How were the studies combined?
Pooled effects estimates, adjusted for study, were calculated using the Mantel-Haenszel method with 95% confidence intervals (CIs).
The authors then used weighted least-squares regression to evaluate variation between studies, to model the RR as a function of aspirin dose in milligrams, to test for trends and to graph the predicted dose-response curve. The meta-regression was done on an intention-to-treat basis.
The authors also explored the possibility of a quadratic relation by adding a squared term to the model. To explore the possibility of a dose-response relation more complex than a quadratic relation, the authors fit a cubic spline regression model.
How were differences between studies investigated?
The authors examined study-specific data graphically for apparent heterogeneity across studies and also tested heterogeneity formally with a direct pooling method (see Rothman, in Other Publications of Related Interest no.1).