Stroke team workforce maldistribution in indonesia: a burden-based policy analysis

Abstract

Background

Stroke is Indonesia's leading cause of disability, yet workforce planning relies on population-based norms, risking territorial inequities in access to essential services.

Objective

To examine the ecological alignment between provincial stroke burden and key stroke team workforce availability, and to test the robustness of a proposed burden-based planning indicator.

Methods

Cross-sectional ecological policy analysis across 34 provinces combining Global Burden of Disease Study 2023 DALYs with 2025 administrative data on neurologists, neurosurgeons, and physiotherapists. Burden-to-workforce ratios assessed service pressure; robustness was tested against an alternative indicator specification, alternative statistical classification methods, and formal comparison with population-based fulfillment metrics.

Results

Substantial geographic mismatches exist between need and availability. Eight provinces (23.5%) were classified High Pressure for each cadre (neurologists, neurosurgeons, physiotherapists) individually, while the overall bottleneck classification (worst cadre per province) flagged 11 provinces (32.4%) as High Pressure. Four provinces (North Kalimantan, West Sulawesi, Gorontalo, West Papua) reported zero practicing neurosurgeons. Provinces with comparable stroke burden displayed up to 26.3-fold differences in neurologist workforce capacity. Burden-based classification correlated weakly with population-based fulfillment metrics (Spearman's ρ = 0.35, p = 0.044) and showed only 41.2% categorical agreement, indicating the two approaches frequently identify different priority provinces.

Conclusions

Indonesia's stroke workforce distribution shows a structural, ecologically observed mismatch with subnational disease burden. Cautious integration of burden-based indicators such as DALYs alongside conventional population-based metrics may support more equitable workforce planning; prospective validation and multi-country testing are warranted before broader adoption.

Highlights

• •Introduces a Burden-to-Workforce Ratio (BWR) linking stroke DALYs to workforce supply.
• •Neurologist workforce differs up to 26.3-fold between provinces with similar burden.
• •Four provinces report zero practicing neurosurgeons for stroke care.
• •Burden- and population-based classifications agree on only 41.2% of provinces.
• •Cautious integration of DALYs may complement population-based workforce planning.

Research in Context:

• •What is already known about the topic? Health workforce planning typically relies on population-based norms, ignoring subnational epidemiological variation. Evidence linking workforce distribution directly to disease burden (DALYs) remains scarce, particularly in transitioning economies, and no burden-to-workforce indicator has yet been validated or stress-tested for a specific tracer condition in an LMIC setting.
• •What does this study add to the literature? This study introduces a Burden-to-Workforce Ratio (BWR) across 34 provinces, shows that eight provinces face high neurosurgical service pressure even though the aggregate national neurologist density figure looks adequate in isolation, and, going beyond a purely descriptive account, formally tests the robustness of the indicator and benchmarks it quantitatively against conventional population-based planning.
• •What are the policy implications? Policymakers should consider complementing, rather than replacing, supply-driven workforce planning with burden-based evidence. Integrating DALYs into workforce allocation, alongside conventional standards, may support more equitable access; broader adoption should proceed through phased piloting given the descriptive, single-country evidence base presented here.

1. Introduction

Indonesia is undergoing a rapid epidemiological transition, where non-communicable diseases (NCDs) have replaced infectious diseases as primary causes of morbidity and mortality 1., [2]. Stroke is the leading cause of death in the country [3], and according to Global Burden of Disease (GBD) 2023 estimates, it accounts for approximately 7.1% of Indonesia's total disease burden, equivalent to roughly 6.7 million DALYs [4]. This burden places substantial pressure on the national health system, particularly the National Health Insurance (JKN), where stroke is reported to rank among the costliest catastrophic conditions [5].

Health workforce planning in Indonesia and many LMICs, however, continues to rely largely on population-based norms (e.g., neurologists per 1000 population) [6], [7], [8]. This approach is administratively simple, but it fails to account for substantial subnational variation in disease epidemiology. It assumes health needs are uniform across the archipelago, even though stroke burden is often highest in regions with the least capacity to manage it. Effective stroke care requires a coordinated stroke team spanning the continuum from acute diagnosis to long-term rehabilitation [9].

Composite indicators that aggregate burden-of-disease metrics with workforce counts, however, require explicit theoretical justification and empirical stress-testing before they can be recommended as planning tools. To date, no burden-to-workforce indicator of this kind has been benchmarked against alternative specifications, alternative classification methods, or conventional population-based standards for a single tracer condition in Indonesia or a comparable LMIC setting. This study addresses that evidence gap.

Addressing current maldistribution requires moving from supply-driven planning toward a need-based approach informed by epidemiological evidence: integrating DALYs into workforce distribution offers a complementary method to identify service gaps and allocate resources where they are most needed [9]. Empirical analyses linking workforce distribution directly to subnational disease burden remain scarce in the Indonesian context. Using stroke as a tracer condition, we quantify the supply-need mismatch, test the robustness of the proposed indicator against alternative specifications and classification approaches, benchmark it against conventional population-based planning, and propose a policy framework for more equitable, evidence-based workforce planning to inform the National Medium-Term Development Plan.

