Purpose: There is an unmet clinical need for an accurate and objective diagnostic tool for early detection of acute exacerbation of chronic obstructive pulmonary disease (AECOPD). DETECT (NCT03556475) was a multicenter, observational, cross-sectional study aiming to develop and validate multivariable prediction models for AECOPD occurrence and severity in patients with chronic obstructive pulmonary disease (COPD) in China.
Patients and Methods: Patients aged ≥ 40 years with moderate/severe COPD, AECOPD, or no COPD were consecutively enrolled between April 22, 2020, and January 18, 2021, across seven study sites in China. Multivariable prediction models were constructed to identify AECOPD occurrence (primary outcome) and AECOPD severity (secondary outcome). Candidate variables were selected using a stepwise procedure, and the bootstrap method was used for internal model validation.
Results: Among 299 patients enrolled, 246 were included in the final analysis, of whom 30.1%, 40.7%, and 29.3% had COPD, AECOPD, or no COPD, respectively. Mean age was 64.1 years. Variables significantly associated with AECOPD occurrence (P < 0.05) and severity (P < 0.05) in the final models included COPD disease-related characteristics, as well as signs and symptoms. Based on cut-off values of 0.374 and 0.405 for primary and secondary models, respectively, the performance of the primary model constructed to identify AECOPD occurrence (AUC: 0.86; sensitivity: 0.84; specificity: 0.77), and of the secondary model for AECOPD severity (AUC: 0.81; sensitivity: 0.90; specificity: 0.73) indicated high diagnostic accuracy and clinical applicability.
Conclusion: By leveraging easy-to-collect patient and disease data, we developed identification tools that can be used for timely detection of AECOPD and its severity. These tools may help physicians diagnose AECOPD in a timely manner, before further disease progression and possible hospitalizations.
Keywords: chronic obstructive pulmonary disease, acute exacerbation, prediction model, diagnosis