Promoting hygienic weaning food handling practices through a community-based programme: Intervention implementation and baseline characteristics for a cluster randomised controlled trial in rural Gambia

Objective Contamination of weaning food leads to diarrhoea in children under 5 years. Public health interventions to improve practices in low-income and middle-income countries are rare and often not evaluated using a randomised method. We describe an intervention implementation and provide baseline data for such a trial. Design Clustered randomised controlled trial. Setting Rural Gambia. Participants 15 villages/clusters each with 20 randomly selected mothers with children aged 6-24 months per arm. Intervention To develop the public health intervention, we used: (A) formative research findings to determine theoretically based critical control point corrective measures and motivational drives for behaviour change of mothers; (B) lessons from a community-based weaning food hygiene programme in Nepal and a handwashing intervention programme in India; and (C) culturally based performing arts, competitions and environmental clues. Four intensive intervention days per village involved the existing health systems and village/cultural structures that enabled per-protocol implementation and engagement of whole villager communities. Results Baseline village and mother’s characteristics were balanced between the arms after randomisation. Most villages were farming villages accessing health centres within 10 miles, with no schools but numerous village committees and representing all Gambia’s three main ethnic groups. Mothers were mainly illiterate (60%) and farmers (92%); 24% and 10% of children under 5 years were reported to have diarrhoea and respiratory symptoms, respectively, in the last 7 days (dry season). Intervention process engaged whole village members and provided lessons for future implementation; culturally adapted performing arts were an important element. Conclusion This research has potential as a new low-cost and broadly available public health programme to reduce infection through weaning food. The theory-based intervention was widely consulted in the Gambia and with experts and was well accepted by the communities. Baseline analysis provides socioeconomic data and confirmation of Unicefs Multiple Indicator Cluster Survey (MICS) data on the prevalence of diarrhoea and respiratory symptoms in the dry season in the poorest region of Gambia. Trial registration number PACTR201410000859336; Pre-results.

Villages were the unit of randomisation for this parallel cRCT. The 4-day community intervention was followed by a reminder visit after 5 months. Two cross-sectional samples were taken to measure baseline characteristics and outcomes: one before randomisation and the other 6 months postintervention roll-out. There were no changes to the protocol after commencement. The cRCT was conducted in the Central River Region (CRR), one of the Gambia’s administrative regions. CRR is 48 000 km2 in area, organised into 11 districts with 659 villages, and has a population of 201 506 of which 41 334 (20%) are children under 5 years of age.10 CRR was selected for the intervention as it has the highest incidence diarrhoea in the Gambia, particularly in children aged 6–24 months (26.5% of children under 5 years had diarrhoea in the 2 weeks preceding the Unicef Multiple Indicator Cluster Survey (MICS) in 2010 vs 17% nationally.11 The rates for acute respiratory infection (ARI) of children under 5 years were 14.2% in CRR compared with 6% nationally). CRR is rural, with low literacy, and is economically the poorest region in the Gambia. Villages in the region differ in their access to water supply and healthcare. A typical village has a head and a religious leader, but the size of settlements registered on the national population census (in 2013) ranges from as few as 27 to 1800 population per village, giving mean village size for CRR of 357 (SD±59).10 As with the other regions in the Gambia, Unicef and the Gambian Ministry of Health and Social Welfare (MOH) have selected a number of villages (158 in CRR) to become primary healthcare (PHC) villages where they have trained (for 4 weeks) a village health worker (VHW) and a traditional birth attendant (TBA) to provide health promotion and basic health support to the villagers.12 Inclusion criteria for study villages for the intervention were PHC villages in CRR with a population of 200–450. It was felt that such villages, with lay health workers, would be best able to support the programme given the available resources. The 200–450 population criteria per village was decided on three grounds: the requirement for a minimum of 20 families with children aged 6–24 months, a population close to the mean village size in CRR (357) and the need to avoid villages that were too large given the size of the team implementing the intervention. Exclusions for the villages were those that were within 5 km of already selected villages. Inclusion criteria for households within the villages for the baseline were mothers with children aged 6–24 months; exclusions were those expecting not to be resident in the village for the following 6 months. There were no other exclusions. Sample size calculation (online supplementary file 1) was based on data from formative research investigating behaviours and testing food and