Background

Maternal mortality is an essential indicator of the quality of health care. The Sustainable Development Goals (SDGs) call for a global maternal mortality ratio (MMR) of less than 70 maternal deaths per 100,000 live births (LB) and an MMR of less than 140 per 100,000 live births in every country by 2030 [1]. The MMR in South-East Asia was estimated at 134 women per 100,000 live births in 2020. Indonesia has the fourth highest MMR (173/100,000 LB) in the region, following Timor-Leste, Cambodia, and Myanmar [2]. To achieve the SDGs target, the WHO recommends each country implement Maternal Death Surveillance and Response (MDSR) [3]. MDSR is a continuous cycle of identification and notification of maternal death, review of maternal death by a local maternal death review committee, analysis and interpretation of the findings, and response and monitoring of response [3].

In 1994, Indonesia implemented a policy of maternal death notification and maternal death reviews [4]. Subsequently, several programs have been implemented throughout the years to reduce maternal death in Indonesia, though no studies have evaluated the effect of these interventions on maternal outcomes [4]. The Ministry of Health (MoH) in Indonesia introduced comprehensive MDSR in 2016 [4]. Although data on all maternal deaths have been collected, no reports have been written, and little is known about the implementation and effectiveness. Reliable data on national and regional trends of maternal deaths, the causes, and understanding why the women die are essential to identify the main issues and lessons learned and assist policymakers in developing effective strategies to reduce the number of maternal deaths further.

Therefore, the primary objective of this systematic review is to analyse the national and regional MMR in Indonesia and report the MMR trend in the period before 1990 up to 2022. The secondary objective is to report the underlying causes of maternal deaths in Indonesia and the trend in causes of death in the before mentioned time frame.

Methods

This systematic review was conducted using the Cochrane Collaboration principles and the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines (PRISMA) (see supplementary file 1) [5]. We registered the study protocol in PROSPERO [CRD42022320213] on May 25th, 2022.

Setting

Indonesia is an ethnically, culturally, and economically diverse country with a population of more than 270 million people, broadly dispersed over 16,000 islands. For this review, we divided Indonesia into five regions: Java and Bali, Sumatera, Sulawesi, the Eastern part of Indonesia, and Kalimantan, each with several provinces (Fig. 2) [6].

The population density differs across the regions, as well as the density of health facilities and services, with the most health facilities in Java and Bali and the least in Sulawesi and the Eastern part of Indonesia [7, 8]. Per 100,000 people, there are 2–7 primary health centers (hereafter referred to as Puskesmas), one hospital, 50–60 doctors, two gynecologists, and 70–160 midwives. The number of health services per number of inhabitants in Java and Bali is similar to that of other regions in the country. Yet, the density of health facilities per square kilometer (km), and thus the distance and time people must travel to a health center, is significantly different [9]. The average distance to a primary health center is 3 km, and to a hospital is 12 km (0.5 km in Java and Bali, 29 km in Sulawesi) [10].

Indonesia has national health coverage, currently covering 90% of the total population [11]. The remaining 10% have private health insurance or no insurance. People with a high-middle income pay a monthly fee, while poor people are fully subsidized. Insurance covers almost all health services, including antenatal care and delivery [12].

Health care in Indonesia is based on a primary health care concept, where Puskesmas are the basic health care facilities, supported by hospitals and other community-based health facilities, from village to national level. Puskesmas offer antenatal care (ANC) and assistance with uncomplicated births and refer patients with complications to the hospital. All maternal deaths in the community and in health facilities are registered at the district health office and sent to the provincial office, and finally to the national office [13]. No data is available on the proportion of births and maternal deaths in the Puskesmas specifically.

Search criteria

We conducted electronic searches in Pubmed, Embase, Global Health, CINAHL, Cochrane library, and Google Scholar. The search strategy consisted of keywords and MESH terms of “maternal mortality” and “Indonesia”. To ensure the inclusion of local studies, an additional search was conducted in Bahasa Indonesia in Google Scholar and Portal Garuda (supplementary file 2).

Eligibility criteria & study selection

Studies were included if they reported the number of maternal deaths per number of live births (MMR) at the national or regional level in Indonesia and/or if they reported the causes of maternal deaths. We included all eligible studies published before the day of the electronic search on April 29th, 2023. The type of studies included in this review are quantitative observational studies, including cross-sectional (descriptive study), case-control, case-series, prospective study, retrospective study, and mixed method.

The exclusion criteria were: (1) secondary research (systematic review, commentaries, books, regulatory or committee guidance); (2) the full text unavailable; or (3) case reports or case series with less than ten maternal deaths.

