Brugh 2018.

Study characteristics
Methods	Study design: cRCT Study grouping: parallel How were missing data handled? 162 HHs (4.6%) LTFU and 70 HHs with missing data were excluded from the analysis. Randomisation ratio: 1:1 Recruitment method: NR Sample size justification and outcome used: evaluation team calculated number of VCs that would need to be visited based on the number of total eligible. HHs in VCs. Additionally, the statistical power of the study was based on having a minimum number of VCs included. There needed to be ≥ 29 VCs included in the study. Sampling method: in the first sampling stage, 2 traditional authorities were randomly selected from each of the 2 study districts. The second sampling stage consisted of randomly selecting 14 VCs within the 2 traditional authorities in Mangochi and 15 VCs from Salima for a total of 29 study VCs. The final sampling stage was at the HH level; in Mangochi VCs, where the number of eligible HHs tended to be high, 125 eligible HHs were randomly selected in each of the 14 study VCs. All eligible HHs were selected for interview in Salima study VCs. Study aim or objective: evaluation of the expansion of the Malawi's Social Cash Transfer Program on HH food insecurity and dietary diversity. Study period: June–September 2013 to November 2014–January 2015 Unit of allocation or exposure: VCs
Participants	Baseline characteristics Intervention or exposure Age: HH head, years, mean: 58.8 (SD 19.45); number of members in age group (mean): aged 0–5 years: 0.68 (SD 0.91); 6–11 years: 1.17 (SD 1.04); 12–17 years: 0.94 (SD 0.95); 18–64 years: 1.17 (SD 1.02); ≥ 65 years: 0.63 (SD 0.64) Place of residence: Salima district, mean: 0.36 (SD 0.47) Sex: female‐headed HHs, mean: 0.83 (SD 0.37) Ethnicity and language: NR Occupation: engaged in wage employment, %: 4.8; engaged in ganyu labour, %: 55.5; non‐farm enterprise, %: 23.8; selling any crops, %; 21.4 Education: HH head, any schooling, mean: 0.29 (SD 0.45) SES: own any land, %: 90.0; own < 1 acre, %: 25.8; HH size, mean: 4.59 (SD 2.20) Social capital: received from non‐HH members: cash, mean: 0.66 (SD 0.46); food/other consumables, mean: 0.90 (SD 0.29); labour or time, mean: 0.49 (SD 0.49); agricultural inputs, mean: 0.31 (SD 0.45) Nutritional status: food security: worried not enough food, mean: 0.84 (SD 0.36); > 1 meal a day, mean: 0.79 (SD 0.40); proportion of HH expenditure on food, mean: 0.77 (SD 0.11); kcal per capita, mean: 1831.03 (SD 1220.90); food energy deficient, mean: 0.62 (SD 0.48); depth of hunger, mean: 464.10 (SD 491.02); HDDS, mean: 5.63 (SD 1.78). Children aged 0–5 years: stunted, %: 49.8; wasted, %: 4.3; underweight, %: 18.0 Morbidities: HH head: chronically ill, mean: 0.47 (SD 0.49). Children aged 6–59 months: diarrhoea past 2 weeks, %: 16.6; fever past 2 weeks, %: 24.0; cough past 2 weeks, %; 25.8. Adults, aged > 50 years: morbidity, %: 56.0; Concomitant or previous care: participation in other social programmes: food/cash programme, mean: 0.15 (SD 0.35); mother/child feeding programme, mean: 0.15 (SD 0.35) Control Age: HH head, years, mean: 56.86 (SD 19.68). Number of members in age group: 0–5 years, mean: 0.68 (SD 0.90); 6–11 years, mean: 1.23 (SD 1.12); 12–17 years, mean: 0.93 (SD 0.97); 18–64 years, mean: 1.18 (SD 1.02); ≥ 65 years and older, mean: 0.56 (SD 0.65) Place of residence: Salima district, mean: 0.41 (SD 0.50) Sex: female‐headed HHs, mean: 0.85 (SD 0.36) Ethnicity and language: NR Occupation: engaged in wage employment, %: 6.2; Engaged in ganyu labour, % 58.5; non‐farm enterprise, %: 22; selling any crops, %: 24.05 Education: HH head, any schooling, mean: 0.30 (SD 0.46) SES: own any land, %: 89.6; own < 1 acre, %: 23.7; HH size, mean: 4.58 (SD 2.28) Social capital: received from non‐HH members: cash, mean: 0.71 (SD 0.46); food/other consumables, mean: 0.94 (SD 0.24); labour or time, mean: 0.55 (SD 0.51); agricultural inputs, mean: 0.34 (SD 0.48) Nutritional status: food security: worried not enough food, mean: 0.83 (SD 0.38); > 1 meal a day, mean: 