Peer Reviewed Articles Maternal Morality on Household Social Determinants

• Loading metrics

Open up Access

Peer-reviewed

Research Article

Understanding the determinants of maternal mortality: An observational study using the Indonesian Population Census

• Lisa Cameron,
• Diana Contreras Suarez,
• Katy Cornwell

x

• Published: June three, 2019
• https://doi.org/ten.1371/journal.pone.0217386

Figures

Abstruse

Background

For countries to contribute to Sustainable Development Goal iii.1 of reducing the global maternal mortality ratio (MMR) to less than 70 per 100,000 live births by 2030, identifying the drivers of maternal mortality is critically important. The ability of countries to place the central drivers is even so hampered by the lack of data sources with sufficient observations of maternal death to let a rigorous analysis of its determinants. This newspaper overcomes this problem by utilising census information. In the context of Indonesia, we merge individual-level data on pregnancy-related deaths and households' socio-economic condition from the 2010 Indonesian population census with detailed data on the availability and quality of local health services from the Hamlet Census. We use these data to test the hypothesis that health service access and quality are important determinants of maternal death and explicate the differences between loftier maternal mortality and low maternal bloodshed provinces.

Methods

The 2010 Indonesian Population Census identifies 8075 pregnancy-related deaths and 5,866,791 live births. Multilevel logistic regression is used to analyse the impacts of demographic characteristics and the being of, distance to and quality of health services on the likelihood of maternal expiry. Decomposition analysis quantifies the extent to which the difference in maternal mortality ratios between high and depression performing provinces can be explained by demographic and health service characteristics.

Findings

Health service admission and characteristics business relationship for 23% (CI: 17.2% to 28.5%) of the departure in maternal mortality ratios between high and low-performing provinces. The virtually important contributors are the number of doctors working at the community health centre (8.six%), the number of doctors in the village (vi.9%) and distance to the nearest hospital (5.ix%). Altitude to health clinics and the number of midwives at community health centres and village health posts are not significant contributors, nor is socio-economic condition. If the same level of access to doctors and hospitals in lower maternal mortality Java-Bali was provided to the higher maternal bloodshed Outer Islands of Indonesia, our model predicts 44 deaths would be averted per 100,000 pregnancies.

Conclusion

Indonesia has employed a strategy over the past several decades of increasing the supply of midwives every bit a manner of decreasing maternal mortality. While at that place is prove of reductions in maternal mortality standing to accrue from the provision of midwife services at village health posts, our findings suggest that farther reductions in maternal mortality in Republic of indonesia may require a change of focus to increasing the supply of doctors and access to hospitals. If information on maternal death is collected in a subsequent demography, future inquiry using two waves of demography data would bear witness a useful validation of the results found here. Similar research using demography data from other countries is as well likely to be fruitful.

Citation: Cameron Fifty, Contreras Suarez D, Cornwell K (2019) Understanding the determinants of maternal bloodshed: An observational report using the Indonesian Population Demography. PLoS ONE xiv(6): e0217386. https://doi.org/10.1371/journal.pone.0217386

Editor: David Hotchkiss, Tulane University, United states

Received: June 28, 2018; Accepted: May 11, 2019; Published: June iii, 2019

Copyright: © 2019 Cameron et al. This is an open up access commodity distributed under the terms of the Creative Commons Attribution License, which permits unrestricted utilize, distribution, and reproduction in whatever medium, provided the original writer and source are credited.

Data Availability: In relation to information admission, nosotros are legally restricted from providing access to the data. Both the Indonesian Population Census Data and the Village Census Data are collected by the Indonesian Statistical Agency (Badan Pusat Statistik, BPS). The information sets are available for buy from BPS (which is the way in which we accessed them). Purchasers of the information are legally prohibited from sharing the information. Enquiries tin can be directed via their online system at: https://www.bps.become.id/bill of fare/three/cara-mendapatkan-information-bps.html.

Funding: An initial version of the paper was supported by funding from the Australian Department of Foreign Affairs and Trade through the Australia-Indonesia Partnership for Gender Equality and Women's Empowerment (MAMPU) programme (MAMPU-MONASH-2014-i to LC). The funder had no role in study design, information collection and analysis, decision to publish, or preparation of the manuscript. The views expressed in this newspaper are the authors' alone and are not necessarily the views of the Australian Authorities, MAMPU or other partner organisations.

Competing interests: The authors have declared that no competing interests be.

Introduction

The Sustainable Development Goals aim to reduce the global maternal mortality ratio (MMR) to less than seventy per 100,000 live births by 2030 [1]. Identifying fundamental determinants of maternal bloodshed and their relative importance is critical to priority setting in policy development, yet a surprisingly pocket-sized number of studies have quantified the role of such determinants. That very modest numbers of maternal deaths are observed in even large random samples of the population presents challenges for analyses of determinants. For case, the Indonesia Demographic and Health Survey which, owing to the absence of authentic civil registration and expiry reporting, is used to generate the official estimates of maternal mortality rates, surveyed 45,607 women and estimated an MMR of 359 deaths per 100,000 live births in 2012 on the basis of reports of a full of only 92 maternal deaths over the preceding v year menstruum. It is not possible to estimate a model of the determinants of maternal mortality with such data as also few deaths are captured.

To overcome this problem, studies either: 1) examine cantankerous-country or regional data [2–4] or combine household surveys from several countries [5], both of which tin can make it difficult to conspicuously identify determinants equally at that place is a not bad deal of heterogeneity across regions/countries; 2) examine factors affecting uptake of maternal health services rather than maternal death directly [6–13]; or 3) identify cases of maternal death and then append these with a random sample of births (controls) that have not resulted in death [xiv–20]. Collection of the case information is costly and time-consuming and often simply viable over a limited geographic range and many of these studies are restricted to cases and controls admitted to hospitals or health centres, which are likely to be biased to item demographics. This is where demography data can show useful. If suitable questions around the maternal status of recently deceased women are included in census questionnaires, the resultant data offering a solution to the small number of observations in other data sources and allow an examination of determinants of maternal mortality at the national and sub-national levels.

