Global progress in reducing maternal mortality requires enhancing and incorporating effective midwifery services within primary health services, complemented by timely, competent, and efficient referral to hospital care. Interventions need to be targeted toward the most vulnerable groups such as rural and poor populations, ethnic minorities, and other underprivileged groups.1 In resource-constrained settings, trained midwives are desperately needed for areas with extreme health needs,2 and many countries implement innovative strategies to improve access to maternal health services.3 Questions, however, are raised about ways of ensuring the sustainability of such schemes.4
In Vietnam, there are 52 vulnerable ethnic minority groups, comprising 13.3% of the population. About 91.5% inhabit rural and mountainous areas, live in relative poverty, and have low levels of education.5 Low utilization of maternal health care among ethnic minority groups has been highlighted in several studies. The risk of not giving birth in a health facility among ethnic minority women in 2012 is nearly 20 times higher than among women of majority ethnicity.6 Disparities in access to safe maternal services are likely to increase over time7 and therefore need urgent attention.
Research shows that childbearing practices within ethnic minority groups are highly influenced by cultural values and beliefs. Cultural traditions and local customs can constrain the utilization of maternal health care.8 Modern health care provision in Vietnam is less culturally adapted to ethnic minorities. This is because most health staff members are Kinh, the majority ethnic group, who rarely speak local languages and often do not understand traditions and values of ethnic minorities. Negative attitude and discrimination by health staff discourage women from ethnic minorities to seek health care.9 These intrinsic and contextual factors are interlinked and constitute a complex web of influences that determines the low accessibility of maternal health care by the ethnic minority groups.10
To respond to these challenges and improve utilization of maternal services among ethnic minorities, since 2012, the Ministry of Health in Vietnam has been scaling up the scheme of recruiting and training village-based ethnic minority midwives (EMMs).11 In this scheme, young (under 35 years) ethnic women, nominated by their communities, participate in a 6-month training program on midwifery, which includes attending at least 20 deliveries under the supervision and mentorship by experienced midwives. Afterward, the newly trained EMMs are expected to return to their villages to provide free maternal and neonatal counseling, antenatal checkups, assist normal deliveries at home, provide postnatal and newborn care, “first aid” for complicated deliveries, and referrals to a community health center or a district hospital.11
Similar schemes, targeting hard-to-reach areas, were implemented elsewhere, for example, in Mali, Somalia, and Sudan.3 Studies suggest that many schemes faltered after initial support had ended.4 This article contributes to a growing, although still limited, knowledge on experience of implementing, scaling up, and sustaining such schemes, drawing on our assessment of the EMM scheme in Vietnam. We report results from implementation research,12 sponsored by the World Health Organization, which included designing and implementing interventions package to improve acceptability of EMMs by ethnic minorities and ultimately improve utilization of their services. Objectives of this article are (1) to report changes in acceptability and utilization of EMM services following implementation of interventions and (2) to identify key facilitators of and barriers to utilization of EMM services. We do not examine the effectiveness of EMM services on maternal and child health (MCH) outcomes but report changes in the use of services and key factors that determine the utilization of MCH care provided by the EMMs. Our results and experience should be of interest and relevance to different stakeholders in many countries who are interested in, or are in the process of, implementing innovations to improve access to maternal health care among vulnerable groups.
The study was carried out in 2 provinces, representing 2 main highland regions in Vietnam, Dien Bien and Kon Tum. A quasi-experimental survey with pretest/posttest design was adopted, with the caveat that because women delivered only once during the project time frame, we did not examine changes in responses of the same respondents. A pretest survey, which was conducted in September 2015 and reported elsewhere,13 informed the design of interventions package. The intervention was implemented for 6 months (October 2015-April 2016), with support from and involvement of village leaders, commune health workers, and district and provincial supervisors. The posttest survey was conducted in May 2016. We now describe the study setting and intervention and evaluation components of the study.
The 2 provinces, Dien Bien and Kon Tum, were selected for their high proportion of ethnic minorities and low utilization of maternal health care services. In each province, we selected 2 districts and 2 communes within each district with all corresponding villages. A total of 31 villages from 8 communes with all EMMs (N = 31) were included in the study.
