PURPOSE: The Severity-of-Asthma (SOA) score is a validated disease-specific severity score based on frequency of asthma symptoms, use of systemic corticosteroids, use of other asthma medications, and history of hospitalization/intubation. Our objective was to further validate the SOA score in EXCELS, a 5-year prospective observational cohort study evaluating the long-term clinical safety and effectiveness of omalizumab (Xolair).
METHODS: EXCELS enrolled patients treated with/without omalizumab aged ≥12 years, had moderate to severe persistent allergic asthma, and a positive skin test or in vitro aeroallergen reactivity. We analyzed subjects (N=2878) in the non-omalizumab reference group. The ability of the SOA score to predict adverse asthma outcomes was compared with the Asthma Control Test (ACT), FEV1, and Work Productivity and Impairment Index (WPAI). Five clinical outcomes were analyzed reflecting adverse asthma outcomes: emergency department visits, unscheduled doctor visits, steroid bursts, serious adverse events (SAEs) classified as asthma exacerbations, and SAEs leading to hospitalizations. Data were analyzed using logistic regression with assessments from highest area under receiver-operator-characteristic curves (AUC), and by classification and regression tree (CART) analysis using variable importance rankings.
RESULTS: The SOA score was the most powerful predictor of clinical outcomes. The logistic regression AUC values for the SOA score ranged from a minimum of 0.68 for unscheduled visits to a maximum of 0.77 for SAE exacerbations. In the CART analysis, the SOA was consistently the most important variable, with the next best variable being a minimum of 26% as important for steroid bursts, to a maximum of 64% as important for SAE exacerbations. For some endpoints, the SOA score was a stand-alone predictor, though for others it needed to be coupled with ACT, WPAI, and FEV1 to optimally predict the outcome.
CONCLUSION: In the EXCELS cohort, the SOA score was the most powerful predictor in the database for all 5 adverse clinical outcomes, both when using logistic regression and CART methods.
CLINICAL IMPLICATIONS: The SOA score is a valid measure of asthma severity that has potent predictive capacity.
DISCLOSURE: Mark Eisner, Employee Dr. Eisner was a full time faculty member at University of California, San Francisco when the study was designed and analyzed. Beginning March 1, 2010, he became a full time employee of Genentech, Inc and retains a volunteer faculty position at UCSF. Dr. Yegin and author Trzaskoma are full time employees of Genentech, Inc.; No Product/Research Disclosure Information