Validation of self-reported chronic obstructive pulmonary disease in a cohort study of nurses

Abstract

Chronic obstructive pulmonary disease (COPD) is the fourth leading cause of death, and its prevalence is increasing; however, few strategies are available for disease prevention or modification besides smoking interventions. To facilitate examination of modifiable risk factors for COPD in the Nurses' Health Study, the authors validated a questionnaire-based definition of COPD in a subset of this US cohort. Participants were surveyed biennially about lifestyle factors, including smoking, since 1976 and physician diagnosis of COPD since 1988. Self-reported cases were defined as reporting COPD on both the original (1988-1996) and supplemental (1998) questionnaires. The authors requested medical records for a 10% random sample of 2,790 cases and reviewed these records in a systematic, blinded fashion. Validated cases required obstructive spirometry, emphysema on chest radiograph or computed tomography, or physician diagnosis. COPD was confirmed for 78% of 273 cases. Spirometry or radiographic results were available for 84%; when available, mean forced expiratory volume in 1 second was 51% predicted (standard deviation, 19). Applying these results to a hypothetical cohort, the authors estimated the degree to which disease misclassification biases relative risks toward the null value, confirming that questionnaire-based COPD research should focus on minimizing false positives rather than false negatives. In conclusion, COPD can be studied in large, questionnaire-based cohorts of health professionals.