Key Points
Question
Was nationwide resignation of residents in South Korea associated with changes in the resuscitation practices for in-hospital cardiac arrest?
Findings
In this cohort study of 383 patients with 469 in-hospital cardiac arrest events at a tertiary teaching hospital in South Korea, the crisis period was associated with fewer cardiopulmonary resuscitation attempts exceeding 35 minutes and a higher hazard of termination without return of spontaneous circulation. Despite this, return of spontaneous circulation rates and overall survival outcomes were not significantly different.
Meaning
These findings suggest prolonged workforce shortage may be associated with changes in resuscitation termination decisions, supporting the need for standardized termination of resuscitation decision-making during a health care crisis.
This cohort study examines associations between a nationwide resident resignation in South Korea and differences in in-hospital cardiac arrest characteristics and resuscitation practices.
Abstract
Importance
Health care crises may be associated with changes in in-hospital cardiac arrest (IHCA) care and resuscitation practices. Whether prolonged health care workforce shortage is associated with termination of resuscitation (TOR) practices remains unclear.
Objective
To determine whether a nationwide resident resignation in South Korea was associated with differences in IHCA characteristics, resuscitation trajectories, and TOR practices.
Design, Setting, and Participants
A retrospective cohort study conducted at a single tertiary teaching hospital in South Korea of adults aged 18 years or older who underwent cardiopulmonary resuscitation (CPR) for IHCA between March 1, 2023, and February 28, 2025.
Exposure
IHCA occurring before vs during a workforce crisis precipitated by mass resident physician resignations.
Main Outcomes and Measures
Primary outcomes were return of spontaneous circulation (ROSC) and TOR without ROSC, analyzed as competing events over CPR duration. Secondary outcomes included CPR duration, 24-hour survival, 30-day survival, favorable neurological status, and survival to discharge.
Results
A total of 383 patients with 469 IHCA events were included (median [IQR] age, 69 [61-79] years; 238 [62.1%] male). Overall IHCA incidence did not differ between periods (2.30 vs 2.50 incidents per 1000 admissions; P = .36). The median CPR duration was 8 minutes in both periods. CPR exceeding 35 minutes was less frequent during vs before the crisis (12 of 180 events [6.7%] vs 38 of 289 events [13.1%]; P = .03). ROSC, neurological outcomes, and survival outcomes did not differ statistically. In adjusted competing-risk analyses, the crisis period was not associated with ROSC (adjusted hazard ratio [aHR], 0.93; 95% CI, 0.72-1.19) but was associated with higher hazard of TOR without ROSC (aHR, 1.66; 95% CI, 1.14-2.42). Standardized cumulative incidence analyses showed a higher cumulative incidence of TOR without ROSC, with an absolute risk difference of 4.9 percentage points at 35 minutes.
Conclusions and Relevance
In this retrospective cohort study, the crisis period was associated with fewer prolonged CPR attempts, with no significant differences in ROSC or survival outcomes. These findings suggest that prolonged health care workforce shortage and associated resuscitation-system adaptations may have influenced TOR practices, underscoring the need for standardized TOR decision-making.
Introduction
Effective management of in-hospital cardiac arrest (IHCA), including high-quality cardiopulmonary resuscitation (CPR), depends on rapid recognition, coordinated multidisciplinary teams, rapid response systems, and adequate hospital resources.1,2,3 These systems may be particularly vulnerable during periods of severe health care strain. Previous health care crises, such as the COVID-19 pandemic, have demonstrated how sudden workforce and resource shortages can disrupt resuscitation efforts and clinical outcomes after IHCA.4,5
Most prior analyses of IHCA during health care crises have focused on overall outcomes, such as return of spontaneous circulation (ROSC) or survival to discharge.6,7,8 Less attention has been given to whether the process of resuscitation itself changes under strained conditions.5,7,8 In particular, the decision to terminate resuscitation (TOR) is a complex clinical judgment. CPR duration and TOR may reflect not only the patient’s likelihood of recovery but also team capacity, staffing constraints, and changing thresholds for clinical futility under institutional pressures.8,9,10 While survival-based outcomes are fundamental, evaluating changes in the resuscitation process is critical to capture the qualitative associations of resuscitation care during a health care crisis.
In South Korea, policy changes in early 2024 triggered an 18-month health care workforce crisis,11,12 beginning with the resignation on February 19, 2024; by July 8, 2024, 12 600 medical residents (92%) maintained their resignations,13 with the resident workforce subsequently recovering by September 2025.14 This disruption was associated with reduced hospital utilization, constrained intensive care unit (ICU) capacity, and an increased clinical burden on the remaining staff.15,16,17 However, its associations with IHCA resuscitation practice remain unknown.
We therefore examined IHCA characteristics, resuscitation trajectories, TOR practices, and clinical outcomes before and during this prolonged health care workforce crisis at a tertiary referral teaching hospital. We hypothesized that the crisis period would be associated with differences in resuscitation practices, particularly during prolonged resuscitation, and examined whether these differences were accompanied by changes in ROSC or survival outcomes.
Methods
Study Design and Population
We conducted a retrospective cohort analysis of patients aged 18 years or older who underwent at least 1 episode of CPR at Seoul National University Hospital between March 1, 2023, and February 28, 2025. Events after a documented prearrest do-not-resuscitate order were excluded. IHCA events were identified through mandatory institutional CPR reports. The study period was divided into a precrisis period (March 2023 to February 2024) and a crisis period (March 2024 to February 2025) following the mass resignation of resident physicians. This study was approved by the institutional review board of Seoul National University Hospital (2409-032-1567), with the requirement for informed consent waived because of the retrospective study design. The study was conducted in accordance with the Declaration of Helsinki and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Changes in Staffing and Resuscitation Team Composition
The workforce crisis substantially changed both the resuscitation team and the rapid response team (RRT). Before the crisis, resuscitations were managed exclusively by a CPR team of 4 ICU-based resident physicians. During the crisis period, however, the mass resignation–induced staffing shortage necessitated their replacement with 4 board-certified faculty physicians trained in advanced cardiac life support. The RRT continued to operate on weekdays from 7 am to 7 pm with trained nurses and intensivists, but without resident participation. General ward and ICU bed capacity was partially reduced, and board-certified faculty physicians assumed direct responsibility for both routine ward care and on-call clinical coverage in place of residents.
