Differential Impact of Risk Factors on Stroke Occurrence among Men vs. Women in West Africa: the SIREN Study

Abstract

Background and Purpose:

The interplay between sex and the dominant risk factors for stroke occurrence in sub-Saharan Africa has not been clearly delineated. We compared the effect sizes of risk factors of stroke by sex among West Africans.

Methods:

SIREN is a case-control study conducted at 15 sites in Ghana and Nigeria. Cases were adults aged >18 years with CT/MRI confirmed stroke and controls were age-and sex-matched stroke-free adults. Comprehensive evaluation for vascular, lifestyle and psychosocial factors was performed using validated tools. We used conditional logistic regression to estimate odds ratios (OR) and reported risk factor specific and composite population attributable risks (PAR) with 95% CIs.

Results:

Of the 2,118 stroke cases, 1,193 (56.3%) were males. The mean ± SD age of males was 58.1±13.2 versus 60.15±14.53 years among females. Shared modifiable risk factors for stroke with adjusted ORs (95% CI) among females versus males respectively were hypertension [29.95(12.49-71.77) vs 16.10(9.19-28.19)], dyslipidemia [2.08(1.42-3.06) vs 1.83(1.29-2.59)], diabetes mellitus [3.18(2.11-4.78) vs 2.19(1.53-3.15)], stress [2.34(1.48-3.67) vs 1.61(1.07-2.43)] and low consumption of green leafy vegetables [2.92(1.89-4.50) vs 2.00(1.33-3.00)]. However, salt intake and income were significantly different between males and females. Six modifiable factors had a combined PAR of 99.1%(98.3-99.6) among females with 9 factors accounting for 97.2%(94.9-98.7) among males. Hemorrhagic stroke was commoner among males (36.0%) than females (27.6%) but stroke was less severe among males than females.

Conclusions:

Overall, risk factors for stroke occurrence are commonly shared by both sexes in West Africa favoring concerted interventions for stroke prevention in the region.

INTRODUCTION

Recent secular trends indicate an unequivocal surge in stroke incidence, prevalence, morbidity and mortality within low-and-middle income countries in sub-Saharan Africa (SSA). ^1–3 Combating this surge will require the identification and targeting of population subsets in SSA, which are susceptible to stroke through particular biological or social characteristics interacting with established stroke risk factors. A key demographic factor contributing to potentially different stroke risk is sex. We have previously reported an association of interleukin–6 (IL-6) rs1800796 and cyclin dependent kinase inhibitor (CDKN2A/CDKN2B) rs2383207 with ischemic stroke in indigenous West African males but not females.⁴ The interaction between sex and the dominant risk factors underpinning stroke among Africans have not been clearly deciphered thus undermining efforts at controlling the burden of stroke.^3,5,6 Studies have identified sex differences in risk factor profile ^7–12, stroke presentation and severity ^11,13,14, choice and response to therapy ^7,15–18

Reasons for these sex-related differences are multifactorial and have been the subject of many studies. ^11,18–21 Understanding what reduces or eliminates sex differences is valuable because it can point to the underlying mechanism for the disparity^8,19,21 and can lead to the identification of modifiable factors and potential interventions. The effect of sex on stroke risk has not been well characterized among Africans and existing studies have not provided conclusive evidence.^20,22–24 Therefore, we sought to compare the effect sizes of vascular risk factors of stroke by sex among West Africans.

METHODS

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Design

The Stroke Investigative Research and Educational Networks (SIREN) study is a multicenter case-control study involving 15 sites in Ghana and Nigeria (Supplementary Table I). The study commenced in August 2014 and the study protocol has been published.²⁵ Briefly, stroke cases included consecutive consenting adults aged >18 years with first clinical stroke within 8 days of current symptom onset or ‘last seen without deficit’ with neuroimaging confirmation with Computerized Tomography (CT) or Magnetic Resonance Imaging (MRI) scan within 10 days of symptom onset (Table II for eligibility criteria).

