We used the data from Indonesian Health Survey (SKI) 2023 and basic health survey (Riskesdas) in 2013 as well as in 2018 — those regularly conducted by Ministry of the Health through the agency of health development policy. All three surveys are comparable across years due to the common methods of data collection, measurement, analysis and indicator evaluation used for each. At each stage of the two-stage stratified cluster sampling procedure used in the survey, two sampling frames were employed. In the 2013 Riskesdas, data were collected from 294,959 households and 1,027,763 individuals across 11,986 census blocks in 497 districts and cities within 33 provinces. ²⁸ The 2018 Riskesdas expanded its scope, covering all 34 provinces and 514 districts and cities, with a total of 282,654 households and 1,017,290 individuals sampled from 29,821 census blocks. ²⁹ For SKI 2023, data were successfully recorded from 315,646 general households and 284,178 households with children under-five. At the individual level, interviews were conducted with 1,191,692 individuals, while blood sampling and dental and oral examinations were performed on 37,443 participants. This study encompassed 38 provinces and 514 districts and cities across Indonesia. ³⁰ The inclusion criteria consisted of all women aged 15–49 who were eligible for interviews. Testing on anemia was done on eligible pregnant women who provided informed consent. Rapid Diagnostic Test (RDT) kits were used to test for malaria on children between the ages of 6 and 59 months. Height and weight measurements were recorded for children aged 0–59 months and women aged 15–49 years.

The data encompasses various aspects, including and demographic and socioeconomic traits of women and children, household conditions, MCH status indicators, and healthcare service utilization, collected through structured questionnaires. Additionally, physical assessments such as measurements of women’s and children’s height and weight, as well as anemia and malaria testing, were conducted. The other category encompasses healthcare service utilization variables, including antenatal care (ANC), skilled birth attendance (SBA), delivery in healthcare facilities, complete postpartum care visits, complete vaccination coverage, medical care for childhood diarrhea, and the prevalence of contraceptive, as well as the use of Long-Lasting Insecticide Nets (LLINs) for children.

The study analyzed indicators in two main categories: MCH outcome indicators, includes under-five mortality rate (U5MR), infant mortality rate, stunting, underweight children, prevalence of postpartum complication among women, prevalence of ARIs, and diarrhea in children under-five. Additionally, maternal health was assessed using low Body Mass Index (BMI) in women. Children were classified as stunted or underweight based on WHO reference standards. A height-for-age Z-score that was less than −2 standard deviations (SD) from the median was considered stunting, while underweight was determined using a weight-for-age Z-score below −2 SD. ^{31 , 32} Measurements were taken using a digital weighing scale (accuracy: 0.1 kg) and a height/length measuring instrument (accuracy: 1 mm). Children’s nutritional status was evaluated using the 2006 WHO growth criteria. Women aged 15 to 49 years had their height and weight recorded. BMI was calculated as weight (kg) divided by height squared (m ²). A BMI below 18.5 kg/m ² was used to determine whether this group is thin or suffering from acute malnutrition. ³³ Diagnosis of childhood diarrhea by health workers within the previous 1 month. Malaria testing was conducted only on children between the ages of 6 and 59 months using a RDT and thick blood smear examination, with diagnoses confirmed by a doctor within the previous 12 months. Testing was performed on children who had experienced fever within two days prior to data collection for the SKI 2023.

Indicators were harmonized across survey years to ensure conceptual and operational comparability, despite minor differences in wording and definitions. Differences in wording and indicator structure across surveys were aligned based on core concepts and consistently available indicators. ^{30 , 34 , 35} Variable definitions were also adjusted to reflect the regulations in force during each survey period, such as the completeness of ANC visits defined as ≥4 visits in Riskesdas 2013 and 2018 and ≥ 6 visits in SKI 2023. ^{36 , 37} Despite variations in the categorization of variables among surveys, this approach allows for accurate trend analysis.

Quintile ratios and concentration indices were used to measure the dependent variable, which was inequality levels. The quintile ratio assesses health outcomes and access to health services across economic classes by measuring the disparity in health outcomes for the richest and the poorest. ³⁸ The 2013 Riskesdas and 2023 SKI considered household asset ownership data for this purpose. Household economy status was assessed based on ownership of durable assets such as houses, vehicles, motorcycles, bicycles, refrigerators, and others. Using Principal Component Analysis (PCA), several durable asset indicators were combined to create a single index that classified households into five wealth quintiles (from poorest to richest). ^{30 , 34} The data indicate that the publicly available dataset for the 2018 Riskesdas did not include a wealth index at the individual level, unlike the 2013 and 2023 datasets. The concentration index, in contrast, provides a relative measure of health inequity and captures the level of socioeconomic stratification by health indicators across the five wealth quintiles. ³⁸ This study focused on the absolute percentage difference in the average population between the richest and poorest, and also between rural and urban populations. For the purpose of understanding the patterns of inequalities in detail, the study also focused on measuring the concentration index and the quintile ratio of the variable to identify the positive or negative changes in the variable over time.

