The burden of acute myocardial infarction after a regional cardiovascular center project in Korea

Transcription

1 International Journal for Quality in Health Care, 2015, 27(5), doi: /intqhc/mzv064 Advance Access Publication Date: 13 August 2015 Article Article The burden of acute myocardial infarction after a regional cardiovascular center project in Korea ARIM KIM 1, SEOK-JUN YOON 2, YOUNG-AE KIM 3, and EUN JUNG KIM 4 1 Graduate School of Public Health, Korea University of Seoul, Seoul , South Korea, 2 Department of Preventive Medicine, College of Medicine, Korea University, Seoul , South Korea, 3 Cancer Policy Branch, National Cancer Center, Goyang , South Korea, and 4 Economic Research Institute, Korea University of Seoul, Seoul , South Korea Address reprint requests to: Seok-Jun Yoon, Department of Preventive Medicine, College of Medicine, Korea University, 73 Inchonno, Seongbuk-Gu, Seoul , South Korea. Tel: ; Fax: ; yoonsj02@korea.ac.kr Accepted 27 July 2015 Abstract Objective: The aim of this study was to examine the impact of a government-directed regional cardiovascular center (RCVC) project on the length of stay (LOS) and medical costs due to acute myocardial infarction (AMI). Design: A retrospective claim data review. Setting: Forty hospitals including four RCVCs in Korea. Participants: A total of 1469 AMI patients who visited a RCVC in two regions between February 2009 and December Intervention(s): RCVC project has been fostering specialized center by region for management of cardiovascular disease. It has built a system that could receive intensive care quickly within 3 h when disease occurred. Main Outcome Measure(s): Changes in the LOS and cost were evaluated using the difference-indifferences (DIDs) method combined with propensity score matching (1:1) and multilevel analysis with adjustment for patient s and institutional factors. Results: The net effect of RCVC project implementation showed decline of LOS ( 0.71 days) and total medical costs ( 797 US dollars) by DID. After the RCVC project, the LOS for patients with AMI hospitalized in a RCVC was decreased by 8.9% (β = 0.094, P = 0.041) compared with patients hospitalized in a hospital not designed as a RCVC. Compared with costs before the RCVC project, they were decreased by 11.5% (β = 0.122, P = 0.004). Conclusions: We provided evidence regarding the change in the societal burden due to AMI after regionalization. Although there was a reduction of LOS and direct medical costs reported in limited number of regionalized hospitals, in the long term we can anticipate an expanding impact in all regionalized hospitals. Key words: health policy, cardiovascular diseases, care pathways/disease management, evaluation methodology, health services research Introduction Cardiovascular disease (CVD) is a major health problem across the world, and disability and productivity loss due to poor health are important global health issues. Globally in 2010, cardiovascular and circulatory diseases, such as ischemic heart disease and stroke, were the leading cause of death (29.59%) and had the highest associated disability-adjusted life years (11.89%) [1]. The burden of CVD in South Korea is similar to the global trend. CVD is the most common The Author Published by Oxford University Press in association with the International Society for Quality in Health Care; all rights reserved 349

2 350 Kim et al. cause of death in Korea after cancer and stroke, accounting for 19.5% of all deaths [2]. Other studies also have reported a remarkable increase in the social and economic burden of CVD compared with the past [3, 4]. Because of the significant burden of CVD and the importance of tertiary prevention strategies, including prompt treatment and prevention of recurrence or complications in the management of CVD, many countries have specialized regional health service systems and evaluate their performance regularly. For instance, a healthcare program for patients with acute myocardial infarction (AMI) in England achieved a change in the clinical practice that was swift and consistent on a national scale. This change in the practice brought immediate benefits for patients in terms of reduced mortality and shortened hospital stays [5]. Additionally, prior studies demonstrated that regional systems or a high quality of hospital performances achieved the improvement during the 30 days in mortality and timeliness of reperfusion therapy and closed the gap in mortality between rural and metropolitan patients [6 10]. However, there is also the opinion that there is an insufficient evidence of an advantage of regional hospitals for the treatment of AMI [11]. In Korea, three Regional Cardio-cerebrovascular Centers (RCCVCs) were established for the prevention and treatment of CVD and initially funded by the Ministry of