In this paper we utilize the 2010 Indonesian demography information to identify key determinants of maternal bloodshed. More than than 8000 maternal deaths were reported past household members in 2010. 1 draw-dorsum of census data is that information technology frequently provides scant data on potential determinants of maternal mortality. The Indonesian census provides information on socio-economic status. Nosotros enrich this past merging information technology with village level data on the availability and characteristics of health and other community infrastructure from the 2011 Village Census (Potensi Desa, PODES). Previous studies accept used the 2010 Indonesian census to calculate regional MMRs simply have not used the census data to investigate the potential determinants of maternal death [21,22]. We know of simply two other studies that accept used population demography data to report determinants of maternal mortality–in the context of Tanzania and Due south Africa [23–24]. Different this study, neither of these studies had access to detailed wellness service information. We estimate a model of maternal mortality and so implement a decomposition to examine to what extent differences in socio-economical status and access to health services between loftier MMR and low MMR provinces explain the differences in maternal bloodshed.

Methods

Information sources

The Population and Village Censuses are conducted by the Indonesian Statistical Agency (Badan Pusat Statistik, BPS). Further details on the data sources are presented in S1 Appendix. The 2010 Population Demography allows identification of pregnancy-related deaths. It asks:

Has there been a death in this household since 1 Jan 2009? If yes, and the person who died was female and over ten years old: Did [proper name] die while significant, during commitment or the 2 months after birth?

Restricting the sample to women anile xv–49 to friction match the standard definition of the MMR, the resultant group of 8075 deceased women class our cases of maternal decease. To these data we appended the 5,866,791 women anile fifteen to 49 years who had a alive nascence since i Jan 2009 and who form our sample of surviving women (controls) who we will refer to equally at-risk women. While item on the private characteristics of the surviving women is recorded in the census, only historic period and information on other living household members is available for the deceased women. In the analysis below we examine the function of maternal age and household head characteristics. We also utilize information from the census on household housing conditions—floor type and sanitation facilities–every bit command variables. The 2010 census is the only Indonesian population census to date which has collected information on maternal bloodshed. For this reason we bear a cross-sectional analysis.

The Village Demography is a three-yearly survey of village officers in each of Indonesia's over lx,000 villages. It collects information on village characteristics including the chief source of income in the village; health services available in the hamlet (village basic health posts; maternal and child health posts; and village birthing centres); numbers of doctors in the village; number of midwives working at the village wellness post; distances to the nearest infirmary and health centre; and transport infrastructure. This information is collected from village staff, including health staff. In 2011 the Village Census included an additional module which nerveless detailed data on the characteristics of health services from interviews with health service staff. From these data we synthetic variables on the number of doctors and midwives working at the health center; indicators of whether the community wellness service has an inpatients service, and whether the village birthing heart (if it exists) has an inpatients service.

The Hamlet Census aims to provide a tape of infrastructure across all of Republic of indonesia's villages. We were able to match 95% of the villages beyond the Village Census and the Population Census.

The Indonesian health sector

In Indonesia, basic primary health care, including ante-natal intendance, is provided at Village Wellness Posts (Pos Kesehatan Desa, Poskesdes). Villages also have maternal and child health posts (Pos Pelayanan Terpadu, Posyandu) which open periodically (ordinarily once a month) and conduct basic maternal and kid health checks such as monitoring kid growth and providing nutritional advice. They are mainly operated by volunteers under the supervision of the sub-district Community Health Centre (Pusat Kesehatan Masyarakat, Puskesmas). Some villages have hamlet birthing centres (Pos Bersalin Desa, Polindes). Wellness issues that cannot be handled at the village level are referred to the Community Health Centre which, in turn, makes referrals to hospitals.

Maternal health has been a priority expanse in Indonesia'south health policy calendar since the late 1980s [25]. The village midwife program (bidan di desa) was introduced in 1989 with the aim that a trained midwife and nativity facility (polindes) would be placed in every hamlet, aslope engagement of volunteers within the village (kaders) to promote wellness service utilisation [26]. The goal of a midwife in every village was largely achieved and the plan has been shown to accept increased access to skilled nascency attendants [8]. However, concerns most quality quickly surfaced as midwife training falls brusk of the WHO training requirements for a skilled nascence bellboy and graduates take been institute to score poorly in skills tests [27, 28]. Various other strategies accept followed (detailed in S2 Appendix). However, despite the diverse iterations of national strategies, setting of targets, introduction of schemes, and big investments in expanded service provision, quality of and access to maternal health services remain at relatively low levels in Indonesia. The current reality for Indonesia is well short of the goal of universal access to health services that meet minimum standards. Reports by both the World Bank and the Indonesian National University of Sciences place a great deal of emphasis on the geographical inequities in access to wellness care services, especially college level care, and the importance of strategies for reaching the outer islands (non-Java) [26, 29].

An analysis of 6 rounds of Indonesian Demographic and Wellness Survey (DHS) data finds that the charge per unit of maternal wellness care has increased over fourth dimension. In 2012 73% of women gave birth in a health facility, compared to 22% in 1986 [30]. The use of whatever dues-natal intendance increased from 81% to 95% over the same period. The study finds meaning geographic variation in service utilisation with utilisation being highest in Java and Bali. Socioeconomic characteristics are a strong determinant of maternal health service utilisation, with a female parent from a richest quintile household being v.45 times more likely to requite birth in a wellness facility than a mother from a poorest quintile household. However, inequality in access to health care was decreasing over this catamenia with the greatest increases in utilisation being among poorer households.

Conceptual framework

A number of authors have developed frameworks to capture the key factors contributing to maternal mortality. Most are based on a framework based on a sequence of events (pregnancy, complications thereof, and death), to which efforts to reduce maternal mortality must reply. The outcomes associated with these events are each adamant by afar/contextual factors (socio-economic and cultural factors), intermediate factors (wellness condition, reproductive status, access to wellness services, health care behaviour and utilisation) and unknown or unpredicted factors [31]. An extension of this framework includes health system aspects, such equally referral paths for managing obstetrics complications [32] and "access" to services through all stages for a safe maternity [33].

