Development and validation of a clinical prediction model for concurrent pulmonary infection in convalescent patients with intracerebral hemorrhage.

Abstract

Objectives: This study aimed to develop and validate a clinical prediction model for assessing the risk of concurrent pulmonary infection (PI) in patients recovering from intracerebral hemorrhage (ICH).

Methods: In this retrospective study, we analyzed clinical data from 761 patients in the subacute recovery phase of ICH, of whom 504 developed PI and 257 did not. Univariate logistic regression was initially used to identify potential risk factors, followed by variable selection through the Least Absolute Shrinkage and Selection Operator (LASSO) regression. Predictors selected by LASSO were entered into a multivariate logistic regression to establish a final model. A nomogram was constructed based on the significant variables. The model's discrimination was evaluated using the area under the receiver operating characteristic curve (AUC), and its calibration was assessed using calibration plots and the Hosmer-Lemeshow goodness-of-fit test. Clinical utility was evaluated via decision curve analysis (DCA). Positive predictive value (PPV) and negative predictive value (NPV) were also calculated at the optimal threshold.

Results: Eight independent predictors were identified: age, prophylactic antibiotic use, disturbance of consciousness, tracheotomy, dysphagia, duration of bed rest, nasal feeding, and procalcitonin level. The model demonstrated excellent discriminative ability with an AUC of 0.901(95%CI 0.878-0.924) and good calibration (Hosmer-Lemeshow test, P = 0.982). At the optimal cut-off point, the PPV was 92.6% and the NPV was 68.0%. DCA indicated favorable clinical benefit across a wide range of threshold probabilities.

Conclusion: We developed a nomogram-based prediction model that accurately identifies the risk of pulmonary infection in patients recovering from ICH. This model offers valuable support for early clinical decision-making and targeted preventive strategies.