Olney 2016.

Study characteristics
Methods	Study design: cRCT Study grouping: parallel group How were missing data handled? analytic sample was restricted to HHs or mothers with complete data at baseline and endline for a given indicator. To address possible attrition bias, attrition weights were calculated and applied to the sample descriptive statistics and impact estimates." Randomisation ratio: approximately 1:1 (initially 25:15:15 for control, intervention led by OWLs, and intervention led by HC members; intervention data were combined as no differences were observed for OWL‐led and HC‐led with regards to outcomes, and to increase statistical power) Recruitment method: HHs were invited to participate. Trained field workers explained the study to eligible HHs, and informed consent was obtained from either the HH head or the mother of the selected child. Sample size justification and outcome used: the study was restricted to 55 villages that met minimum programme eligibility criteria in the 4 departments because of logistics and cost. We estimated a sample size of 30 children per cluster with statistical significance of 0.05, power of 0.80 and an intracluster correlation of 0.02. This sample size provided the ability to detect minimum differences between treatment groups of a change of 0.25 in mean HAZ and WHZ and a 0.3 g/dL change in Hb. The sample also permitted the estimation of changes of 10 pps in the prevalence of wasting, anaemia and diarrhoea and 15 pp in stunting." Sampling method: before the baseline evaluation, villages within 4 'departments' in the province of Gourma were identified for possible inclusion in the EHFP programme; participating villages needed to have water sources to support production during the dry season. 55/181 eligible villages were identified for randomisation and were stratified by commune/department and village size before randomisation into 1 of 3 groups: 1. control group, which received no interventions from HKI (25 control villages), 2. EHFP programme with BCC led by OWLs (15 OWL villages – nutrition and health education done by OWLs), or 3. EHFP programme with BCC led by HC members (15 HC villages – nutrition and health education done by HC members). Within the selected villages, all HHs with a mother who had a child aged 3–12 months at the time of the baseline survey (February–May 2010) were invited to participate in the impact evaluation. Study aim or objective: to assess the impact of the EHFP programme on child nutritional status and health as primary impact measures, as well as on mothers' nutritional status and empowerment as secondary impact measures. Study period: February 2010 to June 2012. Unit of allocation or exposure: villages
Participants	Baseline characteristics Intervention or exposure Age: children aged ≤ 6 years in HH, mean: 2.6 (SD 1.42); mothers, mean, years: 28.1 (SD 6.74); child, mean, months: OWL: 7.14 (SD 2.6); HC: 7.21 (SD 2.71) Place of residence: rural area in Burkina Faso; mean Housing Quality Index Score: –0.1 (SD 1.27) Sex: female, %: 100 (Olney 2016). Boys, n (%): OWL: 217 (49), HC: 218 (50.5) Ethnicity and language: NR Occupation: NR Education: HH head had any formal education, n (%): 97 (11); OWL: 49 (11.3); HC: 42 (9.9). Woman/mother had any formal education, n (%): 62 (7); OWL: 26 (5.9); HC (31 (7.1) SES: HH size, mean, n: 7.5 (SD 3.59); OWL: 7.78 (SD 3.64); HC: 7.24 (SD 3.52). Housing quality index factor score, mean: 20.1 (SD 1.27). Dirt floor in primary house, n (%): OWL: 273 (61.6), HC: 252 (57.1); children aged 6 years, mean, n: 2.6 (SD 1.42); mean asset value for men: XOF 74,312 (SD 61,035); mean asset value for women: XOF 42,225 (SD 64,193); female‐headed HHs, n (%): 62 (7) Social