• Quality Improvement in Health Care
• /
• v.10 no.2
• /
• pp.216-231
• /
• 2003

Objectives: To develop a model that predicts a death probability of acute myocardial infarction(AMI) patient, and to evaluate a performance of hospital services using the developed model. Methods: Medical records of 861 AMI patients in 7 general hospitals during 1996 and 1997 were reviewed by two trained nurses. Variables studied were risk factors which were measured in terms of severity measures. A risk model was developed by using the logistic regression, and its performance was evaluated using cross-validation and bootstrap techniques. The statistical prediction capability of the model was assessed by using c-statistic, $R^2$ as well as Hosmer-Lemeshow statistic. The model performance was also evaluated using severity-adjusted mortalities of hospitals. Results: Variables included in the model building are age, sex, ejection fraction, systolic BP, congestive heart failure at admission, cardiac arrest, EKG ischemia, arrhythmia, left anterior descending artery occlusion, verbal response within 48 hours after admission, acute neurological change within 48 hours after admission, and 3 interaction terms. The c statistics and $R^2$ were 0.887 and 0.2676. The Hosmer-Lemeshow statistic was 6.3355 (p-value=0.6067). Among 7 hospitals evaluated by the model, two hospitals showed significantly higher mortality rates, while other two hospitals had significantly lower mortality rates, than the average mortality rate of all hospitals. The remaining hospitals did not show any significant difference. Conclusion: The comparison of the qualities of hospital service using risk-adjusted mortality rates indicated significant difference among them. We therefore conclude that risk-adjusted mortality rate of AMI patients can be used as an indicator for evaluating hospital performance in Korea.

• The Korean Journal of Health Service Management
• /
• v.9 no.2
• /
• pp.81-88
• /
• 2015

The purpose of this study was to compare the risk-adjusted in-hospital mortality for craniotomies between logistic regression and multilevel analysis. By using patient sample data from the Health Insurance Review & Assessment Service, in-patients with a craniotomy were selected as the survey target. The sample data were collected from a total number of 2,335 patients from 90 hospitals. The sample data were analyzed with SAS 9.3. From the results of the existing logistic regression analysis and multilevel analysis, the values from the multilevel analysis represented a better model than that of logistic regression. The intra-class correlation (ICC) was 18.0%. It was found that risk-adjusted in-hospital mortality for craniotomies may vary in every hospital. The agreement by kappa coefficient between the two methods was good for the risk-adjusted in-hospital mortality for craniotomies, but the factors influencing the outcome for that were different.

• Journal of Preventive Medicine and Public Health
• /
• v.46 no.2
• /
• pp.74-81
• /
• 2013

Objectives: The objective of this study was to evaluate the performance of risk-adjusted mortality models for colorectal cancer surgery. Methods: We investigated patients (n=652) who had undergone colorectal cancer surgery (colectomy, colectomy of the rectum and sigmoid colon, total colectomy, total proctectomy) at five teaching hospitals during 2008. Mortality was defined as 30-day or in-hospital surgical mortality. Risk-adjusted mortality models were constructed using claims data (basic model) with the addition of TNM staging (TNM model), physiological data (physiological model), surgical data (surgical model), or all clinical data (composite model). Multiple logistic regression analysis was performed to develop the risk-adjustment models. To compare the performance of the models, both c-statistics using Hanley-McNeil pair-wise testing and the ratio of the observed to the expected mortality within quartiles of mortality risk were evaluated to assess the abilities of discrimination and calibration. Results: The physiological model (c=0.92), surgical model (c=0.92), and composite model (c=0.93) displayed a similar improvement in discrimination, whereas the TNM model (c=0.87) displayed little improvement over the basic model (c=0.86). The discriminatory power of the models did not differ by the Hanley-McNeil test (p>0.05). Within each quartile of mortality, the composite and surgical models displayed an expected mortality ratio close to 1. Conclusions: The addition of clinical data to claims data efficiently enhances the performance of the risk-adjusted postoperative mortality models in colorectal cancer surgery. We recommended that the performance of models should be evaluated through both discrimination and calibration.

• Journal of Preventive Medicine and Public Health
• /
• v.34 no.1
• /
• pp.21-27
• /
• 2001

Objectives : To develop a model that will predict the mortality of patients undergoing Coronary Artery Bypass Graft (CABG) and evaluate the perfermance of hospitals. Methods : Data from 564 CABGs peformed in six general hospitals were collected through medical record abstraction by registered nurses. Variables studied involved risk factors determined by severity measures. Risk modeling was performed through logistic repression and validated with cross-validation. The statistical performance of the developed model was evaluated using c-statistic, $R^2$, and Hosmer-Lemeshow statistic. Hospital performance was assessed by severity-adjusted mortalities. Results : The developed model included age, sex, BUN, EKG rhythm, Congestive Heart Failure at admission. acute mental change within 24 hours, and previous angina pectoris history. The c-statistic and $R^2$ were 0.791 and 0.001, respectively. Hosmer-Lemeshow statistic was 10.3(p value=0.2415). One hospital had a significantly higher mortality rate than the average mortality rate, while others were net significantly different. Conclusion : Comparing the quality of service by severity adjusted mortality rates, there were significant differences in hospital performance. The severity adjusted mortality rate of CABG surgery may He an indicator for evaluating hospital performance in Korea.