An culling framework in which to frame maternal mortality is the "3 delays": delay in the decision to seek care, filibuster in the arrival of/to intendance, and filibuster in receipt of quality care [34]. We infringe from these existing models to develop the framework shown in Fig ane around which nosotros construction our empirical inquiry. Nosotros distinguish between the socio-economical and the wellness service factors that interact with each other and contribute to maternal mortality.

In the socio-economical context, factors such equally economical status, levels of education and other wellness and demographic characteristics influence maternal health outcomes at both an private and broader (national/community) level. For example, a woman's education, income and admission to household resources have effects on her health and reproductive behaviour and so volition influence her historic period of marriage, family planning services and health care utilize, her power to identify signs of risk and brand use of public health information. Customs economic status influences how resources are used to make good-quality care accessible to those most in need. This includes development of infrastructure (roads, send and communications) to link women with services. While each of these factors may not directly affect the pregnancy, they set the socioeconomic context for a woman's pregnancy and birth experience.

The health services context includes services along the pregnancy cycle–before, during and later that shape maternal health outcomes including death. The supply side is represented by the availability of services and their quality and the demand side is reflected in individual's use of the services. For example, availability of family planning services may influence women's decisions about the timing of their first pregnancy, spacing between pregnancies and/or access to condom abortions. Women'south empowerment, educational activity, and cultural factors bear on their power to seek information and use it and so shapes women'due south utilization for those services.

During pregnancy, the mother's health status is determined by women's access to antenatal care via availability of health clinics and qualified personnel that provide examinations and identify risky pregnancies. Access to specialized care and a referral system allows women to receive advisable treatment during pregnancy and delivery, for case a planned assisted delivery at a infirmary. During delivery, the cardinal factors are who carries out the commitment, where is takes place and preparedness in case of complications. While a skilled nascence attendant (SBA) should exist able to deal with normal deliveries, doctors may be required in case of complications and the availability of advisable equipment or medical supplies (e.m. blood transfusions) is key for survival in emergency situations. The use of antenatal intendance or utilise of skilled support during delivery faces several barriers on the need side, like the affordability of the service or women'southward perception of need and value.

Empirical bear witness has found household income, women's education, fertility rates, level of urbanization, household wealth and women's wellness status to be stiff predictors of maternal mortality [35–38]. Studies identifying determinants of maternal bloodshed at the individual level likewise find that wellness service access and apply play a dominant function. Risk factors (e.yard. mother's age, starting time-births, health status) are exacerbated in areas under-served past health facilities [39] or when there are delays in seeking and reaching care [40].

While in general it has been established that access to ante-natal intendance improves maternal health [41], at that place is show that deliverying in a health facility is a more critical factor [42]. Birth location and available health personnel has been proven vital. In cross-countries studies, the decrease in the MMR has been associated with higher rates of SBAs [43]. There are important aspects of the provision of SBAs that touch on their effectiveness, including quality of training, proximity to clients, access to the necessary drugs and equipment, and coordination in the overall health arrangement [44]. Research also highlights that location of birth is important to the effectiveness of SBAs. Traditional nativity attendants and SBAs have comparable adverse outcomes for births that take place exterior health facilities, suggesting the benefits of SBAs cannot be experienced unless the delivery takes place in a well-resourced health facility, and such care is not reached too late [45].

A systematic review of global causes of maternal deaths estimates that around 83% of maternal deaths in South-East asia were a result of direct obstetric causes, while 17% were due to indirect causes, such as pre-existing medical conditions. Among the direct obstetric causes, haemorrhage was most common (36% of direct causes), followed past hypertensive disorders such every bit eclampsia (17%) [46,47]. Mortality rates are highest at the time of nativity and the 24 hours postpartum [48], suggesting that interventions nearly effective at saving lives ought to ensure emergency care at the time of birth—the presence of skilled attendants at births and timely access to emergency obstetric intendance [49–52]. The barrier of distance for uptake of health services such as utilize of SBAs or specialized obstetric care [53] and deliveries in facilities [54,55] is well-evidenced.

Statistical analysis

Nosotros kickoff past estimating a multilevel multivariate logistic regression model with maternal death as the dependent variable and with a wide range of potential explanators. The full set of explanatory variables includes the age of the women (in years), urban/rural status, household caput characteristics (education and employment status) which as well serve every bit a proxy for income, housing information (sanitation access, floor blazon) as a proxy for wealth, hamlet socio-economical characteristics (majority access to improved water and sanitation, main source of incomes, road type and status) and admission to and characteristics of health services (distance from village to nearest hospital and wellness centre (kms); whether the wellness eye has an inpatient facility; number of doctors and midwives working at the health centre; and inside the village—number of doctors, number of midwives working at the village health postal service; whether the village has a birthing station; and whether the birthing station has an inpatients facility). Variables like the number of doctors or midwifes are a proxy for access to skilled birth support during the commitment. Distinguishing betwixt the number of doctors/midwifes in the village and at health centres helps separating the effect of potentially having skilled support during delivery with or without access to a well-resourced health facility. Distance and availability of inpatient facilities are used as proxies for access to specialized care, relevant in cases of complications during commitment.

Wellness service utilisation data is non available in the information sources used. Although the uptake of wellness services is what ultimately affects maternal mortality, including wellness service utilisation variables equally explanators is anyway problematic as the utilise of wellness services besides reflects health status. Mothers in poorer health or at higher gamble are more likely to utilise the services of health professionals, other things equal. Hence, the coefficient on utilisation of health services is likely to be biased downwards, and the endogeneity of these variables causes all coefficient estimates in the model to exist biased and inconsistent. Methods such as instrumental variables can be used to overcome the endogeneity problem, however they rely on there being appropriate instruments that tin can explain utilisation but do not directly bear on maternal bloodshed. Most variables that bear upon health service utilisation, for example, didactics levels, income etc, would be expected to directly affect maternal wellness. Distance to health services is ofttimes used to instrument for utilisation, which then produces results largely akin to controlling for distance to services equally we practise here. Sometimes studies use community averages of health utilisation to lessen the endogeneity problem simply to the extent that community health service utilisation reflects village health weather condition, the endogeneity problem persists.