capital: social support score for women, mean: 3.2 (SD 1.37) Nutritional status: HHs experiencing food shock in the last 12 months, mean, n: 0.4 (SD 0.63); mother's BMI, mean: 20.2 (SD 2.22); underweight mothers, n (%): 203 (23); DDS for mothers, mean: 1.8 (SD 1.09); DDS for HHs, mean: 5.6 (SD 1.93) Morbidities: NR Concomitant or previous care: NR Control Age: children aged ≤ 6 years in HH, mean: 2.7 (SD 1.53); mothers, mean, years: 28.3 (SD 6.86); child's age, mean, months: 7.4 (SD 2.64) Place of residence: rural area in Burkina Faso; mean housing quality index score: 0.1 (SD 1.26) Sex: female, %: 100 (Olney 2016). Boys, n (%): 292 (50.6) Ethnicity and language: NR Occupation: NR Education: HH heads with formal education, n (%): 58 (10); women with formal education, n (%): 42 (7) SES: HH size, mean: 8.0 (SD 3.82). Housing quality index factor score, mean: 0.1 (SD 1.26). Dirt floor in primary house, n (%): 213 (35.7); mean number of children aged 6 years: 2.7 (SD 1.53); mean asset value for men: XOF 72,689 (SD 54,694); mean asset value for women: XOF 44,294 (SD 36,923); number of female‐headed HHs (%): 42 (7) Social capital: social support score for women, mean: 3.0 (SD 1.44) Nutritional status: HHs experiencing food shock in the last 12 months, mean, n: 0.3 (SD 0.58); mother's BMI, mean: 20.6 (SD 2.27); underweight mothers, n (%): 90 (15); DDS for mothers, mean: 1.8 (SD 1.07); DDS for HHs, mean: 5.8 (SD 1.70) Morbidities: NR Concomitant or previous care: NR Overall Age: child's age, mean, months: 7.26 (SD 2.65) Place of residence: rural area in Burkina Faso Sex: boys, n (%): 727 (50.1) Ethnicity and language: NR Occupation: NR Education: HH head had any formal education, n (%): 148 (10.2). Women/mother had any formal education, n (%): 96 (6.6) SES: HH size, mean: 7.72 (SD 3.69). Dirt floor in primary house, n (%): 148 (10.2) Social capital: NR Nutritional status: NR Morbidities: NR Concomitant or previous care: NR Inclusion criteria: villages located in the 4 selected departments, had access to water in the dry season to enable participation in the agricultural intervention, and met the population size criteria (≤ 4000 inhabitants) (Olney 2015). Mothers with a child aged 3–12 months. Exclusion criteria: NR. Pretreatment: mothers' BMI was lower in the treatment group compared to control, and more underweight mothers (BMI 18.5 kg/m2) were present in the treatment group. Attrition per relevant group: total attrition 16% of the original sample. Attrition: control villages: 19% (113/597) for HHs and 29% (148/510) for mothers; intervention villages: 14% (124/884) for HHs and 22% (173/787) for mothers. Description of subgroups measured and reported: NR. Total number completed and analysed per relevant group: total 1481 completed the endline survey; intervention: 884 HHs (441 in HC group and 443 in OWL group). 376 children (HC group) and 395 children (OWL group) with complete observations. Control: 597 HHs. 511 children with complete observations. Mothers' BMI: control 510, intervention 787; mothers' consumption of individual food groups and mean DDS: control 506, intervention 766; HH consumption of individual food groups and mean DDS: control 596, intervention 880; women's empowerment: control 517, intervention 781. Total number enrolled per relevant group: total: 1767 HHs. Intervention: 15 villages in HC group and 15 villages in OWL group. 514 HHs in HC group and 512 HHs in OWL group. Control: 25 villages, 741 HHs Total number randomised per relevant group: control: 25 villages; intervention: 30 villages (15 OWL villages and 15 HC villages analysed together in Olney 2015). Total group: 1767 HHs with intervention HHs 514 in HC group and 512 in OWL group. Control HHs were 741.
Interventions	Intervention characteristics Intervention or exposure group: integrated agriculture and nutrition programme Food access intervention category: increase buying power Intervention type: income generation. Description: integrated agriculture and nutrition programme aimed at improving maternal nutrition outcomes through increased production and consumption of nutrient‐rich food. Set of agricultural production and nutrition interventions targeted to mothers with children aged 3–12 months at baseline. Agricultural production interventions centred on dedicating land to women's production during the secondary agricultural season and distributing inputs and training to increase production and consumption of nutrient‐rich foods and to generate additional income (and control over that income) through the sale of surplus production. Programme beneficiaries were provided with saplings, cuttings and seeds of nutrient‐rich fruits (e.g. mangoes and papayas) and vegetables (e.g. orange‐flesh sweet potatoes, dark green leafy vegetables and carrots) and small gardening tools (e.g. hoes, shovels and watering cans). Beneficiaries were also given chicks to increase production of animal source foods (i.e. eggs and meat from the chickens). In addition, they received training in optimal agriculture and poultry‐raising practices to help them establish their homestead food production activities. Nutrition intervention: used a BCC strategy known as the Essential Nutrition Actions framework, which focuses on 7 primary health and nutrition behaviours. Twice a month, all beneficiary mothers were visited in groups or individually by 1 of 2 different types of community volunteers trained by the programme; either OWLs or HC members. This component aimed to improve mothers' own health and nutrition by enabling them to adopt optimal health and nutrition practices for themselves and their young children. Duration of intervention period: NR Frequency: frequency of distribution of productive assets and training regarding agricultural intervention NR. For the nutrition intervention, beneficiary mothers were visited twice a month by community volunteers (either OWLs or HC members). Number of study contacts: 2; baseline (February–May 2010); and 2‐year follow‐up (February–June 2012). Providers: NPO (HKI) Delivery: agricultural interventions first rolled out to female village farm leaders, who in turn trained other mothers in their communities. Primary health and nutrition education was through BCC led by OWLs or HC members. Co‐interventions: NR but authors mentioned in discussion that (quote) "it is possible that other unmeasured factors also contributed to reducing underweight such as changes in use of health care services, morbidity, or workload." Resource requirements: NR Economic indicators: NR Control group: no intervention
Outcomes	Dietary diversity: HDDI, MDD; proportion of mothers consuming individual food groups in past 7 days Dietary intake: energy gap Anthropometry: adult BMI; prevalence of underweight among adults
Identification	Sponsorship source: USAID, Office of US Foreign Disaster Assistance through HKI; European Commission/International Fund for Agricultural Development; the Gender, Agriculture and Assets Project, supported by the Bill Melinda Gates Foundation; the CGIAR Research Program on Agriculture for Nutrition and Health led by the IFPRI; and the USDA (AD). Country: Burkina Faso Setting: homesteads in agricultural areas Comments: trial registration: NCT01825226 Author's name: Deanna K Olney Email: d.olney@cgiar.org Declarations of interest: yes; 5 authors had no conflicts of interest. 1 author (A Pedehombga) worked for HKI. Study or programme name and acronym: Helen Keller International (HKI) enhanced‐homestead food production (EHFP) programme Type of record: journal article
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Unclear risk	Stratified randomisation was performed (by commune and village size), but no information provided on how the randomisation sequence was generated.
Allocation concealment (Selection bias)	Low risk	Allocation to intervention groups was done at the same time for all clusters/villages.
Baseline characteristics similar (Selection bias)	Low risk	All baseline characteristics (with the exception of BMI outcomes) were similar across groups. Many of these were also adjusted for in the analysis.
Baseline outcome measurements similar (Selection bias)	Low risk	Prevalence of diarrhoea was lowest in control group, and, for mothers, BMI and prevalence of underweight was lower in control group. However, these were adjusted for in the analysis.
Blinding of participants and personnel (Performance bias)	Low risk	Blinding not possible, but this is unlikely to have influenced participants' or staff's behaviour or experience.
Blinding of outcome assessment (Detection bias)	High risk	Unclear who collected outcome data. Some outcomes were based on women's self‐reports, which could have been influenced by lack of blinding, e.g. the self‐reported consumption by food group.
Protection against contamination (Performance bias)	Low risk	Allocation was by village and it was unlikely that the control group received the intervention.
Incomplete outcome data (Attrition bias)	Unclear risk	Overall attrition was fairly low (16%); however, differential attrition across groups resulted in significantly higher attrition among control HHs and mothers. This problem was addressed by including attrition weights in analyses. Differential attrition also occurred by HH size and composition, the presence of polygamy, female‐headed HHs and female formal education. Furthermore, women with a higher BMI were more likely to attrite resulting in an attenuation of effect size.
Selective outcome reporting (Reporting bias)	Low risk	No protocol available. All outcomes in the trial registry reported in publications. Some stated measures such as mothers' health and knowledge of nutrition and hygiene were NR.
Other bias	Unclear risk	Misclassification bias: unlikely. Measurement bias: unlikely. Incorrect analysis: low risk. Analyses adjusted for clustering, it is unclear whether these findings were comparable with an individually randomised trial. Recruitment bias: high risk. Randomisation was performed before recruitment was done, significant baseline imbalance was present for important outcomes of interest, there was a considerable risk of bias from loss of clusters as 1 intervention cluster attrited before follow‐up.