Development and testing of a systemic lupus-specific risk adjustment index for in-hospital mortality

Valid comparison of patient outcomes among hospitals requires adjustment for differences in the severity of patients' illness. Disease-specific indexes of severity of illness may permit more accurate risk adjustment than generic indexes. The objective of this study was to develop a systemic lupus-specific risk adjustment index for in-hospital mortality, and to compare its performance to that of the generic Charlson index.

A systemic lupus-specific risk adjustment index was developed using discharge abstract data from a 50% random sample (n = 4994) of patients with systemic lupus erythematosus (SLE) hospitalized on an emergent or urgent basis in California from 1991 to 1994 (n = 9989). The index was tested on the remaining members of the sample. Candidate variables for the index were the diagnoses included in the original Charlson index, and nephritis, chronic renal failure, pericarditis, pleuritis, psychosis, seizures, hemolytic anemia, and thrombocytopenia. Multivariate logistic regression analysis was used to identify the set of variables that best differentiated those patients who died in the hospital from those who survived, and to provide the weights for construction of the index.

In the derivation set of patients, the SLE-specific index accurately predicted in-hospital mortality (area under the receiver operating characteristic curve c = 0.79). In the test set, the SLE-specific index (c = 0.72) was similar to the original Charlson index (c = 0.74) in its ability to predict in-hospital mortality (p = 0.32). However, the SLE-specific index accounted for substantially more variation in the risk of mortality among patients (R2 = 0.069) than did the Charlson index (R2 = 0.036). Use of the SLE-specific index rather than the Charlson index for risk adjustment did not alter the association between hospital experience and the probability of in-hospital mortality. Results were similar in the subgroups of patients with emergent hospitalizations and those with emergent hospitalizations due to SLE.

The SLE-specific risk adjustment index developed from diagnoses recorded in administrative discharge abstracts performed similarly to the generic Charlson index in correctly classifying mortality outcomes, but the SLE-specific index stratified patients by their level of risk of mortality better than the Charlson index. Adjustment for SLE-specific risks of mortality did not alter the association between hospital experience and the risk of in-hospital mortality.