Differences in the distribution of stroke subtypes in a UK black stroke population – final results from the South London Ethnicity and Stroke Study

In this prospective study with recruitment from a contiguous geographical area of
South London, UK, we found significant differences in the distribution of stroke subtypes
between black and white stroke patients, with an increase in small vessel disease
stroke, and a reduction in large artery and cardioembolic stroke. These differences
persisted after controlling for traditional cardiovascular risk factors and degree
of social deprivation.

A major finding from the SLESS study is the increase in small vessel disease seen
in the black stroke patients. Small vessel stroke was 2.6 times more common in black
patients, after controlling for risk factors and deprivation. Consistent with this,
leukoaraiosis on brain imaging, which is a radiological marker of small vessel disease,
was increased in black patients after controlling for risk factors and deprivation.
An increase in small vessel disease in black stroke patients has been reported in
some, but not all, previous studies comparing stroke subtypes between the two ethnic
groups 7]–11]. A major strength of our study was the degree of investigation of patients, which
allowed detailed subtyping. This is particularly important for the small vessel disease
subtype for which diagnosis, based on clinical syndrome alone using classifications
such as the Oxfordshire Community Stroke Project Classification 23], can be inaccurate and may differentially affect prevalence rates of the small vessel
disease subtype in different ethnic groups. In the interim SLESS analysis, we demonstrated
that stroke subtyping using the clinical Oxfordshire Community Stroke Project Classification
resulted in a higher frequency of small vessel disease particularly in the white patients,
and an underestimation, compared with results based on the pathophysiological TOAST
classification, of the OR of small vessel disease stroke in black versus white stroke
patients (OR, 1.95 vs. 2.93) 5]. This difference was primarily accounted for by misdiagnosis of white stroke patients
with a clinical lacunar syndrome and carotid stenosis as small vessel disease. Epidemiological
studies which do not perform extracranial imaging in all patients will therefore underestimate
the difference in small vessel disease between the two ethnic groups. In SLESS, 95 %
of all ischemic strokes had this imaging and there was no difference in its use in
the two ethnic groups. The increased small vessel disease we found in the black stroke
patients is consistent with population-based studies looking at subclinical markers
of small vessel disease, namely small deep infarcts and white matter hyperintensities.
These have been shown to be increased in black individuals in both the US and UK 24], 25].

The reason for the increased small vessel disease seen in black populations is uncertain.
Hypertension and diabetes are major risk factors for small vessel disease stroke and,
in the present study, both were more common in the black stroke patients, but the
increase in small vessel disease persisted after controlling for these risk factors
5]. It is also possible that increased severity of hypertension in the black patients
could contribute to the increased risk of small vessel disease. There was a suggestion
that the severity of hypertension was greater in black patients, with a trend for
an increased number of antihypertensive agents before stroke. However, a recent analysis
from the REGARDS study has suggested an alternative explanation – that there are ethnic
differences in the impact of elevated blood pressure on stroke risk with similar levels
leading to increased risk in black, compared with white, individuals 26]. The basis for such differences, and whether it represents biological or genetic
differences or possibly residual confounding, is unknown.

Although large artery stroke was more common in white stroke patients, this was accounted
for by a marked increase in stroke due to extracranial large artery stenosis, which
was 2.6 times more common, on fully adjusted analysis. In contrast, stroke secondary
to intracranial stenosis appeared to be more common in black stroke patients, with
a 70 % increase although this did not reach significance on adjustment for risk factors
and deprivation index. Consistent with this, the presence of any intracranial stenosis
on intracranial angiography was more common in black stroke patients.

Cardioembolic stroke was more common in white stroke patients. This is likely due
to the lower incidence of atrial fibrillation in black individuals, which has been
the subject of recent interest. This observation has been reported in both classic
epidemiological studies 27], and more recently in studies using implantable cardiac devices 28]. The reasons for this difference remain uncertain, but it was not explained by classical
atrial fibrillation risk factors, which were more common in black individuals despite
the reduction seen in atrial fibrillation prevalence 27].

The percentage of other defined causes was similar among black and white patients
(4–5 %). However, age- and sex-adjusted analyses showed that this subtype was associated
with white patients versus black patients and the prevalence of other defined causes
among those younger than 50 years was almost double in white patients versus black
patients. A likely explanation for this is the higher prevalence of classical vascular
risk factors like hypertension and diabetes in the black population at younger ages
than the white population, leading to more classical strokes at a younger age in black
stroke patients versus white stroke patients 4].

