Attending Home vs. Clinic-Based Deliveries: Perspectives of Skilled Birth Attendants in Matlab, Bangladesh

Abstract

In an effort to make skilled attendance at birth more accessible, some countries in Asia have begun major initiatives to promote the option of home delivery with a midwife. Yet there is little empirical evidence from the region to suggest that home-based care is as safe or effective as care in medical facilities. Qualitative research involving key informant and in-depth interviews and group discussions was carried out in 2003 and 2004 in Matlab, a rural area of Bangladesh, to examine the feasibility of home- vs. facility-based delivery from the perspective of 13 skilled birth attendants. The findings illuminate major constraints encountered during home deliveries, including poor transportation, inappropriate environment for delivery, insufficient supplies and equipment, lack of security, and inadequate training and medical supervision, which may prevent the provision of skilled care. Most difficult was the pressure by families to adhere to traditional childbirth norms and convincing families to accept the need for referral. The advantages highlighted of attending births in a health facility were the safe, clean environment, availability of supplies, ability to accommodate other work activities and make quick referrals, and higher coverage. The study illuminates practical, cultural and medical issues that need to be taken into consideration when choosing between home- and facility-based strategies and designing safe motherhood interventions.

Résumé

Afin de faciliter l'accès à une assistance qualifiée pendant l'accouchement, certains pays d'Asie encouragent la naissance à domicile en présence d'une sage-femme. Pourtant, rares sont les données empiriques qui indiquent que les soins à domicile sont aussi sûrs et efficaces que les soins dans des centres de santé dans cette region. Une recherche qualitative, avec la participation d'informateurs clés, des entretiens approfondis et des discussions de groupe, a été réalisée en 2003 et 2004 à Matlab, région rurale du Bangladesh, pour examiner les avantages de l'accouchement à domicile par rapport à l'accouchement en milieu hospitalier dans la perspective de 13 accoucheuses qualifiées. Les conclusions recensent des obstacles majeurs rencontrés pendant les accouchements à domicile, notamment les transports médiocres, l'environnement inadapté, l'insuffisance des fournitures et de l'équipement, le manque de sécurité, et l'inadéquation de la formation et de l'encadrement médical, qui peuvent contrarier la prestation de soins qualifiés. La principale difficulté était de résister aux pressions des familles demandant d'adhérer aux normes traditionnelles et de convaincre les familles d'accepter le transfert de la parturiente. Un environnement propre et sûr, la disponibilité de fournitures, la capacité de mener d'autres activités professionnelles et d'assurer une prise en charge rapide, et la couverture élevée figuraient au nombre des avantages cités pour l'accouchement en milieu hospitalier. L'étude cerne les questions pratiques, culturelles et médicales dont il faut tenir compte lors du choix entre des stratégies à domicile ou hospitalières et de la conception d'interventions de maternité sans risque.

Resumen

En un esfuerzo por ampliar el acceso a una atención competente durante el parto, algunos países asiáticos han emprendido importantes iniciativas para promover la opción de dar a luz en el hogar con una partera. No obstante, poca evidencia empírica indica que los cuidados hogareños son tan seguros o eficaces como la atención médica recibida en un establecimiento de salud en la región. En 2003 y 2004, se llevó a cabo una investigación cualitativa con informantes clave, entrevistas a profundidad y discusiones en grupo, en Matlab, una zona rural de Bangladesh, a fin de examinar la viabilidad del parto en el hogar o en un servicio de salud, desde la perspectiva de 13 parteras calificadas. Los resultados esclarecen las principales limitaciones encontradas durante los partos en el hogar, como transporte deficiente, un ambiente inadecuado para dar a luz, insuficientes suministros y equipo, falta de seguridad, y capacitación y supervisión médica inadecuadas, lo cual podría impedir el suministro de atención competente. Lo más difícil fue la presión de las familias para observar las normas tradicionales referentes al parto y convencer a las familias de aceptar la necesidad de remitir a la mujer a otros servicios. Entre las ventajas de atender el parto en un establecimiento de salud se mencionaron: un ambiente seguro y limpio, la disponibilidad de suministros, la capacidad de realizar otras actividades laborales y hacer referencias rápidas, y más cobertura. El estudio destaca los aspectos prácticos, culturales y médicos que deben tomarse en cuenta al escoger entre estas dos estrategias y al diseñar intervenciones para una maternidad sin riesgos.