2. Methods

2.1. Study design and setting

We conducted a cross-sectional analysis to assess alignment between stroke disease burden and health workforce availability across Indonesia. The unit of analysis was the province (n = 34). Indonesia currently has 38 provinces; however, four new autonomous provinces formed in the Papua region after 2022 were excluded due to unavailability of disaggregated longitudinal burden of disease data in the GBD 2023 dataset. This study integrates epidemiological data representing health needs with administrative workforce data representing supply.

As an ecological (province-level) study, all associations reported below describe area-level patterns. They should not be interpreted as evidence of individual-level relationships, nor as establishing that workforce supply causally determines patient-level access or outcomes. We elaborate on this distinction in Section 4.1 and Section 5.

2.2. Data sources

Epidemiological estimates were obtained from the Global Burden of Disease (GBD) Study 2023, using stroke (ICD-10 I60-I69) as the tracer condition. The primary metric extracted was Disability-Adjusted Life Years (DALYs), expressed as absolute numbers and rates per 100,000 population by province.

Health workforce data were extracted from the National Health Workforce Registry (SISDMK) of the Ministry of Health, accessed in 2025. SISDMK is a mandatory, real-time web-based information system recording licensure and practice location of all health professionals in Indonesia. We extracted total counts of active personnel for three key healthcare professionals representing the stroke care continuum: (1) Neurologists (diagnosis and medical management); (2) Neurosurgeons (acute surgical intervention); and (3) Physiotherapists (post-stroke rehabilitation). Demographic data for population denominators were sourced from Statistics Indonesia (BPS).

Both existing workforce counts (Table 1) and population-based demand targets (Table 2) were sourced from Indonesia's National Planning for Medical and Health Personnel: Regional-Based Projection of Needs 2023–2032 (Ministry of Health, 2026) [11], which applies a national supply–demand–needs model to the SISDMK/SIP licensure registry: demand is estimated from procedure volumes and standard completion times, adjusted for JKN coverage and utilization; supply is projected from active-license headcounts plus incoming graduates, net of fixed annual attrition (i.e. 1.5% for most cadres, 3.5% for nurses and midwives).

Table 1.

Provincial stroke-specific Burden-to-Workforce Ratios and service pressure classification.

