A Risk Prediction Model for Prolonged Length of Stay in Patients with Acute Exacerbations of Chronic Obstructive Pulmonary Disease: A Retrospective Study of 225 Patients in a Single Center in Kunming, China

• Published in: Medical science monitor 2022-02, Vol.28, p.e934392-e934392
• Main Authors: Yang, Lifen, Li, Meihua, Shu, Jingkui, Yang, Yanxia, Huang, Qian
• Format: Article
• Language: English
• Subjects: Acute Disease; Adrenal Cortex Hormones - administration & dosage; Aged; Blood Platelets; China; Clinical Research; Female; Humans; Hydrogen-Ion Concentration; Length of Stay - statistics & numerical data; Male; Middle Aged; Procalcitonin - blood; Pulmonary Disease, Chronic Obstructive - blood; Pulmonary Disease, Chronic Obstructive - physiopathology; Retrospective Studies; Risk Assessment
• ISSN: 1643-3750; 1234-1010; 1643-3750
• DOI: 10.12659/MSM.934392

BACKGROUND We aimed to develop an effective prediction model of prolonged length of stay (LOS) in patients with acute exacerbations of chronic obstructive pulmonary disease (AECOPD). MATERIAL AND METHODS We systematically enrolled 225 patients admitted for AECOPD to our hospital and divided them into a normal LOS group (≤7 days) and prolonged LOS group (>7 days). To analyze differences in laboratory data at different times, 3 logistic regression models were established. To develop the prediction model, all variables with statistical significance were included in the model. The area under the curve (AUC) was used to evaluate discrimination, and the Hosmer-Lemeshow test was used to assess the calibration of the model. RESULTS Factors found to be independently associated with the increased risk of prolonged LOS included the use of corticosteroids during hospitalization, elevated HCO₃⁻, decreased pH, and reductions in platelets (PLTs) and procalcitonin (PCT) between the fourth and first day of hospitalization. The risk prediction model including these factors had an AUC of 0.795, suggesting the good discrimination of our model. The Hosmer-Lemeshow test also showed good calibration of the model, which confirmed its good predictive performance. CONCLUSIONS A clinical prediction model was developed with good predictive performance, which could help clinicians identify patients with a higher risk of prolonged LOS, help shorten hospital stay, reduce the disease burden of patients, and improve the outcomes of AECOPD.