Ciliberto 2005

Methods	Study design: stepped wedge design treated as quasi‐randomised cluster trial Study duration: stepwise enrolment to a 8‐week treatment period (or until reaching WHZ > 0); those who reached WHZ > ‐2 were followed up 6 months after end of intervention period Recruitment date: December 2002 to June 2003
Participants	Children: MAM and SAM, but only data for SAM children are used in this review. Definitions for MAM and SAM were not provided Total number randomised: 645 SAM children (532 in the experimental and 113 in the control group); effective sample size 352 (290 in the experimental and 62 in the control group) Country and setting: southern Malawi; outpatients to NRUs Inclusion criteria: aged 10‐60 months; attending 1 of 7 NRUs as in‐ or outpatient; wasting (WHZ < ‐2), mild oedema (< 0.5 cm pitting oedema on the dorsum of the foot), or both, and a good appetite Exclusion criteria: severe oedema (> 0.5 cm pitting oedema on the dorsum of the foot); systemic infection; anorexia Baseline characteristics of experimental group (including MAM children): 526/992 males; mean age 23 SD 10 months; oedema 434/992; mean weight 7.7 SD 1.7 kg; mean length 74.8 SD 6.6 cm; WAZ ‐3.5 SD 1.0; HAZ ‐3.0 SD 1.5; WHZ ‐2.2 SD 0.8; mean MUAC 11.6 SD 1.4 cm; children still breastfeeding 505/992; mean age when breastfeeding stopped 21 SD 7 months; 347/992 were hospitalised prior to the study (mean 11 SD 9 days) Baseline characteristics of control group (including MAM children): 98/186 males; mean age 24 SD 12 months; oedema 86/186; mean weight 7.6 SD 1.9 kg; mean length 75.0 SD 7.6 cm; WAZ ‐3.7 SD 1.0; HAZ ‐3.2 SD 1.6; WHZ ‐2.5 SD 0.9; mean MUAC 11.6 SD 1.5 cm; children still breastfeeding 72/186; mean age when breastfeeding stopped 21 SD 8 months; 186/186 were hospitalised prior to the study (mean 22 SD 14 days) Stabilised before start of study: yes, the "very ill" received F75 containing 75 kCal/100 mL and 0.9 g protein/100 mL; parenteral antibiotics
Interventions	RUTF: locally produced by the study team and Tambala Foods (Blantyre, Malawi); ingredients were 25% peanut butter, 28% sugar; 30% full‐fat milk; 15% vegetable oil; 1.4% imported micronutrients (Nutriset); 260 g daily portion provided 175 kCal/kg/day and 5.3 g/kg/day protein Standard diet: F100 when in the NRU and maize/soy blended flour supplemented with micronutrients at home; blended flour (80% maize, 20% soy) prepared by carer and to be consumed 7 times/day; the family each participant received 50 kg of flour Concomitant treatment: not reported
Outcomes	Recovery (the attainment of a WHZ > ‐2 while remaining free of oedema) Relapse during intervention period (recurrence of oedema or systematic infection requiring readmission to NRU) Relapse after the study (WHZ < ‐2 or oedema 6 months after recovery) Mortality (all reported child deaths were considered to be a consequence of malnutrition) Weight gain (g/kg/day during the first 4 weeks of the intervention period) MUAC gain (mm/day during the first 4 weeks of the intervention period) Height gain (length/height in mm/day over 8 weeks of treatment) Diarrhoea (days with diarrhoea as reported by carer)
Notes	Ethics approval: College of Medicine Research and Ethics Committee of the University of Malawi; Human Studies Committee of Washington University in St Louis Informed consent: obtained; not reported whether it was orally or in writing Financial contributors: Doris Duke Clinical Scholars Programme; St Louis Children's Hospital Foundation; World Food Programme; Valid International; US Agency for International Development Reference standard for anthropometrical data: NCHS reference population (EPI 2002 version 1.1.2, Centers for Disease Control and Prevention, Atlanta, USA) Quality of anthropometrical measurements: unclear as no relevant information was reported Tested for peanut allergies: no
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"...systematic allocation with a stepped wedge design..."
Allocation concealment (selection bias)	High risk	Quasi‐randomised study, therefore, prediction of next allocation possible
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Children across groups received the same contact time with study personnel, thus low risk for performance bias. No blinding, thus unclear risk for detection bias
Incomplete outcome data (attrition bias) All outcomes	Low risk	Loss to follow‐up not differential and small: 98/992 (9.9%) from the experimental group and 15/186 (8.1%) from the control group dropped out of the study
Selective reporting (reporting bias)	Unclear risk	Protocol not available; primary and secondary outcome prespecified in the Methods section and addressed in the Results section
Other bias	Unclear risk	At baseline, the standard diet group had worse WHZ and longer prior hospital stay