There is a long tradition of community-based rehabilitation for treatment of severe malnutrition: the question is whether it is effective and whether it should be advised for routine health systems.

To examine the effectiveness of rehabilitating severely malnourished children in the community in nonemergency situations.

A literature search was conducted of community-based rehabilitation programs delivered by day-care nutrition centers, residential nutrition centers, primary health clinics, and domiciliary care with or without provision of food, for the period 1980–2005. Effectiveness was defined as mortality of less than 5% and an average weight gain of at least 5 g/kg/day.

Thirty-three studies of community-based rehabilitation were examined and summarized. Eleven (33%) programs were considered effective. Of the sub-sample of programs reported since 1995, 8 of 13 (62%) were effective. None of the programs operating within routine health systems without external assistance was effective.

With careful planning and resources, all four delivery systems can be effective. It is unlikely that a single delivery system would suit all situations worldwide. The choice of a system depends on local factors. High energy intakes (> 150 kcal/kg/day), high protein intakes (4–6 g/kg/day), and provision of micronutrients are essential for success. When done well, rehabilitation at home with family foods is more cost-effective than inpatient care, but the cost effectiveness of ready-to-use therapeutic foods (RUTF) versus family foods has not been studied. Where children have access to a functioning primary health-care system and can be monitored, the rehabilitation phase of treatment of severe malnutrition should take place in the community rather than in the hospital but only if caregivers can make energy- and protein-dense food mixtures or are given RUTF. For routine health services, the cost of RUTF, logistics of procurement and distribution, and sustainability need to be carefully considered.