Introduction

Stroke is the second leading cause of disability and mortality worldwide, with its highest burden observed in low- and middle-income countries1. In 2016, 13.7 million new cases of stroke were reported worldwide, with approximately 87% involving ischemia, 10% involving intracranial hemorrhage, and 3% involving subarachnoid hemorrhage2,3. In response to the growing burden of acute stroke, Taiwan launched the Emergency Department Quality Improvement Program (EDQIP) in 2012 to enhance the quality of emergency care, particularly for time-sensitive conditions such as acute ischemic stroke (AIS). The EDQIP incorporated performance-based financial incentives and aimed to improve key metrics such as timely thrombolysis, emergency department (ED) overcrowding, and critical care transfers4. In 1995, the National Institute of Neurological Disorders and Stroke Trial revealed that patients with AIS who received an intravenous recombinant tissue plasminogen activator (rt-PA) within 3 h of symptom onset experienced a significant reduction in disability at 90 days, with 50% of those treated having minimal or no disability as compared with 38% of the control group5. Since 2010, various treatments for stroke have been developed to extend the therapeutic time window for reperfusion therapy and improve patient outcomes. These treatments include endovascular thrombectomy and neuroprotective agents6. In 2020, the American Heart Association and American Stroke Association endorsed the use of intravenous rt-PA within an extended time window of 4.5 h7,8. Despite the efficacy of intravenous rt-PA in improving outcomes, fewer than 5% of patients with AIS worldwide receive intravenous thrombolytic therapy within the selected therapeutic time window, and fewer than 100,000 endovascular thrombectomy procedures were performed worldwide in 20163. In Taiwan, the utilization rate of thrombolytic therapy increased from 0.03% in 2003 to 1.51% in 2010, with only 0.6% of 394,988 patients with AIS receiving thrombolytic therapy during that period9. More broadly, recent studies have highlighted the importance of optimizing ED performance to improve time-sensitive care delivery, including predictive modeling for ED visit patterns10strategies to improve care for older adults in the ED11and the association between ED boarding delays and 30-day mortality12. These findings underscore the critical role of ED systems in improving outcomes for patients with acute conditions such as AIS.

The EDQIP was designed to improve emergency care quality by reducing overcrowding in EDs, increasing the efficiency of patient transfer and ED management, and improving the outcomes of patients with critical conditions such as out-of-hospital cardiac arrest, acute myocardial infarction, AIS, major trauma, acute aortic disease, and severe sepsis13. As part of the program, patients with AIS who received thrombolytic therapy within 60 min of arrival at the ED were considered eligible for additional financial incentives. EDQIP offers a distinct model that emphasizes emergency department performance and time-sensitive financial incentives. Despite these benefits, the efficacy and impact of the EDQIP on the prognosis of patients with stroke has not yet been evaluated, and little studies have examined ED-based stroke quality improvement programs using nationwide population data in Asian healthcare systems. Therefore, this study aimed to examine the impact of the EDQIP on the overall prognosis of patients with stroke.

Results

Comparison of the EDQIP and non-EDQIP groups after the EDQIP’s implementation

A total of 169,434 patients who visited an ED with a diagnosis of stroke between 2013 and 2015 were initially identified. After patients with a history of stroke (n = 54,161), patients with incomplete data (n = 2,406), patients under 20 years of age (n = 314), and patients who were not admitted to a hospital were excluded, 64,574 patients were ultimately included for analysis (Fig. 1). Among these patients, 1,414 applied for additional financial incentives within the EDQIP, whereas 63,160 did not. The majority of the patients were men, with a large proportion aged above 75 years. Compared with the patients in non-EDQIP group, those in the EDQIP group were more likely to be aged between 55 and 74 years. In addition, patients in the EDQIP group had a higher monthly income and were more likely to reside in highly urbanized areas compared with those in the non-EDQIP group. They also required more intensive care unit (ICU) care compared with their non-EDQIP counterparts. Table 1 presents a detailed comparison of patient characteristics between the two groups. After 1:4 propensity score matching was performed, no significant differences in baseline characteristics were observed between the EDQIP and non-EDQIP groups.

Fig. 1
Fig. 1
Full size image

Patient enrollment flowchart.

