Article Text
PDF
Abstract
Background Evidence suggests that women gave birth in diverse types of health facilities and were assisted by various types of health providers. This study examines how these choices are influenced by the Indonesia national health insurance programme (Jaminan Kesehatan Nasional (JKN)), which aimed to provide equitable access to health services, including maternal health.
Methods Using multinomial logit regression models, we examined patterns and determinants of women’s choice for childbirth, focusing on health insurance coverage, geographical location and socioeconomic disparities. We used the 2018 nationally representative household survey dataset consisting of 41 460 women (15–49 years) with a recent live birth.
Results JKN coverage was associated with increased use of higher-level health providers and facilities and reduced the likelihood of deliveries at primary health facilities and attendance by midwives/nurses. Women with JKN coverage were 13.1% and 17.0% (p<0.05) more likely to be attended by OBGYN/general practitioner (GP) and to deliver at hospitals, respectively, compared with uninsured women. We found notable synergistic effects of insurance status, place of residence and economic status on women’s choice of type of birth attendant and place of delivery. Insured women living in Java–Bali and in the richest wealth quintile were 6.4 times more likely to be attended by OBGYN/GP and 4.2 times more likely to deliver at a hospital compared with those without health insurance, living in Eastern Indonesia, and in the poorest income quantile.
Conclusion There are large variations in the choice of birth attendant and place of delivery by population groups in Indonesia. Evaluation of health systems reform initiatives, including the JKN programme and the primary healthcare strengthening, is essential to determine their impact on disparities in maternal health services.
- Health insurance
- Public Health
- Maternal health
- Health policy
Data availability statement
Data may be obtained from a third party and are not publicly available. The datasets are accessible after registration and submission of written request from the Indonesia Bureau of Statistics (https://microdata.bps.go.id/mikrodata/index.php/catalog/SUSENAS).
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ABOUT THIS TOPIC
Health insurance programmes improve healthcare utilisation, including for maternal health services across socioeconomic groups and geographical regions by affecting women’s choice in skilled birth attendants and facility-based delivery.
However, higher-level care with specialist birth attendants is mostly available in areas with good access to health services and to more affluent populations.
There is insufficient evidence on the effect of health insurance, socioeconomic status and geographical region on women’s choice of type of birth attendant and place of delivery in Indonesia and other low-income and middle-income countries.
WHAT THIS STUDY ADDS
We found that high-level maternal health services were mainly enjoyed by women living in the most developed regions in Indonesia and from the wealthiest economic group.
There was a notable synergistic effect of insurance status, place of residence and economic status in women’s choice of type of birth attendant and place of delivery.
Women with health insurance coverage, living in Java–Bali and in the richest wealth quintile, have the highest probability to be attended by OBGYN/general practitioner during delivery and to deliver at a hospital.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The expansion of the Indonesia national health insurance programme might inadvertently widen geographical inequities in maternal healthcare use and outcomes.
Increased hospital-based delivery among those insured could pose higher costs to the insurance programme.
Policy development should be geared towards improving the quality of care in the primary healthcare setting to ensure the quality of care and financial sustainability of the health insurance programme and reduce health inequities.
