Changes in the Distribution of General Physicians in Korea During the 2024–2025 Medical Crisis: A Descriptive Analysis Using Nationwide HIRA Data

Abstract

In February 2024, approximately 90% of resident physicians in Republic of Korea (= South Korea) collectively resigned from training hospitals in protest of the government’s plan to reform healthcare system, creating an abrupt and nationwide supply shock in the physician workforce. Using publicly available Health Insurance Review and Assessment Service data across eight timepoints (December 2022–March 2026), we describe how this shock reshaped the distribution of general physicians—licensed physicians without specialty board certification—across institution types, regions, urban–rural areas, and clinic specialties. Between December 2023 and December 2024, trainees at training hospitals declined from 13,001 to 1,208 (−90.7%), while general physicians at non-training institutions rose from 5,557 to 9,157 (+64.8%). The increase was concentrated in urban areas, capital-area clinics, and a small set of specialties dominated by dermatology, internal medicine, and aesthetic-related fields; per-population rates confirmed that Gangwon showed the smallest reallocation. Trainee numbers partially recovered by March 2026. Approximately 8,200 (≈ 69%) of the displaced trainees were not accounted for by the concurrent rise in general physicians at non-training institutions. This study is descriptive and does not itself perform causal inference; rather, it documents how a sudden physician supply expansion was absorbed by the existing healthcare market and provides an empirical foundation for future causal analyses treating this shock as a natural experiment.

Graphical Abstract

Republic of Korea (= South Korea) has long debated whether physician shortages, particularly in essential specialties and rural areas, could be resolved by expanding the supply of doctors.1 The country has the lowest physician-to-population ratio among high-income OECD countries, with sharp regional disparities concentrated in non-capital areas.1 In February 2024, the government announced a healthcare reform that included a plan to increase the annual medical school admission quota from 3,058 to 5,058, triggering an unprecedented protest. More than 90% of resident physicians across the country submitted resignations, and the collective walkout persisted for the remainder of 2024 and much of 2025.1,2,3,4 This left teaching hospitals operating with a fraction of their usual trainee workforce while, crucially, the broader market for physician labor became available to these young doctors outside of formal training programs. This situation, although it arose from a healthcare labor dispute, inadvertently created a natural experiment: a sudden increase in the number of physicians available to non-training institutions without any corresponding expansion of healthcare infrastructure. While previous studies have described the consequences for hospital utilization, emergency care, and academic output,5,6,7 evidence on the basic question of where these displaced physicians went—across institution types, regions, and specialties—remains limited. We therefore provide a descriptive nationwide analysis of how the 2024–2025 medical crisis reshaped the distribution of general physicians in Korea, defined as licensed physicians without specialty board certification.

We used the Health Insurance Review and Assessment (HIRA) Service Medical Institution Information dataset, which is publicly released on a quarterly basis and contains institution-level counts of physicians by training status.8,9 Data were obtained for eight timepoints spanning pre-crisis to recovery: 2022.12, 2023.12, 2024.12, 2025.3, 2025.6, 2025.9, 2025.12, and 2026.3. “Trainees” were defined as the sum of interns and residents. “General physicians” were defined as licensed physicians who had neither completed specialty board certification nor were currently enrolled in specialty training (in contrast to board-certified specialists, who had completed residency and passed the Korean specialty board examination). Training hospitals were operationally defined as institutions that had at least one trainee at the December 2023 baseline among hospital types eligible for graduate medical education (advanced general, general, standard, or psychiatric hospitals). We classified all healthcare institutions into eight mutually exclusive groups: training hospital, non-training general hospital, non-training hospital, clinic, long-term care hospital, non-training psychiatric hospital, public health institution, and other. We stratified analyses by region (six regions defined as Capital, Chungcheong, Honam, Yeongnam, Gangwon, and Jeju) and by urbanicity, in which subdistricts ending with “si” or “gu” were classified as urban and those ending with “gun” as rural. For each region we obtained the corresponding registered population as of December 2023 from the Ministry of the Interior and Safety Resident Registration Statistics to compute per-100,000 population rates (Supplementary Table 1). For clinics, we assigned a representative specialty per institution using the HIRA Specialty Information dataset: if a single specialty was declared it was assigned as representative; if multiple specialties were declared with at least one specialist, the specialty with the greatest number of specialists was assigned; if multiple specialties were declared without any specialist, the clinic was classified as a “non-specialist multi-specialty clinic.” All analyses were descriptive; no statistical inference was performed. The study used publicly available, de-identified institution-level data and was exempt from institutional review board review.

