Hypercapnia, Prognostic Nutritional Index and Length of Stay in Acute Exacerbation of COPD: A Two-Variable Admission Framework

Background/Objectives: Established AECOPD prognostic tools (DECAF, BAP-65, PEARL) predict mortality or readmission rather than length of stay (LOS), and no admission-based instrument specifically targets prolonged hospitalization. We tested whether admission PaCO2 and the Prognostic Nutritional Index (PNI), reflecting ventilatory failure and nutritional–immune reserve, are independently associated with prolonged LOS and examined their interaction. Methods: In this single-center retrospective cohort, 213 adults hospitalized exclusively for AECOPD were analyzed after excluding concomitant pneumonia, pulmonary embolism, decompensated heart failure, and in-hospital deaths. Prolonged hospitalization was pre-specified as LOS > 7 days. Multivariable logistic regression evaluated admission PaCO2 (per +10 mmHg) and PNI (per +5 units) with a PaCO2 × PNI interaction; continuous LOS was modeled by Gamma regression. Discrimination was compared with DECAF using DeLong’s test. Results: Prolonged hospitalization occurred in 83 patients (39.0%). Admission PaCO2 was independently associated with prolonged LOS (OR 1.52, 95% CI 1.25–1.88; p < 0.001), and PNI showed a borderline association (OR 0.84, 95% CI 0.71–1.00; p = 0.049); their interaction was significant but exploratory (OR 1.16, 95% CI 1.02–1.32; p = 0.025). In Gamma regression, PaCO2 (RR 1.18 per 10 mmHg) and PNI (RR 0.92 per 5 units) remained associated with LOS. The two-variable model achieved an AUC of 0.682, showing discrimination similar to DECAF in this cohort (AUC 0.695; DeLong p = 0.76), with optimism-corrected AUC 0.672 and calibration slope 0.96. Within moderate hypercapnia (PaCO2 45–60 mmHg), the prolonged-LOS rate was 44.4% in low-PNI versus 15.6% in high-PNI patients. Conclusions: In this single-center retrospective cohort of AECOPD patients surviving to discharge, admission PaCO2 and PNI were jointly associated with prolonged hospitalization, reflecting acute ventilatory burden and nutritional–immune reserve. Using only two admission inputs, the framework showed discrimination similar to DECAF without meaningful reclassification gain (IDI −0.02; NRI 0.02). Given only moderate discrimination (AUC ~ 0.68), external validation is required before clinical use, with the main practical value likely in complementary stratification within moderate hypercapnia.