Agricultural evidence synthesis at a critical juncture: Quality gaps and implications for evidence-based agriculture

Systematic reviews and meta-analyses have become essential tools for synthesizing evidence to guide agricultural policy and practice, yet whether these syntheses are themselves conducted with sufficient rigor has rarely been systematically assessed. This study provides a field-wide assessment of reporting quality in agricultural evidence synthesis, examining whether existing reviews are conducted with sufficient transparency and rigor to support the policy and practice decisions they are meant to inform. We drew a random sample of 244 systematic reviews and meta-analyses from a bibliometric dataset of 1709 articles published between 1997 and 2023, evaluating reporting quality using a 12-indicator rubric across four domains: reproducibility, dual-purpose infrastructure, bias mitigation, and best practices. Articles achieved only 32% of the maximum possible quality score. Dual screening appeared in just 4% of articles, only 18% assessed included-study quality, and only 12% registered a protocol in advance. Systematic reviews substantially outperformed meta-analyses across nearly every domain, suggesting many researchers treat meta-analysis as a standalone statistical procedure rather than a component embedded within systematic review. Quality improved between 2009 and 2016 but has since plateaued despite rapid publication growth. Guideline adherence was the strongest predictor of quality, explaining 20% of observed variance, a far larger effect than journal prestige or international collaboration, indicating that the deficiencies identified are remediable rather than structural and carry real consequences for the reliability of evidence informing agricultural policy. We propose the CREDIT framework, targeting Collaboration, Resources, Education, Documentation, Integration, and Tools, as a coordinated response to the barriers underlying these gaps.