Severity Adjustment

History

The first attempts to measure patient severity took place in the 1970s. Later, particular attention was paid to severity adjustment in 1983, when the nation's Medicare program adopted a Diagnosis Related Group (DRG)–based prospective payment system (PPS) for hospitals. Hospitals were concerned that the new system would not pay for the provision of care to “sicker” patients. There was also concern about the accuracy of the DRG concept, since it assigns patients based mainly on principal diagnosis codes. Since compensation levels were at stake, critics argued that diagnosis severity could be exaggerated by hospitals in an attempt to improve their “bottom lines.” These issues prompted debate over the use of code-based versus medical record surveys to assess patient complexity; thus, there were considerable efforts by developers of severity measures to explore and test a number of systems, such as disease staging, severity scores of Patient Management Categories (PMCs), and All Patient Refined Diagnosis Related Groups (APR-DRGs).

Selection of Performance Outcome

In the process of severity adjustment, healthcare managers and researchers must decide on a performance outcome of interest. Outcome measures that could be attributable to healthcare quality include morbidity, mortality, readmission rates, complication rates, functional status, and patient satisfaction. By far, mortality has been the most frequently used generic measure of hospital performance. Advantages of using mortality include the wide availability of this information, its clearly definable end point, and its importance. Further decisions about the mortality outcome measure may include whether to use in-hospital mortality, 10-day mortality, 30-day mortality, or 1-year mortality.

While a generic measure of hospital performance such as all-cause mortality may be used, it may be of greater interest to evaluate a disease- or procedure-specific mortality (e.g., in-hospital mortality among patients with congestive heart failure or in-house mortality after abdominal aortic aneurysm repair). Instead of mortality, healthcare managers and researchers may be interested in performance indicators such as the number of cesarean deliveries among high-risk women, urinary-catheter-associated infections, or postoperative sepsis. Sometimes, selection of performance indicators may be hampered by small sample size, leaving inadequate statistical power to properly assess priority outcomes.

...