We illustrate the predictive power of the model by computing the predicted probability of maternal death for women with differing circumstances. We do this by substituting the characteristics of different types of women into the estimated regression equation. The function of possible confounding of impacts of observable and unobservable variables due to the cross-sectional nature of the data is explored through the inclusion of regional fixed effects. The regions are defined in line with Indonesia'south master island groupings–Java/Bali; Sumatra; Sulawesi; Kalimantan; Eastward and West Nusa Tenggara, Maluku and Papua. The region stock-still furnishings absorb the effect of non-time-varying unobservable regional characteristics. Any unobserved regional characteristics which are correlated with, for case, admission to health care, and the probability of maternal decease will bias the estimates of the effect of health care on maternal death. Past absorbing the outcome of unobserved not-time-varying regional characteristics, the regional fixed effects reduce the probability of the estimates being biased. Further, the sensitivity of the results to the inclusion of the fixed effects provides a sense of the extent to which the results may be driven by confounding of observable and unobservable variables.

Nosotros then conduct a Blinder-Oaxaca decomposition [56, 57] to investigate the extent to which differences in demographics and access to health services drive the differences betwixt low maternal bloodshed provinces and high maternal bloodshed provinces. Specifically, the decomposition allows us to examine to what extent maternal mortality in the poorer performing provinces (those with a higher MMR than the national MMR- high MMR) would improve if wellness admission was similar to that in the ameliorate performing provinces (MMR beneath the national MRR- depression MMR). To practise this nosotros split up the sample into those with above and below average maternal bloodshed. We then gauge a parsimonious model which includes all statistically significant health service variables plus variables widely idea to be key in the Indonesian context, such as the number of midwives, meet S2 Appendix. This parsimonious model is estimated over the sample of observations from provinces which take a MMR college than the national MMR. The decomposition procedure then substitutes the means of the variables in the amend performing regions into the estimated human relationship between the variables and maternal bloodshed for the poorer performing regions and so predicts the MMR if these provinces had the same characteristics as the amend performing provinces. The difference between the bodily MMR in the higher MMR provinces and the predicted MMR for these provinces using all the characteristics of the lower MMR provinces is called the "Total Explained Component".

All statistical assay is conducted using Stata, version 15.0. Standard errors are amassed at the village level. The decomposition is conducted using the Oaxaca command in STATA with the logit option.

Ideals statement

As this study uses secondary data which were fully anonymized earlier they were accessed, no ethics applications were submitted.

Results

The national MMR implied from the census information of 8075 maternal deaths and 5,866,791 live births, is 137 deaths per 100,000 live births. Fig ii shows the variation in MMRs across the Indonesian archipelago as calculated from the census. MMRs are lower in the less-remote and more economically-adult regions of Java and Bali.

Fig 3 ranks the provinces by their provincial MMR from the province with the lowest MMR to the highest MMR and shows the national MMR (cherry-red line) and also the average number of doctors in the hamlet per head of village population. Bali has the lowest provincial MMR of 47. All of Java and islands to the westward accept rates below 200 per 100,000 alive births. Gorantalo province in the north of the Outer Island of Sulawesi has the highest provincial MMR, more than seven times college than that of Bali at 371 deaths per 100,000 live births. Very high rates are also observed in other Outer Island locations. Doctors per head of population are generally college in the depression MMR provinces and lower in the very high MMR provinces (with West Sumatra and North Sulawesi being notable exceptions). Tabular array A in S3 Appendix presents the means of other central variables—health services, education levels and livelihoods—by province.

Tabular array one provides descriptive statistics for our analysis sample. After dropping observations for which explanatory variables on wellness service admission were missing, our final analysis sample consists of five,567,029 women of reproductive historic period. Most women come from households where the household head is engaged in some form of work for income or family gain (93%). Education levels are low with 44% of household heads having attended no higher than primary education. Agriculture is the main source of income in the majority (63%) of communities. Twenty-3 percentage of households accept rudimentary floor material (clay, bamboo or wood) and 18% exercise not have a toilet. Eleven pct come from communities where open defecation is the norm. Road quality and access is limited for simply a small proportion of women (4% live in villages where the widest road is unpaved, and 3% where the road to the village cannot be passed year circular).

Admission to health services and health personnel vary substantially across the archipelago. The average woman lives 12 kilometres from a hospital and three kilometres from a wellness center. Average distances to a infirmary vary widely from 0.five kilometre in Djakarta to 29.0 kilometres in Central Sulawesi (see Table A and Figure A in S3 Appendix). Health centres on average take 10.half dozen midwives and ii.8 doctors. Once more this is highly variable with the median number being two doctors and the highest lx. About all villages accept a health post only many are non staffed by a midwife (average number of midwives is 0.43). Thirty pct of villages take a birthing station, but but six% accept an inpatient facility.

Determinants of maternal mortality

Tabular array 2 reports adapted odds ratios and 95% CIs from the multilevel multivariate logistic regression model including the total range of command variables. Controlling for other factors, the probability of maternal death increases, at a slightly increasing rate, with the age of the mother. Likelihood of maternal expiry is negatively associated with the didactics of the household head and the head being employed. An at-risk adult female from a household whose head has completed secondary education is 63% less probable to die from maternal causes than a adult female from a household whose head has no education.

Wellness service access strongly reduces the gamble of maternal death. Every extra 10 kms a woman is from the nearest hospital is associated with a 3.9% increase in the likelihood of maternal expiry. Although distance to a wellness centre is non associated with maternal death (adj. OR 0.989, 95% CI 0.943–1.038), each boosted dr. at the wellness centre reduces the probability of maternal death by 3.2%. Additional midwives working at health centres are not associated with a decreased run a risk of maternal death but the number of midwives working in the village at the village health post is protective, reducing the likelihood of maternal death past four.8%. The number of doctors in the village is a strongly statistically significant determinant, although the magnitude of the effect is pocket-size (adj. OR 0.990 95% CI 0.984–0.997). Robustness results reported in Tabular array C in S3 Appendix show that the results when regional stock-still effects are included are very like to the model above.