Table 1 Characteristics of patients with acute ischemic stroke from 2013 to 2015 who received additional financial incentives within the emergency department quality improvement program and those who did not.
Full size table

Table 2 presents the prognosis of patients with AIS. Compared with the non-EDQIP group, the EDQIP group had 9% reduction of 30-day mortality rate (6% vs. 15%, adjusted hazard ratio[aHR] = 0.4, 95% confidence interval [CI] = 0.32–0.50, p < 0.001). Figure 2 presents the multivariate-adjusted 30-day cumulative survival probabilities of the EDQIP and non-EDQIP groups. In addition to more favorable survival outcomes, the EDQIP group exhibited an absolute reduction of 6 days in hospital length of stay (LOS) than that of the non-EDQIP group (12 [8–34] vs. 18 [7–27] days, ratio = 0.82, 95% CI = 0.78–0.87, p < 0.001). Similarly, the median ICU LOS was 2 days shorter for patients in the EDQIP group than for those in the non-EDQIP group (2[2–4] vs. 4[2–7] days, ratio = 0.69, 95% CI = 0.65–0.72, p < 0.001). Patients in the EDQIP group demonstrated a 2% lower rate of revisiting the ED within 72 h after discharge compared with those in the non-EDQIP group (5% vs. 7%, aOR = 0.74, 95% CI = 0.57–0.94, p = 0.016).

Table 2 Prognosis of patients with acute ischemia stroke who received additional care incentives within the emergency department quality improvement program and those who did not.
Full size table
Fig. 2
Fig. 2
Full size image

Multivariate adjusted 30-day cumulative survival probabilities of patients with acute ischemic stroke between 2013 and 2015, following the launch of the Emergency Department Quality Improvement Program in 2012. Patients who were eligible for additional EDQIP care incentives (EDQIP group) demonstrated significantly higher survival probabilities compared with those who were not (non-EDQIP group).

Comparison of patients with stroke before and after the EDQIP’s implementation

A total of 55,604 patients presented to EDs with a diagnosis of stroke in 2011. Of these patients, 14,581 with a history of stroke, 151 under 20 years of age, 981 with incomplete data, and 18,842 not admitted to a hospital were excluded from the study. Ultimately, 21,049 patients were included for analysis (Fig. 1). Compared with the EDQIP group, the pre-EDQIP group comprised a higher proportion of patients aged 75 years and above. In addition, the pre-EDQIP group exhibited higher CCI scores than those of the EDQIP group. After 1:2 propensity score matching, no significant differences in baseline patient characteristics were observed between the pre-EDQIP and EDQIP groups (Table 3).

Table 3 Characteristics of patients with acute ischemic stroke in 2011 and those who received additional care incentives within the emergency department quality improvement program from 2013 to 2015.
Full size table

Table 4 presents a comparison of the patients’ prognoses after stroke for the pre-EDQIP (2011) and EDQIP group. Patients in the EDQIP group exhibited an absolute 9% reduction in 30-day mortality rates compared with those in the pre-EDQIP group (6% vs. 15%, aHR = 0.39, 95% CI = 0.31–0.49, p < 0.001). Figure 3 presents the multivariate-adjusted 30-day cumulative survival probabilities of patients with AIS in the two groups. The EDQIP group exhibited an absolute reduction of 5 days in hospital LOS (12 [8–33] vs. 17[7–27] days, ratio = 0.83, 95% CI = 0.78–0.89, p < 0.001) and 2 days in ICU LOS (2 [2–4] vs. 4 [2–7] days, ratio = 0.70, 95% CI = 0.66–0.73, p < 0.001) compared with the pre-EDQIP group. In addition, the proportion of patients returning to the ED within 72 h after discharge was 2% lower in the EDQIP group than in the pre-EDQIP group (5% vs. 7%, aOR = 0.71, 95% CI = 0.54–0.92, p = 0.011).

Table 4 Prognosis of patients with acute ischemic stroke before and after the implementation of the emergency department quality improvement program.
Full size table
Fig. 3
Fig. 3
Full size image

Multivariate adjusted 30-day cumulative survival probabilities of patients with acute ischemic stroke in 2011 and those who received additional care incentives within the Emergency Department Quality Improvement Program between 2013 and 2015. The EDQIP group exhibited significantly more favorable survival outcomes compared with the pre-EDQIP cohort.