Data Collection, Variables, and Outcomes
Baseline demographics and comorbidities were collected, and the Charlson Comorbidity Index was subsequently calculated.18 Event-specific clinical characteristics were obtained, including the location of the IHCA, RRT activation, the highest National Early Warning Score 2 (NEWS2) within 24 hours preceding the IHCA,19 concurrent acute medical conditions, and the provision of organ support therapies. Resuscitation-related characteristics were documented in general accordance with the Utstein resuscitation registry template for IHCA.20 These data comprised the presumed cause of IHCA, witnessed status, electrocardiographic monitoring, initial documented rhythm, time to initiation of resuscitation and code blue activation, epinephrine use, defibrillation, and CPR duration. In cases where extracorporeal cardiopulmonary resuscitation (ECPR) was deployed, CPR duration was defined as the time until conventional resuscitation efforts, including chest compressions, were discontinued. To account for potential shifts in inpatient severity during the crisis, the departmental case-mix index (CMI) was collected. Values reported as less than 1.0 were treated as 1.0 for continuous analyses.
Clinical outcomes included ROSC and survival metrics, which were evaluated as favorable neurological status, 24-hour survival, 30-day survival, and survival to discharge. Patients discharged alive before 30 days were considered 30-day survivors. Favorable neurological outcome was defined as maintaining a cerebral performance category (CPC) of 1 to 2 or, for those with a baseline cerebral performance category of 3 to 4, no worsening of their score by 30 days or discharge.20 Prolonged CPR was primarily defined as CPR duration greater than 35 minutes, reflecting the time point in previous literature where the time-dependent probability of survival after IHCA drops below 1%.21 Additional cutoffs of greater than 20 and 30 minutes were examined descriptively.
Statistical Analysis
For patients with multiple IHCAs during a single admission, all events were included in time-to-event analyses with adjustment for within-patient correlation, whereas baseline characteristics were analyzed per patient. Categorical and continuous variables were compared using Fisher exact test, Student t test, or the Mann-Whitney U test, as appropriate. IHCA incidence rates (per 1000 admissions) were compared using a 2-sample Poisson rate test.
Multivariable logistic regression was used to evaluate factors associated with prolonged CPR. The model adjusted for age, sex, hospital period (precrisis vs crisis), overall inpatient severity measured by CMI, Charlson Comorbidity Index, NEWS2, initial documented rhythm, location of IHCA, time from admission to IHCA, prior out-of-hospital cardiac arrest (OHCA), witnessed status, and monitored status.
Successful ROSC and TOR without ROSC were analyzed as competing events over CPR duration to characterize the resuscitation process. Unadjusted cumulative incidence functions were compared using the Gray test. After evaluating the proportional hazards assumption (eFigure 1 in Supplement 1), cause-specific Cox models were fitted to estimate the association of the crisis period with each event, adjusting for age, sex, Charlson Comorbidity Index, CMI, initial rhythm, IHCA location, NEWS2, time from admission to IHCA, and prior OHCA. To account for within-patient correlation from multiple IHCA events in the same patient, cluster-robust standard errors were calculated at the patient level in both the multivariable logistic regression and cause-specific Cox regression models. To complement the relative hazard estimates from cause-specific Cox models, the parametric G formula (regression standardization) was used to estimate absolute risk metrics, including standardized cumulative incidence and absolute risk differences for ROSC and TOR, with 95% CIs based on 500 patient-level bootstrap samples.
Subgroup analyses were conducted according to the ROSC status, prolonged CPR, and comorbidity burden assessed by the Charlson Comorbidity Index. The optimal cutoff value for the Charlson index was determined using Youden index from receiver operating characteristic analysis for survival to discharge (eFigure 2 in Supplement 1). We considered a 2-tailed P < .05 as statistically significant. All analyses were performed using R version 4.5.0 (R Project for Statistical Computing).
Results
A total of 383 patients who experienced 469 IHCA events were included, comprising 289 IHCA events from 232 patients in the precrisis period and 180 events from 151 patients during the crisis. The median (IQR) age of the patients was 69 (61-79) years, and 238 patients (62.1%) were male (Table 1). Baseline patient characteristics were generally similar between periods. However, the Charlson Comorbidity Index was lower during the crisis compared with the precrisis period (median [IQR], 5 [4-9] vs 7 [4-9]; P = .03), whereas other comorbidities and prior OHCA were comparable. In contrast, the CMI, reflecting overall inpatient severity, did not differ between periods (median [IQR], 1.15 [1.00-1.61] vs 1.17 [1.00-1.64]; P = .17). There was no significant difference in the overall IHCA incidence between the precrisis and crisis periods (2.30 vs 2.50 per 1000 admissions, P = .36) (Figure 1A).
