Final Report: February 9, 1996

UNRAVELING THE MYSTERIES OF MATERNAL DEATH IN WEST JAVA Reexamining the Witnesses

Meiwita B. Iskandar Budi Utomo Terence Hull Nick G. Dharmaputra Yuswardi Azwar

•Center for Health Research - Research Institute University of Indonesia in Collaboration with •Directorate of Family Health Directorate General of Community Health Ministry of Health Republic of Indonesia

UNRAVELING the mysteries of maternal death in West Java: reexamining the witnesses / [written by Meiwita B. Iskandar ... et al.]. -- Depok : Pusat Penelitian Kesehatan UI, 1996. xii, 135 hlm. ; 29 x 21 cm. Diterbitkan atas kerjasama dengan Direktorat Bina Kesehatan Keluarga, Dirjen Pembinaan Kesehatan Masyarakat ISBN 979-8232-06-2 I. Kehidupan keluarga - Jawa Barat I. Iskandar, Meiwita B. II. Universitas Indonesia. Pusat Penelitian Kesehatan 646. 780 959 821

ISBN : 979-8232-06-2

First edition, February 1996 © Center for Health Research - Research Institute University of Indonesia, 1996

Published by Center for Health Research University of Indonesia Research Institute University of Indonesia Komp. Rektorat, Gd. LPUI, Kampus UI - Depok, Indonesia Telp. (21) 727 0154, Fax. (21) 727 0153

No part of this book may be reproduced in any form-- except for brief quotation (not to exceed 1,000 words) in review or professional work--without permission in writing from the publishers. This study is a collaborative study between the Center for Health Research University of Indonesia and the Directorate of Family Health, Directorate General of Community Health, Ministry of Health, Republic of Indonesia The study is funded by the Third Community Health and Nutrition Project (IBRD Loan 3550-IND), under the Contract with the MCH-FP Project for the fiscal year 1994/1995 No. 40/BM/00.PRO/I/95, 6 January 1995 The study report is written by the following research personnel: Meiwita B. Iskandar Budi Utomo Terence Hull Nick G. Dharmaputra Yuswardi Azwar Support for related data and material processing are provided by the following research assistants: Herna Lestari Eti Sulaeha Tirta Yenti Lukman Rulhakim

iv

Acknowledgments Our major debt of gratitude is to the families of the West Javan women who experienced obstetric emergencies in 1994 and 1995 -- the 53 who died, and the 10 survivors -- and who agreed to share their stories with us. Though selected from among many hundreds of cases using criteria which were not always obvious to them, the family members, neighbors, village officials, TBAs and health personnel were very open, very serious, and very generous with their time. We only hope that the descriptions related here, and our analysis of their experiences will help prevent a repetition of the difficulties they faced. The fieldworkers committed six months of difficult and often arduous effort to ensure that the data were collected accurately and well. They deserve a great deal of credit for hard work and loyalty under pressure. This study would never have been carried out if it were not for the steady encouragement -- and the courage -- of Dr. Nardho Gunawan, Dr. Rachmi Untoro, Dr. Ardi Kaptiningsih, Dr. Ina Hernawati of the Department of Health. From the outset they grasped the major purposes of the innovative methodologies applied here, and recognized that the results would undoubtedly present important challenges to the structure and function of the maternal health care delivery systems nationally. It is indicative of their own pragmatism and professional commitment that they urged the team to push the analysis as deeply as possible in order to discern practical issues of policy, training and budgeting which could be addressed by the government. For them the stories revealled via Rashomon techniques are not unusual. As administrators and policy implementors they daily face the challenge of overcoming the national tragedy of maternal and child mortality.

x

TABLE OF CONTENTS

Acknowledgments........................................................................................................ vi List of Acronyms & Terms ....................................................................................... vii Table of Contents ........................................................................................................ x List of Tables .............................................................................................................. xii Chapter 1. Background ............................................................................................... 1 1.1 Major issues covered in the study ....................................................... 1 1.2 The Rashomon Technique: an experiment in qualitative research . 3 1.3 Reexamining the witnesses: cases of maternal deaths ...................... 6 1.4 Selection of a comparison group: lucky survivors ............................ 7 1.5 Selecting witnesses of relevance .......................................................... 8 Chapter 2. Review of the literature on maternal mortality ................................... 10 2.1 Maternal mortality levels and trends ............................................... 10 2.2 Obstetric causes of death ................................................................... 13 2.3 Determinants of maternal mortality ................................................. 15 2.4 Health services and programs to reduce levels of maternal mortality ............................................... 17 Chapter 3. The contexts of maternal mortality in West Java ............................... 22 3.1 Demographic context ......................................................................... 22 3.2 Fertility and marriage patterns in West Java .................................. 24 3.3 Levels of education and labor in West Java .................................... 26 3.4 Transport and access to existing primary health care .................... 27 3.5 The traditional context ....................................................................... 29 3.6 The context of contrasts in West Java .............................................. 37 Chapter 4. Fieldwork: the search for cases and the questioning of witnesses ..... 39 4.1 Definition and number of cases to be included ................................ 39 4.2 Research personnel: forming the teams and specifying the jobs ... 39 4.3 Searching for maternal deaths and survivors of obstetric emergency cases ................................................................. 41 4.4 Field operations .................................................................................. 41 Chapter 5. Results: background characteristics of cases and witnesses .............. 44 5.1 Number of maternal deaths ............................................................... 44 5.2 Characteristics of cases and witnesses .............................................. 47 Chapter 6. Cross-examining the witnesses .............................................................. 53 6.1 Overview ............................................................................................. 53 6.2 Themes ................................................................................................. 55

xi

Chapter 7. Discussions and Conclusions ................................................................. 86 7.1 Provider ............................................................................................... 86 7.2 Community/client ............................................................................... 90 Chapter 8. Recommendations .................................................................................. 93 8.1 Supervision for service and care ....................................................... 93 8.2 Training on essential emergency obstetric care (POED) ................ 94 8.3 Training on life saving skills (LSS) and provision of essential comprehensive obstetric care (or POEK) 95 8.4 Quality of care and supply/availability of basic medicines (first and second choice medications) ............................................. 96 8.5 On-going promotion of safe motherhood by local government ..... 98 8.6 Reproductive rights and poverty: issues of political and economic commitments .................................................................................................. 99 References ............................................................................................................... 103

Appendices Appendix 1. In a grove: an example of the "Rashomon Technique" ................. 110 Appendix 2. Examples of cross-witness examination for Mysteries of Maternal Death in West Java .................................. 114 Appendix 3. Map identifying location of the case in the Mysteries of Maternal Death in West Java ............................. 130

xii

LIST OF TABLES

Table 2.1.

Estimates of maternal mortality ratio (MMR) in Indonesia .......... 11

Table 3.1.

Quality of Life Index of research Kabupaten and Selected Urban Sites, 1995 ................................................................ 23

Table 3.2.

Number of community health centers and village level integrated service posts, in West Java, 1993 ...................................................... 28

Table 3.3.

On-going culinary beliefs and practices on maternal care in West Java, 1995 .............................................................................. 32

Table 3.4.

On-going cultural beliefs and practices of Dukun/ Paraji concerning maternal care in West Java, 1995 ................................. 34

Table 3.5.

Number of registered nurses, dukun/TBAs, midwives, and village kaders, in West Java, 1993 ................................................................ 36

Table 5.1.

Reported number of birth deliveries and maternal deaths in hospitals during 1993, and number of maternal deaths in the previous 12 months reported through case searching and fieldwork ........... 45

Table 5.2.

Expected and reported number of maternal deaths in the study areas ............................................................................... 47

Table 5.3.

Number and type of cases and witnesses by Kabupaten ................. 49

Table 5.4.

Demographic characteristics of cases ............................................... 50

Table 5.5.

Health seeking behavior characteristics of cases and probable main cause of maternal death ..................................................................... 52

Table 6.1.

Some common practices which carry risks of infection ................. 60

LIST OF ACRONYMS & TERMS

Anaemi Angkot ASEAN BinKesGa BoTaBek CBS CCSUI CFR CHRUI CMR CWR Desa DirjenYanmed DOA EDD EOC Eclampsia Ectopic pregnancy

Ergometrine GOI-UNICEF Gotong Royong Hemorrhage IDHS IEC IM IMR Infancy IU IUD IUFD IV Jamu Jamu Peluntur Jimat

A disease in which the blood gets thin for lack of red blood cells. Signs include pale skin and lack of energy. Angkutan kota (City transportation) Association of South-East Asia Nations Bina Kesehatan Keluarga (Family Health Care) Bogor, Tangerang, dan Bekasi (Industrialized areas of Bogor, Tangerang and Bekasi) Center Bureau Statistics Center for Child Survival University of Indonesia (See Puska) Case Fatality Rate Center for Health Research University of Indonesia (Pusat Penelitian Kesehatan Universitas Indonesia) Child Mortality Rate Child-Women Ratio Villages Direktorat Jenderal Pelayanan Medik (Directorate General for Medical Services, Ministry of Health). Death on Arrival Estimated Delivery Date Emergency Obstetric Care A potentially fatal condition in which a woman with pre-eclampsia (see below) develops convulsions A pregnancy that occurs outside the uterus, usually in one of the fallopian tubes. Ectopic pregnancy can be fatal if not detected early. Injected medication for uterus contraction to control severe bleeding Government of Indonesia-UNICEF Collective Support Severe or dangerous bleeding. Indonesian Demographic Health Survey Information, Education, Communication Intramuscular Infant Mortality Rate The period of time from birth to one year of age. International Unit Intra-Uterine Device Intra-Uterine Fetal Death Intra-Venous Herbal Medicine Abortifacient medicine Amulet

14

Kabupaten Kanwil Kecamatan LE LMP LR LSS Mantra Mati Sahid Mati Suri MCH MCSDP MI MMR MOH Musyawarah NFPCB NGOs NHS Ob-Gyn Obstructed labor

PCSMR PELITA Pemda PHC PID

PKD PMI PO POED POEK POGI Polindes Posyandu

District Regency Kantor Wilayah (Regional Office) Subdistricts Life Expectancy Last Menstrual Period Literacy Rate Life Saving Skills Incantation Martyr's death State of coma, unarousable unresponsiveness Maternal & Child Health Maternal Child Survival, Development and Protection Macro International Inc. Maternal Mortality Ratio Ministry of Health Collective decision making National Family Planning Coordinating Board Non-Governmental Organizations National Health Survey (National Household Health Survey) Obstetric-Gynecology A problem that occurs when the space in the bony birth canal of the mother is either too small or too distorted by disease to permit easy passage of the head of the baby during labor. If the condition is not dealt with in early stages of labor, obstruction can result in death of the mother through infection and exhaustion, and death of the fetus through birth injury and lack of oxygen. Proportional Cause Specific Mortality Rate Pembangunan Lima Tahun (Five Years Development) Pemerintah daerah (Regional/Local Government) Primary Health Care Pelvic Inflammatory Disease: infection and inflammation of the upper reproductive tract (uterus and fallopian tubes) caused by sexually transmitted disease. PID can lead to infertility. Petugas Koordinator Desa (Village Health Officer) Palang Merah Indonesia (Indonesian Red Cross) Given Orally Pelayanan Obstetrik Essensial Dasar (Essential emergency obstetric care) Pelayanan Obstetrik Essensial Komprehensif (Essential comprehensive obstetric care) Persatuan Obstetrik Ginekologi Indonesia (Indonesian Ob-Gyn Association) Pondok Bersalin Desa (Maternity Hut) Pos pelayanan terpadu (Village integrated service post)

15

Pre-eclampsia

Puskesmas Pustu QLI SDKI Sepsis SMAM SOP Susenas TBA Toxaemia WHO YLKI

A serious condition sometimes occurs during pregnancy characterized by development of high blood pressure, protein in the urine, and swelling of the feet, hands, and face. Also known as toxaemia gravidarum. Pusat Kesehatan Masyarakat (Community health center) Puskesmas Pembantu (Sub health center) Quality of Life Index Survey Demografi dan Kesehatan Indonesia (Demographic & Health Survey). Blood poisoning due to severe infection Singulate Mean Age at First Marriage Standard of Procedures (PROTAP) Survei Sosial Ekonomi Nasional (National Social Economic Survey) Traditional Birth Attendant See pre-eclampsia. World Health Organization Yayasan Lembaga Konsumen Indonesia (Foundation for the Indonesian Consumer's Institute).

16

Chapter 1. Background OBJECTIVE: This study investigates maternal mortality in West Java on two levels. First, a review is made of the literature on maternal mortality internationally and with specific focus on West Java. Second, using an innovative method of qualitative data collection (the Rashomon Technique), the importance of different people in dealing with cases of maternal illness and death is analyzed to determine how interventions might change behavior in ways which would better preserve the lives of pregnant and postpartum women in Indonesia.

