In several occasions, a systematic review of study reports called meta-analysis is necessary in
order to make a conclusion on the benefits and risks of drug therapy. Clarity on how the data are
retrieved and analyzed statistically is as important as the pooled estimate generated by statistics in
the meta-analysis. This article presents the issues needed to be considered when reading a report
on meta-analysis. Readers should pay attention to selection criteria of the reviewed studies and
data retrieval process. For statistical consideration, an estimation model: fixed vs. random effects
is the key factor to differences in confidence interval and point estimate of the summary result. If
heterogeneity across studies did exist, factors affecting such heterogeneity should be further
examined. Through subgroup analysis, the meta-analysis is very useful in explaining variations in
the outcomes of drug therapy as related to characteristics of the study population and drug
regimens.