The concentration index and quintile ratio both measure inequality; however, these techniques serve distinct, yet complementing purposes. For example, analyzing health status and health care utilization between the wealthiest quintile and the poorest quintile employs the quintile ratio. This measure describes the difference between the richest and the poorest. However, it, fails to describe the health inequalities that may exist among the three middle quintiles and takes the other three middle quintiles into consideration. This creates a limited perspective on the inequality of the entire population. ³⁹ By contrast, the concentration index provides a better measure of the magnitude of socioeconomic inequality across the five quintiles. This index reflects the extent to which a health indicator is concentrated among the better or worse off. ^{38 , 39} A lower socioeconomic status health is lagging behind the other groups and is considered an unbalanced advantage while higher economic status health is better off. The index is between −1 & +1. The higher the inequality, the wider the gap. A concentration index of 0 is a sign of no socioeconomic inequality. The concentration index (C) of grouped data is given an equation as follows: C = ( P 1 L 2 − P 2 L 1 ) + ( P 2 L 3 − P 3 L 2 ) + ⋯ + ( P T − 1 L T − P T L T − 1 ) where T represents the total number of wealth groups, as is the case here, five. Lt refers to the associated Lorenz curve ordinate and Pt to the wealth-ranking cumulative percentage of the population divided in groups. This method captures the distribution of health measures within and across different social and economic strata, and offers a refined measurement of the trend of inequalities. ³⁸ The wealth index was used to rank individuals from the poorest to the wealthiest so as to create the concentration curves. The cumulative share of the population was plotted on the x-axis, while the cumulative share of each health indicator calculated using survey sampling weights was plotted on the y-axis. Each curve was compared with the 45-degree line of equality to assess socioeconomic inequality. For each indicator, curves for 2013 and 2023 were displayed on the same graph to allow comparison of changes in inequality over time. In 2018, no socioeconomic data were available, therefore, no analysis was performed for that year.

Ethical review board approval was obtained from the Research and Community Engagement Ethical Committee, Faculty of Public Health, Universitas Indonesia, with reference number 562/UN2.F10.D11/PPM.00.02/2025. Written informed consent was obtained from all participants prior to their participation in the study.

Table 1 presented the total number of respondents included in this analysis. A total of 44,174 children between ages 12 to 23 months, 242,768 children under-five years old, 393,907 women, and 464,253 men aged 15 to 49 were interviewed. Additionally, data from 104,714 live births that occurred in the five years preceding the survey were considered, based on paired data from mothers and their children.

Table 2 summarizes rural-urban ratios, wealth quintile ratios, and concentration indices for MCH outcomes, revealing overall improvement alongside persistent socioeconomic gradients. In general, nearly all measured indicators showed a marked improvement over the ten-year period. However, one notable exception is the increase in the proportion of mothers experiencing postpartum complications, which rose from 6.2% to 12% in rural areas and from 5.4% to 19.7% in urban areas. By contrast, other indicators seem to follow a similar pattern of decrease in proportion between 2013 and 2023. The infant mortality rate in the lowest quintile decreased by 5 deaths per 1,000 live births, compared with a reduction of 1 deaths per 1,000 live births in the richest quintile. Similarly, stunting declined by 14.5 percentage points in the lowest quintile, whereas the richest quintile experienced a 11.6 percentage point reduction. However, both the lowest and richest quintile groups showed an increase in the proportion of mothers with postpartum complications. In the richest quintile, the percentage of mothers with postpartum complications rose sharply from 5.3% in 2013 to 19.3% in 2023. There were inequalities that favored the rich over the poor, as seen by the quintile ratios of almost all variables being more than one. Additionally, for the majority of indices, the quintile ratios increased with time, indicating a widening wealth disparity. The greatest ratios, and therefore the largest disparities, were observed in underweight among children, the prevalence of malaria among children under-five, and mothers with low BMI. The underweight prevalence among children in the lowest quintile, for instance, was 2.20 times greater in 2023 and 2.07 times higher in 2013 compared to the wealthiest. Additionally, indicators for the prevalence of ARI and diarrhea in children under-five showed a similar pattern. The largest increases in ratios were observed for stunting among children under-five and mothers with low BMI, which stayed fairly high, rising from 1.74 to 2.14 and from 2.58 to 3.32, respectively. Almost all the poor were more likely to suffer from sickness, as seen by the negative concentration indices for all health outcome variables. Concentration indices that increased were those for the prevalence of ARI, the prevalence of malaria, and low maternal BMI, whereas concentration indices that decreased were those for the infant mortality rate, under-five mortality rate, prevalence of stunting, prevalence of underweight, maternal postpartum complications, and prevalence of diarrhea among children under-five. The childhood malaria concentration index became more negative, changing from −0.49 in 2013 to −0.57 in 2023. The upward trend in concentration indices indicates that health disparities are expanding, further marginalizing individuals from lower socioeconomic backgrounds.

