Brunie 2014.

Study characteristics
Methods	Study design: PCS Study grouping: parallel group How were missing data handled? attrition between baseline and endline was 10.1% overall. Missing data were excluded from analyses. 38 HHs declared participating in activities that did not correspond to their group – these observations were excluded from analyses. Randomisation ratio: 2:2:2:2 – VSL only; VSL + AM; AM only; control; authors mentioned factorial randomised design but it was not really randomised. Recruitment method: districts were randomised to 1 of 2 intervention groups or a control group. Intervention HHs declared interest in participating and the control group comprised sample HHs taken from the general population. Sample size justification and outcome used: sample size was calculated based on approximate estimates from past studies on food sufficiency, HDDS, IDDS and WAZ. These calculations indicated that 20 primary sampling units, comprising 14 HHs, per district was sufficient; resulting in 560(?) HHs per group. Sampling method: 2‐stage sampling approach employed. First, 8/18 districts purposefully selected, and each intervention group was allocated 2 districts. Sample of HHs that declared interest in participating in VSL/AM activities was drawn and a sample from general population was drawn for the control group. Then PSUs were selected: in intervention group, PSUs were VSL groups, and in the control group the PSUs were the EAs from the 2007 Census. Second, HHs were selected: eligible HHs within selected PSUs were enumerated and random sample drawn. HH participation was voluntary. PSUs were randomly selected with probability proportionate to size (in terms of number of HHs represented). Study aim or objective: to evaluate the impact of participation in an economic‐strengthening initiative, namely VSL groups, on HH and child nutritional outcomes, with an additional focus on identifying possible gaps and areas for complementary programming. Study period: August 2009–August 2012 Unit of allocation or exposure: provincial districts allocated to intervention groups
Participants	Baseline characteristics Intervention or exposure Age: mean number of children aged < 5 years: VSL: 0.84 (SE 0.87); VSL + AM: 1.04 (SE 0.87); mean age of HH head, years: VSL 43.19 (SE 14.28); VSL + AM: 38.77 (SE 11.43) Place of residence: NR Sex: male‐headed HHs, %: VSL: 90.54; VSL + AM: 86.90 Ethnicity and language: NR Occupation: agriculture primary occupation, %: VSL: 87.72: VSL + AM: 90.18 Education: mean education of HH head, years: VSL: 3.28 (SE 3.00); VSL + AM: 3.90 (SE 2.92); mean education of spouse, years: VSL 1.83 (SE 2.04); VSL + AM 2.49 (SE 5.58) SES: mean HH size: VSL 4.70 (SE 2.03); VSL + AM: 5.47 (SE 1.97); annual HH income: VSL: MZM 9398 (SE 18,088); VSL + AM: 11,525 (SE 22,482) Social capital: NR Nutritional status: mean months of food sufficiency: VSL: 10.41; VSL + AM: 9.27; mean HDDS: VSL: 4.06; VSL + AM: 4.20; mean IDDS: VSL: 2.51; VSL + AM: 2.99; mean WAZ: VSL: –1.21; VSL + AM: –0.96 Morbidities: NR Concomitant or previous care: SANA (Seguranca Alimentar de Nutricao e Agricultura) – number NR Control Age: mean number of children aged < 5 years: 0.62 (SE 0.78); mean age of HH head, years: 43.34 (SE 13.93) Place of residence: NR Sex: male‐headed HHs, %: 90.30 Ethnicity and language: NR Occupation: agriculture primary occupation, %: 87.55 Education: mean education of HH head, years: 3.33 (SE 3.21) years; mean education of spouse, years: 1.44 (SE 1.84) SES: mean annual HH income: MZM 8843 (20,354) Social capital: NR Nutritional status: VSL matched mean months of food sufficiency: 10.58; VSL + AM matched mean months of food sufficiency: 10.47; VSL matched mean HDDS: 3.73; VSL + AM matched mean HDDS: 3.82; VSL matched mean IDDS: 2.87; VSL + AM matched mean IDDS: 2.82; VSL matched mean WAZ: –1.25; VSL + AM matched mean WAZ: –1.15 Morbidities: NR Concomitant or previous care: SANA (Seguranca Alimentar de Nutricao e Agricultura) – number NR Overall Age: mean number of children aged < 5 years: 0.84 (SE 0.85); mean age of HH head, years: 41.85 (SE 13.46) Place of residence: NR Sex: % male‐headed HHs: 89.3 Ethnicity and language: NR Occupation: agriculture primary occupation, %: 88.43 Education: mean education of HH head, years: 3.50 (SE 3.07); mean education of spouse, years: 2.03 (SE 5.18) SES: mean HH size: 4.94 (SE 2.02); mean annual HH income: MZM 9858 (SE 20,412). Social capital: NR Nutritional status: NR Morbidities: NR Concomitant or previous care: SANA – number NR Inclusion criteria: HHs expressing interest in participating in the activity their district was randomised to. Exclusion criteria: NR. Pretreatment: NR. From baseline characteristics table there seems to be differences across intervention groups regarding annual HH income, education of spouse, age of HH head and number of children aged < 5 years, but whether this was statistically significant was NR. Attrition per relevant group: no attrition per group was reported, but total attrition was 10.1% and an additional 38 (2.7%) HHs did not participate in the intervention their district was randomised to. Description of subgroups measured and reported: no subgroups were measured, but there were 3 intervention groups namely VSL alone, AM alone and VSL + AM. Only VSL alone and VSL + AM were analysed in this study. Total number completed and analysed per relevant group: 395 HHs in the VSL group, 401 in the VSL + AM group and 480 in the control group. Note: check numbers – estimated impact have higher HH numbers Total number enrolled per relevant group: VSL: 395 HHs; VSL + AM: 401 HHs; control: 480 HHs Total number randomised per relevant group: N/A