Table 1. Baseline Characteristics of the Patients With In-Hospital Cardiac Arrest.
| Characteristic | Participants, No. (%) | P value | ||
|---|---|---|---|---|
| Total (N = 383) | Precrisis period (n = 232) | Crisis period (n = 151) | ||
| Age, median (IQR), y | 69 (61-79) | 71 (62-79) | 69 (60-77) | .19 |
| Sex | ||||
| Male | 238 (62.1) | 138 (59.5) | 100 (66.2) | .20 |
| Female | 145 (37.9) | 94 (40.5) | 51 (33.8) | |
| Hypertension | 194 (50.7) | 117 (50.4) | 77 (51.0) | .92 |
| Diabetes | 141 (36.8) | 89 (38.4) | 52 (34.4) | .45 |
| Myocardial infarction | 42 (11.0) | 27 (11.6) | 15 (9.9) | .74 |
| Heart failure | 110 (28.7) | 69 (29.7) | 41 (27.2) | .64 |
| COPD | 32 (8.4) | 22 (9.5) | 10 (6.6) | .35 |
| Liver cirrhosis | 46 (12.0) | 23 (9.9) | 23 (15.2) | .15 |
| Chronic kidney disease | 101 (26.4) | 69 (29.7) | 32 (21.2) | .08 |
| Any malignant neoplasm in a solid organ within 5 y | 135 (35.2) | 86 (37.1) | 49 (32.5) | .38 |
| Any hematologic malignant neoplasms within 5 y | 31 (8.1) | 17 (7.3) | 14 (9.3) | .57 |
| Charlson Comorbidity Index | 6 (4-9) | 7 (4-9) | 5 (4-9) | .03 |
| Prior OHCA before admission | 31 (8.1) | 17 (7.3) | 14 (9.3) | .57 |
Abbreviations: COPD, chronic obstructive pulmonary disease; OHCA, out-of-hospital cardiac arrest.
Figure 1. Line and Bar Charts of Monthly Trends in In-Hospital Cardiac Arrest (IHCA) Incidence and the Proportion of Prolonged Cardiopulmonary Resuscitation (CPR; >35 Minutes) Before and During the Medical Crisis.
The overall incidence of in-hospital cardiac arrest was 2.295 per 1000 admissions during the precrisis period and 2.501 per 1000 admissions during the crisis period.
Clinical and Resuscitation Characteristics of IHCA
Time from admission to IHCA was shorter during the crisis than precrisis period (median [IQR], 2 [0-8] vs 4 [1-16] days) (Table 2). Although the prevalence of specific conditions such as sepsis (62 of 180 [34.4%] vs 128 of 289 [44.3%]) and pneumonia (26 of 180 [14.4%] vs 85 of 289 [29.4%]) differed, major organ insufficiencies—including respiratory, kidney, and hemodynamic (hypotension)—were comparable between the crisis and precrisis periods. Prearrest severity also showed no significant difference, as reflected by NEWS2 scores, RRT activation rates, and organ support therapies. Measured early response metrics, witnessed and monitored status, initial rhythm, presumed IHCA cause, and use of ECPR or defibrillation did not differ statistically between periods (Table 3).
Table 2. Clinical Characteristics at IHCA Onset.
| Characteristic | Cases, No. (%) | P value | ||
|---|---|---|---|---|
| Total (N = 469) | Precrisis period (n = 289) | Crisis period (n = 180) | ||
| Admission to IHCA event, median (IQR), d | 3 (1-13) | 4 (1-16) | 2 (0-8) | .005 |
| Location of cardiac arrest | ||||
| Medical ward | 81 (17.3) | 48 (16.6) | 33 (18.3) | .65 |
| Surgical ward | 45 (9.6) | 30 (10.4) | 15 (8.3) | |
| Emergency department | 115 (24.5) | 64 (22.1) | 51 (28.3) | |
| Intensive care unit | 183 (39.0) | 119 (41.2) | 64 (35.6) | |
| Operating room | 8 (1.7) | 5 (1.7) | 3 (1.7) | |
| Others | 37 (7.9) | 23 (8.0) | 14 (7.8) | |
| Time of cardiac arrest occurrence | ||||
| Daytime (8 am-6 pm) | 246 (52.5) | 159 (55.0) | 87 (48.3) | .18 |
| Nighttime (6 pm-8 am) | 223 (47.5) | 130 (45.0) | 93 (51.7) | |
| Cardiac arrest during RRT operating time | 180 (38.4) | 109 (37.7) | 71 (39.4) | .77 |
| RRT activation before arrest | 70 (45.5) | 44 (46.3) | 26 (44.1) | .87 |
| NEWS2, median (IQR) | 8 (4-11) | 7 (4-11) | 8 (4-12) | .39 |
| Low risk (0-4) | 125 (26.7) | 78 (27.0) | 47 (26.1) | .95 |
| Medium risk (5-6) | 67 (14.3) | 42 (14.5) | 25 (13.9) | |
| High risk (>7) | 277 (59.1) | 169 (58.5) | 108 (60.0) | |
| Active problem at the time of cardiac arrest | ||||
| Respiratory insufficiency | 348 (74.2) | 206 (71.3) | 142 (78.9) | .08 |
| Kidney insufficiency | 284 (60.6) | 177 (61.2) | 107 (59.4) | .70 |
| Hepatic insufficiency | 111 (23.7) | 75 (26.0) | 36 (20.0) | .15 |
| Hypotension | 270 (57.6) | 171 (59.2) | 99 (55.0) | .39 |
| Sepsis | 190 (40.5) | 128 (44.3) | 62 (34.4) | .04 |
| Heart failure | 132 (28.1) | 78 (27.0) | 54 (30.0) | .53 |
| Pneumonia | 111 (23.7) | 85 (29.4) | 26 (14.4) | <.001 |
| Electrolyte abnormality | 208 (44.4) | 113 (39.1) | 95 (52.8) | .004 |
| Acute stroke | 18 (3.8) | 9 (3.1) | 9 (5.0) | .33 |
| Acute nonstroke CNS event | 85 (18.1) | 48 (16.6) | 37 (20.6) | .32 |
| Vasopressor use | 220 (46.9) | 134 (46.4) | 86 (47.8) | .78 |
| Mechanical ventilation | 178 (38.0) | 103 (35.6) | 75 (41.7) | .20 |
| Kidney replacement therapy | 125 (26.7) | 84 (29.1) | 41 (22.8) | .16 |
| ECMO support | 14 (3.0) | 12 (4.2) | 2 (1.1) | .09 |
Abbreviations: CNS, central nervous system; ECMO, extracorporeal membrane oxygenation; IHCA, in-hospital cardiac arrest; NEWS2, national early warning score 2; RRT, rapid response team.
