Severity of Asthma Score Predicts Clinical Outcomes in Patients With Moderate to Severe Persistent Asthma

Mark D. Eisner, MD, MPH; Ashley Yegin, MD; Benjamin Trzaskoma, MS


CHEST. 2012;141(1):58-65. 

In This Article

Abstract and Introduction


Background: The severity of asthma (SOA) score is based on a validated disease-specific questionnaire that addresses frequency of asthma symptoms, use of systemic corticosteroids, use of other asthma medications, and history of hospitalization/intubation for asthma. SOA does not require measurements of pulmonary function. This study compared the ability of SOA to predict clinical outcomes in the EXCELS (Epidemiological Study of Xolair [omalizumab]: Evaluating Clinical Effectiveness and Long-term Safety in Patients with Moderate to Severe Asthma) patient population vs three other asthma assessment tools. EXCELS is a large, ongoing, observational study of patients with moderate to severe persistent asthma and reactivity to perennial aeroallergens.
Methods: Baseline scores for SOA, asthma control test (ACT), work productivity and impairment index-asthma (WPAI-A), and FEV1 % predicted were compared for their ability to predict five prespecified adverse clinical outcomes in asthma: serious adverse events (SAEs) reported as exacerbations, SAEs leading to hospitalizations, the incidence of unscheduled office visits, ED visits, and po or IV corticosteroid bursts related to asthma. Logistic regression analysis, area under receiver operating characteristic curves (AUCROCs), and classification and regression tree (CART) analysis were used to evaluate the ability of the four tools to predict adverse clinical outcomes using baseline and 1-year data from 2,878 patients enrolled in the non-omalizumab cohort of EXCELS.
Results: SOA was the only assessment tool contributing significantly in all five statistical models of adverse clinical outcomes by logistic regression analysis (full model AUCROC range, 0.689–0.783). SOA appeared to be a stand-alone predictor for four of five outcomes (reduced model AUCROC range, 0.689–0.773). CART analysis showed that SOA had the greatest variable importance for all five outcomes.
Conclusions: SOA score was a powerful predictor of adverse clinical outcomes in moderate to severe asthma, as evaluated by either logistic regression analysis or CART analysis.


Asthma remains a major public health concern and is associated with significant loss of productivity, increased health-care use, and substantial costs.[1] From 2002 to 2007, the annual US economic cost of asthma totaled $56.0 billion, including $50.1 billion for direct health care.[1] Patients with severe or difficult-to-treat asthma have been shown to account for a large percentage of the morbidity, mortality, and costs.[2,3] Preventing asthma-specific morbidity and mortality depends on properly identifying and treating high-risk patients.[4,5]

Asthma severity and asthma control are related but separate clinical constructs.[6] Asthma control includes the domains of impairment (measured by the asthma control test [ACT] and FEV1) and risk of future adverse health outcomes (measured by the frequency of exacerbations). Poorer levels of asthma control have been associated with increased risk of severe asthma-related events;[6–9] however, assessment of individual components of asthma control may not accurately predict adverse asthma outcomes.[10] Although severe exacerbations are more common in patients with poorly controlled asthma, exacerbations also occur in patients with well-controlled asthma or asthma that is mild in severity. Moreover, certain asthma medications provide short-term control of symptoms and lung function without apparent effects on inflammation or airway hyperreactivity.[11]

The severity of asthma (SOA) score was developed as a tool to identify asthma patients at risk for adverse clinical outcomes. Previous efforts have established the reliability,[8] concurrent validity,[5,6,8] and predictive validity[7,8] of the 13-item disease-specific SOA score that was designed for use in survey research. A notable feature of the SOA instrument is that it does not require measurement of pulmonary function. The score is based on the frequency of current asthma symptoms (daytime or nocturnal), use of systemic corticosteroids, use of other asthma medications (besides systemic corticosteroids), and history of hospitalization and intubation for asthma.[7,8] The SOA score has been designed as a composite score and does not have statistically validated subscales based on the individual score components. Although the SOA instrument has been validated in patient populations treated by both pulmonary and allergy specialists[12,13] as well as family practice physicians,[14] current lung function data were not available for many of the patients examined in these prior validation studies.

The Epidemiologic Study of Xolair (omalizumab): Evaluating Clinical Effectiveness and Long-term Safety in Patients with Moderate to Severe Asthma (EXCELS) is a large, ongoing prospective observational study in patients with moderate to severe asthma and reactivity to a perennial aeroallergen.[15] It was initiated to primarily monitor the safety of omalizumab over a 5-year period and included a large referent group of patients who had not been treated with omalizumab. EXCELS provides an opportunity to longitudinally validate the SOA score in a cohort of patients with severe asthma in whom spirometry was performed concurrently with the ACT[16] and the asthma-specific adaptation of the work productivity and activity impairment-asthma instrument (WPAI-A).[17,18]

The objective of the current analysis was to validate the SOA score in EXCELS and compare it with the ACT, WPAI-A, and FEV1 percent predicted for the ability to predict five prespecified adverse clinical outcomes during a 1-year follow-up. These validated, clinically relevant assessment tools were chosen for comparison in order to broadly characterize the current impact of asthma on symptoms, lung function, activity limitation, and disability related to asthma.[16,19,20] Two distinct statistical methods were applied in modeling the data: logistic regression analysis and a classification and regression tree (CART) analysis. CART,[21,22] also known as binary recursive partitioning, has been used in clinical studies to efficiently identify the major factors associated with a clinical outcome.[23,24] CART analysis is complementary to logistic regression analysis, but identifies interactions among clinical variables without making assumptions about the data distribution; it is also capable of producing an intuitive visual representation of clinical interactions.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: