JMIR Publications

ABSTRACT

Abstract: The proliferation of overlapping and redundant secondary evidence syntheses has emerged as a critical meta-research concern: multiple review teams independently addressing near-identical clinical questions using overlapping sets of primary studies waste finite research resources and risk distorting the perceived weight of clinical evidence. This review develops a three-dimensional analytical framework—spanning methodological origin, execution-level primary study overlap, and systemic proliferation drivers—to characterize the sources, scale, and consequences of redundancy in evidence synthesis, and critically evaluates proposed solutions against their implementation barriers. Drawing on meta-research studies, meta-epidemiological investigations, and policy analyses published between 2018 and 2026, the review integrates empirical findings on primary study overlap quantified through the corrected covered area metric, meta-research on PROSPERO registration duplication, and analyses of the incentive structures that drive redundant synthesis production. The evidence indicates that methodological redundancy originates from inadequate pre-protocol scoping searches and the absence of automated duplicate detection in registration platforms such as PROSPERO, and that meta-epidemiological studies consistently document moderate to high primary study overlap across systematic reviews on the same clinical question, with overlap rates frequently exceeding 20%. The COVID-19 pandemic functioned as a natural experiment demonstrating that, without coordination infrastructure, surge research conditions amplify redundancy dramatically. Downstream consequences include artificial inflation of evidence certainty through pseudo-replication,direct resource waste with estimates that vary widely depending on costing methodology, and erosion of methodological trust in evidence synthesis as a scientific method. Three unresolved challenges that cut across this analysis demand priority attention: establishing consensus on acceptable redundancy thresholds that distinguish warranted replication from wasteful duplication, developing validated and reproducible methods for AI-assisted overlap detection and evidence synthesis automation, and constructing resource-waste estimates grounded in transparent, empirically defensible costing frameworks rather than extrapolation from small convenience samples.