Health and Welfare in The province was divided into nine zones, except for the metropolitan areas, and then national hospitals or private university hospitals were designated as RCCVC in each region. The program was continually expanded to up to 2012, and 11 RCCVCs are now in operation. The government builds and operates a system in RCCVC that could receive intensive care quickly within 3 h when the disease occurred. The goal of this project was to minimize the incidence of complications through the provision of timely medical services anywhere in the country and to ultimately reduce medical and social costs due to the disease by facilitating earlier return to society after complete recovery [12]. The government supports an expansion of the medical infrastructure of teaching hospitals that are assigned to be RCCVC and operation professional practices, such as the expansion and new construction of facilities, a relocation and centralization center space through renovation, major equipment complement and replacement of aging equipment, staffing and reorganization, maintenance care system through improved treatment processes, and the deployment varied care activities. The project is made through a specialized operation of organization in the region center only. The RCCVC is made up of three clinical centers, such as Cardiovascular Center, Cerebrovascular Center, Cardiocerebral Rehabilitation Center and one center as Prevention and Management Center (Fig. 1). Former project had supported only the expansion of the medical infrastructure of hospitals. Unlike other project, the RCCVC project was creating synergy of the disease management by introducing the hardware construction as supporting facilities and equipments with software development like operational business [13]. Recently, there are few studies that have comprehensively evaluated the performance of RCCVC after the project implementation using indices of organization, process and outcomes and showed positive effects on various aspects like other countries [8, 14]. A number of previous studies on the effect of regional cardiac hospitals that exist to provide early specialized intervention have estimated change in the performance of reperfusion intervention or mortality, targeting only regionalized hospitals [5 11]. These studies have been limited by comparing the net effect of regionalization with non-regionalized hospitals in the same region [15], so it was impossible to know whether findings were actually results of regionalization or were due to other factors, like secular trends. There is a need to conduct comparative analyses of regionalized and non-regionalized hospitals to assess the impact of regionalizing AMI care across the country. Furthermore, it is important to examine changes in total medical costs resulting from improved coordination of health care and length of stay (LOS) due to improved clinical outcomes with regard to the socioeconomic burden of disease. LOS and total medical costs are mainly used as an indicator to assess the performance of medical practices as well as indicators to measure the cost-effectiveness of healthcare Figure 1 Organizational structure and major functions of RCCVC [13].

3 Regional cardiovascular center project in Korea Quality Assessment 351 services. Accordingly, the aim of this study was to examine the impact of a government-directed Regional Cardiovascular Center (RCVC) project on the LOS and total medical cost of patients with AMI. Methods Study sample We estimated the difference in LOS and total medical costs before and after RCVC project implementation in regionalized and nonregionalized hospitals using claims data from the Korean Health Insurance Review and Assessment Service (KHIRA). The unit of analysis was the claim. We identified patients with AMI as a principal or secondary diagnosis by searching International Classification of Diseases-10th Revision (ICD-10) codes. The definition of a patient with AMI was based on the following criteria: (i) aged 18 years or older; (ii) ICD-10 codes of AMI, namely I21 and (iii) inpatient who were admitted through the emergency department to a hospital that was a general hospital or a specified general hospital. We considered as severity of patient, scale of hospital and quality of healthcare service and possibility to select comparable control group in same region. Finally, we have selected four RCVCs in two regions as case groups. Two RCVC designated at 2012 were excluded from the analysis, because the data have not been completed, and thus, it will not be compared with those after implementation. The controls were selected from claims for patients with AMI admitted to nonregionalized hospitals in the same regions during the same periods and matched to cases in a 1:1 ratio. Our final number of claims for