Skip to main content

Table 2 Frequency of reasons for judgements of high risk of selective reporting bias

From: Rethinking the assessment of risk of bias due to selective reporting: a cross-sectional study

Reason

Number (%a) of 1055 studies

Concerns about outcome non-reporting bias

819 (78)

 Not all of the study’s pre-specified outcomes have been reported

387 (37)

 One or more outcomes of interest in the review are partially reported so that they cannot be entered in a meta-analysis

364 (35)

 The study report fails to include results for a key outcome that would be expected to have been reported for such a study

188 (18)

Concerns about the documents available for assessment (e.g. no protocol was available or the only available report is a conference abstract)

59 (6)

Concerns about reporting of only a subset of measurements, analysis methods or subsets of the data that were pre-specified (e.g. data were reported for only some of the pre-specified time points)

58 (6)

Concerns about post-hoc reporting of outcomes, measurements, analysis methods or subsets of the data (e.g. one or more reported outcomes were not pre-specified in a protocol or trial registry)

56 (5)

Concerns about how outcome data were analysed (e.g. a continuous/ordinal outcome was dichotomised or adjusted effect estimates were not reported)

28 (3)

Concerns about discrepant reporting (e.g. outcome data differed across multiple reports for a particular study)

9 (1)

Other concerns (e.g. only adverse events occurring in at least 5 % of participants were reported, trialists emphasised statistically significant results even though these were less relevant/secondary outcomes)

31 (3)

Concerns that are not relevant to the selective reporting domain (e.g. not all randomised participants were analysed, baseline data were not reported, blinding of participants was unclear)

73 (7)

Unclear reason (e.g. stated that “All pre-specified outcomes were reported” or no reason stated)

69 (7)

  1. aPercentages do not sum to 100 as some trials had more than one reason for a high-risk judgement. Review authors stated one reason in the majority of cases (817/1055, 77 %), two reasons for 209/1055 (20 %) studies and three reasons for 29/1055 (3 %) studies