Integrated package of interventions included 3 elements: “launch” meeting to introduce EMMs in their respective villages, EMM monthly performance review and supervision meetings with staffs at community health centers, and 5-day refresher training for EMMs; each is discussed next.
In each village, an introductory or “launch” meeting was organized by the leader or head of the village and community health workers, which involved between 30 and 60 participants including village women's union representatives and, most importantly, the couples or individual women of reproductive age. These meetings were aimed to improve local knowledge of, and attitudes toward, the EMMs, and enhance perceptions of benefits of having EMMs' services, thus increasing utilization of services provided by the EMMs. The costs of each meeting were 500 000 VND (about $25), which included logistics management and snacks and refreshments. Each meeting lasted between 60 and 90 minutes. After initial introductions by the village leader, the head of commune health center (CHC) introduced EMMs, updated participants on their training, and their key competencies and responsibilities. An EMM then introduced herself, reiterating the services she can provide and how she can be reached. The district health center representative, head of the village, and women's union representative then each provided further information about the benefits of the scheme. Following subsequent answers to participants' questions, meetings were usually concluded with some entertainment activities.
Monthly review and supervision meetings with community health center staffs
The monthly CHC meetings were aimed to improve EMM services (number/type/quality) and enhance their job satisfaction. The meetings during the intervention period included EMMs, health center leadership and staff, the district health center representative (attended bimonthly), and the Provincial Reproductive Health Center (attended once).
During these meetings, an EMM reported and reflected on the activities conducted during the past month in her village to a midwife at the CHC who supervised EMM work in each province. Where required, the information was triangulated with reports from the midwife and other village health workers. The subsequent discussion was then aimed to provide further guidance to the EMM by the midwife and the head of CHC and agree to the plan for next month.
All EMMs received 5-day refresher training on postnatal and newborn care to improve their competencies and confidence in providing services. The training was designed by the Department of MCH of the Ministry of Health and conducted by the Provincial Reproductive Health Center under the National Targeted Programme for Health.
We adopted a mixed-methods approach, combining quantitative and qualitative data collection and analysis. Quantitative data were collected using a structured questionnaire with pregnant mothers to determine utilization of services provided by EMMs and key influences on this utilization. All mothers in EMMs' villages with an infant younger than 12 months (during the pretest survey) or younger than 6 months (during the posttest survey) were invited. The response rates were 91.4% and 96.6% in the pre- and posttest surveys, respectively. Data from 244 mothers in the pretest survey and 236 mothers in the posttest survey, who knew about EMM services, were included in the analysis. These sample sizes were calculated on the basis of the single proportion sample size formula with a level of confidence of 95%, sampling error was tolerated at 4%, the proportion of women who use any EMM services was 50% in the pretest survey and 58.6% in the posttest survey (recalculated using result from the pretest survey), and design effect was 2.
Qualitative data were collected, using in-depth interviews (IDIs) and focus group discussions (FGDs), to understand stakeholders' views on determinants of performance and acceptability of EMMs. Key informants who were involved in training, supervision, and other support for EMMs at the provincial, district, and community levels were purposively selected. These included (1) EMMs; (2) mothers-in-law and husbands of women having a child younger than 1 year (who accompanied women to CHCs, in consideration that women already spent significant time completing structured questionnaire); (3) village leaders, village health workers, women's union representative in the village; and (4) health managers at the commune, district, and provincial levels. In total, 10 IDIs and 17 FGDs were conducted during both the pre- and posttest surveys.
A structured questionnaire was developed and used in the pre- and posttest surveys, which included questions on sociodemographic characteristics, knowledge and attitude about EMM services, and utilization of services provided by EMMs. Socioeconomic characteristics included variables on age, ethnicity, educational level, household economic status (poor or nonpoor, using government poverty threshold of monthly household income of 400 000 VND/US $20), road condition surrounding residence, and distance to a nearest health facility. Knowledge about EMM services included knowing that EMMs were trained and how to reach an EMM whenever required. The attitude toward EMM services was measured through perceived trust in each of the services provided by EMMs, which was self-rated using a 5-point Likert scale. Utilization of EMM services was defined as receiving any of the following services: counseling, receiving “first aid” for complicated case, referral to a health facility, antenatal checkup, delivery support, and postnatal care during pregnancy and intrapartum or postnatal periods.
