Comparing uninsured and privately insured hospital patients: admission severity, health outcomes and resource use.

This paper compares uninsured hospital patients with privately insured patients in terms of severity of illness on admission, emergency department use, leaving the hospital against medical advice, length of stay, and in-hospital mortality and morbidity rates. This cross-sectional study includes 29,237 admissions to 100 US hospitals in 1993 and 1994. We found that uninsured patients are sicker, indicating that hospitals should expect uninsured patients to have increased service needs. Our results indicate that the uninsured exhibit higher likelihood of leaving against medical advice, shorter lengths of stay and poorer health outcomes suggest that the uninsured may not be receiving necessary care. Further studies are needed.

Linking health outcomes and resource efficiency for hospitalized patients: do physicians with low mortality and morbidity rates also have low resource expenditures?

This study addresses the question of whether physicians with better health outcomes for their patients spend more or less to accomplish these results. Several studies have examined this outcome-cost relationship at the hospital level, but the results are conflicting. The study sample (using an administrative database [1995 MQPro Comparative Database, MediQual Systems, Inc., Westborough, MA, USA]) comprised 175,249 adult medical service admissions to 100 hospitals in 25 states spanning 26 diagnosis-related groups (DRGs) during 1993 and 1994. Logistic regression models were used to estimate the expected probability of in-hospital mortality or morbidity; age, sex, severity of illness on admission, year of admission, insurance status and hospital were controlled for. The regression residuals were employed as quality indicators. Residual charges and length of stay (LOS) were estimated for each patient using an ordinary least squares regression model and were employed as resource efficiency indicators. A positive, statistically significant association at the physician level was found between mean morbidity residuals and each of the three mean resource efficiency residuals (LOS, 1.42 beta coefficient; ancillary charges, 1.78; and total charges, 1.27, all significant at the P < 0.001 level). The same positive and significant association was found between mortality residuals and each resource efficiency residual (LOS, 0.77 beta coefficient; ancillary charges, 0.80; and total charges, 0.68, all significant at the P < 0.01 level) when patients staying only one or two days were excluded. The results support our hypothesis that, on average, physicians with lower adjusted mortality or morbidity rates also have lower adjusted resource expenditures.

Toward a systems quality paradigm: relating health outcomes, resource expenditures, and appropriateness of cholecystectomy patients.

This study addresses the question for cholecystectomy patients of whether there is an association among manifesting better health outcomes and the quantity of hospital resources consumed when the appropriateness of this surgery is also considered. 10,043 cholecystectomies performed by 218 surgeons in 43 Pennsylvanian hospitals are analysed using data from an administrative data set. Performance measures are adjusted for admission severity of illness and other patient variables. The results demonstrate a statistically significant positive association between adjusted hospital total charges and adjusted morbidity controlling for whether specified clinical criteria are met that validate the need for a cholecystectomy. This study illustrates a systems quality paradigm wherein performance is examined in terms of health outcomes, appropriateness and resource expenditures, as well as the relationships among these three dimensions.

Occurrence of validating clinical findings in the hospital record for appendectomy, cholecystectomy and intervertebral disc excision patients: comparisons among hospitals and among surgeons.

This study examines the variation among 36 Pennsylvania hospitals, and the individual surgeons practicing in them, in the proportion of appendectomy, cholecystectomy and intervertebral disc excision patients with clinical findings in the hospital record that validate the need for surgery. Using admissions from January 1990 through June 1991, we performed logistic regressions on the probability of validating clinical findings controlling for patient age, sex, admission severity of illness, and Medicaid and Health Maintenance Organization membership. Our results show that hospitals, and surgeons, vary significantly in their validation rates for cholecystectomy and disc surgery and, to a lesser extent, appendectomy. We also found that increased procedure-specific volume at both the hospital and surgeon levels is not related to the odds of validating clinical findings. We define a future research agenda to investigate the reasons for the observed differences among hospitals and among surgeons.

Approaches to predictive modeling.

