Abstract

Background: Relapse among children treated for wasting is a major concern.
Objectives: We estimated the frequency and determinants of relapse to wasting in 2 populations exposed to PROMIS, an integrated wasting prevention and screening program.
Methods: Using longitudinal data from PROMIS trials in Burkina Faso and Mali, we calculated the incidence rate and period prevalence of relapse to wasting within 6 mo in children who had ≥1 wasting episode ending when they were ≥6 mo old for which they were treated and recovered (N Burkina Faso = 247; N Mali = 220). We used backward elimination to select a multivariable model of sociodemographic, nutrition-related, and health-related determinants of relapse. We also evaluated whether prevention interventions [behavior change communication (BCC) and/or small-quantity lipid-based nutrient supplements (SQ-LNS)] were associated with relapse, adjusting for confounders and trial arm.
Results: Relapse incidence was 2.6 per child-year in Burkina Faso (n = 291 episodes) and 1.6 per child-year in Mali (n = 300 episodes). In both countries, being fed the recommended food frequency or iron-rich foods after recovering from wasting was associated with lower risk of relapse. In Mali, longer wasting episodes, lack of minimally diverse diet consumption, and several caregiver/household characteristics were associated with higher risk of relapse. In both countries, receipt of BCC after recovery from wasting was associated with lower risk of relapse [incidence rate ratio (IRR)Burkina Faso: 0.51; 95% CI: 0.30, 0.86; IRR Mali: 0.26; 95% CI: 0.11, 0.65), as was receipt of SQ-LNS (IRR Burkina Faso: 0.33; 95% CI: 0.16, 0.70; IRR Mali: 0.43; 95% CI: 0.19, 0.94), after adjustments.
Conclusions: Children being discharged from wasting treatment are a well-defined vulnerable population who stand to benefit from targeted postdischarge preventive interventions. BCC that includes advice on optimal infant and young child feeding practices and SQ-LNS may help prevent wasting relapse in at-risk children.