Detailed question guides for IDIs and FGDs were developed to aid qualitative data collection, which had the following sections: work performed by EMMs and their confidence, acceptability of EMMs, facilitators and barriers in providing services in the villages, and support available to EMMs from local community authorities and health sectors.
Data were collected by 2 researchers from the Hanoi University of Public Health (HUPH) and 2 health staff members from the Provincial Reproductive Health Centers. All data collectors were trained beforehand and were female. Community health workers sent out invitation letters to all eligible respondents via village health workers or EMMs. Most mothers were interviewed at community health centers except in the 2 remote villages where interviews were conducted in their villages. Interviews were conducted in Vietnamese (an official language in Vietnam). For those who could not understand Vietnamese (about 3% of participants), interpretation was done by the other ethnic minority women, women village health workers, or community health workers. Completion of questionnaire took about 30 minutes.
Qualitative data collection started with 3 IDIs and 8 FGDs in Dien Bien Province during the pretest survey. Insights gained from previous interviews informed the subsequent interviews, including the selection of further informants. Thereafter, 2 IDIs and 6 FGDs were conducted in Kon Tum, using the revised interview guide. In addition, researchers observed 8 monthly review and supervision meetings at CHCs. Five IDIs and 3 FGDs were held at the posttest survey to compare results with the pretest survey and triangulate quantitative posttest results. Each IDI or FGD lasted 45 to 60 minutes. All IDIs and FGDs were audio-recoded following informed consent. All IDIs and FGDs during the pre- and posttest surveys were transcribed and analyzed by 2 researchers from HUPH, using the thematic framework approach.14
Quantitative data were entered using EpiData and double-checked before cleaning and analysis using SPSS v20. Because of the small number of EMMs and participants in each province, we were not able to compare and analyze the data by each province. To test for differences between the pre- and posttest surveys, the χ2 test and the independent-samples t test were applied. Multivariate logistic regression using the Enter method was applied to identify factors associated with using any EMM services. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were estimated using the logistic regression coefficient. The goodness-of-fit of the model was assessed using the Hosmer-Lemeshow statistics. All statistical tests were 2-sided and considered significant at P < .05.
Qualitative data were analyzed using the thematic framework approach, which includes stages of familiarization, development of coding framework, mapping, and interpretation.14 The coding framework was developed from topic guides, study objectives, results from quantitative analysis, and themes emerging from the data. All transcripts were coded, data charts were developed for each theme, and the charts were used to describe the themes. Given the manageable volume of qualitative data, analysis was conducted mostly manually and results were summarized using the MindMapping software.
In this section, we describe respondent characteristics, followed by changes in utilization of EMM services and key determinants of this utilization.
As shown in Table 1, distributions of mothers' age group and distance to the nearest health facility were significantly different between the pre- and posttest surveys. Compared with the pretest survey, in the posttest survey, there were more participants in Dien Bien than in Kon Tum, more participants younger than 20 years, and those who were living further than 5 km from a nearest health facility. All other socioeconomic characteristics were similar.
Changes in utilization of EMM services
Figure 1 compares utilization of services provided by EMMs between the pre- and posttest surveys.
After the intervention, the proportion of mothers who used any EMM services significantly increased from 58.6% to 87.7%. The biggest increase was identified for counseling, antenatal checkups, postnatal care services, and home delivery assistance. Utilization of all these services increased by more than 30% and reached 84.3%, 70.3%, 66.9%, and 60.9% in the posttest survey, respectively. The smaller significant changes were observed for supporting mothers with danger signs (20%). Utilization of other services (referral and support during referral) also increased, although the changes were not statistically significant.
Determinants of utilization of services provided by EMMs
Three determinants of utilization of services provided by the EMMs emerged from our analysis, which we discuss next.
Knowledge of, and trust in EMM services
Knowledge of, and trust in, maternal health services provided by EMMs were associated with utilization of EMM services after adjusting for other socioeconomic variables in bivariate logistic regression and multivariate logistic regression (see Table 2).
Two factors were significantly associated with the use of any EMM services in bivariate analysis: (1) knowledge of EMM services; and (2) being able to contact an EMM. After adjusting for other potential factors in multivariate logistic regression using the Enter method, mothers who thought that they could contact an EMM whenever they needed were 4.24 times more likely to use EMM services than those who did not (aOR = 4.24; 95% CI, 2.33-7.69); mothers who knew about EMM services were 1.22 times more likely to use EMM services (aOR = 1.22; 95% CI, 1.02-4.06).
