This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
This study examined the population-level factors associated with the under-five mortality rate (U5MR) across 34 provinces in Indonesia using an ecological study design.
Methods
U5MR data were obtained from BPS-Statistics Indonesia. Demographic, healthcare, and fertility-related factors were included as independent variables. An ecological study design was applied, with 34 provinces serving as the units of analysis. Hierarchical Poisson regression was used to assess the associations between the variables. All variables were log-transformed prior to the analysis.
Results
The U5MR in 2020 was 19.83 per 1,000 live births. Fertility and demographic factors were significantly associated with changes in U5MR. A 1% increase in the provincial gross domestic product (P=0.042) and the doctor-to-population ratio (P<0.001) was associated with 3% and 6% reductions in U5MR, respectively. Conversely, a 1% increase in the bed-to-population ratio (P<0.001), age-specific fertility rate (P=0.004), and total fertility rate (P<0.001) was associated with increases of 19%, 5%, and 117% in U5MR, respectively. The midwife ratio (P=0.433) and vaccine availability at primary health centers (P=0.851) were not significantly associated with U5MR.
Conclusion
Improving the U5MR requires equitable healthcare access, addressing doctor shortages with incentive-based placements, enhancing healthcare facilities, and implementing policies to reduce teenage pregnancies through sex education and contraceptive promotion.
The ongoing efforts to integrate Maternal and Child Health into the national strategy of Indonesia and reach an under-five mortality rate (U5MR) as low as 25 deaths per 1,000 live births (according to the Sustainable Development Goals [SDGs] set by the United Nations) have been relatively successful in Indonesia [1]. The latest data from the 2020 Population Census showed that the 2020 U5MR was 19.83 [2]. However, the number should not be overlooked, as many provinces showed little to no improvement in the U5MR from the past until now, especially in the eastern region. Additionally, the number exposed huge disparities in the U5MR across the 34 provinces in Indonesia.
Previous studies have identified several factors influencing U5MR—sociodemographic factors, such as wealth quintile and parity, and medical factors such as maternal comorbidities [3,4]. All studies were performed at the individual level and utilized a cross-sectional design. While individual-level data may provide various advantages, such as the ability to control more biases and confounding factors, they often depend on the availability of granular health data, which remains limited in Indonesia [5]. The 2020 Population Census also did not include health-related data, focusing more on the sociodemographic characteristics of the Indonesian population. In this context, an ecological study design, particularly a cross-sectional study design, offers a more appropriate approach to avoid data scarcity by utilizing routine and aggregate data across provinces in Indonesia [6,7].
This study employed a cross-sectional ecological design to examine the association between population-level characteristics and the U5MR across 34 Indonesian provinces. This approach is appropriate for identifying provincial-level characteristics such as sociodemographic factors that may be associated with the U5MR, particularly when individual-level health data are unavailable. Ecological studies may be suitable for generating early hypotheses and policies by examining patterns captured in the population, although individual-level causal inferences cannot be inferred owing to ecological fallacies [8].
Although ecological research is not uncommon, its application in Indonesia remains underutilized. The novelty of this study lies in its application of an ecological perspective to examine U5MR disparities at the provincial level in Indonesia, identifying the aspects the government should prioritize. Thus, this study addresses data availability challenges, literature scarcity, and a policy-relevant gap in Indonesia’s U5MR health indicators.
Methods
Data sources
This ecological study was conducted across 34 Indonesian provinces in 2020. All data were obtained from open sources, such as the BPS-Statistics Indonesia website and the Ministry of Health’s report. The website provided various datasets from multiple years, and 2020 was selected because it coincided with the 2020 Population Census, marking the latest available data regarding Indonesia’s demographic landscape [9]. The 2020 Population Census, conducted in accordance with Indonesian Law Number 16 of 1997, was carried out online (February to May 2020) and offline (September 2020). The online method (self-administered) was initially used to reach high-mobility individuals who could not be contacted offline by a surveyor. Those who were unable to participate online were later surveyed using offline methods. The questionnaire remained the same across both modes and covered all 34 provinces in Indonesia, establishing it as a representative dataset. From the 2020 Population Census, the dataset of U5MR was collected. Additionally, BPS-Statistics Indonesia collected annual monetary data from the Ministry of Finance, from which the provincial gross domestic product in this study was obtained.
