Mitigating Publication Bias - Views and experiences of principal investigators based in South Africa on interventions to reduce publication bias

Randomised trials are essential for evidence-based healthcare, yet an estimated 25–50% remain unpublished—publication bias that undermines research integrity and decision-making. This project investigates whether COBWEB—an online manuscript-writing aid aligned with CONSORT—can help principal investigators (PIs) of randomised trials reduce non-publication by streamlining write-up and improving reporting quality. Purpose. To explore South African trialists’ views on COBWEB and to map those views to the GRADE Evidence-to-Decision (EtD) criteria most relevant to research integrity (desirable/undesirable effects, values, equity, acceptability, feasibility). Findings will inform the design of an LMIC-adapted version of the tool and a plan to pilot it. Methods. We conducted two semi-structured online focus groups with 12 PIs of completed trials from diverse disciplines and institutions in South Africa. Sessions were recorded, transcribed, anonymised, and thematically analysed in NVivo using a primarily deductive approach guided by the GRADE EtD framework. Analysis examined perceived benefits/harms, alignment with ethical duties to publish, implications for equity (e.g., language and regional bias), acceptability of structured reporting, and feasibility for under-resourced teams. Data management, coding frameworks, and de-identified excerpts will be shared where permissible. Expected outcomes. A nuanced understanding of COBWEB’s anticipated effects (e.g., reduced cognitive load, earlier CONSORT alignment, clearer justifications for pragmatic decisions) and potential downsides (e.g., checklist fatigue, anxiety for early-career authors, plagiarism-flag concerns). Concrete design requirements for an LMIC-adapted COBWEB (plain-language and multilingual prompts, LMIC-relevant exemplars, guidance on paraphrasing/originality, and optional mentorship linkages). An implementation and evaluation plan to pilot COBWEB with African trial teams, including candidate success metrics (completion rates, time-to-submission, CONSORT adherence, and eventual publication). Open materials (topic guide, codebook, and de-identified data where allowed) to support transparency and reuse. Anticipated impact. By improving completeness and usability of trial manuscripts—especially in settings with constrained supervisory capacity—COBWEB could increase submission and publication rates, mitigate regional and language biases, and strengthen the evidence ecosystem. Results will be disseminated via an open report on OSF, a journal manuscript, and presentations to research-integrity and clinical-trials audiences. Limitations include a single-country sample; the subsequent pilot will test generalisability across additional LMIC contexts.