No	Province	Population (2025)	DALYs Stroke (2023)	Existing Workforce			Stroke-specific Burden-to-Workforce Ratio (BWR) by workforce						Overall Stroke Service Pressure Category⁎⁎
							BWR			Quartile Category
				Neurologists	Neurosurgeons	Physiotherapists	Neurologists	Neurosurgeons	Physiotherapists	Neurologists	Neurosurgeons	Physiotherapists
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)
1	Aceh	5,626,000	3045	74	9	537	41.1	338.3	5.7	Moderate	Moderate	Moderate	Moderate
2	North Sumatra	15,785,800	2938	109	24	559	27.0	122.4	5.3	Low	Low	Low	Low
3	West Sumatra	5,914,300	3534	51	6	369	69.3	589.0	9.6	Moderate	Moderate	Moderate	Moderate
4	Riau	6,811,200	2781	54	6	473	51.5	463.5	5.9	Moderate	Moderate	Moderate	Moderate
5	Jambi	3,768,500	3134	24	2	231	130.6	1567.0	13.6	Moderate	Moderate	Moderate	Moderate
6	South Sumatra	8,928,500	3439	69	6	414	49.8	573.2	8.3	Moderate	Moderate	Moderate	Moderate
7	Bengkulu	2,138,000	3660	12	2	74	305.0	1830.0	49.5	High	Moderate	High	High
8	Lampung	9,522,900	3360	46	6	308	73.0	560.0	10.9	Moderate	Moderate	Moderate	Moderate
9	Bangka Belitung Islands	1,550,800	3509	16	2	106	219.3	1754.5	33.1	High	Moderate	Moderate	High
10	Riau Islands	2,213,500	2326	24	7	160	96.9	332.3	14.5	Moderate	Moderate	Moderate	Moderate
11	Jakarta	10,678,000	2895	320	86	1766	9.0	33.7	1.6	Low	Low	Low	Low
12	West Java	50,759,000	3287	346	75	2094	9.5	43.8	1.6	Low	Low	Low	Low
13	Central Java	38,233,900	4141	316	38	2262	13.1	109.0	1.8	Low	Low	Low	Low
14	Yogyakarta	3,781,500	3624	76	7	534	47.7	517.7	6.8	Moderate	Moderate	Moderate	Moderate
15	East Java	42,089,300	4188	373	63	1680	11.2	66.5	2.5	Low	Low	Low	Low
16	Banten	12,537,400	3245	110	35	712	29.5	92.7	4.6	Low	Low	Low	Low
17	Bali	4,461,300	2835	97	13	570	29.2	218.1	5.0	Low	Low	Low	Low
18	West Nusa Tenggara	5,731,100	3776	22	4	299	171.6	944.0	12.6	Moderate	Moderate	Moderate	Moderate
19	East Nusa Tenggara	5,742,600	3176	19	1	209	167.2	3176.0	15.2	Moderate	High	Moderate	High
20	West Kalimantan	5,766,000	3256	32	6	215	101.8	542.7	15.1	Moderate	Moderate	Moderate	Moderate
21	Central Kalimantan	2,845,000	2448	19	1	131	128.8	2448.0	18.7	Moderate	High	Moderate	High
22	South Kalimantan	4,323,300	5183	40	5	288	129.6	1036.6	18.0	Moderate	Moderate	Moderate	Moderate
23	East Kalimantan	4,267,600	3916	53	14	382	73.9	279.7	10.3	Moderate	Moderate	Moderate	Moderate
24	North Kalimantan	749,400	3259	10	0⁎	56	325.9	3259.0	58.2	High	High	High	High
25	North Sulawesi	2,721,400	4014	31	4	208	129.5	1003.5	19.3	Moderate	Moderate	Moderate	Moderate
26	Central Sulawesi	3,156,100	3383	27	2	139	125.3	1691.5	24.3	Moderate	Moderate	Moderate	Moderate
27	South Sulawesi	9,563,100	3168	114	14	682	27.8	226.3	4.6	Low	Low	Low	Low
28	Southeast Sulawesi	2,836,700	3345	22	3	169	152.0	1115.0	19.8	Moderate	Moderate	Moderate	Moderate
29	Gorontalo	1242,200	3706	10	0⁎	40	370.6	3706.0	92.7	High	High	High	High
30	West Sulawesi	1,525,300	2794	7	0⁎	62	399.1	2794.0	45.1	High	High	High	High
31	Maluku	1,970,600	4136	12	2	69	344.7	2068.0	59.9	High	High	High	High
32	North Maluku	1,373,800	2380	8	1	71	297.5	2380.0	33.5	High	High	High	High
33	Papua	1,073,600	3892	20	6	102	194.6	648.7	38.2	Moderate	Moderate	High	High
34	West Papua	587,600	2681	10	0⁎	45	268.1	2681.0	59.6	High	High	High	High

Note: BWR (Burden-to-Workforce Ratio) represents the relative stroke burden per available health worker. BWR values were categorized into Low, Moderate, and High using quartile-derived thresholds for each health workforce group. For neurologists, BWR values of ≤ 38.236 were classified as Low, >38.236–≤200.778 as Moderate, and > 200.778 as High. For neurosurgeons, values of ≤ 266.357 were classified as Low, >266.357–≤1889.5 as Moderate, and > 1889.5 as High. For physiotherapists, values of ≤ 5.567 were classified as Low, >5.567–≤33.200 as Moderate, and > 33.200 as High. Higher BWR values indicate a greater stroke burden relative to the available health workforce, reflecting greater potential workforce pressure.

^⁎For computational purposes, a denominator of 1 was assigned when the actual workforce was zero. The assigned value was used only to calculate BWR and does not represent actual workforce availability.

^⁎⁎The Overall Stroke Service Pressure Category was determined based on the highest BWR category among neurologists, neurosurgeons, and physiotherapists. A province was classified as High if at least one workforce group had a High BWR, Moderate if no workforce group had a High BWR but at least one had a Moderate BWR, and Low if all three workforce groups had Low BWR.

Table 2.

Stroke team workforce fulfillment based on population-based standards.

