Tonguet Papucci 2015.

Study characteristics
Methods	Study design: cRCT Study grouping: parallel group How were missing data handled? influence of missing data assessed using a sensitivity analysis using multiple imputation to account for missing values. Children without ≥ 2 measurements or excluded from the analysis. Authors employed ITT analysis. Randomisation ratio: 1:1 (16 control villages and 16 intervention villages) Recruitment method: initial recruitment by approaching a village representative (normally village head and his/her committee) to obtain consent for participation of village following an explanation of aims of research. Individual informed consent then sought from individual HH representatives (usually mothers) following explanation of research aim, sequence of activities and procedures, and risks and benefits of participation. Sample size justification and outcome used: Houngbe: type I error of 5%, a statistical power of 90% and a minimum follow‐up time of 24 months, assuming a 33% reduction in the cumulative incidence of wasting, a coefficient of variation K of 0.25 and an anticipated 25% dropout, 16 clusters with 50 children were required in each study group. Tonguet‐Papucci: to detect a decrease with 33% in the cumulative incidence of wasting assuming a baseline incidence rate of wasting of 0.26 per child‐year with a Type I error of 5%, a statistical power of 90% and a minimum follow‐up of 24 months, assuming a coefficient of variation K of 0.25, 16 clusters of 50 HHs per cluster were necessary per study group. This calculation accounted for an anticipated 25% drop‐out. Sampling method: villages randomly assigned to intervention and control groups during a ceremony to keep the allocation of cash transparent and fair. Representatives of each of 32 villages drew blindly from a bag 1 of the 32 identical papers with 'cash' or 'no cash' written on it. Within villages, HH participation in study was voluntary and based on inclusion criteria. How 32 villages in 3 municipalities were selected was NR. Study aim or objective: Houngbe: "to assess the impact of a cash transfer programme in reducing the incidence of acute malnutrition and morbidity and the prevalence of stunting in children aged 36 months. Study period: June 2013 to October 2015 Unit of allocation or exposure: villages
Participants	Baseline characteristics Intervention or exposure Age: child, mean, months: 6.83 (SD 3.29); children, n (%): aged 6 months: 236 (37.5); aged 6–11 months: 358 (56.8); aged 12–15 months: 36 (5.7) Place of residence: rural area in eastern Burkina Faso Sex: children, n (%): boys 349 (55.4); girls 281 (446) Ethnicity and language: predominantly Gourmanche people, Gulmancema is the predominant local language Occupation: NR Education: NR SES: HHs, n (%): low SES: 288 (45.7); medium SES: 224 (35.6); high SES: 118 (18.7) Social capital: NR Nutritional status: mean: WHZ: –1.24 (SD 1.23); HAZ: –1.18 (SD 1.44); MUAC, mm: 131.3 (SD 12.8); stunted children, n (%) (HAZ2): 175 (27.7) Morbidities: NR Concomitant or previous care: none reported but authors mentioned that a national social protection policy that promoted social transfer mechanisms to the poorest and most vulnerable was adopted in 2012 to enhance food security among the population. Control Age: child, mean, months: 7.79 (SD 2.93); children, n (%): aged 6 months: 161 (26.0); aged 6–11 months: 396 (63.8); aged 12–15 months: 63 (10.2) Place of residence: rural area in eastern Burkina Faso Sex: children, n (%): boys 313 (50.5); girls: 307 (49.5) Ethnicity and language: predominantly Gourmanche people, Gulmancema is the predominant local language Occupation: NR Education: NR SES: Number (%) of HHs: low SES 248 (40.1) middle SES 205 (33.1) high SES 166 (26.8) Social capital: NR Nutritional status: mean: WHZ: –1.07 (SD 1.12); HAZ –1.33 (SD 1.24); MUAC, mm: 133.1 (SD 11.7) mm; stunted children, n (%) (HAZ2): 169 (SD 27.2) Morbidities: NR Concomitant or previous care: none reported but authors mentioned that a national social protection policy that promoted social transfer mechanisms to the poorest and most vulnerable was adopted in 2012 to enhance food security among the population. Overall: NR Inclusion criteria: HHs classified as poor or very poor according to the Household Economy Approach; with ≥ 1 child aged < 1 year at time of inclusion, regardless of nutritional status; children with ≥ 2 follow‐up measurements. Exclusion criteria: NR Pretreatment: overall, baseline characteristics were balanced between groups. Children in intervention group were more likely to be 1 month younger and more wasted than children in the control group. Attrition per relevant group: Houngbe 2017: intervention group: 2.2% (14/644) of children; control group: 2.2% (14/634) of children. Intervention: 14 children LTFU at visit 2 (12 deaths and 2 left study area); excluded from analysis. Additional LTFU: 43 children LTFU between visits 3 and 9 (reasons: 35 deaths; 8 left study area); control: 14 children LTFU at visit 2 (10 deaths, 4 wrongly included); excluded from analysis. Additional LTFU: 28 LTFU between visits 3 and 9 (reasons: 22 deaths; 6 left study area). Description of subgroups measured and reported: NR Total number completed and analysed per relevant group: intervention: 630 children analysed; control: 620 children analysed Total number enrolled per relevant group: Houngbe 2017: intervention: 644 children from 602 HHs; control: 634 children from 583 HHs. Total number randomised per relevant group: total 1278 children from 32 villages randomised; intervention: 644; control: 634.
