Transparency without reproducibility? Open science practices in systematic reviews from 2014 to 2024: a cohort study of 300 systematic reviews

ObjectivesOpen science practices, including preregistration and sharing of data and code, aim to improve transparency and reproducibility. However, systematic reviews often remain difficult to replicate without access to underlying materials. We aimed to estimate the prevalence of key open science indicators in interventional systematic reviews and assess temporal changes by comparing a 2024 cohort with previously published cohorts from 2014 and 2020.DesignCross-sectional meta-research study.Data sourcesMEDLINE, Embase, and CINAHL were searched in June 2024 for English-language records indexed between 1 January and 1 February 2024.Eligibility criteriaInterventional systematic reviews with human participants, a clearly defined PICO question, a systematic search, risk-of-bias assessment, and at least one meta-analysis.MethodsFollowing a registered protocol and STROBE guidance, records were randomly sampled and screened in duplicate until 300 reviews were included. Data extraction was conducted in duplicate. We assessed protocol registration and availability, data sharing, code sharing, and preprint dissemination. Preprints were identified through manual searches across >40 servers and author contact. Temporal changes were analysed by comparing the 2024 cohort with 2014 and 2020 cohorts using pairwise risk ratios (RRs) with predefined equivalence ranges.ResultsAmong 300 reviews, cancer was the most common topic (17%), and pharmacological/biological interventions the most common type (35%). The median number of included studies was 14 (IQR 9–23). Most reviews (73%) were published in journals with data/code-sharing policies.In 2024, 70% reported a registered protocol and 61% included a data availability statement. However, only 24% shared underlying data and 1% shared analytical code. Preprints were rare (2%).Protocol registration increased over time (2024 vs 2014: RR 2.03, 95% CI 1.55 to 2.65; 2024 vs 2020: RR 1.67, 1.43 to 1.95), as did data availability statements (2024 vs 2020: RR 1.98, 1.63 to 2.40). Data sharing showed a nonlinear pattern, decreasing from 2014 to 2020 (RR 0.22, 0.13 to 0.36) and increasing from 2020 to 2024 (RR 3.71, 2.30 to 6.00), but remained uncommon. Code sharing showed no detectable change (OR 1.00, 0.15 to 6.51). In 2024, 3% of protocols were registered after submission.ConclusionsTransparency indicators have improved, yet materials required for reproducibility are rarely shared. Reusable data, analytic code, and preprints remain uncommon outside pandemics. These findings highlight a persistent gap between declared openness and practical accessibility, limiting independent verification and reuse of systematic review evidence and underscoring the need for stronger journal and funder requirements.