Evaluation of mediators of seasonal variation in fatal strokes: a multicentre registry-based cohort study.

Objectives: How seasonal variations influence the incidence of fatal stroke remains unclear. Here, we examined potential mediators of the association between seasons and fatal strokes.

Methods: Data were prospectively collected (2014-2016) from the Sentinel Stroke National Audit Programme for admissions to four UK hyperacute stroke units. Relationships between variables were assessed by multivariable logistic regression, adjusted for age, sex and intracranial haemorrhage, and presented as odds ratios (OR) with 95 % confidence intervals (CI). The criteria for conducting a mediation analysis were met if the mediator was a continuous variable and was significantly associated both with the predictor and outcome variables.

Results: A total of 3,309 patients (50 % men, mean age=76.7 yr, SD=13.4) were admitted in spring (n = 830), summer (n = 733), autumn (n = 865) and winter (n = 881). There were no seasonal differences in age, sex distribution, co-morbidities (except congestive heart failure in summer) or stroke management. Compared to patients admitted in autumn, those admitted in winter had a greater risk of severe stroke: OR=2.22 (1.50-3.30), fatal stroke: OR=1.48 (1.13-1.93), palliative care: OR=1.85 (1.21-2.81) and pneumonia: OR=1.36 (1.00-1.85). Stroke severity was associated with a greater risk of fatal stroke: adjusted OR=1.14 (1.12-1.15). Mediation analysis revealed that the relationship between winter and fatal stroke was indeed mediated by stroke severity: indirect effect size=0.040 (0.012-0.068).

Conclusions: Our findings support evidence of increased incidence of fatal stroke in winter, which could be explained by greater stroke severity at this time. Further research is necessary to elucidate the underlying reason for greater severity of stroke in winter.