inpatients with AMI included 735 claims for patients admitted to two regionalized hospitals in Region A where the RCVC project was implemented from February 2009 to March 2010 and 734 claims for patients admitted to two regionalized hospitals in Region B where the RCVC project was conducted from May 2010 to December Measures Demographic measures included age, gender, type of medical coverage, comorbidity, use of percutaneous coronary intervention (PCI) or coronary artery bypass graft surgery (CABG), and death. Comorbidities were assessed using the Charlson Comorbidity Index (CCI) for each claim. CCI was calculated for all the comorbidities according to the scoring system established by Charlson et al. [16]. Characteristic hospital measures included geographic region, institution type, type of foundation and number of beds. Measures related to the RCVC project included regionalized or non-regionalized hospital and before or after the project. A regionalized hospital was defined as a hospital assigned as an RCVC funded by the Ministry of Health and Welfare in 2008 and The non-regionalized hospitals were hospitals in the same regions that were not affected by the RCVC project. Before or after the project was classified by the beginning of the project that was completed in each region (September 2009 in Region A, March 2011 in Region B). For patients who were defined as having AMI, we analyzed average LOS and total medical costs by case. Total medical costs by case were adjusted for the increase in the medical insurance fee in 2012 and converted into US dollars using an exchange rate of Korean won to US $1 (exchange rate on 16 July 2014). using difference-in-difference (DID) analysis. This analysis was able to measure the actual effect of the RCVC project among the total effect of being a regionalized hospital by removing the natural effect of the non-regionalized hospital. We estimated a DID random intercept and slope model of the following form: Y i ¼ α þ β Time i þ γ Center project i þ δ Time i Center project i þ ε Z i þ ϵ i where Y i was a continuous variable for LOS or total medical cost i; Time i was an indicator variable for the onset date of the project; Center project i was an indicator variable indicating whether the hospital was a regional center; Time i Center project i was an indicator variable for the interaction between the onset date of the project and whether the hospital was a regional center; Z i was a vector of covariates, including patient age group, gender, medical coverage, CCI, whether CABG was performed, whether PCI was performed, location of the institution, institution type, foundation entity and number of beds; and ɛ i was the error term. The actual effect of the RCVC project was obtained using the indicator variable for the interaction between the onset of the project and whether a hospital was a regional center. The control group, which is not affected by the intervention, was expected to have similar characteristics to cases in the group affected by the intervention except whether the hospital was designated as regional center or not in DID analysis. Therefore, we also used the propensity score matching method (PSM) to reduce selection bias between cases and controls. PSM produces a sample more similar to a randomized trial design and reduces systematic differences between the case and control groups [17]. We used logistic regression models, including variables for age group, gender, medical coverage, CCI, whether CABG was performed and whether PCI was performed, to calculate the propensity scores. We used the greedy matching algorithm [18] with matching from best to next-best. Differences between the case and control groups were compared using the χ 2 test for categorical variables and t-test for continuous variables after PMS. We conducted DID analysis to verify significant changes in LOS and total medical costs before and after the RCVC project for AMI patients at regionalized and non-regionalized hospitals. Additionally, we performed multilevel analysis to determine the impact of the RCVC project on outcome variables controlling for other variables, such as patient demographics (Level 1) and characteristics of hospital measures (Level 2). Outcome variables were log-transformed for the multilevel analysis, as LOS and total costs are not usually normally distributed. Because all of final models including the random effect of the risk factor have a smaller Akaike Information Criterion (AIC) value than random intercept model, we confirmed the random intercept and slope model has a better fit. To explore this further, the effects of policy according to severity of comorbidity were analyzed. Severity of comorbidity was scored according to CCI. Patients were divided into two groups: mild and moderate, with CCI scores of 1 