Controls were consenting stroke-free adults recruited via robust control recruitment from the community, and participating hospitals. Stroke-free status was confirmed using the 8-item questionnaire for verifying stroke-free status (QVSFS) validated in 3 major languages spoken in West Africa (Ashanti, Yoruba and Hausa).²⁶ Controls were matched by age (+/− 5 years), sex and ethnicity to minimize the potential confounding effect of these variables on the relationship between stroke and the main environmental risk factors (Table II, III and IV). Ethical approval was obtained for all study sites and informed consent was obtained from all subjects. To minimize investigation bias, cost of neuroimaging, echocardiography, carotid Doppler, lipid profiling and other investigations were covered for all eligible patients who could not afford these procedures. Ghana has universal health coverage while in Nigeria, patients make out-of-pocket payments for all investigations and treatments.

Stroke diagnosis and phenotyping (Figure I) were based on clinical evaluation and brain neuroimaging (CT or MRI), electrocardiography (ECG), transthoracic echocardiography, and carotid Doppler ultrasound performed according to the standard operating procedures (SOP). Ischemic stroke was sub-typed clinically using the Oxfordshire Community Stroke Project (OCSP) criteria²⁷ and presumed etiological sub-types were defined using the Trial of Org 10172 in Acute Stroke Treatment (TOAST)²⁸ and the Atherosclerosis, Small vessel disease, Cardiac source, and Other (ASCO)²⁹ criteria. Intracerebral hemorrhage was classified etiologically into Structural, Medication-related, Amyloid angiopathy, Systemic/other disease, Hypertension and Undetermined causes (SMASH-U).³⁰ Stroke severity was measured by the modified National Institute of health stroke scale and the Stroke levity scale.³¹

Data collection

We collected basic demographic and lifestyle data including ethnicity and native language of the subjects and their parents, socioeconomic status, cardiovascular risk profile and dietary patterns. We used validated instruments to assess physical activity, stress, depression, cigarette smoking, and alcohol use.³² We collected blood samples early in the morning after overnight fast in cases (post-acute phase when fasting is feasible) and controls for measurement of blood glucose, lipid profile (total cholesterol (TC), Low Density Lipoprotein-cholesterol (LDL-C), High Density Lipoprotein-cholesterol (HDL-C), Triglyceride (TG) and Glycosylated haemoglobin (HbA1c) using a uniform standard operating procedure across all study sites.

Definition of risk factors

• Hypertension: Blood pressure was recorded at baseline and daily for 7 days or until death. A cutoff of ≥140/ 90 mmHg for up to 72 hours after stroke, a history of hypertension, or use of antihypertensive drugs before stroke or >72 hours after stroke were regarded as indicators of hypertension. Adjustments to systolic BP based on reported associations between pre-morbid BP and acute post-stroke BP in the Oxford Vascular Study (OXVASC) were also applied in sensitivity analyses.³³ Definition of hypertension in controls was self-reported history of hypertension or use of antihypertensive drugs or average BP at first clinical encounter ≥140/90mmHg.

• Diabetes mellitus was defined based on history of diabetes mellitus, use of medications for DM, an HBA1c >6.5% or a fasting blood glucose (FBG) levels > 7.0mmol/l at first encounter in controls or measured after the post-acute phase in cases due to the known acute transient elevation of glucose as a stress response after stroke.³⁴

• Dyslipidemia was defined as TC ≥5.2mmol/L, HDL-C ≤1.03mmol/l, TG ≥ 1.7mmol/l or LDL-C ≥ 3.4mmol/l according to NCEP guidelines or use of statin prior to stroke onset.³⁵

• Cardiac disease was defined based on history or current diagnosis of atrial fibrillation, cardiomyopathy, heart failure, ischemic heart disease, rheumatic heart disease. Cases had ECG and echocardiography done to ascertain diagnosis where feasible.