1.1 Major issues covered in the study Over the past decade the issue of maternal mortality has rapidly risen to the top of the health care agenda in Indonesia. The factors behind this growing community awareness have much to do with the collection and analysis of data on the levels and trends of maternal death. It is now believed that as many as 450 women die for every 100,000 live births in Indonesia, and some regions, like West Java, West Nusatenggara, and Irian Jaya, are thought to have particularly high rates of maternal death. However, like so many results of demographic research these facts may be less truths than artifacts, the unintended consequence of misrecorded information misinterpreted by misguided observers. The operative word here is may, since there is little in the public record which would allow us to test such an allegation. This research into maternal mortality in West Java is not intended to determine the level, or even the medical causes of maternal deaths in the province. Instead, we want to outline some reasons why it might be better to treat the mortality estimates with a high degree of skepticism. Though a shift away from quantitative measures may lose some precision in problem-identification and monitoring, it still allows us to seek the explanation for levels and trends with a broader framework of understanding the nature and process of such deaths. The justification for taking a new and untried approach to policy analysis and planning is two-fold. First, it is evident that despite over seventy years of attempts to determine maternal mortality levels and trends, the data available today are as bad or worse than the results of studies in the thirties, because registration systems are totally inadequate to the task of measuring rates, and indirect estimation techniques are fraught with biases. The latter are also inherently based on experiences many years prior to the data collections. Second, even if we knew maternal mortality levels with precision, it is

17

unlikely that the accuracy of measurement would make any contribution to formulating more relevant or focused initiatives to control the conditions producing the deaths. The determination that maternal mortality ratio is 300 or 400 or 500 per 100,000 births will not affect the knowledge that the causes of death are relatively specific, and the setting of the mortality is shaped by geographic challenges and economic barriers. Suggestions for controlling the mortality rest less with the levels than with the different roles that families, communities, and the medical profession play in caring for pregnant women in different ecological settings. If many of the estimates of maternal mortality ratio1 (MMR) at the provincial level are invalid the problem is compounded at the kabupaten and kecamatan level where difficulties of small sample size and incorrect study design make estimates unreliable. Yet the demand for mortality estimates by small administrative units is fueled by the tendency for projects to be designed and implemented in small ‘experimental’ areas. In the absence of complete and accurate registration statistics, project staff often conduct their own surveys, but funding considerations frequently make these efforts inadequate to measure the relatively small changes which are likely to occur over short periods of time, much less divine the relative contribution of project activities and autonomous changes to the differences in measured mortality. Added to this is the tendency for the staff responsible for the ‘baseline’ to have moved on to other jobs before the final evaluation is designed and implemented, so frequently incomparable methods are applied to what is already a complex evaluation task. Changes in mortality are likely to be related to a variety of factors beside the specific intervention introduced by a single project, but project staff and policy makers persist in looking for impacts in terms of ‘ultimate’ mortality objectives rather than being content with the establishment of ‘intermediate’ successes in improved processes of health services or social behavior. Moreover, projects which target specific interventions (e.g. GOBI-FF of the WHO) commit financial resources which need to be matched by organizational and staff attention -- and this carries a cost for the operation of other, nonselected interventions (such as malaria, adult tuberculosis or rational drug policies), which may also have important impacts on maternal mortality. Again, the pressure for short term, selective intervention project-based strategies arises when mortality becomes a political issue transcending the narrow confines of public health concerns. To call for a non-quantitative approach to the study of maternal mortality implies that the focus of attention should shift to the social process producing deaths, but this is not an easy issue for study. For data to be of optimal accuracy, they must refer to the relatively recent past -- certainly no more than a year retrospectively. Community studies, and the classic anthropological method of participant observation, can be undertaken to show how death occurs in various physical and economic environments, but they risk the problem that even high levels of mortality produce relatively few deaths each year in a 1

The Maternal Mortality Ratio compares the annual number of maternal deaths to 100,000 live births, while the Maternal Mortality Rate calculates the number of deaths among a group of 100,000 women of childbearing ages.

18

village or manageable group of villages. Clinical studies can be designed to cover a larger number of deaths, but they would tend to overlook women who gave birth and died outside the medical service system. Thus to design a qualitative study which covers enough cases, with an appropriate consideration of the normal setting of deaths, there is a need to look at fairly large regions, and ensure that the information on those cases is as full and representative as possible. It is to achieve these characteristics that the present study proposes an innovative approach to data collection and analysis focusing on a variety of versions of stories of maternal death in West Java. 1.2 The Rashomon Technique: an experiment in qualitative research The Rashomon Technique takes its name from the famous 1950 Japanese film directed by Akira Kurosawa. The plot of the film was based on two short stories (‘In a Grove’ and ‘Rashomon’) written in the early part of the century by Ryunosuke Akutagawa, exploring the dilemmas of reconciling different ‘truths’ related by a number of participants in an emotionally disturbing experience. (See Appendixs1 for an exposition of the Rashomon story). In Rashomon the incident was a rape and murder, and the viewer is impressed with how each participant's story has a vivid ring and detail which makes it believable, yet each is so contradictory to the others that the notion of one being ‘true’ is untenable. Readers of the story -- and viewers of the film -- come away unsettled by the notion that there may not be any single truth, just different perceptions and interpretations of realities that preclude any single determination of truth (see Appendix 1 for a detailed analysis of the original Rashomon case). But the story can be interpreted in a different way. Not only can the reader takes some confidence in assuming that there is a real truth to be explored -- at the most basic the truth is that a woman was raped and a death did occur -- but also the fact that different participants have different perspectives on the meaning and causes of the events does not mean that there is no possibility of understanding the events in terms of some externally formulated judgmental perspective. Akutagawa challenges the literalist reader to abandon attempts at such external evaluation by throwing the whole story into the supernatural realm, and presenting one witness speaking from the afterlife. However, even this need not prevent us from seeing that the basic lesson is that all the witnesses -both living and dead -- have unique perspectives on important shared events, and these perspectives can be understood, compared, and analyzed in the context of an externally imposed, ‘objective’ (or at least purposive) set of criteria representing the reader's own personal search for an important truth. We have developed this notion into an approach to the collection and analysis of qualitative data related to a dramatic shared event, in this case a maternal death, or lifethreatening situation. The setting of our application of the approach is rural West Java, an area with high maternal mortality. The basic notion is that for each case, there are a number of ‘witnesses’ or participants who can be relied upon to provide a detailed account of what they experienced and observed, and what they interpret as the causes

19

and responsibilities involved in producing the particular outcome -- a maternal death, or survival. The heart of the technique lies in four steps: 1.

Identification of the 4-6 witnesses for each case. These normally include the husband; the birth attendant ( either the traditional dukun bayi or the modern bidan); close relatives such as mother or grandmother, uncle or aunt; close neighbors or friends; village officials directly involved in attempts to transport the woman to health facilities; and medical personnel directly involved in treating the patient.

2.

Taping and preparation of transcripts of detailed, depth interviews of their stories. Normally the transcript is of between 7 and 20 pages in length, and is an accurate record of every word and phrase used by the respondent -- in the respondent’s chosen language of conversation. It is necessary to annotate the transcript with Indonesian translations of phrases, and to make a complete translation of interviews conducted in the indigenous languages of West Java including Bahasa Sunda or Jawa.

3.

Analysis of the meaning of their stories, both in terms of what they say about their own roles in the events, and how they interpret the roles of other witnesses, and what they interpret as the cause of death. The most important insights afforded by the Rashomon approach are based on analysis of the differences of perception and interpretation among the witnesses.

4.

Analysis of the meaning of the full set of interviews, and summary analysis of the whole case, from the perspective of the investigators. This analysis includes an assessment of the roles of delays in various stages of the process of seeking medical services; and a judgment of the role of individual witnesses, and interactions among witnesses, in producing the outcome.

In developing an analytical strategy to process both the full transcripts and our summaries of various perceptions and reports by witnesses, we have not attempted to prove any particular witness right or wrong, nor are we attempting to assign individual blame for the sequence of events. Rather in stage (4) above, we develop an additional analysis based on our (external) perspective, with a concentration on the medical issues surrounding the events. The material used to develop this assessment includes the full transcripts of the witnesses' stories, any other evidence the research team can collect from the field (such as clinic records, police reports, death certificates and other documentary evidence). These lead to a set of conclusions -- and a set of themes -- about both the relative importance of specific events and the dynamics of specific relations among witnesses, to reach an external assessment about why the death occurred, and what policy and procedural changes implementable by health service and local government agencies could prevent such events in future.

20

It must be stressed that the two stage analytical strategy -- assessing the witnesses perceptions and developing an external assessment -- are not designed to develop judgments on the witnesses' behavior, or to question or denigrate their perceptions of the causes of events. Rather the external assessments are directed at translating that evidence into an unashamedly medicalised, policy-oriented set of perceptions which will inform improvements to medical services, public health campaigns and local government administration in line with the objective of reducing maternal mortality as fast and as far as possible. In other words, the research takes a specifically emic procedure in data collection, and a specifically etic perspective in data analysis. It is not a classic anthropological attempt to study the indigenous concepts of health and illness, though it does borrow some techniques from anthropology in the use of in-depth interview techniques. Instead it is a policy-oriented research approach attempting to understand the variety of perceptions which must be addressed in any intervention to prevent maternal morbidity and mortality. One of the foundations of the Rashomon techniques is the idea that there are already substantial amounts of data available on the epidemiology of maternal mortality. The major causes of maternal mortality in West Java are the classic triad of bleeding, infection, and eclampsia. Many mothers have inadequate ante-natal care and are in a poor nutritional state during the course of their pregnancy. This means that their babies are often born underweight for gestational age. A major factor in maternal mortality is the delay which prevents early detection and correct treatment of problems. The major distinctions of the Rashomon Technique compared to other forms of qualitative social research are: 1. 2.

3.

Concentration on a dramatic shared event, Non-judgmental depth interviews of a limited number of witnesses who have clear and specific roles in the event, to establish their personal interpretation of the causes of the event, On the basis of the analysis of the various ‘truths’ obtained from the witnesses, and supporting information collected from a variety of sources, the analyst develops an ‘external’ evaluation of the evidence, to draw conclusions with direct policy relevance.

None of these things will be surprising to medical personnel familiar with the problem of maternal mortality in Indonesia. What is still at issue -- and what we cannot learn through conventional surveys or clinical studies -- is the question of why delays occur in particular cases, and how circumstances converge to produce a death in some cases, but survival in others. Of particular importance is the need to understand what the family members and health personnel perceive as problems, and whether they perceive delays in treatment as matters over which they have any responsibility or control. The need to accurately discover the participants' perspectives means that the data collection instruments should, as much as possible, be free of assumptions concerning division of responsibility or the causes of events in particular cases.

21

1.3 Reexamining the witnesses: cases of maternal deaths In this study the perspective of Rashomon is applied to obtain a new understanding of the conundrum of field work on maternal mortality. When investigating the setting, background and reasons for a maternal death the investigator is faced with the prospect that each ‘witness’ will have a vivid but selective memory of events. Often, too, the stories are inconsistent, and at times the witnesses are very defensive, because they want to avoid any hint that they might personally bear some responsibility for the death of a young woman in childbirth. Accepting this fact from the outset, the investigator is drawn to collect and analyze field data in a more open, sensitive and detached way -- adopting the perspective honed by Akutagawa in his stories (see appendix 1 for a summary of the story). Rather than checking each witness' story with an eye to ‘rejecting’ things authorities or other witnesses claim to be untrue, we are bound to accept each witness' account as representing their ‘personal, public, truth’ at least at the time they are interviewed. The Interview Guide is simple: Who, What, When, Where, How and Why? The task of the Interviewer is to fully and faithfully record the story of the maternal death or crisis, from the personal viewpoint, and in the precise words of the witness. Leading questions should be avoided, though the interviewer should always ensure that the major issues of the case are covered by the witness, to the limit of their perceptions and abilities. Then the task of analysis is an exercise in comparison of different perspectives among witnesses, and development of the external perspective of the case (the investigators’ perspective) as a means of drawing conclusions which are both independent and systematic, and in this way have the best chance of being policy-relevant. Of course the standards and perspectives and judgments of the principal research team may not always be the same as those of readers of the research report or of policy-makers, so it is important that sufficient information from each case is recorded and preserved to allow the possibility of reinterpretations of the results. The basic approach is to record the story of each witness fully, accurately and without challenging their personal ‘truth’. Then it is important to compare and contrast the different versions to try to develop a reasonable summary of events surrounding the death. The analysis of the data should consider the content and meaning of each witness' story to see where there are patterns of belief related to social and professional roles, and draw conclusions as to where and how intervention programs can change the circumstances and information environment in order to prevent a repetition of decisions and actions (or inaction) which led to the death. Preliminary tests with the approach undertaken in Pasar Minggu, Jakarta, in December 1994 revealed that there is often some difficulty locating and interviewing key witnesses. In one case family members refused to speak about the death of a young

22

mother at all until the forty-day post-death ritual had been undertaken to mark the transition of the soul to heaven. Prior to that ceremony, talk of the circumstances surrounding the death was thought to upset the soul, but after forty days family members spoke freely, frankly, and passionately about all the events leading up to the death of the mother.

1.4 Selection of a comparison group: lucky survivors Just as a maternal death is a shocking occurrence affecting each witness and participant in dramatic ways, so the events surrounding an obstetric emergency which ends with the survival of the mother and child can be seen as a dramatic success story. Similarly the stories of each witness will differ, but now in the direction of embellishments to stress the way each participant contributed to survival and what they saw as the important elements of success. Again there is a need to record, analyze and evaluate the different versions to gain an understanding of how events leading to survival differed from those leading to death, and what decisions were of crucial importance in the different outcome. For this purpose a small number of comparison cases were studied using the same data collection techniques as those applied to the cases of maternal death.

1.5 Selecting witnesses of relevance Specific efforts were made in each case to analyze the links between different ‘witnesses’ defined in terms of their social roles, relation to the victim, and potential role for saving the situation under different circumstances. Because each case was ‘unique’ in terms of the number and nature of people taking specific actions, it is not possible to specify a full set of relations to be analyzed, but normally four key actors were examined in depth to gain an understanding of their decisions, or indecisions, relating to the case. These were the Husband, the Mother or other female relative of the woman and a key modern health provider (doctor or midwife) or traditional health workers (dukun). In many cases the potential witnesses can be seen as falling into two groups. First those who are immediately related to the victim, and who might be expected to have been involved over the full course of events surrounding the death, and second, those who may have been called in later in the course of events, and who might have limited, though possibly strong, knowledge of the events. Generally deliveries in West Java occur at home, so the most proximate witnesses would be family members and traditional birth attendants. In cases of deliveries in a village birthing house (polindes) or a health center (puskesmas), the midwife or doctor might be regarded as the most proximate witness to the events. Diagram 1 is suggestive of the selection criteria. If one or more of these was not present, or if other actors had much more important roles, the team was forced to consider the degree to which a particular line of enquiry would be pursued. Decisions to include supplementary witnesses were made by the research team on the basis of the basic story presented by family members and health

23

personnel, and the need for elaboration or confirmation of elements of the stories. In some cases new witnesses were sought to confirm problematic accounts, in other cases witnesses were re-visited to request further information. In a few cases it was clear that the pursuit of further information would not necessarily resolve differences among the witnesses, and the team accepted the confusion of accounts as inevitable consequences of the method. In a sense, this was simply another issue to be analyzed, rather than a problem to be resolved. Diagram 1. Dramatic, emotional event

Maternal Death

Family members or close neighbors or friends. Most proximate witnesses

Medical or bureaucratic personnel Secondary witnesses

Husband Mother Neighbor(s) Sister Dukun Bayi

Specialist Doctor Midwife Nurse Village Official

To summarize, the Rashomon Technique constitutes a new means of collecting and analyzing qualitative data in demographic studies of mortality. In the case of maternal mortality in West Java, the approach used the focus of a single dramatic event, a maternal death, to draw out a variety of perspectives from different members of the community. By selecting witnesses who were both close to the event, and potentially crucial in determining the outcome, the research team was able to open to question a variety of assumptions underlying policy interventions. The metaphor of Rashomon is a powerful tool for training interviewers, desensitizing respondents and disciplining analysis. By directing attention away from the threatening notions of guilt and innocence, and legitimizing the idea of different, but equally valid, truths, the technique opens the door to ideas which might not otherwise be countenanced by medical practitioners, government officials, or community leaders. These are ideas surrounding the notion that the perspectives of the parents and traditional healers need to be fully analyzed and fully integrated into policy discussions if we are to effectively address the problem of maternal mortality.

