Nutritional status and early warning of mortality in southern Ethiopia, 1988-1991

Ethiopian Government guidelines on nutritional survey data and relief provision call for intervention once mean weight-for-length (W/L) in an area falls below 90% of reference, on the basis that mortality is unlikely to rise until this level is reached. In this paper the appropriateness of the 90% cut-off is examined using data from Wolayita, southern Ethiopia. SURVEY DESIGN AND SUBJECTS: Fifteen to 25 villages are selected at random for survey each year, with all children 70-110 cm in length being followed up every 2 months. During the 3 years covered by these analyses 21,701 W/L measurements were made on 5455 children from 65 villages. 126 of the children died.

In the first and third survey years, rapid declines in mean W/L were recorded, with mortality increasing very roughly three-fold (compared to year 2, P < 0.01), even though mean W/L remained at or above 90% of reference at all times. A logistic regression analysis relating mortality to W/L indicates that between 20% and 35% of the greater mortality in years 1 and 3 can be explained by the observed changes in W/L. The remainder occurred because of an increase in underlying or background risk (which might perhaps be expected in the circumstances of generally deteriorating nutritional status).

The results suggest that child mortality is likely to increase before area mean W/L falls to 90% of reference, indicating that emergency interventions should be triggered earlier than at present.