Hjelm 2017.

Study characteristics
Methods	Study design: cRCT Study grouping: parallel group How were missing data handled? subsample of HHs with female carers. No systematic difference in treatment group, age, education or marital status between women who responded and those with missing values. Due to their relatively small number, these observations were dropped from the analysis. Randomisation ratio: 1:1 Recruitment method: CWACs were randomly selected from 2 districts by the Zambian Ministry of Community Development Mother and Child Health. Each CWAC identified eligible HHs meeting ≥ 1 inclusion criterion, and 33 HHs were approached from these. Sample size justification and outcome used: reported that a power analysis was conducted to determine a sample size large enough to detect meaningful effects, also among subgroups. Outcomes used in their calculation NR. Sampling method: total sample: 46 CWACs out of approximately 100 CWACs from each district (Luwinga, Serenje districts) were included through a lottery held at Ministry headquarters in June 2010. Thereafter, 33 HHs per CWAC were randomly selected (out of approximately 100 eligible HHs per CWAC), resulting in 3077 HHs (15,630 people). Subsample: 2490 HHs with female carers were included in the secondary analysis of the outcome perceived stress. Study aim or objective: impact evaluation of programme related to changes in 5 primary areas: income, education, health, food security and livelihoods. Study period: December 2011 to December 2014 Unit of allocation or exposure: communities through CWACs
Participants	Baseline characteristics Intervention or exposure Age: total sample: children, mean, years: 14.88 (SD 1.50); subsample of HHs with female carer: carer mean, years: 51.98 Place of residence: HH distance to food market, mean: 34.15 (SD 31.77); HH distance to health facility mean: 13.02 (SD 17.55) Sex: adolescents in HH, dichotomised to male = 1, female = 0, mean: 0.54 (SD 0.50) Ethnicity and language: NR Occupation: NR Education: subsample of HHs with female carers: ever attended school, proportion: 0.60; highest grade completed, mean: 2.98 SES: total sample: HH size, mean: 4.98 (SD 2.47); total HH expenditure per person per month, mean (SD): ZMW 50,832.42 (47,438.87); subsample of HHs with female carer: HH size, mean: 5.14; children aged 0–5 years, proportion, n: 0.77; assets owned (clock, watch, mobile phone, radio, sofa, table, mattress), proportion: 0.51 Social capital: NR Nutritional status: total sample: HFIAS, mean: 14.78 (SD 5.49); expenditure on food per person per month, mean: ZMW 38,641.75 (SD 36,237.80); share of total expenditure on food per capita, mean: ZMW 0.74 (SD 0.16); subsample of HHs with female carer: HFIAS, mean: 14.75 Morbidities: NR Concomitant or previous care: total sample: whether HH received a food security pack, mean: 0.01 (SD 0.09) Control Age: total sample: children, mean, years: 14.86 (SD 1.44); subsample of HHs with female carer: age of carer, mean, years: 51.26 Place of residence: HH distance to food market, mean: 27.51 (SD 30.67); HH distance to health facility mean: 11.91 (SD 15.55) Sex: adolescents in HH, dichotomised to male = 1 female = 0, mean: 0.53 (SD 0.50) Ethnicity and language: NR Occupation: NR Education: subsample of HHs with female carers: ever attended school, proportion: 0.63; highest grade completed, mean: 3.09 SES: total sample: HH size, mean: 5.01 (SD 2.50); total HH expenditure per person per month, mean: ZMW 51,843.45 (SD 42,876.01); subsample of HHs with female carer: HH size, mean: 5.18; people aged 0–5 years, proportion, n: 0.73; assets owned (clock, watch, mobile phone, radio, sofa, table, mattress), proportion: 0.58 Social capital: NR Nutritional status: total sample: HFIAS, mean: 14.68 (SD 5.71); expenditure on food per person per month, mean: ZMW 40,367.87 (SD 35,290.74); share of total expenditure on food per capita, mean: ZMW 0.77 (SD 0.15); subsample of HHs with female carer: HFIAS, mean: 14.61 Morbidities: NR Concomitant or previous care: total sample: whether HH received a food security pack, mean: 0.01 (SD 0.10) Overall Age: subsample of HHs with female carer: years, mean: 51.62 Place of residence: NR Sex: NR Ethnicity and language: NR Occupation: NR Education: subsample of HHs with female carers: ever attended school, proportion: 0.61; highest grade completed, mean: 3.03 SES: total sample: per capita share of expenditure on food: 0.754; subsample of HHs with female carer: HH size, mean: 5.16; people aged 0–5 years, proportion, n: 0.75; assets owned (0–7), proportion: 0.54 Social capital: total sample: any NGOs operating in community, % (n/N): 32.6 (30/92) Nutritional status: subsample of HHs with female carer: HFIAS, mean: 14.68 Morbidities: CES‐D short form mean score, for adolescents only: 17.89; mean proportion of adolescents depressed (CES‐D score ≥ 20): 0.33 Concomitant or previous care: NR Inclusion criteria: HHs that are female‐headed and caring for orphans, had a disabled member, were elderly headed (> 60 years) and caring for orphans, or are special cases (cases that are critical, but did not qualify under other categories; e.g. 2 elderly people unable to care for themselves). Exclusion criteria: NR Pretreatment: no differences reported between the groups in the total sample (3077 HHs), or those in the subsample of HHs with female carers (2490 HHs). Attrition per relevant group: overall attrition was 106/3076 (3.4%) HHs. Per‐group attrition was 70/1561 (4.5%) HHs (4.2% in Serenje district and 4.9% in Luwingu district) in the intervention, and 35/1515 (2.3%) (2.3% in Serenje district and 2.4% in Luwingu district) in the control group. Description of subgroups measured and reported: HHs caring for orphans, female‐headed HHs and HHs with adolescents. Total number completed and analysed per relevant group: 2970/3076 (96.6%) HHs overall, with 1490/1561 (95.5%) in intervention group and 1480/1515 (97.7%) in control group. Total number enrolled per relevant group: 3076 HHs; 1561 in intervention group and 1515 in control group. Subsample of HHs with female carers: NR. Total number randomised per relevant group: 3076 HHs in 92 CWACs; 1561 from 46 CWACs in intervention group and 1515 from 46 CWACs in control group. Subsample of HHs with female carers: NR.