24

Chapter 2. Review of the literature on maternal mortality 2.1 Maternal mortality levels and trends It is widely agreed that maternal mortality rates and ratios in Indonesia are unacceptably high. Ironically the data on which that judgment rests is remarkably unreliable. Different studies report a wide range of estimates of the maternal mortality ratio from a couple hundred to four times that level, but an estimate of 400 maternal deaths per 100,000 live births has been generally accepted as the prevailing level (see CBS, NFCB, MOH, and MI, 1995: 211; see also Table 2.1). This estimate suggests that about 19,000 Indonesian women die every year from complications of pregnancy, abortion attempts, and childbirth.2 This consensus understanding rests on studies carried out over the last decade. These studies use a wide variety of data sources and calculation methodologies. The 1986 National Health Survey (GOI-Unicef, 1988) and the 1992 National Health Survey (Soemantri, 1994), reported 450 and 425 maternal deaths per 100,000 live births respectively. The 1994 National Demographic Health Survey (IDHS) reported estimates of 326 for 1981-82, 360 for 1984-88, and 390 for 1989-94 (CBS, NFCB, MOH, and MI, 1995: 211). This pattern of apparently rising mortality based on one data source is confused by the fact that different techniques are used to calculate the first and the latter two estimates, and the sisterhood method in general has been called into question. 2

This estimate is based on 1993 official statistics: Indonesian population of 190 millions and crude birth rate of 25 per 1,000 live birth (MOH, 1994). According the World Health Organization (Royston and Armstrong, 1989: 11), a maternal death is defined as the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and the site of pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes.

25

Table 2.1. Estimates of maternal mortality ratio (MMR) in selected areas of Indonesia Locationa

MMRb

Time reference

Type/area of study

Indonesia NHS 1985

370 450

1978-1980 1985

NHS 1992

455

1991

404

1991

390

1989-1994

360 326

1984-1988 1981-1982

West Java

490

1977

Sukabumi, West Java Tanjungsari, West Java Kuningan, West Java Central Java

470

1982-1983

490

1988-1989

360

1977

360

1977

Central Java

340

1987

Mojokerto, East Java

397

1977

Bali

331

1982

Bali

230

1985

12 hospitals Chen et al. (1981) Retrospective Budiarso (1986) (direct), 7 provinces Prospective Kosen and Soemantri (1994) Retrospective Badan Litbang and LDUI (1994) Sisterhood (direct) CBS, NFCB, MOH, and MI (1995) Sisterhood (direct) Sisterhood (indirect) Sisterhood Budiarso (1991) (indirect), 8 regencies Prospective, rural Budiarso (1989) areas Prospective, rural Ngantung (1990) areas Sisterhood Budiarso (1991) (indirect) Sisterhood BKKBN (1990) (indirect) Prospective, rural Agoestina (1989) areas Sisterhood Soemantri (1989) (indirect), 19 subdistrict Sisterhood Wirawan and (indirect) Linnan (1994) Retrospective Budiarso (1986) (direct)

IDHS 1994

Sourcec

Source: modified from Soemantri (1995: 10). a NHS = National Health Survey; IDHS = Indonesia Demographic Health Survey. b Number of maternal deaths per 100,000 live births. c Complete references for Kosen and Soemantri (1994) and Ngantung (1990) were missed in Soemantri (1995).

26

Two different studies conducted in Bali reported very different estimates of maternal mortality; Fortney (1985) reported 718 maternal deaths per 100,000 live births, but, using the indirect sisterhood method, Wirawan and Linnan (1994) reported an estimate of only 331 maternal deaths per 100,000 live births in 1982. On the other hand, studies conducted in various parts of West Java reported 450 to 650 maternal deaths per 100,000 live births (Budiarso and Setyowati, 1991; CCSUI, 1991). The latest estimate of maternal mortality using an internationally based regression estimation procedure, with inputs from the 1994 IDHS was 647 per 100,000 live births (Stanton et al. 1995: 20). In short, estimates of maternal mortality ratios in Indonesia are all high, but there is no consistency in the levels which would allow any speculation to be made about trends. In contrast, maternal mortality ratios in developed countries are very low, 7 to 15 maternal deaths per 100.000 live births (Lettenmaier et al., 1988: 2; Rosenfield and Main, 1985: 83). Thus, compared with developed countries, the risk of maternal death in developing countries is very high, about 45 times higher, but, on the other hand, the risk of infant death is only 10 times higher (see Bryant, Khan, and Thaver, 1990: 86; McCarthy and Maine, 1992: 23). As women in developing countries become pregnant more often, the average lifetime risk of dying a maternal death for a woman in developing countries is even greater than in developed countries; 1 in 21 in Africa, but 1 in 9,850 in Northern Europe (Maine and Allman, 1990: 31). For one maternal death, many more suffer from chronic, even serious maternal morbidity complications (Lettenmaier et al., 1988: 3; McCarthy and Maine, 1992: 23). Studies in India and Bangladesh suggested 17 to 70 episodes of serious illnesses related to pregnancy and childbirth for every maternal death (cited in Goodburn, Gazi, and Chowdury, 1995: 22). Maternal death is not only a tragedy for the victim, but also has a devastating effect on the remaining members of her family (Abbas and Walker, 1986: 404), especially among her children. In many cultures, including most cultures of Indonesia, a mother occupies a pivotal position in the family (Soekirman, 1994). In Indonesia, the survival and the fate of young children often depends on their mothers (Leimena, 1955: 77). A maternal death is often closely followed by the death of the infant. Two studies in Bangladesh, one by Chen et al. (1974, cited in Winikoff and Sullivan, 1987: 128) and another by Koenig et al. (1988, cited in Tinker and Koblinsky, 1993: 4) reported that 90 to 95 per cent of newborn infants whose mothers died in childbirth, themselves died within one year of birth. Further it is estimated that for every dead mother, on average two children will be left motherless (Winikoff and Sullivan, 1987: 128). Documentation of the fate of motherless children is scanty, but motherless children will obviously receive much less optimal care and health protection than children whose mothers survive (Leimena, 1955: 77; Winikoff and Sullivan, 1987: 128). Thus, one potential impact of the prevention of maternal deaths is the better care and more ensured survival of infants.

27

2.2 Obstetric causes of death One-fourth to one-half of deaths among women in developing countries are maternal deaths (Lettenmaier et al., 1988: 2), compared with less than one per cent in the United States (Rosenfield and Maine, 1985). Most maternal deaths in Indonesia, 75 to 85 per cent, are related to one or in combination of the following three maternal morbid conditions: hemorrhage, infection, and eclampsia (GOI-UNICEF, 1988). These maternal morbid conditions are normally classified as direct obstetric causes of maternal deaths (WHO, 1985: 75). Underlying causes of hemorrhage, infection, and eclampsia may differ from one community to another. In communities where many young women are pregnant, long labor or obstructed labor may play as the main underlying factor. On the other hand, in communities where family planning services are not available or under utilized, illegal abortion may act as the main underlying factor. One study in Bali (Fortney, 1985) showed that 67 per cent of maternal deaths were related to maternal hemorrhage (bleeding). Most cases of maternal hemorrhage are postpartum hemorrhage, especially found in women with high parity (Lettenmaier et al., 1988: 3). The condition is often related to improper uterine contraction such that the placenta is not completely peeled off and bleeding continues after labor. Other causes of postpartum hemorrhage include long labor, uterine rupture, cervical or vaginal laceration, and placenta previa. If not immediately stopped, bleeding that occurs during or around labor can quickly result in maternal death. Because bleeding often cannot be handled by community members, the occurrence of hemorrhage is obviously dangerous for expecting mothers who are far from modern obstetric care and blood transfusion facilities. The fatality from hemorrhage may be related to anemia; hence, the inability or inadequacy of red blood cells to carry oxygen. The type of anemia frequently found in pregnant mothers is a nutritional anemia in which the nutrients, particularly iron, folate, and/or B12, are inadequate. Nutritional anemia inhibits the production of red blood cells. A pregnant woman is considered 'anemic', if her hemoglobin level is less than 11 gram per cent (WHO, 1972). The lower the hemoglobin level of the pregnant mother the higher the risk of maternal morbidity and mortality. One study in Indonesian hospitals shows that the risk of maternal death is 4 times greater for anemic woman than for nonanemic woman (Chi, 1981). If the hemoglobin level is less than 8 gram per cent, the risk of maternal death is about 8 times higher than for non-anemic woman (Belsey and Royston, 1987). The prevalence of anemia among pregnant mothers in Indonesia is very high, higher than the reported figures of ASEAN countries (Singapore, Malaysia, Thailand, and Philippines) (World Health Statistics, 1982). The 1986 National Health Survey reports that 75 per cent of pregnant women are anemic. The average hemoglobin level of pregnant women is reported at 8.7 gram per cent (GOI-UNICEF, 1988); thus, the significant portion of pregnant women, at least 50 per cent, have hemoglobin under the safety level of 9 gram per cent (see WHO, 1972). The higher the age or the parity, the higher the level of anemia, thus compounding the risks of death should complications arise.

28

Unless the mothers suffer from pre-established sexually transmitted diseases, such as gonorrhea or chlamydia, infections are unlikely to occur in the course of normal uncomplicated delivery. However, long labor, early rupture of amnionic membrane, frequent vaginal examinations, use of unsterilized instruments and unhygienic environment of labor increase the risk of infections. Illegal abortion is particularly associated with the high incidence of infections. In Indonesia, the reported rate of complications due to abortion vary from study to study, ranging from 1.9 per cent to 29.4 per cent (Lim, 1989; Rattu, 1973; Sopachua, 1974; Manuaba, 1979; Effendi, 1980; and Manuaba, 1980). Complications are characterized by bleeding and fever, which indicate the occurrence of infections, such as endometritis, adnexitis, pelvic inflammatory diseases or septicemia. Other forms of complications include tetanus, uterine perforation, and laceration of cervix. It is highly possible that abortion complications lead to maternal deaths. One study in Manado reported that 0.74 per cent of all abortions and 3.28 per cent of illegal abortions ended into maternal deaths (Rattu, 1973). Pre-eclampsia is a condition of pregnancy characterized by high blood pressure, edema of lower extremities, and proteinuria. This condition normally occurs in the second or third trimester of pregnancy. If not properly managed, pre-eclampsia could lead to eclampsia, a condition of pregnancy characterized by very high blood pressure and convulsions or cerebral hemorrhage or premature separation of the placenta. These pathologic conditions not only threaten the life of the mother, but also the growth and survival of the fetus. Pre-eclampsia and eclampsia are often found in first pregnancies. Reported rates of pre-eclampsia in first pregnancies are in the range of one per cent to 31 per cent (Lettenmaier et al., 1988: 4). This variation may relate to genetics and nutritional patterns. The overall rate of eclampsia in pregnant women, however, is reported to be less than one per cent (Lettenmaier et al., 1988). Once a woman has entered into the eclampsia condition, treatment and delivery should take place immediately. Even if already treated, the prognosis for eclampsia is usually not good. Both in developed and developing countries, five to 17 per cent of mothers with eclampsia will die. Those who are able to survive may suffer from chronic illness, such as paralysis, blindness, high blood pressure, or kidney malfunction (Lettenmaier et al., 1988). One study in 12 teaching hospitals in Indonesia showed that eclampsia is third among the causes of maternal mortality with PCSMR (proportional cause specific mortality rate) of 24.4 per cent (Chi, 1981). A study in Cipto Mangunkusumo Hospital - Jakarta, reported eclampsia PCSMR of 31.1 per cent and eclampsia CFR (case fatality rate) of 11.1 per cent (Cited from Gani, 1987). 2.3 Determinants of maternal mortality To avoid confusion, the term ‘causes’, which refer to immediate obstetric causes of maternal death, is differentiated from the term ‘determinants’, which refer to factors determining maternal mortality level in a population at a given time or place (Loudon, 1992: 43). There have been two different views on maternal mortality determinants. The first view gives much emphasis on clinical standards of care, suggesting that maternal mortality level is determined most of all by the quality of obstetric care. Thus, high