0.82 (SD 0.39); proportion of HH expenditure on food, mean: 0.77 (SD 0.11); kcal per capita, mean: 1894.32 (SD 1224.05); food energy deficient, mean: 0.60 (SD 0.50); depth of hunger, mean: 420.75 (SD 490.88); HDDS, mean: 5.64 (SD 1.87). Children aged 0–5 years: stunted, %: 45.5; wasted, %: 3.5; underweight, % 17.3 Morbidities: HH head, chronically ill, mean: 0.41 (SD 0.50). Children aged 6–59 months: diarrhoea past 2 weeks, %: 16.5; fever past 2 weeks, %: 28.2; cough past 2 weeks, %; 26.2. Adults, aged > 50 years: morbidity, %: 50.2 Concomitant or previous care: participation in other social programmes: food/cash programme, mean: 0.20 (SD 0.41); mother/child feeding programme, mean: 0.16 (SD 0.37) Overall Age: NR Place of residence: NR Sex: NR Ethnicity and language: NR Occupation: engaged in wage employment, %: 5.5; engaged in ganyu labour, % 57.0; non‐farm enterprise, %: 23.1; selling any crops, %; 22.7 Education: NR SES: own any land, %: 89.8; own < 1 acre, %: 24.7 Social capital: NR Nutritional status: children aged 0–5 years: stunted, %: 47.9; wasted, %: 3.9; underweight, % 17.6 Morbidities: children aged 6–59 months: diarrhoea past 2 weeks, %: 16.6; fever past 2 weeks, %: 26.1; cough past 2 weeks, %; 26.0. Adults, aged > 50 years: morbidity, %: 53.2 Concomitant or previous care: NR Inclusion criteria: HHs in Mangochi and Salima districts that were either ultra‐poor (unable to meet the most basic urgent needs, including food and essential non‐food items, e.g. soap and clothing) or labour‐constrained (HH had no 'fit to work' members or the ratio of 'unfit' to 'fit' > 3; HH members were 'unfit if aged < 18 years or > 64 years, or if they aged 18–64 but had chronic illness, disability or are otherwise unable to work). Beneficiary selection through a community‐based approach with oversight provided by the local District Commissioner's Office and the District Social Welfare Office. Exclusion criteria: NR Pretreatment: no baseline differences. Attrition per relevant group: intervention group: 70 HHs LTFU (4.2%); missing outcome data (2.8%). Control group: 92 HHs LTFU (4.9%); missing outcome data 32 HHs (1.7%). HHs that were LTFU did not differ from HHs that remained in the study (analysis reported by Abdoulayi et al. 2015) Description of subgroups measured and reported: level of HH's transfer share (low vs high: > 30%; 20–30%; 15–20%; ≤ 15%) Total number completed and analysed per relevant group: intervention group: 1561 HHs; control group: 1729 HHs; total 3290 Total number enrolled per relevant group: intervention group: 1678 HHs; control group: 1853 HHs; total: 3511 Total number randomised per relevant group: intervention group: 14 VCs; control group: 15 VCs
Interventions	Intervention Food access intervention category: increase buying power Intervention type: UCT Description: transfer amount based on HH size and number of children enrolled in primary and secondary school. A single‐person HH received MWK 1000, a 2‐person HH received MWK 1500, a 3‐member HH received MWK 1950, and HHs with ≥ 4 members received MWK 2400. HHs received an additional MWK 300 for each member aged ≤ 21 years enrolled in primary school and MWK 600 for members aged ≤ 30 enrolled in secondary school. Duration of intervention period: 1 year Frequency: monthly. Since the first payments only started March/April 2014, bi‐monthly payments were made in an attempt to interpret the study results as 1‐year impact of the intervention. Number of study contacts: baseline (June–September 2013); follow‐up (November 2014–January 2015) Providers: administered by the Malawi Ministry of Gender, Children, and Social Welfare with additional oversight provided by the Ministry of Economic Planning and Development and technical support from UNICEF Malawi (Abdoulayi et al. 2015). Delivery: NR Co‐interventions: NR Resource requirements: NR Economic indicators: NR Control: no intervention for the duration of the study; implementation of the cash transfers in the control group was delayed until study was completed.
