Sex Differences in Long-Term Quality of Life Among Survivors After Stroke in the INSTRUCT

Hoang T. Phan, PhD; Christopher L. Blizzard, PhD; Mathew J. Reeves, PhD; Amanda G. Thrift, PhD; Dominique A. Cadilhac, PhD; Jonathan Sturm, PhD; Emma Heeley, PhD; Petr Otahal, GDipSc; Peter Rothwell, PhD; Craig S. Anderson, PhD; Priya Parmar, PhD; Rita Krishnamurthi, PhD; Suzanne Barker-Collo, PhD; Valery Feigin, PhD; Seana Gall, PhD


Stroke. 2019;50(9):2299-2306. 

In This Article


Women had poorer HRQoL at 1 and 5 years after stroke than men. Although the effect estimates varied by outcome measure (EQ5D, SF36, and AQoL), the direction of the sex difference was relatively consistent across studies. The greatest contributors to the worse HRQoL in women were advanced age, prestroke functional limitations, and stroke severity. These same factors also accounted for women's worse survival and functional outcomes in the long-term following stroke.[5,7] The presence of poststroke mood disorders also accounted for some of the sex difference. In pooled analyses, the aforementioned covariates accounted for much (eg, 54% at 1-year), but not all, of the sex differences in long-term HRQoL. In study-specific analyses, women still had worse HRQoL than men based on the fully adjusted models. These differences were generally below the clinically meaningful threshold.

Age played an important role in the association between sex and HRQoL after stroke, potentially because of frailty and comorbidities.[18] A strategy to address this complexity may be better access to evidence-based therapies, such as poststroke rehabilitation to improve HRQoL in the long-term for the elderly.[19] Also, the worse HRQoL in women compared with men were more likely because of age and prestroke factors (ie, existing dependency) that are likely associated with multimorbidity, rather than due to stroke treatment and management. More effective primary prevention strategies in the elderly (eg, promotion of healthy aging), targeting multiple behavioral and biomedical risk factors[20] should be a priority to reduce multimorbidity in older people, many of whom are women.

Another main determinant of poorer HRQoL in women was the presence of prestroke functional limitations. The poorer prestroke function in women reflects correlations between sex and age at stroke onset,[5] again highlighting the importance of improving health for older people. In our sensitivity analyses, further adjustment for functional outcome removed the residual sex differences. This suggests that if we could improve function and maximize recovery from stroke for both sexes, the sex differences in HRQoL would be attenuated. More effective rehabilitation programs to increase participation, particularly for women after stroke, are needed to achieve this goal.

More severe strokes in women contributed to their worse HRQoL compared with men. An implication of this finding is that management of modifiable factors of stroke severity such as cardiovascular diseases,[21] and cardioembolic strokes[21] could help mitigate poorer outcomes in women. Increasing access to endovascular therapies in both sexes with ischemic stroke may be important as this clearly reduces stroke severity.[22]

The presence of poststroke mood disorder (eg, anxiety and depression), assessed using the Irritability, Depression and Anxiety Scale (Melbourne) or depression subscore of the General Health Questionnaire (Auckland; Supplement I in the online-only Data Supplement), was a contributing factor to the sex differences for the Melbourne study, but not the Auckland study. This may be because of the variation in assessment scale between studies with both instruments having their own limitations.[23,24] Whether there are sex differences in the diagnosis, response to treatment for poststroke depression and its association with HRQoL are uncertain. There are some sex differences in coping strategies[25] that may affect self-reported poststroke depression. The presence of prestroke depression may be also relevant to assess in future work as it is generally more prevalent in women and depressed patients face poorer outcomes after stroke.[26] This suggests that mood disorders should be assessed as part of a clinical diagnostic interview. Social and family participation[27] in the prevention of poststroke depression in the elderly, many of whom are women, may provide opportunities to address the sex differences in HRQoL. Better detection and treatment of poststroke depression, particularly early after stroke, are likely to improve HRQoL for both sexes.