• Obesity: We assessed both waist-to-hip ratio (WHR) and body-mass index. Individuals were classified using the WHO guidelines using cutoffs of 94cm (men) and 80cm (women) for waist circumference, 0.90(men) and 0.85(women) for WHR and 30kg/m² for BMI (Obesity).³⁶

• Individuals were classified as physically active if they were regularly involved in moderate exercise (walking, cycling, or gardening) or strenuous exercise (jogging, football, and vigorous swimming) for 4h or more per week.³²

• Dietary history included regularity of intake of food items such as meat, fish, green leafy vegetables, addition of salt at table, nuts, sugar and other local staple food items. Regular intake was defined as intake on daily, weekly or at least once monthly versus none in a month.

• Alcohol use was categorized into current users (users of any form of alcoholic drinks) or never/former drinker while alcohol intake (or drinking) was categorized as low drinkers (1–2 drinks per day for female and 1–3 drinks per day for male) and high drinker (>2 drinks per day for female and >3 drinks per day for male).

• Smoking status was defined as current smoker (individuals who smoked any tobacco in the past 12 months) or never/former smoker.³²

• For psychosocial risk factors, we adapted measures of psychosocial stress and depression in the INTERSTROKE study.³²

• Family history of cardiovascular risk/diseases was defined based on self-reported history of any of hypertension, diabetes, dyslipidemia, stroke, cardiac disease or obesity in participants’ father, mother, sibling or second degree relative.

Statistical Analysis

We assessed the bivariate association between risk factors and stroke status (case versus control) using McNemar test for paired categorical outcomes with stratification by sex (Male versus Female). Mantel Haenszel Chi-square is used to compare categorical variables. Further analysis to determine the adjusted associations between the risk factors and stroke occurrence for the total sample and stratified by sex were made using conditional logistic regression with adjustment for potential confounders that were not used in the matching except baseline age was included to adjust for residual confounding due to the non-exact age matching. We have also tested for the interaction between sex and each of the covariates. The adjusted models included selected covariates depending on whether they are confirmed confounders in the bivariate analysis and considerations from the literature on stroke risk factors. Additionally, the final adjusted models were assessed for collinearity using variance inflation factor (VIF) and goodness of fit using residual analysis, Pearson chi-square and deviance statistics. We fixed the type I error rate at 5% and no adjustment was made for fitting multiple models to arrive at the final model.

The odds ratio (OR) and 95% Confidence Intervals (CI) in the final models were estimated using conditional likelihood. We calculated the adjusted Population Attributable Risks (PARs) with their respective 95%CI for each exposure variable included in the best-fitted adjusted models and a composite PAR for all risk factors. The PARs were estimated as the proportion of the risk of the stroke in the population that is attributable to the individual risk factors (i.e. the proportion of cases that would not occur in the population if the factor were eliminated).³⁷ The 95%CI for the PAR were obtained using the AF R-package³⁸ where the variance is estimated via the delta method. The advantage of the AF package is it allows for empirical variance estimator to be used in building the 95%CI. Composite PARs for the dominant risk factors for stroke and sex were calculated using the ATTRIBRISK R package with its 95% CI computed via the bootstrap method. All statistical tests of hypotheses are two-sided. Statistical analyses and graphics were produced with SAS 9.4 and R statistical program (version 3.4.2)

RESULTS

Demographic and clinical characteristics

Out of 2,118 stroke cases, males comprised 1,193 (56.3%). The mean ±SD age of males compared with females was 58.09±13.16 versus 60.15±14.53, p≤0.0001. Compared with males, females had lower educational attainment, were less likely to earn more than $100 a month, and used alcohol less as shown in Table 1.

Table 1.

