APPENDIX B Examples of Programs to Promote Safe Pregnancy and Delivery
This appendix describes several projects that involve interventions at each step of the process of care for obstetric complications outlined in Chapter 5. We present these descriptions because they illustrate concretely some of the principles discussed in the chapter, because many of the reports are not widely available, and because the published literature still contains few evaluations of outcomes.
STEP 1: RECOGNITION OF THE PROBLEM
Can behaviors be changed by making families aware of the signs of complications and where to go in response to them? In one project, women's groups and community strategy meetings were organized in rural areas of Bolivia. Two other examples focus on information, education, and communication efforts in urban areas in Latin America where women are already more familiar with the available services.
Women's Groups in Inquisivi, Bolivia, 1989-1993
A population of 15,000 people, primarily of Aymara heritage, and living in a remote and difficult area, separated by high plains, high Andean valleys and subtropical valleys, is served by three health posts that lack basic equipment, supplies, and skills to manage obstetrical or newborn complications. The maternal mortality ratio (over 1,000 maternal deaths per 100,000 live births) and perinatal mortality ratios (103 perinatal
deaths per 1,000 live births) reflect the isolation and poverty of these people (Howard-Grabman, Seoane, and Davenport, 1993). Women with complications would have to be referred to La Paz or Oruro, the two closest cities, which takes 4 to 6 hours over difficult roads. Traditional birth attendants do not exist, although traditional healers are sometimes consulted when complications arise. Husbands are the main providers of assistance in labor and delivery.
A simple model for community-level problem solving was implemented, consisting of four steps:
identification and prioritization of maternal and neonatal health problems ("auto-diagnosis") by 50 women's groups;
development of strategies and actions to solve the problems by the women's groups and members of their communities and local authorities;
implementation of the groups' plans which involved: training 45 parteras (empirical midwives); educating both women and men in management of hemorrhage; providing family planning services in seven communities in collaboration with a local nongovernmental organization; developing and distributing a home-based women's health card, a manual for parteras, four booklets for women on reproductive health and five radio programs with a local nongovernmental organization;
Over one-half of women in the 52 communities participated in this process between 1991 and 1993. As they and their families became more involved, health practices and service utilization changed: tetanus toxoid immunization coverage, iron tablet distribution, immediate breastfeeding, consumption of iodized salt, prenatal and postnatal care visits, and the number of women attended by a trained birth attendant increased.
The intervention was evaluated using a pre- and postintervention treatment and comparison-area design, with data collected from a village-based information system. The most striking result was a reduction in the number of perinatal deaths, from 75 to 31 deaths over a 2-year period. This decrease was likely due to more immediate neonatal care. For each delivery, a person was designated to take responsibility for drying and warming the child before the placenta was delivered and placing the baby on the mother's breast within the first hour of life. (Previously, the newborn baby was set aside until after delivery of the placenta.) Use of modern contraceptives rose from 0 to 27 percent, reflecting that several of the groups had identified "too many children" as their main health concern (Howard-Grabman, Seoane, and Davenport, 1993).
Cochabamba Reproductive Health Project, Bolivia, 1990-1993
Just over one-half the population of Cochabamba, the third most populous department of Bolivia, is urban. Rapid growth of the urban population is due to the migration of Quechua and Aymara people from both rural parts of the department and the rest of the country. The maternal mortality ratio is very high (480 deaths/100,000), as are the infant mortality rates (126/1,000) and perinatal mortality rates (110/1,000) (Bower and Perez, 1993). The major causes of maternal mortality are complications of induced abortions and the poor conditions under which women have home births and even institutional deliveries.
Traditional health beliefs are dominant: only 13 percent of the urban population exclusively use the Western medical system. In 1991, only 34 percent of pregnant women ever received prenatal care during their pregnancies, and only 13 percent had four prenatal visits, the number stipulated by national norms.
Working with the local public health authority and several local nongovernmental organizations offering health services, the MotherCare program sought to increase demand for reproductive health services, improve the quality of those services and, consequently, contribute to an eventual decline in maternal and neonatal mortality in an area of approximately 500,000 people. A formative study of women's reproductive health knowledge, attitudes, and practices was conducted. The findings of the study were used to develop intervention strategies to improve home practices, increase the appropriate use of formal health services, and train health care providers to offer services more respectful of women's needs and wishes.
A health communication plan was initiated in three phases: (1) sensitization, aimed at creating awareness among policy makers and health providers of the problems of maternal and neonatal health and the differences in perspectives on health care between Quechua-Aymara peoples and the formal health system; (2) prenatal care, to create an awareness of the importance of routine and emergency prenatal care; and (3) safer/cleaner home delivery, promoting the use of sterile materials to cut and tie the umbilical cord, the recognition of and response to obstetrical complications, the avoidance of labor augmenters, and safe delivery of the placenta.
Each phase lasted 3 months and used educational video and radio programs covering the general theme, with television and radio spots for each major subtheme and one flip-chart per subtheme (with instructional guides) that was provided to families. Representatives of each participating agency attended workshops in health communications skills to institutionalize capability to develop and implement the various phases.
Training was conducted with providers to upgrade their skills and to adapt case management procedures to the needs of clients.
According to a household survey, in participating clinics prenatal care attendance rose by 17 percent to more than 100 percent, compared with a baseline period 2 years earlier, although there was no significant increase in the Cochabamba department as a whole (Bower and Perez, 1993). Women who could identify one of the danger signs increased from 26 percent to 43 percent. The percentage of women who could identify edema (traditionally thought to be a positive sign promising an easy birth) as a danger sign rose from 2 percent to 64 percent.
