PMC patient severity scale: derivation and validation - Patient Management Category

Health Services Research, August, 1994 by Wanda W. Young, Susan Kohler, Jeanne Kowalski

A two-part strategy was designed to focus first on clinical and empirical distinctions among diseases (PMCs), and then on an assessment of the impact of multiple diseases on a particular patient's hospitalization. The first stage of analysis resulted in the assignment of each PMC to one of four levels of severity. In the second stage, this scale was expanded to a seven-point scale and redefined to incorporate the effects of comorbidity and complications. Thus, the final PMC Severity Scale measures the severity of the patient's illness episode rather than the severity of each individual disease.

RESULTS

Assignment of Severity Levels to PMCs

The first step in the process of deriving a relative numerical value to measure the severity of each patient's hospitalization was to focus on diseases as if they were managed singly, that is, without regard to comorbidity or other patient characteristics. Thus, a rank of 1 to 4 was assigned to each of the 830 PMCs, based on the subjective clinical judgment of the nurse researcher/author. Level 1 represented the lowest expected severity, that is, those patients who could potentially be treated in an ambulatory setting as well as those with other minor medical problems. Level 4 represented the highest expected severity and life-threatening situations. This was a way to derive a preliminary scale that could be analyzed empirically.

Since severity is not directly quantifiable, surrogate outcomes (death rates and LOS) were used to assess and adjust this initial scale. Most investigations of severity classify disease categories by average death rate to reflect the probability of death and risk of organ failure. Since relatively few diseases typically result in death, however, in this analysis the LOS of each disease-specific hospital stay was also chosen as a surrogate for morbidity.

Average lengths of stay for single diseases (PMCs) were derived from the all-payer 1989 Maryland statewide database of approximately 600,000 patient discharge records. Patients who were transferred at discharge to other short-term hospitals, were discharged against medical advice, or had a hospital stay greater than 105 days were excluded from these calculations, leaving 568,762 cases in the analysis.

Recall that a Patient Management PATH, associated with each PMC, provides physician-specified information about the nature of the types of admissions associated with the particular disease (emergent, urgent, or elective) as well as physician-specified expectations regarding the need for diagnostic and therapeutic interventions and continuous monitoring. These clinical expectations, combined with data on the probability of hospital death and LOS associated with each PMC, were used to adjust the initial severity level assignments made for each PMC and to ensure that these assignments were made consistently across diseases and body systems. Although the resultant disease-specific severity scale (from 1 to 4) explained approximately 28 percent of the variation in LOS for single-disease patients in the developmental database, this was only an intermediate step toward deriving an overall score reflecting the severity of all of the clinical conditions managed in a patient's hospitalization.