No	Province	Population 2025	Existing Workforce			Demand			Fulfillment Percentage (%)			Overall Stroke Service Pressure Category⁎
			Neuro-logists	Neuro-surgeons	Physio-therapists	Neuro-logists	Neuro-surgeons	Physio-therapists	Neuro-logists	Neuro-surgeons	Physio-therapists
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
1	Aceh	5,626,000	74	9	537	32	17	220	231.25	52.94	244.09	Moderate
2	North Sumatra	15,785,800	109	24	559	114	37	1383	95.61	64.86	40.42	High
3	West Sumatra	5,914,300	51	6	369	70	38	1319	72.86	15.79	27.98	High
4	Riau	6,811,200	54	6	473	65	24	1162	83.08	25.00	40.71	High
5	Jambi	3,768,500	24	2	231	25	10	297	96.00	20.00	77.78	High
6	South Sumatra	8,928,500	69	6	414	119	28	795	57.98	21.43	52.08	High
7	Bengkulu	2,138,000	12	2	74	9	8	126	133.33	25.00	58.73	High
8	Lampung	9,522,900	46	6	308	70	31	1242	65.71	19.35	24.80	High
9	Bangka Belitung Islands	1,550,800	16	2	106	9	4	225	177.78	50.00	47.11	High
10	Riau Islands	2,213,500	24	7	160	33	10	309	72.73	70.00	51.78	Moderate
11	Jakarta	10,678,000	320	86	1766	258	56	4142	124.03	153.57	42.64	High
12	West Java	50,759,000	346	75	2094	615	131	6260	56.26	57.25	33.45	High
13	Central Java	38,233,900	316	38	2262	373	104	7846	84.72	36.54	28.83	High
14	Yogyakarta	3,781,500	76	7	534	54	11	792	140.74	63.64	67.42	Moderate
15	East Java	42,089,300	373	63	1680	314	114	5187	118.79	55.26	32.39	High
16	Banten	12,537,400	110	35	712	138	60	2177	79.71	58.33	32.71	High
17	Bali	4,461,300	97	13	570	99	29	1439	97.98	44.83	39.61	High
18	West Nusa Tenggara	5,731,100	22	4	299	73	21	760	30.14	19.05	39.34	High
19	East Nusa Tenggara	5,742,600	19	1	209	23	8	214	82.61	12.50	97.66	High
20	West Kalimantan	5,766,000	32	6	215	29	11	173	110.34	54.55	124.28	Moderate
21	Central Kalimantan	2,845,000	19	1	131	9	32	309	211.11	3.13	42.39	High
22	South Kalimantan	4,323,300	40	5	288	42	9	906	95.24	55.56	31.79	High
23	East Kalimantan	4,267,600	53	14	382	87	18	879	60.92	77.78	43.46	High
24	North Kalimantan	749,400	10	0	56	4	4	81	250.00	0.00	69.14	High
25	North Sulawesi	2,721,400	31	4	208	36	12	559	86.11	33.33	37.21	High
26	Central Sulawesi	3,156,100	27	2	139	32	9	246	84.38	22.22	56.50	High
27	South Sulawesi	9,563,100	114	14	682	176	20	1931	64.77	70.00	35.32	High
28	Southeast Sulawesi	2,836,700	22	3	169	18	9	75	122.22	33.33	225.33	High
29	Gorontalo	1242,200	10	0	40	5	3	100	200.00	0.00	40.00	High
30	West Sulawesi	1,525,300	7	0	62	5	2	92	140.00	0.00	67.39	High
31	Maluku	1,970,600	12	2	69	9	8	72	133.33	25.00	95.83	High
32	North Maluku	1,373,800	8	1	71	26	4	259	30.77	25.00	27.41	High
33	Papua	1,073,600	20	6	102	11	11	203	181.82	54.55	50.25	Moderate
34	West Papua	587,600	10	0	45	5	7	88	200.00	0.00	51.14	High

Source: Ministry of Health (MoH), 2026.

Note: Red indicates low fulfillment (0–50%); yellow indicates moderate fulfillment (51–80%); and green indicates high fulfillment (>80%).

^⁎Overall Stroke Service Pressure Category: red indicates High Stroke Service Pressure; yellow indicates Moderate Stroke Service Pressure; and green indicates Low Stroke Service Pressure.

To support replicability: province identifiers were harmonized across the GBD, SISDMK, and BPS datasets; all 34 provinces had complete, non-missing values, so no imputation was required. Descriptive statistics, BWR calculations, quartile classification (using the weighted average method), and the robustness analyses in Section 2.7 were performed using Microsoft Excel and Python 3.9+. The analytical dataset and analysis procedures are available from the corresponding author upon reasonable request.

2.3. Measures: Burden-to-workforce ratio

To quantify the mismatch between disease burden and workforce capacity, we developed the Burden-to-Workforce Ratio (BWR), an ecological indicator expressing the epidemiological load of stroke carried, on average, by each available specialist in a province:

$${\mathsf{BWR}}_{\mathsf{ip}}={\mathsf{DALYrate}}_{\mathsf{ip}}/{\mathsf{W}}_{\mathsf{ip}}$$

where DALYrate_ip is the province-level stroke DALY rate per 100,000 population (Global Burden of Disease Study 2023), and W_ip is the total count of active health workforce of a given cadre in province p. A higher BWR indicates greater service pressure, implying that the available specialists face a proportionally larger burden of disease and disability relative to the province's population.

We selected DALYs rather than single-dimension indicators (incidence, prevalence, mortality, or hospital admissions) because they integrate fatal and non-fatal outcomes, better approximating the combined clinical and rehabilitative workload generated by stroke across the full continuum of care, consistent with the tracer-condition, multi-cadre design of this study. We explicitly acknowledge that this choice embeds an assumption of rough proportionality between population-level disability burden and individual-provider workload; this assumption has not been directly validated against facility-level workload data (e.g., patient volumes, procedure counts) in Indonesia. We therefore treat BWR as a population-need proxy for service pressure, not as a validated workload or productivity metric. Comparable burden-adjusted supply concepts have been proposed in the broader health workforce planning literature [9].