By various means (mass media, group participation, interpersonal communication), pregnant women were empowered to go to their prenatal care providers knowing specifically why they were there and what questions to ask about their own babies' health. This approach was developed in response to the need for information that many pregnant women complained about when interviewed in the qualitative and baseline studies. Women had pointed out that during prenatal care they did not receive all of the information that they wanted in order to leave confident that their pregnancy was progressing normally.
Psychosocial Support for High-Risk Pregnancies, 1989-1991
A highly structured home visit program in four Latin American cities—Rosario, Argentina; Pelotas, Brazil; Havana, Cuba; and Mexico City, Mexico—evaluated the effect of psychological and social support and health education for women at high risk for poor pregnancy outcomes (Villar et al., 1992). A total of 2,235 women at higher-than-average risk for poor pregnancy outcomes were recruited before their 20th week of pregnancy. All women planned to and did have an institutional delivery.
Through random assignment to an intervention or control group, women received either (1) four to six home visits from a nurse or social worker in addition to routine prenatal care or (2) only routine prenatal care (with a mean of eight prenatal visits). The main objective of the home visits was the strengthening of a pregnant woman's social network. Beginning with the first visit, a support person (husband or partner, mother, sister, friend, or neighbor) was selected by the patient to share all intervention activities. The support person was strongly encouraged to remain involved with the woman throughout the pregnancy, to participate in the decision-making process, to help the woman resolve personal problems, to promote healthful behavior, and to encourage attendance at visits for prenatal care. A second objective of the study was to provide health education during the home visits, including education on nutrition, reducing smoking, alcohol, or drug use, and complications during
pregnancy, labor, and delivery; preparation for delivery and the postpartum experiences; and what to do in case of emergencies. A third objective was to reinforce adequate health services utilization for the pregnant woman and her support person. Complementing these activities were efforts to make the hospital more accessible to women—a hot-line to respond to questions, no appointment required for visits, and a guided tour of the obstetric facility to familiarize the client with labor, delivery, and emergency services.
The result of the program was virtually no change in behavior or outcomes. The women who received the home visits as well as routine prenatal care had outcomes that differed little from those of the women who received only routine care. The risks of low birth weight, preterm delivery, and intrauterine growth retardation were similar in the two groups (Villar et al., 1992). There was significant improvement in the knowledge of seven of nine danger signs considered and in two of three labor onset signs. But there was no evidence that the intervention had any significant effect on specific behaviors—the type of delivery, length of hospital stay, or use of health services. There was no protective effect of the psychosocial support program, even among the mothers at highest risk.
That improvements in knowledge did not translate into changes in health-related behaviors, such as improved use of health services, may have been due to the short duration of the intervention—3 months—and to continued barriers to access (Belizan et al., 1995). Women's behaviors during pregnancy and delivery may not be easy to change with short interventions.
STEP 2: MAKING THE DECISION TO SEEK CARE
The Perinatal Regionalization Project, Tanjungsari, West Java, Indonesia, 1989-1993
The Perinatal Regionalization Project aimed to develop a comprehensive maternal health program to improve maternal and perinatal health outcomes. Located in a rural subdistrict of West Java, Tanjungsari has a population of 90,000; it is served by three government health centers with midwives (one of which has delivery beds for normal births) and a district hospital 40 kilometers away in an adjacent subdistrict. Villages in this mountainous area are connected by roads, but some are impassable even for a four-wheel-drive ambulance regardless of season. While the health infrastructure is well utilized by women for prenatal care and children's health, traditional birth attendants are the primary providers during the intrapartum and postpartum periods.
The program consisted of three interventions:
The first part was to train traditional birth attendants in the recognition and referral of maternal and newborn complications.
The second part was to establish community birthing homes (polindes ) at the village level in 10 of 27 villages in the intervention subdistrict. Each birthing home was to serve as the practice base for a community nurse-midwife and as residence for her if at all possible. (The Indonesian government has mandated the training of one midwife, a bidan di desa , to be available per village throughout the country.) Equipped with beds, scales, instruments and medical supplies, prenatal and delivery services for normal births were available in the polindes; on a monthly basis, child health, immunization, and family planning services were also provided. A two-way radio was installed in each of the ten polindes , the three health centers in the subdistrict and the district hospital. In case of a delivery, the traditional birth attendant and a community worker might manage the woman together at the birthing home unless a midwife was available or needed by the traditional birth attendant. The radios could be operated by the traditional birth attendants, community workers, or village headman and his wife to call for assistance from a doctor or midwife or for an ambulance.
The third part was to train for maternal providers at all levels of the system, informal and formal. The traditional birth attendants had been previously trained, but their knowledge and skills were reinforced through practice with the midwives and monthly discussions at the health centers with doctors and midwives. Once maternal and perinatal death audits began, a physician visiting the families in the village would also discuss cases with the traditional birth attendants. Both the health center doctors and midwives had continuing education courses provided by consultants and senior midwives in the district hospital. The specialist obstetric staff of the nearby medical school collaborated with the consultant obstetrician and pediatrician of the district hospital in the formulation of management protocols for hospital, health center, and the midwives in the polindes.
Following the establishment of the birthing homes, an information, education, and communication strategy was implemented in the last 6 months of the project to improve the awareness and responsiveness of women, their husbands, community leaders, and traditional birth attendants concerning the purpose and use of the birthing huts (for routine care) and timely recognition of danger signs and referral. Parades, contests of knowledge among traditional birth attendants, and "open house" days at the polindes, where the ambulance and two-way radio services
were demonstrated, brought messages directly into villages. Promotional leaflets for pregnant women, husbands, and traditional birth attendants addressed several issues:
the importance of attending prenatal care, especially when feeling sick or experiencing one of the pictured conditions (bleeding, swollen face, feet, and hands, fever with chills, headache, and breech or transverse lie);
where to get prenatal care;
why to use a birthing hut for prenatal and delivery care; and
when to take the first step to a place where a pregnant woman has decided to deliver.
