in the systems of care available and to a large gap between what facilities and life-saving skills are available and what is needed.
Indicators of the quality of emergency care for obstetric complications at a facility include the time interval between admission to treatment, facility trends in case-fatality rates for all complications, the caesarean section rate, and trends in numbers of deaths (maternal and perinatal); the proportion of perinatal deaths contributed by stillbirths or early neonatal deaths and proportion contributed by full-term babies may also be useful. Such data have only recently begun to be collected in most places. Medical records are typically the source of such information, and they are notoriously lacking or incomplete. Medical or verbal autopsies of maternal deaths, in which causes of death and avoidable factors are determined by a medical team that reviews each death, have proven to be a very useful way to monitor deficiencies.
Trends in the time from admission to treatment for emergency obstetric cases were observed in three referral sites in two African countries, Ghana and Nigeria. From a review of medical records, it was found that waiting time had increased from an average of 5 hours in 1983 to 15.5 hours in 1988 in Calabar (Ghana) for women who died of hemorrhage. In Zaria (Nigeria) the waiting increased from 3.5 to 6.9 hours between 1983 and 1988 (Prevention of Maternal Mortality Network, 1995).
Although data on admission to treatment for most complications is often lacking, caesarean sections require an operating theater; typically, the dates and times of procedures are recorded. The average admission-to-treatment interval for emergency caesarean sections for six referral facilities in three districts of India was 1.5 to 5 hours. These long intervals were caused by the need to locate doctors who were on call and time for them to come to the facility. In another setting, a delay was caused by a patient's relatives having to purchase anesthetics outside the facility. Case fatalities in these sites ranged from 2.8 percent to 6.9 percent. Low case fatality rates were found in some cases because the most serious cases were referred to another facility or left the facility against medical advice.
In Quetzaltenango, Guatemala, only 8 percent of patients had no wait upon arrival at the referral hospital; 45 percent waited up to 1 hour, and 47 percent waited more than 1 hour. Women in a study indicated that they would not accept referrals from traditional birth attendants to hospitals because of long waiting periods (O'Rourke, 1995).
Most incriminating of the quality of care is a nationally representative study of 718 maternal deaths in Egypt in 1992 (Ministry of Health, 1994). Avoidable factors were assigned by a local advisory group of a panel of doctors for each governorate that met weekly to review each maternal death in that area; their decisions were reviewed by a central advisory group at the national level. The leading avoidable factor was poor management