Demographic and variables for Stroke by Sex (Cases vs Controls)

Cases				Controls
Variable	Women	Men	p-value	Women	Men	p-value
	N (%)	N (%)		N (%)	N (%)
Total	925	1193		925	1193
Age <50	209 (22.6)	306 (25.6)	0.1041	238 (25.7)	341 (28.6)	0.1440
No education	252 (27.2)	91 (7.6)	<0.0001	295 (31.9)	117 (9.8)	<0.0001
Income <100$	499 (53.9)	379 (31.7)	<0.0001	548 (59.2)	588 (49.3)	<0.0001
Hypertension	872 (94.3)	1125 (94.3)	0.7645	570 (61.6)	637 (53.4)	<0.0001
Dyslipidaemia	743 (80.3)	915 (76.7)	0.0527	574 (62.1)	723 (60.6)	0.5276
Diabetes	377 (40.7)	419 (35.1)	0.008	132 (14.3)	150 (12.6)	0.2559
Cardiac Disease	120 (12.9)	128 (10.7)	0.1131	54 (5.8)	55 (4.6)	0.2068
WH raised	716 (77.4)	821 (68.8)	<0.0001	632 (68.3)	657 (55.1)	<0.0001
No physical activity	52 (5.6)	45 (3.8)	0.0444	28 (3.0)	21 (1.8)	0.0553
Tobacco use in 12mths	7(0.8)	60(5.0)	<0.0001	1 (0.1)	26 (2.2)	<0.0001
Used alcohol before	174 (18.8)	583 (48.8)	<0.0001	152 (16.4)	514 (43.1)	<0.0001
Stressed	185 (20.0)	247 (20.7)	0.7249	113 (12.2)	163 (13.6)	0.3415
Depressed	70 (7.6)	88 (7.4)	0.8229	59 (6.4)	70 (5.8)	0.5883
Cardiovascular disease in family	366 (39.6)	481 (40.3)	0.6518	273 (29.5)	330 (27.7)	0.3883
Added table salt very often	62 (6.7)	108 (9.0)	0.0471	33 (3.6)	83 (6.9)	0.0006
Green vegetable consumption ≤ 1 per month	313 (33.8)	399 (33.4)	0.7865	219 (23.7)	289 (24.2)	0.3303
Greens weekly	367 (39.6)	454 (38.0)		286 (30.9)	366 (30.6)
Greens daily	168 (18.1)	208 (17.4)		343 (37.1)	406 (34.0)
Confectionary consumption	239 (25.8)	351 (29.4)	0.0113	263 (28.4)	411(34.4)	0.0002
Meat consumption	692 (74.8)	905 (75.8)	0.0606	624 (67.4)	894 (74.9)	<0.0001

Risk factors for stroke by sex

The five shared modifiable risk factors associated with stroke occurrence with adjusted ORs (95% CI) among females and males respectively were hypertension [29.95 (12.49–71.77) versus 16.10 (9.19–28.19)], dyslipidemia [2.08 (1.42–3.06) versus 1.83 (1.29–2.59)], diabetes mellitus [3.18 (2.11–4.78) versus 2.19 (1.53–3.15)], stress in the preceding 2 weeks of stroke [2.34 (1.48–3.67) versus 1.61 (1.07–2.43)] and low consumption of green leafy vegetables [2.92 (1.89–4.50) versus 2.00 (1.33–3.00)], Table 2, Figures II and III. Furthermore, cardiac disease [ 1.82(1.00–3.27 versus 1.75(0.97–3.170] for stroke occurrence did not show a statistically significant difference, while cigarette smoking, high salt, higher income, and meat consumption were independently associated with stroke among males. Compositely, 6 modifiable factors- hypertension, dyslipidemia, diabetes mellitus, cardiac diseases, stress and low consumption of green leafy vegetables were associated with a combined PAR of 99.1% (96.0–99.8) among females. While 9 factors-hypertension, dyslipidemia, diabetes, physical inactivity, tobacco smoking, stress, table added salt, low consumption of green leafy vegetables and regular meat consumption accounted for a PAR of 98.3% (97.1–99.2) among males. Tests for interactions between sex and individual risk factors was significant only for monthly income and table added salt (Table 2).

