In 2024 (ROC 113), Taiwan's ER visits totaled 7,281,960, with 267,960 cases boarding in the ER for over 24 hours, a rate of 3.68%. 1
This rate is the highest in three years. 2022 (ROC 111) was 2.56%, 2023 (ROC 112) was 3.32%, and 2024 rose again — meaning 1 in every 27 ER patients waited more than a day. 1
Three Years, One Upward Line
The NHIA lists the "ER patient boarding over 24 hours rate" as an efficiency monitoring indicator, with the numerator being cases boarding over 24 hours and the denominator being total ER visits, measured semi-annually. 1
Overall, 2024 total ER visits actually decreased from 2023 (7.28 million vs. 7.48 million), but cases boarding over 24 hours rose from 248,378 to 267,960, pushing the rate from 3.32% to 3.68%. 1
Presentations down, boarding up — two opposite directions. The NHIA explained in the report that ER overcrowding involves three facets: presentation, length of stay, and inpatient admission; a bottleneck in any segment shows up in boarding time. 1
| 2022 | 2.56 | Overall |
|---|---|---|
| 2023 | 3.32 | |
| 2024 | 3.68 |
Source: NHIA "2024 National Health Insurance Monitoring Indicator Results Report"
Medical Centers at 7.73%
By hospital tier, pressure concentrates in medical centers. In 2024, medical centers had 1,913,390 ER visits with 147,905 boarding over 24 hours (7.73%); regional hospitals 3,472,336 visits, 107,040 cases, 3.08%; district hospitals 1,896,234 visits, 13,015 cases, 0.69%. 1
Notable is district hospitals. Their rate fell from 0.97% in 2023 to 0.69% in 2024, the only tier to decline. 1
In other words, the problem is not uniformly worsening across Taiwan, but concentrating in large hospitals.
Source: NHIA "2024 National Health Insurance Monitoring Indicator Results Report"
Even Level-1 Triage Patients Are Stuck
Looking only at the overall rate might suggest those boarding are low-acuity. Triage data overturns that assumption.
In 2024, the overall boarding-over-24-hours rate for level-1 triage patients was 11.96%, level-2 8.52%, level-3 3.16%; for medical centers alone, level-1 was 19.30%, level-2 13.10%, level-3 6.33%. 1
Level-1 denotes most urgent. At medical centers, 1 in 5 level-1 patients waited over a day in the ER — showing boarding time reflects not just patient streaming but critically ill patients unable to access ICU or inpatient beds. 1
The NHIA's Three Causes
The report's NHIA explanation is direct, breaking causes into three parts. 1
First, some non-urgent patients flood ERs, compounded by insufficient accessibility of primary clinics on nights and holidays, causing ER overload.
Second, ER patient acuity has risen with more complex conditions; elderly patients with multiple chronic diseases require longer observation stays, extending length of stay.
Third, insufficient inpatient beds and nursing staff, plus inter-hospital transfer mechanisms needing better communication and information referral, make timely inpatient admission difficult for ER patients.
These three correspond to three distinct solutions: presentation end needs referral system and primary care accessibility; length-of-stay end needs internal ER process improvements; admission end needs beds and staffing.
Five Meetings and a String of Measures
In 2025 (ROC 114), the NHIA held five consultation meetings on ER overcrowding on February 26, March 7, April 29, June 17, and July 21, inviting the Emergency Medicine Society, Emergency Management Society, Trauma Society, Hospital Association, and Nurses Union Federation. 1
Resulting measures include: adjusting ER, ICU, and nursing payment standards; increasing major trauma emergency surgery and anesthesia add-ons; revising the ER Quality Improvement Program; adding a Sunday and national holiday Urgent Care Center (UCC) pilot program; expanding home-based acute care; adding Outpatient Parenteral Antibiotic Therapy (OPAT). 1
The payment standard adjustments took effect May 1, 2025, tied to specific indicators — hospitals meeting targets receive increased individual budget caps, with indicators including nursing staff raises, ER overcrowding improvement, ICU efficiency, and ER patient transfer rates. 1
The referral system also moved. The report notes that from September 1, 2025, referral-out and referral-acceptance payment points were raised: referral-out +200 points to 700 per case; referral-acceptance +300 points to 500 per case. 1
How to Read the Numbers
One methodological limitation must be clear: boarding time comes from hospital-reported data, reflecting "rate of person-visits remaining in ER," not individual wait-time distribution, and does not show whether patients were ultimately admitted, transferred, or discharged. 1
But it suffices to answer a key question: 2024's ER pressure was not caused by record-high presentations, because total visits fell; pressure appeared in the back end — patients get in but can't get out. 1
Whether 2025's payment adjustments and pilot measures bend this curve will only be known when the next annual monitoring results arrive. Until then, only one thing is certain: 2024's 3.68% is the three-year peak.
Further Reading:
- Taiwan Healthcare and National Health Insurance — Context on ER payment and global budget system
- Taiwan Life Expectancy — How population aging changes ER and inpatient demand
- Taiwan Long-Term Care System Development — Where post-discharge care goes next
- Healthcare and Long-Term Care Special Topics — ER boarding, nursing staff, and physician counts by county on one page, with stakeholder positions on nurse-patient ratio legislation
References
- 2024 National Health Insurance Monitoring Indicator Results Report (Ministry of Health and Welfare, National Health Insurance Administration) — Published November 2025, contains 2022–2024 ER total visits and boarding-over-24-hours rates (overall 2.56%, 3.32%, 3.68%; medical centers 6.36%, 7.18%, 7.73%; regional hospitals 1.76%, 2.69%, 3.08%; district hospitals 0.77%, 0.97%, 0.69%), level-1 to level-3 triage rates, NHIA explanation of three major causes, and 2025 five consultation meetings with payment standards, UCC, OPAT, home-based acute care, and referral point adjustments.↩2345678910111213141516