Hoffmann 2018b.

Methods	Study design: cluster‐randomised controlled trial Country: Kenya Setting: villages within two maize‐producing counties (Meru, Tharaka‐Nithi) in Eastern Kenya Study aim: to evaluate the impact of a package of post‐harvest technologies appropriate for use by smallholder farmers on contamination of maize with aflatoxin Study dates: June 2013 to October or November 2013 Recruitment: farmers from households in intervention villages were visited at their homes and invited to attend an information meeting in the village about aflatoxin prevention and access to post‐harvest technologies. Farmers from households in control villages were approached by a trained farmer in their village to attend a training session on aflatoxin prevention. Sampling: random sampling of villages Sample size justification: no Unit of randomisation: villages Total number randomised: 30 villages (15 intervention; 15 control); 679 households (350 intervention; 329 control) Unit of analysis: households; standard errors were bootstrapped to account for the effect of clustering at the village level Intention‐to‐treat analysis: Yes. Quote: “Analysis of the intervention’s impact was conducted using multivariate linear regression based on an intent‐to‐treat approach…” Missing control variables were imputed for use in multivariate linear regression." Attrition: 139 households (82 intervention; 47 control) for intent to pay for treatment technology and reported post‐harvest losses, but 508 households (277 intervention; 231 control) for aflatoxin contamination of household stored maize Relevant study limitations as reported by study authors: farmers who still had maize in their households three months after harvest were relatively more well‐off, on average, than those who did not. Small sample sizes limited conclusions on the relative contributions of training and technology. Aflatoxin contamination varies considerably by year; research spanning several seasons would have been ideal.
Participants	Inclusion criteria: households with women who were at least 18 years of age and in their fifth to final month of pregnancy, as estimated by the woman. Infants delivered from the pregnancy for which their mothers were enrolled; if an infant died or moved outside the study area, a child aged between 12 months and 35 months, residing in the household, was taken as a replacement child. Exclusion criteria: NR Group differences: more control households had heard of aflatoxins, compared to intervention households (P = 0.01) Subgroups: subsistence farmers vs market producers Baseline characteristics of participants Intervention group (N = 329) Mother Mother, age: NR Marital status: NR Mother, educational level: proportion with any secondary school: 0.20 Occupation: NR Mother, nutritional status: NR Mother, dietary diversity or food security measure: NR Mother, biomarker of aflatoxin exposure: NR Child Child, age: NR Child, gender: NR Child, nutritional status: NR Child, dietary diversity or food security measure: NR Child, morbidity: NR Child, biomarker of aflatoxin exposure: NR Household Place of residence: NR Race, ethnicity, and religion: NR Socio‐economic status: household size, mean (SE): 5.1 (0.10); head employed as farm labourer in past month, mean (SE): 0.21 (0.02); monthly consumption expenditure per adult equivalent in Kenyan Shillings (KSh), mean (SE): 3709 (137); number of non‐animal assets, mean (SE): 1.7 (0.08); acres of land owned, mean (SE): 1.6 (0.09) Social capital: NR Household head, age: in years, mean (SD): 37 (11.97) Household head, gender: proportion female, mean (SD): 0.12 (0.37) Household head, educational level: proportion with primary school, mean (SD): 0.67 (0.56); secondary school, mean (SD): 0.15 (0.37) Household, food security, or dietary diversity measure: NR Prior knowledge of aflatoxin: heard of aflatoxin, mean (SE): 0.59 (0.03); describes aflatoxin as mould, mean (SE): 0.38 (0.03); describes aflatoxin as toxin, mean (SE): 0.05 (0.01); describes aflatoxin as both mould and toxin, mean (SE): 0.09 (0.02); knows drying maize before storage prevents aflatoxin, mean (SE): 0.83 (0.02); knows drying maize off bare ground prevents aflatoxin, mean (SE): 0.22 (0.02) Control group (N = 350) Mother Mother, age: NR Marital status: NR Mother, educational level: proportion with any secondary school: 0.18 Occupation: NR Mother, nutritional status: NR Mother, dietary diversity or food security measure: NR Mother, biomarker