Keywords:

• childbirth
• safe motherhood
• midwifery
• qualitative research
• Bangladesh

In an effort to make skilled attendance at birth more accessible, some countries in Asia have recently begun major initiatives to promote the option of delivery at home with a skilled attendant.¹² Yet there is little empirical evidence to suggest that home-based strategies are as effective as those in medical facilities. Research into home-based strategies has largely focused on the training of traditional birth attendants (TBAs),³⁴ whereas home birth with a skilled attendant has received little attention.

In Bangladesh, a national level initiative is underway to post skilled birth attendants in rural communities⁵ in response to persisting high rates of maternal mortality, now reported at 320 per 100,000 live births.⁶ Currently, 91% of births still take place at home with only 13% attended by a health professional.⁷ The strong preference for home birth is commonly associated with restrictions on female mobility and cultural norms.⁸⁹

A critical element of delivery care relates to the ability of skilled attendants to perform effectively. Key aspects of safe delivery practices include the physical setting, availability of supplies and equipment, level of skill and self-confidence of the provider and a supportive environment.¹⁰ While the constraints faced by women living in developing countries in accessing professional birthing care and quality of care issues are well researched,^1112131415 the views and experiences of skilled attendants assisting childbirth in women's homes have not been available.

This paper examines the experiences of skilled birth attendants carrying out home-based deliveries in a rural setting in Bangladesh, in comparison with deliveries attended by the same attendants in health facilities. The research raises critical questions and concerns about the feasibility of a professional home-based delivery approach in this part of the world.

Background

The Matlab Maternity Care Programme was initiated in 1987 at a time when the majority of maternal deaths in Bangladesh occurred at home and were directly related to complications during labour or the post-partum period, which could have been prevented.¹⁶ These findings led to the hypothesis that posting skilled birth attendants at the community level would have the best potential for saving lives, provided these workers were trained and equipped to treat obstetric complications and were backed by an effective chain of referral. The aim was for skilled birth attendants to live near pregnant and delivering women to facilitate rapid intervention at the onset of a complication.

The programme was designed so that two skilled birth attendants were based in each of the four health facilities that exist in the ICDDR,B International Centre of Health and Population Research service area of Matlab and would be available to respond to calls 24 hours a day for deliveries at home. Each health centre served a population of approximately 25,000 with 90% of households living within 3 km of the facility. Two additional skilled birth attendants provided care in the ICDDR,B maternity clinic, and three attendants were on standby to fill in for absence on leave or sickness, if the attendant(s) had spent the previous night assisting a delivery or were out of the health centre when a delivery call came. Periodic transfers were made so that the attendants rotated posts.

The attendants were nurse–midwives who typically had four years of training (three years of basic nursing and a fourth year devoted to childbirth) or lady family planning visitors (LFPVs), who were paramedics with at least six months midwifery training. Both types of attendants were instructed to provide skilled obstetric care, including management of normal labour, medical treatment of problem pregnancies (including iron and antihypertensive drugs) and neonatal care, but excluding assisted deliveries. Attendants were also responsible for organising referrals and accompanying patients when a home birth was deemed unsafe, first to the ICDDR,B basic emergency obstetric care (EOC) clinic in Matlab or, if needed, to a comprehensive EOC facility (public or private) in Chandpur, the district headquarters. Other primary responsibilities included antenatal and post-natal visits in women's homes. Additional safe motherhood activities involved visiting families to encourage the use of the midwifery services and training TBAs on safe delivery practices and referrals.

These attendants relied heavily on women community health workers (CHWs), who were local residents engaged in maternal and child health and family planning activities. The CHWs played a critical role in establishing social connections between the skilled birth attendants, who were comparatively better off and better educated than their patients, and community members. In order to avail themselves of delivery services offered by the skilled birth attendants, families were instructed to inform the CHW about the pregnant woman's condition first. The CHW would fill out a form that the family member subsequently delivered to the skilled attendant stationed in the health centre. The attendant was then required to travel by whatever means available – boat, rickshaw or on foot – to the household of the delivering woman.

Each of the attendants had a male porter to carry supplies and assist with transport, who also assumed the role of security guard when deliveries occurred at night. A speedboat was available 24 hours a day for emergency transfers.

