk for those aged 85 years and over. The Atherosclerosis Risk in Communities (ARIC) study showed a black-to-white incidence ratio of 2.77 (95% confidence interval [CI], 1.37–5.62) for those aged 45–54 at baseline, and 2.23 (95% CI, 1.66–3.00) for those aged 55–65 at baseline. This finding from ARIC was updated to cover the period from 1990 through 2011, with an overall incidence rate of 2.96/1000 (95% CI, 2.74–3.20) person-years in whites and 6.02/1000 (95% CI, 5.47–6.62) person-years for blacks. In contrast, there is little evidence of a higher case fatality among blacks relative to whites. For example, the GCNKSS showed very similar 30-day, 90-day, and 1-year mortality following stroke for blacks and whites, and this similarity has been remarkably consistent over time. In REGARDS, there was also no evidence of a black-white difference in 30-day case fatality. In the ARIC study, between 1999 and 2011, the 30-day case-fatality rate was 0.11 (95% CI, 0.09–0.14) for whites and 0.09 (95% CI, 0.07–0.12) for blacks. In general, studies have shown blacks to have lower or similar in-hospital or 30-day mortality than whites. ^, As such, it seems likely that the black-to-white disparity in stroke is largely attributable to a racial difference in stroke incidence.

Reflecting the temporal changes in stroke mortality for blacks and whites, the GCNKSS has also reported temporal decreases in stroke incidence rates, significant for whites, but nonsignificant for blacks, over the periods 1993–94 and calendar years 1999 and 2005. While strokes among those aged 20–44 represent a relatively small portion of the total stroke burden, there appears to be a disturbing increase in stroke incidence only in this age strata from 12/100,000 (95% CI, 9–15) in 1993/1994 to 26/100,000 (95% CI, 20–31) in whites ( P = .02), and from 46/100,000 (95% CI, 31–62) in 1993/1994 to 58/100,000 (95% CI, 40–75) in blacks ( P > .05). The ARIC study found a significant decrease in stroke incidence for both blacks and whites over the age of 65 for the period 1987–2011.

Black-white differences in risk factors could be contributing to the racial disparity in stroke incidence through several pathways.

• 1.

  The possibility that a higher prevalence of major stroke risk factors could be contributing to a difference in stroke risk was first assessed using National Health and Nutrition Examination Survey (NHANES) data, showing that approximately one-third of the racial differences in risk for cerebral infarction could be attributed to racial differences in risk factor prevalence. ARIC also showed that adjustment for hypertension and diabetes attenuated the black-to-white stroke incidence ratio 60% from 2.41 (95% CI, 1.85–3.15) to 1.57 (95% CI, 1.18–2.09), and further adjustment for educational status attenuated the disparity an additional 13% to 1.38 (95% CI, 1.01–1.89). Finally, REGARDS showed that adjustment for the Framingham stroke risk factors and SES (as indexed by income and education) was associated with approximately a 50% mediation of the increased risk of stroke; for example, at age 45, the black-to-white increased risk of nearly three times (hazard ratio [HR], 2.90; 95% CI: 1.72–4.89) was decreased to two times the risk (HR, 2.01; 95% CI, 1.16–3.47). Collectively these reports suggest that 30%–60% of the black-white difference in stroke risk may be attributable to the prevalence of “traditional” risk factors. Even if well-controlled, a higher prevalence of risk factors in the black population can contribute to an increase in stroke risk. For example, pharmacologically treated hypertensive REGARDS participants with systolic blood pressure (SBP) treated to below 120 mm Hg were at 1.42-times (95% CI, 0.94–2.15) increased risk if 1 medication was required, 1.60-times (95% CI, 1.06–2.42) if 2 medications were required, and 2.48-times (95% CI, 1.63–2.77) if 3 or more medications were required to achieve control.

• 2.

  Both NHANES and REGARDS have documented that while blacks with hypertension are significantly more likely to know of their condition, and blacks who are aware of their hypertension are significantly more likely to be treated for it, blacks who are treated for hypertension are significantly less likely to have their blood pressure treated to guidelines. For example in REGARDS, after adjustment for SES and other risk factors, the odds of adequate control were 35% less in blacks than whites (odds ratio [OR], 0.65; 95% CI, 0.57–0.72).

• 3.

  Contributions of other or “novel” risk factors that are both more prevalent in the black population and strongly associated with stroke risk may be substantial contributors to disparities in stroke incidence. For example, blacks are at a disadvantage for many measures of SES, and lower SES has been shown to be strongly associated with stroke risk (and more weakly associated with case-fatality). Much work remains to be done to investigate to what extent SES (and psychosocial factors such as stress, depression, discrimination, etc.) attenuates the racial disparity in stroke, and whether it has a direct effect or whether its impact is through increasing the prevalence of traditional risk factors such as hypertension and diabetes. There could also be novel physiologic risk factors, such as C-reactive protein (CRP) and lipoprotein(a), that have higher average levels in the black than the white population ^, and have been shown to be related to stroke risk ^, ; however, much work also remains to assess whether such novel risk factors would attenuate the black-to-white disparities in stroke. Likewise, higher levels of interleukin-6 in blacks have been shown to mediate approximately 28% of the black-white differences in stroke risk.