Outcomes	Proportion of HH expenditure on food Food security: worried not enough food; having > 1 meal per day Dietary diversity: HDDS Adequacy of dietary intake: proportion food energy deficient; depth of hunger
Identification	Sponsorship source: The Malawi Social Cash Transfer Program Impact Evaluation was contracted to the University of North Carolina at Chapel Hill and the Center for Social Research at the University of Malawi. The baseline and first follow‐up (midline) were funded by UNICEF, the German Government through KfW, Irish Aid and FAO; the International Initiative for Impact Evaluation (3ie) and the EU provided additional funding for the second follow‐up (endline) survey. The Government of Malawi provided significant in‐kind contributions and support to all 3 rounds. Country: Malawi Setting: ultra‐poor and labour constraint HHs, Mangochi and Salima districts Comments: no trial registry number Authors' names: Kristen Brugh, Gustavo Angeles, Peter Mvula Email: knbrugh@gmail.com; gustavo_angeles@unc.edu; petermvula58@yahoo.com Declarations of interest: NR Study or programme name and acronym: Malawi Social Cash Transfer Program Type of record: journal article, study reports
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Low risk	Coin tossing was conducted for each TA in each district whereby half of the VCs were allocated to the intervention or control group (baseline report p. 9; Appendix C1).
Allocation concealment (Selection bias)	Low risk	VCs in each of selected traditional authorities were selected (from a hat) and an ordered list of clusters was created. At 2 district meetings, the allocation of VCs to the intervention or control was conducted for each TA, according to the ordered list (baseline report page 7, 9, Appendix C1).
Baseline characteristics similar (Selection bias)	Low risk	HH baseline characteristics similar in both groups.
Baseline outcome measurements similar (Selection bias)	Low risk	HH outcome measurements (e.g. measures of food security and dietary diversity) were similar in both groups at baseline.
Blinding of participants and personnel (Performance bias)	Low risk	Blinding of HHs or personnel was not possible; however, performance bias was unlikely as the intervention was implemented by government structures.
Blinding of outcome assessment (Detection bias)	High risk	Not stated whether the field teams collecting the outcome data at the end of the study were blinded. However, measurements of HH food security and dietary diversity were self‐reported.
Protection against contamination (Performance bias)	Low risk	The geographical location of the VCs selected for the intervention or control groups were in the same district. However, a geographical map of the study villages suggest that the majority of the intervention villages were not close to any of the control villages (see Abdoulayi 2014 Baseline report).
Incomplete outcome data (Attrition bias)	Low risk	Low attrition due to LTFU and missing outcome data (intervention group: 117/1678 (7%) vs control group: 124/1853 (6.7%). No evidence of differential attrition between groups.
Selective outcome reporting (Reporting bias)	Unclear risk	Study protocol N/A. Although the cash transfer programme had several aims, the paper by Burgh reported on the outcome domains of food security and dietary diversity. The study authors described 7 outcomes in their methods (e.g. consuming > 1 meal a day, per capita caloric availability, HDDS) and they provided outcome data for all of these variables.
Other bias	Unclear risk	Misclassification of exposure: low risk. Measurement bias: unclear risk. Detailed description of measurements NR. Incorrect analysis: low risk. Effects adjusted for clustering.