Confidential inquiries into maternal deaths: Modifications and adaptations in Ghana and Indonesia

2.1. Variations in confidential inquiries 

Two main variations in the confidential inquiries were implemented in accordance with the country context. In Ghana, most births are attended by health professionals in facilities; therefore, clinical case notes were used and events within the health facility were the focus. The confidential inquiry was modified to encourage the clinical panel to identify both adverse and favorable factors associated with each case.

In Indonesia, the traditional confidential inquiry process was adapted more extensively. The assessment was of community factors, not facility care, as a different approach was required to suit the context. Because village-based midwives do not usually record clinical events, care provided in the community was reviewed through the use of interview transcripts rather than clinical case notes. The investigation of events in the community surrounding death was extended to include cases of severe maternal complications, and both adverse and favorable factors were included in the assessment.

2.2. Study population, sample, and data sources 

Within the 6 months available for the study, we estimated that a maximum of 20 cases could be reviewed. The series of cases in each country was selected to represent a range of the common types of complication. In Ghana, the selection was taken from across the time period of the study, from a range of direct and indirect obstetric complications and representing different facilities. In Indonesia, cases were selected from the most common obstetric complications where a midwife was in attendance. Cases occurring before 28 weeks of gestation were excluded because the focus was on care provided during labor and delivery.

Both studies were linked with other research. In Ghana, the linked study identified 150 maternal deaths in health facilities between 2003 and 2005 from the admissions and delivery registers of 10 government and mission district and regional hospitals. The details of these deaths were assessed by 2 clinically trained researchers to fit the selection criteria described above, thus identifying 20 cases for panel review. The full set of case notes for this sample of 20 was found in the medical records office, as well as additional records such as outpatient cards, nursing notes, laboratory forms, partographs, and charts for monitoring vital signs.

In Indonesia, the linked study was conducted from 2004 to 2005 in a rural district of 1 million people in Banten, West Java, with 334 villages, 30 health centers, and 1 district hospital. Cases of death or severe obstetric complications from the study area in which the patient was admitted to the district hospital were identified by a physician by reviewing records and holding discussions with healthcare providers where necessary. After they were discharged, and having given their consent, the identified (surviving) women were accompanied to their homes by a research team member to ascertain the address for the follow-on visit for interview. Interviews in 13 cases of women receiving direct care from the village midwife were completed within the study period. In-depth interviews were conducted with people involved in each case, including the village midwife, health center personnel, traditional-birth attendants, the pregnant woman herself (for cases of complications without mortality), family members, and lay members of the community such as the village head and the driver of the vehicle used for transportation. Interview guides were used to explore the barriers to and the facilitating factors for accessing effective care and referral. In addition to the interview guide, a list of clinically relevant prompts such as preparatory actions for labor, how the midwife was notified of complication, and how the delivery was managed was used during the interviews. The interview transcripts were used to review the cases.

2.3. Review panels 

Review panels were set up in Ghana and Indonesia. In Ghana, 8 members were selected from maternity service providers with obstetrics, midwifery, anesthesiology, pathology, and public health skills. In Indonesia, we set up 2 separate 7-member panels: a specialist panel of doctors based primarily in urban secondary and tertiary referral centers; and a community practitioner panel comprising midwives posted in villages, and doctors and midwives based in community primary-level facilities. Locally based researchers felt that a mixed panel could lead to limited participation by some individuals, and it was expected that the focus of analysis of the 2 panels would differ in accordance with their backgrounds. Differences in approach to inquiry were observed between the 2 groups. Both panels reviewed the same cases using the same process.

In both countries, panel members reviewed the documentation, synthesized information, completed assessment forms for each case, and met regularly over 6 months to review cases as a group. All panel members worked outside the study areas and were not familiar with the cases discussed. Copies of the case notes or interview transcripts were distributed to the panel members 2 weeks before the meetings.

