Fernald 2011.

Study characteristics
Methods	Study design: cRCT How were missing data handled? missing data excluded from analysis Randomisation ratio: 2:1 (79 parishes in intervention: 39 parishes in control) Recruitment method: NR Sample size justification and outcome used: NR Sampling method: stratified random sampling. Parishes stratified into rural and urban groups and from each group, treatment and comparison parishes were randomly selected. Study aim or objective: to analyse the impact of a programme that transfers cash to women in rural Ecuador on measures of ECD (Paxson 2010). First objective was to use a randomised effectiveness trial in Ecuador to address the question of whether very young children (aged 12–35 months) benefit in terms of health outcomes or language development if their families receive a cash transfer (Fernald 2011; study included a subset of younger children only). Study period: duration of intervention during which participants received transfers was unclear. Rural families in treatment group were eligible for the transfer for 17 months prior to the follow‐up survey. Rural families became eligible for transfer from June 2004, and urban families in November 2006 and follow‐up survey was conducted between September 2005 and January 2006. Unit of allocation or exposure: parishes allocated to intervention and control groups but certain HHs only selected for inclusion in study.
Participants	Baseline characteristics Intervention or exposure Age: mother, years, mean: 26.51 (SE 7.22); child, months, mean: 38.82 (SE 13.13) Place of residence: urban: 46% (365) Sex: child male, mean: 0.494 (SE 0.5) Ethnicity and language: NR Occupation: NR Education: mother's education, years, mean: 6.88 (SE 2.94). SES: number of family members, mean: 4.78 (SE 2.19) Social capital: mother living with husband, mean: 0.696 (SE 0.460) Nutritional status: mother's elevation‐adjusted Hb, mean: 11.64 (SE 1.44). Child HAZ (US norms), mean: –1.22 (SE 1.51). Child's elevation‐adjusted Hb (g/dL), mean: 10.38 (SE 1.46). Child TVIP standardised score, mean: 82.45 (SE 13.63) Morbidities: NR Concomitant or previous care: NR Control Age: mother, years, mean: 26.47 (SE 7.19); child, months, mean: 35.38 (SE 12.63) Place of residence: rural Ecuador Sex: child male, mean: 0.540 (SE 0.499) Ethnicity and language: NR Occupation: NR Education: mother's education, years, mean: 6.78 (SE 2.68) SES: number of family members, mean: 4.72 (SE 2.12) Social capital: mother living with husband, mean: 0.695 (SE 0.461) Nutritional status: mother's elevation adjusted Hb, mean: 11.53 (SE 1.59). Child HAZ (US norms), mean: –1.20 (SE 1.68). Child's elevation‐adjusted Hb (g/dL), mean: 10.30 (SE 1.52). Child TVIP standardised score, mean: 84.27 (SE 13.86) Morbidities: NR Concomitant or previous care: NR Overall Age: 1196 mothers; 697 children. Mother's age, years, mean: 22.6 (SD 3.8); child's age, months, mean: 6.6 (SD 4.2) Place of residence: urban: 45% (542) Sex: child male: 52% (361) Ethnicity and language: mother speaks indigenous language: 3% (30) Occupation: NR Education: mother's completed schooling (grades), mean: 6.9 (SD 2.9) SES: Asset Index, mean: 0 (SD 2.3) Social capital: NR Nutritional status: mother's adjusted Hb level, mean: 11.3 (SD 1.5). Child's HAZ, mean: 0.6 (SD 2.1). Child's adjusted Hb, mean: 9.6 (SD 1.3) Morbidities: NR Concomitant or previous care: NR Inclusion criteria: primary sample of HHs drawn for this study included only families in the first or second Selben quintiles who had children aged 0–6 at baseline, had no older siblings and had not received the Bono Solidario programme Exclusion criteria: NR Pretreatment: no significant difference between intervention and control parishes. Differences in baseline characteristics between HHs in the treatment and control groups were small and are never significant at conventional levels. This was true for the sample as a whole, as well as for families and children in the poorest quartile of per capita expenditures." Attrition per relevant group: total: 163/2748 children were LTFU (belonging to 77/1642 HHs). Attrition per group NR. Description of subgroups measured and reported: baseline expenditure (bottom quartile, top 3 quartile); age (young vs old); gender (boys vs girls) Total number completed and analysed per relevant group: total completed: 2585 children, 1565 HHs. Total number of children per group NR. Total number parishes analysed: varied per outcome due to missing data. Total number enrolled per relevant group: 77 parishes enrolled: 51 treatment; 26 control. Total enrolled: either 2748 or 2069 children (numbers in table A2 and 2 differed). Total HHs enrolled: 1642; 1388 children in intervention; 681 children in control. Total sample at baseline consisted of 3426 HHs and 5547 children aged < 72 months. Fernald 2011 focused only on children aged < 36 months at follow‐up (included children in urban and rural areas whereas Paxson 2010 only reported results for rural areas). Total number randomised per relevant group: 77 parishes randomised: 51 treatment; 26 control. Total enrolled: either 2748 or 2069 children (numbers in table A2 and 2 differed); 1388 children in intervention; 681 children in control.