Table 3. Resuscitation Characteristics and In-Hospital Cardiac Arrest Outcomes.
| Characteristic or outcome | Cases, No. (%) | P value | ||
|---|---|---|---|---|
| Total (N = 469) | Precrisis period (n = 289) | Crisis period (n = 180) | ||
| Time from arrest recognition to CPR initiation, median (IQR), min | 0 | 0 | 0 | .25 |
| Time from arrest recognition to code blue activation, median (IQR), min | 0 | 0 | 0 | .27 |
| Code team arrival time from code activation, median (IQR), min | 0 (0-2) | 0 (0-2) | 0 (0-1) | .37 |
| Witnessed event | 428 (91.3) | 260 (90.0) | 168 (93.3) | .24 |
| Monitored event | 344 (73.3) | 210 (72.7) | 134 (74.4) | .75 |
| Initial documented rhythm | ||||
| Asystole | 96 (25.3) | 64 (26.6) | 32 (23.2) | .17 |
| Pulseless electrical activity | 184 (48.5) | 111 (46.1) | 73 (52.9) | |
| Pulseless ventricular tachycardia | 55 (14.5) | 41 (17.0) | 14 (10.1) | |
| Ventricular fibrillation | 44 (11.6) | 25 (10.4) | 19 (13.8) | |
| Defibrillation | 101 (21.5) | 64 (22.1) | 37 (20.6) | .73 |
| Defibrillation in shockable initial rhythma | 68 (68.7) | 43 (65.2) | 25 (75.8) | .36 |
| Time from arrest recognition to defibrillation, median (IQR), min | 2 (1-4) | 2 (0-4) | 2 (1-4) | .93 |
| >2 minb | 27 (39.7) | 17 (39.5) | 10 (40.0) | >.99 |
| Epinephrine use | 369 (78.7) | 228 (78.9) | 141 (78.3) | .91 |
| Epinephrine use in nonshockable initial rhythmc | 257 (91.8) | 162 (92.6) | 95 (90.5) | .65 |
| Time from arrest recognition to first epinephrine use, median (IQR), min | 2 (1-4) | 2 (1-4) | 1 (1-3) | .22 |
| >5 mind | 35 (13.6) | 22 (13.6) | 13 (13.7) | >.99 |
| ECPR | 26 (5.5) | 12 (4.2) | 14 (7.8) | .10 |
| Duration of CPR, median (IQR), min | 8.0 (3.0-20.0) | 8.0 (3.0-20.0) | 8.0 (3.0-20.0) | .62 |
| >20 min | 113 (24.1) | 71 (24.6) | 42 (23.3) | .83 |
| >30 min | 75 (16.0) | 52 (18.0) | 23 (12.8) | .16 |
| >35 min | 50 (10.7) | 38 (13.1) | 12 (6.7) | .03 |
| >35 min (non-ECPR cases) | 37 (8.4) | 30 (10.8) | 7 (4.2) | .02 |
| Cause of cardiac arrest | ||||
| Cardiovascular | 139 (29.6) | 94 (32.5) | 45 (25.0) | .13 |
| Respiratory | 93 (19.8) | 55 (19.0) | 38 (21.1) | |
| Traumatic | 12 (2.6) | 7 (2.4) | 5 (2.8) | |
| Neurologic | 18 (3.8) | 9 (3.1) | 9 (5.0) | |
| Hypovolemia, including nontraumatic bleeding | 55 (11.7) | 26 (9.0) | 29 (16.1) | |
| Metabolic | 111 (23.7) | 74 (25.6) | 37 (20.6) | |
| Anaphylaxis | 2 (0.4) | 2 (0.7) | 0 (0.0) | |
| Others or unknown | 39 (8.3) | 22 (7.6) | 17 (9.4) | |
| Successful ROSC | 351 (74.8) | 221 (76.5) | 130 (72.2) | .33 |
| 24-h Survival after IHCA | 252 (53.7) | 159 (55.0) | 93 (51.7) | .51 |
| 30-d Survival after IHCA | 152 (32.5) | 98 (34.0) | 54 (30.2) | .42 |
| Favorable neurological status | 117 (24.9) | 77 (26.6) | 40 (22.2) | .32 |
| Survival to discharge | 132 (28.3) | 82 (28.5) | 50 (27.9) | .92 |
Abbreviations: CPR, cardiopulmonary resuscitation; ECPR, extracorporeal cardiopulmonary resuscitation; IHCA, in-hospital cardiac arrest; IQR, interquartile range; ROSC, return of spontaneous circulation.
Percentages were calculated among patients with a shockable initial rhythm (n = 99).
Percentages were calculated among patients with a shockable initial rhythm who underwent defibrillation (N = 68).
Percentages were calculated among patients with a nonshockable initial rhythm (n = 280).
Percentages were calculated among patients with a nonshockable initial rhythm who received epinephrine (n = 257).
Distribution of CPR Duration and Characteristics of Prolonged Events
The median duration of CPR was 8 minutes in both periods. Nevertheless, the upper tail of the CPR duration distribution shifted leftward during the crisis period, with fewer resuscitations exceeding 35 minutes (12 of 180 [6.7%] vs 38 of 289 [13.1%]; P = .03) (Table 3 and Figure 2), a finding that persisted after excluding ECPR cases. In the adjusted multivariable logistic regression model, the crisis period was significantly associated with lower odds of CPR lasting longer than 35 minutes (adjusted odds ratio, 0.48; 95% CI, 0.24-0.96; P = .04).