A specially designed case assessment form with guidelines was used by the review panels for all cases. The form was based on the South African confidential inquiry [5], and the variations introduced and the accompanying methodology were developed into a tool called TRACE [11]. Even in the case of an adverse outcome such as a maternal death, a basic premise of TRACE is that it is important to identify positive aspects of care, which can help to build on strengths within the health system and prevent disillusionment. The form prompts panel members to use case notes or interview transcripts to assess systematically the factors related to the standard of health care provided, and the availability and quality of information in the documentation provided by the study. To foster ownership and engagement in the process, the case assessment form was finalized only after input from the review panels at a special meeting before the start of the confidential inquiries.

All features that might locate or identify a particular case were removed from the case notes and interview transcripts. Approval from ethical committees in each country was obtained for the study. Following up cases in the community, especially when a maternal or neonatal death was experienced, was approached in a sensitive way, and informed consent was obtained from district authorities and community members.

At least 2 researchers, 1 with a clinical background, were present for all panel meetings to facilitate the process and clarify uncertainties. Observations of the researchers and ongoing feedback from the panel members were recorded at each panel meeting. Additional meetings were held with the panel to elicit feedback on the findings, method, and process.

Data were analyzed by identifying themes relating to the care provided across the cases, using the structure of the case assessment forms. In addition, we reviewed the feedback from panel members and the observations of researchers during the panel meetings, and organized the information into topic areas relevant to the study hypothesis that confidential inquiries are not widely conducted because of their demotivating effect, the unavailability of sufficient recorded information, and a lack of time and resources.

3.1. Inclusion of favorable factors 

It was possible to identify negative and positive factors in all cases. In Ghana, for example, midwifery care was found to be adequate in several cases and superior to the care provided by doctors [12]. Referral and availability of emergency care facilities were adequate and did not contribute to adverse events. In Indonesia, the village midwives were effective in terms of mobilizing referral, but were less capable in the clinical management of complications [13]. Panel members identified adverse factors easily; however, when attempting to identify favorable factors in the first few cases, they needed prompting and reminding. Later, a balance between ascertaining favorable factors and ascertaining adverse factors was reached. The discussion and critique were equally thorough for both types of factor. In the feedback sessions, panel members indicated their preference for the inclusion of favorable factors, which they believed made the inquiry more balanced, objective, and comprehensive. They felt that the learning benefits from each case were enhanced by identifying positive events (alongside the deficiencies), thus showing that some appropriate actions were taken by health professionals in every case.

3.2. Quality of clinical case notes 

The clinical case notes available in Ghana were mostly from district hospitals, with only 2 sets of notes from regional hospitals. The panel noted that the availability of information in regional hospital notes was generally more comprehensive, especially for results of laboratory investigations and anesthesia notes. However, the comprehensiveness of district hospital notes was found to be sufficient for the panel to make a clear decision on cause of death and to identify specific positive and negative factors associated with each case. Information regarding admission, condition of patient, time intervals, and routine care procedures was generally available in the notes. Information about anesthesia, surgery, and postpartum care was often inadequate.

3.3. Use of interview transcripts 

In Indonesia, the specific adaptation of using interview transcripts was necessitated by the lack of written records of cases seen by village midwives. Between 2 and 6 interviews were conducted for each case. The panels found sufficient information in the interview transcripts to complete an assessment form in all cases, although the desire for more details, especially medically relevant information, was expressed. For example, there were 2 cases in the community in which the midwives “failed” to remove retained placentas. The panels felt that a better judgment of the midwives' skills could have been made if the subsequent clinical events or the information from the hospital about the cause of retained placenta had been made available. The amount of time required to complete the interviews and review the transcripts was considerable, and was one reason why only 13 cases of the planned 20 were reviewed. Panel members in Indonesia found that reading through several interview transcripts for each case was time consuming and constituted a heavy workload. Although the researchers offered to summarize the interviews, panel members decided that they preferred to read the transcripts themselves.