The program was evaluated through interviews with all 2,275 women who had a singleton birth during the 15-month project period in 1992-1993 in the intervention area and all 1,000 new mothers in the control subdistrict of Cisalak where public services continued as usual.
Approximately 32 percent of women reported complications during the antepartum and intrapartum periods, while 29 percent suffered problems during the puerperium. The vast majority of deliveries in both the study and control areas were attended by traditional birth attendants (86 and 87 percent, respectively). The site of delivery was, however, different: in Tanjungsari, 85 percent delivered at home, compared with 96 percent in Cisalak. Much of this difference had to do with where women with complications sought care: in Tanjungsari, 31 percent (120 of 390 women) with delivery complications delivered in a health facility; in Cisalak, 11 percent did so.
How much of this use of health facilities for delivery because of complications was due to the recognition and referral of traditional birth attendants, the primary care providers? Unfortunately, the overall rate of intrapartum referral of women with complications during either the prenatal or intrapartum periods was low—13 percent in Tanjungsari and 6 percent in Cisalak. For those who suffered in the intrapartum period, 18 percent were referred in Tanjungsari and 9 percent in Cisalak. More women with ''no complications" as perceived by themselves, or with only prenatal complications, were referred in Tanjungsari as well, possibly a precautionary measure made by traditional birth attendants trained in recognizing danger signs of complications.
Compliance with referral in these areas was relatively high, with the rate of compliance for intrapartum referral higher in Tanjungsari (87 percent) than in Cisalak (69 percent). Approximately 65 percent (98 cases) of complicated cases that were delivered in facilities in Tanjungsari had been recognized and referred by a traditional birth attendant, compared
with only 14 complicated cases delivered in facilities in Cisalak. The traditional birth attendants' initial enthusiasm for making referrals dissipated over time. In focus groups, some of them expressed concern about undermining their status by referring cases to higher levels of care. They were afraid that women or other traditional birth attendants would say that they were unable to handle complications on their own. Several said they would refer only if they felt the family would comply.
How much did the program change the birthing pattern? The evaluation showed that women's knowledge of complications during pregnancy, labor, and delivery was higher in the intervention area than in the control area and that knowledge was more complete following the communications campaign. Use of birthing huts for prenatal care also increased following the campaign, from 9 to 26 percent (Hessler-Radelet, 1993). But use of delivery care proved more resistant to change. Focus group discussions and observations revealed that:
Women did not care for take-home materials that depicted danger signs and complications, as they believed these illustrations would be prophetic.
Women also felt that knowing about danger signs was important, but thought, if the danger sign (e.g., edema) didn't hurt or was a common occurrence during pregnancy and disappeared after delivery or didn't inhibit everyday chores, it could not be very dangerous.
Traditional birth attendants felt uncomfortable referring more than the "usual" number of clients to a birthing hut; they felt it undermined their authority in the eyes of the community. However, the community generally respected their decisions in terms of referral for danger signs and complications (Winnard, 1993).
Indonesian women do not like to think about negative events or plan ahead for them. They believe that planning for a negative event will disrupt their sense of inner calm and may cause that negative event to occur. Cost, distance, and the desire to remain privately at home near family members are the reasons most cited by women for not accepting referral for a delivery place outside the home. Changing time-tried patterns of birthing will require more than 6 months of information and education, although use of prenatal services appears fairly easy to increase.
In villages in Tanjungsari with polindes, 19 percent of all pregnant women (and 37 percent of those with complications) delivered in health facilities, compared with 12 percent (and 26 percent of those with complications) in villages without them. Women do not appear to use the birthing huts for delivery; rather, the effect of the huts may have been to
familiarize women with formal health sector personnel and to provide the means to better access other health facilities during labor and delivery (Kwast, 1995).
In summary, women with complications in Tanjungsari were more likely to be referred and to deliver in a health facility with a doctor than their counterparts in Cisalak, although most women in both sites, with or without complications, still delivered at home with a traditional birth attendant. Perinatal death rates were not significantly different between the two sites, though the numbers of births were small enough that this statistical test had very low power (99 perinatal deaths and 2,275 births in Tanjungsari; 37 perinatal deaths and 1,000 births in Cisalak). More women with intrapartum complications (especially prolonged labor) in Tanjungsari than in Cisalak gave birth in the hospital with a doctor. Yet deaths in the hospital actually increased, which suggests either that cases arrived too late for care or that the care was not adequate.
The Quetzaltenango Project, Guatemala, 1989-1993
In rural Guatemala, traditional birth attendants deliver 70-90 percent of infants. Obstetric skills are only available at hospitals at the department level, although prenatal and child care are available at health posts and centers. Midwifery training was abolished 20 years ago in Guatemala, so delivery care is primarily in the hands of traditional birth attendants.
In the intervention area of rural Quetzaltenango, the Institute of Nutrition of Central America and Panama (INCAP) trained government nursing staff to train over 400 traditional birth attendants in 3-day sessions in the recognition and timely referral of maternal and newborn complications, including bleeding, swelling of hands and face during pregnancy, malpresentation, prolonged labor, retained placenta, depressed newborn, postpartum bleeding, and maternal and neonatal infections. Working with doctors at the referral hospital and health personnel at health centers and posts, protocols for recognition and referral or management of complications were developed and implemented. A strong emphasis was placed on fostering good working relationships between the different levels of health care and improving humane treatment of both women and traditional birth attendants by health staff. Posters were also placed in all health facilities reminding health care providers that everyone in the community is trying to save the lives of women and their newborns on the difficult road from home to the formal health care facilities.