This strategy continued until 1996, when there was a shift in approaches in Matlab. Home deliveries were gradually phased out and replaced by facility-based care. Between 1996 and 2000, all four health centres were upgraded and equipped to perform obstetric care. The same midwives were asked to stop attending home births and to perform deliveries in the health facilities.

Research methods

The research was carried out between September 2003 and June 2004 in the Matlab sub-district of Bangladesh. Matlab is located in the delta region 30 km southeast of Dhaka. The population in this rural area is generally poor, with farming and fishing the primary source of income; literacy rates are low, particularly among women.¹⁷ The majority of inhabitants are Muslim, with a minority Hindu population, and female mobility is restricted. Typical of this region of Bangladesh, the area is flooded several months during the year.

The sampling frame was exhaustive, including all 13 skilled birth attendants who attended home deliveries. As three of them requested a transfer before 1996, when the facility-based approach was introduced, only their input relating to home deliveries was included in the research. The manager, who played a key role in the development and implementation of the programme, was also included as a respondent.

The research strategy included key informant interviews with four of the skilled attendants and the programme manager on an ongoing basis throughout the study. They were selected based on seniority and their ability to describe the experiences of the skilled attendants more broadly. In-depth interviews involving a series of repeat visits were carried out with the other nine skilled attendants. These interactions generated an additional set of research questions, which guided group discussions. Respondents were divided into two groups of six and seven, and a series of six group sessions were held until we were satisfied that all pending questions were addressed. A final set of interviews, designed to validate the findings and provide additional insights and interpretations, were carried out with key informants.

Two senior research investigators with extensive experience in qualitative data collection administered the interviews. Informed consent was taken from all respondents before the interview. The researchers took notes during the interviews and expanded upon the information immediately afterwards. Interviews were conducted in Bangla and translated into English.

For the analysis, a coding system was developed, capturing the main themes and concepts. Data triangulation identified only those concepts that could be validated through a combination of data sources, such as multiple interviewees, key informants, and group discussions.

Findings: home-based deliveries

The findings reveal a broad set of difficulties that the attendants confronted when assisting home deliveries. From their perspective, the health facility offered an environment more conducive to effective skilled attendance at births.

Transportation

Respondents uniformly complained that transport was extremely difficult, especially in the dry season when the only option was to go by foot or rickshaw. Due to the poor roads, particularly at the initiation of the programme, the use of the rickshaw was limited and a good deal of travel on foot was required to reach households. In the monsoon season a boat was used which, at the height of the rains, made access easier. However, in the early and late monsoon season travel by boat was often not possible.

Delays in obtaining the speedboat for emergencies were also common. One of the midwife informants recounted a story about a woman who suffered a uterine prolapse after a TBA attempted to remove the placenta manually. The midwife was summoned and the woman's life-threatening state made evident the need for immediate hospital care. The midwife notified the boat service, but the woman died before the boat arrived more than an hour later.

Lack of a proper environment for delivery

In village settings, families typically allocated a dirty, private corner of the house to serve as the place for delivery.⁸¹⁸¹⁹ In addition, as families placed a strong emphasis on removing the woman from the prey of evils spirits, a dark area was preferred. The lighting was generally poor, as most rural households did not have electricity. When the delivery occurred at night, there was only the light from a kerosene lamp, which was insufficient, especially if an episiotomy was needed.

As childbirth is traditionally considered to be polluted by Bangladeshis, the delivery bed materials had to be discarded after the birth. Hence, an old blanket or jute sack was placed on the floor, covered by a plastic sheet. The attendants found it arduous to deliver from the floor, which they felt compromised their skills.

In households, bamboo screens commonly divide rooms. During labour, groups of female relatives and neighbours would gather on the other side of the screen, peering through the slits of bamboo, conversing about the progress of the labour and delivery, laughing amongst themselves, and frequently questioning the skilled attendant about the delivery. These interruptions disturbed the attendants' concentration and interfered with their work. At times, this was so disruptive that the attendants said they were tempted to leave.

The delivery environment in Hindu households, where a temporary, separate room (aus ghor) was constructed for the delivery at minimal cost, was described as particularly problematic. The room typically had adequate space only for the mother and baby, forcing the skilled attendant to position herself partially outside the structure when assisting the delivery. During the monsoon season, the attendant was unprotected from the rain. Due to the association of the blood and fluids with pollution, the structure was typically constructed in a “dirty” area.