29

maternal mortality is related to poor obstetric practice, and low maternal mortality is related to good obstetric practice (see McCarthy and Maine, 1992: 28). The second view, on the other hand, emphasizes standards of living. This view argues that concentration only on the quality of obstetric care may obscure the real or underlying causes of maternal mortality, notably, poor maternal health as a result of social, economic, and nutritional deprivation (Loudon, 1992: 44). Several authors systematically address the process that culminates in maternal death or serious maternal morbidity. Fathala (1987) described ‘the long road to maternal death’ that women follow, from underlying socio-economic conditions of life to demographic and health service factors that contribute to death. Thaddeus and Maine (1990) believes that ‘delay’ is the most pertinent factor contributing to maternal deaths. Hence, delay could be divided into three phases: the phase 1 delay, that is the delay in the decision to seek care (modern maternal health care); the phase 2 delay, that is the delay on arrival to the care facility; and the phase 3 delay, that is the delay in receiving the adequate care (Thaddeus and Maine, 1990: 3). The patient’s decision to seek care, the availability of transportation, the condition of the roads, the facility’s capabilities to deal promptly with obstetric complications, can all be implicated in a maternal death by causing delay. Thus, delay is a concept that unifies a number of seemingly disparate factors, such as distance, women’s status, distribution of facilities and shortage of hospital or medical supplies. Recently, McCarthy and Maine (1992) and then Tinker and Koblinsky (1993) proposed a conceptual framework outlining the linkages between maternal mortality and morbidity and the proximate, intermediate, and distant causes, which include aspects of the social, cultural, and economic environment. In that framework, healthy women, healthy newborns, and maternal or perinatal morbidity and mortality are considered as maternal health outcomes. The proximate determinants, that are closest to maternal health outcomes, include pregnancy, the development of pregnancy related complications, and the management of pregnancy, delivery, and the postpartum period. These proximate determinants are affected by intermediate determinants, which include access to quality of family planning and maternal health care, women’s reproductive and health behavior, and women’s health and nutritional status. The intermediate determinants, in turn, are influenced by distant determinants, which include women’s status, political commitment, socio-economic development and cultural background factors (McCarthy and Maine, 1992: 25-26; Tinker and Koblinsky, 1993: 11). Thus, the framework suggests that programs to reduce levels of maternal mortality must operate through sequences focused on three intermediate outcomes. These sequences aim to: reduce the likelihood that a woman will become pregnant; reduce the likelihood that a pregnant woman will experience a serious complication of pregnancy or childbirth; and improve the outcomes for women with complications (McCarthy and Maine, 1992). The health status of woman prior to and during pregnancy can have an important influence on her chances of developing and surviving a complication. Pre-existing morbid conditions like malaria, hepatitis, anemia, and malnutrition may worsen during

30

pregnancy and aggravate the seriousness of pregnancy complications (Royston and Armstrong, 1989: 107-136). Reproductive status, such as age and pregnancy order, is known to have ‘j-shaped’ relation with maternal mortality ratio, with risks that are high for young women, older women, women with no children, and those with many children, but are lower for women in between (McCarthy and Maine, 1992: 27). Access to health services that can provide safe and successful births is generally limited. In many settings, especially rural areas, the physical distance between services and women in need of obstetric care is considerable. Not only long physical distances, but also financial barriers, shortages of trained personnel, and poor performance in the part of trained personnel contribute to high maternal mortality in developing countries (cited in McCarthy and Maine, 1992: 27). Use of health services and other health care behavior are also likely to have important influences on pregnancy outcomes. For health services to be effective, women have to use them. Many women in developing countries, especially in rural areas, often prefer using traditional than the available modern health services (Goodburn et al., 1995; Utomo et al., 1992). In West Java, for example, most of birth deliveries occur at home attended by traditional birth attendants (CBS, NFCB, MOH, and MI, 1995: 157-158). Despite a number of rational measures devised to ease and facilitate labor and delivery, certain traditional health care practices are potentially dangerous and can imperil the life of both mother and child (Bathia, 1981; Utomo et al., 1992; Goodburn et al., 1995). Other variables, such mother’s education, workload, and morbid condition during referral and transportation, are also related to maternal mortality (Lettenmaier et al., 1988; Okafor and Rizzuto, 1994: 353-354).

2.4 Health services and programs to reduce levels of maternal mortality Health technologies for preventing deaths from obstetric complications have existed for decades: antibiotics for infections, cesarean section for obstructed labor, blood transfusion and oxytocin for hemorrhage, and sedatives and other drugs for eclampsia, but for various reasons they are not accessible to most women in poor countries (Maine and Allman, 1990: 34). Thus, the problem is not that the technology is not known; rather it is that health services are ‘underdeveloped’ (Bryant et al., 1990: 88). Health services in developing countries usually follow the models of Western health systems, which concentrate substantial capital in urban medical centers, relying on patients to somehow make their way to the specialised facilities. Such systems tend to consume a large portion of a poor nation’s health budget, yet serve only a minute portion of its people, because complementary non-health infrastructure is so weak (Rosenfield and Maine, 1985: 83).3 Because of chronically low health spending levels, Indonesia continues to lag behind countries of similar per capita income as regards government outlays. For example, per capita government health expenditure in China and India exceeded that in 3

A similar statement on low level of health budgeting was given by the Indonesian Minister of Health (Kompas, 4 May, 1995: 3).

31

Indonesia, while outlays in Thailand and Malaysia are large multiples of the Indonesian figures (World Bank, 1994: 3). As a move away from Western models, the increased interest in Primary Health Care (PHC) since the Alma Ata Conference in 1978 offers a major improvement in the health system of developing countries. Unfortunately PHC, as currently defined and in the current setting, will have little impact on maternal mortality (Rosenfield and Maine, 1985: 83). The preventive measures adopted in the PHC, including screening during antenatal care, will not significantly avert hemorrhage and obstructed labor, the main causes of maternal deaths (Rosenfield and Maine, 1985: 84). In Indonesia, training traditional birth attendants in the absence of skilled backup support did not decrease women’s risk of dying once pregnant (Alisjahbana, 1991, cited in Tinker and Koblinsky, 1993: 7). A woman with obstructed labor who cannot get appropriate medical care, i.e., cesarean section, will probably die whether she is malnourished or not (Maine and Allman, 1990: 33). Although severe postpartum hemorrhage is commonest among high parity women, it is difficult to identify those women among the high parity group in which this complication will occur. Moreover, a significant portion of serious complications occur among women with no recognizable risk factors (McCarthy and Maine, 1992: 28). Even if community-based health services are able to prevent some obstetric complications and to manage the less severe cases (Rosenfield and Maine, 1985: 84), onward referral is still required for more serious maternal health problems such as eclampsia, obstructed labor, and severe hemorrhage (Bryant et al., 1990: 88). In many cases, in fact, the PHC does not reach the social and geographic periphery to identify women at risk, and the referral channels to life saving secondary care facilities are inadequate (Bryant et al., 1990: 88). Moreover, hospitals where emergency cases are referred often have chronic shortages of trained staff and essential supplies (Rosenfield and Maine, 1985: 84). While the number of health personnel and facilities is still considered inadequate, the available health services are, paradoxically, underutilized (World Bank, 1994: 4). Moreover, the posyandu, the services which reach deepest into the communities, are regarded by many clinic staff as heavy burdens. Rather than being seen as community initiated and run activities backed up by health personnel, the staff regard the activities as yet another thing they must supervise (Iskandar et al. 1994: 5). Socioeconomic development alone cannot reduce maternal mortality (Tinker and Koblinsky, 1993: 6). Obstetric problems can only be effectively managed by continuous, community-based medical interventions. The health services must be able to use potent drugs such as oxytocics and antibiotics, provide blood transfusions, perform obstetric surgery, and handle life-threatening complications (Tinker and Koblinsky, 1993). Because of life-threatening obstetric complications are often unpredictable, qualified maternal health services must be widely and rapidly accessible (Tinker and Koblinsky, 1993: 10). The important principles are to bring life saving obstetric functions as close to people’s homes, and to carry out these functions at the most peripheral level at which they can be undertaken safely and effectively. The Safe Motherhood Initiative launched at a conference in Nairobi in 1987 proposed a short-term strategy to make family planning

32

and maternal health care more effective - by improving quality, increasing access, and educating the public about the importance of such services and how they can best be used. A more comprehensive plan calls for improving women’s socioeconomic status through health, education and other factors (Tinker and Koblinsky, 1993: xiii). Programming for safe motherhood, however, requires local flexibility and initiative in planning, combined with strong national and local political support (Tinker and Koblinsky, 1993). The interventions being promoted by various safe motherhood projects include those that encourage change in the status of women; programs that offer family planning, prenatal care, nutritional supplementation, and tetanus immunization; programs that provide more effective linkages between traditional birth attendants and the modern health system; and programs that undertake overall improvements in access to and the quality of emergency obstetric care (cited in McCarthy and Maine, 1992: 23-24). In Indonesia, traditional birth attendants have long been the object of ‘training’ to teach them rudimentary principles of hygiene and provide them with a basic kit for assisting in deliveries. The first attempts at such training were in 1807 (Hull, 1988: 135), suggesting that the underlying notions of the safe motherhood initiative was not totally new for Indonesia. Until recently, however, the impact of such TBA training on the reduction of maternal mortality is questioned due to lack of continuity and lack of backup support (Alisjahbana, 1991, cited in Tinker and Koblinsky, 1993: 7). Traditional birth attendants have not been integrated into the health care system as acknowledged members of the health service team, nor have the positive or benign elements of their practices and beliefs been accepted as valid elements of maternity care procedures in the local setting (Hull, 1988: 136). By the isolation of the TBA not only is the potential for them to do good inhibited, but the possibility of controlling any harm they might do is also limited. Prompted in part by the Safe Motherhood Initiative Conference in Nairobi in 1987 and the flaws of TBA training program, the Indonesian government initiated the village midwife program in 1989 (DepKes R.I., 1989; Gunawan, 1993). Under that program, village midwives were to be posted in villages, especially those areas not easily accessible by health services. Basically, the village midwife program is designed to gradually improve the accessibility and the quality of maternal health services. It was planned that 18,900 village midwives would be posted during 1989 to 1994, and another 34,000 village midwives would be posted during 1995 to 1996 (Gunawan, 1993: 7).4 It was targeted that by 1995/96 there would be one village midwife per one village (Gunawan, 1993: 1). It was also expected that a village midwife should stay in the village. In terms of qualifications, the village midwife is a paramedic school graduate with one additional year midwifery training. To adequately perform their jobs, the village 4

It seems difficult to produce so many village midwives in a relatively short time. In West Java, for example, not only public, but also private hospitals are utilized to provide training for the procurement of new village midwives (Kompas, 13 November 1993). Such a quantitative orientation would be likely to compromise the training quality. In realizing the plan, a study by Pusat Penelitian Pranata Pembangunan (1993) showed that 4,000 new village midwives were produced in 1989/1990 and another 4,760 new village midwives were produced in 1990/1991.

33

midwives are equipped with various manuals of MCH-Family Planning programs5, a midwifery kit, and an IUD kit. In addition to financial support for accommodation, transport, and field activities, the village midwives should also receive technical supervision from the Ministry of Health, and administrative supervision from the local government (Gunawan, 1993).6 Ideally, village midwives should have the following midwifery skills: treating anemia cases, managing preeclampsia and eclampsia cases, conducting episiotomy, suturing birth canal laceration, managing hemorrhage cases by performing bimanual placental extraction, stabilizing cases before referral, resuscitating shock cases, managing infection cases, performing intravenous infusion, performing vacuum extraction, preventing hypothermia of newborns, basic care of newborn babies, and resuscitating the low APGAR babies (DepKes R.I., 1994b). The real impact of this village midwife program, however, is not yet known. More researches are needed to see whether the program is implemented as planned, and whether the program has an effect on the reduction of maternal mortality. The literature thus gives a comprehensive, pragmatic, yet proactive perspective on the medical service needs of pregnant women in poor economic and social environments. While arguing that poverty and ignorance can provide the conditions for high maternal mortality, the thrust of the research specifically addressing maternal mortality stresses that most of the direct causes of maternal death are conditions needing emergency medical treatment. Prevention has an important role to play, but this role does not guarantee a substantial reduction in maternal deaths in the abscence of commensurate improvements in the speed and quality of case referral. The basic strategy of the Indonesian program seems to fit with the findings of international policy argumentation -work on a variety of fronts at once, improving the status of women (through education), improving the biological condition of mothers (through nutrition and ante-natal care), improving the quality of delivery assistance (through training TBAs and village midwives) and improving the availability of emergency obstetric care (through the establishment of local hospitals with specialist doctors). The question is whether such a strategy works well enough, and is efficient enough, to reduce maternal mortality quickly. That is the question we have sought to answer through the study of 63 cases of obstetric emergency -- 53 of which resulted in death -- in West Java. 5

In 1994, a packet of new manuals for village midwives and new modules of life saving skills for midwives were produced (see Depkes R.I., 1994c; 1994d). 6 A report from the field, however, showed that the practice was not always in line with the plan. For example, some village midwives in north Sumatra who had been posted for one year never received any financial supports for their accommodation and transportation (Kompas, 8 April 1995: 15). In remote villages in West Java, some village midwives reported that they had to provide extra outlay to cover their transportation expenses (Kompas, 24 Mei 1994).

34

The low quality of health center services was observed in three recent micro-studies: One study in ten health centers in Lombok, NTB, by Neilan (cited in World Bank, 1994: 10) showed that doctors were not satisfied with puskesmas conditions, and complained about inadequate water supply, equipment and furniture. Yet the most salient issues were various deficiencies in the process of providing care to health center visitors, such as not washing hands before or after a delivery; frequent reuse of non-sterilized needles; perfunctionary physical examination and inconsistency between diagnosis and prescription, especially by nurses and paramedics; overuse of drugs and injections; limited attention of staff in instructing patients about appropriate use of medication, possible side effects and so forth. Another study in several health centers in Central Java by Sciortino (cited in World Bank, 1994: 10) showed that the effectiveness of health center staff was disrupted and undermined by several operational factors: ‘flexible’ personal interpretation of working hours; inappropriate deployment of personnel resources; haphazard and personally determined ways of conducting activities; conflicts between staff factions; inadequate biomedical knowledge among nurses; over emphasis on injections and less emphasis on examinations and communications; unwillingness to become involved in various community-based health delivery activities and failure to view villagers as partners in a health development process; arbitrary increases of fees; heavy requirements for duplicative and inconsistent reporting and recording activities; emphasis of quantitative target on service provision; supervision which was more ceremonial than functional. In a different light, a study conducted by Iskandar et al. (1994) in samples of health centers and posyandu in West Java, South Sulawesi, and East Java provinces also discovered deficiencies in service provision and supervision of services. However clients generally expressed satisfaction, though they said that they really had no way of knowing if the service was adequate or not. They merely trusted the health center personnel.