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power Intervention type: UCT Description: seasonal UCTs provided monthly July–November over 2 years (2013 and 2014). Monthly allowance of XOF 10,000 (USD 17) was given by mobile phone (offered by the project) to participating HHs. Specifically designated mothers were the primary recipients of the transfer because they were usually in charge of child care. Duration of intervention period: 5 months (July–November) in 2013 and 5 months (July–November) in 2014. These months represent the 'lean season' in Burkina Faso. Frequency: monthly Number of study contacts: 9 visits. Baseline data collected 1 month earlier in the intervention group than in the control group in order to enable cash transfer to start on time. Follow‐up visits performed at the same time in the 2 groups. Data collection lasted 29 month (June 2013 to October 2015). Providers: ECHO trained project staff Delivery: a dedicated team supervised and followed up cash transfer activities jointly with the research team. A partnership with a mobile telephone company enabled cash distribution via mobile telephone. Before the intervention, all mothers in the intervention group received an identity card provided by the field teams, a mobile telephone and a subscriber identification module card linked to an electronic account. At the time of distribution, mothers received a text message providing a code and notifying them that their account was credited. Mothers were thus invited to visit cash withdrawal points to collect their money. Presentation of the identity card and the code granted access to the money. Mothers confirmed the cash withdrawal by signing follow‐up lists. All study participants in the intervention group (100%) received their monthly allowance within 1 week. Operational constraints such as mothers' limited knowledge about the use of mobile telephones, difficulty charging the mobile telephones because of the lack of electricity and low literacy rate among mothers were encountered during the delivery of cash. Sessions demonstrating basic uses of a mobile telephone, home visits by cash transfer supervisors to inform HHs about the scheduled dates for cash transfers, switching subscriber identification module cards from 1 telephone to another at cash withdrawal points, and direct transfers in remote villages were mitigation strategies put in place to tackle these difficulties. Co‐interventions: none reported but authors mentioned that a national social protection policy that promotes social transfer mechanisms to the poorest and most vulnerable was adopted in 2012 to enhance food security among the population. Resource requirements: NR Economic indicators: over 1 year, a total of XOF 50,000 (USD 85) was transferred to each eligible HH, representing; 33% of the 2014 national poverty line, estimated at XOF 153,530 (USD 260). Control Food access intervention category: non‐active control Intervention type: N/A Description: mothers of children in control group did not receive a cash grant. Incentives (e.g. a cooking kit, fabrics) were given to HHs in the control group to compensate for the time they spent answering the MAM'Out questionnaires. Duration of intervention period: N/A Frequency: unclear how often compensation was given. Number of study contacts: 9 visits. Baseline data collected 1 month earlier in intervention group than in control group in order to enable cash transfer to start on time. Follow‐up visits performed at the same time in the 2 groups. Data collection lasted 29 months (June 2013 to October 2015). Providers: N/A Delivery: N/A Co‐interventions: 9 visits. Baseline data collected 1 month earlier in intervention group than in control group in order to enable cash transfer to start on time. Follow‐up visits performed at the same time in the 2 groups. Data collection lasted 29 months (June 2013 to October 2015). Resource requirements: N/A Economic indicators: N/A
Outcomes	Dietary diversity: DDS; MDD; minimum acceptable diet; proportion consuming iron‐rich or iron‐fortified foods Anthropometry: incidence of wasting; HAZ; WHZ; incidence of stunting; MUAC Morbidity: incidence diarrhoeal disease; incidence ARIs
Identification	Sponsorship source: Houngbe et al 2017: Action Against Hunger France and the CDC, European Commission's Humanitarian Aid and Civil Protection department, USAID (through the Technical Operational Performance Support programme) and Foundation Action Against Hunger (France) for research and innovation supported research uptake and the dissemination of results. The cash transfer programme was funded by European Commission's Humanitarian Aid and Civil Protection department. 1 study author supported by the CGIAR Research Program on Agriculture for Nutrition and Health (A4NH), led by the International Food Policy Research Institute. Tonguet‐Papucci: study funded by Actino Contre la Faim – France and the Center for Disease Control. The cash transfer programme was made possible thanks to ECHO funds. The cost‐effectiveness analysis is co‐funded by Action Contre la Faim and the Nutrition Embedding Evaluation Program (NEEP, PATH‐DFID). Country: Burkina Faso Setting: poor and very poor rural HHs in eastern Burkina Faso Comments: registered at clinicaltrials.gov as NCT01866124. Authors' names: F Houngbe; A Tonguet‐Papucci Email: fhoungbe@actioncontrelafaim.org; apapucci@actioncontrelafaim.org Declarations of interest: yes; Houngbe: "J‐FH, LH, and PK, no conflicts of interest. FH, AT‐P, CA, and MA‐A are employed by Action Against Hunger France, which implemented the MAM'Out study." Tonguet‐Papucci: "The authors declare that they have no competing interests." Study or programme name and acronym: Moderate Acute Malnutrition Out (MAM'Out) research project/study Type of record: Houngbe: journal article. Tonguet‐Papucci: study protocol
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Low risk	Quote: "Villages were randomly assigned to the intervention and control groups during a ceremony in order to keep the allocation of cash transparent and fair. Representatives of each of the 32 villages drew blindly from a bag one of the 32 identical papers with "cash" or "no cash" written on it."
Allocation concealment (Selection bias)	Low risk	Allocation to intervention groups was by village at the beginning of the study.
Baseline characteristics similar (Selection bias)	Low risk	Quote: "Overall, baseline characteristics were balanced between the intervention and the control groups (Table 1)." Comment: intervention group had more young children (aged < 6 years) compared to control group but all analyses adjusted for child's age at baseline. In addition, Houngbe 2017 reported that adjustment had been done for important prognostic covariates.
Baseline outcome measurements similar (Selection bias)	Low risk	Children in intervention group were more likely to be wasted. No morbidity outcomes measured at baseline. Analysis adjusted for morbidity status and WHZ at baseline.
Blinding of participants and personnel (Performance bias)	Low risk	No blinding possible but this was unlikely to affect participant behaviour.
Blinding of outcome assessment (Detection bias)	High risk	Outcome assessors were not blinded to treatment assignment. Morbidity outcomes were recalled by mothers and these could have been influenced by knowledge of treatment allocation.
Protection against contamination (Performance bias)	Low risk	Allocation was by village and it was unlikely that control villages received treatment. Given the cluster randomised design of the study combined with poor and very poor participants who likely did not travel to other villages often, the risk of contamination was not judged to be appreciable.
Incomplete outcome data (Attrition bias)	Low risk	Attrition was not high: 99/1278 (7.8%) children enrolled LTFU, due to either death or leaving the study area. 57/644 (8.9%) children in intervention group and 42/634 (6.6%) children in control group LTFU from visits 1 to 9. Data from children with < 2 measurements were excluded: 14 (2.2%) children in intervention group (2.2%) and 14 (2.2%) children in control group were excluded.
Selective outcome reporting (Reporting bias)	High risk	Some morbidity outcomes reported in the protocol were NR in published paper (oedema and measles).
Other bias	Unclear risk	Misclassification bias: unlikely. Measurement bias: unlikely. Incorrect analysis: unclear. Adjusted for clustering. Poisson regression model adjusted for clustering by village, household and child. The payment ('compensations') for the time control participants spent on the study may have resulted in volunteer bias, which may have altered effect sizes. Furthermore, the payment in itself was not meaningfully different from the intervention (though no values were specified) and may have influenced outcomes as well.