Factor analysis in predominantly severe COPD: Identification of disease heterogeneity by easily measurable characteristics

Background The clinical and demographic variables defining the heterogeneity of chronic obstructive pulmonary disease (COPD) are unclear. A post-hoc analysis of five randomised studies in patients with a history of previous exacerbations examined the clinical and demographic characteristics describing moderate-to-very-severe COPD. Methods Factor analysis was performed on all continuous baseline demographic and clinical data, without variable selection. Analyses were based on the full cohort and on stratifications by pack-years smoked, smoking status, gender, and comorbidities; patient exacerbation history was analysed in two of the five studies. Findings 6162 COPD patients were evaluated (70% male; 40% current smokers; mean pre-bronchodilator forced expiratory volume in 1 s [FEV₁] 35.2% predicted). Baseline clinical and demographic variables loaded differentially on six factors with minimal overlap, explaining 60.4% of the heterogeneity: 1) symptoms (cough, dyspnoea, sleep disturbance), health status, reliever use; 2) pre-bronchodilator FEV₁, FEV₁/forced vital capacity, morning peak expiratory flow (PEF), body mass index (BMI); 3) blood pressure; 4) age, months since first COPD symptoms; 5) PEF variability; 6) pulse, FEV₁ reversibility. Most factors loaded similarly in stratified and exacerbation analyses. BMI loaded with reversibility in females, and with age and months since first COPD symptoms in ex-smokers. Exacerbations loaded to factor 6. Interpretation Readily available data can explain ∼60% of COPD heterogeneity in a large dataset of predominantly severe COPD patients. Factors were robust over determinants of disease outcome; gender, smoking status, pack-years smoked, and comorbidities. The main factors were largely unchanged by adding exacerbations. Only BMI loaded to other factors.