The review question and inclusion criteria were generally clear, although the criteria for study design were only stated in the abstract. The authors searched a range of sources including grey literature. The review was limited to English language studies, so relevant studies published in other languages could have been overlooked. Publication bias was assessed, but the results were not reported. Other than data extraction, the number of reviewers involved in various stages of the review was not reported, which meant that the risk of reviewer errors or bias affecting the review was uncertain.
Study quality was assessed, although details of criteria and results were not reported in the paper. Relevant details of included studies were reported. Studies were pooled by meta-analysis and heterogeneity was assessed.
It was clear that the included studies were at high risk of bias. Small sample sizes, short observation periods and non-concurrent collection of data made it difficult to be sure that differences in accident risk were caused by CPAP treatment. The authors acknowledged these limitations, but based their conclusions on the large and statistically significant reductions seen in almost all the studies. The weaknesses of the evidence base suggest that the authors' conclusions should be interpreted with some caution.