Fig. 1 shows workforce changes by institution group (Supplementary Table 2). Trainee numbers at training hospitals decreased from 13,001 at the December 2023 baseline to 1,208 at December 2024 (−90.7%), with only a gradual increase through June 2025 (2,274) before a larger recovery by March 2026 (11,520, or 88.6% of baseline). During the same period, general physicians at non-training institutions rose markedly, from 5,557 to 9,157 (+64.8%). Clinics showed the largest absolute increase (3,915 → 6,436 at December 2024, +64.4%, with a further rise to 7,098 at June 2025 before declining as trainees returned). Non-training general hospitals, non-training standard hospitals, and long-term care hospitals all showed similar trajectories with 2- to 4-fold increases from baseline. Public health institutions declined steadily, consistent with the wind-down of the mandatory public health doctor system. Training hospitals themselves also showed a transient increase in general physicians (215 → 497 at December 2024), suggesting some resigned trainees may have been re-employed by their prior institution as licensed general physicians outside the training track.

Fig. 1. Physician workforce changes during the 2024–2025 medical crisis. (A) Number of trainees (interns + residents) at training hospitals across eight timepoints (2022.12–2026.3). (B) Number of general physicians by institution group, shown as independent vertical scales for each group. Training hospitals were operationally defined as institutions with at least one trainee at 2023.12 among types eligible for graduate medical education. “Non-training” institutions include all others.

LTC = long-term care.

Fig. 2 presents regional and urban–rural patterns (Supplementary Tables 3 and 4). Trainee decreases at training hospitals and general-physician increases at non-training institutions were both concentrated in the Capital region: trainees there fell from 8,372 to 788 at December 2024 (a loss of 7,584, accounting for 64.3% of the national decrease), while general physicians at non-training institutions in the Capital area rose from 2,577 to 4,720 (a gain of 2,143, accounting for 59.5% of the national increase). All other regions (Chungcheong, Honam, Yeongnam, Gangwon, Jeju) showed the same directional patterns but at much smaller magnitudes. Because absolute counts partly reflect underlying population size, we also examined general-physician increases per 100,000 population using December 2023 registered population as the denominator. The Capital region remained highest at 8.24 per 100,000, followed by Jeju (6.27), Yeongnam (5.89), Honam (5.84), and Chungcheong (5.78); Gangwon was the clear outlier at 3.25 per 100,000—approximately 40–55% of the level observed in every other non-capital region (Supplementary Table 1). The urban–rural comparison (panels C and D) revealed that rural areas (“gun” subdistricts) were almost entirely uninvolved: rural trainee numbers dropped from 96 to 17 in absolute terms, and rural general-physician numbers at non-training institutions remained essentially flat (842 → 887 at December 2024). In other words, even after normalizing for population size, the redistribution was almost exclusively an urban, capital-region phenomenon, and the underserved rural areas that the medical school quota expansion was nominally intended to benefit experienced no meaningful reallocation of physician labor.

Fig. 2. Regional and urban–rural patterns of physician workforce changes. (A) Number of trainees at training hospitals stratified by six regions. (B) Number of general physicians at nontraining institutions stratified by six regions. (C) Number of trainees stratified by urbanicity. (D) Number of general physicians at non-training institutions stratified by urbanicity.

Urban = subdistricts ending in “si” or “gu”, Rural = subdistricts ending in “gun.”

Fig. 3 shows the distribution of general-physician increases across clinic representative specialties (Supplementary Tables 5 and 6). Between December 2023 and December 2024, the three largest absolute increases were in dermatology (+480), non-specialist multi-specialty clinics (+432), and internal medicine (+393). Substantial increases were also observed in orthopedics (+228), family medicine (+140), plastic surgery (+122), and otolaryngology (+111). By June 2025, the peak of the crisis, dermatology had increased by +714 and orthopedics by +279 compared with baseline. This pattern indicates that the redistribution was not uniform across specialty fields: while primary care fields such as internal medicine and family medicine absorbed a substantial portion of displaced physicians, aesthetic- and cosmetic-oriented specialties such as dermatology and plastic surgery also grew disproportionately. Notably, as trainees returned through September 2025 and beyond, general-physician counts declined correspondingly in most specialties, although dermatology and non-specialist multi-specialty clinics retained a substantial portion of their crisis-era growth.