of aflatoxin exposure: NR Child Child, age: NR Child, gender: NR Child, nutritional status: NR Child, dietary diversity or food security measure: NR Child, morbidity: NR Child, biomarker of aflatoxin exposure: NR Household Place of residence: NR Race/ethnicity and religion: NR Socio‐economic status: household size, mean (SE): 5.1 (0.10); head employed as farm labourer in past month, mean (SE): 0.17 (0.02); monthly consumption expenditure per adult equivalent in KSh, mean (SE): 3648 (150); number of non‐animal assets, mean (SE): 1.8 (0.09); acres of land owned, mean (SE): 1.9 (0.10) Social capital: NR Household head, age: in years, mean (SD): 37 (11.06) Household head, gender: proportion female, mean (SD): 0.09 (0.36) Household head, educational level: proportion with primary school, mean (SD): 0.70 (0.54); secondary school, mean (SD): 0.17 (0.36) Household, food security or dietary diversity measure: NR Prior knowledge of aflatoxin: heard of aflatoxin, mean (SE): 0.73 (0.02); describes aflatoxin as mould, mean (SE): 0.46 (0.03); describes aflatoxin as toxin, mean (SE): 0.08 (0.01); describes aflatoxin as both mould and toxin, mean (SE): 0.11 (0.02); knows drying maize before storage prevents aflatoxin, mean (SE): 0.87 (0.02); knows drying maize off bare ground prevents aflatoxin, mean (SE): 0.27 (0.02)
Interventions	Intervention: Type: package of post‐harvest technologies plus standard of care for aflatoxin prevention Description: aflatoxin prevention training: standard of care aflatoxin prevention training using the 'training the trainers' approach. Intervention: farmers attended village meeting during which they heard recommended post‐harvest practices for aflatoxin control and the use of a mobile maize dryer. Post‐harvest agricultural intervention: households with an expected harvest of > 45 kg randomly assigned to one of the following groups in terms of access to a mobile maize dryer and hermetic storage bags: 1. Full discount (free access); 2. Partial discount (150 KSh per 90 kg bag), or 3. No discount (350 KSh per 90 kg bag). Intervention households were also randomly assigned (1:1) to a market incentive or no market incentive. Separate meetings conducted in each village with farmers assigned to the market incentive payment vs those not offered payment. Aflatoxin prevention training by 'training the trainers' approach: at least one farmer from each of the study villages was selected, in consultation with community leaders, to receive training re: the causes and consequences of aflatoxin contamination in maize and recommended practices for aflatoxin prevention. Delivery: intervention: additional training conducted at village meetings with eligible farmers. Quotation: "more detailed than the training of trainers offered to the villages". A booklet given to each household describing recommended post‐harvest practices and how to access a mobile maize dryer. Post‐harvest agricultural technology: each household received free plastic sheeting. Appointments via telephone for the use of the dryer within a village. A study dryer transported to a central location within the village. The drying service included transportation of farmers and their maize from their homestead to the dryer location, measurement of initial grain moisture content, use of the flatbed dryer, and post‐drying moisture testing. Aflatoxin prevention training: elected representative farmers trained and asked to pass the information on to other farmers in their village Providers: study staff (intervention); master trainers (aflatoxin prevention training) Duration of intervention: five months Duration of follow‐up: five months Co‐intervention: NR Cost: standard of care for aflatoxin prevention: cost of master trainers (who facilitated the training of one farmer from each village): approximately USD 1.70 per farmer (two meetings with 25 farmers each per day, transport rental was USD 60 and trainers' wages were USD 25). Compensation for trained farmers (who facilitated training of farmers in their own community) not reported by study authors. Intervention: provision of drying sheets: USD 5 per farmer; training: USD 1.70 per farmer (included vehicle rental cost of USD 60 per day and trainer wage of USD 25 for two meetings of 25 farmers per day); mobile drying service: 2.93 Ksh per kg (fully subsidised); 0.73 Ksh per kg (partially subsidised); hermetic storage bags: 220 KSh per bag. Exchange rate 101.2 KSh = USD 1 (March 2018) Resource