Systematic reviews (SRs) are very labor intensive, depending on the topic and scope of literature search. We have previously reported the general efficiency of the process [number of authors required, ratio of papers included to those retrieved in the databases searched (SEff)] in a sample of 195 reviews on a wide variety of human health topics as a mean number of authors = 5 and SEff<3%. Using recently published SRs, we aimed to compare relative efficiency with these metrics across obesity-related topics to determine whether there are systematic differences between topics that make some areas of literature more burdensome to synthesize than others.
We searched PubMed for papers published in 2016 with a title containing 'systematic review'. Of these, we selected papers that contained any of the following terms in the title: 'obesity', 'diabetes', 'cardiovascular', or 'nutrition' to code as categories of the papers. We extracted the following data from each source as available (full text or abstract): Category of review topic, N authors, and the numbers of each literature selection stage reported in the PRISMA diagram (if present): N screened (total found in database searching) and N studies included in the review. Between the categories we compared mean # of authors and SEff using ANOVA.
Data was available to analyze 519 SRs for at least one outcome. Some with multiple categories of focus were coded as a separate group (N=86). SRs with nutrition focus was the smallest group (N=37). Overall, M authors=5.6 and M SEff = 2.8%. Significant group differences were observed in mean N authors/review, F(4,518)=3.69,p=.006. Using a Bonferroni post hoc test, obesity and diabetes SRs were significantly lower in mean # of authors than CVD SRs. For SEff (N=297), there were no group differences, p=.657; range=0.03-44.7%.
SRs of these topics do not greatly differ in efficiency or people required compared to other medical topics. CVD reviews may involve relatively more team members.