In subdomain analyses, the impacts of stroke were greater in women compared with men in physical functioning, independent living, social relationships, and psychological well-being. These were mainly explained by prestroke factors or stroke severity. The domains that showed no difference between sexes included illness, physical senses (AQoL), or bodily pain, general health, and role emotional (SF36). A potential limitation of these analyses is that the available instruments are generic scales. In a study of sex differences using a stroke-specific instrument, women had poorer HRQoL in some other domains (ie, language, thinking, and energy) and these differences existed after adjustment for confounding factors.[28] The residual sex differences may be further accounted for by unmeasured or poorly measured factors, such as psychosocial functioning or mood disorders.

Authors of previous research have reported sex differences in utility scores, but less often considered whether the differences are clinically meaningful.[23,24] This is important as women in the general population may also have poorer HRQoL than men, particularly in older age groups.[16,29] However, there is a lack of comparison of sex difference in HRQoL between people with stroke and the general population. In our analyses, sex differences in utility scores between stroke and general populations existed in some age groups but varied between studies. The variation may be attributable to the specific utility scores used, with opposing results by age for the EQ5D and the other instruments. This possibly relates to the scale discrepancy whereby some social aspects of HRQoL measured in the SF36 or AQoL (eg, social functioning, family role; Table XII in the online-only Data Supplement) that may contribute to the greater health loss for younger women than younger men, were not captured by the EQ5D. Another possibility is the impact of variability in self-reported HRQoL because of different demographic, economic, cultural, and social factors across populations.[30] Future studies of sex differences in HRQoL should consider these cultural and contextual factors by comparing findings in patients with stroke to population norms, and determining whether the differences are clinically meaningful. We found that stroke caused a substantial HRQoL loss for both men and women, consistent with a previous report. [31] Addressing this impaired HRQoL should be a priority with targets including access to evidence-based care, these being associated with better HRQoL.[32]

Our research has several strengths. We used individual long-term outcome data from high-quality population-based studies across countries. To our knowledge, this is the largest study ever performed to comprehensively examine the contributing factors to the sex difference in HRQoL using a common metric of utility scores. The use of 2-stage meta-analysis overcame the variability in HRQoL measures and covariates between studies. We also compared the net mean differences in HRQoL between people with stroke and the general population to determine how stroke impacts on men and women's health.

We acknowledge several limitations in this study. We performed multiple imputation with available baseline covariates to replace missing data in the studies with large proportions of loss-to-follow-up. However, the possibility of selection bias cannot be eliminated due to the potential differences developing after discharge, and the likelihood that the data are not missing at random. We found that the direction of sex differences in baseline characteristics and clinical factors were similar between people with and without HRQoL assessment (Tables III and VI in the online-only Data Supplement). However, those lost-to-follow-up, compared to those assessed (Table VII in the online-only Data Supplement), were generally more likely to be women, (except for Melbourne) older, dependent before stroke, and had more severe strokes. Thus, they may face poorer HRQoL. The sex differences in HRQoL among all stroke survivors may, therefore, be different to our estimates. The HRQoL assessment was only available in 3 studies at each time point (1- and 5-year). The included cohorts were mostly conducted in high-income countries so the results might not be generalizable to low- and middle-income countries. We identified some other stroke incidence studies that we did not include without sex-specific results or HRQoL assessment.[7] We advocate the inclusion of longer-term patient-reported outcome measures in such studies to assess sex differences in HRQoL, particularly in low- and middle-income countries. We did not have measures of all potentially important covariates. For example, hormonal factors and cognitive status could also impact sex differences in HRQoL[3,33] but we lacked these details across our various studies. Stroke care and poststroke factors (eg, mood disorders) were not measured in all studies and so there is a risk of residual confounding. We found that stroke care did not affect the sex difference in HRQoL, but the investigated studies were conducted a long time ago. Further work should confirm whether the difference in contemporary processes of hospital care could have an impact on sex differences in HRQoL and associated factors.