Yevgenia D. Mackiernan

Judging hospitals by severity-adjusted mortality rates: the influence of the severity-adjustment method.

OBJECTIVES This research examined whether judgments about a hospitals risk-adjusted mortality performance are affected by the severity-adjustment method. METHODS Data came from 100 acute care hospitals nationwide and 11880 adults admitted in 1991 for acute myocardial infarction. Ten severity measures were used in separate multivariable logistic models predicting in-hospital death. Observed-to-expected death rates and z scores were calculated with each severity measure for each hospital. RESULTS Unadjusted mortality rates for the 100 hospitals ranged from 4.8% to 26.4%. For 32 hospitals, observed mortality rates differed significantly from expected rates for 1 or more, but not for all 10, severity measures. Agreement between pairs of severity measures on whether hospitals were flagged as statistical mortality outliers ranged from fair to good. Severity measures based on medical records frequently disagreed with measures based on discharge abstracts. CONCLUSIONS Although the 10 severity measures agreed about relative hospital performance more often than would be expected by chance, assessments of individual hospital mortality rates varied by different severity-adjustment methods.

Predicting in-hospital deaths from coronary artery bypass graft surgery. Do different severity measures give different predictions

OBJECTIVES Severity-adjusted death rates for coronary artery bypass graft (CABG) surgery by provider are published throughout the country. Whether five severity measures rated severity differently for identical patients was examined in this study. METHODS Two severity measures rate patients using clinical data taken from the first two hospital days (MedisGroups, physiology scores); three use diagnoses and other information coded on standard, computerized hospital discharge abstracts (Disease Staging, Patient Management Categories, all patient refined diagnosis related groups). The database contained 7,764 coronary artery bypass graft patients from 38 hospitals with 3.2% in-hospital deaths. Logistic regression was performed to predict deaths from age, age squared, sex, and severity scores, and c statistics from these regressions were used to indicate model discrimination. Odds ratios of death predicted by different severity measures were compared. RESULTS Code-based measures had better c statistics than clinical measures: all patient refined diagnosis related groups, c = 0.83 (95% C.I. 0.81, 0.86) versus MedisGroups, c = 0.73 (95% C.I. 0.70, 0.76). Code-based measures predicted very different odds of dying than clinical measures for more than 30% of patients. Diagnosis codes indicting postoperative, life-threatening conditions may contribute to the superior predictive power of code-based measures. CONCLUSIONS Clinical and code-based severity measures predicted different odds of dying for many coronary artery bypass graft patients. Although code-based measures had better statistical performance, this may reflect their reliance on diagnosis codes for life-threatening conditions occurring late in the hospitalization, possibly as complications of care. This compromises their utility for drawing inferences about quality of care based on severity-adjusted coronary artery bypass graft death rates.

Differences in procedure use, in-hospital mortality, and illness severity by gender for acute myocardial infarction patients: are answers affected by data source and severity measure

OBJECTIVES According to some studies, women with heart disease receive fewer procedures and have higher in-hospital death rates than men. These studies vary by data source (hospital discharge abstract versus detailed clinical information) and severity measurement methods. The authors examined whether evaluations of gender differences for acute myocardial infarction patients vary by data source and severity measure. METHODS The authors considered 10 severity measures: four using clinical medical record data and six using discharge abstracts (diagnosis and procedure codes). The authors studied all 14,083 patients admitted in 1991 for acute myocardial infarction to 100 hospitals nationwide, examining in-hospital death and use of coronary angiography, coronary artery bypass graft surgery (CABG), and percutaneous transluminal coronary angioplasty (PTCA). Logistic regression was used to calculate odds ratios for death and procedure use for women compared with men, controlling for age and each of the severity scores. RESULTS After adjusting only for age, women were significantly more likely than men to die and less likely to receive CABG and coronary angiography. Severity measures provided different assessments of whether women were sicker than men; for all cases, clinical data-based MedisGroups rated womens severity compared with mens, whereas four code-based severity measures viewed women as sicker. After adjusting for severity and age, women were significantly more likely than men to die in-hospital and less likely to receive coronary angiography and CABG; women and men had relatively equal adjusted odds ratios of receiving PTCA. Odds ratios reflecting gender differences in procedure use and death rates were similar across severity measures. CONCLUSIONS Comparisons of severity-adjusted in-hospital death rates and invasive procedure use between men and women yielded similar findings regardless of data source and severity measure.