The prevalence of diabetes was substantially higher in black compared to white patients
(43 % in Black Caribbean patients, 36 % in Black African patients and 18 % in white
patients). Further, in the multivariable analyses, diabetes was strongly associated
with black versus white patients (no difference between Black Caribbean and Black
African), independently of other demographic or vascular risk factors. Possible explanations
for this difference might include differing rates of control of diabetes and other
risk factors, undetermined environmental risk factors or differences in genetic susceptibility.

We also found differences in risk factor profiles and stroke subtypes between Black
African and Black Caribbean stroke patients. The former are primarily first generation
immigrants from Africa, while the latter are primarily first generation immigrants
from the Caribbean islands. Black African patients were younger, more likely to be
men, hypertensive, had higher deprivation index and were less likely to be smokers.
While small vessel disease stroke was increased to a similar degree in both Black
African and Caribbean individuals, intracranial large vessel disease was more common
in Black African compared with Black Caribbean stroke patients, and this difference
remained after controlling for risk factors. Cardioembolic stroke was less associated
with the Black African group but this was not significant after adjustment for risk
factors.

There are similarities, but some differences, between the pattern of stroke subtypes
seen in this UK population and inner city US populations. While the increase seen
in small vessel disease and intracranial large vessel disease is common across all
studies, the US studies have shown equivalent or higher levels of extracranial large
vessel disease in blacks and similar levels of cardioembolic stroke 10], 11]. Prevalence of hypertension, diabetes and hypercholesterolemia are similar amongst
the two populations although the US populations were slightly older 11].

The mechanisms underlying the difference in stroke subtypes between the ethnic groups
remain uncertain. One possibility is that genetic or other factors modulate the way
in which conventional risk factors, such as hypertension, result in end organ damage.
However, although we controlled for all risk factors and also a measure of social
deprivation, it is also possible that there remain unadjusted differences in risk
factor profiles that contribute to the differences in stroke subtypes. In addition,
rates of suboptimal control of vascular risk factors might differ between groups.
The stroke subtype profile in Black Caribbean patients appeared in some ways intermediate
between that found in Black African and white stroke patients. Whether this reflects
different risk factor exposure, with African patients having been exposed to a “western”
risk factor environment for shorter durations or differing genetic admixture is unknown.
The technique of genetic admixture data using genome wide association data may help
resolve the relative role of genetic versus environmental influences 29].

Our study has a number of strengths. It included a large number of well-phenotyped
black patients with stroke and had a much larger number of black strokes than previous
studies looking at stroke subtypes. All patients were prospectively recruited and
all stroke subtyping was performed by the same individual with review of original
brain imaging. There was a high rate of investigation for causes of stroke, with all
patients having brain imaging and 96 % having either carotid duplex or magnetic resonance
angiography to image the extracranial vessels.

A potential limitation is that SLESS was not truly population based. However, the
catchment areas of the three hospitals covered a contiguous geographical area. Patients
were recruited not only from hospital admissions but also from outpatient stroke services.
The community-based South London Stroke Register is nested within the geographical
catchment area of SLESS, which allowed us to determine the proportion of our black
and white patients with stroke in the community who were admitted to hospital. A detailed
analysis of the first 600 patients found the proportion of both black and white patients
admitted was very similar to that found in the South London Stroke Register population
over the same period, suggesting at most a very small case ascertainment bias in our
study population 5].

Moreover, the determination of self-reported ethnicity using the UK Census 2001 definition
has limitations. Although patients reported of mixed-ethnicity were not included,
there is still the possibility of admixture. In addition, ethnicity does not only
represent genetic background, but also cultural and behavioural differences which
can evolve within individuals over time and between generations 30], all of which are important in the risk of vascular disease.

Black patients were recruited from three acute hospitals within a contiguous geographical
area in South London, whereas the white patients were only recruited in one of these
three hospitals. This possibly might have induced bias in the comparison of black
patients with white patients. However, we performed a sensitivity analysis, comparing
stroke subtypes between black and white patients, in which we only included the black
patients that were recruited from the same hospital as the white patients. The results
of this analysis were similar to the full analysis and did not alter our conclusions.
Therefore, we think that, if there would be any bias due to different inclusion sites,
the effect would only have been small.

Another possible limitation is the slightly longer time period over which black stroke
patients were recruited compared to white stroke patients. This might possibly have
induced bias due to, for example, changing diagnostic techniques and treatment protocols.
We performed a sensitivity analysis in which we repeated all analyses, only including
the black stroke patients recruited from 2003 (75 % of all black patients). The sensitivity
analyses did not show any changes in results compared with the original analysis.