The outcome of this interaction was mixed. One year after the traditional birth attendant training, community-level behaviors had not
changed significantly. Through interviews with a sample of recently delivered women in target communities, levels of complications and use of health services were determined retrospectively in both intervention and control areas (total population 165,000) and before and after the interventions. Levels of complications were the same before and after the interventions, except for postpartum complications, which decreased from 6 percent to 2 percent. Approximately 20 percent of all women reported they suffered complications during the prenatal, labor and delivery, and puerperium, and one in ten newborns was said to have a complication. Nearly one-half of the women with maternal complications used health services even before the interventions, and this percentage did not significantly increase after the interventions (49 percent vs. 52 percent) (Bailey, Szaszdi, and Schieber, 1994). Approximately one-fifth of the women with complications who used health facilities did so after referral by traditional birth attendants. After the interventions, this proportion decreased to 13 percent (Bailey, Szaszdi, and Schieber, 1994), but more women were going to health facilities directly after the interventions. Traditional birth attendants did not recognize complications (recognition declined from 81 percent to 68 percent), did not refer the women, or the woman stated she did not accept her referral but went to a health facility anyway.
More than four out of five women saw a traditional birth attendant for prenatal care, regardless of whether they experienced a complication. Intrapartum and postpartum complications seemed to signal more danger to families: 52 percent and 65 percent of women with such complications, respectively, who used services did so directly without consulting a traditional birth attendant. Fetal and newborn complications caused women to go directly to health facilities (Bailey, Szaszdi, and Schieber, 1994).
The effect of the community-based interventions appears negligible because traditional birth attendants were already referring a significant proportion of complicated cases they saw before the interventions and because women with complications were already going to health facilities without consulting a traditional birth attendant. These data suggest an independence on the part of the woman and her family in seeking out services when they determine there is a complication.
Women in this study area are not that distant from the referral hospital; approximately 42 percent are within 60 minutes of the hospital. Roads are relatively good, and transportation is available. Thus, the use of services for women with maternal complications appears to be fairly normal behavior, although uncomplicated births at home with a traditional birth attendant is the common procedure in Quetzaltenango.
The interventions had more effect at the hospital. Patient satisfaction (measured as the percentage stating they would return) significantly increased.
Delays between admission and treatment decreased significantly. A medical audit revealed that the case management in hospitals of newborns had improved over the life of the project. Most women admitted with prolonged labor from home were subsequently managed correctly in the hospital. But late referral by a traditional birth attendant or late recognition and decision making by the family, and to a lesser extent lack of hospital supplies (lack of oxygen during one of the project years), continued to plague the project (Schieber et al., 1995). Perinatal mortality also decreased, though the difference was not statistically significant.
Program for Reduction of Maternal and Perinatal Deaths, Brazil, 1975-1984
A program in rural Northeast Brazil was intended to reduce maternal and perinatal deaths by ensuring prompt referral of complicated pregnancies and deliveries to a teaching hospital in the city of Fortaleza. Mini maternity units were set up throughout the region and traditional birth attendants were trained, some of whom were chosen to work in the units. The 40 units varied in size, resources, and services: the largest had 8-10 beds and could offer a wide variety of services, including prenatal, delivery, and postpartum care; the smallest consisted of a single room adjoining traditional birth attendants' homes where services were limited to normal delivery and postpartum care. Training of all traditional birth attendants consisted of five 1-hour meetings and practical experience at a maternity unit to ensure that each could provide prenatal care, identify problem pregnancies, and assist in normal deliveries and postpartum care for mother and newborn. They were taught to refer women who had a prenatal problem, especially eclampsia or hemorrhage; a complicated labor, including placenta previa or abruptio; a malpresentation and cases of cord or limb prolapse; and to consider for referral women under age 19 or over 35. The program included supervision and instruction by the teaching hospital staff as well as local hospital staff.
In one of the program counties, Trairi, which is far from Forteleza with difficult transportation and impassable roads in the rainy season, four mini maternity units were established and 78 traditional birth attendants trained for a population of 30,000. In 1984 (10 years after the program was begun), 64 percent of deliveries took place at home, primarily with a traditional birth attendant, and 36 percent in hospitals. Of the 10 percent of all women with singleton births who suffered complications in labor, 93 percent delivered in a hospital, with 49 percent referred by traditional birth attendants and 51 percent self-referred (Bailey et al., 1991). The traditional birth attendant referrals were more often linked with a
complication: 52 percent of their referrals were diagnosed by hospital personnel as having a complication, but only 18 percent of the "walk-ins" actually had a medically diagnosed complication (Bailey et al., 1991). This may have been because women who themselves identified a complication were more likely to go to a traditional birth attendant, who then referred them to the hospital (Janowitz et al., 1988). That trained traditional birth attendants could distinguish severity of conditions and refer appropriately is indicated by the prenatal pathologies referred: 10 of 15 women with hemorrhage or hypertension were referred, while only 1 of 12 women with fatigue, nausea, dizziness or vomiting was sent to hospital (Janowitz et al., 1988).
Assistance at birth did not have a significant impact on infant survival in home deliveries. However, the odds of dying for a baby delivered by a traditional birth attendant with a high caseload (>29 births per year) were 0.6 that of a home birth not delivered by a traditional birth attendant. Although there was also no significant association between survival and attendant or place of delivery, perinatal mortality was lowest for deliveries at the mini maternity units with trained traditional birth attendants.
The authors concluded (Janowitz et al., 1988:56):
Providing one-bed obstetric units for the busiest traditional birth attendants appears to be an intervention that can reduce mortality. The construction of additional units would further concentrate deliveries and make supervision easier than overseeing the work of 78 traditional birth attendants, some of whom attend only a few deliveries a year.
