Validation study of a verbal autopsy method for causes of childhood mortality in Namibia.

Verbal autopsy uses a caretaker interview to determine the cause of death. We conducted a study of the major causes of child death in Namibia to determine the validity of this method. A questionnaire, including signs and symptoms of the diagnoses of interest was administered to the caretaker in 135 deaths of children < 5 years old who were identified from hospital records. The 243 diagnoses included malnutrition (77), diarrhoea (73), pneumonia (36), malaria (33), and measles (24). Sensitivity and specificity of various algorithms of reported signs and symptoms were compared to the medical diagnoses. An algorithm for malnutrition (very thin or swelling) had 73 per cent sensitivity and 76 per cent specificity. An algorithm for cerebral malaria (fever, loss of consciousness or convulsion) had 72 per cent sensitivity and 85 per cent specificity, while for all malaria deaths the same algorithm had low sensitivity (45 per cent) and high specificity (87 per cent). For diarrhoea, loose or liquid stools had high sensitivity (89 per cent), but low specificity (61 per cent). Cough with dyspnoea or tachypnoea had 72 per cent sensitivity and 64 per cent specificity. An algorithm for measles (age > or = 120 days, rash) had 71 per cent sensitivity and 85 per cent specificity. The study results suggest verbal autopsy data can be useful to ascertain the leading causes of death in childhood, but may have limitations for health impact evaluation.