requirements: personnel: trainers (specialists in food quality and post‐harvest handling); equipment: plastic sheeting (500 gauge plastic); mobile maize dryer (EasyDry500; www.acdivoca.org/easydry), hermetically seal able storage bags (Purdue Improved Cowpea Storage (PICS) bags) Control Type: standard of care for aflatoxin prevention Description: aflatoxin prevention training by 'training the trainers' approach: at least one farmer from each of the study villages was selected, in consultation with community leaders, to receive training re: the causes and consequences of aflatoxin contamination in maize and recommended practices for aflatoxin prevention Delivery: elected representative farmers trained and asked to pass the information on to other farmers in their village Providers: master trainers Duration of intervention: five months Duration of follow‐up: five months Co‐intervention: NR Cost: standard of care for aflatoxin prevention: cost of master trainers (who facilitated the training of one farmer from each village): approximately USD 1.70 per farmer (two meetings with 25 farmers each per day, transport rental was USD 60 and trainers' wages were USD 25. Compensation of trained farmers (who facilitated training of farmers in their own community) not reported by study authors. Resource requirements: personnel: trainers (specialists in food quality and post‐harvest handling)
Outcomes	Outcomes: Aflatoxin content of maize (ppb) at 3.5 months Post‐harvest practices
Notes	Sponsorship source: Ministry for Foreign Affairs of Finland through the FoodAfrica Programme, UK aid from the British people, the CGIAR Research Program on Agriculture for Nutrition and Health (A4NH) led by the International Food Policy Research Institute, and the Bill and Melinda Gates Foundation Grand Challenge Explorations Award Study title or name (study acronym): Mitigating aflatoxin exposure to improve child growth in Eastern Kenya (MAICE) Author’s name: Vivian Hoffmann; Alexia Pretari Email: v.hoffmann@cgiar.org; alexia.pretari@gmail.com; Declarations of interest: yes. Quotation: "Declarations of interest: None." Trial registration: AEARCTR‐0000105
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation sequence
Allocation concealment (selection bias)	Low risk	Clusters (villages) were simultaneously assigned to intervention or control, therefore, allocation could not be foreseen in advance of enrolment.
Baseline outcome measurements similar	Low risk	Study did not report any relevant review outcomes, and therefore, this domain does not apply.
Baseline characteristics similar	Low risk	More households in control group had heard of aflatoxin prior to the study, compared to intervention households (P = 0.01). However, study authors controlled for baseline characteristics in their analysis.
Incomplete outcome data (attrition bias) All outcomes	High risk	High total attrition. Attrition was a bit higher in the intervention group (23.4%), compared to the control group (17.3%). Specifically, for the measurement of aflatoxin in stored home‐produced maize, data were collected from only 171 households (no. per group not reported; attrition of 74.8%).
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel not reported.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Unclear whether the assessors who conducted the follow‐up survey were blinded. While maize contamination assays comparing intervention and control were objective, self‐reported post‐harvest practices were not.
Protection against contamination	Low risk	Intervention and control villages were approximately 4 km apart. Study staff held meetings with eligible households in each cluster. Access to mobile maize dryer in intervention clusters restricted to households that could produce a voucher.
Selective reporting (reporting bias)	Low risk	The study protocol is available at www.socialscienceregistry.org/trials/105, and all prespecified outcomes were reported.
Other bias	Unclear risk	Recruitment bias (cluster‐RCT): randomisation appears to have followed recruitment as the train‐the‐trainer approach was conducted before allocation to trial arms of interest for this study: low risk. Incorrect analysis (cluster‐RCT): low risk; standard errors were bootstrapped to correct for clustering at the village level. Other bias: unclear risk. It is unclear how the study authors selected the 15 control clusters in this study from the 25 control clusters that were generated by the randomisation sequence.