“…2 They raised concerns regarding the methodology of our study, arguing that the use of adjusted effect estimates derived from the multivariable Cox proportional hazards model and/or propensity-score matching might lead to a biased and potentially misleading result and that a meta-analysis of the unadjusted effect estimates should have been undertaken along with that of the adjusted ones. We agree with Dr Ackerman et al 1 that using adjusted effect estimates for a meta-analysis of observational studies is imperfect, when the variables adjusted for by the included studies vary; however, using unadjusted effect estimates for meta-analysis would be neither less biased nor be justifiable. The limitation of our meta-analyses is intrinsic to observational studies.…”