“My head was drenched by rainwater and my bottom was soaked with cow urine because the aus ghor was built just beside the cow shed.” (Midwife)

Hindus also subscribe to choa-chuee, a belief system which teaches that anything an impure person touches is also impure and must be avoided. As a result, during delivery family members would refuse to provide assistance or materials to the attendant because she was regarded as impure. She was even required to pump the tubewell herself if she needed water, or somebody from the family would pour water into her cupped hands, invariably maintaining a distance while doing so. These practices caused frustration for respondents, who expected cooperation and respect from the delivering woman's family.

Both Hindu and Muslim families were said to show disdain towards midwifery and its association with impurity. Religious differences also caused conflict between the attendants and the families.

“In the morning I heard the husband of the woman loudly order one of the maidservants, ‘Give that Charal (one of the lowest castes) woman some puffed rice.’ I felt so bad that I could not drink even a sip of water there.” (Hindu midwife attending a birth in a Muslim household)

Lack of acceptability of procedures: delivery position and episiotomies

Traditionally, women deliver in a kneeling position with the genitals covered, which meets cultural norms. TBAs would not visually examine the birth canal or cervix; rather, they frequently placed their hands in the birth canal to feel the position of the baby and assess the progress of labour. In contrast, the skilled attendants were trained to perform deliveries with the woman lying on her back which, for them, had the advantage that it allowed the midwife to assess the progress of labour more easily and ease the baby out to avoid tearing. Women often objected to using the supine position for delivery or spreading their legs, and resisted if the attendant tried to lift their clothes during delivery, as causing extreme shame. While some of the respondents allowed the delivering woman to choose her position, others were uneasy and unwilling to compromise. In the latter case, extended negotiations between the attendant and the woman's family would be required. The attendants faced problems particularly with the older women who were commonly present and were the primary decision-makers regarding childbirth.

Fears about episiotomy created tremendous barriers to any willingness to seek skilled assistance for home-based delivery, and convincing families about the need for an episiotomy was difficult. Older women held many misconceptions about episiotomies and were frequently opposed, maintaining that they had delivered babies without this invasive practice and that tears were a natural outcome of delivery that healed on their own. They additionally expressed concerns that the procedure would have a negative effect on sexual relations. Stories of women who developed infections after episiotomy or suffered from pain associated with the procedure were apparently circulated in the community, creating much resistance.

A widespread traditional practice during the post-partum period was for women to expose the opening of the birth canal to heat, by squatting over hot coals for long periods several times a day; this was believed to heal the area, particularly if a tear had occurred. Even after episiotomy this practice was followed, and the attendants reported cases when the stitches melted, leading to additional problems. It was also difficult for women to maintain appropriate hygiene following an episiotomy, as recommended by the birth attendants. Women did not have sanitary pads but used pieces of cloth to absorb post-partum discharge. These cloths were generally re-used without being washed with soap, and due to the association with pollution, they were typically dried in a hidden, dark and damp corner of the house away from men's view. This put women at risk of infection.

Lack of necessary supplies and equipment

The skilled attendants always kept two bags containing instruments and supplies in the health centre ready to attend a normal delivery. This included a blood pressure machine, stethoscope, fetoscope, two pairs of disposable gloves, one safe delivery kit, a thermometer, scissors, disposable syringes and a flashlight, which was needed for night calls. Basic drugs were ampicillin, diazepam, paracetamol, gentian violet, tetracycline eye ointment, metronidazol and saline. Respondents often required additional equipment and supplies but due to the limitations of transport, were forced to restrict what they carried. If an unforeseen need arose, they frequently found that their supplies were inadequate and they would have to return to the health centre for them. At times, when the newborn required special care, they were forced to carry the infant back to the health facility as well.

Resistance to referrals

Respondents indicated that the most difficult part of their work was convincing families to accept the need for referral. The skilled attendants were instructed to attend normal deliveries and refer complicated cases that could not be managed at home to the Matlab clinic. However, particularly initially, there was an expectation that the attendants could handle all cases; family members could not understand why some deliveries might be beyond their capabilities. A common reaction was that in referring, the attendants were simply avoiding their responsibilities. Many families, mainly the mothers-in-law, expressed grave concerns over transporting a delivering woman to Matlab¹ or allowing her to be subject to the procedures used in the maternity clinic. The skilled attendants often spent hours or even full days trying to convince families to let them take the woman to Matlab.

