Association between the prognostic nutritional index and frequent acute exacerbations in patients with COPD: development of a predictive model

The present study sought to evaluate the relationship between the Prognostic Nutritional Index (PNI) and frequent acute exacerbations in patients with chronic obstructive pulmonary disease (COPD), and to establish a PNI-based nomogram for predicting the risk of recurrent exacerbations in this population. This retrospective study included 976 patients diagnosed with COPD at the Affiliated Hospital of Guangdong Medical University between June 2018 and December 2019. Logistic regression analyses were conducted to determine independent predictors of frequent acute exacerbations. Candidate variables were selected using a bidirectional Wald chi-square approach and subsequently incorporated into a predictive nomogram. Model fit was assessed using Akaike Information Criterion (AIC) and deviance statistics, and multicollinearity was evaluated with generalized variance inflation factor (GVIF).The logistic regression model demonstrated good fit (AIC = 763.64; deviance reduction = 112.30, df = 7, P < 0.001). No significant multicollinearity was observed [all GVIF (1/(2 × Df)) < 2 ].No significant multicollinearity was detected, supporting the stability of the regression model. The model’s discrimination and calibration were further examined, followed by internal validation. For external validation, a prospective cohort of 487 COPD patients admitted to the Second Affiliated Hospital of Guangdong Medical University between March 2024 and December 2024 was enrolled. A total of 976 patients were analyzed, including 683 in the training cohort and 293 in the internal validation cohort. Multivariate logistic regression analysis revealed that PNI and hyperlipidemia are protective factors in patients with obstructive pulmonary disease, while K + , HDL-C, total number of hospitalizations (5–10 times; > 10 times), and diabetes were identified as independent risk factors for frequent acute exacerbations in patients with chronic obstructive pulmonary disease. The final nomogram incorporated PNI, serum potassium (K +), high-density lipoprotein cholesterol (HDL-C), diabetes, hyperlipidemia, and total hospitalization frequency. The model demonstrated satisfactory discriminative ability and calibration performance. In the training cohort, the optimal cut-off value of PNI for predicting frequent AECOPD was 42.93. The area under the receiver operating characteristic curve (AUC) (95% CI) was 0.74 (0.69–0.78) in the training cohort, 0.70 (0.63–0.76) in the internal validation cohort, and 0.73 (0.68–0.78) in the external validation cohort. The study results demonstrate that the PNI-based column chart model exhibits excellent discriminative capacity, calibration performance, and clinical net benefit. PNI serves as a protective factor, while hospitalization frequency is a significant risk factor, providing a scientific basis for individualized risk assessment and comprehensive intervention in COPD patients. The complex interactions of metabolic-related factors warrant further investigation.