A systematic review and meta-analysis of randomized controlled trials to reduce burden, stress, and strain in informal stroke caregivers.

Only articles published in English were included and no grey literature was searched, even though about 68% of the included trials were run in non-English-speaking countries, so relevant work may have been missed. The trials were very different from one another, so the pooled figure hides the fact that support meant something different in every study. For burden the heterogeneity was 89.2% and nothing the authors tested explained it. Only results measured straight after the intervention were pooled. Four trials also measured at six months and three at 12 months, but the long-term effect could not be established from this analysis. Fourteen of the 19 trials were rated at some concerns or high risk of bias, and 17 of the 19 could not blind participants or the people delivering the intervention. The trials with positive results tended to be the weaker ones. Five of the 19 trials were pilot or feasibility studies with too few participants to detect a change, which pulls a pooled estimate toward showing no effect. Retention was poor in places, falling to 39% of the intervention group in one trial and 50% of the control group in another, and most studies did not report using intention-to-treat analysis. Most caregivers were women (mean 72%), so these results say nothing about what works for male caregivers.

The authors declared no conflicts of interest. The work was supported by an Australian Government Research Training Program Scholarship, which funded one author.

Do not read the six trials with positive results as proof that these programmes work. The pooled analysis of all 14 burden trials found no significant effect, and when the authors re-ran the analysis by study quality it was the trials at higher risk of bias that showed the more favourable results.