Discussion

In this study, patients with stroke who received additional incentives within the EDQIP exhibited significantly more favorable outcomes compared with those who did not. Specifically, the EDQIP group exhibited a lower risk of 30-day mortality compared with the non-EDQIP group. Additionally, the EDQIP group experienced shorter hospital and ICU LOS and a lower likelihood of returning to the ED within 72 h after discharge compared with the non-EDQIP group. To the best of our knowledge, this is the first nationwide, population-based study in an Asian context to evaluate the impact of an ED-focused quality improvement program on stroke outcomes.

Evidence suggests that the EDQIP improves the outcomes of patients with AIS. Transferring patients with AIS to a stroke center for management can reduce their mortality rates and increase their likelihood of receiving t-PA treatment14. Appropriate quality improvement initiatives have been found to enhance the effectiveness of emergency treatment for patients with AIS15. Additionally, pay-for-performance incentives in health insurance systems have been found to be associated with improved outcomes in the management of various conditions, such as diabetes and chronic kidney disease, by providing financial incentives for high-quality care16,17,18. Within the EDQIP, incentives are provided for AIS cases with National Institutes of Health Stroke Scale (NIHSS) scores of 4–25 who received r-tPA injections13. In line with these findings, our study demonstrated that patients with stroke who received EDQIP incentives exhibited more favorable outcomes compared with those who were not eligible for additional incentives within the EDQIP. This improvement may be attributed to several mechanisms. First, the financial incentives tied to timely r-tPA administration encouraged hospitals to streamline AIS workflows and improve compliance with evidence-based AIS treatment guidelines. Second, because most hospitals treating acute conditions in Taiwan are contracted with the National Health Insurance (NHI) program, they tend to actively align their clinical practices with NHI policy initiatives. Studies have shown that healthcare providers in Taiwan exhibit high compliance with pay-for-performance programs under the NHI19. Third, EDQIP participation is often linked to hospital accreditation criteria, which further drives institutions to monitor and improve metrics such as the number of rt-PA administrations and the door-to-needle time. Collectively, these factors may have contributed to faster treatment, more standardized care, and improved coordination in emergency settings for AIS.

Our findings not only underscore the effectiveness of the EDQIP in improving the outcomes of patients with AIS but also lay the foundation for broader implications in healthcare practices worldwide. Unlike similar programs, such as the Get With The Guidelines—Stroke program in the United States, the EDQIP demonstrates a uniform pattern of improved patient outcomes through well-structured quality improvement initiatives20,21. Comparable efforts have also been implemented internationally, including the Canadian Alteplase for Stroke Effectiveness Study, which evaluated thrombolysis performance across multiple centers7; the United Kingdom’s Sentinel Stroke National Audit Programme, which emphasized multidisciplinary care benchmarks and continuous audit feedback22,23; and Australia’s Stroke Foundation National Audit, which focused on system-wide care quality indicators and adherence to clinical guidelines24. Compared to these programs, EDQIP offers a distinct model by incorporating time-based financial incentives within an emergency department–centered framework. It also demonstrated a major economic impact, with major reductions in hospital and ICU LOS resulting in substantial cost savings for the healthcare system25. These savings can be redirected to other critical areas, further enhancing the overall quality of healthcare delivery. Future studies should prioritize examining long-term outcomes and patient-reported measures to provide a more comprehensive assessment of the program’s sustained impact on patient health.