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power Intervention type: UCT Description: monthly transfer of ZMW 55,000 (USD 11) irrespective of size of HH. This amount was chosen as it is considered sufficient to provide 1 meal a day for each HH member over the course of 1 month. Duration of intervention period: 36 months (December 2011 to December 2014); for as long as criteria were met. Frequency: monthly transfer Number of study contacts: 3 contacts: baseline: November/December 2011, follow‐up: November/December 2013 and final follow‐up: November/December 2014. Subsample of HHs with female carer: baseline: November/December 2011 and follow‐up: November/December 2014. Providers: Zambian government: Ministry of Community Development, Mother and Child Health (MCDMCH). Delivery: payments were made every other month through a local paypoint manager. Co‐interventions: during the baseline survey HHs were questioned regarding receipt of a food security pack; unclear whether this was a potential co‐intervention. Resource requirements: NR Economic indicators: NR Control: no intervention Co‐interventions: during the baseline survey HHs were questioned regarding receipt of a food security pack; unclear whether this was a potential co‐intervention.
Outcomes	Proportion of per capita expenditure spent on buying food Food security: HFIAS Anxiety/depression: Cohen PSS; CES‐D short form; depression (based on cut‐off value for the CES‐D)
Identification	Sponsorship source: consortium of donors including DfID, UNICEF, Irish Aid, and the Government of Finland. Palermo, Handa, and Hjelm received additional funding from the Swedish International Development Cooperation Agency (G41102) to the UNICEF Office of Research – Innocenti for analysis of the data and drafting of the manuscript. Country: Zambia Setting: socially vulnerable HHs in 2 rural districts with extreme poverty (Luwinga, Serenje) Author's name: Lisa Hjelm Email: lhjelm@unicef.org Declarations of interest: no Study or programme name and acronym: Zambia Multiple Category Cash Transfer Program (MCP) Type of record: journal article
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Unclear risk	Randomisation by coin toss with local officials, ministry staff and community members present. Randomisation appeared to have occurred by a single coin toss, allocating one half of the list of CWACs to intervention or control – unclear how this may have biased the process.
Allocation concealment (Selection bias)	Unclear risk	NR whether the allocation outcome of the randomisation process (a single coin toss) was protected from the person performing the randomisation (the Ministry's permanent secretary) beforehand.
Baseline characteristics similar (Selection bias)	Low risk	HH characteristics were similar for both groups at baseline in the total sample, as well as in the subsample of HHs with female carers.
Baseline outcome measurements similar (Selection bias)	Unclear risk	Total group: balanced at baseline with no statistically significant differences for outcome measures. Subsample of HHs with female carers: indices of HH food insecurity were similar at baseline; however, parameters of perceived stress were only measured at the end of the study period.
Blinding of participants and personnel (Performance bias)	Low risk	Participants and study staff do not appear to have been blinded to assignment, but it is unlikely that this influenced the intervention received.
Blinding of outcome assessment (Detection bias)	High risk	Participants did not appear to have been blinded to assignment and were the outcome assessors as outcomes were self‐reported.
Protection against contamination (Performance bias)	Low risk	Serenje and Luwingu are both large, geographically discrete districts. In addition, communities were the unit of randomisation. It is unlikely that contamination would present considerable bias.
Incomplete outcome data (Attrition bias)	Unclear risk	Missing values were dropped from the analysis, but there was low overall attrition of 3.4% (with 4.5% in the intervention and 2.3% in the control group); with no evidence of selective attrition. Attrition in HHs with female carers, however, was not described.
Selective outcome reporting (Reporting bias)	Unclear risk	Outcomes for the overall evaluation were not clearly stated, but all outcomes appeared to have been addressed in the evaluation report.
Other bias	Low risk	Recruitment bias, low risk: randomisation followed recruitment and baseline survey. Incorrect analysis, low risk: OLS regression with cluster robust SEs were used to account for the clustered nature of the data. Loss of clusters, low risk: no loss of clusters reported. Seasonality bias, low risk: unlikely as the baseline and follow‐up data were conducted at the same time of the year.