Figure 2. Line Chart of Cumulative Distribution of Cardiopulmonary Resuscitation (CPR) Duration Among In-Hospital Cardiac Arrest (IHCA) Events.
Empirical cumulative distribution functions of CPR duration are shown. The vertical dashed line indicates 35 minutes from CPR initiation.
Despite differences in prolonged resuscitation efforts, overall clinical outcomes did not differ statistically (Table 3). ROSC was achieved in 130 of 180 events (72.2%) during the crisis period and 221 of 289 events (76.5%) before the crisis (P = .33). Neurological outcomes and survival at 24 hours, 30 days, and discharge also showed no significant differences (eg, survival to discharge: 50 [27.9%] vs 82 [28.5%]; P = .92).
Exploratory subgroup analyses suggested that the lower frequency of CPR lasting longer than 35 minutes was concentrated among events without ROSC (eTables 1 and 2 in Supplement 1). In cases where ROSC was not achieved, baseline comorbidity burden remained consistent across both time periods (median Charlson Comorbidity Index of 7 in both groups; P = .75). However, prolonged resuscitations—specifically those exceeding 35 minutes—were less frequent during the crisis period (8 of 50 [16.0%] vs 25 of 68 [36.8%]; P = .01), and a similar pattern was observed among non-ECPR cases (7 of 47 [14.9%] vs 24 of 67 [35.8%]; P = .02). This trend persisted in the subgroup with high comorbidity burden, defined as a Charlson Comorbidity Index of 5 or higher; among IHCA events without ROSC, events with CPR duration exceeding 35 minutes were less frequent during the crisis period (6 of 38 [15.8%] vs 21 of 55 [38.2%]; P = .02), whereas no significant difference was observed among events with ROSC (1 of 69 [1.4%] vs 8 of 158 [5.1%]; P = .28) (eTable 3 in Supplement 1).
Among events with CPR duration greater than 35 minutes, clinical outcomes were poor in both periods (eTable 4 in Supplement 1). ROSC was achieved in 4 of 12 prolonged events (33.3%) during the crisis period compared with 13 of 38 (34.2%) before the crisis (P > .99), and survival to discharge occurred in 1 event in each period. There were no survivors to discharge among patients who underwent prolonged CPR without ECPR.
Competing-Risk Analysis of ROSC and TOR
To account for the competing risks of achieving ROSC vs the termination of resuscitation efforts, cause-specific Cox models were employed (eTable 5 in Supplement 1). The crisis period was not associated with the hazard of ROSC (adjusted hazard ratio [aHR], 0.93; 95% CI, 0.72-1.19). In contrast, the crisis period was associated with a higher cause-specific hazard of TOR without ROSC (aHR, 1.66; 95% CI, 1.14-2.42).
Standardized cumulative incidence analyses using the G formula showed a similar pattern (eFigure 3D-3F in Supplement 1). The standardized cumulative incidence of ROSC was similar between the precrisis and crisis periods across CPR duration time points, whereas the incidence of TOR without ROSC was higher in the crisis period. The standardized absolute risk difference for TOR without ROSC was 0.027 (95% CI, 0.001-0.101) at 20 minutes and 0.049 (95% CI, 0.002-0.149) at 35 minutes (eFigure 4 in Supplement 1).
Discussion
In this study, we identified changes in IHCA characteristics and resuscitation practices before and during a nationwide resident resignation in South Korea. The crisis period was associated with fewer prolonged resuscitations and a higher cause-specific hazard of TOR without ROSC, while no statistically significant differences were detected in ROSC, neurological outcomes, or survival. These findings suggest that the workforce crisis and related resuscitation-system adaptations may have altered decision-making regarding the cessation of extended resuscitation efforts.
Although longer CPR duration is consistently associated with lower IHCA survival,21 the optimal timing of TOR and the definition of prolonged CPR remain uncertain.22 In line with current American Heart Association guidelines,23 multiple studies have used a 20-minute cutoff,24,25 whereas others have proposed 30-minute22,26 or 35-minute27,28 thresholds. Based on data from the Get With The Guidelines–Resuscitation registry showing that favorable neurologic outcomes and survival after IHCA decline below 1% after 32 and 39 minutes, respectively,21 we adopted a 35-minute threshold, beyond which outcomes in this study were uniformly unfavorable. A recently proposed IHCA TOR rule incorporates variables such as unwitnessed arrest, absence of monitoring, asystole, and resuscitation duration of at least 10 minutes.29 However, direct application of this rule to our cohort was limited because only 14 cases across both periods met the combined criteria of unwitnessed and unmonitored asystole, and 11 of these received at least 10 minutes of CPR, precluding meaningful interperiod comparisons. Two of these 11 patients achieved ROSC, but none had a favorable neurologic outcome or survived to discharge, supporting the poor prognosis associated with these indicators.