We exported the yielded articles to the Mendeley reference manager and removed duplicates. Two authors (MS, BSW) screened every article independently in three steps: (1) title screening, (2) abstract screening, and (3) full-text screening. Rayyan was used for the screening process. In the case of disagreement in study selection, other authors (KV, DN) were consulted. The authors recorded the reasons for excluding studies (Fig. 1 and supplementary file 3).

Fig. 1
figure 1

PRISMA flowchart of study selection (see supplementary file 1 for the PRISMA checklist)

Full size image
Fig. 2
figure 2

Distribution of the studies

Full size image

Data extraction

We collected data on study characteristics (title, year of publication, design study, location of study, study setting: hospital or community-based, and study period), number of maternal deaths, number of live births, MMR per 100.000 live births, and underlying causes of deaths. The included studies were divided into three groups: studies that reported the MMR and the causes of maternal deaths, studies that reported only the MMR, and studies that reported only the causes of maternal deaths.

Quality assessment

To assess the quality of evidence, we used two assessment tools: the AXIS tool (2016) for cross-sectional studies and critical appraisal tools (JBI, 2020) for case-control studies and case series [14, 15]. Two authors (MS & BSW) independently assessed the methodological quality and risk of bias of each article. The AXIS tool consists of 20 items, and the JBI tool consists of 10 items. Each item contains one point (0 or 1). If an item was not applicable (NA) for the study, it was not scored. The obtained score was divided by the total score possible after excluding the NA items. The final score ranged from 0 to 1, categorized into very weak (0–0.20), weak (0.21–0.40), moderate (0.41–0.60), strong (0.61–0.80), and very strong (0.81–1) (see supplementary file 4).

Definitions and summary measures

A maternal death is a woman who dies during pregnancy or within 42 days of the termination of pregnancy, irrespective of the duration and site of pregnancy, and includes both direct and indirect causes of death [16]. Coincidental deaths (such as an accident) were not considered maternal deaths. The maternal mortality ratio is the number of maternal deaths per 100,000 live births. Direct maternal deaths result from obstetric complications of pregnancy or its management, while indirect maternal deaths are those resulting from pre-existent diseases or conditions aggravated by the physiological effects of pregnancy. The causes of maternal deaths were reported according to the WHO, ICD-MM [16]. The studies reporting maternal deaths according to other classification criteria were reclassified according to the ICD-MM guideline.

We reported the data using three time frames: (1) before the Millennium Development Goals (MDGs) i.e., < 1990, (2) during MDGs from 1991 to 2015, and (3) during the SDGs from 2016 up until now. Studies with multi-year observation time, falling into two different time frames, were put in the group with the longest observation time.

Data analysis

Descriptive statistics (percentages) were used to present the trend of maternal deaths and the causes of maternal deaths. We calculated the MMR by summing up the number of maternal deaths divided by the total live births for each time frame at a national and regional level. We used Microsoft Office Excel (Microsoft 365, version 16.70) to synthesize the data in tables, graphs, and text boxes.

Results

We retrieved 3249 studies with our search and selected 63 studies for final analysis. Figure 1 illustrates the selection of studies throughout the different phases of this review. We found 35 studies that reported the MMR and the causes of death, four studies on only the MMR, and 24 studies on only the causes of death in Indonesia. The 63 studies included a total of 254,796 maternal deaths (from 1970 until 2022), with information on the MMR in 244,377 (96%) cases and information on the causes of deaths in 12,893 (5%) cases. Fifty-eight studies were conducted at the regional level (n = 43 in Java & Bali, n = 6 in Sumatera, n = 4 in Sulawesi, n = 3 in the Eastern part of Indonesia, and n = 2 in Kalimantan), and five studies at the national level. The distribution of the studies is shown in Fig. 2.

Table 1 provides an overview of all the characteristics of the included study, the reported number of maternal deaths, the MMR, and the quality assessment. A more elaborate description of the included studies and their methodological assessment can be found in supplementary files 4 and 5, respectively. Thirty-nine studies (62%) were community-based (n = 253,069 maternal deaths, n = 11,219 the causes of maternal deaths), and 24 studies (38%) were hospital-based (n = 1727 maternal deaths, n = 1674 the causes of maternal deaths). The eleven studies (18%) with a very weak or weak-rated methodological quality all had a small number of participants and comprised 0.7% of the total study population (supplementary file 4).

Table 1 Study characteristics reported maternal deaths and MMR, and quality assessment
Full size table

Figure 3 illustrates the trends in MMR per region and time frame. The MMR in Sulawesi of 1811 per 100.000 live births is based on one study conducted in a tertiary facility and is not representative of the entire region [17]. While the MMR declined rapidly throughout all regions in Indonesia in the previous century, it has stagnated since 2015.

Fig. 3
figure 3

The MMR trend in Indonesia in the three-time frames, nationally and per region

Full size image

Table 2 demonstrates the causes of maternal deaths according to the ICD-MM. The three most common causes were non-obstetric complications (n = 3705, 29%), obstetric hemorrhage (n = 3178, 25%), and hypertensive disorders in pregnancy and childbirth (n = 3013, 23%). Among the 3705 cases of non-obstetric complications, 47% concerned other non-obstetric complications (n = 1745), 45% were non-pregnancy related infections (n = 1650), and 8% were cardiovascular diseases (n = 310) (supplementary file 6). The cause of maternal deaths was unknown/unspecified in 12% (n = 1525).

Table 2 Underlying maternal death causes in Indonesia according to the ICD-MM
Full size table

Figure 4 shows the trend of the causes of maternal deaths in the three timeframes. The proportion of maternal deaths due to obstetric hemorrhage and pregnancy-related infections has nearly halved in the past 30 years, respectively from 48 to 18% and from 15 to 5%. Maternal deaths due to hypertensive disorders and non-obstetric causes increased from 8 to 19% and 10–49%. A maternal death with unspecified cause was reported in 12 to 8% of cases, a proportion relatively similar throughout the years.

Fig. 4
figure 4

Trend of maternal death causes in Indonesia in the three-time frames, according to the ICD-MM

Full size image

Discussion

The MMR in Indonesia has declined over the past decades. The national MMR between 2016 and 2020 was 249 per 100.000 live births: a decline of 45% since 1990 and before. This systematic review illustrates a shift in the underlying causes of maternal deaths in Indonesia. While obstetric hemorrhage was the cause of almost half of the maternal deaths two decades ago, non-obstetric complications and hypertensive disorders are currently the most important causes.

The MMR in this review is higher than the estimated number from WHO (173 in 2020) and the Ministry of Health (91 in 2018–2020), most likely due to underreporting resulting from a lack of a robust maternal death surveillance system [2, 18, 19]. The underreporting is found mostly among women who died due to non-obstetric complications, often outside of a maternity ward, and misclassified as ‘no maternal deaths’ [20, 21]. The risk of an MMR estimate lower than the actual MMR is that (1) health workers and policymakers are not as aware and engaged as they can be, (2) the MMR decline (SDGs) may seem less than it actually is, and (3) if surveillance were to be improved, it will initially seem like the MMR increases, while actually, it is the reporting that is more accurate.

Most of the studies in this review were conducted in Java-Bali, a region with a lower MMR than the rest of the country. This is likely a consequence of a higher population density and a higher concentration of tertiary hospitals and research facilities. This overrepresentation may have also caused the national MMR to be underreported in our review. At the same time, Java-Bali is a referral province for complex patients, leading to a high concentration of high-risk pregnancies compared to other regions. Lack of data in some regions, such as Sulawesi and the Eastern part of Indonesia, could hinder the understanding of maternal health problems at the national level, leading to less effective policies as well as interventions in underrepresented areas.

Health facilities and medical staff are concentrated in cities. Yet, except for in Java and Bali, the most of people in Indonesia live in rural areas [8]. Access to health facilities is particularly challenging in an archipelago such as Indonesia, where the average distance to a hospital is almost 60 times more on one island than the other (Java island 0.5 km, vs. Sulawesi island 29 km), which is likely an important contributor to the disparities in pregnancy and childbirth health services and outcomes seen [10].

The high MMR in Sulawesi and the Eastern part of Indonesia can be explained by their sparse population, the lower socio-economic status of its people, the poorer infrastructure with less access to healthcare facilities, and the limited resources for medical staff (e.g. training, protocols and guidelines) [8, 22,23,24]. Additionally, the referral conditions vary and are generally of poor quality due to the highly decentralized health system, poor coordination and communication between primary care and referral hospitals, and lack of standardized protocols [25, 26]. A centralized national referral system can help improve uniformity, equality, and effectiveness of maternal and neonatal care among regions [25].

Implementing MDSR is especially important in Sulawesi and the Eastern part of Indonesia, as the data in these regions is the least reliable, while the MMR is the highest, and reduction strategies will be most effective. Nonetheless, multisectoral engagement, such as improving infrastructure and educational opportunities, is equally crucial to achieving maternal death reduction [27].

The MMR in Indonesia is considerably high for an upper middle-income country, highest in the Southeast Asian region, and the decline in MMR is one of the least in the region [2]. Indonesia is not likely to achieve SDG goal 3 by 2030 unless maternal death reduction strategies are prioritized, including accessible care to all, elimination of health disparities, and improved MDSR to monitor trends. The recent WHO progress report on improving maternal and newborn health indicates that globally, maternal mortality has remained stagnant or even increased in some regions. The report highlights the importance of access to high-quality and respectful maternal and newborn care to reduce maternal mortality [28]. The findings of our review align with this WHO report, as maternal mortality in Indonesia remains high despite the majority of women attending the recommended number of ANC visits and giving birth with a skilled birth attendant [29]. Strengthening the quality of care by increasing the number of facilities providing Basic Emergency Obstetric and Newborn Care and Comprehensive Emergency Obstetric and Newborn Care, especially in rural and underserved areas, in addition to referral system strengthening and expanding research capacity, is key in reducing maternal mortality in all parts of Indonesia.

The fact that the proportion of maternal deaths due to hemorrhage and sepsis decreased is in line with the increasing development of the country and the achievement of quality health care and health workers [30]. Non-obstetric complications are the most frequent causes of maternal deaths in Indonesia, similar to global maternal death proportions [31]. Similar high proportions for non-obstetric complications are reported in Sub-Sahara Africa. However, non-obstetric infections (HIV/TBC) make up the largest share there, while non-communicable diseases make up the largest in the past years [32]. Non-communicable diseases (NCD), including hypertension and obesity, are associated with increasing population density and lifestyle factors, such as low physical activity, high alcohol consumption, smoking, and an unhealthy diet. This growing impact of NCD on maternal health calls for lifestyle intervention and prevention strategies [33,34,35]. The recent COVID-19 pandemic has had its toll on maternal deaths in Indonesia as well, contributing to a quarter of maternal deaths due to non-obstetric complications (see supplementary file 6) due to the disease itself, and most likely even higher due to obstetric complications as the overburdened healthcare system could not provide the same level of quality care and less access to care [36, 37].

In this review, we found that pregnancy with abortive outcomes contributes to 2% (234 cases) of maternal deaths in Indonesia. However, this is likely underreported as approximately 79% of the estimated two million abortions in Indonesia are unsafe, as the practice is not legalized unless it concerns special cases (health emergencies, victims of rape) [38,39,40]. Providing legal, accessible, and free safe abortion services is essential to reduce maternal mortality and morbidity [39].

12% of the maternal deaths were due to unspecified causes. Attributing a cause of maternal death can be challenging, especially if multiple comorbidities are present [41]. The studies included in this review classified maternal deaths heterogeneously, complicating the comparison. Similarly, it was unclear whether the cause of death defined the underlying cause or the mode of death [42]. The implementation of maternal death audit by the local team or community, and at a national level, is crucial to improve maternal data registration, the attribution of causes, and the lessons learned to reduce maternal deaths in a ‘blame-free, shame-free’ manner.

To the best of our knowledge, this is the first and most comprehensive systematic review of trends and causes of death in Indonesia using both international and local journal databases. This study has some limitations. Firstly, the studies included in this review were very small or conducted within a certain context. Secondly, some studies might have reported the same data, considering that certain studies were conducted with a national scope within the same time frame. Thirdly, there is significant heterogeneity between the studies, for example, the use of the MMR denominator (number of live births) and causes attribution (classification system) with no specification of whether the underlying cause or mode of death is reported, limiting us in conducting a meta-analysis [42]. While reclassifying the deaths using ICD-MM enhances local and global comparability, we may have turned in the accuracy of death attribution (as we did not assess each case but only had the locally reported derivative cause).

Conclusion

Maternal deaths in Indonesia nearly halved between 1990 and 2020, though it seems much higher than WHO estimations and considerably higher than other upper-middle income and other Southeast Asia countries. The decline of MMR is most likely a consequence of the decrease in maternal deaths due to hemorrhage and sepsis as a result of improved quality of basic obstetric care. Yet, the decline is too slow to achieve SDG 3.1 by 2030. The disparities in MMR between regions in Indonesia call for reducing health inequalities, social and economic, and travel distance. Great efforts are needed to strengthen the quality of maternal and newborn care, particularly in rural and underserved areas, improving infrastructure and accessibility, developing a national centralized referral system, and implementing incentive programs to address the unequal distribution of healthcare professionals. The share of hypertensive disorder and non-obstetric complications increased, emphasizing the importance of policymakers to steer maternal death reduction strategies more towards the prevention of non-communicable diseases and the promotion of health-related lifestyle interventions. National implementation of MDSR with a strong commitment of all stakeholders and policymakers is necessary to eliminate preventable maternal deaths and achieve the desired SDG maternal death reduction in Indonesia.