A community-level complementary-food safety and hygiene intervention improves family-food preparation behaviours in rural Gambia: a follow-up of a cluster randomised controlled trial

Discussion We present the results of a 32-month post-intervention evaluation of the MaaChampion cluster-RCT for its additional benefit on improved food hygiene behaviours of caregivers while preparing family food. The MaaChampion intervention improved overall adherence to a composite measure of five behaviours associated with preparing and cooking family food 32 months after intervention delivery. This was mainly driven by improvements in the practice of three behaviours relating to handwashing with soap while preparing food, and was supported by the finding that a greater proportion of the intervention villages had soap for handwashing compared with the control villages. To the best of our knowledge, this is the first RCT to examine the effect of an intervention on household family-food hygiene practices in LMICs. Our systematic review (publication in draft) highlights five former studies25–29 that reported gains in knowledge, attitudes and practices concerning family-food hygiene in response to a food hygiene intervention. These studies took place in LMIC settings (Bangladesh, Iran, Cambodia and Laos) and assessed food hygiene interventions of varying complexity from a single small group teaching session in Cambodia and Laos29 to eight visits held across 8 months in Bangladesh.25 However, studies were generally of poor quality: no trial involved a control group and the majority did not assess the statistical significance of the primary outcome.25 27–29 In all included studies, post-intervention follow-up did not exceed 1 month, and in one study, follow-up took place immediately after intervention delivery25, impairing conclusions regarding lasting behaviour change. Furthermore, those delivering the intervention also performed outcome assessment which risked observation bias.25 28 29 Encouragingly, intervention development in three sites was informed by the results of a baseline outcome assessment,27–29 or by formative research in the study site (Bangladesh),25 but only one intervention in Bangladesh25 targeted emotional drivers of behaviour change (eg, fear or disgust) over ‘Germ Theory’. Our intervention targeted complementary food but also improved food hygiene behaviours for the wider family that was still evident 32 months later. Improvements in complementary-food hygiene behaviours at 6 months more pronounced than at 32 months follow-up;24 hence, we would expect the same for family-food hygiene behaviours at 6 months (although this was not assessed) with possible health benefit to family members. At 32 months, the intervention significantly improved three handwashing behaviours among five tested family-food hygiene practices. Internal validity is provided through increased soap availability noted in intervention homes, despite no soap distribution or any messages to enforce having soap in the kitchen. This may have been due to particularly effective intervention components that specifically addressed food-hygiene related handwashing: the ‘Glow-Germ’ game and/or the ‘SuperAmma’ handwashing animation videos.17 Alternatively, handwashing-related interventions by other organisations (including non-governmental organisations) in the region since our intervention may have inadvertently had a greater impact among our intervention villages over those allocated to the control group. The intervention did not sustain benefits in reheating leftover food for family consumption, although the study found limited opportunities for reheating family food in both intervention and control groups as both tended to cook fresh dishes for adult meals, likely reducing the power to detect any differences. In The Gambia, formative research indicated that adult leftovers were usually reheated before eating anyway (not commonly done for child’s complementary food).24 Differences in the practice of washing pots and utensils were minimal between groups. At the 6-month assessment, this behaviour, regarding complementary food, was uncommon only in control villages24; similar findings were observed during formative research.8 By 32 months, adherence to this practice was high (around 80%) among intervention and control groups regarding complementary food, attributed to cross-contamination.24 Additionally, qualitative data suggested that villagers took pride in sharing adopted behaviours with others as MaaChampions and MaaSupervisors,24 and as washing pots and drying on a clean surface was the most visible behaviour to outsiders, this may have been adopted more than handwashing behaviours, affecting both child and family food preparations similarly. Globally, children under 5 years of age have the greatest rates of diarrhoea morbidity of any age group30 and this intervention reduced diarrhoea among children aged 6–36 months.24 This was probably because, in addition to poor complementary-food hygiene practices, poor family-food hygiene behaviour also places children under 5 years of age at risk of diarrhoea, as family food is also fed to the majority of children aged 18 months to 5 years in LMICs,8–13 with children aged 6–24 months being frequently fed family food in West Africa.10 Additionally, complementary food is at risk of contamination from poor family food hygiene practice as caregivers in The Gambia may prepare both foods simultaneously.8 Therefore, improvements in handwashing behaviour when preparing family food, as well as subsequent reduced diarrhoea in family members