• Health Policy and Management
• /
• v.17 no.3
• /
• pp.87-105
• /
• 2007

The purpose of this study was to analyze whether nonemergency, isolated coronary artery bypass graft (CABG) surgery for high- or low-risk patients biases the assessment of the risk-adjusted mortality rates of hospitals. This study used 2002 National Health Insurance claims data for tertiary hospitals in Korea. The study sample consisted of 1,959 patients from 23 tertiary hospitals. The risk-adjustment model used the patients' biological, admission, and comorbidity data identified in the claims. The subjects were classified into high- and low-risk groups based on predicted surgical risk. The crude mortality rates and risk-adjusted mortality rates for low-risk, high-risk, and all patients in a hospital were compared based on the rank and the four intervals defined by quartile. Also, the crude mortality rates of the three groups were compared with their 95% confidence intervals of predicted mortality rates. The C-statistic (0.83) and Hosmer-Lemeshow test ($X^2$=11.47, p=0.18) indicated that the risk-adjustment model performed well. Presenting crude mortality rates with their 95% confidence intervals of predicted rates showed higher agreements among the three groups than using the rank or intervals of mortality rates defined by quartile in the hospital performance assessment. The crude mortality rates for the low-risk patients in 21 of the 23 hospitals were located on the same side of their 95% confidence intervals compared to that for all patients. High-risk patients and all patients differed at only one hospital. In conclusion, the impact of risk selection by hospital on the assessment results was the smallest when comparing the crude inpatient mortality rates of CABG patients with the 95% confidence intervals of predicted mortality rates. Given the increasing importance of quality improvements in Korean health policy, it will be necessary to use the appropriate method of releasing the hospital performance data to the public to minimize any unwanted impact such as risk-based hospital selection.

• Journal of Digital Convergence
• /
• v.11 no.5
• /
• pp.371-384
• /
• 2013

The purpose of this study was to develop the risk-adjusted mortality model using Korean Hospital Discharge Injury data and US National Hospital Discharge Survey data and to suggest some ways to manage hospital mortality rates through comparison of Korea and United States Hospital Standardized Mortality Ratios(HSMR). This study used data mining techniques, decision tree and logistic regression, for developing Korea and United States risk-adjustment model of in-hospital mortality. By comparing Hospital Standardized Mortality Ratio(HSMR) with standardized variables, analysis shows the concrete differences between the two countries. While Korean Hospital Standardized Mortality Ratio(HSMR) is increasing every year(101.0 in 2006, 101.3 in 2007, 103.3 in 2008), HSMR appeared to be reduced in the United States(102.3 in 2006, 100.7 in 2007, 95.9 in 2008). Korean Hospital Standardized Mortality Ratios(HSMR) by hospital beds were higher than that of the United States. A two-aspect approach to management of hospital mortality rates is suggested; national and hospital levels. The government is to release Hospital Standardized Mortality Ratio(HSMR) of large hospitals and to offer consulting on effective hospital mortality management to small and medium hospitals.

• Journal of Preventive Medicine and Public Health
• /
• v.39 no.1
• /
• pp.13-20
• /
• 2006

Objectives: To propose a risk-adjustment model with using insurance claims data and to analyze whether or not the outcomes of non-emergent and isolated coronary artery bypass graft surgery (CABG) differed between the low- and high-volume hospitals for the patients who are at different levels of surgical risk. Methods: This is a cross-sectional study that used the 2002 data of the national health insurance claims. The study data set included the patient level data as well as all the ICD-10 diagnosis and procedure codes that were recorded in the claims. The patient's biological, admission and comorbidity information were used in the risk-adjustment model. The risk factors were adjusted with the logistic regression model. The subjects were classified into five groups based on the predicted surgical risk: minimal (<0.5%), low (0.5% to 2%), moderate (2% to 5%), high (5% to 20%), and severe (=20%). The differences between the low- and high-volume hospitals were assessed in each of the five risk groups. Results: The final risk-adjustment model consisted of ten risk factors and these factors were found to have statistically significant effects on patient mortality. The C-statistic (0.83) and Hosmer-Lemeshow test ($x^2=6.92$, p=0.55) showed that the model's performance was good. A total of 30 low-volume hospitals (971 patients) and 4 high-volume hospitals (1,087 patients) were identified. Significant differences for the in-hospital mortality were found between the low- and high-volume hospitals for the high (21.6% vs. 7.2%, p=0.00) and severe (44.4% vs. 11.8%, p=0.00) risk patient groups. Conclusions: Good model performance showed that insurance claims data can be used for comparing hospital mortality after adjusting for the patients' risk. Negative correlation was existed between surgery volume and in-hospital mortality. However, only patients in high and severe risk groups had such a relationship.