• 4.

  It is also possible that risk factors could be contributing to the disparities in stroke incidence through a mechanism where the impact of the same level of risk factors could have a larger impact in the black population than the white population. For example, REGARDS data have shown that a 10 mm Hg difference in SBP is associated with an 8% (HR, 1.08; 95% CI, 1.00–1.16) increased stroke risk in the white participants; however, a similar 10 mm Hg difference in SBP in the black cohort was associated with a three-times larger, 24% increase in risk (HR, 1.24; 95% CI, 1.14–1.35); this suggests that blacks are more susceptible to the failure to control elevated SBP.

Stroke disparities for the Hispanic population are more complex, as the mortality data described above show that in 1999, Hispanics were at approximately 30% lower risk of death from stroke than whites, an advantage that declined to approximately 14% lower risk by 2017 (see Fig. 15.1 ). This is in contrast to 2000–2010 incidence data from the Brain Attack Surveillance in Corpus Christi (BASIC) study showing that for ages 45–59 the risk of incident stroke was 1.94 (95% CI, 1.67–2.25) times higher in the Hispanic population than the white, for ages 60–74 1.50 (95% CI, 1.35–1.67) times higher, and no difference in incident stroke risk for ages 75 and above (relative risk [RR], 1.00; 95% CI, 0.90–1.11). NOMAS data from 1993 to 1997 showed higher risk in Hispanics than whites for incident intracranial atherosclerotic strokes (RR, 5.00; 95% CI, 1.69–14.76), extracranial atherosclerotic strokes (RR, 1.71; 95% CI, 0.80–3.63), lacunar strokes (RR, 2.32; 1.48–3.63), cardioembolic strokes (RR, 1.42; 95% CI, 0.97–2.09), cryptogenic stroke (RR, 1.44; 95% CI, 1.09–1.92), and other strokes (RR, 2.27; 0.42–51.91). New data from NOMAS through April 2019 found the highest incidence rate in blacks (13/1000 person-years), followed by Hispanics (10/1000 person-years) and lowest in whites (9/1000 person-years.)

The lower mortality but higher incidence in Hispanics could be a product of several factors. First, as described above, there could be misreporting of race/ethnicity in the vital statistics and census. However, it is also possible that there is confounding of ethnicity and geography. Alternatively, there is the possibility that Hispanic-white ethnic differences in stroke mortality are confounded with Hispanic-white geographic differences in stroke mortality. Specifically, New York County (largely Manhattan, the location of the NOMAS Study) was in the 9th percentile of counties reporting Hispanic stroke mortality but the 0th percentile of counties reporting white stroke mortality. Likewise, Nueces County, Texas (the location of the BASIC Study), is in the 88th percentile of counties reporting Hispanic stroke mortality but the 53rd percentile of counties reporting white stroke morality. Hence, another possible explanation of the paradox of high stroke incidence but low stroke mortality for Hispanics is that the NOMAS study is being conducted in a county with low Hispanic stroke mortality but strikingly lower white stroke mortality, while the BASIC study is being conducted in a county with high Hispanic stroke mortality but average white stroke mortality. With this uncertainty regarding the existence and magnitude of disparities in the Hispanic populations coupled with the remarkable growth in the Hispanic population in the United States, it is critical to continue the efforts of studies such as BASIC and NOMAS and expand both surveillance and longitudinal efforts to understand Hispanic-white differences. There are fewer studies examining disparities in stroke among Hispanics.

Similar to Hispanics, national mortality data suggest stroke mortality is lower for Native Americans than for whites; however, as noted above, the classification of race in mortality data is particularly problematic for Native Americans. The relatively low reported stroke mortality for Native Americans stands in very stark contrast to reported stroke incidence from the Strong Heart Study. With the exception of women born between 1939 and 1947, the stroke incidence for the other age and sex strata was 1.62–3.52 times higher for American Indians in the Strong Heart Study than for whites in the ARIC study. Much of this excess risk in American Indians was mediated by adjustment for risk factors with racial differences becoming nonsignificant. The same report contrasted risk in American Indians to blacks in ARIC, showing similar risk and a relatively small impact of adjustment for risk factors. The interpretation of these observations is complex, since in addition to the racial differences in populations there is also the possibility that differences in incidence rates could be attributed to (1) differences in study methods between the various sources, (2) confounding with geographic disparities in stroke, and (3) temporal changes in stroke risk. Regardless, this remarkable disparity in stroke incidence in Native Americans clearly warrants further study.