3; severe, with CCI scores of 4. Our final results are displayed as rate of change on LOS and total medical costs, which were calculated as 100 * [exp(β) exp(0)] (%) with estimates (β) from the random intercept and slope model. Statistical analyses were performed with SAS Enterprise Guide, version 4.3 (SAS Institute Inc., Cary, NC, USA) and P < 0.05 was considered statistically significant. Statistical analysis To evaluate the impacts of the RCVC project, average LOS and total medical costs by case among patients with AMI presenting to a regional hospital before and after project implementation were compared Results Between February 2009 and December 2011, we identified 1469 claims for patients admitted to four regionalized hospitals in Regions

4 352 Kim et al. A and B, which comprised 716 claims before and 753 claims after project implementation, and 1469 claims from patients admitted to 36 non-regionalized hospitals, which comprised 750 claims before and 719 claims after project implementation. Patients who were between 51 and 80 years admitted to a regionalized hospital accounted for >75% of the patients, and 70% were male. Overall, 1.2% of patients underwent CABG, and PCI was performed in 75% of patients in regionalized hospitals, respectively. A simple check to evaluate the success of PSM is to look at differences in case mix before and after for each group of hospitals. After PSM, comparisons between the intervention and control groups suggested no significant differences at baseline. However, because the RCVC project is exclusively running at teaching hospitals, there was a significant difference in institution type, type of healthcare facilities foundation and number of beds (Table 1). The simple net effect of the RCVC project is presented in Table 2. The average LOS of patients admitted to regionalized hospitals was decreased by 0.71 days, and total medical costs by case were declined Table 1 General characteristics of AMI patients Case (group affected by the center project) by 797 US compared with non-regionalized hospitals after the RCVC project implementation. To characterize the actual effects of the RCVC project on LOS and costs, we conducted multilevel analysis controlling for demographic characteristics, which are listed below in Table 3, including age, gender, medical coverage, CCI, CABG, PCI, death, location of facilities, institution type, types of healthcare facilities foundation and number of beds. Overall, average LOS and total medical costs by case were lower for AMI patients admitted to regionalized hospitals after RCVC project implementation than for patients admitted at nonregionalized hospitals. After the RCVC project, the LOS was decreased by 8.9% (β = 0.094, P =0.041)comparedwithpatientshospitalizedin a hospital not designed as a RCVC. Compared with costs before the RCVC project, they were decreased by 11.5% (β = 0.122, P =0.004). The result of Table 4 indicates that the rate of change in LOS and costs for cases with severe comorbidity were greatly decreased in comparison to those with mild and moderate comorbidity after RCVC project implementation. LOS in severe comorbidity group showed the most Control (group that is not affected by the center project) Before After Before After 716 (51.8) 753 (51.2) 750 (51.1) 719 (48.9) Age (y), n (%) a (14.5) 113 (15) 99 (13.2) 105 (14.6) (21.8) 177 (23.5) 158 (21.1) 163 (22.7) (26.3) 171 (22.7) 208 (27.7) 182 (25.3) (29.6) 204 (27.1) 209 (27.9) 200 (27.8) (7.8) 88 (11.7) 76 (10.1) 69 (9.6) Gender, n (%) b Male 483 (67.5) 526 (69.9) 529 (70.5) 490 (68.2) Medical coverage, n (%) b Health insurance 659 (92) 677 (89.9) 695 (92.7) 664 (92.4) Medical aid 57 (8) 76 (10.1) 55 (7.3) 55 (7.6) Charlson score, n (%) a (36.2) 251 (33.3) 263 (35.1) 254 (35.3) (28.8) 222 (29.5) 223 (29.7) 224 (31.1) (15.4) 124 (16.5) 119 (15.9) 102 (14.2) 4 53 (7.4) 62 (8.2) 55 (7.3) 61 (8.5) 5 88 (12.3) 94 (12.5) 90 (12) 78 (10.9) No. of patients with CABG, n (%) 8 (1.1) 10 (1.3) 9 (1.2) 3 (0.4) b No. of patients with PCI, n (%) 553 (77.2) 549 (72.9) 556 (74.1) 554 (77.1) b No. of death, n (%) 47 (6.6) 54 (7.2) 73 (9.7) 61 (8.5) b Location of facilities, n (%) b A regions 353 (49.3) 382 (50.7) 410 (54.7) 325 (45.2) B regions 363 (50.7) 371 (49.3) 340 (45.3) 394 (54.8) Institution type, n (%) <0.001 b Specified general hospital 698 (97.5) 726 (96.4) 548 (73.1) 514 (71.5) General hospital 18 (2.5) 27 (3.6) 202 (26.9) 205 (28.5) Types of healthcare facilities foundation, n (%) <0.001 b Academy 363 (50.7) 371 (49.3) 444 (59.2) 397 (55.2) Private 353 (49.3) 382 (50.7) 286 (38.1) 303 (42.1) Public 20 (2.7) 19 (2.6) Number of bed, n (%) <0.001 b (6.3) 44 (6.1) (2.5) 27 (3.6) 62 (8.3) 51 (7.1) (97.5) 726 (96.4) 595 (79.3) 580 (80.7) (6.1) 44 (6.1) P-value a P-values provided from t-test. b P-values provided from χ 2 test.