BWR is calculated separately for each of the three specialists (neurologists, neurosurgeons, physiotherapists) rather than as a single pooled headcount, since these healthcare professionals differ in role, training, and workload intensity and are not clinically interchangeable. The ratio tabulated is the stroke DALY rate (per 100,000 population) divided by workforce count. The composite ‘Overall Service Pressure’ category (Table 1) is derived using a bottleneck principle: a province is classified according to its worst-performing workforce, so a shortage in any single workforce is not masked by adequacy in the others.

2.4. Service pressure classification.

To translate BWR into actionable policy categories, provinces were classified into three Service Pressure levels based on the quartile distribution of BWR scores, calculated separately per cadre. We adopted fixed quartile thresholds, rather than a pooled national or distribution-derived breakpoint, because quartiles guarantee a fixed, policy-actionable proportion of provinces (the top 25%) flagged for priority intervention each cycle, which matters where realistic allocation capacity is itself constrained. Section 2.7 formally compares this choice against z-score and Fisher–Jenks natural-breaks classification.

High Pressure (Red Zone) means provinces falling into the top quartile (>75th percentile) of BWR distribution, or provinces with zero availability of the specific specialist. Moderate Pressure (Yellow Zone) means provinces with BWR scores between the 25th and 75th percentile. Low Pressure (Green Zone) means provinces with BWR scores below the 25th percentile.

In the dataset analyzed, four provinces recorded a literal zero neurosurgeon count (North Kalimantan, West Sulawesi, Gorontalo, and West Papua) and are classified High Pressure by the explicit zero-count rule, irrespective of the quartile distribution (a denominator of 1 was substituted for BWR computation only, so the ratio reduces to the DALY rate itself; Table 1). No province recorded a zero count for the neurologist or physiotherapist cadres; the minimum nonzero neurosurgeon count elsewhere was one (North Maluku, East Nusa Tenggara, Central Kalimantan), each independently classified High Pressure through the ordinary quartile rule given the resulting extreme BWR.

2.5. Workforce fulfillment assessment (Population-based).

Secondary to burden-based analysis, we evaluated workforce quantity fulfillment against existing national population-based standards. Fulfillment was calculated as current active workforce relative to target requirements (Real Supply/Target × 100%), with targets drawn from the same National Planning for Medical and Health Personnel: Regional-Based Projection of Needs 2023–2032 [11]. Provinces were stratified into Green Zone (>80%), Yellow Zone (51–80%), and Red Zone (0–50% fulfillment), per the SISDMK dashboard [10]. As with Table 1, an Overall Priority Pressure Category was derived using the same bottleneck principle: the worst-performing cadre determines each province's overall classification (Table 2).

2.6. Ethics statement.

This study utilized publicly available, aggregated secondary data from the GBD study and national administrative dashboards. No individual patient or health worker data were accessed. Formal ethical review was waived by the institutional review board.

2.7. Robustness and sensitivity analyses.

To assess the robustness of the indicator, classification method, and validation against conventional planning, we conducted four supplementary analyses (n = 34 provinces). First, re-specifying BWR with absolute rather than rate-based DALYs and re-classifying by the same rank-based rule yielded only 41.2% agreement (14/34) with the rate-based classification, since large-population provinces shift toward higher pressure purely because of population size; we therefore retain the rate-based specification and flag this sensitivity as a limitation. Second, classification was highly robust to the choice of cut-off rule: agreement with z-score classification was 97.1% (33/34) and with Fisher–Jenks natural breaks [12] were 76.5% (26/34); we retain the quartile method because it guarantees a fixed, policy-interpretable proportion of provinces each cycle.

Third, to formally compare the discriminative properties of the two approaches, burden-based and population-based classifications were weakly correlated (Spearman's ρ = 0.35, p = 0.044) and showed only 41.2% categorical agreement, indicating the two approaches are related but frequently flag different priority provinces. Fourth, a concentration index [13] for neurologist workforce relative to province-level stroke burden was −0.07 — a near-zero national average that is compatible with severe, policy-relevant misalignment concentrated in specific outlier provinces (e.g., Maluku vs. Central Java), which is the localized inequity that province-level BWR classification, rather than a single aggregate index, is designed to detect.

3. Results.

3.1. Provincial variation in stroke burden.

Substantial heterogeneity in stroke burden was observed across 34 provinces. Stroke DALY rates ranged from 2326 (Riau Islands) to 5183 per 100,000 population (South Kalimantan). Elevated rates were not confined to provinces outside Java: South Kalimantan recorded the highest rate nationally, and East Java and Central Java, Indonesia's most populous provinces, also ranked among the highest, alongside Maluku, North Sulawesi, and East Kalimantan. DALY rates therefore do not follow a simple Java/outside-Java divide; as shown in 3.2, 3.3, it is workforce capacity, not disease burden, that is most sharply concentrated in Java and Bali, which produces the geographic inequity documented below (an ecological, province-level association).

