Hierarchical logistic regression modeling was conducted to examine the effects of hospital and patient levels on mortality risk. After analyzing the impact of hospital levels on mortality risk, this study substituted the resulting estimated intercepts into the hierarchical regression model for patient-level analysis. Table 5 indicates that a significant difference was observed between hospital levels and mortality risk (estimate: −2.2964; P = .0072; 95% CI: 3.9385 to −0.6543), that no significant differences were observed after hierarchical logistic regression modeling was conducted on the effects of admission time, that higher ages were associated with higher mortality risks (estimate: 0.041; P <.0001; 95% CI: 0.02711–0.05488), that higher CCI values were associated with higher mortality risks (estimate: −0.1192; P = .0311; 95% CI: −0.2275 to −0.01086), and that reduced mortality risks were observed in patients who did not receive TPA treatment (estimate: −1.737; P = .0002; 95% CI: −2.6521 to −0.822), suggesting higher mortality risks for those who received TPA treatment. In addition, no significant differences were observed between mortality risk and the following variables: adoption of PTCA and CABG treatment, emergency admission, and sex.
4.1 Clinical implication
This study compared the mortality risk of patients with first AMI onset in 2000 to 2009 who were admitted on weekends with the mortality risk of those admitted on weekdays. The 1-day and 2-week mortality rates of the weekend patients were found to be higher than those of the weekday patients, suggesting that the weekend effect may have affected the mortality risk of the patients for 2 weeks after admission. The results of the Cox regression analysis showed that the mortality risk of the weekend patients was 1.363 times that of the weekday patients, similar to the results of previous studies.[5–8] Li (2012) empirically demonstrated that the 7-, 30-, and 180-day mortality rates of patients admitted on weekends or Chinese New Year holidays were higher than those of patients admitted on weekdays and that this negative weekend effect mostly occurred in nonmedical centers rather than in medical centers. Thus, the hierarchical logistic regression modeling results generated in the present study revealed that hospital levels not only directly affected the mortality rate of patients admitted on weekends and weekdays but also influenced patient levels.
To examine the effect of admission method, this study conducted a univariate correlation analysis, showing that the 1-day and 2-week mortality rates of the weekend patients were higher than those of the weekday ones. Therefore, the weekend effect may have affected the 2-week mortality risk of weekend samples admitted through the EDs. This can be attributed to the fact that the patients with less severe symptoms generally visited outpatient departments, whereas those with more severe symptoms were admitted through EDs. In addition, this study conducted a Cox logistic regression analysis to calculate the hazard ratio (HR) of the weekend samples (HR: 0.797; 95% CI: 0.631–1.006), and a performed hierarchical logistic regression analysis on the same samples (P = .0524; 95% CI: −0.5737 to 0.003). The inferential statistical results indicated that no significant differences were observed in mortality risk between outpatient and emergency admission, rendering the study unable to confirm the existence of the weekend effect in the context of weekend emergency admissions. Sharp et al (2013) reported that, in the context of emergency admissions, the mortality risk of adults admitted on weekends through EDs was significantly higher than that of adults admitted on weekdays (odds ratio [OR]: 1.073; 95% CI: 1.061–1.084). After controlling for patient characteristics, the following results were obtained: an adjusted OR (AOR) of 1.026 (95% CI, 1.005–1.048). Sharp et al (2013) did not identify the existence of the weekend effect even after adjusting patient income, insurance status, hospital ownership, ED volume, and hospital teaching status.
