Abstract:

Objective To systematically evaluate the risk for readmission in patients with chronic obstructive pulmonary disease (COPD) and provide references for the construction and optimization of prediction model.Methods The literatures on the risk prediction model for COPD readmission were independently screened in China National Knowledge Infrastructure (CNKI), VIP, WanFang Data, Cochrane Library, PubMed, Embase databases from database inception to November 22, 2023 by two researchers. After extracting data, the bias risk and applicability of the models were evaluated using the PROBAST tool. Results Twelve cohort studies representing 21 models were finally included. Eleven studies reported the area under a receiver operator characteristic (ROC) curve (AUC) and one study for the C-index. The AUC ranged from 0.603 to 0.917, with AUC>0.7 for 16 models. Six studies conducted model calibration and eight studies for internal or external validation. The overall applicability of the 12 studies was good, but with a high risk of bias, mainly in the analysis domain. The included studies had significant differences in the predictive factors, with the most common predictive factors of lung function indicators, Charlson comorbidity index, times of hospitalization due to a history of acute exacerbation during the previous year, eosinophil levels, and inhaled drug therapy. Conclusion The performance of the included models varied greatly, with good applicability but high risk of bias. Due to the incomplete screening method, and there were significant differences in predictive factors of the included studies. Future prediction models should focus on lung function, Charlson comorbidity index, times of hospitalization due to a history of acute exacerbation during the previous year, eosinophil levels, and inhaled drug therapy.

Key words: pulmonary disease, chronic obstructive, rehospitalization, prediction model, systematic review