As shown in Figure 2, women's awareness and knowledge of EMM services significantly increased from the pre- to posttest surveys, particularly in relation to knowledge of home delivery assistance (69%-76.5%), antenatal checkups (46%-62%), and postnatal care (40%-50%). The knowledge of other services such as counseling and neonatal care also increased, although not significantly.
Utilization of EMM services increased proportionately with level of trust in EMM services. Mothers who scored trust in EMM services higher were 1.1 times more likely to use any EMM services than those who scored trust lower (aOR = 1.1; 95% CI, 1.05-1.15). Mean of trust in all EMM services significantly increased between the pre- and posttest surveys (3.58 vs 3.88, respectively; t test for equality of means P < 0.001).
Increased knowledge of, and trust in, EMM services was also evident in qualitative findings:
[People in the village] trust me because of the monthly women's union meeting talk about me. Because of this meeting, they trust me more and I am not feeling shy in providing services at their house. (IDI_2_KT_EMM2)
The introductory launch meetings were perceived as being particularly useful. In addition, EMMs were also introduced in some other meetings such as meetings with women's union representatives.
EMMs' confidence in providing services
During the intervention period, the monthly meetings, supportive supervision by the CHC staff, and the refresher training program provided an opportunity for EMMs to improve their skills and plan for delivery of services in the villages, which perceived by some as ultimately contributing to improved health outcomes.
I want to have EMM working at the village. EMM working at the village is good for people and contributed to the work of community health centers ... [which helps] provide maternal health care services better. (IDI_2_DB_HW1)
However, qualitative results showed that not all EMMs were fully confident in providing services. This was due to their perceived lack of knowledge, skills, and experience. One EMM reflected that she:
...cannot do much, because of my competence, my competence is low, I am afraid of not being trusted by mothers... I don't know all of pregnant women in the village, those whom I know are mainly because mothers' belly is visible. Because I am a newbie, I am so embarrassed to ask if they are pregnant. I don't dare to ask directly... I am not yet married so I feel so embarrassed. (IDI_2_KT_EMM1)
In postintervention assessment, we found that EMMs particularly lacked confidence in assisting home deliveries, and many EMMs preferred not to provide obstetric services for mothers having their first child. Instead, they referred women to deliver in health facilities.
We found that 2 contextual factors determine utilization of services provided by the EMMs: distance to a nearest health facility and their allowances.
Distance to a nearest health facility was associated with using any EMM services. Table 2 showed that mothers who lived less than 5 km to a nearest health facility were 2.3 times more likely to use any EMM services than those who lived further away (aOR = 2.3; 95% CI, 1.19-4.46). Distance is also an important determinant for EMMs' ability to attend review meetings at CHCs:
I need to visit CHC twice per month, once to participate in monthly CHC meeting, another time is for submitting my report. However, my place is quite far from CHC, about 9 km, road condition is bad, and I have difficulty to afford petrol cost. (FGD_1_DB_EMM1)
In 2013, the Ministry of Health recognized the EMMs as a health cadre.11 EMMs started receiving monthly salary equivalent to village health workers (about US $25-$30) in return for provision of services free of charge. However, payments to EMMs depended on availability of resources within the provincial budget, which meant delayed and insufficient payments. In both Dien Bien and Kon Tum, EMMs received supplementary allowances from the National Targeted Programme (about $10). Small and irregular allowances discouraged EMMs from performing their responsibilities. Their husbands and family members also objected to their continuation of their responsibilities, which led to some EMMs quitting their jobs and seeking other employment:
My husband is angry when I am going out for work as EMM frequently. He scolds me that I go for work all day but have no money, even for buying petrol....(FGD_2_DB_EMM2)
However, although EMM services are free of charge, some people provided the EMMs with small amounts of money as an appreciation of their work or invited them to attend a ceremony of giving name to a newborn. This can build EMMs' confidence and encourage EMMs to continue with their work.