The Ministry of Health’s report is an annual report that provides information regarding the social determinants of health at the provincial level. This information is derived from routine data collection at the district level and aggregated at the provincial level. Data obtained from the reports were (1) total fertility rates (TFR), (2) midwives ratio per 1,000 population, (3) doctor ratio per 1,000 population, (4) bed ratio per 1,000 population, (5) percentage of puskesmas (primary healthcare centers) with vaccine availability for basic immunization, and (6) age-specific fertility rates (ASFR) for those aged 15 to 19 years. All data were collected at the provincial level, and the units of analysis were 34 provinces in Indonesia.
Outcomes
The outcome of this study was U5MR, which measured the number of children aged 0 to 59 months divided by the number of live births. The U5MR was one of the main data interests of the 2020 Population Census and can serve as a proxy indicator for health improvement in Indonesia.
Predictors
The independent variables in this study included various factors: (1) healthcare facilities; (2) economic factors; and (3) fertility factors. Healthcare facilities included the doctor ratio per 1,000 population, the midwives ratio per 1,000 population, and the bed ratio per 1,000 population. All variables were obtained at the provincial level.
The bed ratio per 1,000 population was obtained from the 2020 Indonesian Health Profile. The bed ratio was obtained directly from the report, without requiring further calculations. Conversely, the midwives and doctor ratio per 1,000 population was obtained from the report, but in the respective absolute numbers. The report also included the percentage of puskesmas with all basic vaccines in stock, as defined by the Ministry of Health. These vaccines (hepatitis B, BCG [Bacille Calmette-Guérin], DPT-HBHib [diphtheria–pertussis–tetanus, hepatitis B, and Haemophilus influenzae type b], Polio, and Measles/Rubella) reflect facility availability and not actual child vaccination coverage. To transform the rate form, we divided them by the total population in 2020 and multiplied by 1,000.
Economic factors included the provincial gross domestic product in 2020 (current price). Finally, fertility factors included the ASFR and TFR. Early adolescent ages (15–19 years) were chosen for the ASFR because they were related to teenage pregnancy, one of the most influential factors for stillbirth. Economic and fertility factors were obtained from the BPS-Statistics Indonesia website [10]. The selection of independent variables was justified through a previous study [11]. As this study utilized a secondary dataset, and all data are available on the BPS-Statistics Indonesia website, ethical approval was not required for this research. This waiver corresponded to the agency’s website policy, which permitted unrestricted analysis of their data at no cost, provided that it was properly cited and responsibly interpreted [10].
Statistical analysis
Bivariate and multivariate analyses were performed. Univariate analyses were performed by calculating the mean and standard deviation (SD) for each variable. Additionally, a heat map to examine the rate distribution of U5MR was produced using Tableau (Salesforce Inc.). For multivariate analysis, we applied a hierarchical generalized linear model using a Poisson distribution. The hierarchical model produced three different models: the first included economic factors, the second included the first model and healthcare facilities, and the third included the second model and fertility factors, producing adjusted relative risk. The hierarchical approach allowed the estimation of the effects of explanatory factors by entering them progressively into the model, producing a more structured inclusion of variables. All variables in the model were log-transformed. The exponentiated coefficients represented elasticities, where a 1% increase in the predictor is associated with β% change in U5MR.
The model specifications included (1) multicollinearity, (2) overdispersion, and (3) homoscedasticity. Multicollinearity was assessed using the variance inflation factor (VIF). A VIF score above 10 indicated the presence of multicollinearity. Overdispersion was assessed using Pearson goodness-of-fit (Pearson GoF). A statistically significant result for GoF indicated that overdispersion was present. Finally, to address potential issues with heteroscedasticity, robust standard errors were employed to ensure reliable coefficient estimates. Statistical significance was set to 0.05. All statistical analyses were performed using Stata ver. 17.0 (StataCorp.).
Results
Distribution of U5MR in Indonesia
The U5MR of Indonesia was 19.83, while its mean was 23.58 (SD=9.28). Papua showed the highest rate of U5MR (49.04), contrary to the lowest in DKI Jakarta (12.02). We observed a provincial cluster of high U5MR rates in Eastern Indonesia, including Papua, West Papua (47.23), and Maluku (36.54). Notable U5MR was also seen in provinces in Central Indonesia, especially in the Sulawesi and Nusa Islands (West and East Nusa Tenggara). Provinces in Kalimantan Island showed moderate to little U5MR, and similar rates were also seen in provinces in Sumatra Island, except for Aceh provinces and Bengkulu. Provinces with low U5MR were seen in Java and Bali islands, showing inequalities in Indonesia. Details of U5MR distribution can be observed in the map seen in Figure 1 and Supplement 1.
