Assessments of mortality, morbidity, and nutritional status in Somalia during the 1991-1992 famine. Recommendations...

To evaluate the various survey methods used in Somalia between 1991 and early 1993 while assessing documentation of mortality and malnutrition rates and common causes of morbidity and mortality.

Twenty-three population surveys were identified from the Center for Public Health Surveillance for Somalia, the United Nations Children's Fund, and other humanitarian organizations.

Only surveys with defined populations and apparently systematic methodology that focused on mortality, morbidity, and/or nutritional status were included.

Extensive methodological differences were found among the 23 surveys. Target populations and sampling strategies varied widely. Twelve studies were considered not reproducible. Of the 16 studies assessing mortality, only eight assessed cause of death. Use of units of measurement and inclusion of denominators in rate calculations were inconsistent. None of the studies provided confidence intervals around the point estimates of the rates. Of the 11 studies providing information on morbidity, none provided case definitions. And in the 16 studies reporting nutritional status, a variety of measurement methods and definitions of malnutrition were used. Three studies presented information based on mid-upper-arm circumference measurements, and 10 presented weight-for-height data below 70% and 80% of the reference median; only four studies presented z scores.

While the results of some studies may have influenced policy and program management decisions, their effects may have been limited by failure to adequately document results and by differences among studies in objectives, design, parameters measured, methods of measurement, definitions, and analysis methods. We recommend that agencies conducting population studies in emergency situations define clear study objectives, use standard sampling and data collection methods, and ensure precise written documentation of study objectives, methods, and results.