The initial search in PubMed resulted in 337 articles. Using Covidence, we excluded 79 articles during the pre-screening of title and abstracts. During the coding process, there were 76 articled excluded (Figure 1), including 2 unable to be retrieved. The final analysis comprised 182 articles, resulting in 102 systematic reviews from the Journal of Clinical Oncology, 56 from Lancet Oncology, 18 from Clinical Care Research, and 1 from Cancer Research.
Across all journals, components of the PRISMA extension varied greatly in terms of reporting (Fig 1). The majority of systematic reviews or meta-analyses incorporated these terms into the title (149/182). Additionally, the number of included studies (146/182), information regarding main outcomes (165/182), and general interpretation of results (174/182) were described in the majority of abstracts in our analysis. Other components were reported less frequently. Information related to the search, such as the databases (89/182) and dates searched (86/182), were reported in only about half of abstracts. O.
In general, frequency of reporting of many components was consistent across journals (Table 2). (It should be noted that our search yielded only one study from Cancer Research, so we omitted this study from journal comparisons). While overall percentages on particular components across journals did vary, the reporting of certain items appeared with greater frequency across the 3 journals. Perhaps the most notable difference was in funding. The Lancet Oncology reported funding sources with the greatest frequency (20/58) which was much higher than what was reported in Journal of Clinical Oncology (5/106) and Clinical Cancer Research (0/17).
Finally, we examined the mean score of each abstract by journal by year. For each abstract, one point was assigned if a particular component had been addressed (or 0 otherwise), and a summary score was calculated for each abstract. Means of overall scores are plotted in Figure 3. As shown, there is not detectable trend across any journal. The publication of the PRISMA extension for abstracts in 2013 did not produce a detectable upward trend in mean scores.