The success of the EDQIP may serve as a valuable model for enhancing the quality of care for patients with AIS in developing countries. In these regions, challenges such as limited access to imaging technology for rapid diagnoses, underdeveloped transportation networks, and insufficient emergency medical services often delay patient access to specialized stroke centers, thereby hindering the timely administration of thrombolytic therapy26,27. With the implementation of quality improvement policies, such as financial support for the stroke care infrastructure, these barriers can be addressed, leading to improved outcomes for patients with AIS. To achieve these goals, governments and international organizations should explore alternative funding models, including public–private partnerships and tiered financial incentives, to facilitate access to stroke care28. In addition to financial investment, other policy strategies may include establishing regional AIS referral networks to facilitate timely transfers, supporting the development of certified stroke centers across geographic zones, and expanding the use of telemedicine to assist with real-time stroke diagnosis and treatment, particularly in rural or underserved areas29,30. These strategies could serve as scalable components of a broader stroke care framework adapted to different healthcare infrastructures. Overall, our findings provide valuable insights for future policy development aimed at enhancing the quality of AIS treatment in resource-limited settings. However, it is important to note that only a small proportion of AIS patients received EDQIP incentives during the study period, reflecting the early-stage adoption of thrombolysis and stringent eligibility criteria. As such, while the program showed promise in improving outcomes among this selected subgroup, the generalizability of our findings to the broader AIS population may be limited. Future evaluations incorporating broader program participation and real-world implementation data will be essential to assess its impact at scale.

Although we used a nationwide database to evaluate the effectiveness of the EDQIP, this study has several limitations. First, despite the effectiveness of ICD-9-CM codes in the identification of stroke31the severity of AIS could not be directly measured using the National Health Insurance Research Database (NHIRD). Although the EDQIP group exhibited a shorter hospital LOS and a lower proportion of ICU admissions compared with the non-EDQIP group, the effectiveness of the EDQIP may have been underestimated because patients with an NIHSS score below 4 did not receive r-tPA and may have also been excluded from the EDQIP. In addition, because the EDQIP incentives only applied to patients with NIHSS scores between 4 and 25, the study population likely excluded those with very mild or very severe strokes, thereby biasing the EDQIP group toward patients with moderate stroke severity. To partially account for stroke severity, ICU admission status was included in the propensity score matching and multivariate models as a proxy variable. Furthermore, since the EDQIP group was defined based on receipt of rt-PA therapy with associated financial incentives, the observed effects may reflect both treatment receipt and program participation, potentially confounding the impact of the EDQIP itself. Second, physician blinding was not feasible in this study, raising a concern that physicians may have been more attentive to patients eligible for EDQIP incentives. Third, although this study was retrospective in nature, selection bias could not be completely avoided. Notably, patients in the EDQIP group were more likely to be aged between 55 and 74 years, whereas those in the non-EDQIP group were more likely to be aged 75 years or above. In addition, a greater proportion of patients in the non-EDQIP group exhibited a CCI score of 3 or higher. To minimize possible selection bias, propensity score matching and multivariate analysis were used to adjust for potential confounders, including ICU admission status as a proxy for severity. Nevertheless, residual confounding from unmeasured variables—such as stroke severity scores (e.g., NIHSS) and patient functional status—may still exist. This limitation should be considered when interpreting the observed associations. Finally, because we used data only from the NHIRD, our findings may not be generalizable to populations from other countries. In addition, because only 2.2% of patients with AIS received EDQIP-related incentives, the study population represents a highly selected AIS patient group. Therefore, the findings should not be generalized to all patients with AIS but rather interpreted as the potential benefit among patients who met EDQIP eligibility and thrombolysis therapy treatment criteria. Future evaluations incorporating broader program participation and real-world implementation data will be essential to assess the program’s effectiveness at scale. Additionally, subgroup analyses stratified by age, comorbidity burden, or stroke subtype should be explored to better identify which populations benefit most from ED-based quality improvement programs such as EDQIP.

In conclusion, EDQIP was associated with improved the outcomes of patients with AIS in Taiwan, including lower 30-day mortality rates, reduced hospital and ICU LOS, and 72-h ED revisits. While these findings support the potential value of quality improvement initiatives in enhancing stroke care, they should be interpreted with caution due to possible confounding by indication. Nevertheless, the results highlight the promise of structured, incentive-based programs for integration into broader health-care systems.

Methods

Study design and data sources

This retrospective study was conducted to explore the prognosis of patients with AIS, comparing those who received care incentives within the EDQIP with those who did not. Data was obtained from the NHIRD, which is maintained by the Health and Welfare Data Science Center. Launched in 1995, Taiwan’s National Health Insurance Program covers 99.93% of the entire population32,33. The NHIRD has been widely used in research, and the accuracy of its data has been validated in previous studies34.

