The review objectives and inclusion criteria were clear. The search covered a range of relevant databases and included both published and unpublished trials. Limiting the search by language meant that some relevant trials could have been missed. The authors did not assess trial quality, so the risk of bias in the included trials was unclear, but they were all RCTs.
It was unclear if the patients in the included trials were representative of those for whom the drugs are considered in UK practice. Limited review methods were reported, so reviewer error and bias cannot be ruled out. The network meta-analysis used standard methods and an appropriate method was used to assess inconsistency.
The authors' conclusions reflect the results of the network meta-analysis and are broadly in line with those of previous meta-analyses, but they do not reflect the high probability of liraglutide 1.8mg being the most effective treatment, which was not discussed by the authors. The authors investigated some sources of heterogeneity and inconsistency, but some issues, as they acknowledged, were not considered. This, together with the uncertain quality of the included trials, suggests that the conclusions may not be fully reliable.
This review was funded by two drug manufacturers, and three of the authors were employed by these two companies.