Table 3 summarizes the rural/urban ratios and quintile ratios and concentration index of various health service indicators. Evidence shows a general increase in the utilization of health services across all wealth quintiles. However, there were exceptions to this positive trend. Specifically, the utilization of services such as full immunization rates among children aged 12–23 months and antenatal visits to trained personnel saw a decline over the ten-year period. Conversely, the rate of full postpartum visits showed stagnation. Similar to health status metrics, health service utilization metrics also indicated that the poorest group made greater progress more than the richest over the ten-year period. For example, the percentage of births assisted by skilled birth attendants in the lowest quintile increased by 19.8 percentage points, compared to an increase of only 2.3 percentage points in the richest quintile. Furthermore, similar patterns were observed in several indicators from 2013 to 2023. Conversely, concentration indices that decreased included those for births assisted by skilled birth attendants, births delivered in health facilities, children under-five years with diarrhea who received medical treatment, contraceptive prevalence rate, and complete postpartum visits. The concentration index for LLINs use in children under-five years became more positive, changing from −0.12 in 2013 to 0.11 in 2023. These rising concentration indices suggested that inequality gaps widened, to the disadvantage of individuals with low socioeconomic status.

Figure 1 plotted concentration curves for the change in each health indicator between 2013 and 2023, with the 45° dashed line representing an equal distribution of improvements across wealth quintiles. Overall, most indicators showed curves that lay below the line of equality for the lower quintiles, indicating that much of the improvement in health status was captured by the better-off. This pattern was especially marked for under-five mortality (U5MR), infant mortality, prevalence of underweight, prevalence of malaria, and low maternal BMI. Conversely, the curves for stunting and, to a lesser extent, prevalance of ARI and maternal postpartum complications lay closer to or slightly above the equality line, suggesting relatively more pro-poor or more evenly shared improvements for these outcomes.

Figure 2 presents data from 2013 to 2023, showing overall improvements in health service utilization. However, these gains were more pronounced among wealthier populations. The concentration curves indicate that services such as antenatal visits, skilled birth attendance, and immunization became more concentrated in the higher quintiles, widening the gap between the rich and poor. In contrast, services such as births delivered in health facilities and children receiving treatment for diarrhea showed a more equitable distribution. Despite progress in expanding access to basic services, the data highlights increasing inequality in healthcare, with wealthier groups benefiting more from the improvements.

Regional health inequalities in Indonesia: Figure 3 and Figure 4 illustrated concentration curve that highlighted regional disparities in stunting rates and births assisted by skilled birth attendants in Indonesia from 2013 to 2023. Regions such as Java show lower stunting rates, while others, such as Papua and Maluku, exhibit higher rates, particularly among the lower population quintiles. Additionally, the poorest quintiles contribute very little to total SBA coverage, while access increases sharply in the middle quintiles and becomes almost universal among the richest groups. Maluku and Papua show the greatest inequality, with the curve far below the equality line in the lower quintiles.

The wealthiest quintiles continue to have better health outcomes and more service consumption, according to our analysis of data from 2013 to 2023. This is known as the “inverse care law.” Importantly, the increasing concentration indices for a number of indicators point to the entrenchment of socioeconomic differences, especially in preventive care and nutrition. Using national averages could conceal serious vulnerabilities among the impoverished and those living in distant areas. These growing disparities jeopardize Indonesia’s capacity to meet its SDG targets and guarantee the long-term viability of its healthcare system if they are not addressed. There is an indication that the available policies have not adequately met the needs of the most vulnerable groups. The sharp rises in maternal and postpartum complications in all quintiles, especially in the urban quintiles, point to the urgent need for improving the quality of clinical health services and the emergency obstetric care systems. While universal MCH coverage has expanded, these findings suggest that universalist approaches alone are insufficient. Indonesian health policy has to shift from emphasizing universal coverage to equitable quality. Future policies should incorporate explicit equity targets, prioritize underserved regions such as Papua and Maluku, and strengthen the quality of primary and emergency obstetric care, particularly through improved specialist distribution and the reduction of non-financial barriers.