Previous studies from the COVID-19 pandemic showed that health care strain may alter the incidence, characteristics, and outcomes of IHCA through resource limitations and hospitalized population shifts.4,8,30,31 In contrast, the recent South Korean health care crisis was characterized primarily by an abrupt shortage of resident physicians, with no major disruptions to other medical resources and minimal impact on other health care personnel. Because resident physicians play a central role in inpatient care in South Korea, their sudden shortage may have contributed to changes in the composition and characteristics of the hospitalized patient population during the crisis.15,16,32 Despite these contextual differences, IHCA resuscitation practices should remain consistent, as resuscitation is a core, life-critical hospital function.3 This aligns with COVID-19 resuscitation guidance emphasizing the preservation of high-quality resuscitation whenever resources permit.33
TOR decisions may be influenced by a wide range of elements, ranging from patient-level characteristics, workforce-related factors, and material resources, to broader contexts including ethical considerations, family communication, and operational pressures or institutional expectations.10,22,34,35,36,37 In our setting, the resident physician resignation primarily affected workforce-related factors, including physician staffing, CPR team composition, overall workload, and supervision structure.17 While characteristics of patients with IHCA differed during the crisis, the crisis period remained independently associated with fewer prolonged CPR attempts and higher TOR without ROSC after adjustment, suggesting that these differences alone do not fully explain the observed change in TOR practice. Nonpersonnel resources, such as resuscitation equipment, remained relatively unchanged, as reflected by the comparable use of defibrillation, epinephrine and ECPR between periods. Nonetheless, ICU operational restructuring may have indirectly affected the decision-making environment.
Trainee physicians, including residents, play important roles in resuscitation teams.38,39,40 Following their resignation, many South Korean hospitals redistributed residents’ clinical responsibilities to attending physicians,16,17,32 potentially resulting in a transition from resident-led to attending-led cardiac arrest teams. However, how these changes may influence TOR decision-making remains poorly understood. Several studies suggest that clinicians’ experience and seniority may influence TOR decisions, with junior physicians tending to continue CPR longer.34,35,36 Given that TOR decisions are high-stakes and effectively irreversible, their potential variation according to team composition, leadership structure, or institutional context carries important clinical implications.10,34,39 Ultimately, structured TOR decision-support frameworks29 and postevent review3 may help reduce unwarranted variation and maintain consistency in decision-making during health care crises.
Limitations
Several limitations should be acknowledged. First, we did not exclude patients admitted following OHCA and allowed multiple IHCA events per patient. However, prior OHCA was relatively uncommon (8.1%) and similarly distributed across periods (Table 1), and within-patient correlation was accounted for in the cause-specific Cox and standardized cumulative incidence analyses. Second, because this study had a modest sample size, the absence of statistically significant differences in ROSC, survival, or neurological outcomes should not be interpreted as evidence of equivalent outcomes. Some subgroup analyses, particularly those involving patients with a high comorbidity burden or those who had a shockable initial rhythm, were based on small sample sizes and therefore should be interpreted with caution. Third, because this was a single-center before-and-after observational study, residual confounding and temporal confounding, including the possibility of increasing team experience over time, cannot be excluded. Although monthly IHCA incidence (Figure 1A) and the proportion of prolonged CPR (Figure 1B) did not show a clear seasonal or monotonic temporal pattern, these descriptive findings cannot fully exclude unmeasured temporal associations. Fourth, we could not directly measure detailed indicators of CPR quality, clinician judgment and postresuscitation care, so our interpretation of changes in TOR practice remains inferential. Additionally, because this was a single-center study conducted in a tertiary teaching hospital during a unique nationwide workforce crisis, generalizability may be limited. Multicenter studies are needed to clarify how workforce shortages may be associated with resuscitation practices across diverse hospital settings.
Conclusions
In this single-center before-and-after cohort study of patients with IHCA, the crisis period was associated with fewer CPR attempts exceeding 35 minutes and a higher cause-specific hazard of TOR without ROSC, while no statistically significant differences were detected in ROSC or major survival outcomes. These findings suggest that sustained workforce disruption may alter TOR practices and support the need for optimized and standardized TOR decision-making during health care crises.
eTable 1. Clinical and resuscitation characteristics of in-hospital cardiac arrest events without ROSC
eTable 2. Clinical and resuscitation characteristics of in-hospital cardiac arrest events with successful ROSC
eTable 3. Resuscitation characteristics and outcomes of in-hospital cardiac arrest events among patients with high comorbidity burden (Charlson Comorbidity Index ≥5)
eTable 4. Resuscitation characteristics and outcomes of in-hospital cardiac arrest events with CPR duration exceeding 35 minutes
eTable 5. Cause-specific Cox model for ROSC and termination of resuscitation (TOR)
eFigure 1. Scaled Schoenfeld residual plots for the primary exposure (crisis period) across cause-specific Cox models
eFigure 2. Receiver operating characteristic (ROC) analysis of the Charlson Comorbidity Index for discriminating survival to discharge
eFigure 3. Unadjusted and standardized cumulative incidence of ROSC and termination of resuscitation according to CPR duration before and during the medical crisis