3.4. Differences between specialist and community practitioner panels in Indonesia 

The overall conclusions of the 2 panels regarding the care provided by midwives in Indonesia were similar. However, the processes that took place in the 2 panels differed. The community practitioner panel spent less time debating each case and challenged colleagues less, thus going into less depth in analyzing the factors relating to each case. This panel referred more frequently to national standards and guidelines when assessing the cases. The specialist panel had long debates on controversial areas and referred to their own clinical experience extensively. The community panel spent much more time discussing community-related factors such as referral problems and decision making, whereas the specialist panel tended to focus on medical care, despite sections in the assessment form prompting full assessment of both community and medical factors. The specialist panel also expressed astonishment at the complexity of the community factors and the problems faced by the village midwife.

3.5. Resources required 

Reviewing cases individually was reported to take 30–60 minutes per case. Panel members in Indonesia reported a longer time for individual review of cases, which was attributed to the many transcripts that they had to read for each case. With an average of 2.4 hours spent discussing each case (Table 1), the total estimated time required to review each case was 3.0–3.5 hours (meetings plus individual review time). The perception of all panel members was that the time was well spent, and crucial for improving personal and nationwide clinical practice.

Table 1. Time costs of conducting panel meetings.

		Cases reviewed	Meetings	Total meeting time, h	Average time spent discussing each case, h
	Ghana	20	8 h per month for 6 months	48	2.4
	Indonesia specialist panel	13	2 h per week for 5 months	40	3.1
	Indonesia community practitioner panel	13	3.5 h per month for 6 months	21	1.6
	Total				2.4

The commitment and interests of all panel members were clear; an average of 80% attendance was recorded for panel meetings, despite the heavy workloads of members. A contingency plan to send a named alternate if the main representative could not attend was hardly used. Panel members overstayed scheduled times of panel meetings to complete discussions, and made arrangements for panel meetings to be held regularly during personal leisure time so that interruptions were minimized.

The total costs of conducting the inquiries ranged from US $4000 to US $11000 (Table 2). These costs included the meeting venue, refreshments, reimbursement of time (for individual reviews and for panel meetings), and transportation. In Ghana, the cost per case reviewed per panel member was US $68. Reimbursement of time was 74% of total costs. In Indonesia, the total cost per case reviewed per panel member was US $99 in the specialist panel and US $46 in the community practitioner panel (Table 2). No data were available for differentiating between costs of time reimbursement and other costs, although such costs were expected to be small because the meeting room was free and all panel members were based relatively close to the meeting venue.

Table 2. Monetary costs of conducting confidential inquiries: Ghana and Indonesia.

		US $	Cases reviewed	Panel members	Cost per case per panel member, US $
	Ghana
	Expenses	2827	20	8	18
	Reimbursement of panel members' time	8040	20	8	50
	Total costs	10867	20	8	68
	Indonesia specialist panel
	Total costs: expenses and reimbursement of panel members' time	9000	13	7	99
	Indonesia community practitioner panel
	Total costs: expenses and reimbursement of panel members' time	4197	13	7	46

The above estimates include only the costs of the panel. Other costs included the time spent by the research team selecting cases, and logistical arrangements such as identifying a venue, photocopying case notes and interview transcripts, and co-ordinating suitable meeting times. In Indonesia, 3–5 2-hour interviews per case were undertaken by 1 qualitative researcher, 1 physician, and 1 research assistant, in addition to time spent traveling to villages.

3.6. Dissemination 

Dissemination of the synthesized findings of the review panels on the state of maternity care and the experience of conducting the adapted confidential inquiries was conducted in Ghana and Indonesia. A summary of the panel reviews was produced for distribution and was presented at research and government meetings, with recommendations for improvement in maternity care and identification of strengths within the health system. Meetings were held with clinicians and managers in the regions and districts involved in the study to review the findings and the experience.

In Ghana, officials from the Ministry of Health and other development partners involved in maternity care were invited to observe the panel meetings in action and to discuss the findings with the panel members. The experience gained in Ghana has given impetus to the revival of interest in setting up a routine national confidential inquiry into maternal deaths. After the completion of the work in Indonesia, the Ministry of Health adapted the methodology described here to conduct routine reviews of maternal deaths in health facilities. In both countries, members of the panels involved in the present study were invited to participate in the discussions.