To illustrate the magnitude of these results, consider two 22 year old women whose household and village characteristics are very dissimilar. The first woman lives in a neighbourhood typical of Jakarta. She lives with her husband who has completed secondary school and is employed. Their business firm has a tiled floor and a toilet, with water piped to the house. Near people in their neighbourhood accept like quality housing, the roads are all paved and accessible twelvemonth circular. There is a infirmary one kilometre away and she lives next to a wellness centre which has an inpatient service, and is staffed by v doctors, and viii midwives. In addition, viii doctors live in her neighbourhood. The predicted probability of maternal death for women in these circumstances is 51 deaths per 100,000 live births.

The second woman lives in a small agronomical village in Papua with unpaved roads that wash out in the moisture season leaving the village inaccessible. She lives in a house with no toilet, a dirt floor and an unprotected water source like most others in her village. Her husband is not working and has only completed principal school. There are no health facilities in her village. The nearest hospital is 14 kilometres away, and it is 3 kms to the nearest health centre where in that location is an inpatients service with one doctor and 2 midwives. In that location is no health postal service, no doctors in her hamlet and no birthing station. For the woman in these circumstances the predicted probability of maternal death is 345 per 100,000. The adult female in Papua is more 6 times more than likely to experience a pregnancy-related decease than the adult female in Dki jakarta.

Factors contributing to the difference between high MMR and low MMR provinces

Descriptive statistics on key variables for low and high maternal mortality provinces are provided in Table i. Wellness service availability is worse in college maternal bloodshed provinces, especially with respect to distance to hospitals (10.7 kms versus 17.7 kms). Although the average number of doctors differs only slightly (2.9 versus two.4), Fig iv shows that the distribution of doctors varies markedly with high MMR provinces having a much greater number of clinics with zippo doctors. The vertical axis shows the per centum of clinics (in high MMR or low MMR provinces) which accept the number of doctors shown on the horizontal axis. For example, approximately 8% of clinics in loftier MMR provinces accept no doctors working there versus merely about 1% of clinics in low MMR provinces. The modal number of doctors working at clinics is 2 in low MMR provinces and one in high MMR provinces.

Table iii presents the estimation results for high MMR provinces that underlie the decomposition. This parsimonious model includes mother's historic period, household caput characteristics and central health service characteristics.

Tabular array 4 reports the decomposition results. The maternal bloodshed ratio in the low-performing provinces (191 per 100,000 pregnancies) is 66 percent higher than in the loftier-performing provinces (115 per 100,000 pregnancies). The explanatory variables explain 23.two% of this gap. The average historic period at which women give birth only explains a very modest per centum (1.7%) of the departure in MMRs. If educational attainment and employment rates in high MMR provinces became more like depression MMR provinces the assay predicts that MMRs would worsen by a small amount (0.2% and 1% respectively). Well-nigh all the explained difference is accounted for by differences in access to health services. The number of doctors working at the sub-district wellness centre and in the village are the most important contributors (explaining 8.6% and 6.9% of the gap respectively), followed past admission to a infirmary (explaining 5.nine% of the gap). The number of midwives working at the wellness middle and at the village health mail service and whether the hamlet has a birthing station are not significant contributors to the gap.

Word and conclusions

The national MMR implied from the demography data is lower but not so different from the World Banking concern estimate for 2010 of 165 deaths per 100,000 live births [59] which is much lower than that calculated from the 2012 DHS of 359 per 100,000 live births [60]. Calculating an MMR for Indonesia is complicated as there is not reliable civil registration and death reporting and differing information sources and modelling assumptions produce very different results. Further, estimates using methods that allow an examination of trends over time differ wildly—declines in the MMR of 8% (using estimates from the DHS) and 52% (Maternal Mortality Interpretation Inter-Agency Group modelling). The challenges of calculating an authentic MMR for Indonesia, and using demography data more generally, are discussed in more item in S4 Appendix.

At that place is not bad variation in maternal mortality ratios and access to health services across Indonesia. In densely populated Coffee-Bali (and a modest number of other provinces) maternal mortality ratios are considerably lower than in other areas. Our results show that individual socio-economic status has a strongly protective upshot, consistent with the previous literature [5, 10, 12], but explains very little of the difference in maternal mortality rates across provinces. Health service access, especially to doctors and hospitals, explains a big function of the difference in maternal mortality, consistent with assistance by a health professional previously being found to reduce the risk of maternal decease [fourteen]. Although in Java the average density of midwives has been found to be a strong determinant of assistance past a health professional during birth [12] and distance to a health centre found to be a determinant of maternal expiry (for women who were assisted by a wellness professional) [14], nosotros find that health services provided at the village level, including the number of midwives at village health posts, and the number of midwives at sub-district customs wellness centres are not significant explanators of the difference between high and low MMR provinces. This is likely due to the widespread access to these services throughout Republic of indonesia, in part due to the village midwife program.

Our results show that investment in hospitals and doctors in the Outer Islands would significantly reduce maternal mortality. Indonesia has the lowest doc-population ratio in Due south-Eastern asia [61] and, different some other countries in the region such every bit Thailand, where the provision of skilled birth attendants was followed by increased medical facility chapters and which experienced rapid down trends in maternal mortality, in Indonesia not all health centres tin provide basic obstetric care [62]. Doctors in Republic of indonesia are disproportionately clustered in urban centres on Java, and are often inaccessible to the poor [viii, 30]. If the same level of admission to these services as is available in Coffee-Bali was provided to the low-performing regions in the Outer Islands of Republic of indonesia, our model predicts that the gap between maternal mortality ratios across these regions would drop past just over 20%. That is, in the provinces in which the maternal mortality ratio is currently in a higher place the mean of 191 deaths per 100,000 pregnancies, we would expect 44 deaths to exist averted per 100,000 pregnancies. Policy prescriptions are however complicated by the observation that although maternal mortality ratios are high in the more far-flung provinces, due to its big population, Java/Bali accounts for 46% of all maternal deaths in Indonesia (Table A and Figure B in S3 Appendix).