35

Chapter 3. The contexts of maternal mortality in West Java 3.1 Demographic context The region of West Java, as well as its Sundanese people, are paradoxical when seen in terms of the current national development process. It is a region where major centers of Indonesian population and economic growth are located, and at the same time it is a region where various socio-cultural and demographic indicators7 of poor local areas are at odds with the general trends of development. West Java covers an area of about 32.2 percent of the Island of Java, or about 2.4 per cent of Indonesia. The province is administratively divided into 20 Kabupaten (districts or regencies) with 467 Kecamatan (subdistricts), consisting of a total of 7,101 desa (villages). Most of these desa are distributed near major growth areas such as the regencies of Bogor (530 desa + 22 in the vicinity of the city); Kuningan (569 desa), Bandung (448 desa + 135 in the vicinity of the city), Cirebon (424 desa+ 22 in the vicinity of the city), or Sukabumi (356 desa+ 15 in the vicinity of the city) where major socio-economic access, services, and facilities are better. In 1990, the total population of West Java was 35,385,000, or about one fifth of the total population of Indonesia. The population growth rate of the region seems to be falling, where between 1971 and 1980 it was 2.66 per cent, and between 1980 and 1990, it was reduced to an estimated 2.57 per cent8, which is still comparatively higher than the 1.98 per cent of the national average (CBS, 1994: 41). As of 1994 its total population was 37,918,186 (Pemda Jabar-UNICEF, 1995:11). This means that, currently, the total population of the province has an average density of 765 person per square kilometer (compared with the national average of 93 person per km square). The area indicating the lowest growth rates is Ciamis (0.8 per cent) while Bekasi has the highest growth (6.3 per cent). However, these figures seem to be rising with more influx of rural population, especially into urban major areas such as Bogor, Tanggerang, and Bekasi (the areas of Botabek). Population growth in the Botabek area was 5.1 per cent between 1980 and 1990. Overall in West Java there is an increasing trend toward urbanization since the 1970s. In 1971 12.4 per cent of the population was counted in urban areas, while in 1990, the figure rose to 34.5 per cent. Nevertheless, the larger part of West Java population live in rural areas. 7

Most of the data, unless indicated, are quoted from CBS (1993). The Government of West Java, based on Susenas data, estimated population growth between 1990 and 1994 to reach 1.73 percent per annum (Pemda Jawa Barat and UNICEF, 1995). 8

36

The West Java region is actually endowed with some of the most fertile land in the country and with ample natural resources such as hydro-power and natural gas. In fact, the multitude of rivers flowing in the region provides about 23 per cent of national electrical power output. In terms of its gross regional product, West Java also indicates higher rates of growth (8.2 per cent gross regional product; and 6.0 per cent gross product per capita) compared to national figures (average of 5.6 percent gross national product, and 3.4 per cent gross product per capita), while in terms of its overall economic growth rate the figure has been at a relatively constant 7.8 per cent per annum since 1988. Our research sites are all located in the hinterlands of West Java, i.e. in the Kabupaten (regencies) of Lebak, Sukabumi, Ciamis, Kuningan, and Purwakarta. Most of these sites are located within the southern mountainous geographical region, where their population densities are comparatively lower than other locations in the West Java region. In these areas too, poverty seem to characterize the population, with a quality-of-life index (based on infant mortality rate, life-expectancy rate, and literacy rate) comparatively worse than West Java as a whole, and some of its urban areas in particular as indicated below: Table 3.1 Quality of Life Index of Research Kabupaten and Selected Urban Sites, 1995 Sites

IMR

LE

LR

PQLI Index

59.00 55.67 59.03 62.97 58.17

85.90 91.78 91.56 87.97 91.91

72.88 71.40 74.66 79.84 73.64

Research Kabupaten Lebak Sukabumi Ciamis Kuningan Purwakarta

74.00 79.33 77.00 53.67 80.67

Urban Sites Bogor Sukabumi Bandung

31.37 74.87 99.61 96.60 46.35 65.84 99.12 86.13 45.13 69.07 98.97 89.08

West Java

69.67 60.10 89.58 75.65

Source: Pemda Jabar-UNICEF (1995:18) based on Indeks Mutu Hidup Jabar, Kantor Statistik Jabar, 1994. Notes: IMR= Infant mortality rate (per 1,000 live births) , LE= life expectancy rate in years, LR= literacy rate in per cent, and QOL=quality of life index.

37

Certainly, these figures have improved in the last decade. In 1980, for example, the quality of life index for West Java only reached 54.4 (compared to the national average of 57.0), while in 1990 the figure has improved up to 68.6, compared to the constant figure of the national average. Despite recent recorded improvements in the indicators of Infant Mortality Rate, Life Expectancy, and Literacy Rate, West Javanese still maintain strong traditions, as reflected in the number of people who still use of their mother-tongue, Sundanese (86.6 per cent in 1990). Only those in or around the Botabek areas have indicated high number of Bahasa Indonesia speakers. These ties to traditions, as we will latter see on in the following chapters, contributing to the comparatively low levels of the quality of life maintained and reproduced at the village level. It would thus be misleading to assume that West Java is a homogeneously well developed region when comparing its socio-economic indicators to other regions. This has been in part attributed to the uneven geographic and climatic conditions of the region with mountain areas dominating (about 70 per cent of the region) the southern region and while lowlands are characteristic of the northern coastline. Geography and climate create the variation in agricultural productivity and population concentration. This is also reflected in the uneven distribution of the population of West Java where higher percentages occur in, or near, the northern areas while the southern parts (Sukabumi, Lebak, etc.) are less densely populated. The population is also concentrated in larger urban centers of Bogor, Bandung, Bekasi, and Tanggerang, all of which are located in the northern low lands, and less so in the mountain areas of our research sites. 3.2 Fertility and marriage patterns in West Java Major indicators of demographic change can be reflected in the estimates of fertility, mortality, and migration. In West Java, fertility continues to be high relative to the rest of the country. In the span between 1967-1970, its Total Fertility Rate (TFR) amounted to 6.3 (compared to the national level of 5.6), while between 1981 and 1984, it fell to 4.3 (compared to the national level of 4.1). Although reported figures are higher, they nevertheless are actually declining, in part due to effectiveness of the national family planning programs. Similarly, child-woman ratio (CWR)9 has declined from 701 in 1971, to 636 in 1980, and down to 474 in 1990, indicating decreases of fertility rates in the successive five year periods prior to decennial censuses. Fertility trends are also influenced by the number of women ever married in West Java, and by this measure the proportion of women ever married between 1980 and 1990 have significantly declined by 4.4 per cent, i.e. from 74.2 per cent in 1980 to 69.8 percent in 1990. This decline was even more pronounced among those women in the 15-24 years old age group. 9

CWR indicates the number of children aged 0-4 years per 1,000 women in their reproductive ages of 1549 years.

38

The age at first marriage of West Java women has been increasing. In 1980, the median age at first marriage was 16.2 years, while in 1990 this figure rose to 16.9. However, by using Singulate Mean Age at First Marriage (SMAM) to calculate the mean age at marriage, the estimated average of age at first marriage in West Java in 1990 was still lower than the national average. For example, in urban areas, age at first marriage in West Java was 22.7 years (compared to national average of 24 years) while in rural areas it is 18.9 years (compared to national average of 20.4 years) for 1990 (CBS, 1993a: 2122). Other indicators of the impact of regional development relating to demographic changes are the infant mortality rate (IMR) and the childhood mortality rate10 (CMR). These figures are also indicative of a region’s overall health service delivery. In West Java, IMR is still comparatively higher than the national average. In 1980, the estimated IMR was 128 per 1000 live-births compared to the national average of 112. In 1990 IMR in West Java had declined to 83, but remain comparatively higher than the national average of 70 per 1000 live-births. As we scan the figures, it becomes apparent that the higher IMRs mostly occur in those socio-economically isolated areas, including in our five research areas of Lebak, Sukabumi, Ciamis, Kuningan, and Purwakarta. Childhood mortality rates (CMR) too in West Java have been declining, but are still comparatively higher than the national figures. In 1980 the West Java CMR accounted for 207 deaths per 1000 live-birth (compared to the national average of 165), while in 1990 the figure has declined to 134 (compared to the national average of 104). Furthermore, urban-rural disparity in CMR for West Java remained high (95 urban vs. 151 rural in 1990). Figures for IMR and CMR have shown declines, but it has not been so with the maternal mortality. There are no official figures on how maternal mortality rate currently stands in West Java, however various studies have indicated still significantly high figures (see Table 2.1). Although it is often blamed upon the individual population, whose health habits and practices are deemed as ‘harmful’ and unchanged, it is also evident that the economic environment has not allowed individuals to change for the better. But the fact that there are significant declines in both IMR and CMR, (though doubtful in MMR), indicates that, to some degree, development programs (including family planning and health services) have had significant impacts upon the population of West Java (even if only for those closer to urban areas). But what is also evident here is that these programs have not been evenly distributed, nor have they borne equal results throughout the region. Their inaccessibility (due to access, ignorance, traditional practices, and most importantly still low levels of education) has been a constant challenge to the government.

3.3 Levels of education and labor in West Java 10

Childhood mortality rate is defined as the probability of dying from birth to the fifth birthday.

39

If level of education reflects how people will be able to absorb new ideas, then the data revealed in the 1990 censuses show that people in rural areas are very much lagging behind their urban counterparts. This urban-rural disparity is due to the fact that schools are mostly located in cities and towns rather than in hinterland villages. In West Java the illiteracy rate was quite high (12.7 percent in 1990). Urban-rural and gender disparities reveal that there were 9.9 per cent of male and 21.4 percent female who are illiterate in rural areas and that there were still about 3.9 male and 10.3 female who are illiterate in urban areas. Data also reveals that the Kabupaten of Indramayu and Cirebon have the highest illiteracy rate in West Java (35.3 per cent and 7.5 per cent respectively) compared to Tasikmalaya (5.4 per cent) or Bandung (2.7 per cent). There is an increasing trend of people to complete primary education. But in rural areas by 1990, about 17.0 per cent have lacked any opportunity to attend school, about 37.2 percent had attended but not completed their primary education, and about 36.3 percent completed their primary education but failed to achieve a higher level of education. Recently the law has aimed to promote compulsory 9 years primary and lower secondary education. Among adults in rural West Java today only 9.5 percent have achieved this goal. These figures, however, are improvements over the previous decade, and reflect increased parental concern for providing basic education for their children. Gender disparities are still evident in rural West Java. There is still a discriminatory attitude toward girls, whereby parents wish to treat girls more as exchange (marriageable) commodities rather than as future insurance (capital or savings). In 1990, as many as 18.4 per cent of girls never attended school, and 32.9 per cent never completed primary education, while only 33.3 percent of the girls completed their primary education. Low levels of education results in low access to modern sector labor markets. In 1990, about a third of the working population (37.2 per cent) were absorbed in the agricultural sector. About 16.3 percent were involved some sort of manufacturing industry, and 18.9 per cent were involved in some sort of trading services, while about 14.1 per cent were involved in public sector services. The poor socio-economic conditions of certain areas of West Java has long been the motivation for women joining the prostitution industry. The areas in and around Indramayu, for example, have long been known as the source of supply of women for sexual services to the court of Cirebon. By this ‘historical accident’ the area has built a reputation for supplying sex workers to Jakarta and throughout western Indonesia. For the most part, the characteristics of poverty, low education, early marriages and high divorce rate, are contributing factors which encourage women into the sex industry in major cities of West Java (Jones, Sulistyaningsih, and Hull, 1995: 30, 39)

3.4 Transport and access to existing primary health care

40

Although road infrastructure has expanded and improved in the recent decades, it nevertheless is mostly concentrated in the northern lowlands while in the southern highlands it is much restricted in terms of length, accessibility, network and quality (Pemda-UNICEF, 1995: 21). As such, communication and service delivery to the southern areas of West Java have been rather slow to develop and have not been in line with the current population and economic growth of the northern areas. Even among better-off farmers in the southern areas, many are limited by the poverty of their environments in terms of access to, and availability of transportation to basic social services. In Lebak, Sukabumi, and Purwakarta, there are many settlements in mountainous areas with severely limited public transportation, causing people to be less inclined to access to the available health services. Most of the areas in both Lebak and Sukabumi are located on highlands, but access to public transportation are better in Sukabumi than in Lebak. Unlike Lebak, Sukabumi used to be a major area of tea plantations and infrastructure from the colonial era provided a foundation for developments in the last few years. Lebak, lacking such foundations has been relatively disadvantaged. In Ciamis too, most of the areas are mountainous, but they may be easily accessed through major (Kabupaten and Provincial) roads and other collector (Kecamatan) roads heading to tourist destinations, especially the Pangandaran Beach. In contrast, Kuningan is located in fertile highland plains of Mt. Ciremai, but access is comparably lower than Ciamis. In general, in many areas of the five selected regencies, regular public transportation (e.g. ‘colts’) cannot reach all villages, but ‘ojeks’ (hired motorcycles) are available to be used to reach most villages. In terms of general social service availability and delivery, access to and utilization of health services has been limited. To serve more than 35 million people, West Java only has 86 public hospitals (both private and government) with about 11,000 beds at district and higher administrative levels, and 1,058 Puskesmas (Public Health Centers) and 1,672 Puskesmas Pembantu (Pustu, or Support Health Centers), and 563 mobile units at subdistrict levels. In terms of utilization, data of hospital visits for 1993 indicate that more than 2.1 million people were treated in hospitals, of whom 78 per cent arrived spontaneously on their own accord, and 22 per cent were referred through health centers or other health facilities (Kanwil DepKes Jabar, 1994: Table 23A). Aggregate numbers can, however, be misleading. The ratio of Puskesmas per 10,000 population stands at only at 0.28 (Kanwil DepKes Jabar, 1994: Table 27A). At the desa levels, there were 47,022 Posyandu (integrated health post) at 1993, but only 511 Pondok Bersalin (Maternity Hut), most of which (273) are located in Bogor Regency (Kanwil DepKes Jabar, 1994: Table 27B). In terms of health personnel, West Java in 1993 only had 4,078 health professionals (specialist, general practitioners, dentists, and others), 2,393 nurses, and 1,059 midwives (Kanwil DepKes Jabar, 1994: Tables 30A; 30B). While at the neighborhood levels, there are 177,983 Kader, giving the ratio of 2 Kader per 100 households (Kanwil DepKes Jabar, 1994: Table 28).

41

Table 3.2 Number of puskesmas (community health centers) and posyandu (village level integrated service posts), in West Java, 1993

Location

No. of Puskesmas Research Areas Lebak 32 Sukabumi 51 Ciamis 55 Kuningan 33 Purwakarta 16

Ratio/ persona

No. of Posyandu

Ratio/ householdb

0.34 0.26 0.37 0.36 0.27

1,310 2,706 1,991 1,326 808

0.7 0.6 0.5 0.6 0.6

Selected Urban Sites Bogor 11 Sukabumi 7 Bandung 64

0.57 0.28 0.26

301 158 1,805

0.6 0.6 0.5

West Java

0.28

47,022

0.6

1058

a Number b

of puskesmas per 10,000 people. Number of posyandu per 100 households

Source: Kanwil DepKes Jabar Jabar (1994), Modified Table collated from Tables 27A, 27B and 28.

Since most hospitals are located in urban areas, i.e. Kabupaten towns or Provincial cities, they are distant in geographic, financial and opportunity cost terms, by most of the people in our research sites. Various health indicators show that health services are very much underutilized by most women and figures for West Java are often worse than national average. For example, the average frequency for pregnant women seeking antenatal care in West Java in 1992 is 3 times compared to the national figure of 4 times. Anemia prevalence among women in West Java in 1991 is 69.5 per cent compared to the national average of 55.0 per cent.11 11

Figures based on CBS (1991, cited in CHRUI, 1995, passim).

42

3.5 The traditional context The majority (98 percent in 1990) of West Javanese adhere to the Islamic faith, while the remainder profess other faiths (Catholic, Protestant, Hindu, Buddhist, and others). By way of tradition, the region can be divided into several distinct culture areas, namely the Parahyangan, or Pasundan (around Bandung), Botabek (industrialized areas of Bogor, Tanggerang, and Bekasi), Cirebonan (or Dermayon, including Indramayu), Banten (including the still traditional group of Baduis), Kuningan and Tasikmalaya (Tasik). These areas, among West Javanese, are regarded as having distinct culture, traditions, and dialects, and own socio-political and historical evolution.12 Banten and Cirebon were two major Islamic Sultanates up to the 15th Century, and the Kingdom of Galuh (Sumedang) may have had strong Islamic Mataram influence. The influence of the Sultanate of Cirebon still holds among the people in the area, while Banten’s influence ceased in the past century. Opposed to these were other groups of people (most likely Hindu descendants from the original Kingdom of Pajajaran) not wishing to be ‘islamised’ who moved deep into the southern highlands (e.g. the Baduis) or to other parts in the hinterland (such as the Tasik). Other areas in the northern coast13 have had an eclectic cultural history and their traditions reflect these mixed elements (e.g. Chinese, Muslim, and Central Javanese influences in Indramayu). While the Botabek areas have been shaped by rapid recent economic development, such that its proximity to Jakarta has made the area the major national center for industrial growth attracting many people from the hinterland. Thus, by virtue of its culture history, the Sundanese have inherited a blend of Islam, Hindu-Javanese, and part-Dutch colonial values, overlain with ethics of priyayi bureaucrats and local forms of petty aristocracy. The Sundanese language, with its three levels of appropriateness of address, is the reflection of this eclectic inheritance which continues to be reproduced among its current generation of people. Personal attributes, based on personal linkages, knowledge, perception and expression of the world (thus on the individual's place in a social cosmos) is the core notion of Sundanese society. Demeanor, wealth, and power (bureaucratic, mystical, or social) are indicators of cosmic powers. But these powers are ascribed mostly to men. The Sundanese culture and Islamic religion have predisposed women to be passive acceptors of fate (takdir, or nasib). What this has produced, ultimately, is the gender disparity seen in most aspects of sociocultural life of the Sundanese. 12

The Parahyangan area is also sometimes referred to as Tatar Sunda. Linguistically, the Pasundanan is divided by the rivers of Cilosari and Citandui. But because of historical process, some of people in the areas in Banten and along the northern coast speak Javanese. Those areas in the Pasundan (Ciamis, Tasikmalaya, Garut, Bandung, Sumedang, Sukabumi, and Cianjur are regarded were linguistic authenticity and purity are maintained, whereas areas in Banten, Krawang, Bogor, and Cirebon are areas where linguistic impurity are located. See Koentjaraningrat (1995: 307ff). 13 Sometimes referred to as the Pantura, literally: northern coast way.

43

The eclectic culture history of West Java has also brought about the production of strong class distinctions, with a ruling-class and aristocratic society, as evinced in the maintenance of stratified language. But the Sundanese have never had the extensive court-culture found in Central Java. As a result, at the community level, there is a sort of horizontal homogeneity in their social actions as reflected in the need to achieve constant social balance and harmony by means of collective decision making (musyawarah) and collective support (gotong royong). Vertical and horizontal opposition is resolved through a complex exchange of patron-client relationship (or bapak-ism), whereby lower-class persons are mobilized by means of vertical orders, instructions, and propagation of support, and by a mutual effort to please (the leaders, or elites) and be pleased (by the leaders). It is by this context of socio-cultural recognition that community participation in health programs have, to some degree, been produced in recent years (Sarwono, 1993). For many villagers, modern medicines are expensive and unavailable, so people have tended to treat their own illness by self-medication with traditional (e.g. use of jamu herbal medicine) to facilitate healing. Where available, people also use modern (off-theshelf) medicine, including penicillin or antibiotics.14 The combination of both modern and traditional medication is part of the effort to achieve harmonious balance of the real/natural and the physical/spiritual aspects of daily life. Various attributes and rules are often affected for this purpose, such as regular life-cycle rites, use of mantras, prayers (do’a), amulets, holy prayer water (air do’a), etc. Despite the inroads of modern medicines and services, traditional healing practices continue to be widely carried out in the population. For example, night blindness is cured by regularly banging the head of the patient, measles is cured by spitting lime or tamarind juice (jeruk nipis) to the eye, fever or convulsion are cured by dukun’s mantras and restrictions (pantangan) of food or actions to release evil spirits. For mothers, traditional obstetric and gynocologic practices include ruruyuk (cleaning of womb by drinking herbal ‘medication’ and applying its grounded residue (sometimes from cement) in the vagina as a sort of tampon), or nyandak (sitting on the wall for 40 days after delivery). Both ruruyuk and nyandak are regarded as complementary practices that contribute to cleansing of mother’s womb and restoring the ‘tightness’15 of her vagina. It is virtually impossible to determine the current spread or intensity of such practices because little quantitative epidemiological research has been done on the 14

see Sudarti, et al. (1988:10, 39); see also Sarwono (1993:128). Because of lack of knowledge of these medications, self-administration often brings serious complications and drug-resistance. The use of Jamu drinks from early childhood is perceived as generic medication and health treatment. But jamu may also induce abortion because it consist of several herbal ingredients including pineapple, chili, alcohol, and carbonated water. Jamu (in liquid or powder) are available widely, and TBAs often recommend its use (Sarwono, 1993:131). 15 Women , and presumably their sexual partners, carry out a variety of measures to ensure that sexual relations are 'dry' or 'tight', presumably to increase friction, and enhance pleasure. It is unclear whether such practices have side-effects which might increase susceptibility to vaginal infection.

44

customs, and in any case people tend to be very hesitant about reporting behavior which might classify them and 'backward' in the eyes of urban researchers. Traditional beliefs and practices pertaining to the low level of nutrition status include taboos against eating fish (may cause hookworm to children and may cause foul odor of lactating breast milk), taboo against consuming eggs (may cause pimples among children), taboos on certain vegetables or fruits (such as cucumbers, bananas, pineapples, all which are deemed to cause reduce women’s libido), rice, spinach, and beancurd (tahu) (for post-partum women for up to 40 days). Men and boys are given preferential treatment in consumption of food over women and girls. These food-related practices contribute to women’s and children’s low nutritional status due to protein and vitamin deficiencies (for various health status indicators affecting women and children, see Bappenas and UNICEF, 1994).

45

Table 3.3. On-going culinary beliefs and practices on maternal care in West Java, 1995 Cultural Habits

Perceived Purpose

Culinary Beliefs on Prenatal and Post Natal Care Pantangan Dahar (Eating Prohibition)

Other Pantangan

During pregnancy or post-delivery, women are not allowed to eat certain foods (pantangan dahar, lit. eating prohibition) by the dukun or paraji. These pantangan include food such as: Tjau (banana): will make the mother ‘quickly fertile’ (enggal subur) Ikan gurameh (cod fish) and goat meat: will cause hemorrhaging during delivery Udang (shrimp): will cause complications during delivery, namely breech position (sungsang) of the infant just as the arched form of the shrimp Madu (honey): will cause hard labor and pain during birth delivery (just as the sting of the bee, which produced the honey, is very painful) So leaf (melinjo muda): will cause mothers to often rest (‘ngaso’) during delivery instead of pushing hard for the baby to come out. Gula Jawa (solid brown sugar): will cause pre-delivery bleeding. N’dog (egg): will cause abscess (bisul) and scab (korengan). Nangka (jack-fruit): will cause difficulty during delivery because it is sticky and fibrous. Hahaseuman (lit. sour): are sour tasting foodstuff that are prohibited by dukun immediately after delivery. Instead the women should take papaitan, or bitter tasting food. Gorengan: are oil-cooked food prohibited by dukun after delivery. Instead mothers can only eat non-oil cooked food (nyangrai), grilled, or baked food. This prohibition may last from 40 days up to three months. These are action-based precautions which are laced with supernaturalistic and symbolically derived meanings such as: covering or closing an open hole, sitting in front of the door, or slaughtering a chicken (meuntjit hayam, or folding a towel around the neck are believed to cause difficulties during delivery, namely obstructed labor. Women are also not allowed to go out far after sundown, or during twilight (maghrib), or throw out garbage through the window, or go to the river, in order to avoid being disturbed by spirits (roh). But if they must do so, then they would be equipped with some sort of supernatural symbolic weapon such as a pin, scissors, and garlic.

46

Table 3.3. (continued) Practical Beliefs on post and antenatal Care

See also Table 6.1. Ruruyuk: is the practice of applying herbal medication into a woman’s vagina (as a sort of tampon) during and after 40 days of delivery. The purpose is cleanse or absorb impurities (kotoran) in the womb after delivery. Some also believe that such application would quicken the process of healing of the vaginal wound. This herbal formula include: chili leaf, sembung, ginger (jahe), kencur, jambu biji, lempayung, and kunyit. The extract of this formula can also be prescribed orally with the belief that the formula will regain the woman’s strength. Nyandak: is the practice of resting on the bed (or pole) with the head and the back on the wall up to 40 days after delivery, and sometimes up to two months. Its purpose is so that ‘impure blood’ (darah kotor) will quickly get out of the woman’s body while ‘white blood’ will not travel up and clog her respiratory system. While sitting in this position, the woman’s legs are also weighted down with rock or a wooden block so that she would not be able to move. It is perceived that the longer the period of nyandak the better its purpose is achieved (as indicated by her swollen legs). The practice of Nyandak is believed to be complementary to the practice of ruruyuk.

Note: These practices were noted as occurring in contemporary rural West Java, but it is impossible to determine the incidence and prevalence of observance, the depth of belief in presumed effects, or the degree to which observance would affect the health and well-bing of a woman. Source: based on transcripts and field notes conducted by CHR-UI staff in West Java, January-September 1995.

Laced with modern medical use are also strong beliefs in supernatural powers related to illness, or means of healing, which can be, or are deemed to be, affected by spirit beings (roh halus) or guardian spirits (roh penjaga, or tuyul). These supernatural beings are thought to be able to bring ill-health, disharmony, chaos, or cataclysm whenever members of a family, or community, disregard or violate their sacred domain.16 In order to safeguard against or to cure a spiritual-caused illness, dukuns provide amulets (jimat) and chant mantras (di jampi-jampi). In rural West Java (and even among urbanites), such beliefs and practices for protection against spirits are still extensive. 16

Eating wrong food, passing ‘evil places’ at the wrong time (e.g. during twilight hours, midnight hour, or during Friday Kliwon, or Thursday Night (malam jum’at) can bring about the wrath of spirits. Quoted in Sarwono (1993) from Raden Ajoe Sangkanningrat, 1927:10, 29, in ‘Iets over Hygiene in verband met adat, geloof, en bijgeloof van het Soendaneesche volk’ in Vereeniging tot Bervodering der Hygiene in Nederlandsch-Indie, Bandoeng: Publicatie no. 10.

47

Table 3.4 On-going cultural beliefs and practices of Dukun/ Paraji concerning maternal care in West Java, 1995 Cultural Habits

Perceived Purpose

Pre-natal community social contract with Dukun or Paraji

Nyangcangkeun (lit. to hook) is a sort of oral contract which socially binds those involved (the husband, wife, and the dukun) to entrust the role of pre- and post natal care to the dukun. The dukun is entrusted by the community as such that they are given the affectionate title of Emak Paraji (lit. mother paraji). The de facto contract is entered when pregnancy reaches seven months, but intensive lobbying between the mother and husband and the preferred paraji begins some months before.

Community deference to the dukun or paraji

Several factors contribute to the significant role of the dukun: - culturally context bound: dukun live in the same area and gained their local knowledge from the same community. - cultural capital: dukun is the nearest, most accessible (economically, socially, and physically) community health resource. Dukun with other ‘medicinal’ know-how (e.g. bone mending, herbal apothecary, etc.) are ascribed higher respect and social status than other dukun. - symbolic capital: dukun is perceived to be endowed with spiritual and supernatural powers as they grow in age. They are symbolically ascribed the role of the (adopted) ‘mother’(Emak). Ngopenan (lit. to care): is the perceived main role of the paraji, i.e. providing affectionate care.

Spirit related difficulties

Santapan is caused by spirits to obstruct the delivery of the placenta. Kasambet is caused by spirits entering the woman which is indicated by symptoms such as eclampsia, anxiety, biting tongue, loss of consciousness. Dido’akeun or disyare’at is a service (i.e. providing water which are enhanced by prayers) given by a dukun or a kyai (sometimes also referred to as orang pintar, or lit. clever man) to relieve pregnant victim of spiritual or supernatural disturbances such as santapan or kasambet.

48

Table 3.4 (continued) Time tales

Time for delivery is determined by several indicators: Kokotor: is sticky liquid substance preceding the amnion (air ketuban) Tjai Kawah or pameuntas jalan: is the term of the excretion of amnion after its membrane rupture. Usually the paraji would give a sort of medication in the form of grated garlic and/or with palm oil applied on the surface of the mother’s stomach or sometimes around the vagina with the aim that the baby may come out easily and quickly. Jukut Fatimah: is a sort of grass, presumably obtained from the holy land and of Mekkah. It is placed in water and waited until it blooms, which indicate the time for birth is near and help must be given immediately. Po’e Weton: is the time of birth based on the birthday of the mother. If the expected weton day is not appropriate, then it is assumed the birth delivery support cannot yet be provided.

Source: based on transcripts and field notes conducted by CHR-UI staff in West Java, January-September 1995.

Because the belief in the supernatural is so enculturated among people, dukuns (and sometimes elders, or kokolot) are held in special regard and esteem, and become first choice as a community’s health resource17 simply because dukuns do provide spiritual (or ‘cosmic’) support and supernatural healing not guaranteed by modern diagnosis and medication. Faith in the efficacy and efficiency of traditional-cum-spiritual healers will continue to remain strong as long as people retain supernatural beliefs which are not accomodated by modern or other alternative forms of healing. In most maternal cases covered by this study, people accessed the nearest and most familiar of health institutions, namely the traditional birth attendants (paraji, or dukun beranak, or ‘mak dukun). Data indicate that in 1991, use of TBAs in West Java reached more than 80 percent, while in 1992 that figure was reported as 74 per cent of maternal deliveries conducted by TBAs.18 The task of dukun beranak or paraji19 is 17

People still seek the advise of dukun as the last resort (or to obtain a ‘second’ opinion) for medical treatment, even after seeking modern (and in many cases along with modern treatment) medical support. 18 Based 1992 National Socio Economic Survey (CBS, 1993b: 63). Comparatively, in 1992, access for deliveries by doctors accounted for only 4.3 per cent, for midwives/nurses 22 per cent, and ‘others’ (i.e. members of the family or known others)amounted to only 0.2 per cent. Figures for doctors and nurses/midwives, have shown steady increases of access to them, but more likely in sub-urban and urban areas. 19 In Java, paraji is usually a woman, and in Bali it is usually a man; thus, it is not gender-specific. They can be women, or men, usually above 40 years old, who were ascribed to the role from their parents or close elder relatives.

49

primarily to help pregnant women with ante-natal care, with the delivery of a baby, and with post-partum care during the first 40 days after the birth. Their services may include regular visits, prayers, and massages. Basically, paraji or dukuns20 provide comprehensive and familiar care for the mothers during the pre-, and post-natal periods, and as such they will be likely to remain in the cultural context of West Java for many more years. It must be remembered that most births attended by dukun are successful, and even when difficulties arise, most mothers and babies survive. As such the community perceptions of the 'average likelihood of success' is quite different from the public health perception of elevated risks of alternative procedures, which operates at the margin of common experience. Table 3.5 Number of registered nurses, dukun/TBAs, midwives, and village kaders, in West Java, 1993 Location

Nurses

Dukun/ TBAs

Percentage Midwives of TBAs Trained

Village kaders

1,290 2,327 1,606 715 626

69.0 60.0 69.9 87.4 86.1

137 171 197 194 82

3,573 10,176 10,242 6,282 4,729

Research Areas Lebak Sukabumi Ciamis Kuningan Purwakarta

23 10 34 26 14

Selected Urban Sites Bogor Sukabumi Bandung

400 16 486

95 76 158

100.0 100.0 89.9

55 38 458

1,906 1,368 13,922

West Java

1,901

26,439

75.3

4,807

17,7983

Source: Kanwil DepKes Jabar (1994). Collated from Tables 28 and 30E.

In return for their services, people provide paraji with loyalty, respect and in-kind payments or rewards (money, clothes, agricultural produce, etc.), which in many cases are negotiable. Beside being familiar and providing individual comprehensive services, paraji’s services are also immediately accessible and affordable. Also, because of their 20

There are several types of dukun acting as community health resource: dukun sunat who performs circumcisions on boys, and dukun patah tulang who specializes in mending or stetting broken bones. Others include the tabib, or (Indian-trained) medicine man, or sinshe (Chinese-trained) medicine man. There is also the more general dukun who provide shamanistic services.

50

age, many are also regarded to hold the all important ‘cosmic’ powers (by virtue of their knowledge of the do’a, or prayers, and knowledge of the all mysterious spirits and supernatural influences on the body), and thus command high respect from the community (Sarwono, 1993: 118,130). But many of these dukun or paraji are untrained and practice unhygienic, and often harmful behavior to women in labor. The West Java MOH data indicate that in 1993 more than 25 per cent of them are still untrained.21 Nevertheless, because of people’s familiarity with dukuns or paraji, they are very much trusting toward dukun, despite the high risk of death in a delivery assisted by these people. In many cases, people believed that death during childbirth is pure (mati sahid) for both the mother and the baby.22 As such, when deaths do occur, people are often resigned to accept the fate (nasib) without question. The paraji or dukun, because of their local context, can only serve a very limited number of people within a limited area coverage. But their numbers are also significantly higher than nurses or village nurses. In 1993 there were 26,439 registered dukun compared to a total of 9,315 health workers (including nurses, nonhealth workers, midwives, mental-health workers) (Kanwil DepKes Jabar, 1994: Table 30B).23 While one TBA serves about 600 to 700 women, one nurse serves 6000 and one midwife serves 7000 women in their work area (Dekker 1989:15, cited in Sarwono, 1993:130).

3.6 The context of contrasts in West Java It is important to recognize the context of contrasts existing in the community level health care and health practices in West Java. On the one hand there are various issues of aggregate (‘community’) programs and implementation of service deliveries which, despite the large number of projects, continue to be ineffective in reducing infant, child, and maternal death, while on the other hand there are various individualistic and context-specific socio-cultural aspects (beliefs and practices) which continued and contribute to impede those community health improvement efforts. The characteristic of West Java, with all its contrasting socio-cultural dichotomies, must be considered as relevant if the hope for improvement in health service delivery is to be affected. These dichotomies include aspects of: urban-rural disparities (economic, access, education); gender dominance and patriachalism (or bapakism); modern-traditional normative values, life-styles, and behavior; vital trends and disparities (marriage, fertility, mortality, morbidity); the reproduction of priyayi attitudes and patron-client relationships in obtaining, providing and maintaining services; the potential for community participation (by ways of musyawarah, or collective decision21

‘Trained’ should be seen with qualification. Often times it is a one-off, rather than continuous, learning and supervision process. In some cases they have not been trained for several years and have reverted to their own traditional practices. 22 Mati sahid, or mati suri, has equal standing with death as a martyr in a holy war (Sarwono, 1993:130). 23 This also indicates a low service ratio of only 25,2 per 100,000 person in the population.

51

making, and gotong-royong, or collective work, and emphasis for keseimbangan, or natural balance) which are yet to be tapped; and local eclectic practices of religion (e.g. Islam laced with mysticism). From the health delivery system, there are also disparities in terms of: the number of programs to be affected and the availability of human resources to achieve those programs. There are also many other issues to be resolved or improved upon such as: in the still low quality of its health services and care of patient; in the low extent and coverage of service deliveries; in the capacity and enthusiasm of the health workers; in the practice of hygiene; in the continuity of supervision; and in the availability of materials for their effective work. These contrasts, which have been evident in the past, and as we will further explore in the following chapters, have shed some understanding into some of the reasons why the issues of maternal mortality in West Java have not progressed as it should have been. We believe that it is important to understand the fundamental world-views and practices of the individual people involved in maternal mortality issues if intervention changes in health delivery system hope to be more effective.

52

Chapter 4. Fieldwork: the search for cases and the questioning of witnesses 4.1 Definition and number of cases to be included Based on the moderate estimate of demographic statistics for West Java: 10,000 households per subdistrict, 5 household members per household, crude birth rate of 25 per 1,000 persons per year, and maternal mortality ratio of 400 per 100,000 live births per year, there would be, on the average, 1,250 live births and five maternal deaths per year per subdistrict or two to three maternal deaths per six months per subdistrict. Due to the relative rarity and the wide geographical distribution of maternal deaths, only 50 cases of maternal death were going to be included in the study. For comparative purposes, 10 obstetric emergency cases which ended up with the survival of the mother and child were also going to be included. It was thought that the limitation of cases to one year in the past be sufficient to overcome the effect of recall memory-lapse errors among the witnesses. After a year, non-family members would be expected to forget many of the details of the case, and even family members might begin to recount the story in a much attenuated fashion. Cases that occurred at the most recent times received priority to be included in the study. Administratively, West Java consists of 20 regencies and 4 municipalities. For practical reasons, the study purposively selected five regencies thought to represent some of the major features of geographic variation of the West Java Province. These selected five regencies were Lebak in the south-west, Sukabumi in the middle-south, Ciamis in the south-east, Purwakarta in the middle-east, and Kuningan in the north-east of the province. For each selected regency, it was planned to study 10 cases of maternal death and 2 cases of obstetric emergency survivors. Thus, there would be 50 cases of maternal deaths and 10 cases of obstetric emergency survivors to be included in the study. 4.2 Research personnel: forming the teams and specifying the jobs Research personnel consisted of principal investigators and fieldwork personnel. The research personnel worked in a team where each member has specified, but complementary job descriptions. The fieldwork team included a field manager, six interviewers, four transcribers, one scout, and one driver. The duties and job descriptions of the various research personnel were as follows: Field Manager: Coordinate the activities of all the field workers. Manage the movement and maintenance of all field equipment. Liaise with local authorities. Liaise

53

with the Principal Investigators and the central office. Prepare the case summaries and field reports. Scouts: Find respondents: Identify potential cases. Make lists of potential witnesses, determine the availability of witnesses, make appointments to interview. Carry out any photo-documentation and copying of secondary statistical data. Interviewer: Interview using a tape recorder and notebook. Check the draft transcript for completeness and accuracy and return to seek clarification of any serious inconsistencies or ambiguities of wording. Prepare the witness summaries based on the transcript and the research guideline. Transcriber/Translator: Transcribe directly onto computer each interview precisely in the language of the interview. Assist in the translation of Bahasa Sunda and Jawa transcripts. Principal Investigators: Analyze and summarize the case reports. Draw up external analyses of the cases and summarize the study results. Draw up policy analysis and recommendations. Prepare preliminary and final reports to the funding agency, and write up results for publication. The principal investigators and the field manager were the research staff of the Center for Health Research University of Indonesia. The other fieldwork personnel, including the scout, interviewers, and transcribers were recruited by the project through several steps of selection. The scout selected was a senior high school graduate who showed the following criteria: fluent in Sundanese, familiar with the geography of West Java, able to communicate with people, enjoys working in the field, and committed to the goals and methods of the project. The interviewers were selected from 80 applicants, all of whom were university graduates, but only 15 fulfilled the selection criteria which included: being currently single, trained in social sciences, experienced in fieldwork, enjoys living in the rural areas, speaks, or at least understands, Sundanese. These 15 selected persons were trained for five days in the study related concepts and technical details of field work activities. Evaluations were conducted to select the best seven training participants to be involved in the pre-test interviewing the witnesses of three cases of maternal deaths in Sukabumi Regency. Finally, six trained persons were recruited as interviewers; while one trained person was dropped due to low motivation and commitment. Four transcribers were selected from nine applicants who fulfilled the criteria: has at least senior high school level of education, enjoys living in rural areas, and familiar with word processing. The selection of transcribers was based on accuracy, tidiness, typing skill, and self confidence shown during test of transcription.

54

4.3 Searching for maternal deaths and survivors of obstetric emergency cases With the help of the Family Health Directorate of the Department of Health, contact was made with the Regency Health Office in five selected regencies to request their assistance in listing maternal deaths and survivors of obstetric emergency cases that occurred during the previous 12 months. The listing format and procedures were guided by the research team. The format included the name of the victim, the name of the household head, the date of event, the complete address, and the names of persons to be contacted. Following to this request, the Regency Health Office instructed the MCH section to organize the task down through subdistrict health centers, then from the subdistrict health center to the village levels utilizing village birth attendant networks, through routine monthly meetings at district and subdistrict levels. In most cases, the networks of village birth attendants have already existed under the coordination of the subdistrict health center, but need to be activated to find cases of maternal deaths, such as in this study, otherwise they are moribund. Operationally, searching for cases of maternal deaths was carried out at the village levels by birth attendants, most of whom were village midwives and traditional birth attendants, under the coordination and supervision of village-health officer24, who reported the findings to the subdistrict health center, and the subdistrict health centers reported to the MCH section of the Regency Health Office. Searching for cases of maternal deaths was also conducted at hospitals and maternity hospitals, and the findings were sent directly to the Regency Health Office. The process from the first contact to the Regency Health Office to obtain the list of maternal deaths took about two months. Some incentives in the form of transportation fee were provided to those who actively involved in case searching activities.