Introduction
Improving access to safe birth settings and skilled professionals is one of the components of health policy to avert maternal morbidity and mortality in many low-income and middle-income countries (LMICs). Existing studies show that 16%–33% of maternal deaths could be reduced through universal coverage of skilled birth attendance.1 The place of delivery is also linked to the quality of maternal care. In the context of LMIC, home deliveries are less likely to have favourable maternal and newborn outcomes compared with facility-based delivery2–4 due to the higher rate of complications caused by poorer hygiene, limited access to skilled birth attendant (SBA), higher probability of infections as well as limited access to referral health facilities.4–7 Interventions that improve the coverage and uptake of high-quality maternal care are key to achieving sustainable development goal target 3, which aims to reduce neonatal and maternal mortality to less than 12 per 1000 live births and 70 per 100 000 live births by 2030, respectively.8
Indonesia, home to 270 million population, has the largest population and third highest maternal mortality ratio (MMR) in the Southeast Asia region. Despite gradual improvement over the last decades, Indonesia’s MMR at 177 per 100 000 in 2017 was higher than countries with similar economic development levels.9 There are also large variations in maternal outcomes across regions in Indonesia. For example, MMR in Java–Bali—the most developed region in Indonesia—is substantially lower (147 per 100 000) compared with the other regions, particularly the Eastern islands (489 per 100 000).10 Such inequities may relate to the fact that rural and less developed areas have significantly lower health workers and health facilities to population ratio, particularly ones that are equipped for safe deliveries and the management of delivery complications.11–13
The Indonesian government has introduced a series of policies for primary care strengthening to improve outcomes and reduce health inequalities. As the primary health centres (PHCs) are available in all Indonesian districts and are relatively more cost-efficient than hospitals, the intention of such policy is to provide good quality health services and reach the majority of the population in the most efficient way. A similar approach was also integrated into the Indonesia national health insurance programme (Jaminan Kesehatan Nasional (JKN)). Introduced in 2014, the programme aims to cover all populations and increase access to quality health services including maternal and child health,14 15 and help achieve universal health coverage (UHC). Under the JKN regulation, all normal deliveries should take place at primary health facilities. Hospitals should only be used for referral cases of pregnancies and deliveries with complications or in areas where there are no available primary health facilities that are equipped to provide the services.16
However, around 20% of delivery-associated JKN reimbursement at the hospital level was for normal deliveries,17 which translates to around US$42 million spent for services that could essentially be delivered at the primary healthcare settings.16 Anecdotal evidence indicated that this might be due to incorrect referrals and bypassing of the referral system due to the lack of confidence of the patients in the quality of care provided at primary health facilities,18–22 particularly among urban population and those from the wealthier economic background. This may be due to perceived low quality among pregnant women towards primary health facilities,23 24 which is in line with a recent World Bank study25 where most Indonesia PHCs have limited basic amenities and SBAs compared with secondary healthcare facilities.25 26
Previous studies have generally shown variations in the clinical competence and quality of maternal care and health outcomes across different types of birth attendants. An evaluation study conducted in 5 LMICs involving 1524 SBAs revealed that physicians scored higher in knowledge and skills tests related to obstetric care compared with other health cadres, including midwives and nurses.27 Similar findings were also shown in countries such as Cambodia, Tanzania and Afghanistan.28–30 However, other studies show favourable maternal health outcomes for obstetric nurses in Mexico.31 In the Indonesian context, previous studies and a World Bank report show that midwives may have limited training in several key competencies32–34 and are usually assigned to health facilities with more limited infrastructure compared with physicians.
Few studies in Indonesia examined factors associated with skilled birth attendance and facility-based delivery.35–38 For instance, Efendi et al 39 found that high wealth quintile, educational attainment and residing in urban predict childbirth in health facilities. However, these studies have not examined the important heterogeneity in different choices of types of providers nor the levels of health facilities, such as delivery at home, primary care or hospital settings. Analysing data from 41 460 women, our study aims to assess the effect of JKN, the largest single-payer public health insurance in the world, on the choice of different types of SBA and levels of health facilities and how the effect may vary across the socioeconomic groups and regions in Indonesia.
Methods
Data
This study used the Indonesian National Socioeconomic Survey (Susenas) 2018, conducted by the Indonesia National Bureau of Statistics. Susenas is a nationally representative, a multistage stratified survey that collects socioeconomic and health information. Importantly, Susenas includes a specific subsection that measures healthcare utilisation of women aged 15–49 years, including healthcare use during pregnancy and delivery. The original sample of the 2018 Susenas included 285 400 households and 1 098 260 individuals. This study focused on women aged between 15 and 49 years who had a delivery within 24 months prior to the survey. After excluding observations with missing values (1.6%), 41 460 most recent birth histories from women from 34 provinces of Indonesia were included in our analysis (see also online supplemental appendix 1 for sample flowchart).
