Kusuma 2017a.

Study characteristics
Methods	Study design: cRCT Study grouping: parallel group How were missing data handled? No missing data reported Randomisation ratio: 1:1 Recruitment method: NR Sample size justification and outcome used: NR Sampling method: First, within each province, the 20% richest districts were excluded for both programmes (based on school transition rates, malnutrition and poverty). Districts who participated in the Kecamatan Development Project (a governmental poverty programme that developed infrastructure and capacity) were eligible for Generasi, from which 20 were selected and stratified by province. In NTT, East Java, and West Java selection was random, in Gorontalo and North Sulawesi all eligible districts were selected. Within the selected districts, subdistricts were not eligible if they had participated in the UPP or where < 30% of the villages (desa) and urban precincts (kelurahan) were considered as rural by the national statistics office. The final screening yielded 300 PNPM Generasi eligible subdistricts. The remaining districts were considered for PKH. The subdistricts that were deemed as 'supply‐side ready' were then randomly assigned to the PKH treatment and control groups. Study aim or objective: Kusuma 2016: to provide evidence on the effects of HH cash transfers (PKH) and community cash transfers (Generasi) on determinants of maternal mortality. Kusuma 2017: to provide evidence on the impact of HH cash transfers (PKH) and community cash transfers (Generasi) on children's food consumption. Study period: from June–August 2007 to October–January 2010 Unit of allocation or exposure: subdistricts
Participants	Baseline characteristics Intervention: NR Control Age: mother, mean, years: 35.77; children aged < 5 years, mean, n: 1.46; child age, mean, months: 30.11 Place of residence: urban subdistricts Sex: child, female, proportion: 0.45 Education: mother's education (1 = ≥ 6 years): 0.71 SES: HH size, mean: 5.8; per capita HH expenditure: IDR 5.36; latrine in house, mean: 0.48; house had electricity, mean: 0.77; wood and coal cooking fuel, mean: 0.77 Nutritional status: children's food consumption (previous week), mean: grain, roots, tubers: 0.97; milk: 0.4; meat: 0.38; fish: 0.76; eggs: 0.66; fruit, vegetables: 0.95 Overall: NR Inclusion criteria: very poor HHs (UCT database) with pregnant/lactating women, children aged 0–15 years in supply‐ready urban subdistricts (based on existing health and education facilities). Exclusion criteria: NR Pretreatment: for PKH there were more girls in the intervention group (P < 0.05). However, the study authors did not report any other differences for a number of maternal, HH and subdistrict characteristics at baseline between the intervention and control groups within each programme. Attrition per relevant group: the PKH children sample can be considered without attrition because it was 1394 at follow‐up out of 1395 at baseline. Description of subgroups measured and reported: NR Total number completed and analysed per relevant group: 1394 children, 1376 HHs in total. Total number enrolled per relevant group: Kusuma 2017 used subsets of data from HHs that participated in PKH and Generasi and reported on food consumption for children aged 24–36 months: 1395 HHs at baseline. Total number randomised per relevant group: 360 subdistricts (Intervention group: 180 subdistricts; control group: 180 subdistricts)
Interventions	Intervention characteristics Food access intervention category: increase buying power Intervention type: PKH (CCT) Description: cash transfers to a HH consisting of a fixed amount of USD 20 per year with the following additions: USD 80 per year if a mother was pregnant or had children aged 0–6 years (or both); USD 40 per year if a mother had 1 child at primary school and USD 80 per year if she had 1 child at secondary school. No rules how the cash should be used. Trained field facilitators advised HHs on conditionalities and cash penalty. Conditionalities: health: 1. 4 antenatal visits, 2. iron tablets during pregnancy, 3. assisted delivery, 4. 2 postnatal visits, 5. complete childhood immunisations, 6. adequate monthly weight increases for infants, 7. monthly weighing for children aged < 3 years and biannually for children aged < 5 years and 8. vitamin A twice a year for children aged < 5 years. Education: 9. primary school enrolment of children aged 6–12 years, 10. minimum attendance rate of 85% for primary school‐aged children, 11. junior secondary school enrolment of children aged 13–15 years and 12. minimum attendance rate of 85% for junior secondary school‐aged children. Duration of intervention period: 2 years Frequency: every 3 months Number of study contacts: 2: baseline survey (June–August 2007); follow‐up survey (October‐December 2009) Providers: government of Indonesia Delivery: collected by mothers through the nearest post office. Follow‐up survey showed that 50% of sampled HHs in the treatment areas said that they ever received PKH. This relatively low PKH coverage rate could partly explain the lack of impact. However, such coverage rates might be due to HHs not really being aware of the various poverty programme cash/subsidy they received (e.g. BLT (UCTs) vs PKH (CCTs) vs BOS (school operational assistance). Co‐interventions: NR Resource requirements: NR Economic indicators: NR Control characteristics: no intervention
Outcomes	Anthropometry: underweight (WAZ < –2SD); severe underweight (WAZ < –3SD); wasting (WHZ < –2SD); severe wasting (WHZ < –3SD); stunting (HAZ < –2SD); severe stunting (HAZ < –3SD)
Identification	Sponsorship source: research fellowship from the Harvard Kennedy School Indonesia Program. Country: Indonesia Setting: urban very poor HHs (PKH) and rural very poor HHs (Generasi) in West Java, East Java, North Sulawesi, Gorontalo, and East Nusa Tenggara provinces Author's name: Dian Kusuma Email: dkusuma@mail.harvard.edu Declarations of interest: none declared. Study or programme name and acronym: Program Keluarga Harapan (PKH) and Generasi. Type of record: journal articles
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Unclear risk	Random sequence generation method NR.
Allocation concealment (Selection bias)	Unclear risk	Allocation concealment methods NR
Baseline characteristics similar (Selection bias)	Low risk	No important differences between groups reported at baseline.
Baseline outcome measurements similar (Selection bias)	Unclear risk	Although food consumption was similar at baseline, growth outcomes at baseline were NR
Blinding of participants and personnel (Performance bias)	Low risk	Blinding of participants and personnel was not possible but unlikely to influence intervention received.
Blinding of outcome assessment (Detection bias)	Low risk	Unclear whether outcome assessors were blinded but key outcomes were objective and unlikely to be influenced by lack of blinding.
Protection against contamination (Performance bias)	Low risk	Allocation was at subdistrict level, which minimised the risk of spillovers.
Incomplete outcome data (Attrition bias)	Low risk	No attrition was reported.
Selective outcome reporting (Reporting bias)	Unclear risk	Protocol N/A.
Other bias	Unclear risk	Misclassification of exposure: low risk. Measurement bias: unclear risk. Incorrect analysis: low risk. analyses adjust for clustering. Recruitment bias: low risk. HHs randomly selected after subdistrict randomisation.