Table 2:

Odds Ratio and Population Attributable Risk with 95% CI estimates of Stroke Risk factors by Sex

Label	Female		Male		Interaction between sex And risk factor
	Odds Ratio 95% CI	PAR(95%CI)	Odds Ratio 95% CI	PAR(95%CI)	P Value*
Age ≥50	7.93(2.09-29.98)	67.5(56.5-78.4)	3.20(0.98–10.46)	51.1(37.9-64.4)	0.13
Education	1.33(0.85-2.09)	18.6(−9.4-46.7)	1.46(0.76–2.80)	29.9(−9.2-69.2)	0.94
Monthly income >$100 (USD)	0.85(0.59-1.24)	−7.4(−35.8-21.1)	1.87(1.35–2.58)	31.4(18.4-44.4)	0.03
Hypertension	29.95(12.49-71.77)	92.7(89.7-95.7)	16.10(9.19–28.19)	89.7(85.2-94.2)	0.21
Dyslipidemia	2.08(1.42–3.06)	41.6(26.6-56.5)	1.83(1.29–2.59)	34.8(19.7-49.8)	0.56
Diabetes Mellitus	3.18(2.11–4.78)	27.2(21.0-33.2)	2.19(1.53–3.15)	18.1(10.9-25.2)	0.41
Cardiac Disease	1.82(1.00–3.27)	5.1(0.30-9.8)	1.75(0.97–3.17)	4.6(−0.9-10.2)	0.61
Raised Waist-to-hip ratio	1.69(1.07-2.68)	36.1(5.8-66.4)	1.35(0.96–1.89)	19.1(4.7-35.1)	0.38
No Physical Activity	2.02(0.90–4.52)	2.8(0.2-5.4)	2.70(0.77-9.46)	2.2(−0.3-4.7)	0.92
Stress	2.34(1.48-3.67)	14.3(6.3-22.2)	1.62(1.07–2.43)	9.2(2.1-16.3)	0.21
Family history of cardiovascular diseases	1.44(0.97–2.14)	11.9(−4.2-28.1)	1.19(0.84–1.68)	6.9(−6.2-20.1)	0.33
Sprinkled salt	6.06(2.23–16.44)	7.4(5.6-9.3)	1.37(0.78–2.40)	2.9(−0.5-6.4)	0.02
Green leafy vegetables	2.92(1.89–4.50)	20.2(14.9-25.4)	2.00(1.33–3.00)	15.6(8.6-22.7)	0.18
Confectionary sugar/syrups	1.34(0.92–1.95)	7.5(−0.4-15.6)	1.07(0.76–1.50)	2.3(−8.5-13.2)	0.30
Meat	1.75(1.17–2.62)	35.4(16.2-54.6)	1.38(0.89–2.14)	23.5(−11.8-58.8)	0.46
Composite PAR		99.1(98.3-99.6)		97.2(94.9-98.7)

PAR: Population Attributable risk; CI: Confidence Interval; P* value from Conditional logistic regression for the interaction between sex and each risk factor

There were inter-country differences in the effect sizes between the sexes, for instance, Nigerian men had higher incomes, consumed more red meat than Ghanaian males. Ghanaian women had higher effect sizes for hypertension, low consumption of green leafy vegetable, low physical activity, and lower effect of stress than Nigerian females (Table V). Hypertension had a greater effect size in females than in males using different definitions in sensitivity analyses (Table VI).

Stroke types by sex

Ischemic stroke was commoner among females at 72.4% versus 64.0% among males, p<0.001. PACI strokes were more common among males (35.7%) than females (27.9%) while lacunar infarctions were more frequent among females (45.7%) than males (38.3%) using the OCSP classification. Etiologic subtypes of ischemic stroke according to ASCO and TOAST classification by sex is shown in Table 3. Hypertension-related hemorrhagic stroke was commoner among males than females. Strokes were severer among women than men.

Table 3.