While supervision may be made easier, mini maternities still do not attract many deliveries: even 10 years after the initiation of the program, only 1 in every 10 women delivered in the mini maternities. Although not statistically significant, mortality among deliveries to women with complications of labor or malpresentations was higher for home than for hospital deliveries. The authors concluded that the mortality could have been lowered with earlier and more frequent referral of problem cases, an improved transportation system, and possibly an improvement in the local hospitals so that they did not have to send out referrals for cesarean section (a little over 1 percent of births were referred from a local hospital to the teaching hospital) (Janowitz et al., 1988).
Other Training Efforts
In contrast to the Fortaleza experience, the Danfa Project in Ghana found that traditional birth attendants were reluctant to refer women to hospitals. The Danfa Project area is located 30 kilometers north of Accra,
a busy metropolis with 6 hospitals and 68 maternity homes. Yet 93 percent of deliveries in the project area were assisted by traditional birth attendants, even 15 years after the project began with an emphasis on recognition of complicated cases and referral. While the trained traditional birth attendants could appropriately identify what constitutes a "high-risk" case (e.g., primiparas; short, previous cesarean section; twins; malpresentation) and stated they would refer them, many reported routinely performing such deliveries themselves. The only major complication which most (74 percent) stated they would refer directly was eclampsia, while bleeding during pregnancy or in the postpartum period would be first treated with herbs in two-thirds of the cases (Eades et al., 1993).
Reasons for women's preference for the traditional birth attendant even when midwives and physicians are available include not only the cost, distance, and lack of supplies and equipment at the center or hospital, but also fear of anticipated treatment in the hands of medically trained staff. Traditional birth attendants reported that patients feared painful and disrespectful treatment from hospital personnel.
In an urban study in Benin, Sargent (1985) found that clients and traditional birth attendants share similar beliefs, values, and ideas about the cause of illness; hence, even in a city, assisted deliveries were more the norm than the exception. The traditional birth attendants' duties included traditional healing activities, which patients found valuable. The perceived sociocultural similarity may explain many women's preference for a traditional birth attendant even when modern facilities are accessible.
STEP 3: REACHING CARE OF ADEQUATE QUALITY
Maternal mortality fell in Sweden in the eighteenth and nineteenth centuries, unlike in other European countries, apparently because of home visits by certified midwives (Hogberg and Wall, 1986). There have also been promising results of prenatal screening of demonstrated risk factors and identification of danger signs by joint efforts of midwives with traditional birth attendants at the health center or community levels in Ethiopia and Nigeria (Brennan, 1992; Poovan, Kifle, and Kwast, 1990). Several attempts have been made to bring services closer to women through midwifery outreach or to bring women closer to services through maternity waiting home located close to hospitals.
The Matlab Project, Bangladesh, 1987-1989
In Matlab, a rural subdistrict of Bangladesh, an effective community-based maternal and child health and family planning project was introduced in late 1977. The key service providers are the female village health
workers. They offer a choice of contraceptive methods at the home of each woman, motivate and counsel mothers for family planning, monitor and manage adverse effects, administer vaccines, promote oral rehydration, distribute vitamin A capsules, provide nutritional education, detect and refer malnourished children, and distribute safe delivery kits and iron tablets to pregnant women. They refer severely sick mothers or children to one of four decentralized outposts staffed by female paramedics or to the central Matlab clinic where at least one female physician is always available.
In 1985, when the 10-year effects of this project were reviewed, the maternal mortality rate in the comparison area was roughly twice that of the treatment area (121 versus 66 per 100,000 women). However, differences in the maternal mortality ratio, which measures only the obstetric risk, were not substantial, remaining around 550 per 100,000 live births (Koenig et al., 1988). This result suggests that the most important reason for the difference in maternal mortality rates between treatment and comparison areas was the intensive family planning program. By 1985, the contraceptive prevalence rate in the Matlab treatment area had reached approximately 44 percent, while the level in the comparison area was close to the national average of 17 percent (Fauveau, 1991).
A community-based maternity care project with referral links and transportation to a local and district hospital was added to the Matlab Project in part of the treatment area in 1987. One trained nurse midwife was posted at each of the two health centers in the study area. Their duties were to attend as many home deliveries as possible and to complement activities of traditional birth attendants, manage obstetric complications at their onset, and accompany patients requiring referral for higher level care to the project's central maternity clinic. The community health and family planning workers identified pregnant women for the midwives, linked them for prenatal visits, and notified the midwives when labor began. The intervention also included arrangements for emergency transportation to the maternity center in simple "ambulances" (country boats dedicated to the project). A female physician was available 24 hours a day at the Matlab Maternity Center. This hospital was equipped to provide vacuum extractions and dilation and curettage and to manage pre-eclampsia and eclampsia, but cases requiring blood transfusion or surgery were referred and transported to the district hospital an hour away (Maine et al., 1996).
Outcomes were measured by comparing maternal mortality ratios in the maternity care intervention area and the rest of the treatment area (that is, both areas benefitted from the intensive home visit program started in 1977). During the 3 years prior to the maternity care intervention (1984-1986), there was no significant difference between the two areas
in maternal mortality rates or ratios. During the 3 years after the expanded project was implemented (1987-1989), the maternal mortality ratio in the intervention area had fallen by 68 percent (which was statistically significant). The causes of death that were reduced by the program were, in order of importance, the complications of abortion, postpartum hemorrhage, postpartum sepsis, and eclampsia. Other causes of adult female mortality were constant over the project period. Although abortion was not a specific focus of the project, a decrease in abortion-related mortality may have been due to provision of early abortions by manual vacuum aspiration ("menstrual regulation," which is legal in Bangladesh) or early intervention in cases of abortion complications (Fauveau et al., 1991; Fauveau, 1991; Maine et al., 1996).
A common problem in evaluations of complex interventions is isolating the "active ingredients." The Matlab maternity care project included the posting of trained midwives to clinics. It also included using well-established family planning and health workers in the community, an ambulance service, a maternal and child health hospital staffed by physicians at all times, a district hospital capable of providing the range of obstetric services discussed in Chapter 5, and good referral links among all the providers. Its apparent success holds promise for the mixed strategy of community- and hospital-based efforts we have recommended, but a great deal of further work is required to determine how to adapt this model in other settings (Maine et al., 1996).
Indonesian Bidan di Desa Program
The government of Indonesia began training and posting certified midwives (bidan di desas) in each of the country's 64,000 villages in 1989. The bidan di desas have multiple functions: provision of health services for the community in the home including handling deliveries, family planning and screening for clinical contraception, provision of assistance to the traditional birth attendants, and detection of complications and referral. Following technical training, a 1-week orientation may be given at provincial level for the bidan di desa followed by a 1-4 month orientation at her assigned health center. The problems in large-scale implementation of this program have become apparent in its first 5 years.
One study found that the bidan di desas are often filling in on administrative tasks for understaffed health facilities. In addition, many of them expressed reluctance to be assigned to remote areas (Radyowiyati and Sequeira, no date). General acceptance of this new cadre into the fabric of village life remains an obstacle, especially for bidans who are not from the areas where they are posted. A successful posting also depends on the enthusiasm and receptiveness of village leaders, of both the men
and their wives. Yet often there is little communication with villagers about the new health worker's role. Not surprisingly, there has been competition between bidan di desas and traditional birth attendants, who currently attend most deliveries in homes. The new bidans are finding it difficult to attract business: they are typically younger and less experienced than are traditional birth attendants; they charge higher fees; and the traditional birth attendants perform many additional valued services that the bidans do not, such as massage and participation in ceremonies and rituals (Radyowiyati and Sequeira, no date).
In a study with direct observation of 58 bidan di desas, they were found to spend most of their time providing care to children under 5 and first aid for adults in Central Java and prenatal care in South Sulawesi; very little of their time was spent attending births. They often had only a blood pressure gauge, stethoscope, and a room with a bed for examinations; only a few had any medications or injection sets needed for stabilizing or managing obstetric emergencies (Achadi et al., 1994). Some stated they did not feel competent to deal with complications of delivery or other emergencies, but their ability to treat minor health problems has opened doors to acceptance by the communities.
Many adjustments to the bidan di desa program are presently being planned, including changes in the initial technical training to incorporate a focus on how to approach communities and possible continuing education to increase skills in managing common diseases such as diarrhea and acute respiratory infections, as well as education to enhance midwifery skills and a rotation for training at an assigned health center or hospital to maintain skills.
Maternity Waiting Homes
Maternity waiting homes are designed to overcome the difficulties of emergency transportation over long distances by women experiencing life-threatening emergencies in labor, by bringing women within reach of a hospital before labor begins. A recent report from Attat in rural Ethiopia shows that, with community consultation and participation, women at high risk will come to stay in a suitable waiting home near the hospital in the last month of pregnancy (Poovan, Kifle, and Kwast, 1990). Of the 151 women who used the home in its first year, 111 were at high risk because of previous cesarian sections or poor obstetric histories, including ruptured uterus and other emergencies: over one-third came from more than 40 kilometers away. Hence, appropriate use was being made of the waiting home.
Another study of more than 5,600 women who delivered singleton full-term infants in a rural district hospital in rural Zimbabwe compared
women who used a maternity waiting home with women who were referred by a traditional birth attendant from home (control 1) or from home or clinic directly (control 2). Women with obstetric risks (such as no births or more than six previous poor outcomes) made up 56 percent of the total study population. Among these women, the perinatal mortality rate of the maternity home group was nearly 50 percent lower than that of control group 1 and even lower compared with group 2 control. However, for all women who stayed at the home, the perinatal mortality rate was not significantly reduced (Chandramohan, 1992).
Financing Maternity Care
Fees for services may be a barrier to using services, especially for the poor. But fees can be manipulated to stimulate desired behaviors among both providers and clients. In Zimbabwe, for example, pregnant women are charged a flat fee for prenatal care, plus a regular delivery and inpatient per diem fee for an in-patient delivery. Kutzin (1993) reports that following intensification of these fee collections, one district hospital had an increase in the numbers of women arriving having just delivered. Staff assumed this was because mothers were waiting until the last minute in order to avoid additional bed fees. Yet in another municipal hospital in Zimbabwe, fees were ''bundled" so that a woman paid a flat fee for prenatal care, delivery, emergency transport, if needed, and postnatal care. This scheme encourages women to seek prenatal care, to come to the maternity center when labor begins, and to follow up with postpartum checkups. The bundling of services for one fee is an incentive for women to use services appropriately and not to delay seeking care (Kutzin, 1993).
STEP 4: IMPROVING QUALITY OF CARE
Refresher training for medical providers, developing and implementing protocols for management of obstetric and neonatal care, and the institution of maternal and perinatal death audits within facilities have been the three primary interventions to improve the quality of maternity and newborn care.
Training Midwives in Life-Saving Skills, Africa
Between 1988 and 1994, midwives were given advanced obstetric training to upgrade skills needed in managing complications through a competency-based curriculum for midwives developed by the American College of Nurse Midwives in several sites in Ghana, Uganda, and Nigeria. The course consisted of 10 modules, covering essential care for obstetric
complications (defined in Chapter 5), plus risk assessment during prenatal care and resuscitation. Typically, training lasts 3 weeks at a high-volume maternity hospital.
In Uganda, both trainees and tutors showed improvement in knowledge immediately after training. A longer term evaluation in 1994 (3 to 12 or more months posttraining) tested a 14 percent sample of midwives purposely selected to be representative of the sites and types of organizations that had participated in the training (Mantz and Okong, 1994). A skills tests on the use of the partograph showed continued understanding of its use by 75 percent of midwives. Fewer midwives passed a test of their ability to use a gestational wheel and the "Handbook for Midwives" as resources in determining how to manage an obstetric emergency. Inspection of facilities where the trained midwives worked revealed that the handbooks were scarce, even though they had been issued to the midwives during their training to provide them with up-to-date information on case management. A record review of partograph use in 36 facilities showed that midwives in most facilities carried out appropriate activities according to partograph guidelines more than 50 percent of the time. There was no real difference in midwives' activities whether they had been trained 3 or 12 or more months prior to the review.
Facilities where the trained midwives worked supported their work with the necessary equipment in 29 of 38 inventoried, but in only 7 was the revised handbook available, and in 7 there was no access to water (including one 100-bed hospital). Frequently, pressure gauges were unavailable or not functioning, and records were often lacking or locked up for safe-keeping. Trained midwives faced a number of other barriers as well, including lack of time to carry out tasks appropriately, especially in the prenatal clinics or labor and delivery wards when there were high numbers of patients and only one or two midwives on the ward (there is a chronic lack of staff); lack of support from the institutional authority to allow them to carry out tasks for which they are trained (e.g., start intravenous fluids, examine the cervix, or take any other active management measures); transferrals to new sites with staffs that were not oriented to the training; and chronic financial constraints that made it impossible for districts to provide continuing education and supervision.
A Nigerian facility that served as a site for similar training recorded nearly a doubling in the number of deliveries by the midwives during the year following the training, apparently due to the improved image and perceived quality of care.
Implementing Standards of Care and Protocols, Quetzaltenango, Guatemala, 1989-1993
The Western General Hospital San Juan de Dios in Quetzaltenango, Guatemala, is a general hospital with an obstetric department that delivers approximately 3,000 newborns per year. Prior to the start of the maternal and neonatal health project described above, the sanitary conditions in the sick baby nursery were poor, overcrowding of incubators was a problem, adequate temperature control was difficult to maintain, resuscitation facilities were suboptimal, and equipment, including oxygen, was in short supply.
In 1990 a neonatologist joined the hospital staff, and protocols and standards of care were developed for normal and abnormal obstetric and neonatal conditions by a specialist team of obstetrician-gynecologists and neonatologists. Training of hospital staff to use these norms took place in March 1991. Simultaneously, changes were made in the temperature control in the nursery with portable heaters, and cot nursing, guidelines for clothing sick neonates, infection control, and handwashing surveillance were initiated.
To evaluate the effect of these changes, a medical audit of case management through record review was carried out retrospectively in 1993 (Schieber et al., 1995). A random sample of cases was drawn from the early neonatal deaths for 1 year before implementation of the interventions and 2 years after. Avoidable factors were categorized as patient or home circumstances; referral by traditional birth attendant or others; service factors due to management by obstetric, pediatric, or nursing staff; or lack of equipment, facilities, and medications.
The hospital neonatal mortality rate decreased from 38.3 in 1989 to 25.9 in 1991, but it increased to 31.9 in 1992. This same pattern was seen in early neonatal mortality—a significant decrease from 32.9 per 1,000 live births in 1989 to 23.6 in 1991, and then an increase to 26.0 in 1992. The arrival of a neonatalogist and revision of standards of both obstetric and neonatal care by 1991 apparently had an effect. The rise in the mortality rates in 1992, however, may have been associated with the shortage of oxygen experienced in that year as determined by the medical audit. Most of the change was due to reductions in mortality due to sepsis and among low birth weight babies: 50 percent of deaths were within the first 24 hours of life throughout the study period, mainly due to asphyxia and hyaline membrane disease, which is associated with prematurity. In spite of intensive training in resuscitation, the low Apgar scores at births (0-3) changed only somewhat at 5 minutes. That the majority of these babies were at-term points to the need for urgent improvement of labor management both in communities (to ensure women go to appropriate facilities
earlier) and in hospitals. But between 1989 and 1991, there was a remarkable drop in avoidable factors assigned to staff as contributors to early neonatal deaths, emphasizing the improvement in the pediatric staff in responding to newborn problems (Schieber et al., 1995).
Instituting the Medical Audit of Maternal and Perinatal Death
Institution of medical audits of maternal and perinatal deaths within facilities may be a powerful intervention to improve the quality of services. One report from India shows a decrease in errors in judgment or delays in treatment at the department level and an increase in the number of high-risk and emergency cases attended by consultant obstetricians, with consequent decreases in maternal deaths from eclampsia and from the development of obstructed labor after admission, 10 years after institution of the medical audit within one facility (Bhat, 1989). Records were designed to document clinical management of all deliveries, confidential records of all maternal deaths were created for review by a committee, and weekly rounds reviewed all emergency cases, maternal and perinatal deaths, operative deliveries, and complicated pregnancies. Since this was a referral hospital, a team of a consultant, a resident, and a technician began visiting the surrounding primary health care centers periodically to provide prenatal care and communicate the outcome of referred cases.
Prior to the institution of the medical audit, a review of maternal deaths revealed that consultants did not always attend the cases that resulted in death and that residents failed to diagnose early enough such problems as malpresentation and concealed accidental hemorrhage and did not always call a consultant at the time of surgery. There was also delay in performing cesarean sections. At times when staffing was inadequate, such as at night and on weekends or holidays, there were more maternal deaths.
Achadi, A., E. Achadi, A. Kusdinar, D. Dachlan, A. Watief, B. Boer, Zahro, and H. Setwiawan 1994 Village Midwife (Bidan Di Desa) in Central Java and South Sulewesi. A collaboration between Center for Child Survival, University of Indonesia, and Directorate of Family Welfare, Ministry of Health, Republic of Indonesia, Jakarta, Indonesia.
Bailey, P.E., R.C. Dominik, B. Janowitz, L. Araujo 1991 Obstetrica e mortalidade perinatal em uma area rural do nordeste Brasileiro. Boletin de la Oficina Sanitaria Panamericana 111(4):306-318.
Bailey, P.E., J.A. Szaszdi, and B. Shieber 1994 Analysis of the Vital Events Reporting System of the Maternal and Neonatal Health Project: Quetzaltenango, Guatemala. Paper #3 prepared for the U.S. Agency for International Development. MotherCare Project #DPE-5966-Z-00-8083-00. Arlington, Va.: John Snow, Inc.
Belizan, J.M., F. Barros, A. Langer, U. Farnot, C. Victora, and J. Villar 1995 Impact of women education during pregnancy on behavior and utilization of health resources. American Journal of Obstetrics and Gynecology 173:894-899.
Bhatt, R.V. 1989 Professional responsibility in maternity care: Role of medical audit. International Journal of Gynecology and Obstetrics 30:47-50.
Bower, B., and A. Perez 1993 Final Project Report: Cochabamba Reproductive Health Project. Report prepared for the U.S. Agency for International Development. MotherCare, Arlington, Va.: John Snow, Inc.
Brennan, M. 1992 Training traditional birth attendants. Postgraduate Doctor Africa 11(1):16.
Chandromohan, D. 1992 Effectiveness of Maternity Waiting Homes in Reducing Adverse Perinatal Outcomes: Experience from Rural Zimbabwe. Unpublished master's dissertation. London School of Hygiene and Tropical Medicine, England.
Eades, C., C. Brace, L. Osei, and K. LaGuardia 1993 Traditional birth attendants and maternal mortality in Ghana. Social Science & Medicine 36(11):1503-1507.
Fauveau, V. 1991 Matlab Maternity Care Program. Review paper prepared for the Department of Population and Human Resources, World Bank, Washington, D.C.
Fauveau, V., K. Stewart, S.A. Kahn, and J. Chakroborty 1991 Effect on mortality of a community-based programme in rural Bangladesh. Lancet 338:1183-1186.
Hessler-Radelet, C. 1993 Lessons Learned About the Impact of the Regionalization of Perinatal Care Project. Field report prepared for the U.S. Agency for International Development. MotherCare Project #936-5966. John Snow Inc., Arlington, Va.
Hogberg, U., and S. Wall 1986 Secular trends in maternal mortality in Sweden from 1750 to 1980. Bulletin of the World Health Organization 64(1):79-84.
Howard-Grabman, L., G. Seoane, and C.A. Davenport 1993 The Warmi Project: A Participatory Approach to Improve Maternal and Neonatal Health: An Implementor's Manual. MotherCare Project. Arlington, Va.: John Snow, Inc.
Janowitz, B., P.E. Bailey, R.C. Dominik, and L. Araujo 1988 TBAs in rural northeast Brazil: Referral patterns and perinatal mortality. Health Policy and Planning 3(1):48-58.
Koenig, M.A., V. Fauveau, A.I. Chowdhury, J. Chakraborty, and M.A. Khan 1988 Maternal mortality in Matlab, Bangladesh: 1976-1985. Studies in Family Planning 19(2):69-80.
Kutzin, J. 1993 Obstacles to women's access: Issues and options for more effective interventions to improve women's health. Human Resources Development and Operations Policy. Human Resources working paper no. 13. Washington, D.C.: World Bank.
Kwast, B.E. 1995 Building a community-based maternity program. International Journal of Gynecology and Obstetrics 48(Suppl.):s67-s82.
Maine, D., M.Z. Akalin, J. Chakraborty, A. de Francisco, and M. Strong 1996 Why did maternal mortality decline in Matlab? Studies in Family Planning 27:179-187.
Mantz, M.L., and P. Okong 1994 Evaluation Report: Uganda Life Saving Skills Program for Midwives, October-November, 1994. Report prepared for the U.S. Agency for International Development. MotherCare Project #5966-C-00-3038-00. Arlington, Va.: John Snow, Inc.
Poovan, P., F. Kifle, and B. Kwast 1990 A maternity waiting home reduces obstetric catastrophes. World Health Forum 11:440-445.
Radyowiyati, A., and T. Sequeira no date Bidan Desa: A Case Study of Two Javanese Villages in Yogyakarta . Center for Women's Studies and Department of Health, Gadjah Mada University, Indonesia
Sargent, C. 1985 Obstetrical choices among urban women in Benin. Social Science and Medicine 20:287-292.
Schieber, B.A., M. Mejia, S. Koritz C. Gonsalez, and B. Kwast 1995 Medical Audit of Early Neonatal Deaths: INCAP: Quetzaltenango Maternal and Neonatal Health Project. Technical Working Paper #1 prepared for the U.S. Agency for International Development. MotherCare Project #936-5966. Arlington, Va.: John Snow, Inc.
Villar, J., U. Farnot, F. Barros, C. Victora, A. Langer, and J.M. Belizan 1992 A randomized trial of psychosocial support during high-risk pregnancies. New England Journal of Medicine 327:1266-1271.
Winnard, K. 1993 Applying Social Marketing to Maternal Health Projects: The MotherCare Experience. MotherCare Project, The Manoff Group. Arlington, Va.: John Snow, Inc.