PREDICTING IN-HOSPITAL MORTALITY FOR STROKE PATIENTS : RESULTS DIFFER ACROSS SEVERITY-MEASUREMENT METHODS

Objective: To see whether severity-adjusted predictions of likelihoods of in-hospital death for stroke patients differed among severity measures. Methods: The study sam ple was 9,407 stroke patients from 94 hospitals, with 916 (9.7%) in-hospital deaths. Probability of death was calculated for each patient using logistic regression with age-sex and each of five severity measures as the independent variables: admission MedisGroups probability-of-death scores; scores based on 17 physiologic variables on admission; Disease Stagings probability-of-mortality model; the Severity Score of Pa tient Management Categones (PMCs); and the All Patient-Refined Diagnosis Groups (APR-DRGs). For each patient, the odds of death predicted by the severity measures were compared. The frequencies of seven clinical indicators of poor prognosis in stroke were examined for patients with very different odds of death predicted by different severity measures. Odds ratios were considered very different when the odds of death predicted by one severity measure was less than 0.5 or greater than 2.0 of that pre dicted by a second measure. Results: MedisGroups and the physiology scores pre dicted similar odds of death for 82.2% of the patients. MedisGroups and PMCs disa greed the most, with very different odds predicted for 61.6% of patients. Patients viewed as more severely ill by MedisGroups and the physiology score were more likely to have the clinical stroke findings than were patients seen as sicker by the other severity measures. This suggests that MedisGroups and the physiology score are more clinically credible. Conclusions: Some pairs of severity measures ranked over 60% of patients very differently by predicted probability of death. Studies of seventy-adjusted stroke outcomes may produce different results depending on which seventy measure is used for risk adjustment. Key words: seventy; risk adjustment; stroke; in-hospital deaths; mortality rates. (Med Decis Making 1996;16:348-356)

Screening inpatient quality using post-discharge events.

BACKGROUND Decreasing hospital lengths of stay (LOS) hamper efforts to detect and to definitively treat complications of care. Patients leave before some complications are identified. OBJECTIVES To develop a computerized method to screen for hospital complications using readily available administrative data from outpatient and nonacute care within 90 days of discharge. DESIGN We developed the Complications Screening Program for Outpatient data (CSP-O) by using diagnosis and procedure codes from Medicare Part A and B claims to define 50 complication screens. Seventeen apply to specific procedural cases, and 33 apply to all adult, acute, medical, or surgical hospitalizations. The CSP-O algorithm examined outpatient, physician office, home health agency, and hospice claims within 90 days following discharge. SUBJECTS Seven hundred thirty nine thousand, two hundred and forty eight discharges of Medicare beneficiaries (age range, > or = 65 years) were admitted to 515 hospitals nationwide in 1994. RESULTS Complete 90-day, post-discharge windows were present for 62.8% of all and 68.5% of procedural cases. The 33 general screens flagged 13.6% of all cases; only 1.8% of procedural cases were flagged by the 17 procedural screens. When we allowed the CSP-O algorithm to scan information from acute hospital readmissions, flag rates rose to 32.8% for general and 8.7% for procedural complications. Controlling for patient and hospital characteristics, flag rates were considerably higher among the very old and at small and for-profit institutions. CONCLUSIONS Whereas several CSP-O findings have construct validity, limitations of claims raise concerns. Regardless of the CSPOs ultimate utility, examining post-discharge experiences to identify inpatient complications remains important as LOSs fall.