eFigure 4. Standardized risk difference in cumulative incidence of ROSC and TOR over CPR duration, comparing during- versus before-crisis periods
Data Sharing Statement
References
- 1.Andersen LW, Holmberg MJ, Berg KM, Donnino MW, Granfeldt A. In-hospital cardiac arrest: a review. JAMA. 2019;321(12):1200-1210. doi: 10.1001/jama.2019.1696 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Dezfulian C, Cabañas JG, Buckley JR, et al. Part 4: systems of care: 2025 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation. 2025;152(16_suppl_2):S353-S384. doi: 10.1161/CIR.0000000000001378 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chan PS, Greif R, Anderson T, et al. ; International Liaison Committee on Resuscitation . Ten steps toward improving in-hospital cardiac arrest quality of care and outcomes. Resuscitation. 2023;193:109996. doi: 10.1016/j.resuscitation.2023.109996 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tong SK, Ling L, Zhang JZ, Yap FHY, Law KL, Joynt GM. Effect of the COVID-19 pandemic on cardiac arrest resuscitation practices and outcomes in non-COVID-19 patients. J Intensive Care. 2021;9(1):55. doi: 10.1186/s40560-021-00570-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Secrest KM, Anderson TM, Trumpower B, et al. Early changes in hospital resuscitation practices during the COVID-19 pandemic. Resusc Plus. 2022;12:100317. doi: 10.1016/j.resplu.2022.100317 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fekete-Győr A, Seckington A, Kiss B, et al. The impact of the SARS-CoV-2 pandemic on in-hospital cardiac arrest: A systematic review and meta-analysis. Resusc Plus. 2024;20:100756. doi: 10.1016/j.resplu.2024.100756 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Vaillancourt C, Charette M, Khorsand S, et al. Impact of a COVID-19 code blue protocol on resuscitation care and CPR quality during in-hospital cardiac arrest. Resuscitation. 2024;198:110172. doi: 10.1016/j.resuscitation.2024.110172 [DOI] [PubMed] [Google Scholar]
- 8.Chan PS, Spertus JA, Kennedy K, Nallamothu BK, Starks MA, Girotra S. In-hospital cardiac arrest survival in the United States during and after the initial novel coronavirus disease 2019 pandemic surge. Circ Cardiovasc Qual Outcomes. 2022;15(2):e008420. doi: 10.1161/CIRCOUTCOMES.121.008420 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Raffay V, Wittig J, Bossaert L, et al. ; ERC Ethics in Resuscitation Collaborators . European Resuscitation Council guidelines 2025 ethics in resuscitation. Resuscitation. 2025;215(suppl 1):110734. doi: 10.1016/j.resuscitation.2025.110734 [DOI] [PubMed] [Google Scholar]
- 10.Wang CH, Chang WT, Huang CH, et al. Factors associated with the decision to terminate resuscitation early for adult in-hospital cardiac arrest: Influence of family in an East Asian society. PLoS One. 2019;14(3):e0213168. doi: 10.1371/journal.pone.0213168 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yoon JH, Kwon IH, Park HW. The South Korean health-care system in crisis. Lancet. 2024;403(10444):2589. doi: 10.1016/S0140-6736(24)00766-9 [DOI] [PubMed] [Google Scholar]
- 12.The Lancet Regional Health-Western Pacific. junior doctor strikes in South Korea: more doctors are needed? Lancet Reg Health West Pac. 2024;44:101056. doi: 10.1016/j.lanwpc.2024.101056 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Seong SJ, Hwang JY, Song JY, et al. Burnout and depression of medical faculty: after mass resignation of junior doctors in Korea. J Korean Med Sci. 2026;41(12):e102. doi: 10.3346/jkms.2026.41.e102 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Murphy F. South Korean doctors end 18 month strike. BMJ. 2025;390:r1876. doi: 10.1136/bmj.r1876 [DOI] [PubMed] [Google Scholar]
- 15.Kim S, Ahn S, Koh K, Lee D. The association of nationwide trainee doctor mass resignations in South Korea with hospital utilization, expenditure, and patient outcomes. Front Public Health. 2025;13:1718795. doi: 10.3389/fpubh.2025.1718795 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Son C, Jung J, Joo H. Resident physician walkout and high-incidence cancer operation volumes in South Korea. JAMA Netw Open. 2026;9(3):e260697. doi: 10.1001/jamanetworkopen.2026.0697 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chang YR, Cho JH, Cho J, et al. Impact of medical crisis on the critical care system in South Korea. Acute Crit Care. 2025;40(3):393-401. doi: 10.4266/acc.000575 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. doi: 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]
- 19.Welch J, Dean J, Hartin J. Using NEWS2: an essential component of reliable clinical assessment. Clin Med (Lond). 2022;22(6):509-513. doi: 10.7861/clinmed.2022-0435 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nolan JP, Berg RA, Andersen LW, et al. Cardiac arrest and cardiopulmonary resuscitation outcome reports: update of the utstein resuscitation registry template for in-hospital cardiac arrest: a consensus report from a task force of the International Liaison Committee on Resuscitation (American Heart Association, European Resuscitation Council, Australian and New Zealand Council on Resuscitation, Heart and Stroke Foundation of Canada, InterAmerican Heart Foundation, Resuscitation Council of Southern Africa, Resuscitation Council of Asia). Circulation. 2019;140(18):e746-e757. doi: 10.1161/CIR.0000000000000710 [DOI] [PubMed] [Google Scholar]
- 21.Okubo M, Komukai S, Andersen LW, et al. ; American Heart Association’s Get With The Guidelines—Resuscitation Investigators . Duration of cardiopulmonary resuscitation and outcomes for adults with in-hospital cardiac arrest: retrospective cohort study. BMJ. 2024;384:e076019. doi: 10.1136/bmj-2023-076019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Goldberger ZD, Chan PS, Berg RA, et al. ; American Heart Association Get With The Guidelines—Resuscitation (formerly National Registry of Cardiopulmonary Resuscitation) Investigators . Duration of resuscitation efforts and survival after in-hospital cardiac arrest: an observational study. Lancet. 2012;380(9852):1473-1481. doi: 10.1016/S0140-6736(12)60862-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wigginton JG, Agarwal S, Bartos JA, et al. Part 9: Adult advanced life support: 2025 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation. 2025;152(16_suppl_2):S538-S577. doi: 10.1161/CIR.0000000000001376 [DOI] [PubMed] [Google Scholar]
