Osei 2017.

Study characteristics
Methods	Study design: cRCT Study grouping: parallel group How were missing data handled? Missing or implausible outcome data were reported in the article and were excluded from the analysis involving the respective outcomes variables (Osei 2017). Attrition not relevant as study data were collected through independent cross‐sectional surveys at baseline (2009) and follow‐up (2012). Therefore, data on the impact of the intervention were not gathered from the same individuals in the 2 surveys. Randomisation ratio: 1:1 (subdistrict level) Recruitment method: recruitment to intervention group: not described. 1 woman per group of intervention villages (about 5 or 6) was selected and trained by HKI in improved gardening and poultry‐rearing practices in preparation to host a VMF. Under the guidance of HKI, this woman then trained 20 other beneficiary women on how to establish their home gardens and poultry production (Osei 2017). Sample size justification and outcome used: calculated to detect an assumed difference in the prevalence of stunting and underweight of 10% as well as a difference in wasting of 5% among children in the treatment compared to the control communities. Using a power of 0.80, a 95% CI (2 sided), an assumed design effect of 2 and an upward adjustment of 10% to account for LTFU, 1970 children were considered adequate for each of the pre‐ and post‐treatment surveys. The baseline sample was increased to 2106 to allow enough participants for a substudy, which involved providing micronutrient powders to a subsample of 110 children in the intervention communities for 6 months. The follow‐up sample was also increased to 2614 to allow sufficient participants for disaggregated analysis (Osei 2017) Sampling method: multistage cRCT. The Baitadi district is administratively divided into 12 subdistricts called 'Ilakas', and each 'Ilaka' is further divided into 'village development committees' (VDCs), which consist of several (about 9) villages. (Osei 2015). Assigned Ilakas (instead of villages or HHs) into each of the 2 study groups. Ilakas were paired on several key socioeconomic indicators, and 4 of the generated 6 pairs of Ilakas were selected for inclusion in study using a simple random sampling procedure. The same procedure was followed to assign 1 Ilaka in each pair as the treatment group and the other as the control. Overall, there were 21 VDCs in the treatment Ilaka and 20 in the control Ilakas (Osei 2017). All the VDCs in the EHFP programme communities received the EHFP intervention. To select families for the pre‐ and postsurveys, VDCs were stratified by treatment (21) or control group (20), and 14 VDCs were selected from each using a simple random sampling procedure to participate in the surveys. However, all the VDCs in the EHFP programme communities received the EHFP intervention, regardless of their participation in the surveys. Within each selected VDC, HHs were selected using the probability proportional to size technique, and in each selected HH, a child aged 12–48 months was chosen together with his/her mother for the assessments. If a HH had > 1 eligible child, the youngest child was chosen (Osei 2017). Study aim or objective: to determine the effect of an EHFP programme consisting of home garden, poultry raising and nutrition education implemented over 2.5 years vs control (no intervention) on anthropometry and anaemia among children (aged 12–48 months) and their mothers (Osei 2017). Study period: 2.5 years from 2009 to 2012. The baseline survey was conducted in August 2009, the follow‐up survey in August/September 2012. Unit of allocation or exposure: subdistricts (Ilakas)
Participants	Baseline characteristics Intervention or exposure Age: mother, mean, years: 27.3 (SD 5.8); child mean, months: 28.3 (SD 10.0); Child aged < 24 months, %: 35.2. Place of residence: NR Sex: child female, %: 46.6. Male head of HH, %: 77.3 Ethnicity and language: upper caste, %: 77.4 Occupation: mother work outside the home, %: 10.1 Education: mother education: none, %: 57.3 SES: Wealth terciles, %: lower 25.5, middle 28.2, upper 46.3. No cash income in last month, %: 46.7. Own land, %: 99.0. HH size, mean: 7.2 (SD 4.0). Mother parity, mean: 2.0 (SD 2.0). Mother ≥ 3 children, %: 48.1. Mother married, %: 98.9. Mother main carer of child, %: 48.8. In debt, %: 76.5 Social capital: NR Nutritional status: food insecure HHs, total, %: 79.7 (77.2–82.0) Morbidities: NR Concomitant or previous care: NR Control Age: mother, mean, years: 27.5 (SD 5.7); child, mean, months: 27.8 (SD 9.8); child aged < 24 months, %: 36.1 Place of residence: NR Sex: child female, %: 46.1; male head of HH, %: 83.8 Ethnicity and language: upper caste, %: 82.3 Occupation: mother work outside the home, %: 15.0 Education: mother education none, %: 80.3 SES: wealth terciles, %: lower 42.2, middle 36.7, upper 21.1. No cash income in last month, %: 49.2. Own land, %: 98.9. HH size, mean: 6.0 (SD 3.0). Mother parity, mean: 3.0 (SD 2.0). Mother ≥ 3 children, %: 53.1. Mother married, %: 99.4. Mother main carer of child, %: 70.3. In debt, %: 76.2 Social capital: NR Nutritional status: food insecure HHs, total, %: 87.4 (85.3–89.3) Morbidities: NR Concomitant or previous care: NR Overall: NR Inclusion criteria: families with children aged 0–23 months in 1 group of communities received the EHFP intervention (treatment group); in each selected HH, a child aged 12–48 months was chosen together with his/her mother for the assessments. Exclusion criteria: NR Pretreatment: baseline intervention sample had significantly more pregnant women, lower maternal parity, larger family size and a lower proportion of male‐headed HHs than the control, although these characteristics did not differ between the groups at follow‐up. Children in the intervention group were significantly older than those in the control group at follow‐up. Other characteristics that were similar among the study groups at baseline but differed significantly at follow‐up included the proportion of HHs with no monthly cash income and those with some financial debts at the time of the surveys. Both surveys showed significantly fewer mothers with no formal education, mothers who worked outside the home, and proportion of families with married HH heads, in the upper caste and lower tercile of wealth among the treatment compared to the control group. Attrition per relevant group: attrition not relevant as baseline and follow‐up samples were independent. NR. Authors reported data excluded from analysis. Of the 2106 mother–child pairs in the baseline sample, 1 child and 2 mothers missed Hb measurements; 8 children had implausible HAZ and 4 children had implausible WHZ values, and 1 mother had implausible BMI (< 12.0 kg/m2). Therefore, these participants were excluded from the analysis involving these outcomes. In addition, 10.9% of the mothers (intervention: 100; control: 129) were pregnant at baseline and were excluded from maternal BMI and underweight analysis. For the 2614 mother–child pairs assessed at follow‐up, 18 had implausible HAZ and 11 children had implausible WHZ values, and 9.7% mothers (intervention: 125; control: 128) were pregnant. These participants were excluded from the analysis involving the respective outcome variables. Description of subgroups measured and reported: NR. Total number completed and analysed per relevant group: total number varied depending on the outcome being analysed. General numbers at follow‐up: intervention: 1037 families (1037 mothers and 1037 children); control: 1037 families (1037 mothers and 1037 children). Total number enrolled per relevant group: intervention: 4 subdistricts; 21 VDC; 1055 families (1055 mothers and 1055 children); control: 4 subdistricts; 20 VDC; 1051 families (1051 mothers and 1051 children). Total number randomised per relevant group: intervention: 4 subdistricts; 21 VDC; 1055 families (1055 mothers and 1055 children); control: 4 subdistricts; 20 VDC; 1051 families (1051 mothers and 1051 children).
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power Intervention type: income generation through EHFP programme consisting of home garden, poultry raising and nutrition education. Description: 1 woman per group of intervention villages (about 5 or 6) was selected and trained by HKI in improved gardening and poultry‐rearing practices in preparation to host a VMF. Under the guidance of HKI, this woman then trained 20 other beneficiary women on how to establish their home gardens and poultry production. Altogether, the project intended to establish 120 VMFs, each of which served as a site for purchasing inputs and ongoing training for all the beneficiary women. For every season (rainy and winter) of the first year, each woman was given a one‐off free supply of seeds, saplings and locally bred chicks to establish their home gardens and poultry production. Throughout the intervention, the women met monthly at the VMF to refresh lessons on agriculture techniques and nutrition through social and BCCs, such as optimal infant and young child feeding practices, cooking demonstrations on how to use the EHFP produce to enrich family meals, and participation in routine public health services (immunisation, growth monitoring, vitamin A supplementation and deworming) offered in their communities. The intervention activities were jointly monitored through monthly home visits by trained project staff, female community health volunteers, and agriculture extension officers. During each home visit, the project staff and agriculture extension officers observed the home gardens and poultry, including the number of different varieties of crops cultivated, whether the garden was fenced, how many of poultry birds the family had, whether the poultry had started laying eggs, etc., and provided advice on any garden or poultry issues the family was facing. On these home visits, the project staff and the female community health volunteers also reinforced the educational messages on breastfeeding and complementary feeding to all mothers (Osei 2017). Duration of intervention period: 2.5 years: 2009–2012 Frequency: For every season (rainy and winter) of the first year, each woman was given a one‐off free supply of seeds, saplings and locally bred chicks to establish their home gardens and poultry production. This was followed by monthly home visits by trained project staff, female community health volunteers,and agriculture extension officers, who observed the home gardens and poultry, and provided advice on any garden or poultry issues the family was facing. Monthly meetings at VMF. Number of study contacts: 2: 1 pre‐intervention survey (August 2009) and 1 post‐intervention survey (August–September 2012) Providers: NGO (Hellen Keller International) project staff Delivery: before study started, there were briefings in all the intervention villages (including meetings with village leaders) to gain their commitment and support. The intervention activities were jointly monitored through monthly home visits by trained project staff, female community health volunteers and agriculture extension officers. During each home visit, the project staff and agriculture extension officers observed the home gardens and poultry, and reinforced the educational messages on breastfeeding and complementary feeding to all mothers. Co‐interventions: NR Resource requirements: NR Economic indicators: NR Control: no intervention
Outcomes	Food security: HFIAS Anthropometry: WAZ, WHZ, HAZ, stunting, wasting, underweight, BMI and underweight of mothers Biochemical: mean Hb concentration (mother/child) Morbidity: anaemia (mother/child)
Identification	Sponsorship source: USAID Country: Nepal Setting: homesteads in Baitadi District, a remote hilly community in the far western region of Nepal. Author's name: Akoto Osei Email: andykofi20@gmail.com; oseia@africa‐union.org Address: Department of Social Affairs, African Union Commission, Room 1216, New Building, PO Box 3243, Addis Ababa, Ethiopia Declarations of interest: no potential conflicts of interest. Study or programme name and acronym: 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	Quote: "Simple random sampling procedures' were performed to select both the sub‐district pairs (Ilakas), allocate one sub‐district in each pair to either the intervention and control groups, and to select village development committees from which families were selected for the pre‐and post intervention surveys. However, no information is provided on which 'simple random sampling procedure' was followed. Households that participated in the pre‐ and post intervention surveys were selected using the probability proportional to size technique."
Allocation concealment (Selection bias)	Low risk	Allocation of subdistricts to study groups were performed for all units at the start of the study.
Baseline characteristics similar (Selection bias)	Low risk	Imbalances at baseline and follow‐up for many variables were present, but these were adjusted for in the analysis.
Baseline outcome measurements similar (Selection bias)	Unclear risk	At baseline, treatment HHs were less food insecure, had a higher mean WAZ for children, had a lower prevalence of underweight and stunting in children, and a significantly lower adjusted mean BMI compared to the control HHs. Unclear to what extend these were adjusted for in the analysis.
Blinding of participants and personnel (Performance bias)	Low risk	No blinding was possible but it was unlikely this influenced the intervention received.
Blinding of outcome assessment (Detection bias)	Low risk	Blinding of participants and personnel was not possible due to the nature of the intervention, but it was unlikely that the outcomes were influenced by the lack of blinding. The primary outcome variables (stunting, underweight, wasting, Hb concentration and anaemia among children and Hb concentration, anaemia and underweight among their mothers) were assessed using objective anthropometric and Hb measurements that were taken at baseline and follow‐up.
Protection against contamination (Performance bias)	Unclear risk	Although the randomisation was done at a subdistrict level, which should have minimised contamination, the authors reported an increase in the same activities of the intervention in the control group during the trial.
Incomplete outcome data (Attrition bias)	Unclear risk	Unclear how the micronutrient powder intervention provided in the substudy (Osei 2015) might have affected the study outcomes. It is NR if it was adjusted for in the analysis. Attrition not relevant: study data were collected through independent cross‐sectional surveys at baseline (2009) and follow‐up (2012), so data on the impact of the intervention were not gathered from the same individuals in the 2 surveys. Outcome data from both intervention and control groups were excluded from the analysis, but the number of exclusions for Hb measurements, HAZ and WHZ values, maternal BMI were small and the reasons for them reported. Exclusion of maternal BMI and underweight analysis due to pregnancy in both the baseline and follow‐up surveys were balanced across groups.
Selective outcome reporting (Reporting bias)	Unclear risk	The stated primary outcomes were reported on in the article, and intermediary outcomes are reported in supplemental tables. However, no protocol was available.
Other bias	High risk	Measured association might be biased. Study data were collected through independent cross‐sectional surveys and assessments at baseline (2009) and follow‐up (2012), so data on the impact of the intervention were not gathered from the same individuals in the 2 assessment periods. Therefore, the observed changes in outcomes cannot be directly associated with the intervention.