Supplemental material
This study did not seek a full review from an institutional review board as this manuscript used secondary data analyses using a dataset that is publicly available and contains no personal identifications of the survey participants.
Our main outcome variables are the choices of delivery made by women in their most recent delivery, including the types of (a) birth attendant who assisted during the delivery and (b) the place of delivery.
For the types of birth attendant, the alternatives were: (1) an obstetrician/gynecologist (OBGYN) or a general practitioner (GP); (2) a midwife or a nurse; or (3) a traditional birth attendant (TBA) or other non-health birth attendant. Other non-health birth attendants could include a family member or no assistance at all. And the alternatives for choice of place of delivery were: (1) hospital (including secondary and tertiary hospitals and either of private or public entity); (2) higher primary health facility (including public PHCs or private delivery clinics); (3) lower primary health facility (including auxiliary PHCs or a midwife clinic); and (4) home or non-health facility delivery. The Susenas did not distinguish between public and private health facilities.
Patient and public involvement
This study used a secondary dataset that is publicly available. No direct patient involvement in the design and implementation of the study.
Statistical approach
In this study, a multinomial logit regression model was used to investigate factors associated with medical provider selection. McFadden40 developed the model to better understand rational consumer choices in the economic literature. In health service research, the multinomial logit model is increasingly being used to understand patient choices and medical decisions. The multinomial model is applied to a discrete dependent variable that accepts multinomial unordered outcomes from individual patients. For the purpose of this study, individual patients’ choices of medical providers were defined in our study as two sets (one for types of birth attendants, another for place of delivery) of binary variables indicating which alternative (j=1,…m) each individual (i=1,…,n) selects. In our analysis, j represents the various options for:
Birth attendant:
And the place of delivery:
The choices made by pregnant women depend on their characteristics and perceived benefits from the different types of delivery care. In other words, a pregnant woman chooses her delivery attendant (or place of delivery) given her insurance coverage status, income, education, her place of residence other features such as cultural preferences.41–43 It is important to note that this study did not attempt to capture the effect of women’s autonomy in choosing and deciding where to deliver, but rather to assess how the status of women’s insurance coverage may affect the utilisation of different types of birth attendants and place of delivery.
As the nature of the alternatives available in this study is multicategory and dissimilar from one another, we applied the multinomial logit regression. This approach has been commonly used in health service research.42–44 The probability of a woman choosing an alternative j can be estimated using the following equation:
(5)
We presented results from multinomial logit using relative risk ratio (RRR), controlling for all other covariates included in the analyses. We applied survey weighting to account for the sampling strategy used in the Susenas. All analyses were conducted using Stata SE V.15.
Independent variable
Our primary independent variables of interest are the enabling variables including women’s insurance status, type and region of residency and economic status. Three main types of health insurance ownership in Indonesia are those covered by government-organised national health insurance (JKN), private insurance providers and no insurance coverage. Women’s economic status is measured by using the household’s total expenditure per capita and converted into quintiles. The type of residency is denoted as urban or rural. We are also interested in assessing how the regions of residence affect the choices made by pregnant women in deciding who would assist with their deliveries and where to give birth. Java–Bali region—the most developed region—has much better access to higher quality health services, as opposed to other regions such as Sulawesi or Eastern Indonesia.11
Other independent variables include the sociodemographic characteristics: maternal age (less than 20 years, 20–25 years, 26–35 years and 35+), marital status (currently married or not), the education level (no education, primary, secondary and postsecondary education) and occupation status (working or not working).
We could not include variables on pre-existing medical conditions of the pregnant women or known risk factors for delivery complications as the Susenas survey does not collect such information.
As this study is a collaboration between researchers from LMICs and high-income countries and to promote equitable authorships, we included our reflexivity statement in online supplemental appendix 2.