Table 1 provides the descriptive analysis for all variables in this study. Some variables exhibited a small SD, except for provincial gross domestic products. This implies inequality in government expenditure and resources that may play a role in health outcomes, such as U5MR. Moreover, bed ratio, midwife ratio, and doctor ratio showed considerable variability among provinces in Indonesia. Table 1 also confirmed that there were more midwives than doctors in the general population.
Table 2 provides the association between predictors with U5MR. We found that a 1% change in provincial gross domestic products caused a 3% (95% confidence interval [CI], 1%–6%) decrease of U5MR, and this effect was significant at the 0.05 level. A 1% change in doctor ratio per 1,000 population also caused a 6% (95% CI, 2%–10%) decrease of U5MR. Interestingly, we found that a 1% increase in bed ratio per 1,000 population increased the U5MR by 19% (95% CI, 11%–28%). Regarding fertility factors, we found that a 1% increase in ASFR increased the U5MR by 5% (95% CI, 1%–9%), while a 1% increase in TFR also increased the U5MR by 117% (95% CI, 71%–175%). Midwives ratio per 1,000 population (P=0.433) and puskesmas percentage with vaccine availability (P=0.851) were not significantly associated with U5MR.
The mean VIF for each model confirmed the absence of multicollinearity, as it remained below 10.00 in all cases (Supplement 2). The P-value for Pearson GoF in each model was insignificant (P>0.05), indicating the absence of overdispersion in our model. Furthermore, the robust standard error utilized in each model seemed to hold the assumption of homoscedasticity and provided a more precise CI.
Discussion
Study highlights
This study successfully addresses its aims. First, it examined the distribution of U5MR across 34 provinces in Indonesia and found that the eastern region of Indonesia was unequally affected by U5MR compared to other regions. Second, the study identified disparities in the distribution of healthcare facilities and provincial budgets across 34 Indonesian provinces. Finally, the study confirmed several factors associated with U5MR—provincial gross domestic product, bed ratio per 1,000 population, doctor ratio per 1,000 population, ASFR, and TFR.
Comparison
The U5MR in Indonesia was 19.83. This number was remarkable because it passed the target set by the SDGs [1]. However, the U5MR in Indonesia appeared to be higher compared to neighboring countries such as Singapore (2.3 per 1,000 live births), Malaysia (8 per 1,000 live births), and Thailand (8.7 per 1,000 live births) in the same year, although it showed a better comparison to Indochina countries such as Cambodia (25.5 per 1,000 live births) and Myanmar (42.9 per 1,000 live births) [12]. This showed that the relative position of Indonesia in Southeast Indonesia was moderate. Moreover, our study found that the U5MR in Indonesia was unevenly distributed across 34 provinces, with the eastern regions most affected by the highest U5MR. This implies unequal opportunities regarding socioeconomic and health outcomes in Indonesia [13]. Our findings, justified by notable gaps in the SD in various variables, confirmed this.
This study found that the doctor ratio per 1,000 population was associated with U5MR and that an increase in the doctor ratio decreased the U5MR, although a similar pattern was not observed for the midwives ratio. The additional input of healthcare professionals affected the U5MR, as shown by a previous study [14]. Specifically, the scope of midwifery in Indonesia was limited to maternal and neonatal mortality, especially during the labor process. This was not until 2019 where the framework of Midwifery Act was ratified that included the following: “potentially increase the health status of the society, primarily maternal, newborn, baby, toddler and pre-school health” [15]. We argued that the effect of such a framework lagged behind because this framework was adopted in 2019, and it might take time to understand the impact of the framework and how it would be able to increase the scope of midwives for the U5MR. Furthermore, the U5MR encompasses more than labor processes. In Indonesia, midwives often manage labor in their private midwifery practices. Nevertheless, when complications arise, these practices lack adequate facilities to handle severe cases, requiring referrals to the nearest available doctor and highlighting the critical role of doctors [16].
Our study also found a counter-intuitive result. An increase in the bed ratio was associated with an increase in U5MR. This finding should be interpreted with caution, as regional disparities in healthcare facilities persisted in our data. Notably, our bed ratio data reflect general hospital beds, without differentiation into Neonatal Intensive Care Unit, pediatric, or obstetric beds, owing to data limitations in Indonesia. Additionally, regional disparities in healthcare facility utilization were dominant in our dataset. West Papua, for example, had a U5MR of 47.23 and 2.3 of bed ratio per 1,000 population. Despite a higher bed ratio compared to other provinces, actual access to hospitals and essential medical care remained limited due to geographic barriers, uneven distribution of beds within facilities, and logistical constraints [17,18]. The utilization of other healthcare facilities, such as Antenatal Care (one of the key components in reducing U5MR) in Eastern Indonesia, also reported similar low utilization—bed availability alone could not reflect effective healthcare access [19]. Previous studies support the role of accessibility in child health outcomes. In Ethiopia, children living more than 1.5 hours away from a health center had a significantly higher risk of death than those living closer [20]. Inaccessibility was also associated with higher U5MR and lower healthcare utilization, such as antenatal visits and skilled assistance during delivery, as found in Malawi [21]. Furthermore, a study in Bangladesh showed that an increased distance between mothers’ homes and the nearest health facility was linked to a 15% to 20% increase in the likelihood of neonatal, infant, and U5MR [22]. Future studies should incorporate more specific bed types and service utilization metrics that account for regional accessibility to better interpret these findings.
This study also found that a change in provincial gross domestic product affected the U5MR. This could be explained by provincial gross domestic products as a mechanism of government investing in healthcare expenditure [23]. More spending on health may imply more investment in health, such as healthcare facilities, competent human resources, and increased accessibility of healthcare, reducing the U5MR. The association between such health outcomes and health expenditure also aligned with a previous study [24]. In sub-Saharan Africa, a 1% increase in health expenditure per capita resulted in a 0.5% reduction in the U5MR [25].
Similarly, this study found that increases in the TFR and ASFR were associated with increases in the U5MR. This finding was consistent with that of a previous study [11]. As the TFR constituted the number of children ever born to a woman, the association between TFR and U5MR could be explained similarly to parity. A previous study showed that parity was associated with U5MR [4,26]. Higher parity was common in low-income settings and poverty, making them vulnerable to more health-threatening situations [27]. Such circumstances also prevent higher parity women from accessing more healthcare due to limited resources that could be allocated to each child. A study found that the higher the number of children a woman has, the lower the likelihood of her accessing maternal and child intervention [26,27]. However, interpreting the TFR in our study should be done cautiously because a 1% increase in TFR was associated with a 117% increase in U5MR, a notably high number, although the model specification showed no multicollinearity or overdispersion. Moreover, we conducted sensitivity analyses using alternative model specifications that excluded the TFR. The results were consistent in direction and significance, and the model fit indices (Akaike information criterion [AIC]/Bayesian information criterion [BIC]) showed only minimal differences (TFR: AIC=3.49, BIC=–91.5; without TFR: AIC=3.44, BIC=–94.8), supporting the robustness of the findings. The likelihood ratio test comparing the two nested models was not statistically significant (P>0.05), further indicating that the strong association between TFR and U5MR was not due to model misspecification.
One possible explanation is that the TFR covers a broader reproductive age span, encompassing diverse biological risks, whereas the ASFR in our model was limited to the 15 to 19 age group. Moreover, as the ASFR used in this study was based on adolescent age, an increase in adolescents’ pregnancy was associated with an increase in U5MR. The association was obvious, as adolescents’ pregnancy is classified as a high-risk birth due to heightened dangers of complications for the mother and the child [28]. Further research is necessary to confirm the TFR findings and explore other mechanisms involved.
Policy implication
This study highlighted the importance of equity and healthcare accessibility in reducing U5MR across Indonesia. Therefore, the small number of doctors in the population should be addressed. Current policies, such as the Ministry of Health’s Penugasan Khusus, aim to place doctors and other health professionals in remote and underserved regions, providing financial and career incentives for services in these areas. Similar to China’s policy, which aims to have two to three qualified general practitioners for every 10,000 residents in urban and rural areas by 2020 [29], such a policy could be strengthened by combining increased staffing with the development of local healthcare facilities equipped for mobility access, particularly in geographically challenging provinces such as Papua and West Papua [30]. Providing an incentive-based practice placement could also help increase the number of doctors, retain doctors, and improve equitable health delivery in geographically challenging areas [31].
Finally, this study calls for a comprehensive population policy to reduce teenage pregnancies. Although school-based counseling and peer counseling programs are already available in Indonesia, their effectiveness varies considerably across regions due to differences in implementation quality and commitment from relevant actors (i.e., local government, schools, and community organization) [32]. To enhance these, such implementation should be complemented by widespread distribution of condoms and contraceptives, community-based counseling, and culturally sensitive awareness campaigns.
Strength and limitation
The strength of our study is the recency of data. The 2020 Population Census was the latest survey compared to other surveys such as the Demographic Health Survey, which was conducted in 2017. Second, the hierarchical model used in this study offers new perspectives on how each factor contributes to the U5MR in Indonesia. Despite the interesting findings of our study, there were several limitations.
First, given the nature of an ecological study, an ecological bias exists. Therefore, the results should be interpreted with caution. Moreover, while the predictors in this study were shown to be associated with U5MR in previous studies, some important variables could not be included because of data availability, as we relied on a secondary dataset. This limitation is partly because certain health indicators were not measured in 2020, as government data collection efforts were primarily focused on COVID-19 (coronavirus disease 2019)-related indicators. Furthermore, we could not provide data at the district level because the Ministry of Health and BPS-Statistics Indonesia only aggregated the data at the provincial level. Additionally, the census relied heavily on the memory recall of respondents, causing memory relapse bias.
Conclusion
This study emphasizes the complexity of the U5MR across provinces in Indonesia. Disparities remain significant, particularly in the eastern regions, where access to healthcare is limited. Several provincial-level factors were significantly associated with the U5MR, including gross domestic product, doctor ratio, bed ratio, TFR, and ASFR. Additionally, higher doctor ratios were associated with lower U5MR, indicating the importance of the equitable distribution of healthcare professionals. Conversely, the counter-intuitive association between higher bed ratios and higher U5MR highlights that accessibility, rather than availability, remains crucial in Indonesia.
The government should adopt more specific interventions and region-sensitive approaches, particularly in Eastern Indonesia. Providing incentive-based placement programs, expanding accessible healthcare infrastructure, and drawing specific benchmark solutions from comparable nations such as China could be beneficial for Indonesia. Moreover, comprehensive sex education and improved access to contraceptives are essential, preventing higher numbers of ASFR and TFR and reducing adolescents’ pregnancy and fertility across the population.
Notes
Conflict of interest
No potential conflict of interest relevant to this article was reported.
Funding
None.
Data availability
Data of this research are available from the corresponding author upon reasonable request.
Author contribution
Conceptualization: FAKM, FE, KS. Methodology: FAKM, KS. Software: FAKM. Validation: FAKM. Formal analysis: FAKM. Investigation: FAKM. Resources: FAKM. Data curation: FAKM. Project administration: FAKM. Visualization: FAKM. Supervision: FE, KS. Writing–original draft: FAKM. Writing–review & editing: all authors. Final approval of the manuscript: all authors.
Values are presented as relative risk (95% confidence interval).
VIF, variance inflation factor.
*P<0.05.
**P<0.01.
***P<0.001.
References
1. Sharrow D, Hug L, You D, Alkema L, Black R, Cousens S, et al. Global, regional, and national trends in under-5 mortality between 1990 and 2019 with scenario-based projections until 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation. Lancet Glob Health 2022;10:e195-206.
3. Andegiorgish AK, Woldu HG, Elhoumed M, Zhu Z, Zeng L. Trends of under-five mortality and associated risk factors in Zambia: a multi survey analysis between 2007 and 2018. BMC Pediatr 2022;22:341.
4. Misau YA, Bashir MF, Dabo MM, Adamu AS, Dattijo LM, Musa AZ, et al. Risk factors associated with under five children mortality in Northeast Nigeria; a retrospective analysis of national demographic and health survey data. West Afr J Med 2024;41:651-8.