3.2. Health workforce availability and maldistribution.

National neurologist density stood at 0.92 per 100,000 population, an aggregate figure that masks profound geographic inequities. Workforce availability remained heavily skewed toward Java and Bali: Jakarta reported the highest neurologist density (3.0 per 100,000) and Yogyakarta the second highest (2.0), whereas East Nusa Tenggara, West Nusa Tenggara, and West Sulawesi reported the lowest nationally, all below 0.5 per 100,000. Similar geographic concentration was observed for neurosurgeons and physiotherapists, leaving a service vacuum across large parts of the eastern and outer archipelago.

Central Java and Maluku share nearly identical stroke burden (4141 and 4136 DALYs per 10,000 population, respectively). Despite this epidemiological parity, their workforce capacity is disproportionate: Central Java is served by 316 neurologists, whereas Maluku relies on only 12, a 26.3-fold disparity. A similar pattern is observed in lower-burden regions; although Bali and West Sulawesi report comparable burden levels (2835 and 2794 DALYs, respectively), Bali benefits from a relative surplus of 97 neurologists compared to just 7 in West Sulawesi, a 13.9-fold disparity.

These comparisons illustrate current allocation mechanisms that appear to prioritize market density and historical supply trends over epidemiological need. As descriptive, ecological comparisons, they do not by themselves establish that workforce numbers are the sole determinant of unmet need; unmeasured factors such as referral pathways, private-sector capacity, and diagnostic infrastructure may also contribute.

3.3. Burden-to-workforce service pressure.

When DALY estimates are integrated with workforce supply, substantial mismatches emerge between epidemiological need and service capacity. Using the Burden-to-Workforce Ratio (BWR), provinces were grouped into high, moderate, and low service pressure categories (Table 1; Fig. 1). For neurological services, eight provinces (23.5%) were classified as High Pressure, with shortages most pronounced in Eastern Indonesia and parts of Sumatra. Eight provinces were categorized as Low Pressure, while the majority (18 provinces; 52.9%) fell within the Moderate category.

Fig. 1.

Equity map of the stroke Burden-to-Workforce Ratio across Indonesian provinces, based on the Overall Stroke Service Pressure Category (Table 1). Red indicates High Pressure, yellow indicates Moderate Pressure, and green indicates Low Pressure. Provinces in red combine high disease burden with low workforce density; these are the priority zones for intervention identified in Section 3.3.

A similar distribution was observed for neurosurgical services, with eight provinces classified High Pressure and eight Low Pressure by the quartile rule. Within the High-Pressure group, four provinces (North Kalimantan, West Sulawesi, Gorontalo, and West Papua) recorded zero practicing neurosurgeons and were classified High Pressure by the explicit zero-count rule (Section 2.4) rather than the quartile rule alone (a denominator of 1 was substituted for BWR computation only; Table 1, footnote). The remaining four High Pressure provinces (Maluku, East Nusa Tenggara, Central Kalimantan, North Maluku) reached High Pressure through the quartile rule based on an extreme but nonzero BWR.

Rehabilitation services demonstrated a comparable distribution, with eight provinces in High Pressure and eight in Low Pressure. The predominance of Moderate Pressure provinces indicates rehabilitation capacity remains uneven and insufficient relative to epidemiological need across the archipelago.

3.4. Population-based workforce fulfillment.

When workforce fulfillment was assessed using population-based standards (Table 2; Fig. 2), a different pattern emerged. Applying the same bottleneck principle used for Table 1 (the worst-performing cadre determines each province's Overall classification), 29 of 34 provinces (85.3%) were classified High Pressure (0–50% fulfillment in at least one cadre), the remaining five (Aceh, Riau Islands, Yogyakarta, West Kalimantan, Papua; 14.7%) were Moderate Pressure (51–80%), and none achieved Low Pressure (>80% across all three cadres). This near-universal pattern offers limited differentiation and fails to capture epidemiological heterogeneity.

Fig. 2.

Fulfillment of stroke team requirements across 34 provinces in Indonesia, based on population-based workforce standards (Table 2). Red indicates High Pressure (0–50% fulfillment) and yellow indicates Moderate Pressure (51–80% fulfillment); no province reached Low Pressure (>80% fulfillment) under this approach. Compared with Fig. 1, the burden-based classification reveals sharper workforce inequities across provinces than the population-based standard, indicating that the disease-burden approach is more sensitive to local context.

