Search

Background: COPD is underdiagnosed, often uncovered during exacerbations or after significant loss of lung function. We are developing a new method (simple questionnaire + peak expiratory flow [PEF]) for identifying patients needing spirometric evaluation for clinically significant COPD (FEV1 Purpose: This study used random forests (RF) to identify the best, smallest set of questions (with and without PEF) capable of case identification. Methods: Prospective, cross-sectional, multi-site, case-control study. Subjects: ≥40 years from US pulmonary and primary care clinics. Cases (clinically significant COPD): ≥1 AECOPD past 12 months or FEV1 60% pred) and AECOPD-free ≥12 months or no COPD. Subjects completed questionnaires, PEF, spirometry (if not on record). Results: N=346; 186 Cases (184 with PEF), 160 Controls. Mean (SD) age=62.7 (10.1) years; 55% female; 86% white; 41% ≤high school; 16% never smoked. Cases: mean (SD) FEV1% pred=42.5(14.2); Controls=82.5(15.7). RF case-control sensitivity/specificity estimates for the final 5-item questionnaire: 83.5/69.7; questionnaire followed by score-driven use of PEF: 90.2/78.8%. Case-no COPD with score-driven PEF: 90.2/92.0. Questions ask about exposure, breathing, tire easily, and acute respiratory illness, and do not ask about smoking, cough, or sputum. Conclusions: These 5 questions + selective use of PEF may be an effective and efficient approach for identifying patients needing diagnostic evaluation for COPD.