After the intervention, the utilization of EMM services increased, with nearly 90% of mothers using at least 1 service provided by the EMMs, with the strongest increases being in assisting home deliveries and antenatal care. Similar results were reported by a similar scheme in Indonesia.15 However, the increase in utilization of services we found in the short term was much higher than what was found in Pakistan, 4 years after the launch of a similar scheme,4 or in Indonesia, 3 years after the introduction of a scheme.15 The different time frames, that is, 6 months in our case versus several years in these 2 studies, raise a question and perhaps the need to sustain significant short-term gains.
A key task for EMMs was to assist women delivering at health facilities because facility-based intrapartum care and emergency obstetric care are effective strategies to save mothers' lives.16 The proportion of home deliveries was reduced between the pre- and posttest surveys, although it was statistically insignificant among the study population. Our findings are similar to results from other studies.4 On reflection, 2 reasons may explain the modest changes. First, the intervention period was relatively short, and most of the EMMs lacked confidence and felt they lacked skills and experience in counseling, particularly taking account of local birth culture and norms in their work.17 Second, strong preference of home delivery needs a possible time lag before substantial changes can be observed. Better understanding of ways of improving motivation of EMMs and aligning their training and work with cultural norms of ethnic minorities is an agenda for future research to help sustain the EMM scheme.
Increase in the referral rate of women by the EMMs to high-level health facilities was also found to be relatively modest. As EMMs' self-confidence grows, they are likely to assist more home deliveries, possibly leading to a reduction in the rate of referrals. The balance between potential assistance of home deliveries and promoting facility births as a safer option represents a question for consideration for policy makers in Vietnam and in other countries implementing similar schemes.
Trust is a key determinant of choice of a health service provider.18 We also found that trust in EMM services represents an important determinant of utilization of EMM services. We recognize, however, that trust is a complex phenomenon that is determined by a multitude of factors. For example, the young age and the limited experience of EMMs can influence the choice of traditional birth attendants by people who rarely use health services. On the contrary, people who are familiar with health care can perceive EMMs as being less competent than nurses or fully qualified midwives.4 , 18 The need to better understand the underlying causes of (mis)trust is evident from our project, is also emphasized in similar studies, and is another area for future research.
We found that EMMs were not self-confident about their skills and expertise, which can be an important contributor to trust and, ultimately, utilization of their services.
Self-confidence of health staff is an important indicator of their ability and competence19 and can determine staff retention. In Taiwan, nearly one-third of newly graduated nurses left their first jobs within 3 months because of lack of self-confidence.20 Although some studies attempt to understand levels of confidence across different health cadre19 or key determinants of staff confidence,21 limited evidence exists in relation to confidence experienced by midwives, particularly those providing services at the community level such as EMMs. Investigating causes of confidence in detail was outside the scope of our study but is an important question to address in further research.
Distance to a nearest health facility was consistently reported as a barrier to the use of MCH in mountainous areas of Vietnam. All the EMMs involved in our study lived in their respective villages and provided services within a walking distance. On the contrary, the longer distances to community health centers represented clear challenges to EMMs. These, combined with limited allowances, contributed toward resistance of EMMs' families to continue with their responsibilities. There is a need to find ways of ensuring easy access of pregnant mothers to EMM services without putting unnecessary strain on EMMs themselves. Possible practical issues in resolving this include aligning the timing of the submission of monthly reports and monthly meetings and varying location of monthly meetings, including considering meetings in EMMs' villages. It would be important, however, to remember the need to integrate the EMMs within the health system as one way of building and maintaining their self-confidence and ensuring appropriate access to peer support.
A key question is “What is the contribution of the implemented intervention to the increased utilization of services provided by EMMs?” Increased utilization of EMM services clearly requires efforts from both the demand side (ie, users) and the supply side (ie, service providers). On the demand side, the intervention included launch meetings in the villages. These meetings, apart from raising awareness about EMMs, were also a possibility for EMMs to increase their self-confidence. The meeting costs were relatively small compared with effects on awareness, trust, and self-confidence. Replication and scaling up of similar schemes need to carefully consider required investments at a larger scale. On the supply side, attending monthly meetings at CHCs became part of EMMs' routine.11 Attendance of monthly meetings and submission of monthly reports faced different challenges (distance, limited allowances), raising the need to possibly align the timing of different visits to CHCs and perhaps even vary locations of such meetings, for example, scheduling some meetings in EMMs' villages.