resulting in a reduced environmental prevalence of pathogens, may have contributed to the reduction in diarrhoea incidence among complementary food aged children as previously reported.24 Furthermore, the importance of the reduction of diarrhoeal disease among those aged over 5 years must not be overlooked. Globally, diarrhoeal diseases remain among the top 12 causes for disability adjusted life years among adolescents, young adults and the elderly.31 Strengths of this study include the intervention’s strong theoretical base, previously used in other successful food and hand hygiene trials.9 17 Powerful motivational drivers were strongly associated with the target behaviours in the stories, drama and songs. The intervention also prepared the social environment by gaining community support through the involvement of key figures like older mothers, fathers and elders, and by encouraging caregivers to publicly commit to adopting the behaviours. These actions aimed to elevate their status as role models, supported by displaying their photos, offering peer education and mobilising the community.23 32 Qualitative data indicate that the MaaChampion programme was seen as a beneficial and joyful local initiative, visibly improving children’s health and reducing family expenses. The physical environment saw minimal change, except for the plastic sheets that were distributed for drying dishes, placing educational posters and danglers in kitchens and rewarding successful caregivers with a bar of soap. The intervention was low-cost,24 33 culturally sensitive, used local healthcare workers and was based on mixed methods formative research in the region. These factors have been called for by Woldt and Moy5, Touré et al and Curtis et al22 34 to encourage intervention delivery and behaviour uptake at scale. Regarding outcome assessment, structured observation sheets were used to assess food hygiene behaviours, rather than self-reported questionnaires, as the latter frequently over-report the incidence of behaviours.35 Moreover, the period of daily observing caregivers was more than twice as long as previous community-based hygiene trials,9 17 36 with family-food preparation being assessed at both breakfast and lunchtime, improving the reliability of the findings. Overall, this is the longest follow-up of any food hygiene intervention conducted in a domestic LMIC setting to date, thus addressing the question of whether behavioural improvements are sustained in the long term. Limitations include the possible presence of reactivity and observation bias.24 Reactivity bias occurs where caregivers may have only demonstrated improved food hygiene practice because they were aware they were being observed.37 However, the underlying purpose of the data collection was concealed from participants and fieldworkers, and we attempted to minimise this bias by only having two follow-ups and not evaluating food preparation repeatedly.17 We also used different teams to deliver the intervention, collect baseline and outcome measurements. The follow-up time meant that the majority of children at the 6-month follow-up were not in the age range at the 32-month follow-up and together with random selection helped to reduce the chance of the same caregivers being assessed at both 6 and 32 months. Outcome evaluation took place in one village per day, ensuring caregivers were not able to discuss the fieldworker observations and questions among themselves between assessment days. There is likely to have been intervention cross-contamination, whereby control caregivers may have adopted family-food hygiene behaviours from intervention villages by the time of 32-month follow-up, which may have diluted effect sizes.24 Observer bias was curtailed in that outcome assessors were not made aware of the intervention, nor of the purpose of the evaluation, and were recruited from non-study villages. Nevertheless, participant caregivers may have informed observers they had MaaChampion in their village, which may have influenced the observer’s assessment of behaviour. It was also not possible to establish whether an increase in handwashing with soap led to a direct reduction in microbiological contamination of family food, as we were unable to measure faecal form counts. However, at 6-month follow-up,24 reductions in laboratory assessed Escherichia coli counts of complementary-food samples were statistically significant and mirrored the statistical significance of behaviour changes observed in the mother’s adherence to complementary-food hygiene behaviours. It is therefore conceivable that the improved adherence to family-food hygiene behaviours we observed also improved microbiological contamination rates of family-food. In summary, the MaaChampion cluster-RCT found significant improvements in caregiver’s family-food handwashing practices nearly 2.5 years following an intervention to improve their complementary-food hygiene practices, without further community visits. This addressed the clear evidence gap we have identified in the literature.25–29 It is significant that our study’s outcome behaviours were not the primary objective of the intervention and were observed with important methodological improvements to reduce reactivity and observation biases. This could have important implications for adults and children over 2 years of age, thus reducing the pool of infection and cases of diarrhoea morbidity and mortality in all age groups. Therefore, this intervention warrants replication and testing in larger trials38 which ought to target both complementary-food and family-food hygiene behaviours.