Construction of nomogram for predicting pulmonary hypertension on echocardiography in chronic obstructive pulmonary...

To identify independent predictors of pulmonary hypertension (PH) on echocardiography in chronic obstructive pulmonary disease (COPD) patients during acute exacerbations.

We retrospectively enrolled 151 patients hospitalized for acute COPD exacerbation. Patients were divided by echocardiography into COPD with PH (n=75) and COPD without PH (n=76) groups. LASSO regression screened 17 laboratory indicators to identify the optimal variable subset. Multivariate logistic regression determined independent PH predictors, and a nomogram was constructed. The discrimination and calibration of model were assessed.

LASSO regression analysis screened out 6 optimal variables (age, albumin, cystatin C, red blood cell distribution width (RDW), pH, and CD4/CD8 ratio). Multivariate logistic regression analysis revealed that serum albumin, cystatin C, RDW, pH, and age were independent predictors of PH. The nomogram constructed using these six variables (age, serum albumin, cystatin C, RDW, CD4/CD8 ratio, and pH) demonstrated a C-index of 0.800 (95 % CI: 0.731-0.869). The calibration curve showed that the predicted probability of the model was in good agreement with the actual observed values.

The nomogram constructed based on 6 variables has good predictive efficacy for PH on echocardiography in COPD patients during acute exacerbations. It can provide a convenient risk stratification tool for clinical practice.