4.4 Field operations Field operations consisted of both sequential and simultaneous steps of field activities. The field activities were running for about seven months from March to September 1995. The field operations began with searching for location of cases done by the Scout and the Field Manager. Based on the list of maternal deaths and obstetric emergency cases received from the Regency Health Office, searching for location of cases was started and conducted sequentially from Lebak, Sukabumi, Ciamis, Kuningan, to Purwarkarta, but, at this stage, limited to five cases per subdistrict. Cases of the most recent times were given priority to be located. These searching activities consisted of checking and mapping, to check whether the list included only maternal deaths and obstetric emergency cases, whether the date of events was correct, and whether the address was correct, and to map the confirmed cases. The map was prepared as to visualize the location of cases so that interviewers, by using the map, could easily find the 24

Village health officer (PKD or Petugas Koordinator Desa) is the subdistrict health center’s employee who has the responsibility to coordinate and supervise health program activities at two to four villages depending on the number of villages per subdistrict.

55

cases. In many instances, one case required one map, but, sometimes, two or three adjacent cases required only one map. For each case located, a prime contact person, usually the victim’s adult household member, was identified as the first witness. The next witnesses were determined through snowballing mechanisms. After this stage, data collection activities started, but the Scout continued searching for location of additional cases. Cases not reported in the listing, but found in the field to occur before six months were also included in the study. A map visualizing the geographical distribution of confirmed cases was also prepared in each of the five selected regencies. At the time of locating and mapping maternal death cases, searching for cases of obstetric emergency survivors was also carried out by the scout and the field manager by asking the health center personnel, particularly the midwives, whether they are aware of the recent occurrence of obstetric emergency survivors in their working areas. Hence, the obstetric emergency survivor was defined as a woman who survived from a pregnancy related complication, such as hemorrhage, infections, or pre-eclampsia. Data collection activities were conducted sequentially from Lebak, Sukabumi, Ciamis, Kuningan, to Purwakarta. For managing data collection activities, two to four temporary parallel, or sometime consecutive, field offices were established in each of the five selected regencies depending on the geographical distribution of cases. The distance from one temporary field office to the location of cases and witnesses, at the most, ranged from 10 to 30 kilometers. The temporary field offices were staffed by Interviewers and Transcribers (Notulis) and equipped by computer note books, tape and cassette recorders, and stationery. The field manager stayed in the field and monitored and controlled the fieldwork activities. For facilitating easy mobilization of temporary offices and field personnel, one van-sized car, including the driver, was provided in the field. In many cases, the location of cases cannot be reached by car. Thus, the car was only used to drop the field personnel in a field meeting-spot thought to be near to the location of cases. From the field meeting-spot, the interviewer walked or used a motor cycle to reach the respective witnesses. For each case located, several witnesses were interviewed in-depth by the interviewer. As far as possible, the flow and contents of the interview followed the simple interview guideline as mentioned in the description of the Rashomon Technique. The interviews began with the first witness identified by the Scout, and then followed sequentially with other witnesses, who were identified from the previous interview. Number of witnesses interviewed for each case located ranged from four to eight persons, normally included husband, adult household member, parents, traditional birth attendant, midwife, physician, other relative, and neighbor. As one witness determined the next witnesses, the witnesses were interviewed sequentially. Every day during fieldwork, four to six interviewers worked independently in the field to interview the witnesses. The length of interview per one witness ranged from one to three hours. The following information were collected from various witnesses based on their knowledge, perception and experiences: socio-economic and demographic characteristics, history of general health and illnesses, history of obstetric illnesses and health care for the last and previous

56

pregnancies, travel history for seeking health care facilities, and history of treatments in health care facilities. In some cases, some predetermined witnesses were dropped as they were not appropriate. In other cases, some witnesses were interviewed more than one time to assure that main issues were covered in the interview. Compared with other types of witnesses, traditional birth attendants and physicians were generally more difficult to be interviewed as they were afraid to be blamed. All interviews were recorded by a tape recorder, but points need to be explored were also documented in the pocket diary notebook that always brought by the interviewer. Each day the transcribers were responsible for typing a complete and accurate transcript of the previous day's interviews, and any notes or reports compiled by the field personnel. Transcribers were responsible for backing up and preserving all computer files produced in the course of the study. The interviewers and sometime also the field manager checked all the work of the transcribers to ensure completeness, accuracy, correct spelling and punctuation. The interviewers also annotated all Indonesian language transcripts to clarify any Sundanese words or phrases, and made full Indonesian translations of all interviews conducted in Sundanese. For each case located, seven to ten days were required to finish with interviewing, transcriptions, witness’s interview summaries, and background documentation. The fieldwork phase thus involved the simultaneous implementation of data collection and data processing and analysis.

57

Chapter 5. Results: background characteristics of cases and witnesses 5.1 Number of maternal deaths As the record of vital events in developing countries is seriously deficient, accurate data on mortality are very difficult to obtain. In Indonesia, vital registration existed as early as 1911, but only selected areas of Java and Madura were included and about 30 to 50 per cent of deaths in the covered areas were missed (Utomo, 1982: 297). Moreover, statistics of registered deaths have not been published regularly. Until today, there is no national vital registration that would yield complete and accurate data on births and deaths. Consequently, mortality can only be estimated from population censuses or surveys, often through indirect methods. Data on number of birth deliveries and maternal deaths in the five selected regencies are available only at hospital levels, but for 1993 at the latest. Even if the hospital data are available for the past 12 months, they are obviously bias since the majority of birth deliveries in West Java occurred at home. The 1993 hospital statistics on maternal mortality ratio vary between the five selected regencies;. while no maternal death was reported in Sukabumi, Ciamis, and Purwakarta, 3 maternal deaths per 851 birth deliveries and 4 maternal deaths per 684 birth deliveries were reported in Lebak and Kuningan respectively (see Table 5.1). In Lebak, for example, the yearly hospital maternal mortality ratios for 1989 to 1992 vary greatly, from 970 to 2806 maternal deaths per 100,000 birth deliveries, but for 1993 fell to 353 maternal deaths per 100,000 birth deliveries (Dinas Kesehatan Kabupaten Lebak, 1994: Table 12). This large variation of hospital maternal mortality ratios from year to year and from regency to regency may in part reflect the data recording inconsistencies. Because of bias, small number, not-recent timing, and possibly inaccuracy of the number of maternal deaths reported, hospital statistics cannot be used for the entry to find cases of maternal deaths that occurred in the population. Consequently, an alternative mean to find cases of maternal deaths in the population need to be developed. The study demonstrated that the existing networks of village birth attendants, which involve the health center personnel, village midwives and traditional birth attendants, could be activated and used to search for the occurrence of maternal deaths in the population, at least in West Java. With the help of the Regency Health Office, case searching through hospitals and networks of village birth attendants in the five selected regencies could find 186 cases of maternal deaths that occurred during the past 12 months. It is interesting to note that hospitals in Ciamis and Purwakarta, which reported no maternal death for 1993, also reported no maternal deaths in the past 12 months since April 1995 (see Table 5.1). 58

Table 5.1. Reported number birth deliveries and maternal deaths in hospitals during 1993, and number of maternal deaths in the previous 12 months (from April 1995) reported through case searching and fieldwork.

Hospital statistics

a

Case searching

Regency Birth deliveries

Deaths

Hospital

Lebak Sukabumi Ciamis Kuningan Purwakarta

851 1053 670 684 772

3 0 0 4 0

Total

4030

7

b

Additional found

cases

Total

Health c center

Not yet d reported

New e cases

11 4 0 2 0

14 61 42 21 31

10 7 19 5 0

2 1 3 2 2

37 73 64 30 33

17

169

41

10

237

f

a

Reported number of birth deliveries and number of maternal deaths at hospitals during 1993 (Dinas Kesehatan Kabupaten Lebak, 1994: Table 12; Dinas Kesehatan Kabupaten Sukabumi, 1994: Table 12; Dinas Kesehatan Kabupaten Ciamis, 1994: Table 12; Dinas Kesehatan Kabupaten Kuningan, 1994: Table 12; Dinas Kesehatan Kabupaten Purwakarta, 1994: Table 12). b Case searching during the study: number of maternal deaths reported through hospitals. c Case searching during the study: number of maternal deaths reported through subdistrict health centers; these subdistricts’ and the hospitals’ reported numbers were made to be mutually exclusive. d Additional cases found incidentally by the field research personnel during field operation: not yet reported through hospitals or subdistrict health centers. e Cases that occurred after the recall specified period of case searching. f Total number of maternal deaths obtained during case searching and fieldwork.

Given that not all subdistrict health centers sent the list of maternal deaths and not all village midwives involved actively in case searching, the real number of maternal deaths should be more than that of reported number of 186 cases. In fact, more additional cases of maternal deaths, which mostly also occurred in the same period as specified by the case searching procedure, were incidentally found by the field manager and the scout during mapping of maternal death cases. Number of additional cases of maternal deaths is found more in Ciamis and Lebak than in other three regencies. In fact, unlike in Ciamis and Lebak, no additional efforts were made in other three selected regencies to find additional cases of maternal death. Thus, such a difference reflects the variation in the degree of efforts to search for additional or new cases of maternal during mapping activities. In fact, the field procedure employed in the study found 237 cases of maternal deaths distributed in the five selected regencies. As the purpose of the study was not to estimate maternal mortality ratio, but rather to find 50 to 60 cases of maternal deaths, no intention on the part of the research personnel to list comprehensively all maternal deaths that occurred in the study areas.

59

It is noteworthy to exercise with a question on how many per cent of maternal deaths were not found by the case searching procedure employed by the study. For this exercise, expected number of births, and then expected number of maternal deaths were calculated for the 35 subdistricts included in the study on the basis of the reported number of population, the crude birth rate of 25 live births per 1,000 people, and the maternal mortality ratio of 400 maternal deaths per 100,000 live births. These expected number of maternal deaths were then compared with the number of maternal deaths found from the case searching procedure. Assuming that the expected number of maternal deaths is the ‘true’ number of maternal deaths, the case searching procedure missed about 30 per cent of maternal deaths. The percentage of missed maternal mortality cases accounted for only 9 per cent in Ciamis, but it accounted for 57 per cent in Purwakarta. As previously indicated, this large variation might be due to length of time and efforts exerted to search for maternal mortality cases.

Table 5.2 Expected and reported number of maternal deaths in the study areas b

Regency

Number of a subdistricts

Population

Expected c births

Expected d deaths

Reported e deaths

Percentage of f difference (%)

Lebak Sukabumi Ciamis Kuningan Purwakarta

7 9 9 5 5

342945 560540 459690 304548 394811

8574 14014 11492 7614 9870

34 56 46 30 39

25 43 42 17 17

-27.1 -23.3 -8.6 -44.2 -56.9

Total

35

2062534

51563

206

144

-30.2

a Include b

only subdistricts where cases were included in the study. Based on the 1994 official statistics of the Regency Health Office (Dinas Kesehatan Kabupaten Lebak, 1994; Dinas Kesehatan Kabupaten Sukabumi, 1994; Dinas Kesehatan Kabupaten Ciamis, 1994; Dinas Kesehatan Kabupaten Kuningan, 1994; Dinas Kesehatan Kabupaten Purwakarta, 1994). c Expected number of births per year based on crude birth rate of 25 live births per 1,000 population; the total subjects to rounding errors. d Expected number of maternal deaths based on maternal mortality ratio of 400 per 100,000 live births; the total subjects to rounding errors. e Reported number of maternal deaths based on case searching and fieldwork. f Percentage of difference between expected number of death and the study’s reported number of maternal death: computed by the formula ((expected number-reported number)/expected number) x 100 per cent.

60

5.2 Characteristics of cases and witnesses Of 237 cases of maternal deaths found through the case searching, 53 of the most recent cases distributed over the five regencies were included in the study. For comparison, ten recent cases of obstetric emergency survivors distributed over the five regencies were also included. These 63 cases were widely scattered covering 35 subdistricts; 7 subdistricts in Lebak, 9 subdistricts each in Sukabumi and in Ciamis, and 5 subdistricts each in Kuningan and in Purwakarta (see Tables 5.2). The number of maternal deaths and survivors of obstetric emergencies studied in each selected regency ranged from 5 to 16 cases and 1 to 4 cases respectively (see Table 5.3). Number of witnesses interviewed in each selected regency ranged from 34 to 129 persons, suggesting that, on the average, 6 witnesses per one case of maternal death or one case of obstetric emergency survivor. These witnesses included husband, mother or mother in-law, father or father in-law, relative, neighbor, traditional birth attendant, midwife or paramedic, physician, and, in the case of survivor, the survivor herself. Almost all cases included the traditional birth attendant and the midwife or the paramedic as the witnesses. All witnesses were adults; most of them aged above 20 years. The witnesses with age above 40 years generally included mother or mother-in law, father or father in-law, and traditional birth attendant. As expected, the majority of the witnesses were females and low educated. The witnesses with low educational level included the victim’s household family members, the neighbors, and the traditional birth attendants. The majority of maternal death and life-threatening events included in the study occurred in the past 6 months of the interview date. Only 30 per cent of the events occurred in the past 6 to 12 months of the interview date (see Table 5.4). Age of the victims covered the range of reproductive age, but the majority of them aged from 20 to 39 years. This is understandable since most of the pregnancies occur in those ages. The victims’ level of education is generally low; most of them had elementary school, and only 25 per cent had junior school education or above. More than half of the victims were reported as non-working women, just worked at home as the housewife. Other victims were reported as working women: laborer, farmer, or small trader. More than half of maternal death and survivor of obstetric emergency cases had 4 or more pregnancies and live births. About 22 per cent of the cases had loss of one or more children.

61

Table 5.3 Number of maternal deaths, survivors, and witnesses and demographic characteristics of the witnesses by regency

Regency

Deaths Survivors Witnesses Type of witnesses Husband Mother/in-law Father/in-law Relative Neighbor TBA Midwife/paramedic Physician The survivor

Age of witnesses (years)