5 Regional cardiovascular center project in Korea Quality Assessment 353 Table 2 Change in medical utilization on patients with AMI before and after RCVC project LOS (days) Cost (dollars a ) Before After Difference Before After Difference Group affected by the regional center project Group that is not affected by the regional center project b Difference b 810 b 797 a 1 dollar = won (16 July 2014). b P-value < Table 3 Multilevel analysis of RCVC project impact on AMI significant decrease by 23.5% (β = 0.286, P = 0.046). The effects of cost saving were remarkable in severe group (23.4%, β = 0.267, P =0.035). Discussion LOS (log) The aim of this study was to estimate the reduction in LOS and costs for patients with AMI after implementation of a RCCVC project by the government to reduce the societal burden of AMI. We performed a DID analysis in combination with PSM, using claims data from KHIRA ( ), for patients with AMI who were admitted to regionalized hospitals as cases and claims for patients with AMI admitted to non-regionalized hospitals in the same region as controls. After adjusting for demographic variables, DID analysis revealed that the average LOS and total medical costs by case were lower for AMI patients admitted to regionalized hospitals after RCVC project implementation than for AMI patients admitted at non-regionalized hospitals, and all differences were statistically significant. The results Cost (log) Estimate a SE P-value Estimate a SE P-value Intercept < <0.001 Group (ref = control) Time (ref = before) <0.001 Group*Time SE, Standard Error. a Adjusted for age, gender, medical coverage, CCI, CABG, PCI, death, location of facilities, institution type, types of healthcare facilities foundation and number of beds. Table 4 Rate of change by severity of comorbidity after RCVC project LOS (%) Cost (%) Rate of change (%) P-value Rate of change (%) P-value Total Mild and moderate a Severe b Rate of change on LOS and total medical costs, which were calculated as 100 * [exp(β) exp(0)] (%) with estimates (β) from the random intercept and slope model according to each severity group. The estimates was adjusted for age, gender, medical coverage, CCI, CABG, PCI, death, location of facilities, institution type, types of healthcare facilities foundation and number of beds. a Mild and moderate: CCI 3. b Severe: CCI > 3. of our study for LOS and costs are consistent with other data, suggesting that regionalization efforts for AMI care focused on improvement of integrated intervention lead to reductions in average LOS and total costs per case [14, 19, 20]. This study revealed little differences in average LOS after the RCVC project by DID method. This may be a result of the cumulative effect of comprehensive quality evaluation of AMI care for the national health insurance program, which limited how much additional improvement could be achieved. The comprehensive quality evaluation of AMI care for the national health insurance program has been conducted by KHIRA since 2005 to reduce variation between hospitals in outcomes of healthcare services and to improve quality of healthcare services. Since 2008, it has expanded to encompass general hospitals all over the country and since 2010 has applied and incentive program according to evaluation outcomes, similar to pay for performance. As a result, the average total evaluation scores have been increasing, narrowing the gap between hospitals and continually decreasing the growth rate of LOS and medical costs [21, 22]. Therefore, both the RCVC program and the comprehensive quality evaluation of AMI care for the national health insurance program have contributed to improvement in clinical outcomes and narrowing the gap between regional hospitals. Meanwhile, there was more effective in total medical costs per case than LOS after the RCVC project. This finding is likely due to efficient operating strategies in regionalized hospitals. Strategies, such as new clinical pathways used actively by specialists and having a specialist on duty, could reduce unnecessary healthcare utilization, such as admissions to intensive care units and additional examinations, interventions, and medications. The importance of integrated intervention is known [7] and strategies to improve clinical outcomes include activation of a specialized intervention team by the emergency team, maintenance of a specialized intervention team that is available at close range, and having a stationed specialist [23]. The RCVC in Korea incorporates these integrated interventions. In the future, to prove influence factors on decreasing costs clearly, we can analyze factors of