The great strength of the census data is that they capture then many maternal deaths. They too take some weaknesses. First, in that location is likely to be some underreporting of maternal deaths by respondents in the census; it does not provide information on births for single women; and it does non identify at risk women whose pregnancy did not progress to a live birth. These omissions could bias our results if the relationship betwixt the household and village characteristics and maternal mortality differ for these cases. Hence the results presented hither should be taken to apply to married mothers whose pregnancy progresses to a alive birth. Second, remote communities are more than probable to take missing values in the Village Census. Equally village level wellness care is likely important in remote areas as other sources of medical care are deficient, our results may understate the boilerplate role of hamlet level health care. Finally, as data on maternal mortality is only bachelor in the 2010 census, we are restricted to using a single cross-section of data so information technology is possible that unobserved household and village characteristics confound some estimates, although the stability of the estimates when region fixed furnishings are included suggests this is not the case. Future inquiry will be able to farther explore this issue by examining changes over time if maternal death questions are included in the 2020 census.

Withal these caveats, census data are likely to exist a valuable source of information in many countries and, when merged to community level information on health and other facilities, have the potential to significantly raise our understanding of how to reduce the occurrence of maternal expiry.

Supporting information

Acknowledgments

We are grateful for comments received on an early on version from Krishna Hort and Alison Morgan from the Nossal Institute for Global Health at the Academy of Melbourne and Rachael Diprose from the School of Social and Political Sciences at the University of Melbourne.