A four-component clinical model for process improvement is presented: (1) patient-related risk factors, (2) clinical processes ordered by the attending physician, (3) the hospital's execution of the physician's plan, and (4) the patient's outcome, or outcomes, resulting from the first three factors. The goal of risk adjustment in the analysis of quality of care is to account for the contribution of patient-related risk factors, so that the patient's outcome can be used as an indicator of the care ordered by the physician and executed by the hospital. Risk adjustment is usually accomplished by comparing the patient's predicted outcome, based on the patient's risk factors, to the observed outcome. Historically, three approaches to the development of prediction models have been used: (1) selection and weighting of risk factors by expert opinion, (2) univariate analyses, and (3) multivariate analyses. Future prediction models will be based on neural network techniques or cluster analysis. As these prediction models have evolved, there has been a steady increase in their predictive power.

Relating hospital health outcomes and resource expenditures.

This study addresses the question of whether hospitals with better health outcomes for their patients spend more or less to accomplish these results. Adult medical service admissions to 43 Pennsylvania hospitals are analyzed. Health outcomes and resource expenditures are adjusted for admission severity of illness and other patient variables. The results demonstrate a positive correlation between adjusted mortality (logit regression) and adjusted total charges, ancillary charges, and length of stay (ordinary least squares regression), but only the mortality/length-of-stay relationship is statistically significant (p < .05). For patients staying at least four days, however, there is a statistically significant, positive relationship between adjusted mortality and all three adjusted measures of resource expenditures. The relationship between the adjusted morbidity and each of these three adjusted resource measures is positive and statistically significant. The positive relationship is largely unrelated to such readily observable hospital characteristics as size, staffing, teaching status, and location in urban areas.

Inter-hospital mortality and morbidity variation in Pennsylvania.

A 1986 Pennsylvania law requires the public disclosure of hospital mortality and morbidity rates. This study of hospital admissions in 1989 and 1990 examines the variation in these health-outcome indicators for the 10 most frequently occurring DRGs in the adult medical service in a sample of 20 Pennsylvania hospitals. These mortality and morbidity rates are adjusted for admission severity, DRG, age, and sex, using a logistic regression model. The null hypothesis of no significant variation among hospitals is rejected by the statistically significant (p < 0.01) results of a likelihood ratio test on the hospital variables in logit models for both mortality and morbidity. Test results also show that 4 (20 percent) of 20 hospitals have statistically significant (p < 0.05) adjusted mortality rates, and 4 (20 percent) of 20 hospitals have significant morbidity rates. Such information may impact hospital management practices in a variety of ways.

Predicted probabilities of hospital death as a measure of admission severity of illness.

This paper evaluates a new method for assessing hospital admission severity of illness based on disease-specific models (logistic regression) of the probability of in-hospital mortality. Results for the 26 disease groups in MDC 4--Diseases of the Respiratory System, MDC 5--Diseases of the Circulatory System, and MDC 6--Diseases of the Digestive System are presented using data on all 1991 admissions from 111 hospitals throughout the United States. These disease models are empirically derived using clinical findings from laboratory, radiology, pathology, diagnostic procedures, patient history and physical exam, as well as patient age and sex. Each predictive algorithm is presented, and the strong predictive performance of these models is indicated by the average C statistic of .870. A predicted probability of death is calculated for each hospital patient in the study sample, and these probabilities comprise a continuous variable that indicates admission severity of illness.

Interhospital variations in admission severity-adjusted hospital mortality and morbidity.

In this study hospital admissions are categorized into admission severity groups based on key clinical findings. Severity of illness is determined again later in the hospital stay after treatment has been initiated. High severity on this second review is labeled major morbidity or morbidity, depending on the severity level, and these rates serve as a health outcome indicator along with in-hospital mortality. This study's findings show, for ten hospitals randomly selected from MedisGroups users, considerable interhospital variation in standardized mortality and morbidity ratios for ten frequently occurring DRGs on the adult medical service. After adjusting for admission severity and case mix, three of the ten study hospitals have a statistically significant (p less than .01) difference between the hospital's standardized mortality ratio and 1.0. Such a significant difference exists for the standardized major morbidity ratio of four hospitals and for the standardized morbidity ratio of three hospitals. At the DRG-specific level, our results show that 8.9 percent, 4.4 percent, and 15.0 percent of the hospital-specific mortality, major morbidity, and morbidity ratios, respectively, are statistically significant. Most hospital outliers have fewer deaths or morbid cases than expected. We caution that the study hospitals may not be representative of a larger group of U.S. hospitals.