The Sisterhood Method of Estimating Maternal Mortality: The Matlab Experience

This study reports the results of a test of validation of the sisterhood method of measuring the level of maternal mortality using data from a Demographic Surveillance System (DSS) operating since 1966 in Matlab, Bangladesh. The records of maternal deaths that occurred during 1976-90 in the Matlab DSS area were used. One of the deceased woman's surviving brothers or sisters, aged 15 or older and born to the same mother, was asked if the deceased sister had died of maternity-related causes. Of the 384 maternal deaths for which siblings were interviewed, 305 deaths were correctly reported, 16 deaths were underreported, and the remaining 63 were misreported as nonmaternal deaths. Information on maternity-related deaths obtained in a sisterhood survey conducted in the Matlab DSS area was compared with the information recorded in the DSS. Results suggest that in places similar to Matlab, the sisterhood method can be used to provide an indication of the level of maternal mortality if no other data exist, though the method will produce negative bias in maternal mortality estimates.The results are reported of a test to validate the sisterhood method of measuring the rate of maternal mortality using data from a Demographic Surveillance System (DSS) operating since 1966 in Matlab, Bangladesh. The records of maternal deaths that occurred during 1976-90 in the Matlab DSS area were used. One of the deceased woman's surviving brothers or sisters, 15 years or older and born to the same mother, was asked if the deceased sister had died of maternity-related causes. The respondents of the field survey came from 3 groups. In the first group respondents were individuals with a sister who had died of maternity-related causes during the period. The second group consisted of siblings of women of reproductive age who had died of nonmaternity-related causes. The third group comprised respondents of both sexes who did not have a sister who died of maternal or nonmaternal causes. Of the 384 maternal deaths for which siblings were interviewed, 305 (79%) deaths were correctly reported, 16 deaths were underreported, and remaining 63 (16%) deaths were misreported as nonmaternal deaths. 70 (18%) of the 384 deaths were attributed to induced abortion, 17 (4%) to spontaneous abortion, 214 (56%) to direct obstetric causes, and 79 (21%) to indirect obstetric causes. 354 (92%) of the 384 deaths were to ever-married women and the remaining 30 (8%) to never-married women. Of 70 deaths related to induced abortion, only 35 were reported to ever-married women, and no such deaths were reported as occurring to never-married women. Information on maternity-related deaths obtained in a sisterhood survey conducted in the Matlab DSS area was compared with the information recorded in the DSS. Findings suggest that in places similar to Matlab, the sisterhood method can be used to provide an indication of the level of maternal mortality if no other data exist, although the method will produce negative bias in maternal mortality estimates, especially on deaths related to induced abortion.