Why perinatal mortality cannot be a proxy for maternal mortality.

In recent years, the perinatal mortality rate (PNMR) has been proposed as a proxy measure of maternal mortality, because perinatal deaths are more frequent and potentially more easily measured. This report assesses evidence for an association between these two statistics. This study, based upon data from Matlab, Bangladesh, shows that the maternal mortality ratio (MMR) and the PNMR do not vary together over time, and that the PNMR does not reliably indicate either the magnitude or the direction of change in the MMR from year to year. Statistical analysis shows that the correlation between the PNMR and the MMR is not significantly different from zero. An examination of the major causes of maternal and perinatal deaths indicates that the two measures cannot be expected to vary together. Almost half of perinatal deaths result from causes that do not pose a threat to the mother's life, and almost half of maternal deaths result from causes that do not lead to perinatal death. Monitoring of the PNMR can give an inaccurate picture of maternal mortality and should not be used as a proxy.

Why did maternal mortality decline in Matlab?

In 1991, an article on the Maternity Care Program in Matlab, Bangladesh, reported a substantial decline in direct obstetric deaths in the intervention area, but not in the control area. The decline was attributed primarily to the posting of midwives at the village level. In this article, data are presented from the same period and area on a variety of intermediate events. They indicate that the decline in deaths was probably due to the combined efforts of community midwives and the physicians at the Matlab maternity clinic. Their ability to refer patients to higher levels of care was important. The data further indicate that the decline in deaths depended upon the functioning of the government hospital in Chandpur, where cesarean sections and blood transfusions were available. Midwives might also have made a special contribution by providing early termination of pregnancy, which is legal in Bangladesh.

PIP: Data were collected during 1993 from the Matlab Demographic Surveillance System, midwives' cards, the Matlab maternity clinic record book, and records at the government's district hospital. This study analyzed these data in order to determine why maternal mortality declined in the intervention area. Direct obstetric deaths declined from 20 deaths during the 3 years before the Maternity Care Program was implemented to 6 deaths during the 3 years after program implementation. In the control area, mortality remained stable at 20 deaths during the same time period. The declines were apparent among diagnoses for induced abortion, eclampsia and pre-eclampsia, and prolonged obstructed labor. There were few changes in causes of death in the control area. During the intervention period, midwives in villages administered services to 49 women with preeclampsia, of whom 6 later developed eclampsia. There were 20 referrals to the maternity clinic for a variety of complications including eclampsia. The Matlab clinic received 54 patients during the intervention period with a primary diagnosis of pre-eclampsia or eclampsia. Midwives provided care for 77 women with prolonged labor. Matlab clinic received 116 patients due to prolonged labor. 4% of the 2364 midwives' cards indicated referral. 83% of referrals were to the Matlab clinic. The Matlab maternity clinic had 300 admissions during the study period, of which 65% (194 women) were from the intervention area. Women from the intervention area were 2.3 times more likely to be treated at the Matlab clinic than women from the control area. 69% of admissions at Chandpur District Hospital were from the Matlab intervention area. Case fatality rates in the hospital did not differ among intervention and control populations. The authors conclude that greater use of midwives, referrals and proper transport, and better service conditions significantly contributed to maternal mortality decline.