Commonly Used Patient-Reported Outcomes Do Not Improve Prediction of COPD Exacerbations: A Multicenter 4½ Year Prospective Cohort Study.

Abstract

Several cross-sectional and short-term studies suggest that patient-reported outcomes (PROs) may be useful to help predict COPD exacerbations, but evidence from long-term prospective cohort studies is lacking. Our aim was to assess if PROs add to the prediction of exacerbations. Four hundred eight primary care patients with COPD participated in the multicenter prospective International Collaborative Effort on Chronic Obstructive Lung Disease: Exacerbation Risk Index Cohorts (ICE COLD ERIC) study. PROs included the Chronic Respiratory Disease Questionnaire, the Hospital Anxiety and Depression Scale, the Feeling Thermometer, and the LASA Physical Activity Questionnaire (LAPAQ). Exacerbations during 4½ years were adjudicated by experts. We used negative binomial regression to assess the association between PROs and exacerbations while considering age, sex, airflow obstruction, previous exacerbations, and smoking status. We then assessed if adding PROs improved the predictive properties of widely used indices. Lower levels of dyspnea, fatigue, and anxiety and higher levels of physical activity were associated with a lower risk of exacerbations (adjusted incidence rate ratio [IRR], 0.90; 95% CI, 0.84-0.97; P = .01; IRR, 0.93; 95% CI, 0.86-1.00; P = .04; IRR, 1.03; 95% CI, 1.00-1.06; P = .03; IRR, 0.93; 95% CI, 0.87-0.99; P = .04, respectively). The best prediction was when LAPAQ was added to the BMI, airflow obstruction, dyspnea, and previous exacerbations (BODEx) index (area under the curve Δ+0.02, from 0.72-0.74). However, this increase was not clinically relevant (net reclassification improvement, 0.03; 95% CI, -0.03 to 0.08). Some domains of PROs are independently associated with exacerbations but do not lead to clearly improved prediction when added to established indices to predict exacerbations. There is still much room for improvement in the prediction of exacerbations. ClinicalTrials.gov; No.: NCT00706602.