water samples for faecal coliforms. bmjopen-2017-017573supp001.pdf The villages were randomly selected by an epidemiologist in the UK, aware of the biases potentially associated with a non-random village sampling, from a list of all villages in CRR after applying the selection criteria. We provided written and oral information and sought informed consent from the village heads for the villagers’ participation in the programme. For the baseline, a list of all mothers with children aged between 6 months and 24 months living in the village at the time was obtained from the maternal child health register, and households were chosen randomly, based on the study criteria. Mothers gave written informed consent. In case of illiteracy, the information was read out (and a written copy left behind), and a thumb print was obtained in the presence of a family witness and the fieldworker. During the initial recruitment visit (December 2014; dry season), after consent, we characterised all 30 villages and 201 randomly chosen mothers within them before randomisation and collected data about socioeconomic background of the families and diarrhoea and respiratory illnesses of the index child over the last 7 days. Randomisation took place after all village heads provided consent and the baseline data collection had been completed. Randomisation was conducted by a statistician in the UK using a computerised random number generator. The villages were grouped and randomised within strata (north or south of the river and by quartiles of the village population) into 15 control and 15 intervention villages. Allocation concealment was not possible because the intervention team had to know which village would receive the intervention before it was implemented. While it was not possible to blind the implementers of the intervention programme or the families who received the intervention, the families exposed to the intervention were unaware of the comparative nature of the intervention with a control village. This article presents the data for the baseline, which are analysed using descriptive summaries. After consent by the head of village and randomisation, the control villages received a 1-day visit by a public health officer (PHO) who, using a flip chart during a village gathering, talked about using water in household gardening. No further visits were made to the control villages. The intervention components and delivery package were theoretically based and informed by the local context from our formative research and by the lessons/tools from community interventions in handwashing studies in India13 and weaning food hygiene in Nepal.7 The latter employed the same theoretical models in similar study questions. The intervention comprised a community participation campaign delivered to all the villages and focused on mothers of weaning babies and those with children under 5 years in whole village. The intervention team visited each village on days 1, 2, 17 and 25 and was delivered between 16 February and 28 April 2015 (the dry season). A set of activities was conducted that involved mothers and other village members in village-wide events, neighbourhood meetings and home visits, with the wider involvement of the village authorities and volunteers.13 We included a fifth visit after 6 months as it was envisaged that were such a programme to be implemented at scale, then for the behaviour change to be sustained, villages would require a reminder visit before or early in the diarrhoea high-risk rainy season.14 Mothers and their families are busy at this time and hence more likely to forget weaning food hygiene behaviour. The programme’s daily schedule and tools and including their links with the motivational theory are summarised in tables 1 and 2. Details of intervention activities and during visits to the intervention villages The idea of a 4-day programme was adapted from the India SuperAmma study.13 However, the details of the activities during each day of the Team’s visit to the villages were adapted from Gautam et al’s Nepal study.13 Itself drawing on aspects from the SuperAmma India study (see footnote to table 3 for source of adapted tools). PHO, public health officer; TBA, traditional birth attendant; TC, traditional communicator; VHW, village health worker. Intervention tools and their application during the intervention *Tool adapted from Weaning-food Hygiene Nepal study.7 †Tool adapted from SuperAmma India Handwashing study.19 PHO, public health officer; TBA, traditional birth attendant; TC, traditional communicator; UV, ultraviolet; VHW, village health worker. We used two theoretical frameworks in designing the intervention. First, Hazard Analysis and Critical Control Points (HACCP),15 16 which are conventionally used in the food processing industry to reduce microbiological contamination. The WHO/FAO Expert Committee on Food Safety has recommended the use of HACCP[s] in homes in LMICs to provide insight into food preparation hazards and remedial preventive measures.16 17 There is also evidence from efficacy and a small population trial that weaning food hygiene activities following the HACCP approach can help identify measures to improve weaning food safety.16 Table 3 summarises the corrective measures that were prioritised following our formative research.8 Critical control points and corrective measures (practices) and handwashing motivational