D. Marsh et al., VERBAL AUTOPSY IN KARACHI SLUMS - COMPARING SINGLE AND MULTIPLE CAUSEOF CHILD DEATHS, Health policy and planning, 10(4), 1995, pp. 395-403
Background: Children dying in developing countries often have more tha
n one serious disease process. Identifying the single most important i
s difficult and risks omitting valuable information. We report pattern
s of childhood death in Karachi slums comparing single and multiple ca
use analysis. Methods: The Aga Khan University has developed primary h
ealth care programmes in Karachi slums where community health workers
maintain monthly surveillance of families with children aged under fiv
e years. We investigated all deaths among children under five from 199
0 to 1992 when the overall IMR was 78. We used a structured verbal aut
opsy questionnaire comprised of verbatim and symptom check-list sectio
ns. Reviewers assigned up to four diagnoses per death. We compared ana
lysis by single (main) cause and multiple (consolidated) causes per ca
se. Results: Overall, the 431 cases had 627 causes of death; 156 cases
(36%) had more than one cause of death. By consolidated analysis, dia
rrhoea, malnutrition, row birth weight (LBW), acute respiratory infect
ion (ARI), and vaccine preventable diseases played roles in 41%, 24%,
22%, 13%, and 2% of deaths, respectively. Proportionate mortality rati
os by main cause (1 diagnosis per child) analysis for the same diagnos
es were: 39%, 1%, 8%, 11%, and 1%, respectively. Half of all child dea
th occurred by age 3 months. Eighty-three per cent (146/176) of fatal
diarrhoea syndromes were due to acute watery diarrhoea, and 41% (72/17
6) were associated with severe malnutrition (vs 21% [12/56] of fatal A
RI cases). Each verbal autopsy cost an estimated US$4.00, an annual co
st of US$0.08 per programme child. Implications: Compared to single ma
in cause analysis, consolidated analysis more completely assesses reas
ons for child death to guide programme response. We propose that healt
h planning for populations with low vaccine coverage or common co-morb
id conditions, such as malaria and anaemia, would particularly benefit
from analysis of death seeking multiple causes.