Barriers and enablers for publishing results of randomized trials: a cross-sectional survey of principal investigators in sub-Saharan Africa

To investigate the views and experiences of principal investigators (PIs) based in sub-Saharan Africa (SSA) regarding publication bias of randomised controlled trials (RCTs) and strategies to mitigate it. Cross-sectional online survey, informed by previous reviews and prior surveys on publication bias. The survey consisted of closed-ended items with Likert scales and optional free text. We emailed invitations to African-based PIs that we identified through the World Health Organization International Clinical Trials Registry Platform. 6056 invitations were sent, 610 opened the survey, 433 were eligible to partake, and 409 consented and proceeded with the survey. Awareness and perceived magnitude of publication bias in SSA, experiences of timely publication, non-publication, delayed publication and their perceived effectiveness of mitigating strategies. Awareness that all RCTs are to be published was 62.7% (227/362). More than 90% agreed that PIs are responsible for ensuring that their RCTs are published (295/323), and 84% (272/323) within 24 months of completion. Publication bias in SSA scored a median of 71/100 (IQR: 50 – 84) for geographical region and 66 (IQR = 40 –79) for the respondent’s specific field of research. Of the respondents, 63.1% (219/347) published the findings of at least one RCT within 24 months, and 23.6% (82/347) were unable to publish at least one complete RCT. Notably, 74% (246/332) indicated not having had prior training in writing RCT-specific manuscripts. Of the 82 that were unable to publish, their reported reasons for non-publication included 30.5% (25/82) journal rejection, 29.3% (24/82) time constraints, 14.6% (12/82) non-significant or 13.4% (11/82) non-positive results, 13.4% (11/82) methodological concerns and 12.2% (10/82) lack of confidence in writing. 41.2% (143/347) of respondents reported delays greater than 24 months, 47.9% (69/347) attributed time constraint, 38.9% (56/347) time-consuming analysis, and 36.8% (53/347) journal rejection. Of the 332 participants that answered questions on mitigating strategies, the following were suggested to be very effective: 56.6%(188/332) dedicated writing time, 37.1% (123/332) guidance on writing a manuscript and 35.8% (119/332) peer-review prior to submission. Furthermore, 52.7% (175/332) rated the CONSORT checklist as effective/very effective, and 37.4% (124/332) noted artificial intelligence tools somewhat effective. SSA PIs note that publication bias is a problem, which is a result of barriers such as time constraints, analysis difficulties, journal rejection, and limited writing support. Enablers include mentorship, dedicated writing time, and structured reporting tools.