When nutrition science fails: A bibliometric analysis of retracted diet-related articles (2000-2025)

IntroductionRetractions aim to remove flawed science, yet delayed withdrawals allow erroneous nutrition research to influence reviews, guidelines, and textbooks. This bibliometric study examined causes, geographic patterns, and post-retraction impact of nutrition articles withdrawn between 2000 and 2025.MethodsPubMed/MEDLINE and PubMed Central were searched (30 April 2025) for records indexed as 'Retracted Publication' using human-nutrition MeSH terms. Inclusion required a human-nutrition focus and formal retraction notice. Scopus provided citation counts, document types, and author affiliations. Notices were coded into eight exclusive categories (κ = 0.86). Time-to-retraction (TTR) was the interval, in years, between publication and withdrawal. Descriptive analyses used JAMOVI.ResultsForty-five retracted articles were identified. Leading causes were methodological/statistical errors (18.9%), lack of ethics approval (15.1%), and data-integrity breaches (11.3%). Mean TTR was 5.7 ± 4.8 years (median = 4.2). China and the U.S. contributed most to absolute numbers (28 % and 14 %), while Greece and Nigeria had the highest retraction densities. The corpus accrued 1155 citations, 53 % post-retraction; 62 % of those appeared in reviews or meta-analyses. Citation half-life post-withdrawal was 3.3 years, and fewer than 10 % included explicit warnings.ConclusionDelayed retractions enable flawed findings to distort pooled estimates in secondary research. Persistent citations expose gaps in alert systems. Most retractions stem from methodological or ethical flaws, yet their influence lingers. To safeguard scientific integrity, retraction notices must be standardised and machine-readable, peer review must include automated checks, and graduate education should incorporate training in citation.