Statistical Analysis of Proposed Pediatric Asthma Screening Survey
BACKGROUND: Poor asthma control is responsible for considerable morbidity and mortality among children (1). Current pharmacotherapy can suppress exacerbations of asthma symptoms. Thus, proper treatment of asthma is imperative in limiting the toll of this disease process on individuals as well as society. Treatment protocols tend to be based on measurement of asthma severity (3), but there are currently no widely accepted guidelines defining efficacy of treatment (i.e.- asthma control). OBJECTIVE: This study aimed to assess the construct validity and reliability of an asthma survey among a group of known asthmatics. Furthermore, we intended to determine the level of correlation between survey responses and asthma severity amongst survey participants as well as to discern the ability of the survey to discriminate between mild persistent, moderate persistent and severe persistent asthma. METHODS: Surveys from 207 parents/guardians of children aged 5-17 with physician-diagnosed asthma were evaluated for construct validity using Principal Components Factor Analysis. Reliability was assessed via Cronbach's alpha coefficient scale. Severity/response correlations were tested by Chi-square exact tests and the strength of each relationship was assessed using Spearman's correlation. Discriminating ability was analyzed by ROC curve, sensitivity, specificity and odds ratio. RESULTS: Construct validity testing showed that the scale is unidimensional with a Cronbach's alpha coefficient of 0.8076, indicating a high degree of reliability. Significant associations between asthma severity and each question were found, indicating that more severe asthmatics reported significantly greater symptom frequency (p-value range <0.001 - 0.019, Spearman's range = 0.152 - 0.396). ROC analysis yielded an area under the curve of 0.728. Analysis of the ROC curve indicated an optimal cutoff score of =6 to indicate moderate-to-severe asthma. This cutoff yielded a sensitivity and specificity of 65.2% and 70.2%, respectively. Odds ratio was 4.407 (95% CI of 2.366 - 8.207). CONCLUSION: Our results indicate that, among asthmatics, the survey is valid and reliable. We also noted more frequent symptoms as severity increased, indicating sub-optimal control among more severe asthmatics. Finally, the ability of the survey to predict asthma severity is not supported as the survey seems to assess asthma control, with higher scores indicating poorer control.