Ensuring sustainability of community health worker schemes is important. Different countries experiment with different approaches. For example, in Indonesia and Pakistan, to sustain the scheme and motivate outreach services provision, out-of-pocket payments for services delivery were introduced. However, in Pakistan, this led to decreased utilization of services,4 and in Indonesia, midwives could not financially sustain their practices and started serving wealthier clients, effectively excluding those in greater need.22 Increased and targeted government subsidies can alleviate such pitfalls and ensure more equitable access to quality services. However, feasibility of such additional funding remains a challenge for many resource-constrained settings. Rapid economic developments in Asia may provide an opportunity for increased government subsidies.
We acknowledge the following study limitations. Our small-scale evaluation was designed without a “control group” for comparison. Increased knowledge, change in level of trust, and increased utilization of EMMs, thus, may be altered by others factors. The age brackets of respondents in the pre- and postintervention periods were different, potentially affecting our results. Furthermore, the participation of women from Dien Bien in the posttest survey accounted for almost two-thirds of all responses during the posttest survey, which may also have introduced respondent bias. However, more participants living within longer distances from health facilities in the posttest survey gave us assurance that our results were not too skewed. The total number of respondents was too small to analyze data by province, thus limiting our ability to explain differences by province. Exploring reasons for different response rates (such as those linked to distance or differences in the implementation of intervention) represents an agenda for a larger follow-on study. We did not explore experiences of prior pregnancies and deliveries, which may affect a woman's decision whether to utilize EMM services. Further studies need to consider this issue. Because interventions were developed as an integrated package, we did not aim to establish any hierarchy of intervention components. We recognize, however, that different combinations and sequence of implementation of intervention components can have varying implications on study results. Finally, assessing the intervention effects on health outcomes (such as maternal mortality) was outside the scope and time frame of this study, as we focused on service utilization. Larger and longer studies can usefully explore the link between the EMM scheme and health outcomes.
Overall, considering the benefits we identified so far, we remain optimistic about the potential for the EMM scheme to improve access to MCH services to ethnic minorities in Vietnam. The 3-component intervention implemented within our study provided useful insights into possible support measures to strengthen, replicate, and sustain the scheme from both the demand and supply components. It also revealed possible adjustments that can improve the scheme, such as alignment of timing of monthly reports and meetings and varying location of monthly reviews. Finally, we identified agenda for further research that should help evaluate, and further strengthen, the EMM scheme.
Implications for Policy & Practice
- Combination of launch meetings, monthly reviews, and supervision meetings had enabled to significantly increase the utilization of services provided by EMMs in the context of Vietnam.
- Utilization of services provided by the EMMs in Vietnam is determined by knowledge of, and trust in, EMM services; confidence of EMMs themselves; and wider context (distance to a nearest health facility and supportive policies).
- Integrated health systems interventions, such as the 3-component intervention used in this study, can help increase the uptake of services provided by EMMs and ultimately ensure sustainability of the EMM scheme in Vietnam and other similar schemes in other contexts.
Utilization of services provided by EMMs had significantly increased following the implementation of the 3-component intervention. Key determinants of service utilization included knowledge of, and trust in, EMM services, EMMs' confidence, distance to a nearest health facility, and allowances provided to EMMs. Providing continuous support and integration of EMMs within frontline service provision and ensuring adequate local budget for monthly allowances should help sustain the EMM scheme and ensure continued access to MCH care to vulnerable populations in Vietnam.
The Authors wish to thank the relevant health staffs and patients from the two Provinces for taking part in this study and Dr. Le Minh Thi for her contribution to research design and data collection. The project was financially supported by the grant provided by the World Health Organization (project HQHSR 1409738 task 2.6, WHO Reference 2015/493716-0) and the Vietnam Ministry of Health under the National Targeted Program for Health. The views expressed in this paper are of the Authors only, and do not necessarily reflect the position of the World Health Organization and the Vietnam Ministry of Health.
Ethics clearances for this study were obtained from the Institutional Ethics Review Boards of the Hanoi School of Public Health (reference number 281/2014/YTCC-HD3) and of the World Health Organization (Protocol ID: RPC-759). Informed consent was obtained from all study participants prior to data collection.