Stem cell therapy of cardiac disease: an update

PS analyses, the authors found that the results betweenthese two techniques were not materially different inmost of these studies. Even if the results were differentbetween the two approaches, one could not necessarilyassume that the PS analysis yielded the ‘true’ answer,at least not without specifying underlying assumptions.In general, the approach of using traditional multi-variate regression adjustment is preferable if the sam-ple size is sufciently large and the outcome of interestis not rare. Only if the outcome is rare relative to thenumber of confounders and the number of studysubjects in the smaller exposure group is sufcientlylarge to warrant multivariable PS estimation, then thisstatistical technique has a legitimate role to potentiallyreduce bias and expand the possibilities in observa-tional outcomes research [6]. Only then, the use of PScan be regarded a substantial help, not just hype.