Study participants

Patients who presented with their first episode of AIS at an ED between 2011 and 2015 were included in the study. AIS was identified using the principal diagnosis code for cerebral artery occlusion (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 433.x to 434.x). Patients under 20 years of age, patients with a history of AIS, and patients with incomplete data were excluded from the analysis. Patients who were not admitted to a hospital were also excluded. Because the EDQIP was launched in 2012, patients with AIS diagnosed in 2011 were assigned to a group called the pre-EDQIP group. Patients with AIS diagnosed between 2013 and 2015 were further assigned to EDQIP and non-EDQIP groups depending on whether they received rt-PA therapy and whether they were eligible for additional financial incentives within the EDQIP (identified by code P4601B).

Relevant variables

To accurately determine the participants’ characteristics, demographic information, including age and sex, was retrieved from the NHIRD. Comorbidities were identified using ICD-9-CM codes for conditions such as hypertension and stroke and diseases listed in the Charlson Comorbidity Index (CCI)35. Patients were considered to have these comorbidities if they had received a primary or secondary diagnosis for a condition at least twice during outpatient visits or once during hospitalization. CCI scores were calculated for all participants and categorized into four groups: 0, 1, 2, and ≥ 3. To determine disease severity, information was obtained on whether the patients were admitted to an ICU. Economic status was categorized into six levels depending on each patient’s monthly income: ≤NT$20,008, NT$20,009–NT$28,800, NT$22,801–NT$36,300, NT$36,301–NT$45,800, and ≥ NT$45,801. The urbanization level of residential areas was divided into seven categories, with level 1 representing the most urbanized areas36. Additionally, healthcare organization characteristics, including hospital level (medical center, regional hospital, or district hospital) and organization type (public or nonpublic), were considered as part of the analysis.

Outcome measurement

The primary endpoint for this study was the risk of mortality within 30 days of admission. The secondary endpoints included hospital LOS, length of ICU stay, and the proportion of patients who required a 72-h ED revisit after discharge.

Statistical analysis

All statistical analyses were conducted using SAS software version 9.1 (SAS Institute, Cary, NC, USA). A p value less than 0.05 was considered statistically significant.

To minimize the potential effect of confounding factors between EDQIP group and non-EDQIP group or between pre-EDQIP group and EDQIP group, propensity score matching was performed using several variables, such as sex, age, monthly income, urbanization level, CCI score, hospital level, hospital type, and ICU admission status. The demographic data of patients with AIS are presented based on their grouping (pre-EDQIP group, EDQIP group, and non-EDQIP group). Comparative analyses were conducted between the pre-EDQIP group and the EDQIP group after matching, as well as between the EDQIP group and the non-EDQIP group. Categorical variables are presented as numbers and percentages. An SMD ≤ 0.1 indicated no significant difference between the groups37.

The 30-day mortality rate and 72-h ED revisit rate of the study participants are presented as crude case numbers and percentages. A log-rank test was conducted to explore differences in the cumulative survival probabilities of the participants. A multivariate analysis was conducted using a stratified Cox proportional-hazards regression model, with sex, age, monthly income, urbanization level, CCI score, hospital level, hospital type, and ICU admission status adjusted for. McNemar’s test was used to evaluate differences in the 72-h ED revisit rate between the participants. To determine the effect of the EDQIP on the frequency of 72-h ED revisits, conditional logistic regression was applied. Hospital and ICU LOS were described using medians and interquartile ranges (IQR). Because the LOS data were not normally distributed, logarithmic transformations were applied for analysis. Wilcoxon’s signed-rank test was used for bivariate analysis, followed by a multiple regression model with generalized estimating equations to account for the aforementioned covariates in multivariate analysis.

Since our study used the secondary dataset (Taiwan National Health Insurance Research Database, NHIRD), and the patient identifications in the NHIRD have been scrambled and de-identified by the Taiwan government for academic research use, informed consent was waived by the Research Ethics Committee. The research was conducted in accordance with the 1964 Declaration of Helsinki and its amendments and was approved by the institutional review board of Jen Ai Hospital (Institutional Review Board No. JAH 111 − 20), Taichung, Taiwan.