Results
Descriptive analysis
Of the 41 460 women who recently gave birth, 12 108 (29.2%) chose to be assisted by OBGYN/GP, 24 862 (59.9%) by midwife/nurse and 4490 (10.8%) chose TBA or other non-skilled attendants. The majority of women chose to deliver at a secondary facility (31.1%) followed by higher primary health facility (30.0%), home delivery (25.5%) and lower primary health facility (14.1%).
Sample characteristics are presented in table 1. A large proportion of women were covered by JKN (63.3%), lived in the Java–Bali region (52.6%), had secondary-level education (57.2%), were not working (66.9%), aged between 26 and 35 (51.5%) and were currently married (98.3%).
Sample characteristics by types of providers and place of delivery, 2018
Factors associated with the choice of type of birth attendant
Table 2 presents the RRRs from the multinomial logit model for each alternative given a particular determinant or women’s characteristics for the choice of birth attendant (see also online supplemental appendix 3 for the logit results). We found an increasing relative risk for choosing OBGYN/GP and midwife/nurse over TBA as wealth status increases. The relative risks of choosing OBGYN/GP and midwife/nurse versus TBA were 12.3 and 3.3, respectively, for women in the richest wealth quintile compared with the poorest. Having insurance coverage was also associated with statistically significant higher likelihood in using SBA over a TBA. Compared with those with no insurance coverage, women who were covered by the JKN were 2.9 (95% CI 2.5 to 3.4) and 1.6 (95% CI 1.4 to 1.8) more likely to choose OBGYN/GP and midwife/nurse, respectively, over TBA. Higher RRR for choosing OBGYN/GP (RRR=6.50, 95% CI 3.1 to 13.5) and midwife/nurse (RRR=2.4, 95% CI 1.2 to 4.9) over TBA was observed for women with private health insurance compared with no insurance.
Relative risk ratio (RRR) (95% CI) for the multinomial logistic regression fitted for the choice of birth attendant
We present the predicted probabilities based on the multinomial logit regression in table 3. Health insurance coverage increased the probability of choosing OBGYN/GP from 22.4% to 35.5% (for JKN) and to 45.3% (for private insurance). Those with any insurance coverage also have a lower probability of choosing a midwife/nurse or TBA compared with uninsured women. For instance, the probability for choosing TBA was 3.0% for JKN and 1.7% for private insurance, compared with 5.6% among uninsured women.
Predicted probability of patient choice of different delivery providers
Women living in urban areas were more likely to be attended by OBGYN/GP over TBA relative to rural residents (RRR=2.9, 95% CI 2.4 to 3.7). The probability of choosing TBA is also higher for women in Eastern Indonesia (10.6%) compared with other regions such as Java–Bali (2.8%) or Sumatra (3.4%).
The combined effect of insurance coverage, region of residence and wealth status on the choice of birth attendant is presented in figure 1. The coverage of health insurance increased the probability of having an OBGYN/GP across the regions. The synergistic effect of the insurance coverage, region of residence and economic deprivation was notably large. For example, women who were insured, living in the Java–Bali region and from the richest wealth quintile had the highest probability of choosing OBGYN/GP (54.7%) compared with uninsured women who live in Eastern Indonesia and from the poorest wealth quintile (8.6%). In contrast, the lowest probability of choosing TBA was among insured women, living in Java–Bali and in the richest wealth quintile (1.5%) as opposed to their uninsured, poorest and living in Sulawesi region counterparts (2.3%). Interestingly, the insurance coverage had little effect in choosing a TBA for women living in Eastern Indonesia.
Predicted probability of patient choice of different birth attendants, by insurance status. OB-GYN, obstetrician/gynecologist; GP, general practitioner; TBA, traditional birth attendant.
Region of residence played a significant role in the choice of types of birth attendant. The probability of choosing an OBGYN/GP is higher for women living in the Java–Bali region and were from the richest wealth quintile (54.7%) compared with women in the richest wealth quintile who were also insured but live in the less-developed region of Eastern Indonesia (36.5%).
