Applying verbal autopsy to determine cause of death in rural Vietnam.

Aims: Verbal autopsy (VA) is an attractive method for ascertaining causes of death in settings where the proportion of people who die under medical care is low. VA has been widely used to determine causes of childhood and maternal deaths, but has had limited use in assessing causes in adults and across all age groups. The objective was to test the feasibility of using VA to determine causes of death for all ages in Bavi District, Vietnam, in 1999, leading to an initial analysis of the mortality pattern in this area.

Methods: Trained lay field workers interviewed a close caretaker of the deceased using a combination closed/open-ended questionnaire.

Results: A total of 189 deaths were studied. Diagnoses were made by two physicians separately, with good agreement (kappa = 0.84) and then combined to reach one single underlying cause of death for each case. The leading causes of death were cardiovascular and infectious diseases (accounting for 20.6% and 17.9% of the total respectively). Drowning was very prevalent in children under 15 (seven out of nine cases of drowning were in this age group).

Discussion: One month seemed an acceptable minimum recall period to ensure mourning procedures were over. A combination VA questionnaire was an appropriate instrument provided it was supported by adequate training of interviewers. Two physicians were appropriate for making the diagnoses but predefined diagnostic methods for common causes should be developed to ensure more replicable results and comparisons, as well as to observe trends of mortality over time. The causes of death in this study area reflect a typical pattern for developing countries that are in epidemiological transition. No maternal deaths and a low infant mortality rate may be the result of improvements in maternal and child health in this study area. Using the VA gave more precise causes of death than those reported at death registration. Although the validity of the VA method used has not been fully assessed, it appeared to be an appropriate method for ascertaining causes of death in the study area.