OBJECTIVE: Severity adjustment is an oft-cited requirement when comparing physicians or medical delivery systems. Each application of severity adjustment, however, has to be tested to validate the need, the method, and its value. We examined the value of severity adjustment for identifying physician outliers when studying length of stay in the hospital. DESIGN: We compared the placement of physicians in an outlier category using a severity-adjusted average length of stay (SLOS) index with their placement using the unadjusted average length of stay (ALOS). Changes in placement of the list were validated by the utilization review coordinators. SETTING: A 614-bed tertiary-care university teaching hospital. SUBJECTS: We analyzed 11,146 discharges from 138 physicians in 1992. RESULTS: The mean ALOS +/- standard deviation was 9.05 + 4.50 days, and the SLOS Index was 7.56 +/- 3.06. There were 120 inliers, 6 high outliers, and 12 low outliers by the ALOS method. Using the SLOS index, 27 of 138 physicians had their categories changed from inlier to outlier or from outlier to inlier. The difference in group changes was more significant for those going from outlier to inlier status (8/120 vs 6/18; P < .001). The patients of the six physicians whose status changed from outlier to inlier status were sicker, as indicated by the comorbidity, complications, and manifestations of disease processes score. The utilization reviewers validated the status changes in 8 of 14 instances. CONCLUSIONS: Severity-adjusted length of stay by the SLOS index appears to provide a more accurate measure than the unadjusted ALOS. The changes, however, were small. It is not clear that the added effort is worthwhile.
OBJECTIVE: Severity adjustment is an oft-cited requirement when comparing physicians or medical delivery systems. Each application of severity adjustment, however, has to be tested to validate the need, the method, and its value. We examined the value of severity adjustment for identifying physician outliers when studying length of stay in the hospital. DESIGN: We compared the placement of physicians in an outlier category using a severity-adjusted average length of stay (SLOS) index with their placement using the unadjusted average length of stay (ALOS). Changes in placement of the list were validated by the utilization review coordinators. SETTING: A 614-bed tertiary-care university teaching hospital. SUBJECTS: We analyzed 11,146 discharges from 138 physicians in 1992. RESULTS: The mean ALOS +/- standard deviation was 9.05 + 4.50 days, and the SLOS Index was 7.56 +/- 3.06. There were 120 inliers, 6 high outliers, and 12 low outliers by the ALOS method. Using the SLOS index, 27 of 138 physicians had their categories changed from inlier to outlier or from outlier to inlier. The difference in group changes was more significant for those going from outlier to inlier status (8/120 vs 6/18; P < .001). The patients of the six physicians whose status changed from outlier to inlier status were sicker, as indicated by the comorbidity, complications, and manifestations of disease processes score. The utilization reviewers validated the status changes in 8 of 14 instances. CONCLUSIONS: Severity-adjusted length of stay by the SLOS index appears to provide a more accurate measure than the unadjusted ALOS. The changes, however, were small. It is not clear that the added effort is worthwhile.