Factors associated with the choice of place of delivery
The RRR for the multinomial logit regressions and the predicted probability for types of place of delivery is presented in tables 4 and 5 (see also online supplemental appendix 4 for the logit results), respectively. Economic status was associated with choosing hospitals and higher primary health facilities over home delivery, with increasing likelihood as the women’s economic status increases. There was no significant association between choosing a lower primary health facility over home delivery and women’s economic status.
Multinomial logit results for choice of levels of facilities
Predicted probability of patient choice of level of health facilities
Women’s insurance status was also associated with the choice of place of delivery. Compared with no insurance, those with insurance coverage (either JKN or private) were more likely to choose hospital over home delivery (RRR 3.3 and 4.8, respectively) (table 4). Those with any insurance coverage have a lower probability of choosing home delivery (10.7% for JKN and 8.9% for private insurance) compared with 19.7% probability for women with no insurance coverage (table 5). Similar to the choice of birth attendant, women with JKN and private insurance coverage had higher probabilities of choosing hospital delivery (38.3% and 45.8%, respectively) compared with only 21.3% for uninsured women. The variations in the predicted probability for choosing higher or lower primary health facilities were smaller across the women’s characteristics.
The RRR for hospital delivery, higher and lower primary health facility over home delivery were 2.9, 2.9 and 1.7, respectively, for women living in urban areas compared with their rural counterparts. Women living in regions outside Java–Bali were less likely to choose hospital delivery over TBA. Home delivery was also more likely for women living in Eastern Indonesia (21.9%) and Kalimantan (23.9%) compared with other regions such as Java–Bali (8.7%), Sumatra (18.0%) and Sulawesi (18.2%).
Figure 2 shows the synergistic effect of insurance status, region and economic status on the choice of place of delivery. Across the regions, insurance coverage increased the likelihood of having hospital-based delivery and in general, decreased the likelihood of choosing primary health facilities as the place of delivery. As wealth status improves, the effect of having insurance coverage on choosing hospital delivery also increased significantly. The combination of insurance coverage, wealth status and region affects the choice of having home delivery. The lowest probability was observed among insured women in the richest quintile who live in urban Java–Bali (2.3%), while the highest was among poor uninsured women living in Kalimantan (48.3%).
Predicted probability of patient choice of different health facilities, by insurance status.
Discussion
Main findings
Using data from 41 460 childbirth delivery from 34 provinces in Indonesia, we found substantial variations in choice for types of birth attendant and place of delivery by socioeconomic status, geographical region and health insurance status in Indonesia. Our results indicate that health insurance coverage (both JKN and private health insurance) was associated with increased use of higher-level health providers and facilities and reduced the likelihood of choosing to be attended by a midwife/nurse or to deliver at primary health facilities. Specifically, women who are enrolled in the health insurance scheme are more likely to be attended by an OBGYN/GP (increases from 22.4% to 35.5% for JKN and to 45.3% for private insurance), while the probability of visiting midwife/nurse decreases from 72.0% to 61.5% (JKN) and to 53.0% (private insurance). Health insurance coverage also increases the probability of having a hospital delivery, from 21.3% to 38.3% (JKN) and to 45.8% (private insurance). Conversely, having insurance coverage also decreases the probability of delivering at a higher primary health facility, from 38.2% to 35.8% (JKN) and 31.5% (private insurance).
Our results also indicate that high-level healthcare utilisation was mainly enjoyed by women living in Java–Bali and from the richest wealth quintile. Living in urban areas or Java–Bali increased the probability of being attended by an OBGYN/GP during delivery and having delivered at a hospital compared with other regions of Indonesia. Furthermore, compared with the poorest wealth quintile, being in the richest wealth quintile increased the probability of having attended by OBGYN/GP from 22.0% to 52.5%. Women from the richest wealth quintile also had higher probability at 51.5% of having hospital-based delivery, compared with only 23.2% among women from the poorest wealth quintile.