The results of the Cox logistic regression analysis and hierarchical logistic regression analysis indicated that the mortality risk of patients who received TPA treatment was 3.12 times that of patients who did not receive the treatment, suggesting an increased mortality risk among the specific group of patients (average age: 71.90 ± 13.14; average CCI: 1.83 ± 1.02). Moreover, such increased risk can be attributed to several factors causing admission delays, including high disease severity, high ED volume, long referral time, and high cardiac catheterization room volume. The 2013 ACCF/AHA Guideline for the Management of ST-Elevation Myocardial Infarction, which emphasizes advances in reperfusion therapy, indicates that, in the absence of contraindications, fibrinolytic therapy (a TPA treatment) should be administered to patients when the anticipated transferal time exceeds 120 minutes. Vora et al (2015) reported that patients treated with fibrinolytic therapy did not have significant mortality difference compared with those treated with primary percutaneous coronary intervention (pPCI) (3.7% vs 3.9%; AOR: 1.13; 95% CI: 0.94–1.36), but that they had a higher bleeding risk (10.7% vs 9.5%; AOR: 1.17; 95% CI: 1.02–1.33) and that, for patients unlikely to receive timely pPCI treatment, pretransfer fibrinolysis and an early transfer for angiography may be a suitable reperfusion option when the potential benefits of timely reperfusion outweigh bleeding risk. Furthermore, most countries have recognized pPCI as an effective method of clearing a blocked coronary artery for patients with AMI. It has a 90% surgical success rate and reduces 24-hour mortality risk and diminishes 30-day reinfarction rate. However, more than one-third of reinfarctions occur within a year after PTCA treatment. Thus, to reduce further stenosis, patients can receive a coronary stent implantation after balloon angioplasty. The stenosis incidence rate can be reduced to 15% to 20% by metal coronary stent implantation and to 5% by drug-eluting stent implantation.
This study observed that, for both weekdays and weekends, more than 80% of the patients received PTCA treatment and more than 20% of the samples received CABG treatment, and that a total of 138 patients received both PTCA and CABG treatment, accounting for 6.8% of all patients. Although the NHIRD did not provide disease severity indicators, this study considered patients who received 2 or more treatments as an indicator of high disease severity because these patients generally experienced more severe symptoms or riskier treatments than the other patients did. The Cox and hierarchical logistic regression results revealed no significant differences between PTCA and CABG treatment and mortality risk, which was similar to the results of other studies. Previous study suggested that the proportion of patients with AMI who received pPCI substantially increased annually, from 12.4% in 1996 to 54.7% in 2007.
The mortality risk of the patients who received pPCI was affected by factors including income, insurance, aspirin intake, admission time (working hours or nonworking hours; weekdays or weekends), hospital ownership, ED volume, number of cardiology physicians, physician experience with pPCI, and door to balloon (D2B) time.[10,14,15] Accordingly, a high-quality cardiology center equipped with a solid network connecting each department or division, comprehensive medical treatment planning, and experienced attending physicians can ensure no significant differences in mortality risk and provide effective pPCI treatments to patients regardless of admission time.
Each year, more than 20% of the patients received CABG treatment, whereas 6.8% of the patients received pPCI followed by CABG (indicating riskier treatment conditions or higher disease severity). Mehta et al (2013) found that, after receiving pPCI with stent implantation, patients with ST-segment elevation myocardial infarction received CABG because of stenosis recurrence and vessel malformation or rupture, exhibiting mortality rates of 10% and 20% respectively. For patients requiring early CABG, monitoring TIMI 3 flow rate is a safe option to enable delivering a timely treatment for early ischemia.
The present study observed that the mortality rates of the patients began to increase substantially 3 months after their first ED admission, the reasons of which may be similar to the findings of a study examining patients with AMI in New Jersey hospitals. In recent decades, inpatient mortality has decreased substantially, whereas the decrease in long-term mortality has been less noticeable, suggesting that the mortality rate increased after the patients were discharged, mainly because of high noncardiovascular mortality, especially from respiratory and renal diseases, septicemia, and cancer in senior patients. Table 6 presents the different treatments for AMI patients in various populations.[1,3–8,10,12,14–20] In addition to diagnostic criteria, this disparity is largely due to the different sources of AMI patients. From 1996 to 2006, the inpatient AMI mortality rate declined annually, a phenomenon similar to the decreasing AMI mortality rate reported by the Ministry of Health and Welfare in their statistics that was corrected by the 2000 to 2025 population-adjusted mortality rate estimated by the World Health Organization. The 2005 to 2008 AMI mortality rate determined in the LHID under the NHIRD showed a similar decreasing trend.
