Fig. 2

One-sample MR estimates of effects of sleep duration on partum-related outcomes in five groups in UK Biobank women. We present MR estimates of a linear effect of increasing duration on the outcome across the length of residual duration (h/day) covered in each group. Further details about identifying the pattern of nonlinear effects are illustrated in Table 1. P-value for Cochran’s Q-statistic testing statistical evidence for between-group heterogeneity. P-value for nonlinearity testing statistical evidence whether the MR estimates are changed as the self-reported sleep duration mean increases. Abbreviation: CI, confidence interval; MR, Mendelian randomization; OR, odds ratio