6 354 Kim et al. decreasing medical costs classified in detail, such as procedural costs, imaging costs, intensive care costs, medicine costs after discharge or another source, using claim data. We revealed that the rate of change in LOS and costs on severe comorbidity group showed more reduction than on mild and moderate group. In other words, RCVC project has made more impact on severe comorbidity group, especially in aspect of cost saving. This may indicate that the project was to minimize the incidence of complications through the provision of timely medical services and to ultimately reduce medical costs. Rapid transfer and timely intensive care of AMI patients are safe and effective using a standardized protocol with an integrated transfer system and empowerment of specialists, which will also translate into improvement in the long-term outcome of AMI patients [24]. However, this outcome should be interpreted with caution, because our study did not include more detailed preprocedural information regarding AMI severity in common with other study [25]. We did not find an improvement in 30-day mortality, despite improved outcomes of LOS and costs. It may also be explained in part by residual confounding [25, 26]. We cannot demonstrate that the shortened hospital stay and saved medical costs caused long-term effect like readmission, revascularization, recurrent MI and so on. The reduction in hospital LOS and medical costs could have reduced the risk of hospital-associated adverse events. Conversely, shortening the time in the hospital could have led to more adverse events in the period early after discharge [27]. It is unknown whether decreases in the LOS and medical costs of patients with AMI who hospitalized in RCVC may have been associated with changes in long-term outcomes. In the future, if building databases are available in the long term, which is more than for 2 years off and on after project implementation (totally >4 years), we can make a more developed comparative study to determine whether there were changes in readmission, discharge to skilled nursing facilities and hospital mortality with trends in 30-day mortality as effect of the project. The long-term effects of project can be proved. The following limitations should be considered when interpreting the results of this study. First, there are many factors related to LOS and medical costs of AMI patients, such as time to arrival at the emergency department, ambulance use, the patient s condition on arrival at the hospital, status of professionals and performance of the intervention; however, it is difficult to get this information from claims data. Second, we included 4 of 11 regionalized hospitals and focused our analysis on short-term effects within 1 year, which may affect the generalizability of our results. Third, our study considered only the effectiveness of health care with regards to quality of health service. The Agency for Healthcare Research and Quality (AHRQ) in the USA has suggested that the dimensions of quality of health service are effectiveness, timeliness, patient safety, patient-centeredness, care coordination, efficiency, infrastructure and access to health care [28]. Considering the purpose of the RCVC project was to provide highquality, timely interventions for AMI patients everywhere, there is a need to estimate performance in aspects other than effectiveness. Despite these limitations, the findings of this study provide valuable insights regarding the actual change in the burden of AMI after RCVC project implementation through analysis using DID in combination with PSM. DID method can show net effects of policy implementation over time and be applied in public health analysis to measure the interaction between intervention and time [29]. Generally, random selection of controls is preferred; however, we minimized selection bias through use of PSM in this study, because it is difficult to select controls through random sampling in an observational study design. Moreover, we used random intercept and slope model from multilevel analysis with adjustment for other potential confounding factors and focused on estimates of interaction as main results. Conclusion To our knowledge, this is the first study to characterize the actual impacts of the RCVC project. In conclusion, despite the limitations of the study, we provided evidence regarding the change in the societal burden due to AMI after regionalization. Although there was a reduction of direct medical costs reported in limited number of regionalized hospitals, in the long term we can anticipate an expanding impact on medical costs in all regionalized hospitals, because costly outcomes are avoided by appropriate, timely intervention. In the future, a prospective analysis at the national level is needed to determine the overall social and economic impact of the RCVC project in Korea. Acknowledgements The authors thank the Korea Health Industry Development Institute for their participation in this project. Funding This work was supported by a grant from the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (HI13C0729). References 1. Global burden of disease compare. Data visualization of Institute for Health Metrics and Evaluation. University of Washington. org/gbd-compare/ (1 August 2014, date last accessed). 