References

1. 1. Sustainable Development Goals. Goal three.1: Ensure healthy lives and promote well-beingness for all at all ages. 2015. http://world wide web.un.org/sustainabledevelopment/health/ Cited 15 May 2019.
2. two. Lan CW, Tavrow P. Composite measures of women's empowerment and their association with maternal mortality in low-income countries. BMC pregnancy and childbirth. 2017 Nov;17(two):337.
   • View Commodity
   • Google Scholar
3. iii. Stover J, Ross J. How increased contraceptive use has reduced maternal mortality. Maternal and child health journal. 2010 Sep 1;14(5):687–95. pmid:19644742
   • View Commodity
   • PubMed/NCBI
   • Google Scholar
4. 4. Bulatao RA, Ross JA. Which health services reduce maternal mortality? Evidence from ratings of maternal health services. Tropical Medicine & International Wellness. 2003 Aug;eight(viii):710–21.
   • View Article
   • Google Scholar
5. 5. Graham WJ, Fitzmaurice AE, Bong JS, Cairns JA. The familial technique for linking maternal death with poverty. The Lancet. 2004 Jan three;363(9402):23–7.
   • View Article
   • Google Scholar
6. six. Titaley CR, Dibley MJ, Roberts CL. Factors associated with underutilization of antenatal care services in Indonesia: results of Republic of indonesia Demographic and Health Survey 2002/2003 and 2007. BMC public health. 2010 December;10(1):485.
   • View Article
   • Google Scholar
7. 7. Frankenberg E, Buttenheim A, Sikoki B, Suriastini Westward. Exercise women increment their utilise of reproductive health intendance when it becomes more available? Show from Indonesia. Studies in Family Planning. 2009 Mar;xl(1):27–38. pmid:19397183
   • View Article
   • PubMed/NCBI
   • Google Scholar
8. viii. Hatt L, Stanton C, Makowiecka K, Adisasmita A, Achadi E, Ronsmans C. Did the strategy of skilled attendance at nascence achieve the poor in Indonesia?. Message of the World Health System. 2007;85:774–82. pmid:18038059
   • View Article
   • PubMed/NCBI
   • Google Scholar
9. ix. Baird J, Ma S, Ruger JP. Effects of the World Bank'due south maternal and child health intervention on Indonesia'south poor: evaluating the condom motherhood projection. Social Science & Medicine. 2011 Jun 1;72(12):1948–55.
   • View Commodity
   • Google Scholar
10. 10. Ronsmans C, Scott S, Qomariyah SN, Achadi E, Braunholtz D, Marshall T, et al. Professional assist during nascency and maternal mortality in two Indonesian districts. Bulletin of the Globe Wellness Organisation. 2009;87:416–23. pmid:19565119
    • View Article
    • PubMed/NCBI
    • Google Scholar
11. 11. Heywood P, Choi Y. Health system performance at the district level in Indonesia subsequently decentralization. BMC international health and human rights. 2010 Dec;x(ane):3.
    • View Commodity
    • Google Scholar
12. 12. Achadi E, Scott South, Pambudi ES, Makowiecka K, Marshall T, Adisasmita A, et al. Midwifery provision and uptake of maternity care in Indonesia. Tropical Medicine & International Health. 2007 December;12(12):1490–7.
    • View Article
    • Google Scholar
13. 13. Achadi EL, Achadi A, Pambudi E, Marzoeki P. A Study on the Implementation of Jampersal Policy in Indonesia, World Bank Health, Nutrition, and Population (HNP) Discussion Paper 91325. 2014 Sep.
14. xiv. Scott S, Chowdhury ME, Pambudi ES, Qomariyah SN, Ronsmans C. Maternal mortality, birth with a wellness professional and distance to obstetric care in Indonesia and Bangladesh. Tropical medicine & international health. 2013 Oct;18(10):1193–201.
    • View Commodity
    • Google Scholar
15. xv. Taguchi N, Kawabata G, Maekawa M, Maruo T, Aditiawarman , Dewata L. Influence of socio-economic background and antenatal intendance programmes on maternal bloodshed in Surabaya, Republic of indonesia. Pregnancy Hypertens. 2003 Jul;two(3):183.
    • View Article
    • Google Scholar
16. 16. Akbar A, Wicaksono B, Dachlan EG. Maternal mortality and its mainly possible cause pre-eclampsia/eclampsia in developing state (Surabaya-Indonesia equally the model). Pregnancy Hypertens. 2012 Jul;2(iii):184
    • View Article
    • Google Scholar
17. 17. Akbar A, Laksana MAC, Dachlan EG, Prasetyo B. Maternal decease risk factor score based on hospital reference pattern and maternal condition of eclamptic woman in Soetomo Hospital, Surabaya, Indonesia. Pregnancy Hypertens. 2012 Jul;2(3):183. pmid:26105231
    • View Commodity
    • PubMed/NCBI
    • Google Scholar
18. 18. Mawarti Y, Utarini A, Hakimi Yard. Maternal care quality in near miss and maternal mortality in an academic public 3rd infirmary in Yogyakarta, Indonesia: a retrospective cohort report. BMC Pregnancy Childbirth. 2017 May;17(1):149. pmid:28532393
    • View Commodity
    • PubMed/NCBI
    • Google Scholar
19. xix. Midhet F, Becker Southward, Berendes HW. Contextual determinants of maternal mortality in rural Pakistan. Social Science & Medicine. 1998 Jun fifteen;46(12):1587–98.
    • View Article
    • Google Scholar
20. 20. Run into 10.
21. 21. Hartanto W. Indonesia Maternal Mortality Estimate from 2010 Population Census. Newspaper presented at PAA-APA Workshop on Data Quality and Accessibility in the 2010 Round Censuses in Asia, 2012, Bangkok, Baronial 25–26.
22. 22. Soemantri S. Mortality Data Quality. Newspaper presented at Workshop and Training on SP 2010: Estimation of MMR 2012, Aston Hotel, Bogor, November 12–fourteen.
23. 23. Hanson C, Cox J, Mbaruku K, Manzi F, Gabrysch S, Schellenberg D, et al. Maternal mortality and distance to facility-based obstetric care in rural southern Tanzania: a secondary analysis of cross-sectional census information in 226 000 households. The Lancet Global Health. 2015 Jul ane;iii(7):e387–95. pmid:26004775
    • View Article
    • PubMed/NCBI
    • Google Scholar
24. 24. Garenne Thousand, McCaa R, Nacro K. Maternal bloodshed in South Africa in 2001: From demographic census to epidemiological investigation. Population Wellness Metrics. 2008 December;6(1):4.
    • View Article
    • Google Scholar
25. 25. Shiffman J. Generating political will for safe maternity in Republic of indonesia. Social Science & Medicine. 2003 Mar i;56(6):1197–207.
    • View Article
    • Google Scholar
26. 26. National Academy of Sciences. Reducing Maternal and Neonatal Mortality in Indonesia: Saving Lives, Saving the Future. Report of the Joint Committee on Reducing Maternal and Neonatal Mortality in Republic of indonesia; Development, Security, and Cooperation; Policy and Global Affairs; National Inquiry Quango; Indonesian Academy of Sciences. Washington DC: National Acadamies Press, 2013.
27. 27. Rokx C, Schieber G, Harimurti P, Tandon A, Somanathan A. Health Financing in Indonesia: A Reform Route Map. Directions in Development: Human Development. Washington DC: The Globe Bank, 2010.
28. 28. Koblinsky G "Republic of indonesia, 1990–1999" in: Koblinsky One thousand. (ed.), Reducing Maternal Mortality: Learning from Republic of bolivia, Cathay, Egypt, Honduras, Indonesia, Jamaica, and Republic of zimbabwe. Washington DC: The World Bank; 2003 May.
29. 29. Kosen Due south, Tarigan I, Usman Y, Suryati T, Indriasih E, Harimat , et al. Universal Maternal Health Coverage? Assessing the Readiness of Public Health Facilities to Provide Maternal Health Care in Republic of indonesia. The World Bank; 2014.
