The effect of malaria on stunting: an instrumental variables approach.

BACKGROUND: Previous studies have found mixed evidence for an effect of malaria on stunting, but have suffered from concerns about confounding and/or power. Currently, an effect of malaria on stunting is not included in the Lives Saved Tool (LiST) model. METHODS: We used instrumental variables regression with the sickle cell trait and random assignment to bednets as instruments in the analysis of data on children aged 0-2 y from a bednet trial in western Kenya. RESULTS: We estimated that one additional clinical malaria episode per year increases the odds of a child being stunted by 6% (OR estimate: 1.06, 95% CI 1.01 to 1.11). CONCLUSIONS: Our finding that malaria affects stunting suggests that an effect of malaria on stunting in young children should be considered in the LiST model.