AFRICA
S31
CVJAFRICA • Volume 26, No 2, H3Africa Supplement, March/April 2015
There were many population-/community-based studies
reporting crude prevalence of stroke survivors with prevalence
rates ranging from 15/100 000 population in Ethiopia in 1988,
8,17,34
to 963/100 000 population in Egypt in 2010 (Table 2).
8,17,34,35
The
low prevalence rate recorded in Ethiopia in 1988, included in
the meta-analysis, may have been due to the high fatality rates
from stroke, which have generally been reported in many parts
of Africa.
8,17,34,36
It may also reflect low stroke incidence in rural
Ethiopia at that period, or simply that patients with mild strokes
who had recovered were not detected. Moreover, the Ethiopian
study was a broad door-to-door survey of neurological disorders
in the community, which could imply that active case recognition
of specific stroke cases may be less rigorous.
8
In 1982, in Igbo Ora, Nigeria, stroke had an estimated
crude prevalence of 58 per 100 000 (Table 2). However, the
denominator population was far too small to establish stroke
prevalence accurately.
37
In 2005 to 2006, another study conducted
in Lagos, Nigeria yielded a crude prevalence rate of stroke of
114/100 000 persons.
38
This may suggest at least a doubling of
the stroke prevalence in Nigeria. As reported by several other
studies, males were more affected (males:female
=
1.51) and age
was a strong risk factor with prevalence of nearly 5% for those
in the ninth decade of life.
38
Stroke-prevalence studies in demographic surveillance sites
that provide an accurate denominator have arguably provided
the most accurate measures of stroke burden in recent years,
despite their limitations.
16
The largest study of the prevalence of
disabling hemiplegic stroke in sub-Saharan Africa was done in
1994 in the rural Hai district of Tanzania (Table 2).
16
It provided
an age-standardised (Segi world population) prevalence of
disabling stroke of 154 per 100 000 in men and 114 per 100 000
in women over 15 years of age.
In 2001, a stroke-prevalence study in Agincourt, rural South
Africa, with diagnosis of stroke based on the WHO definition
of stroke, provided an age-standardised (Segi world population)
stroke prevalence of 290 per 100 000 people over the age of 15
years (male: 281 per 100 000, females 315 per 100 000).
16
The rural Tanzanian (1994) and Agincourt studies (2001)
both have the advantage of accurate denominators and careful
assessment of people who screened positive for stroke. However,
the higher prevalence of stroke in Agincourt may be because
Agincourt is further along the epidemiological transition, or
due to the fact that the Tanzanian study included only disabling
hemiplegic stroke.
16
A repeat rural Tanzanian study
8,17,39
showed
an increase in prevalence per 100 000 population from 127
among people aged 15 years and above in 1994 to 2 300 in 2010
among people aged 70 years and above (Table 2).
8,16,17,39
Similarly,
as shown in Table 2, comparison between studies performed in
Egypt in 1993 and 2009 showed an increase in prevalence per
100 000 population from 508 to 560 (mixed) and 410 to 580
(urban).
Supporting this increase, two recent studies in Egypt
produced crude prevalence rates of 922
40
and 963 per 100 000
population (Table 2), with an age-adjusted local prevalence rate
of 699.2/100 000 and an age-adjusted prevalence relative to
the standard world population of 980.9/100 000.
35
There was a
significantly higher prevalence of ischaemic (895/100 000) than
haemorrhagic (68/100 000) stroke. Stroke prevalence was the
same in rural and urban areas but significantly higher in illiterate
(2 413/100 000) than literate participants (3 57/100 000).
35
Overall in Africa, the observed population-based prevalence
rates of stroke survivors were generally high and rising, with a
pooled crude prevalence rate of 387.9/100 000 population (which
may be an under-estimate due to the inclusion of the Ethiopian
study among the 11 studies used for the estimate
8
) and a range
of up to 963 per 100 000 all population.
8,17,22
This prevalence lies
within the range of that recorded in other LMICs (500–1 000 per
100 000) and is in agreement with that found in India (550 per
100 000), but higher than that recorded in Saudi Arabia (180 per
100 000) and Italy (140 per 100 000).
22
The high prevalence of
stroke in the study population may reflect the increased exposure
to risk factors for stroke due to ongoing epidemiological and
demographic transitions.
Mortality
Cause-of-death data from Africa are usually not from standard
vital registration, but are predominantly gathered from verbal
autopsy studies, police reports, sibling histories, and burial and
mortuary reports. With the exception of a few higher-quality
studies, most data on CVD in Africa are from small community
surveys and hospital-based registries.
3,23
Hospital-based data show that NCDs are the leading cause
of death in Africa. In a rural hospital in Nigeria, NCDs
constituted 63% of deaths, with stroke being the leading NCD
cause.
45,46
Similarly, hypertension-related NCD deaths led by
stroke constituted the leading cause of death in a Tanzanian
hospital from 2009 to 2011.
47
Based on verbal autopsies from burial surveillance of 58 010
deaths in Addis Ababa from 2006 to 2009, about 11% of the
deaths were attributed to stroke. The mortality rate increased
with age (15–34 years: 1%; 35–54 years: 7%; 55–74 years: 16%;
>
74 years: 18%) but there were no differences by gender.
48
The Agincourt community-based study in South Africa found
that stroke caused 6% of all deaths between 1992 and 1995.
16
Stroke was the most common cause of death in the age group
55–74 years, and the second most common cause of death in
the age group 35–54 years and the
>
75 years group.
16
The crude
stroke mortality rate was 127 per 100 000 over age 35 years.
16
In
a verbal autopsy study in Tanzania, stroke caused 5.5% of adult
deaths in three regions [Dar-es-Salaam (urban), Hai (prosperous
rural) and Morogoro (impoverished rural)].
16
Age-specific stroke mortality rates in Agincourt and the
three regions of Tanzania mentioned above may be as high as in
England and Wales, and perhaps higher in younger age groups,
but larger studies based on accurate vital registration data are
clearly needed.
16
Such data will produce evidence of any change
in stroke mortality rate particularly as lifestyle, cardiovascular
risk burden, population age structure, relative stroke incidence
and case fatality rates change in Africa.
The GBD dealt with the problem of absent or low-quality
epidemiological data from sub-Saharan Africa by incorporating
covariates (CVD risk factors, national income, differences in
measurement method) and ‘borrowing strength’ from nearby
regions and years of observation in CODEm and DisMod-MR
models; and using standard assumptions about the relationship
between disease-specific incidence, prevalence, case fatality, and
mortality in DisMod-MR models.
3
The ensemble approach
combined different model results developed with different
combinations of covariates and statistical approaches.
2,7,49