IntroductionBehavioural interventions could improve caregivers’ food hygiene practices in low-resource settings. So far, evidence is limited to small-scale and short-term studies, and few have evaluated the long-term maintenance of promoted behaviours. We evaluated the effect of a relatively large-scale behaviour change intervention on medium and long-term maintenance of household food hygiene practices in Bangladesh. MethodsWe analyse a secondary outcome of the Food and Agricultural Approaches to Reducing Malnutrition (FAARM) cluster-randomised trial and its sub-study Food Hygiene to reduce Environmental Enteric Dysfunction (FHEED), conducted in Habiganj district, Sylhet division, Bangladesh. The FAARM trial used a 1:1 parallel arm design and included 2705 women in 96 settlements: 48 intervention and 48 control. Women in the intervention settlements received training in homestead gardening, poultry rearing and nutrition over three years (2015–2018), complemented by an eight-month (mid-2017 to early-2018) behaviour change component on food hygiene using motivational drivers. Nested within the FAARM trial, the FHEED sub-study evaluated several outcomes along the hygiene pathway. For this article, we evaluated household food hygiene behaviours by analysing structured observation data collected in two cross-sectional surveys, four and 16 months after the food hygiene promotion ended, from two independent subsamples of FAARM women with children aged 6–18 months. We assessed intervention effects on food hygiene practices using mixed-effects logistic regression, accounting for clustering. In exploratory analyses, we further assessed behaviour patterns – how often critical food hygiene behaviours were performed individually, in combination and consistently across events. ResultsBased on the analysis of 524 complementary feeding and 800 food preparation events in households from 571 participant women, we found that intervention households practised better food hygiene than controls four months post-intervention, with somewhat smaller differences after 16 months. Overall, the intervention positively affected food hygiene, particularly around child feeding: using soap for handwashing (odds ratio 5·8, 95% CI 2·2–15·2), cleaning feeding utensils (3·8, 1·9–7·7), and cooking fresh/reheating food (1·8, 1·1–2·8). However, the simultaneous practice of several behaviours was rare, occurring in only 10% of feeding events (intervention: 15%; control: 4%), and the practice of safe food hygiene behaviours was inconsistent between events. ConclusionOur findings suggest that a motivational behaviour change intervention encouraged caregivers to maintain certain safe food hygiene practices in a rural setting. However, substantial physical changes in the household environment are likely needed to make these behaviours habitual. Trial registration numberNCT02505711.