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

Disclosures

Stroke. 2019;50(9):2299-2306. 

In This Article

Results

There were 4228 first-ever stroke cases in the studies conducted between 1993 and 2013 (Table 1). HRQoL outcomes were assessed among 1210 (63%) of 1914 survivors at 1 year (3 studies) and 1057 (58%) of 1837 survivors at 5 years (3 studies; Table 1). Sex differences in baseline factors among survivors are provided in the online-only Data Supplement (Supplement III and Tables IIA and IIB in the online-only Data Supplement).

HRQoL at 1-year

Among 3 studies with 1-year HRQoL, the sample for complete-case analysis was 1116 (8% of available cases were dropped because of missing data on covariates). In pooled unadjusted analyses, women had significantly lower median utility scores (median difference, MDunadjusted −0.147; 95% CI, −0.258 to −0.036; Figure 1, top). However, there was significant between-study heterogeneity (I 2=81.2%, Q=10.6; P=0.005). Study-specific unadjusted female:male MDs varied from −0.069 (Oxford), −0.197 (Melbourne), to −0.210 (Perth). These differences were statistically significant, but only in the Melbourne and Perth studies were they clinically significant based on minimum clinically important difference (MID) values of 0.06 for AQoL, 0.08 for SF6D, and 0.12 for EQ5D (Figure 1, top).

Figure 1.

Median difference in 1-y utility scores between sexes after stroke in unadjusted (top) and adjusted (bottom) models. Negative values to the left of 0 indicate worse outcome in women. AQoL indicates Assessment of Quality of Life; EQ5D, European Quality of Life-5 Dimensions; and SF36, Short form-36 questions.

In the fully adjusted model, the sex differences were substantially attenuated (61.4%) after accounting for covariates (pooled MDadjusted −0.067; 95% CI, −0.111 to −0.022; Figure 1, bottom). We found no statistical evidence of heterogeneity in adjusted MD estimates (I 2=17.7%; Q=2.4, P=0.297) across the studies. Among the 3 studies, the adjusted female:male MD again remained clinically significant based on the MID values in the Perth and Melbourne studies (Figure 1). In study-specific multivariable models, the following factors were found to be important covariates: age (2/3 studies), stroke severity (1/3 studies), prestroke functional outcome (1/3 studies), and mood disorders (Melbourne; Table 2). There were no statistical interactions between sex and covariates. No factors of treatment and management of stroke accounted for the sex differences. In a separate model, further adjustment for poststroke functional outcome removed the residual sex difference (Table 2).

The loss to follow-up was large in the Perth study (72.5%). We identified very few sex differences between those assessed and lost to follow-up in baseline characteristics (Table III in the online-only Data Supplement). Sensitivity analyses using the multiple imputation to account for missing data showed limited differences in the estimated pooled effects (MDunadjusted, −0.104; MDadjusted, −0.066; Table IV in the online-only Data Supplement) compared with complete-case analyses (MDunadjusted, −0.147; MDadjusted, −0.067).

HRQoL at 5-year

Among 3 studies with 5-year HRQoL, the sample available for complete-case analysis was 927 (12% of cases were missing data on confounding factors). In unadjusted analyses, women had significantly lower utility scores (pooled MDunadjusted, −0.090; 95% CI, −0.119 to −0.062; Figure 2, top). There was no evidence of statistical heterogeneity (I 2=39.5%; Q=3.3, P=0.191). All the study-specific female:male MD were statistically significant (Figure 2). Clinical significance was only evident in the Melbourne and Auckland studies based on MID values of 0.06 for AQoL, 0.08 for SF6D, and 0.12 for EQ5D.

Figure 2.

Median difference in 5-y utility scores between sexes after stroke in unadjusted (top) and adjusted (bottom) models. Negative values to the left of 0 indicate worse outcome in women. AQoL indicates Assessment of Quality of Life; EQ5D, European Quality of Life-5 Dimensions; and SF36, Short form-36 questions.

After adjustment, the sex differences were only slightly attenuated (5%; pooled MDadjusted, −0.085; 95% CI, −0.135 to −0.034; Figure 2, bottom) without evidence of statistical heterogeneity (I 2=0%; Q=3.3, P=0.571). Study-specific MDs were statistically significant, but only in the Melbourne study was it clinically significant based on the MID values (Figure 2). Contributing factors to the 5-year sex differences were similar to those of 1-year analyses (Table V in the online-only Data Supplement).

Large loss to follow-up was observed in the Auckland study (61.6%), but few sex differences between those assessed and lost to follow-up in baseline characteristics were identified (Tables VI and VII in the online-only Data Supplement). Sensitivity analyses using the multiple imputation showed no difference between the estimated pooled effects and complete-case analyses (Table IV in the online-only Data Supplement).

Subdomain HRQoL

Generally, compared with men, women had worse HRQoL in several subdomains. They included independent living, social relationships, and psychological well-being (Melbourne; Table VIIIA in the online-only Data Supplement); functioning and vitality (Perth; Table VIIIB in the online-only Data Supplement); physical functioning and mental health (Auckland; Table VIIIB in the online-only Data Supplement). Advanced age, more severe strokes, and prestroke dependency were main contributing factors to women's poorer physical health and mental health (Table VIIIA and VIIIB in the online-only Data Supplement).

Comparison Between Women and Men in Stroke and General Populations

Men and women who survived after stroke had worse utility scores than those in the general population across age groups (Tables IXA through XIB in the online-only Data Supplement). Statistically significant results were more often observed at 1-year compared to 5-year.

As compared with the general population, the reduction in HRQoL among stroke survivors was greater in women than in men. However, the magnitude of the sex difference varied among age groups and outcome instruments and was often below clinically relevant thresholds. For the AQoL, compared with the general population, the differences between women and men survivors of stroke (Melbourne) were greater for those aged <65 than for older people (Tables IXA and IXB in the online-only Data Supplement). The sex differences in SF6D utility scores among stroke survivors (Auckland), compared with the general population, were greatest in the youngest age group (<55; Table X in the online-only Data Supplement). By contrast, the sex differences in EQ5D utility scores between stroke survivors (Oxford) and the general population were greater among those aged 65+ (Table XIA and XIB in the online-only Data Supplement) than younger people.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....