- 24.Yonis H, Andersen MP, Mills EHA, et al. Duration of resuscitation and long-term outcome after in-hospital cardiac arrest: A nationwide observational study. Resuscitation. 2022;179:267-273. doi: 10.1016/j.resuscitation.2022.08.011 [DOI] [PubMed] [Google Scholar]
- 25.Youness H, Al Halabi T, Hussein H, Awab A, Jones K, Keddissi J. Review and outcome of prolonged cardiopulmonary resuscitation. Crit Care Res Pract. 2016;2016:7384649. doi: 10.1155/2016/7384649 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Matsuyama T, Ohta B, Kiyohara K, Kitamura T. Cardiopulmonary resuscitation duration and favorable neurological outcome after out-of-hospital cardiac arrest: a nationwide multicenter observational study in Japan (the JAAM-OHCA registry). Crit Care. 2022;26(1):120. doi: 10.1186/s13054-022-03994-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Goto Y, Funada A, Goto Y. Relationship between the duration of cardiopulmonary resuscitation and favorable neurological outcomes after out-of-hospital cardiac arrest: a prospective, nationwide, population-based cohort study. J Am Heart Assoc. 2016;5(3):e002819. doi: 10.1161/JAHA.115.002819 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Okubo M, Komukai S, Izawa J, et al. Duration of emergency medical services-initiated prehospital cardiopulmonary resuscitation efforts and survival for pediatric patients with out-of-hospital cardiac arrest. Resuscitation. 2026;223:111028. doi: 10.1016/j.resuscitation.2026.111028 [DOI] [PubMed] [Google Scholar]
- 29.Holmberg MJ, Granfeldt A, Moskowitz A, et al. Termination of resuscitation rules for in-hospital cardiac arrest. JAMA Intern Med. 2025;185(4):391-397. doi: 10.1001/jamainternmed.2024.7814 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Miles JA, Mejia M, Rios S, et al. Characteristics and outcomes of in-hospital cardiac arrest events during the COVID-19 pandemic: a single-center experience from a New York City public hospital. Circ Cardiovasc Qual Outcomes. 2020;13(11):e007303. doi: 10.1161/CIRCOUTCOMES.120.007303 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Sultanian P, Lundgren P, Strömsöe A, et al. Cardiac arrest in COVID-19: characteristics and outcomes of in- and out-of-hospital cardiac arrest. A report from the Swedish Registry for Cardiopulmonary Resuscitation. Eur Heart J. 2021;42(11):1094-1106. doi: 10.1093/eurheartj/ehaa1067 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Choi JY, Joo H, Kim JS, et al. Impact of mass resident physician resignation on inpatient mortality and admissions in Korea. Public Health. 2026;254:106225. doi: 10.1016/j.puhe.2026.106225 [DOI] [PubMed] [Google Scholar]
- 33.Nolan JP, Monsieurs KG, Bossaert L, et al. ; European Resuscitation Council COVID-Guideline Writing Groups . European Resuscitation Council COVID-19 guidelines executive summary. Resuscitation. 2020;153:45-55. doi: 10.1016/j.resuscitation.2020.06.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hansen C, Lauridsen KG, Schmidt AS, Løfgren B. Decision-making in cardiac arrest: physicians’ and nurses’ knowledge and views on terminating resuscitation. Open Access Emerg Med. 2018;11:1-8. doi: 10.2147/OAEM.S183248 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.So CW, Lui CT, Tsui KL, et al. Questionnaire survey on medical futility and termination of resuscitation in cardiac arrest patients among emergency physicians in Hong Kong. Hong Kong Med J. 2019;25(3):183-191. doi: 10.12809/hkmj187755 [DOI] [PubMed] [Google Scholar]
- 36.Salleh N, Poh K, Zainal Abidin H, Teh T. Questionnaire survey medical futility and termination of resuscitation in out-of-hospital-cardiac-arrest patients presenting to emergency department in hospitals in Klang Valley. Eurasian Journal of Emergency Medicine. 2023;22:146-160. doi: 10.4274/eajem.galenos.2023.69320 [DOI] [Google Scholar]
- 37.Hua M, Halpern SD, Gabler NB, Wunsch H. Effect of ICU strain on timing of limitations in life-sustaining therapy and on death. Intensive Care Med. 2016;42(6):987-994. doi: 10.1007/s00134-016-4240-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Lauridsen KG, Schmidt AS, Adelborg K, Løfgren B. Organisation of in-hospital cardiac arrest teams - a nationwide study. Resuscitation. 2015;89:123-128. doi: 10.1016/j.resuscitation.2015.01.014 [DOI] [PubMed] [Google Scholar]
- 39.Hejjaji V, Chakrabarti AK, Nallamothu BK, et al. Association between hospital resuscitation team leader credentials and survival outcomes for in-hospital cardiac arrest. Mayo Clin Proc Innov Qual Outcomes. 2021;5(6):1021-1028. doi: 10.1016/j.mayocpiqo.2021.06.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Mitchell OJL, Motschwiller CW, Horowitz JM, et al. Rapid response and cardiac arrest teams: a descriptive analysis of 103 American hospitals. Crit Care Explor. 2019;1(8):e0031. doi: 10.1097/CCE.0000000000000031 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Clinical and resuscitation characteristics of in-hospital cardiac arrest events without ROSC
eTable 2. Clinical and resuscitation characteristics of in-hospital cardiac arrest events with successful ROSC
eTable 3. Resuscitation characteristics and outcomes of in-hospital cardiac arrest events among patients with high comorbidity burden (Charlson Comorbidity Index ≥5)
eTable 4. Resuscitation characteristics and outcomes of in-hospital cardiac arrest events with CPR duration exceeding 35 minutes
eTable 5. Cause-specific Cox model for ROSC and termination of resuscitation (TOR)
eFigure 1. Scaled Schoenfeld residual plots for the primary exposure (crisis period) across cause-specific Cox models
eFigure 2. Receiver operating characteristic (ROC) analysis of the Charlson Comorbidity Index for discriminating survival to discharge
eFigure 3. Unadjusted and standardized cumulative incidence of ROSC and termination of resuscitation according to CPR duration before and during the medical crisis
eFigure 4. Standardized risk difference in cumulative incidence of ROSC and TOR over CPR duration, comparing during- versus before-crisis periods
Data Sharing Statement