2. Korean National Statistical Office. A report on the cause of death statistics in Seoul: Korean National Statistical Office, Kim EJ, Yoon SJ, Jo MW et al. Measuring the burden of chronic diseases in Korea in Public Health 2013;127: Yoon SJ, Kim EJ, Kim HJ et al. A study on measuring the burden of cardiocerebrovascular disease in Korea. Cheong won: Korea Centers for Disease Control and Prevention, NHS Improvement. Growth of Primary PCI for the treatment of heart attack patients in England : the role of NHS Improvement and the Cardiac Networks. Leister: NHS Improvement, Tideman PA, Tirimacco R, Senior DP et al. Impact of a regionalised clinical cardiac support network on mortality among rural patients with myocardial infarction. Med J Aust 2014;200: Ting HH, Rihal CS, Gersh BJ et al. Regional systems of care to optimize timeliness of reperfusion therapy for ST-elevation myocardial infarction: the Mayo Clinic STEMI Protocol. Circulation 2007;116: Chen J, Krumholz HM, Wang Y et al. Differences in patient survival after acute myocardial infarction by hospital capability of performing percutaneous coronary intervention: implications for regionalization. Arch Intern Med 2010;170: De Luca G, Suryapranata H, Ottervanger JP et al. Time delay to treatment and mortality in primary angioplasty for acute myocardial infarction: every minute of delay counts. Circulation 2004;109: De Luca G, Suryapranata H, Zijlstra F et al. Symptom-onset-to-balloon time and mortality in patients with acute myocardial infarction treated by primary angioplasty. J Am Coll Cardiol 2003;42: Rathore SS, Epstein AJ, Nallamothu BK et al. Regionalization of ST-segment elevation acute coronary syndromes care: putting a national policy in proper perspective. J Am Coll Cardiol 2006;47: Ministry of Health and Welfare. Korea Centers for Disease Control and Prevention. Operational guidelines for regional cardio-cerebrovascular center in Seoul: Ministry of Health and Welfare, Kim HS. An overview of cardio-cerebrovascular disease control in domestic and foreign country- Focusing on cardiovascular or cerebrovascular centers in England, Canada and Korea. Public Health Wkly Rep KCDC 2013;6:

7 Regional cardiovascular center project in Korea Quality Assessment Kim HS. Outcomes of Regional Cardiocerebrovascular Center (RCCVC) project during the first phase: focusing on results for severity-adjusted outcome indicators. Public Health Wkly Rep KCDC 2014;7: Lee DH, Seo JM, Choi JH et al. Early experience of Busan-Ulsan regional cardiocerebrovascular center project in the Treatment of ST elevation myocardial infarction. Korean J Med 2013;85: Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40: Smith JA, Todd PE. Does matching overcome Lalonde s critique of nonexperimental estimators? J Econom 2005;125: Parsons LS. Reducing bias in a propensity score matched-pair sample using greedy matching techniques. In: SAS Users Group International (SUGI) 26th conference paper, 2001, pp Khot UN, Johnson ML, Ramsey C et al. Emergency department physician activation of the catheterization laboratory and immediate transfer to an immediately available catheterization laboratory reduce door-to-balloon time in ST-elevation myocardial infarction. Circulation 2007;116: Khot UN, Johnson-Wood ML, Geddes JB et al. Financial impact of reducing door-to-balloon time in ST-elevation myocardial infarction: a single hospital experience. BMC Cardiovasc Disord 2009;9: Health Insurance Review and Assessment Service (HIRA). Comprehensive Quality Report of National Health Insurance Seoul: HIRA, Health Insurance Review and Assessment Service (HIRA). Comprehensive Quality Report of National Health Insurance Seoul: HIRA, Bradley EH, Herrin J, Wang Y et al. Strategies for reducing the door-toballoon time in acute myocardial infarction. NEnglJMed2006;355: Henry TD, Sharkey SW, Burke MN et al. A regional system to provide timely access to percutaneous coronary intervention for ST-elevation myocardial infarction. Circulation 2007;116: Renzi C, Asta F, Fusco D et al. Does public reporting improve the quality of hospital care for acute myocardial infarction? Results from a regional outcome evaluation program in Italy. Int J Qual Health Care 2014;26: Myerson M, Coady S, Taylor H et al. Declining severity of myocardial infarction from 1987 to 2002: the Atherosclerosis Risk in Communities (ARIC) Study. Circulation 2009;119: Bueno H, Ross JS, Wang Y et al. Trends in length of stay and short-term outcomes among medicare patients hospitalized for heart failure, JAMA 2010;303: Agency for Healthcare Research and Quality (AHRQ). National healthcare quality report Washington, DC: AHRQ publication, Ku IH, Lim SH, Moon HJ. The impacts on work, income, and poverty the National Basic Livelihood Security Program-measured using differences-indifferences. Korean Sociol Assoc 2010;44: