A comparison of five surveys that identify individuals at risk for airflow obstruction and chronic obstructive pulmonary disease

Background: The predictive characteristics of different screening surveys for the recognition of individuals at risk for airflow obstruction (AFO) have not been evaluated simultaneously in the same population. Purpose: To compare five AFO/COPD screening questionnaires. Methods: 383 individuals completed the Veterans Airflow Obstruction Screening Questionnaire, Personal Level Screener for COPD (VAFOSQ), the 11-Q COPD Screening Questionnaire (11-Q), the COPD Population Screener (COPD-PS) and the Lung Function Questionnaire (LFQ) and performed spirometry. AFO was defined as forced expiratory volume in one second divided by the forced vital capacity (FEV1/FVC) < 0.7, fixed ratio (FR) or FEV1/FVC < lower limit of normal (LLN). The predictive characteristics of the five questionnaires were calculated and non-parametric receiver operating characteristic (ROC) curves estimated by logistic regression. Results: 376 participants completed at least two of the questionnaires and performed technically acceptable spirometry. AFO was present in 102 (27.1%) and 150 (39.9%) based on LLN and FR, respectively. The number of individuals positively selected by the VAFOSQ was 227, PLS 128, 11-Q 236, COPD-PS 217, and LFQ 328. The area under the ROC curves for the questionnaires was between 0.60 and 0.66 (LLN) and 0.58 and 0.66 (FR). Conclusions: Although these screening surveys have acceptable and similar predictive ability for the identification of AFO, their published thresholds lead to substantially different classification rates. The choice of an appropriate threshold for the identification of individuals with possible AFO/COPD should consider the underlying prevalence of AFO/COPD in the target population and the relative costs of misclassifying affected and unaffected cases. Clinical Trial Registration: None. Primary Source of Funding: Veterans Health Administration. Published by Elsevier Ltd.