The burden-to-workforce ratio approach (Table 1) demonstrated more discriminative power in identifying priority zones than population-based standards (Table 2). Jakarta and West Java appear High Pressure under population-based metrics, driven by physiotherapist shortfalls despite neurologist and neurosurgeon fulfillment above 100% in Jakarta (42.6% physiotherapist vs. 124.0% neurologist and 153.6% neurosurgeon fulfillment; West Java: 33.5% physiotherapist fulfillment), but Low Pressure under burden-adjusted analysis, because their stroke burden is comparatively low. Population-based standards therefore risk over-prioritizing these high-population, lower-burden provinces on the basis of a single cadre's shortfall. Eastern provinces classified as High Pressure under both approaches, by contrast, face compounded challenges requiring immediate intervention. The disease-burden approach thus appears more discriminating for subnational workforce allocation; Section 3.5 reports the formal statistical comparison confirming this pattern and quantifying its limits.

3.5. Sensitivity and comparative validation analysis

Supplementary Table S1 summarizes the four robustness checks from Section 2.7. Reclassification under an alternative, absolute-DALY (population-scaled) BWR specification retained the same category for 41.2% of provinces (14/34) indicated high sensitivity to burden-indicator specification. Classification was highly robust to the choice between quartile and z-score cut-offs (97.1% agreement) and moderately robust to natural-breaks cut-offs (76.5%). Burden-based and population-based classifications were weakly correlated (Spearman's ρ = 0.35, p = 0.044), also showed 41.2% raw categorical agreement, and the national concentration index for neurologist supply relative to burden was −0.07. Robustness and sensitivity analysis results can be found in Supplementary Table S1.

4. Discussion

4.1. Structural misalignment between burden and capacity

This study provides ecological (area-level) evidence of an inverse-care pattern in Indonesia's stroke services: provinces with the highest disease burden paradoxically have the lowest capacity. Aggregate national workforce density figures can mask severe subnational inequities of this kind. Workforce planning based solely on population ratios is standard practice in Indonesia and many LMICs, and it inadvertently reinforces these disparities [6], [8]. Current allocation mechanisms may contribute to underservice of high-burden provinces, though this cross-sectional design cannot establish a direct causal pathway [14].

We emphasize that these findings are ecological associations and should not be interpreted as evidence that workforce distribution alone causes inequitable stroke outcomes, nor that increasing specialist headcount in a given province would proportionally improve outcomes there (the ecological fallacy) [15]. Numerous unmeasured determinants of stroke-care access, including service infrastructure, referral efficiency, and socioeconomic status, were not captured in this analysis; they are enumerated in Section 5, which should be read alongside this discussion. Workforce counts alone therefore represent one component, not a comprehensive measure, of provincial stroke-care capacity.

A further caveat concerns measurement uncertainty in the underlying GBD 2023 DALY estimates, which are modeled quantities with uncertainty intervals rather than direct counts; this analysis treated point estimates as fixed, which may overstate the precision of BWR-based rankings near quartile boundaries. Future iterations should obtain province-level uncertainty intervals from IHME and conduct probabilistic sensitivity analyses to assess how frequently rankings and classifications change.

The divergence between Table 1 and Table 2, already quantified in 3.4, 3.5 (85.3% vs. 32.4% High Pressure; 41.2% categorical agreement), illustrates a broader limitation of population-based planning: it offers limited province-level differentiation and can obscure the inverse-care dynamic that burden-adjusted metrics reveal.

4.2. Bridging the gap: From macro targets to micro planning

Addressing disparities requires a shift in planning methodology, moving beyond exclusive reliance on population-based ratios toward a hybrid, needs-based framework. While WHO population-based ratios remain valuable for setting aggregate national targets [16], they are inherently limited in guiding equitable subnational distribution. Effective workforce allocation demands micro-level planning approaches responsive to local epidemiological and service realities.

We propose an integrated framework combining Burden of Disease (DALYs) data with the Workload Indicators of Staffing Need (WISN) method [17]. DALYs provide macro-epidemiological signals to identify provinces with disproportionate stroke burden 4., [14], while WISN enables facility-level estimation of staffing requirements based on actual workload. Reviewers noted that integrating these two metrics requires an intermediate step, for example using province-level BWR classification to prioritize facilities for WISN assessment, rather than direct mathematical combination; operationalizing and pilot-testing this two-stage sequence is an important next step beyond the scope of the present descriptive analysis. Together, these tools bridge theoretical population-based targets and operational realities across Indonesia's 34 provinces, translating national stroke workforce ambitions into pragmatic, context-sensitive deployment strategies consistent with global recommendations [9].

4.3. Policy implications: Institutionalizing evidence-based policy advocacy

Findings offer a practical tool for national and subnational policy alignment. At the national level, the Ministry of National Development Planning and Ministry of Health may institutionalize DALY-based metrics into the Health Human Resources Long Term Plan and the National Medium-Term Development Plan (RPJMN) [18], prioritizing civil service formation allocation for High Pressure (Red Zone) provinces. This evidence base is equally relevant for calculating stroke-team production capacity from educational facilities and for building collaboration to improve stroke-team competence in high-burden provinces.