drivers that were targeted by our weaning food hygiene intervention Second, we used an applied motivational behaviour change model18 that facilitated the application of identified corrective measures in a way that would add to mother’s knowledge and attitude and would motivate a change in mother’s behaviour. The model draws on psychology research that proposes ways of classifying various drivers of human behaviour. Our formative research found that nurture, disgust, affiliation, status and purity were the strongest motivational drives for our village mothers.8 As with the India and Nepal programmes,7 19 we focused on the use of performing arts (using culturally ingrained styles of drama and songs),20 competitions and environmental cues21 to deliver the HACCP corrective measures and motivational drives. Details of our community weaning food hygiene programme, which was designed by the research team at the University of Birmingham (which included a Gambian PHO from MOH), were widely consulted with expert health promotion agencies who were represented on a Local Scientific Advisory Committee in the Gambia (MOH, Unicef, WHO, University of the Gambia, National Nutrition Agency (NANA) and the MRC Gambia). Subsequently, the material was translated into the three local languages (Mandinka, Wolof and Fula), field-tested and piloted iteratively by the intervention team in the CRR. This team, which also delivered the programme, comprised one literate male and one illiterate female traditional communicators (TCs) with health promotion experience, three PHOs from the local Regional Public Health Department (two with Higher National Diploma from the Gambian College School of Public Health with an additional Masters in Public Health) and an illiterate driver (for 24 days of the 60 days of the village visit, there were two PHOs in the team, while for the remainder days, there were three PHOs). TCs are performing artists who use traditional African drumming, singing and acting to communicate behaviours. The team were deliberately selected from the within existing structures in rural Gambia to demonstrate replicability and scaling. The team conducting the intervention during the 4-day village visits was assisted by a female volunteer (usually a TBA) from each village who received 2 weeks training assisted the work programme during, and in-between, the team visits. The TBAs were encouraged to find one or more assistant volunteers by day 1 of the team’s visit (3 visits in smaller villages ended with no assistants, 11 had one assistant and 1 had three assistants). The assistants were called ‘MaaSupervisors’ and visited the families between team visits to recruit more mothers of young children, reinforce the target practices and hence help ingrain the practices with the cultural norms of the wider community. The intervention focused on a central role model character, the ‘MaaChampion’, a mother who practised the key behaviours used in the messages (table 3) and encouraged other mothers to do the same. Village mothers could achieve ‘MaaChampion’ status if they successfully demonstrate the practices and knowledge and encouraged two other mothers to do so. ‘Funtu’ (a derogatory noun for a discarded useless thing) was another character: a mother who failed to practise any of the target behaviours and reaped the consequences with her family and other villagers. These two characters were described using story drama and songs in the context of an average village. Together they demonstrated all the key behaviours and motivational drives and engendered a wish for behaviour change in village mothers as they identified with the characters’ lifestyles and behaviours. Other components such as competitions (for mothers of children younger than 5 years), environmental cues (for mothers engaged in the competitions) and demonstrations had an important role in embedding behaviour change. The programme’s daily schedule and tools, including their link with the motivational theory, are summarised in tables 1 and 2. Overall, the aim was to apply theory and apply successful elements of two previous studies7 19 while ensuring the intervention was as simple and cost-effective as possible. It also needed to be understandable and replicable by existing local health system/staff in the Gambia. Implementation was staggered over 2 months. During implementation of the intervention, there were no diversions from the protocol. The intervention team logged significant events, comments and the overall participation of villagers/mothers in the programme to enable full evaluation of the intervention implementation. At the end of the intervention implementation, the intervention team were interviewed in a focus group discussion to explore the experience of the team during village visits and implementation and identify successful elements and learning points. These will be reported in a qualitative publication. The details of the intervention were developed in consultation with mothers and villagers during an extensive piloting phase. There were no particular patient advisors. The results will be communicated after the follow-up is complete through the PHOs who visit the villages. There was no other involvement of patients.

• manjang, buba
• hemming, karla
• bradley, chris
• ensink, jeroen h.j.
• martin, james t.
• sowe, jama
• jarju, abdou
• cairncross, sandy
• manaseki-holland, semira