Risk adjustment methods can affect perceptions of outcomes

When comparing outcomes of medical care, it is essential to adjust for patient risk, including severity of illness. A variety of severity measures exist, but perceptions of outcomes may differ depending on how severity is defined. We used two severity-adjustment approaches to demonstrate that comparisons of out comes across subgroups of patients can vary dramat ically depending on how severity is assessed. We stud ied two approaches: model 1 was the admission MedisGroups score; model 2 was computed from age and 12 chronic conditions defined by diagnosis codes. Although common summary measures of model per formance (R-squared and C) both suggested that model 1 is a better predictor of in-hospital death than model 2, the weaker model consistently produced more accurate expectations by payer class and age group. Using model 1 for severity adjustment sug gested that Medicare patients did substantially worse than expected and Medicaid patients substantially better. In contrast, use of model 2 found Medicare patients doing as expected, but Medicaid patients far ing poorly.

Using severity measures to predict the likelihood of death for pneumonia inpatients.

AbstractOBJECTIVE: To see whether predictions of patients’ likelihood of dying in-hospital differed among severity methods. DESIGN: Retrospective cohort. PATIENTS: 18,016 persons 18 years of age and older managed medically for pneumonia; 1,732 (9.6%) in-hospital deaths. METHODS: Probability of death was calculated for each patient using logistic regression with age, age squared, sex, and each of five severity measures as the independent variables: 1) admission MedisGroups probability of death scores; 2) scores based on 17 admission physiologic variables; 3) Disease Staging’s probability of mortality model; the Severity Score of Patient Management Categories (PMCs); 4) and the All Patient Refined Diagnosis-Related Groups (APR-DRGs). Patients were ranked by calculated probability of death; 5) rankings were compared across severity methods. Frequencies of 14 clinical findings considered poor prognostic indicators in pneumonia were examined for patients ranked differently by different methods. RESULTS: MedisGroups and the physiology score predicted a similar likelihood of death for 89.2% of patients. In contrast, the three code-based severity methods rated over 25% of patients differently by predicted likelihood of death when compared with the rankings of the two clinical data-based methods (MedisGroups and the physiology score). MedisGroups and the physiology score demonstrated better clinical credibility than the three severity methods based on discharge abstract data. CONCLUSIONS: Some pairs of severity measures ranked over 25% of patients very differently by predicted probability of death. Results of outcomes studies may vary depending on which severity method is used for risk adjustment.

Does Severity Explain Differences in Hospital Length of Stay for Pneumonia Patients

Objectives: In the USA, the role of patient severity in determining hospital resource use has been questioned since Medicare adopted prospective hospital payment based on diagnosis-related groups (DRGs). Exactly how to measure severity, however, remains unclear. We examined whether assessments of severity-adjusted hospital lengths of stay (LOS) varied when different measures were used for severity adjustment Methods: The complete study sample included 18 016 patients receiving medical treatment for pneumonia at 105 acute care hospitals. We studied 11 severity measures, nine based on patient demographic and diagnosis and procedure code information and two derived from clinical findings from the medical record. For each severity measure, LOS was regressed on patient age, sex, DRG, and severity score. Analyses were performed on trimmed and untrimmed data. Trimming eliminated cases with LOS more than three standard deviations from the mean on a log scale. Results: The trimmed data set contained 17 976 admissions with a mean (S.D.) LOS of 8.9 (6.1) days. Average LOS ranged from 5.0–11.8 days among the 105 hospitals. Using trimmed data, the 11 severity measures produced Rsquared values ranging from 0.098–0.169 for explaining LOS for individual patients. Across all severity measures, predicted average hospital LOS varied much less than the observed LOS, with predicted mean hospital LOS ranging from about 8.4–9.8 days. Discussion: No severity measure explained the two-fold differences among hospitals in average LOS. Other patient characteristics, practice patterns, or institutional factors may cause the wide differences across hospitals in LOS.