For patients who experienced their first AMI episode, the present study found a 1-day mortality rate of 8.07%, a 2-week mortality rate of 10.47%. The mortality rate increased substantially 3 months after AMI onset and the trend continued; the 1-year total mortality rate reached 33.53% (67.63% for weekday patients and 32.37% for weekend patients). In addition, although the number of male patients far exceeded that of female patients and despite the male patients having an AMI incidence rate 2.15 times that of the female patients, no significant mortality rate differences were observed between the sexes in a modeling analysis. After analyzing the age variable through a regression analysis, this study found that older age was associated with higher mortality risks, which can be attributed to the fact that the patients aged >65 years accounted for 68.11% of all patients and that the patients who experienced CCI ≥1 accounted for 71.75% of all patients, conforming to past study results.
4.2 Research limitations
First, this study determined death status according to the patient status codes (ie, TRAN_CODE) recorded in the inpatient claims data (DD; code 4: death, code 5: discharge against advice, and code A: critical discharge against advice) and the insurance status codes (ie, ID_OUT_TYPE) recorded in the NHI underwriting database (code 1: insurance withdrawal and code 5: insurance suspension). National death registry data are more accurate than these data but could not be obtained in this study. Therefore, although death status could have been overestimated, these codes generally represented patient death because of the compulsory nature of Taiwan's NHI program; thus they were accurate indicators of death. Second, this study did not distinguish the nature of holidays, the date of which might have fallen on weekdays or weekends, thereby affecting the study results. Third, this study did not distinguish admission method on Saturdays because some hospitals in Taiwan provide both emergency and outpatient services, which may have affected the results of identifying the weekend effect. Finally, this study did not include various variables not provided in the NHIRD, such as times (eg, working and nonworking hours, D2B time, and referral time), disease severity, number and quality of medical professionals on duty, physician service length and experience, ED service volume, ED overcrowding status, and cardiac catheterization room status, leading to possible bias in the analysis results.
When examining the weekend effect, previous studies have mainly employed single-level analysis and thus have been unable to identify differences between weekdays and weekends. However, this study adopted hierarchical logistic regression analysis for mortality rate stratification, revealing that hospital-level factors could directly affect mortality risks for both weekday and weekend admissions and influence patient-level factors. When examining the mortality risk of the patients admitted on weekdays and weekends, this study conducted a univariate analysis and found that the 1-day and 2-week morality rates of the weekend patients were higher than those of the weekday patients, indicating that the weekend effect may have affected the mortality rates of those admitted on weekends for the 2 weeks after they were admitted. Furthermore, the adopted Cox regression analysis and hierarchical logistic regression analysis generated conflicting results, rendering the study unable to confirm the existence of the weekend effect.
This study was also supported by master thesis from the Cheng-Hua Wang, “A Comparison on Outcomes of Accepted Different Treatments for Acute Myocardial Infaraction Patients via Outpatient Clinics and Emergency Department Admitted between Weekday and Weekend”, Institute of Hospital and Health Care Administration, National Yang-Ming University, Taipei, Taiwan, July 2015.
Conceptualization: Ching-Wen Chien, Cheng-Hua Wang.
Data curation: Pei-En Chen.
Formal analysis: Zi-hao Chao, Song-Kong Huang.
Methodology: Zi-hao Chao, Song-Kong Huang.
Writing – original draft: Cheng-Hua Wang.
Writing – review & editing: Ching-Wen Chien, Cheng-Hua Wang, Tao-Hsin Tung.
. Sattar A, Bari A, Naqvi MA, et al. In-hospital mortality with relation to time of presentation in patients with acute ST elevation myocardial infarction. PJMHS 2014;8:99–104.
