Mediation analyses
| Associations of physical activity ≥4 hour/week | Estimate (95% CI) | P value |
| Admission NIHSS 0–4 | ||
| Average causal mediation effect | 0.10 (0.05 to 0.16) | <0.001 |
| Average direct effect | 0.08 (0.02 to 0.15) | 0.004 |
| Average total effect | 0.18 (0.10 to 0.27) | <0.001 |
| Proportion mediated by haematoma volume | 0.57 (0.31 to 0.87) | <0.001 |
| One-week mRS 0–3 | ||
| Average causal mediation effect | 0.10 (0.05 to 0.15) | <0.001 |
| Average direct effect | 0.07 (0.01 to 0.15) | 0.032 |
| Average total effect | 0.17 (0.09 to 0.25) | <0.001 |
| Proportion mediated by haematoma volume | 0.58 (0.29 to 0.96) | <0.001 |
| 90-day survival | ||
| Average causal mediation effect | 0.05 (0.02 to 0.09) | <0.001 |
| Average direct effect | 0.14 (0.06 to 0.22) | <0.001 |
| Average total effect | 0.19 (0.10 to 0.28) | <0.001 |
| Proportion mediated by haematoma volume | 0.28 (0.14 to 0.48) | <0.001 |
Associations of prestroke physical activity with NIHSS 0–4, mRS 0–3 and 90-day survival were calculated in bootstrapped binary logistic regression analyses with ICH volume as mediator. Proportion (%) denotes the magnitude of the mediation effect (average causal mediation effect divided by the total effect). The models were adjusted for age, sex, birthland, education, income, activities of daily living, living situation, smoking, alcohol, hypertension, hyperlipidaemia, prior stroke or TIA, atrial fibrillation, diabetes mellitus, cardiac disease, heart failure, dementia, cancer, lipid-lowering drugs, antiplatelet drugs, antihypertensive drugs, non-vitamin k oral anticoagulants, warfarin and International Normalised Ratio (INR). Results were obtained from imputed data.
ICH, intracerebral haemorrhage; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; TIA, transient ischaemic attack.