Gertler 2000 (PROGRESA).

Study characteristics
Methods	Study design: cRCT Study grouping: parallel group How were missing data handled? missing data or potentially invalid data were excluded from the analysis (see footnote Table 3; Hoddinott 2003) Randomisation ratio: 60:40 (Gertler 2000) Recruitment method: using door‐to‐door methods to inform HHs about eligibility; PROGESSA achieved a take‐up rate of 97% (Gertler 2000) Sample size justification and outcome used: NR Sampling method: 506 communities randomly sampled from 50,000 eligible PROGRESA communities (matched on initial index level of community poverty) were chosen to participate, with each community randomly assigned to a treatment (n = 320) or control (n = 185) group. Study aim or objective: PROGRESA aimed to improve the nutritional status of poor children in rural Mexico in addition to improving education and health attainments while also reducing consumption poverty. Study period: about 2 years. Summer 1998 to summer 2000 (Gertler 2000). Some studies report < 2 years: from April 1998 (start of benefits) to October 1999 (when control HHs started receiving benefits). Intervention currently ongoing. Unit of allocation or exposure: communities
Participants	Baseline characteristics Intervention or exposure Age: children aged 0–5 years, mean: 2.753 (SD 1.667) (Gertler 2000) Place of residence: poor rural communities Sex: Gertler 2000: % boys aged 0–5 years: 0.394 (SD 0.489) Ethnicity and language: NR Occupation: NR Education: Gertler 2000: father's schooling, mean, years: 1.111 (SD 1.093); mother's schooling, mean, years: 1.047 (SD 1.056) SES: Hoddinott 2003: HH size, mean(?): 5.81. Gertler 2000: labour and non‐labour income, mean: 4.939 (SD 0.896) Social capital: NR Nutritional status: NR Morbidities: Gertler 2000: children aged 0–5 years ill last month, mean, n: 0.306 (SD 0.461) Concomitant or previous care: Skoufias 2007: 1 additional requirement of the PROGRESA programme was that HHs benefiting from PROGRESA were to stop receiving benefits from other pre‐existing programmes such as Ninos de Solidaridad, Abasto Social de Leche, de Tortilla and the National Institute of Indigenous people. Control Age: Gertler 2000: children aged 0–5 years, mean: 2.746 (SD 1.701) Place of residence: poor rural communities Sex: Gertler 2000: % boys aged 0–5 years: 0.376 (SD 0.484) Ethnicity and language: NR Occupation: NR Education: Gertler 2000: father's schooling, mean, years: 1.050 (SD 1.086); mother's schooling, mean, years: 1.016 (SD 1.079) SES: Hoddinott 2003: HH size, mean(?): 5.47. Gertler 2000: labour and non‐labour income, mean: 5.094 (SD 0.814) Social capital: NR Nutritional status: NR Morbidities: Gertler 2000: children aged 0–5 years ill last month, mean, n: 0.298 (SD 0.458) Concomitant or previous care: unclear whether they were still receiving other social support. Overall: NR Inclusion criteria: poor HHs within 506 poor communities with schooling and health infrastructure identified by PMT. List of beneficiary HHs presented to a community assembly for review and discussion and list was changed according to established criteria for the selection of beneficiary families (process called densification). No other criteria reported. Exclusion criteria: NR Pretreatment: overall, communities were comparable but this was not the case at HH or individual level. HHs in the intervention group were bigger, had more children but fewer adults aged > 55 years compared to control group (Hoddinott 2003). No difference in illness rates or number of visits to clinics for nutrition monitoring between control and treatment groups (Gertler 2000). Attrition per relevant group: overall attrition varied between papers and there was no report of attrition per group. 3350/12,291 (27%) intervention HHs did not receive the intervention by March 2000. Of these, 2872 HHs were not incorporated into programme; 478 HHs chose not to participate (no official records) or moved out of locality (Skoufias 2005). Attrition per group NR. In total, 221 HHs excluded as reported no food was consumed within the home, and 7165 HHs excluded with caloric availability per person per day < 875 kcal or > 4.768 kcal (Hoddinott 2003). Description of subgroups measured and reported: NR Total number completed and analysed per relevant group: total 16,614 HHs, which excluded 7386 HHs that were excluded because of inadequate data. Number analysed per group NR and also varied depending on report. Total number enrolled per relevant group: 320 communities allocated to intervention and 186 to control groups, out of 506 communities. 24,000 HHs overall in sample at beginning of study (Hoddinott 2003). 97% of HHs enrolled in programme (Gertler 2000). Baseline sample included 112,319 individuals from 18,795 HHs in 506 experimental communities. Approximately, 60% of sample came from treatment areas and 40% from control (Gertler 2000). Total number randomised per relevant group: intervention group: 320 communities; control group: 186 communities
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power Intervention type: CCT Description: cash transfers to families every 2 months (food and education transfers) if: 1. every family member accepted preventive health services; 2. children aged 0–5 years and lactating mothers attended nutrition monitoring clinics where their growth was measured (every 2 months for children under 24 months), immunisation, obtained nutrition supplements and they received education on nutrition and hygiene; and 3. pregnant women visited clinics to obtain antenatal care (5 visits), nutritional supplements and health education. Linked to children's school enrolment and regular school attendance and clinic attendance. Included in‐kind health benefits; nutritional supplements for children aged ≤ 5 years, and pregnant and lactating women; and instructional meetings on health and nutritional issues. Transfers were targeted to mothers of the families (female head of HH). 