Stroke types and Subtypes, Stroke Levity scale and Severity of Stroke by Sex

Parameters	Female (n %) N= 922	Male (n%) N= 1190	P value*
Stroke type			<0.001
Ischaemic	668(72.4)	762(64.0)
Haemorrhagic	254(27.6)	428(36.0)
OCSP classification			0.0264
Total anterior circulation infarction(TACI)	78(14.1)	91(14.2)
Partial anterior circulation infarction(PACI)	197(35.7)	179(27.9)
Posterior circulation infarction (POCI)	63(11.4)	78(12.2
Lacunar infarction(LACI)	214(38.3)	293(45.7)
ASCO classification			0.1899
Atherosclerosis	109(28.0)	100(20.3)
Small vessel disease	200(51.4)	293(59.4)
Cardioembolic	66(16.9)	87(17.6)
Others	14(3.6)	13(2.6)
TOAST Ischaemic Stroke Subtypes			0.299
Large artery-atherosclerosis	211(37.5)	203(30.8)
Cardio-embolism	39(6.9)	63(9.6)
Small-vessel disease	195(34.6)	261(39.7)
Other determined etiology (Dissection, Vasculitis, Others)	1(0.1)	0(0.0)
Undetermined etiology (Two or more causes identified, negative evaluation, Incomplete evaluation)	117(20.8)	131(19.9)
SMASH-U Haemorrhagic Subtypes			0.0013
Structural	15(6.9)	6(1.6)
Medication related	0(0.0)	3(0.8)
Amyloid angiopathy	5(2.3)	1(0.3)
Systemic disease	0(0.0)	1(0.3)
Hypertension	193(88.9)	335(94.4)
Undetermined	4(1.8)	9(2.5)
Stroke Levity Scale			0.0065
Mild	108(12.9)	202(18.9)
Moderate	287(34.4)	340(31.8)
Severe	439(52.6)	526(49.3)
Modified NIHSS			0.0099
1-5	105(13.9)	182(18.6)
6-14	293(38.9)	382(39.1)
15-25	243(32.3)	287(29.4)
>25	112(14.8)	126(12.9)

OCSP: Oxfordshire Community Stroke Project; TOAST: Trial of Org 10172 in Acute Stroke Treatment ASCO: Atherosclerosis, Small vessel disease, Cardiac source, and Other SMASH-U: Structural, Medication-related, Amyloid angiopathy, Systemic/other disease, Hypertension and Undetermined causes. NIHSS: National Institute of Health Stroke Scale P* Mantel -Haenszel Chi Square

DISCUSSION

We have characterized the similarities and differences in the effect sizes of risk factors associated with stroke occurrence by sex among West Africans in the largest cohort of stroke patients in SSA. Six potentially modifiable risk factors- hypertension, dyslipidemia, diabetes mellitus, cardiac diseases, stress and low consumption of green leafy vegetables were independently associated with stroke occurrence among females. Male West Africans had a wider repertoire of factors associated with stroke occurrence than females with effect sizes of shared vascular risk factors being stronger among females. Overall, hypertension was the most dominant risk factor associated with a high OR of 16.1 among males and 30.0 in females, however our sensitivity analyses using four different definitions for hypertension produced estimates that ranged between 5.3 to 17.4 for males and 4.2 to 32.4 among females. Although effect sizes of risk factors overlapped, Tests for interactions between sex and individual risk factors was significant only for monthly income and added table salt.

Traditional/Sociocultural risk factors

Beyond the differences in the effect sizes, the traditional risk factors of hypertension, diabetes and dyslipidaemia were associated with stroke in both males and females consistent with previous findings.^8,39 The effect size of association between cardiac diseases and stroke occurrence reached statistical significance among females but not in males. There are hints of potential differences in lifestyle and dietary practices by sex that may influence stroke occurrence via a nexus of cultural and socio-economic factors. For instance, male stroke patients reported a higher proclivity to adding salt at table and consuming meat more regularly than females.⁴⁰ In addition, males were more likely to consume alcohol and smoke cigarette. We found associations between higher income among males and stroke while low educational attainment and stroke risk was observed among females. It has been observed that relatively affluent, well-educated population may have difficulty identifying and avoiding high-salt foods even if they perceive it is a health issue.^41,42 Higher salt consumption has been associated with stroke occurrence^5,43 however the mechanistic pathways for this association is not clear but has been posited to be either indirectly via effects on blood pressure or via yet-to-be defined alternative mechanisms.

