The review question was clearly stated and inclusion criteria were defined for participants, intervention, control treatment and outcomes. However, suicide acts were not defined and criteria for study design were broad. Limiting the search to one database plus references raised the potential for publication bias and the omission of other relevant studies. It was not clear if any attempts were made to minimise language bias. Methods were used to minimise reviewer errors and bias in the extraction of data, but it was not clear whether similar steps were taken in study selection.
Study validity was not assessed, so results from these studies and any synthesis may not be reliable. No information was provided about participants. Two of the three RCTs evaluated lamotrigine and all three chart reviews included both valproate/divalproex and carbamazepine. Pooling data from studies of different designs (RCTs and chart reviews) without separately analysing RCTs was not appropriate. In the chart reviews, there may have been differences between treatment groups that influenced the results.
The limited search, absence of a validity assessment and pooling of data from diverse studies mean that the authors’ conclusions may not be reliable.