3 types of monetary transfers: 1. scholarships tied to children's school attendance; 2. money for school supplies and 3. transfers for food. Amount of food transfer was 20% of HH monthly consumption expenditure preintervention. Scholarship transfers occur monthly (?), and varied depending on age and sex of the child, with maximum scholarship cap of MXN 490 pesos per HH (January–June 98) and MXN 625 per HH (July–December 1999). Food transfer also occur monthly(?) and depended on age and sex of children in HH and compliance with PROGRESA requirements. Duration of intervention period: about 2 years: April/May 1998 (benefits started) to November 1999/March 2000 (control HHs start receiving benefits) Frequency: cash transfers every 2 months Number of study contacts: 4: baseline data (survey March 1998) was unusable; October 1998; June 1999; November 1999 Providers: Mexico Government. Health care component administered by Ministry of Health and IMSS‐Solidaridad, a branch of the Mexican Social Security Institute, which provides benefits to uninsured individuals in rural areas. Delivery: every second month. Verification of school attendance relied on completion of forms, which had to be completed and returned to PROGRESA before the initiation of payment. This often led to lags in payments. Every 2 months, confirmation of whether children of beneficiary families attend school > 85% of time was submitted to PROGRESA by school teachers and directors, and this triggered receipt of bi‐monthly cash transfer for school attendance. Receipt of monetary transfers and nutritional supplements were tied to mandatory healthcare visits to public clinics. Healthcare professionals submitted certification of beneficiary visits to PROGRESA every 2 months, which triggered receipt of cash transfer for food support. About 65% of HHs received the PROGRESA cash transfers, due to administrative errors and delays in the final registration of beneficiary HHs (for more details see Skoufias 2005). Another unique feature was that the cash transfers were given to mother of HH, a strategy designed to target the funds within the HH to improve the children's education and nutrition. Fernald 2009: compliance verified by clinics and schools, and about 1% of HHs were denied cash transfers for non‐compliance. Co‐interventions: none reported. 1 additional requirement of the PROGRESA programme was that HHs benefiting from PROGRESA were to stop receiving benefits from other pre‐existing programmes such as Ninos de Solidaridad, Abasto Social de Leche, de Tortilla and the National Institute of Indigenous people. Resource requirements: community health and education facilities Economic indicators: programme's budget was USD 777 million for 1999 (0.2% of GDP). Cost of intervention: mean monthly payments of MXN 197 per beneficiary HH (November 1998); mean of MXN 99 for food and MXN 91 for school attendance. The programme operated in almost 50,000 rural villages in 31 states. PROGRESA's budget was about USD 800 million or 0.2% of GDP (Gertler 2000). Control: no intervention (started receiving benefits 2 years later) Co‐interventions: none reported. Unclear if these HHs were receiving benefits from other pre‐existing programmes
Outcomes	Total HH expenditure; HH expenditure on food Diet intake: HH caloric availability per month per day – all food/individual food groups Anthropometry: height; HAZ, stunting, BMIZ Biochemical indicators: anaemia Cognitive function and development: cognitive test scores (verbal, cognitive, behavioural) Morbidity: child illness rates
Identification	Sponsorship source: Gertler 2000 NR. Gertler 2004 – Mexican government and Mexican Institute of Public Health funded data collection and initial data analysis, US National Institute of Child and Human Development provided research support Country: Mexico Setting: poor rural communities Authors' names: Paul J Gertler; Jere R Behrman; Lia CH Fernald; John Hoddinott; Emmanuel Skoufias Email: Gertler@haas.berkeley.edu; jbehrman@pop.upenn.edu; fernald@berkeley.edu; j.hoddinott@cgiar.org; eskoufias@worldbank.org Declarations of interest: NR Study or programme name and acronym: PROGRESA (Programa de Educacion Salud y Alimentacion)/ Oportunidades (PROGRESA was the original name and currently called Oportunidades) Type of record: IFPRI reports; journal articles
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Low risk	Random assignment generated without weighting with randomisation commands in STATA version 2.0 (Fernald 2009).
Allocation concealment (Selection bias)	Low risk	Allocation done at community level.
Baseline characteristics similar (Selection bias)	Low risk	Study authors reported no baseline differences between groups at community level in terms of age, education, income and access to health care, but there were differences at HH level. They adjusted for HH demographic characteristics (e.g. HH size; proportions of children; and age, gender, education, occupation, ethnicity and marital status of the HH head) in their analyses.
Baseline outcome measurements similar (Selection bias)	Unclear risk	Outcome measurements such as food consumption and child's anthropometry NR by study authors at baseline.
Blinding of participants and personnel (Performance bias)	Low risk	Quote: "No sites were told that they would be participating in the programme, and information about timing of programme roll‐out was not made publicly available." Comment: blinding not possible in this type of intervention but researchers aimed to ensure that participants were not aware of the intervention evaluation. Lack of blinding is unlikely to influence participant or personnel behaviour in this type of intervention.
Blinding of outcome assessment (Detection bias)	High risk	There was no blinding and outcomes were subjective, which could have been influenced by lack of blinding.
Protection against contamination (Performance bias)	Unclear risk	Intervention communities were randomly selected from a set of rural communities in the same geographic region. In a baseline report, the study authors stated that it would be important to monitor individuals leaving or entering the localities in order to assess contamination bias (Behrman 2005), but no further data were provided in the later study report.
Incomplete outcome data (Attrition bias)	Unclear risk	Attrition not clearly reported by study authors. Loss of clusters: NR.
Selective outcome reporting (Reporting bias)	Unclear risk	No protocol available but outcomes outlined in methods section were reported in results section.
Other bias	High risk	Misclassification bias: low risk. Receipt of the programme was verified by examination of PROGRESA records. Measurement bias: unlikely. Incorrect analysis: unclear. Not all papers reported adjusting for clustering. Seasonality bias: unclear. Not controlled for in the analysis. Recruitment bias: high risk. List of beneficiary HHs within a community presented to a community assembly for review and discussion.