method | β1 (se) | β2 (se) | β3 (se) | β4 (se) | β5 (se) |
Ordinary linear regression on the pooled dataset (n = 768 subjects) | 0.067157 (0.057144) | −0.319836 (0.077137) | 0.102682 (0.009834) | 0.771217 (0.144281) | 0.664185 (0.090638) |
Meta-analysis (k = 10 studies), using the actual variance-covariance matrix | 0.077151 (0.066722) | −0.377597 (0.1288864) | 0.108197 (0.012404) | 0.818003 (0.156274) | 0.657785 (0.118143) |
Meta-analysis (k = 10 studies), using the actual variance estimates and assuming zero correlation | 0.089251 (0.071367) | −0.374107 (0.131523) | 0.110964 (0.012804) | 0.819255 (0.156134) | 0.657269 (0.114043) |
Meta-analysis (k = 10 studies), using the actual variance estimates and the alternative model of Riley | 0.024612 (0.055657) | −0.206025 (0.107372) | 0.095521 (0.007975) | 0.880839 (0.155768) | 0.718012 (0.141181) |
Meta-analysis (k = 10 studies), using the actual variance estimates and assuming a plausible correlation | 0.071282 (0.065038) | −0.346298 (0.124092) | 0.107940 (0.011789) | 0.826325 (0.160682) | 0.667961 (0.121064) |
Meta-analysis (k = 10 studies), using a random coefficient model and Individual Patients Data (IPD) | 0.078416 (0.064406) | −0.365146 (0.103916) | 0.105337 (0.010688) | 0.801249 (0.142996) | 0.631803 (0.098856) |