Fig. 3. Changes in GP count by clinic representative specialty, top 15 specialties by 2024.12 absolute increase. Values shown are the difference in general-physician count at each timepoint relative to the 2023.12 baseline. Color intensity was capped at 80% of the maximum observed change to improve discrimination among non-peak values; actual numerical values are printed on each cell. “Non-specialist multi-specialty clinic” refers to clinics declaring multiple specialties without any board-certified specialist.

GP = general-physician, OBGYN = obstetrics & gynecology.

These descriptive findings should be interpreted with several limitations in mind. First, HIRA counts “licensed general physicians” rather than individually identifiable physicians, and we therefore cannot directly confirm that the observed increase represents resigned trainees; nonetheless, the synchronous timing and magnitude of the shifts strongly suggest this link. Second, the increase in general physicians at non-training institutions (+3,600) accounts for only approximately 31% of the observed trainee decrease (−11,793); the remaining ~8,200 physicians (approximately 69% of the decrease) were not captured by the healthcare-market institutions covered in HIRA. Plausible destinations include extended unpaid leaves of absence from training programs, temporary attrition from clinical practice, mandatory or alternative public-service assignments, exit to research or non-clinical positions, and delayed re-entry into the labor market as trainees returned in late 2025. This unaccounted-for gap represents a substantial reservoir of workforce uncertainty and warrants dedicated investigation using individual-level licensing or employment data. Third, institution-level specialty declarations do not capture the actual scope of practice: a clinic declaring “dermatology” may provide general outpatient care, and non-specialist multi-specialty clinics may perform a range of services, including unreimbursed aesthetic procedures. Fourth, HIRA data are restricted to health insurance reimbursable institutions and exclude military and correctional facilities. Fifth, non-training general hospitals may have absorbed excess patient volume from tertiary hospitals, and part of the observed general-physician increase in that group may reflect related staffing changes rather than displaced trainees alone. Finally, the December 2023 baseline is a single time point, and some seasonal variation cannot be fully excluded. Despite these caveats, the present analysis provides, to our knowledge, the first nationwide descriptive picture of how Korea’s physician workforce was redistributed during the 2024–2025 crisis. The pattern—concentrated in capital-area clinics and in specialties spanning both primary care and aesthetic fields, with rural areas almost entirely unchanged even after population normalization—offers empirical evidence relevant to ongoing policy debates about whether expanding physician supply alone, in the absence of structural reforms or geographic incentives, can address regional and specialty imbalances.10,11

In conclusion, beyond its immediate policy relevance, the 2024–2025 Korean medical crisis represents a rare natural experiment in physician labor economics. The abrupt, large-scale, and geographically uneven displacement of young physicians offers an unusual empirical setting in which to inform future causal analyses of how physician workforce composition shapes outcomes that are typically confounded in observational data. These include patient-level outcomes such as unmet medical needs, continuity of care, and access in underserved regions; institutional outcomes such as hospital revenue, service mix, and the substitutability of general physicians for board-certified specialists; and market-level outcomes such as clinic entry, specialty choice, and the pricing of non-reimbursable services. The present study, however, is descriptive in nature and does not itself perform causal inference; rather, it establishes the baseline empirical patterns on which such future work can build. Linking the quarterly workforce shifts documented here to administrative claims, community health survey data, and institutional financial records could yield evidence on how physician supply shapes health system performance that would be difficult to obtain under ordinary conditions. As Korea and other health systems continue to debate the levers of physician workforce policy, making full analytic use of this unplanned experiment should be a priority for future research.

SUPPLEMENTARY MATERIALS

Supplementary Table 1

Regional general physician (GP) increases between December 2023 and December 2024, with population normalization

Supplementary Table 2

Trainees and general physicians by institution group, December 2022–March 2026

Supplementary Table 3

Trainees and general physicians by region, December 2022–March 2026

Supplementary Table 4

Trainees and general physicians by urbanicity, December 2022–March 2026

Supplementary Table 5

General physicians by clinic representative specialty, December 2022–March 2026

Supplementary Table 6

Number of clinics by representative specialty, December 2022–March 2026