Effectiveness of Food-safety and Allergen Training among Food Handlers: A Systematic Review

Abstract

Background: Food-safety and allergen training is widely mandated or recommended, yet training can improve knowledge without reliably changing routine behaviour, organisational controls, or health outcomes. Objective: To determine the effectiveness of training for occupational food handlers in improving food-safety and food-allergen knowledge, attitudes, self-efficacy, observed or reported practices, compliance, microbiological contamination, and incident outcomes. Methods: Openly accessible, source-restricted web searches of PubMed/PMC-indexed records, ClinicalTrials.gov, publisher and DOI records, and backward citation links were completed on 22 August 2026. Primary studies published from 1 January 2015 to 30 June 2026 were eligible when they evaluated prospective training or compared previously trained with untrained occupational food handlers. One reviewer screened, extracted, and appraised the evidence. Randomised studies were assessed with RoB 2; non-randomised intervention studies with ROBINS-I; and training-exposure studies with ROBINS-E. Heterogeneous results were synthesised by outcome domain without pooling. Results: Of 107 records, 24 duplicates were removed, 83 records were screened, 48 reports underwent detailed assessment, and 35 reports representing 34 studies were included. Twenty-six studies addressed general food safety and eight addressed allergens. Training generally improved knowledge and often improved selected practices, particularly when practical demonstration, observation, feedback, repeated reinforcement, or workplace coaching was used. Microbiological improvements were reported in several small uncontrolled studies. Allergen evidence was mostly cross-sectional and inconsistent after adjustment; one large video-based before-after study showed improvements in selected knowledge items. No study directly measured verified foodborne illness or allergic-reaction incidence. Two randomised trials had some concerns; all non-randomised studies had serious risk of bias. Confidence was low for knowledge and observed practice, and very low for microbiological, attitudinal, and allergen outcomes. Conclusions: Training is likely to improve proximal learning and may improve selected behaviours, but education alone cannot be assumed to deliver sustained operational control or prevent illness. Practical, repeated, behaviourally reinforced training embedded in organisational systems is more credible than one-off information delivery. Robust cluster trials with validated observation, microbiological surveillance, longer follow-up, and incident outcomes are needed.