Abstracts

Rationale: In health care research, measuring the comorbidity profile of study subjects is essential to the control of confounding when assessing mortality as an outcome. Several comorbidity risk adjustment indices such as the Charlson/Deyo and Elixhauser have been developed utilizing acute care and cancer patients. While these methods have been widely used for risk adjustment, they do not include some important epilepsy related comorbidities. Our goal was to develop an epilepsy specific risk adjustment comorbidity index. Methods: The following administrative databases were linked (from 1996-97 to 2003-04): a provincial health care insurance plan registry, a hospital discharge abstract database, an emergency room visits database, and a physician claims database in a large Canadian health region (Calgary). Mortality data were obtained from the vital statistics database up until December 2005 to account for delays in mortality coding. A case was defined as anyone who had 2 physician claims or 1 hospitalization or 1 emergency room visit over two years coded for epilepsy. A washout period of at least 3 years (with no epilepsy coding before the index date) was used to ensure that the majority of epilepsy cases were newly diagnosed. A total of 33 comorbidities were assessed for inclusion in the epilepsy specific comorbidity index. A parsimonious model was selected using stepwise logistic regression leaving 14 comorbidities in the final model. Each comorbidity was then assigned a value of 1 to 6 based on the hazard ratio. A total prognostic score was calculated for all eligible subjects using the new epilepsy specific comorbidity index and the Charlson index and crude mortality was compared. Results: 7,253 subjects met the epilepsy case definition. The mean age was 38 years (range: 0.03-96) with males representing 52% of cases. The mortality rate among this group of epilepsy patients with primarily newly diagnosed epilepsy was 7.9%. High rates of chronic pulmonary disease (20.3%), depression (28.2%) and hypertension (19.6%) were noted. Crude mortality among patients was similar for each prognostic score using either the epilepsy specific comorbidity index or the Charlson/Deyo index. However in most prognostic score categories crude mortality scores were higher using the epilepsy specific comorbidity index than the Charlson/Deyo index. The linear slope for crude mortality by prognostics score was also higher for the new epilepsy specific comorbidity index vs. the Charlson/Deyo index. Conclusions: A new comorbidity index for epilepsy specifically designed to include clinically relevant and statistically significant conditions provided better discrimination of crude mortality in a population-based group of epilepsy patients. A disease specific comorbidity index should facilitate better control of confounding, better prediction of mortality and may also be of value for predicting health resource use in epilepsy. However, this new epilepsy specific comorbidity index requires further validation in additional populations with longer term follow up.