6. Suchmacher M, Geller M. Ecological study. In: Suchmacher M, Geller M, editors. Practical biostatistics: a step-by-step approach for evidence-based medicine. Academic Press; 2021. p. 25-8
7. Sedgwick P. Ecological studies: advantages and disadvantages. BMJ 2014;348:g2979.
8. Roumeliotis S, Abd ElHafeez S, Jager KJ, Dekker FW, Stel VS, Pitino A, et al. Be careful with ecological associations. Nephrology (Carlton) 2021;26:501-5.
9. Indonesia National Central Agency. Sensus Penduduk 2020 [Internet]. Badan Pusat Statistik; 2020 [cited 2025 Jan 12]. Available from: https://sensus.bps.go.id/main/index/sp2020
10. Indonesia National Central Agency. Badan Pusat Statistik Indonesia [Internet]. Badan Pusat Statistik; 2023 [cited 2025 Jan 12]. Available from: https://www.bps.go.id
11. Tuzun H. Factors related to infant mortality rate and under-five mortality rate in Turkey: an ecological study with provincial data. J Child 2021;21:221-30.
13. Kusumawardani N, Nambiar D, Hosseinpoor AR. Subnational regional inequality in the public health development index in Indonesia. Glob Health Action 2018;11:1500133.
17. Efriandi T. Spatial access to health care: a case study on community health centers in Asmat District, Papua, Indonesia. J Public Health Policy 2021;42:113-26.
18. Tian F, Pan J. Hospital bed supply and inequality as determinants of maternal mortality in China between 2004 and 2016. Int J Equity Health 2021;20:51.
20. Okwaraji YB, Cousens S, Berhane Y, Mulholland K, Edmond K. Effect of geographical access to health facilities on child mortality in rural Ethiopia: a community based cross sectional study. PLoS One 2012;7:e33564.
21. Quattrochi JP, Hill K, Salomon JA, Castro MC. The effects of changes in distance to nearest health facility on under-5 mortality and health care utilization in rural Malawi, 1980-1998. BMC Health Serv Res 2020;20:899.
22. Islam M, Islam K, Dalal K, Hossain Hawlader MD. In-house environmental factors and childhood acute respiratory infections in under- five children: a hospital-based matched case-control study in Bangladesh. BMC Pediatr 2024;24:38.
23. Morgan D, Mueller M. Understanding international measures of health spending: age-adjusting expenditure on health. OECD Publishing;; 2023.
24. Karaman S, Urek D, Demir IB, Ugurluoglu O, Isik O. The impacts of healthcare spending on health outcomes: new evidence from OECD countries. Erciyes Med J 2020;42:218-22.
25. Amegbor PM, Addae A. Spatiotemporal analysis of the effect of global development indicators on child mortality. Int J Health Geogr 2023;22:9.
26. Sonneveldt E, DeCormier Plosky W, Stover J. Linking high parity and maternal and child mortality: what is the impact of lower health services coverage among higher order births? BMC Public Health 2013;13 Suppl 3:S7.
27. Straneo M. Poverty and high parity in rural settings: accumulation of disadvantage in use of hospitals for childbirth in sub-Saharan Africa with a focus on Tanzania [dissertation]. Vrije Universiteit Amsterdam. 2024.
28. Pietras J, Jarzabek-Bielecka G, Mizgier M, Markowska A. Adolescent pregnancy: medical, legal and social issues. J Matern Fetal Neonatal Med 2024;37:2391490.
29. Kou R, Mei K, Bi Y, Huang J, Yang S, Chen K, et al. Equity and trends in general practitioners’ allocation in China: based on ten years of data from 2012 to 2021. Hum Resour Health 2023;21:61.
30. Zhang B, Ma L, Ma W, Kang D, Mao Y. Incentive affects physicians’ willingness to continue participating in urban–rural medical consortia: a comparative study of county and provincial hospitals in Central China. BMC Health Serv Res 2024;24:1581.
31. Xu J, Tian G, Sun J, Liu J, Chen F, Shi Q, et al. Horizontal inequity trends of health care utilization in rural China after the medicine and healthcare system reform: based on longitudinal data from 2010 to 2018. Int J Equity Health 2023;22:90.
32. van Reeuwijk M, Rahmah A, Mmari K. Creating an enabling environment for a comprehensive sexuality education intervention in Indonesia: findings from an implementation research study. J Adolesc Health 2023;73(1S):S15-20.