30. 30. Nababan H, Hasan M, Marthias T., Dhital R., Rahman A. and Anwar I. (2018). Trends and inequalities in use of maternal health care services in Indonesia, 1986–2012. International Periodical of Women'south Health. 10:11–24. pmid:29343991
    • View Article
    • PubMed/NCBI
    • Google Scholar
31. 31. McCarthy J, Maine D. A framework for analyzing the determinants of maternal bloodshed. Studies in family planning. 1992 January 1;23(1):23–33. pmid:1557792
    • View Article
    • PubMed/NCBI
    • Google Scholar
32. 32. Tinker A, Koblinsky MA. Making motherhood safety. The World Banking concern; 1993 May ane.
    • View Commodity
    • Google Scholar
33. 33. Koblinsky MA, Tinker A. Programming for safe motherhood: a guide to action. Wellness Policy and Planning. 1994 Sep i;9(3):252–66. pmid:10137741
    • View Article
    • PubMed/NCBI
    • Google Scholar
34. 34. Thaddeus S, Maine D. Likewise far to walk: maternal mortality in context. Social science & medicine. 1994 Apr i;38(8):1091–110.
    • View Article
    • Google Scholar
35. 35. Ronsmans C, Graham WJ, Lancet Maternal Survival Series steering grouping. Maternal mortality: who, when, where, and why. The lancet. 2006 Sep 30;368(9542):1189–200.
    • View Article
    • Google Scholar
36. 36. Chowdhury ME, Ahmed A, Kalim Northward, Koblinsky M. Causes of maternal mortality decline in Matlab, Bangladesh. Journal of wellness, population, and nutrition. 2009 Apr;27(2):108. pmid:19489410
    • View Article
    • PubMed/NCBI
    • Google Scholar
37. 37. Shrestha South, Bong JS, Marais D. An analysis of factors linked to the decline in maternal mortality in Nepal. PloS one. 2014 April 4;9(4):e93029. pmid:24705366
    • View Article
    • PubMed/NCBI
    • Google Scholar
38. 38. Hussein J, Bell J, Iang MD, Mesko N, Amery J, Graham W. An appraisal of the maternal mortality decline in Nepal. PloS one. 2011 May 26;vi(5):e19898. pmid:21637836
    • View Article
    • PubMed/NCBI
    • Google Scholar
39. 39. See 19
    • View Article
    • Google Scholar
40. 40. Ganatra BR, Coyaji KJ, Rao VN. Too far, too little, too tardily: a customs-based case-control report of maternal mortality in rural westward Maharashtra, India. Bulletin of the World Health Organization. 1998;76(6):591. pmid:10191555
    • View Article
    • PubMed/NCBI
    • Google Scholar
41. 41. Conway KS, Kutinova A. Maternal wellness: does prenatal intendance make a difference?. Health Economics. 2006 May;xv(five):461–88. pmid:16518834
    • View Article
    • PubMed/NCBI
    • Google Scholar
42. 42. Etuk SJ, Etuk IS, Ekott MI, Udoma EJ. Perinatal outcome in pregnancies booked for antenatal care but delivered outside health facilities in Calabar, Nigeria. Acta tropica. 2000 Feb 25;75(one):29–33. pmid:10708004
    • View Article
    • PubMed/NCBI
    • Google Scholar
43. 43. Scott S, Ronsmans C. The human relationship between birth with a health professional and maternal bloodshed in observational studies: a review of the literature. Tropical Medicine & International Wellness. 2009 December one;14(12):1523–33.
    • View Article
    • Google Scholar
44. 44. See 10, 26 and Harvey SA, Blandón YC, McCaw-Binns A, Sandino I, Urbina 50, Rodríguez C, et al. Are skilled birth attendants really skilled? A measurement method, some agonizing results and a potential way forrard. Message of the World Health Organization. 2007;85:783–90. pmid:18038060
    • View Article
    • PubMed/NCBI
    • Google Scholar
45. 45. Badriah F, Abe T, Baequni B, Hagihara A. Skilled versus unskilled assistance in domicile delivery: maternal complications, stillbirth and neonatal death in Republic of indonesia. Journal of Nursing and Care. 2014;3:198.
    • View Article
    • Google Scholar
46. 46. WHO. Trends in Maternal Mortality: 1990 to 2013 Estimates by WHO, UNICEF, UNFPA, The World Banking company and the Un Population Sectionalisation, Geneva: Earth Health Organization. 2013
47. 47. Say 50, Chou D, Gemmill A, Tunçalp Ö, Moller AB, Daniels J, et al. Global causes of maternal death: a WHO systematic assay. The Lancet Global Health. 2014 Jun ane;two(6):e323–33. pmid:25103301
    • View Commodity
    • PubMed/NCBI
    • Google Scholar
48. 48. Li XF, Fortney JA, Kotelchuck Yard, Glover LH. The postpartum menstruum: the central to maternal mortality. International Journal of Gynecology & Obstetrics. 1996 Jul 1;54(1):1–0.
    • View Article
    • Google Scholar
49. 49. Goldenberg RL, McClure EM. Maternal, fetal and neonatal mortality: lessons learned from historical changes in high income countries and their potential application to depression-income countries. Maternal health, neonatology and perinatology. 2015 Dec;1(one):3.
    • View Commodity
    • Google Scholar
50. 50. Campbell OM, Graham WJ, Lancet Maternal Survival Series steering group. Strategies for reducing maternal mortality: getting on with what works. The lancet. 2006 Oct 7;368(9543):1284–99.
    • View Article
    • Google Scholar
51. 51. Jowett M. Safe motherhood interventions in low-income countries: an economic justification and bear witness of cost effectiveness. Wellness policy. 2000 October 1;53(3):201–28. pmid:10996067
    • View Article
    • PubMed/NCBI
    • Google Scholar
52. 52. Chowdhury ME, Botlero R, Koblinsky K, Saha SK, Dieltiens One thousand, Ronsmans C. Determinants of reduction in maternal mortality in Matlab, Bangladesh: a xxx-year cohort study. The Lancet. 2007 Oct xiii;370(9595):1320–8.
    • View Article
    • Google Scholar
53. 53. Run into 14 and Chowdhury ME, Ronsmans C, Killewo J, Anwar I, Gausia K, Das-Gupta Due south, et al. Equity in use of home-based or facility-based skilled obstetric intendance in rural Bangladesh: an observational study. The Lancet. 2006 Jan 28;367(9507):327–32.
    • View Article
    • Google Scholar
54. 54. Kumar Due south, Dansereau EA, Murray CJ. Does altitude affair for institutional delivery in rural India?. Applied Economics. 2014 November 22;46(33):4091–103.
    • View Article
    • Google Scholar
55. 55. Masters SH, Burstein R, Amofah One thousand, Abaogye P, Kumar S, Hanlon M. Travel fourth dimension to maternity intendance and its effect on utilization in rural Republic of ghana: a multilevel analysis. Social Science & Medicine. 2013 Sep i;93:147–54.
    • View Article
    • Google Scholar
56. 56. Blinder As. Wage bigotry: reduced grade and structural estimates. Journal of Man Resources. 1973 Oct 1:436–55.
    • View Article
    • Google Scholar
57. 57. Oaxaca R. Male-female wage differentials in urban labor markets. International Economic Review. 1973 October 1:693–709.
    • View Article
    • Google Scholar
58. 58. Shapefile http://www.statsilk.com/maps/download-free-shapefile-maps.
59. 59. https://data.worldbank.org/indicator/SH.STA.MMRT
60. 60. Statistics Indonesia. Indonesia Demographic and Wellness Survey 2012 Final Written report, FR275, 2013.
61. 61. National Enquiry Council. Reducing Maternal and Neonatal Mortality in Indonesia: Saving Lives, Saving the Time to come. National Acadamies Press, Washington DC, 2013.
62. 62. Acuin CS, Khor GL, Liabsuetrakul T, Achadi EL, Htay TT, Firestone R, et al. Maternal, neonatal, and child health in southeast Asia: towards greater regional collaboration. The Lancet. 2011 February 5;377(9764):516–25.
    • View Article
    • Google Scholar

miltonforelut.blogspot.com

Source: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0217386

Share this post