. Chang GM, Lau SC, Tung YC. Health system factors associated with outcome of care for acute myocardial infarction
. J Health Sci 2010;12:26–34.
. Pompei P, Charlson ME, Ales K, et al. Relating patient characteristics at the time of admission to outcomes of hospitalization. J Clin Epidemiol 1991;44:1063–9.
. 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;40:373–83.
. Aylin P, Yunus A, Bottle A, et al. Weekend mortality for emergency admissions. A large, multicentre study. Qual Saf Health Care 2010;19:213–7.
. Clarke MS, Wills RA, Bowman RV, et al. Exploratory study of the ’weekend effect
’ for acute medical admissions to public hospitals in Queensland, Australia. Intern Med J 2010;40:777–83.
. Hansen KW, Hvelplund A, Abildstrom SZ, et al. Prognosis and treatment in patients admitted with acute myocardial infarction
on weekends and weekdays from 1997 to 2009. Int J Cardiol 2013;168:1167–73.
. Isogai T, Yasunaga H, Matsui H, et al. Effect of weekend admission for acute myocardial infarction
on in-hospital mortality: a retrospective cohort study. Int J Cardiol 2015;179:315–20.
. Li SF, Lin IC, Tsai WD. Do hospitals provide lower quality care on weekends. J Soc Sci Philosophy 2012;24:233–75.
. Sharp AL, Choi H, Hayward RA. Don’t get sick on the weekend: an evaluation of the weekend effect
on mortality for patients visiting US EDs. Am J Emerg Med 2013;31:835–7.
. O’Gara PT, Kushner FG, Ascheim DD, et al. 2013 ACCF/AHA guideline for the management of ST-elevation myocardial infarction: a report of the American college of cardiology foundation/American heart association task force on practice guidelines. Circulation 2013;127:e362–425.
. Vora AN, Holmes DN, Rokos I, et al. Fibrinolysis use among patients requiring interhospital transfer for ST-segment elevation myocardial infarction care: a report from the US National Cardiovascular Data Registry. JAMA Intern Med 2015;175:207–15.
. Yang YY, Shen MY, Wu PH, et al. A project to reduce the abnormal rate of risk factors in acute myocardial infarction
patients. VGH Nursing 2011;28:275–82.
. Nallamothu BK, Normand SL, Wang Y, et al. Relation between door-to-balloon times and mortality after primary percutaneous coronary intervention over time: a retrospective study. Lancet 2015;385:1114–22.
. Tung YC, Chang GM, Chien KL, et al. The relationships among physician and hospital volume, processes, and outcomes of care for acute myocardial infarction
. Med Care 2014;52:519–27.
. Rathod KS, Jones DA, Gallagher SM, et al. Out-of-hours primary percutaneous coronary intervention for ST-elevation myocardial infarction is not associated with excess mortality: a study of 3347 patients treated in an integrated cardiac network. BMJ Open 2013;3.
. Mehta RH, Harjai KJ, Boura JA, et al. Short-term outcomes of balloon angioplasty versus stent placement for patients undergoing primary percutaneous coronary intervention: implications for patients requiring early coronary artery bypass surgery. Am Heart J 2013;165:1000–7.
. Kostis WJ, Deng Y, Pantazopoulos JS, et al. Trends in mortality of acute myocardial infarction
after discharge from the hospital. Circ Cardiovasc Qual Outcomes 2010;3:581–9.
. Bell CM, Redelmeier DA. Mortality among patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med 2001;345:663–8.
. de Cordova PB, Johansen ML, Martinez ME, et al. Emergency department weekend presentation and mortality in patients with acute myocardial infarction
. Nurs Res 2017;66:20–7.
Keywords:Copyright © 2019 The Authors. Published by Wolters Kluwer Health, Inc. All rights reserved.
acute myocardial infarction; hierarchical logistic regression modeling; weekend effect