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power Intervention type: UCT Description: beginning in mid‐2003, Bono Solidario was gradually replaced with a new programme, the BDH. Eligible families received USD 15 per month. They were not required to withdraw their USD 15 on a monthly basis but could allow transfers to accumulate for up to 4 months. Duration of intervention period: HHs were eligible for 17 months before follow‐up, but it was unclear whether they were receiving transfers during the entire period. Frequency: monthly Number of study contacts: 2: baseline (October 2003–March 2004) and follow‐up (September 2005–January 2006) Providers: transfers administered by the Government of Ecuador and distributed through a large network of private banks (Banred) and the National Agricultural Bank (Banco Nacional de Fomento) (Schady Araujo, 2006) (Fernald 2011). Baseline and follow‐up surveys were conducted by the World Bank and the Government of Ecuador (Fernald 2011) Delivery: transfers were given to mothers rather than fathers and were distributed through the banking system, although beneficiaries did not need to have a bank account to receive them. The fraction of rural families that received transfers among families randomised into the BDH treatment group climbed quickly once the programme became available, reaching 56% by January 2005 and 60% by January 2006. Overall, 75% of sampled families in the treatment parishes received a transfer in ≥ 1 month since June 2004. Mean monthly transfer across all treatment‐group families, between January 2005 and November 2006, was USD 10.51. This was less than amount planned. According to survey response data, there was very little contamination of intervention: take‐up of the BDH programme was 73% for the treatment group and 3% for the comparison group. Co‐interventions: NR Resource requirements: NR Economic indicators: NR Control: no intervention
Outcomes	Anthropometry: HAZ; height Biochemical: Hb Cognitive function and development: language (TVIP score and IDHC‐B score); long‐term memory; short‐term memory; visual integration Anxiety and depression: mother's depression score (CES‐D); mother's PSS
Identification	Sponsorship source: Center for Economic and Policy Studies at Princeton University, the government of Ecuador, and the World Bank Country: Ecuador Setting: rural and urban parishes Author's name: Lia Fernald Email: fernald@berkeley.edu Declarations of interest: NR Study or programme name and acronym: Bono de Desarrollo Humano (BDH) programme Type of record: journal article
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Unclear risk	Study was randomised but authors did not report how random sequence was generated.
Allocation concealment (Selection bias)	Low risk	Allocation was at parish level at beginning of study, and all eligible HHs that were in these parishes were either in intervention or control group. For inclusion in study, HHs had to meet specific criteria and then eligible HHs were randomly selected.
Baseline characteristics similar (Selection bias)	Low risk	Quote: "there is no evidence of significant differences between treatment and control parishes." "The table shows that differences in baseline characteristics between HHs in the treatment and control groups are small in magnitude and are never significant at conventional levels. This is true for the sample as a whole, as well as for families and children in the poorest quartile of per capita expenditures."
Baseline outcome measurements similar (Selection bias)	Low risk	Baseline outcome measurements (HAZ and TVIP score) were similar between children in intervention and control groups.
Blinding of participants and personnel (Performance bias)	Low risk	Blinding was not possible due to the nature of the intervention, but it is unlikely to have influenced participant or personnel behaviour.
Blinding of outcome assessment (Detection bias)	Low risk	Blinding was not possible due to nature of intervention. However, objective outcomes were measured and, thus, it is unlikely that lack of blinding affected outcome assessment.
Protection against contamination (Performance bias)	Low risk	Quote: "According to survey response data, there was very little contamination of the intervention: take‐up of the BDH program was 73% for the treatment group and 3% for the comparison group."
Incomplete outcome data (Attrition bias)	Unclear risk	The numbers varied per outcome reported which indicates that missing data were excluded from the analysis. Total attrition was reported in 1 table as 163/2748 (5.9%) children; however, the total number of children enrolled/randomised was reported as 2069 in another table. Given the unclear reporting of numbers, it is unclear how much missing data there was and how it differed between the intervention and control groups.
Selective outcome reporting (Reporting bias)	Unclear risk	No protocol was available for this study.
Other bias	Low risk	Misclassification bias: unlikely as allocation was not self‐reported. Measurement bias: unlikely; standardised processes and tools used to assess outcomes. Incorrect analysis: SEs were clustered at parish level, therefore, clustering was adjusted for. No other bias identified.