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power (VSL) and social environment/support (AM) Intervention type: VSL groups and rotating labour scheme (AM) Description: VSLs were self‐managed and capitalised microfinance programmes where members pool savings and can borrow from the pool and repay with interest. These programmes work in cycles which terminate in paying out the accumulated savings and interest to members proportional to their initial deposit. The AM rotating labour scheme operates with groups of HHs working together on each family's land or enterprise on a rotational basis. Duration of intervention period: August 2009–August 2012 Frequency: NR Number of study contacts: baseline (August 2009) and 3‐year follow‐up (August 2012) Providers: NGO – Save the Children Delivery: NR Co‐interventions: SANA – Seguranca Alimentar de Nutricao e Agricultura – food security through nutrition and agriculture multiyear assistance programme targeting aspects of food utilisation. Communities mobilised to adopt good nutrition practices and taught pregnant women and cares to prevent malnutrition in young children. AM – rotating labor scheme; groups of HHs came together to work on each family's land or conduct another activity of their choice on a rotating basis. Build system of pooled labour that allows for greater advances in production or other tasks. Resource requirements: NR Economic indicators: NR Control: no intervention
Outcomes	Food security: months of food sufficiency Dietary diversity: HDDS, IDDS Anthropometry: WAZ (underweight)
Identification	Sponsorship source: US Agency for International Development (USAID) under the terms of associate agreement DFD‐AA‐00‐07‐00251‐00, the Supporting Transformation by Reducing Insecurity and Vulnerability with Economic Strengthening (STRIVE) project. Country: Mozambique Setting: HHs in an area with high prevalence of stunted, wasted and underweight children. Author's name: Diana Rutherford Email: drutherford@fhi360.org Declarations of interest: NR Study or programme name and acronym: village savings and loan (VSL) or a combination of VSL and AM. Type of record: journal article
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	High risk	Although the paper described this as a randomised design, there was no randomisation of districts to intervention groups. Quote: "Eight of Nampula's 18 districts were purposefully selected for this study. Two districts were assigned to each intervention and control arms such that, once paired, they formed arms similar with respect to distance from the capital, economic performance, rainfalls, and market activities." CBA design, and randomisation was N/A.
Allocation concealment (Selection bias)	High risk	Allocation was not concealed; CBA study.
Baseline characteristics similar (Selection bias)	Low risk	Unclear if groups were significantly different at baseline but analyses were adjusted for these characteristics.
Baseline outcome measurements similar (Selection bias)	Low risk	Unclear if baseline outcome measurements were significantly different between the intervention and control groups but analyses were adjusted for covariates.
Blinding of participants and personnel (Performance bias)	Low risk	No blinding of participants as this was a cohort study, but it was unlikely that this influenced participant performance.
Blinding of outcome assessment (Detection bias)	High risk	No blinding of participants as this was a cohort study. While some outcomes were objective measurements, others were self‐reported, which could have been influenced by knowledge of treatment allocation.
Protection against contamination (Performance bias)	Unclear risk	No information given on how potential contamination was prevented. As groups were matched inter alia on proximity it is possible that districts may have influenced each other due to relational ties (sending of money to family in a neighbouring district) or migration.
Incomplete outcome data (Attrition bias)	Unclear risk	10% overall LTFU but attrition per group was NR and it was unknown whether attrition was greater in 1 of the groups or if it was equal. Certain observations were excluded from the analyses. These missing data could have influenced the observed effect.
Selective outcome reporting (Reporting bias)	Unclear risk	No protocol available, but all a priori stated outcomes in the methods section were reported on in the results section.
Other bias	Low risk	None identified