The skilled attendants thought referral was the primary reason why they were not called to attend a delivery, particularly women's fear of being sent by ambulance to the EOC facility in Chandpur, accompanied by the attendant, where she would be transferred to the government hospital. This had huge cost implications, particularly with caesarean section, which required at least 10–15,000 taka (approximately US$250–$480 in 1987–96). Moreover, the Matlab and Chandpur hospital staff were not known to villagers and the more urbanised settings were perceived as intimidating.

Working in isolation in these environments, the skilled attendants admitted they frequently lacked the confidence to proceed normally and were quick to refer. When faced with a complicated case, they were often eager to shift responsibility to the medical staff in Matlab as per official policy, while the family often applied overwhelming pressure not to refer. This dilemma caused the attendants a lot of anxiety.

Lack of training for home delivery and medical supervision

Most of the attendants said that lack of training specific to assisting in home deliveries, and being removed from essential supplies and equipment, negatively affected their ability to do their job. Another major limitation was that they lacked medical supervision. Supervision was carried out through monthly meetings where they conferred with other staff in the maternity care programme. This gave them the opportunity to discuss deliveries and stillbirths, complicated cases and referrals, and the conditions under which any deaths occurred. However, they found this system of verbal reporting and indirect supervision inadequate. Many expressed a need for medical supervisors to evaluate their performance directly during field visits and give them advice, thus enhancing their competence and fostering their confidence to cope in difficult conditions. A primary concern was the risk related to home deliveries. In their view, a supervisor would remove the responsibility (and potential blame) from them.

Social pressure

The skilled attendants said they experienced tremendous social pressure during the home delivery period, particularly in life-and-death situations. If anything went wrong, they felt they would be accountable to the family, the community and the ICDDR,B authorities. As a result, they felt vulnerable and were unwilling to make problematic decisions.

“I remember those days when I attended home deliveries. I was so nervous. The fetal heartbeat somehow sounded different in a village setting. Frequently, I referred patients to Matlab because I did not have the courage to maintain full responsibility.” (Lady family planning visitor)

They also said that community members had many expectations which they could not necessarily fulfil. For instance, they were often urged to give an injection to increase labour and speed up delivery, because that is what TBAs and village practitioners frequently recommended, particularly in cases where labour was not progressing. The skilled attendants were expected to use the same approach and deliver the baby quickly, to avoid drawing too much attention to the delivery. In contrast, the skilled attendants had been taught to allow labour to progress naturally. When they refused to give an injection, their credentials and credibility were questioned. They were often suspected of wrongdoing and accused of selling the drugs elsewhere. Family members also questioned why the attendant had come to their home if their solution was to refer the patient, and why they were unwilling to respond to the wishes of the community, which embraced different delivery practices.

Inadequate security

At the inception of the programme, most of the skilled attendants were young and unmarried. In a context where high value is placed on female virginity before marriage and women are vulnerable to sexual and physical abuse,²⁰ the attendants had grave concerns about living and working in the villages, and particularly about travelling at night to attend deliveries. In Bangladesh, female mobility is still restricted and generally unacceptable at night, and women in transit outside the home are considered fair prey to sexual assault.²⁰ Although a porter accompanied them at night, the women remained anxious about their personal safety. Community members were outraged, and complaints submitted to the programme manager after the following incident:

“During that time I was unmarried. Once my boatman told me, ‘Apa (sister), don't attend any calls for the next two to three days after sunset. I've heard that someone is planning to kidnap and marry you.’ That night I received a delivery call and I did not go. The patient died.” (Midwife key informant)

Scheduling difficulties

The attendants were required to remain in the household of a delivering woman until they were confident that neither the mother nor the newborn needed additional assistance. Hence, when an attendant was called to attend a delivery, she was uncertain when she would return to the health centre. Respondents all felt their irregular schedules negatively affected their work satisfaction and interfered with their personal lives. Problems mentioned included leaving a breastfeeding infant for long periods or missing family gatherings. Many were confronted with ongoing opposition by their spouses, often causing conflict in their marriages.

