Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials | Synapse