Beyond incorporating DALY metrics into national plans, concrete instruments could include: (1) differentiated financial incentives and bonded-service placement for specialists serving High Pressure (Red Zone) provinces, building on Indonesia's Nusantara Sehat and PTT-specialist programs; (2) residency and sub-specialty training quotas pre-committing graduates to named high-burden provinces; (3) pilot burden-based performance financing, weighting JKN capitation or DAK transfers by BWR classification alongside population size; and (4) operationalizing the Red/Yellow/Green classification as a living, annually updated dashboard (paralleling the existing SISDMK dashboard) to monitor whether provinces transition from Red to Yellow following intervention.

At the subnational level, Service Pressure categorization (Red, Yellow, Green zones) is an advocacy tool for local governments. Instead of political bargaining, authorities in high-burden regions can use objective epidemiological evidence to negotiate priority budget support and human resource quotas from the central government. This turns planning from a supply-driven exercise into a need-based negotiation.

4.4. Global implications

Beyond Indonesia, this study contributes to global discourse on NCD workforce planning. No universally agreed-upon benchmarks currently exist for stroke workforce density relative to disease burden. We suggest the international health workforce community consider whether disease-burden-adjusted benchmarks (complementing, not replacing, existing population-based ratios) merit further methodological development and cross-country testing, given the limitations of population-only norms documented here. This is presented as a research agenda, requiring prospective multi-country validation, rather than an imminent policy prescription.

This study offers an empirical demonstration of how epidemiological intelligence can be operationalized for subnational workforce planning. To the authors' knowledge, based on a literature review rather than a formal systematic search, this is among the first Indonesian studies to integrate subnational stroke DALYs with workforce-registry data into a burden-adjusted service pressure framework. The approach offers a replicable analytical template for other countries facing rapid epidemiological transition, though this potential should be treated as a hypothesis for future testing rather than an established conclusion.

5. Limitations

This study has several limitations that should inform how the findings and recommendations are used. First, the cross-sectional, ecological design captures a single time point and province-level aggregates; it cannot establish temporal trends, individual-level relationships, or causal pathways between workforce distribution and patient outcomes, and is subject to the ecological fallacy [15]. Second, the analysis considered only three specialist cadres and did not incorporate stroke units, emergency medical services, CT/MRI or thrombolysis/thrombectomy capacity, nursing and allied-health workforce, or geographic accessibility, all of which may substantially affect true service capacity (Section 4.1).

6. Conclusions

Indonesia's current health workforce distribution remains structurally misaligned with the epidemiological transition of stroke burden. Continued reliance on uniform population-based ratios risks perpetuating service inequities where disease burden is disproportionately high. A cautious shift toward complementing population-based planning with burden-informed indicators is therefore warranted, subject to the robustness checks and prospective validation discussed above.

Institutionalizing a hybrid allocation model that integrates national population targets with provincial DALY-to-workforce ratios and workload-based staffing assessments (WISN) offers a promising, hypothesis-generating pathway toward more equitable deployment of specialized health personnel. As Indonesia navigates its non-communicable disease transition, aligning workforce capacity with epidemiological need may support meaningful, geographically equitable improvements in stroke care, though whether such improvements occur requires prospective evaluation, since descriptive cross-sectional evidence alone cannot demonstrate downstream impact.

Declaration of generative AI and AI-assisted technologies

During the preparation of this work, the author(s) used Claude (Anthropic) to improve the language clarity of the original draft and to verify internal consistency between the narrative text, tables during revision. After using these tools, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article.

CRediT authorship contribution statement

Asnawi Abdullah: Writing – review & editing, Writing – original draft, Supervision, Resources, Methodology, Investigation, Conceptualization. Lupi Trilaksono: Writing – original draft, Visualization, Investigation, Formal analysis, Data curation, Conceptualization. Sri Wahyuni: Writing – review & editing, Investigation, Formal analysis, Data curation, Conceptualization. Hardini Kusumadewi: Writing – review & editing, Resources, Project administration, Methodology, Investigation. Agus Dwi Harso: Writing – review & editing, Visualization, Methodology, Investigation. Adinda Desyana Putri: Writing – review & editing, Project administration, Investigation, Formal analysis, Data curation. Siswanto: Writing – review & editing, Investigation.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material: Supplementary Table S1. Robustness and sensitivity analysis of province-level service-pressure classification and neurologist supply concentration.

Data availability

Stroke burden data used in this study are publicly available from the Global Burden of Disease (GBD) 2023 database via the Institute for Health Metrics and Evaluation (IHME). Workforce data were derived from the National Health Workforce Registry (SISDMK) of the Ministry of Health of Indonesia. Aggregated data supporting the findings are available in Table 1 and Table 2 within this manuscript. The derived province-level analytic dataset and analysis scripts underlying Supplementary Table S1 and Section 2.7 are available from the corresponding author upon reasonable request.