• Quality Improvement in Health Care
• /
• v.10 no.2
• /
• pp.164-175
• /
• 2003

Objectives: To evaluate the performance of models to predict AMI patients death using severity adjustment measures in Korea. Methods: Medical records of 861 patients treated by AMI in 7 general hospitals during 1996 and 1997 were reviewed by trained nurses. We measured the severity of patients by APACHE III, MedisGroups, CSI and DS. Using each severity method a predictive mortality for each patient was calculated from a logistic regression model including the severity score. The statistical performance of each severity method model was evaluated by using c-statistics and R2. For each hospital, z scores compared actual and expected mortality rates. Results: The overall in-hospital mortality was 14.5%, ranged from 10.0% to 22.2%. The distributions of severity scores for each method was significantly different by hospitals. The four severity-adjusted models to predict AMI patients death varied in their statistical performance for discrimination power of patients death. Order of Severity-adjusted mortality rates and z scores by four severity measures was different. Conclusion: Severity-adjusted mortality rates of AMI patients might be applied as an indicator for hospital performance evaluation in Korea. Because different severity methods frequently produce different impressions about relative hospital performance, more studies has to be done to use it as quality indicator and more attention should be paid to select appropriate severity measures.

• Journal of Preventive Medicine and Public Health
• /
• v.40 no.1
• /
• pp.29-35
• /
• 2007

Objectives : To assess whether the risk-adjusted in-hospital mortality rates for non-emergent and isolated coronary artery bypass graft surgery (CABG) patients exhibited a consistent trend from 2001 to 2003. Methods : The data used in this study came from CABG claims that were submitted to a Korean Health Insurance Review Agency (HIRA) in 2001, 2002, and 2003. Study datasets included data from 17 tertiary hospitals, which had at least 25 claims each year over 3 years. The inter-hospital differences in patients' risk-factors were identified and controlled in the risk-adjustment model. Actual and predicted mortality rates for each hospital were calculated in 2001, 2002, 2003, and 2001+2002, and were then examined to identify consistent rate patterns over time. Kappa analysis was applied to assess the agreements between rates. Results : Hospitals with lower-than-expected inpatient mortality rates showed more consistent rates than those with higher-than-expected mortality rates. The mortality rates that were calculated based on data obtained over multiple years had less variation among hospitals than rates based on single year data. Based on the Kappa score, the highest agreement was found when the rates were compared between the 2-year combined data (2001+2002) and 2003. Conclusions : Consistent patterns over 3 years were most evident for hospitals which had lower-than expected mortality rates. Policy makers can use this information to identify the degree of outcomes in hospitals and help motivate or channel the behaviors of providers.

• Journal of Preventive Medicine and Public Health
• /
• v.34 no.3
• /
• pp.244-252
• /
• 2001

Objective : Health insurers and policy makers are increasingly examining the hospital mortality rate as an indicator of hospital quality and performance. To be meaningful, a risk-adjustment of the death rates must be implemented. This study reviewed 5 severity measurement methods and applied them to the same data set to determine whether judgments regarding the severity-adjusted hospital mortality rates were sensitive to the specific severity measure. Methods : The medical records of 584 patients who underwent coronary artery bypass graft surgery in 6 general hospitals during 1996 and 1997 were reviewed by trained nurses. The MedisGroups, Disease Staging, Computerized Severity Index, APACHE III and KDRG were used to quantify severity of the patients. The predictive probability of death was calculated for each patient in the sample from a multivariate logistic regression model including the severity score, age and sex to evaluate the hospitals' performance, the ratio of the observed number of deaths to the expected number for each hospital was calculated. Results : The overall in-hospital mortality rate was 7.0%, ranging from 2.7% to 15.7% depending on the particular hospital. After the severity adjustment, the mortality rates for each hospital showed little difference according to the severity measure. The 5 severity measurement methods varied in their statistical performance. All had a higher c statistic and $R^2$ than the model containing only age and sex. There was a little difference in the relative hospital performance evaluation by the severity measure. Conclusion : These results suggest that judgments regarding a hospital's performance based on severity adjusted mortality can be sensitive to the severity measurement method. Although the 5 severity measures regarding hospital performance concurred, more often than would be expected by chance, the assessment of an individual hospital mortality rates varied by the different severity measurement method used.