Comparison with facility-based deliveries

In 1996, a gradual policy shift began from promoting home-based to facility-based deliveries, and midwives and paramedics were made responsible for conducting deliveries in the health centres.²¹ The community health workers were said to have played a key role, informing community members through their door-to-door activities that there would be no more home deliveries and that women in labour should be taken to the health centre. During antenatal visits in the health centre, the skilled attendants showed the delivery room and the available instruments to the pregnant woman and anyone accompanying her (e.g. mother or sister-in-law). In the initial phase of the transition, they also showed village leaders the delivery room and asked them to discuss the benefits of health centre deliveries with the community.

The attendants welcomed the change, which from their perspective improved their ability to perform, citing the following positive outcomes:

•	more rapid transfer of patients from the health centre to the Matlab maternity clinic
•	ability to continue doing other work while monitoring labour
•	ability to provide care for more than one woman at a time
•	care available 24 hours a day
•	secure working environment
•	ability to assist with complicated cases
•	procedures like episiotomies easier to perform
•	assistance available from other skilled attendants if needed
•	electricity 24 hours a day
•	equipment and medications always available
•	controlled, hygienic working environment
•	more regularity in the work schedule.

At the same time, some of the attendants admitted that when working in the facility they were less concerned about their relations with the community. This was reflected in their behaviour during one-on-one interactions with women and family members, which we often found through informal observations to be abrasive and demeaning. Specifically, the midwives reprimanded women for arriving late for delivery, taking too long to deliver, screaming during contractions, refusing to expose their genitalia and having too many children.

Discussion

This study has highlighted a series of difficulties confronted by skilled birth attendants attending home births. A preliminary evaluation of the government's skilled birth attendance programme in Bangladesh had uncovered some of the same constraints, such as inadequate supervision, difficulties travelling to households and lack of support from community members.⁵ In-depth, qualitative research allowed us to identify additional factors, many of which were culturally rooted, that need to be taken into consideration if the Bangladesh government continues to invest in home-based delivery care at the national level.

The rationale for instituting skilled attendance for home-based deliveries was associated with cultural norms favouring home births. Yet many of these same cultural values were found to create conflict between the skilled attendants and pregnant women and their families, thus impeding the aim of safe delivery in the home. The attendants expected cooperation from and acceptance of medical practices by families when they were called to assist a delivery. However, in the home environment families expected relative control over the birth, an event that typically is guided by elderly women relatives and encompasses a mix of traditional and modern practices to ensure a successful outcome. Perhaps the starkest incongruity between traditional and medical practice was that the attendants were expected to deliver all births at home irrespective of the woman's condition.

This divergence of opinions and expectations needs to be considered when providing obstetric care, whether in the home or the clinic setting. Other researchers have described similar clashes between nurse-midwives, who tend to align themselves with modernity and clinical approaches, and the recipients of their care.²²²³ Thus it is important to identify which aspects of professional delivery care can be modified to meet cultural expectations without jeopardising the health of the woman. For instance, in Matlab the non-supine position or adequate cover of the woman to prevent exposure may increase the acceptability of services by women and their families and community support.

Our findings highlight how other constraints found in the home setting could potentially be addressed through improved programmatic strategies. These include training specific to home deliveries and supervision that involves on-site visits by medical personnel. Community education designed to enhance the general understanding of the care and procedures offered by the skilled attendant may alleviate concerns and establish a sense of trust among community members. There remain, however, other obstacles such as difficulties getting to households, security, the time needed to assist a delivery, and conditions within the home, all of which are more difficult to resolve. Another stark issue relates to the conduct of the skilled attendants. We found that in the health facility the midwives often showed disrespect for the pregnant woman and her accompanying family members. Such behaviour was linked to a desire on the part of the attendants to assert their higher social status and the fact that they were in charge. In an ethnographic study of a Bangladeshi hospital obstetric unit, it was also reported that nurse-midwives scolded and humiliated patients to reinforce their dominant status.²⁴ Presumably, in a culture where hierarchical designations govern relations between individuals and groups,²⁵ the tendency to subordinate patients also took place in their homes. Thus, in either setting, safe motherhood programmes must recognise this as a problem in need of redress. Efforts aimed at improving the behaviour of midwives will require culturally appropriate training, taking into account societal views on status and adequate supervision.

Although there are several organisational models for delivery care,²⁶²⁷ the global priority is to ensure skilled attendance for all births²⁸ and access to emergency obstetric care for complications. There remains an ongoing debate among national and international stakeholders about whether skilled birth attendants should be posted at the domiciliary or facility level. Few rigorous evaluations have been carried out examining skilled attendance in the home setting. Probably the most cited example comes from Malaysia where, under a home-based programme, maternal mortality decreased significantly.²⁶²⁷ Explanations for the success of this initiative are attributed to the posting of a sufficient number of well-trained midwives, ongoing access to essential drugs, adequate means of referral to comprehensive care, free care at all levels and recognition of and respect for local customs.²⁶ In Indonesia, however, a home-based delivery policy brought a dramatic increase in skilled attendance at birth, but failed to improve access to life-saving care for women in need.²⁹

There is general agreement that a constellation of conditions is required for professional attendants to function effectively. These include adequate personnel with the appropriate skills to provide maternity care, an enabling physical environment involving sufficient supplies and equipment and efficient referral systems, and ongoing training and supervision.¹⁰³⁰³¹ Less emphasis has been placed on the extent of local acceptability of medical practices and the importance of community support. In a context like rural Bangladesh, the obstacles illuminated in our research raise important questions about whether quality of delivery care can be assured in the home setting.

Whereas in the past it has been assumed that cultural issues would prevent women from utilising clinic-based obstetric services, the experience in Matlab illustrates that a well-organised maternity programme offering comparatively good quality services can help to overcome these barriers. Data collected through the Matlab health and demographic surveillance system suggest that in the health facility the skilled attendants were able to assist more deliveries than in the home setting. During the home-based strategy, the proportion of women seeking skilled delivery attendance at home peaked at 19.6% in 1992, while the rise in facility births was impressive, increasing from 6% in 1996 to 26.8% in 2001, only a year after all the facilities were fully functioning.²¹ At the height of the home-based strategy in 1992, each pair of birth attendants assisted on average 133 home births per year, compared to 183 births per year in each facility in 2001. Socio-economic inequalities in the use of a skilled attendant were substantial for both home- and facility-based care, although inequalities in access were somewhat greater for births in health facilities.²¹ Since 2001, there has been a gradual increase in facility-based delivery care, with 35.9% and 40.1% of women giving birth in a facility in 2003 and 2004, respectively.

While some may argue that the long duration of its safe motherhood programme makes Matlab highly unusual in the Bangladesh context, data show that the Matlab service area is not as unique as many people suggest.³² In fact, over time maternal mortality data from Matlab have been similar to nationwide trends identified through demographic and health surveys. Furthermore, even if Matlab were unique, the conditions would be even more difficult in other parts of the country where safe motherhood programmes and the health infrastructure are less developed.

National governments should be encouraged to clarify what their policies and aspirations are in terms of where women deliver, and either commit to a facility-based strategy or make explicit the rationale for choosing other alternatives. It is generally argued that home births increase coverage with skilled care in rural areas and respond to women's demands for home-based care. However, midwives deployed at the community level may not actually provide much outreach, instead serving those closest to them.²¹ Moreover, the basic conditions in homes in rural areas may prevent midwives from delivering skilled care. Home delivery care is also inefficient in terms of the midwife's time and ability to cope with emergencies, and home-based provision is less likely to be sustained in the long term. Before embarking on major initiatives to scale up skilled attendance at home, our results underscore the importance of further examining the safety and effectiveness of home birth strategies to ensure skilled birthing care for women.

Acknowledgements

This research was funded under Cooperative Agreement #388-A-00-97-00032-00 with the United States Agency for International Development (USAID), ICDDR,B grant number GR-00089. ICDDR,B acknowledges with gratitude the commitment of USAID to the Centre's research efforts. Carine Ronsmans is funded by the Department for International Development, UK. We thank the skilled attendants for the time they gave to share their valuable experiences.

Notes

1 Pregnant women are believed to be highly vulnerable to evil spirits, one reason why their mobility was restricted.⁸⁹¹¹ Susceptibility was considered to be at its pinnacle during labour and delivery, which were to take place in utmost privacy to keep them a secret.