Role of Stress

Stress was independently associated with stroke occurrence in both sexes. However, the effect size and PARs were higher among females than males. Despite the prevalence and potency of this risk factor, little is known about the mechanisms that links stress with stroke.⁴⁴ Interestingly, a recent study has shed light on the role of chronic stress and creation of an atherosclerotic milieu via elaboration of vasculotoxic and pro-atherogenic cytokines.⁴⁵ The resting metabolic activity within the amygdala is significantly associated with the risk of developing cardiovascular disease independently of established cardiovascular risk factors. Furthermore, the link between amygdala activity and cardiovascular disease events is posited to be mediated by arterial inflammation.⁴⁵

Stroke type/subtypes

There were differences in proportions of primary stroke types by sex. The female participants were older and more likely to have ischemic stroke. With advancing age, ischemic stroke is more likely than hemorrhagic stroke and vice versa.⁴⁶ Males significantly had more haemorrhagic strokes causally associated with hypertension than females but no significant differences in etiologic subtypes of ischemic stroke were observed. Intriguingly, although hemorrhagic strokes, which are often severer, were commoner among males than females and the usually less severe lacunar ischemic strokes were commoner among females, we found overall that females had more severe strokes at presentation. The striking differences observed between males and females with regards to primary stroke types, OCSP stroke classification, etiologic subtypes of hemorrhagic strokes and stroke severity are quite significant. Firstly, differential distribution and impact of risk factors may account for the differences in primary stroke types and severity.⁴⁷ There is preliminary evidence^4,48 in support of a genetic basis for the sex disparity in stroke occurrence thus further studies are needed to elucidate the sex-specific genetic mechanisms underlying the pathobiology of stroke and its different subtypes.^11,19 Several studies have shown differing incidences for ischemic versus haemorrhagic stroke by sex.^11,19 Secondly, preventive measures with their associated economic impacts might depend on the specific strokes being targeted for prevention. For instance, given that females tended to have more severe strokes in our study, it might be useful to explore further and identify sex-specific risk associations for severe strokes for evidence-based prevention strategies.

Biological differences

The biological and social explanations for these observations require further investigations. However, the influence of estrogen and testosterone on the endothelium and the vascular system, the role of risk factors unique to women such as the use of oral contraceptives, hormone replacement therapy, and pregnancy, systemic delays in the recognition, and insufficient treatment of conventional stroke risk factors in women have all been considered as probable explanations.¹⁹ Efforts to characterize the possible role of these different factors have been hampered by the paucity of data on sex-differences in age-specific stroke incidence, as outlined in systematic reviews.^11,19 The inherent difficulties in conducting long-term longitudinal follow-up cohort incidence studies and the persistent misperception that stroke is a rarer disease in women may in part be responsible for the paucity of available data.

Strengths and Limitations

This is one of the largest studies to examine the impact of sex on factors associated with stroke risk among west Africans. Previous studies in this population have been limited by sample size and had no control group. A limitation of the case-control design is that causality between putative risk factors and event/outcome outcomes cannot be established. However, because control participants were recruited predominantly from the community, a health volunteer effect cannot be entirely ruled out as influencing the effect sizes observed. We performed individual matching of cases to controls (age, sex and ethnicity not risk factor status) in a 1:1 fashion and used conditional logistic regression analysis to attain unbiased ORs. Due to severity of strokes, responses to questions on lifestyle and dietary behavioral information were obtained from 1621 valid proxies with the remainder from patients themselves. We have previously reported that the associations observed among proxies were in the same direction as for patients with direct assessment.⁵

Conclusion:

Overall, risk factors for stroke occurrence are commonly shared by both sexes in West Africa favoring concerted interventions for stroke prevention in the region.