We also found a notable synergistic effect of insurance status, place of residence and economic status in women’s choice of type of birth attendant and place of delivery. Women with health insurance coverage, living in Java–Bali and in the richest wealth quintile have the highest probability of being attended by OBGYN/GP during delivery and to deliver at a hospital. Conversely, women who are in the poorest wealth quintile, without any health insurance coverage and living in Eastern Indonesia had the highest probability of being attended by a non-skilled provider. Home delivery was most likely for women living in Kalimantan, not insured and from the poorest wealth quintile.
Comparison with literature
Our results confirm the findings of previous studies about the impact of the health insurance policies on the choice of birth attendance and place of delivery in Indonesia and other LMICs.41 45–48 Our results showing variations in the choice of birth attendant and place of delivery across socioeconomic groups and geographical regions support the notion that socioeconomic deprivation manifest as inequity in accessing high-quality maternal health services. More deprived groups tend to deliver at home or at lower primary health facilities with less-competent providers,48 which potentially lead to poorer health outcomes. According to a recent report by the World Bank, the ability of public PHCs in Indonesia in providing basic emergency obstetric and neonatal care services is very low.25 For instance, only 48% of the sampled health centres could perform the procedure to remove retained products or tissues, meaning that if postpartum haemorrhage occurred and was caused by retained placental fragments, half of these facilities would not be able to handle the complication. Furthermore, the survey also found urban–rural variations in the PHCs’ capacity in managing maternal delivery services.
Our findings also support previous studies in Indonesia showing that health insurance coverage expansion alone may not be sufficient in improving equitable utilisation of quality skilled birth attendance, particularly for people residing in rural areas.49–51 Indirect costs of treatment, such as travel costs and long travel time to health facilities, posed additional burdens to not only patients but also their families.50 51 Erlyana et al, using the third wave of the Indonesia Family Life Survey, found that the uninsured rural population was more sensitive to out-of-pocket costs related to commuting distance than the medical cost itself.50 A qualitative study in rural India also reveals that the introduction of the Janani Suraksha Yojana (JSY) programme was ineffective to motivate women to deliver at health facilities due to the distance to the health facility, which incurred additional transportation costs and attendant charges.50 52 Moreover, the complex administration of health insurance and the lack of knowledge of how to use it in emergency situations were also identified factors that hinder women from using health services.53
Policy implications
Since the early 1990s, Indonesian policies have been focused on improving the coverage of skilled birth attendance and facility-based deliveries. Resources have been allocated for meeting these targets: establishing the village midwife programme, village birthing posts, equipping public healthcare facilities (Puskesmas) with delivery apparatus, upgrading Puskesmas with Basic Emergency Obstetric Care facilities and building maternity waiting homes to improve physical access to facility delivery.54–56 However, the latest evidence shows positive associations between a lower MMR and a higher number of doctors and the shortest distance to hospitals but not with the number of midwives or distance to PHCs.12 This implies that services offered at the primary healthcare setting might be lower compared with high-level healthcare facilities. Further, the Java–Bali region on average had higher availability of hospital beds and health workers compared with Eastern Indonesia. For instance, districts in the Java–Bali region on average had more than 1400 hospital beds compared with only 238 beds in Eastern Indonesia districts.11 The low confidence level among patients in the quality of primary health facilities has also resulted in bypassing of the referral system and higher use of hospital services,21 22 particularly among the richer population and those living in urban areas. These would influence the high hospital utilisation in the developed region of Java–Bali. And thus, stronger affirmative health system reforms that focus on improving equitable access to high quality maternal services are warranted.
Our results revealed that insurance coverage also shifted the preference for delivering at primary health facilities to higher-level facilities, that is, hospitals and that the richer population consistently had higher levels of hospital-based deliveries. These warrants continued policy discussions for several important reasons. First, the use of hospitals and medical specialists is skewed towards the richer population and women living in Java–Bali region. This is also partly caused by the significant deficit in clinicians and hospital beds, especially in less developed regions.11 57 58 And as such, beneficiaries living in areas with a higher level of economic development will continue to have better access to a greater number of medical and health service providers if the JKN removes the financial barrier to receiving those services. This could result in a widening of the existing regional disparities. On the other hand, residents of less developed regions would remain to have limited access to high-quality health services due to shortages in medical personnel and facilities in those areas.
Second, hospital-based delivery can be much more costly compared with deliveries in primary healthcare setting. As mentioned above, the JKN reimbursement for hospital-based normal delivery is tripled compared with the reimbursement for primary healthcare settings. Based on 2014–2019 JKN reimbursement data, the cost of normal deliveries occurring at the hospital level increased significantly over the year, from US$26.9 million in 2014 to US$64.2 million in 2019, or a 137% increase.17 59 60 This increase is significantly higher than the reimbursement for catastrophic medical conditions such as cardiovascular diseases, which only increased by around 75% within the same period. It is also important to note that the JKN programme has been experiencing funding deficits since the start of its implementation in 2014, with its cumulative deficit now standing at US$1.6 billion.60 Considering the points above, we do not propose to increase the use of high-level health facilities and specialised healthcare providers but rather to promote policies that encourage normal deliveries to occur at the primary healthcare setting and with good quality of care. This is particularly important to improve efficiency, reduce inequities and ensuring the financial sustainability of the JKN programme. A strong health system underpinned by primary healthcare is also supported by the WHO as the core component in the effort to attain UHC in LMICs.61–63 Further, there is a need to strengthen studies that focus on monitoring the effect of JKN on equity in access to quality maternal care.
Achieving equitable access to quality maternal care should be part of the JKN targets. And as the findings suggest, health insurance coverage alone will not address the ongoing inequities in maternal healthcare utilisation. And thus, the JKN programme would benefit from reforms in the health supply side63 64 and stronger strategic purchasing functions, particularly to improve quality maternal health service delivery to the poorest population and those living in the most deprived regions.65 Other financial barriers such as transportation costs should also be included as part of the JKN scheme to improve access in remote areas.
Limitation
There are several important caveats to this study. First, Susenas dataset did not include the specific features of the type of birth attendant or the place of delivery, such as proximity, equipment or drug availability, and costs associated with the different levels of care. This limits our ability to capture other factors that may directly influence women’s choices. Another limitation is that we could not control for the women’s medical conditions or the known risks factors for delivery complications. Women who, for example, have been identified to have high-risk pregnancies may have opted or been instructed to deliver at a hospital and attended by an OBGYN/GP. And thus, cautious interpretation is warranted. For instance, anaemia is more prevalent among the poorest Indonesian population and may increase the likelihood of postpartum haemorrhage and infection and the need for hospital delivery. In this case, our results may underestimate the association between insurance coverage and the choice of place of delivery.
Conclusion
Insurance coverage, type and region of residency as well as the economic status of women determined the choice of birth attendant and place of delivery. If not addressed, the expansion of the JKN programme could result in widening the existing socioeconomic and regional disparities in the use of specialised and high-level maternal healthcare services. Strengthening the primary healthcare sector by ensuring the availability of qualified health workers and facility readiness should be prioritised in the effort of achieving the UHC and to help ensure the financial sustainability of the JKN programme.
Data availability statement
Data may be obtained from a third party and are not publicly available. The datasets are accessible after registration and submission of written request from the Indonesia Bureau of Statistics (https://microdata.bps.go.id/mikrodata/index.php/catalog/SUSENAS).
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Lei Si
Twitter @timarthias
Contributors The study was conceived by TM and JTL. The methodology development and analysis were conducted by TM and JTL. JTL and TM drafted the manuscript and LPP, KA, DAP, and BM contributed to all sections. JTL is responsible overall content. All authors reviewed, edited, and commented on multiple versions of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.