Establishment of a Risk Prediction Model for Pulmonary Infection in Patients with Advanced Cancer.

OBJECTIVE

Based on clinical data, the risk prediction model of pulmonary infection in patients with advanced cancer was established to predict the risk of pulmonary infection in patients with advanced cancer, and intervention measures were given in advance.

METHODS

The clinical data of 2755 patients were divided into infection group and control group according to whether they were complicated with lung infection. 1609 patients' data from January 2016 to December 2018 served as the training set, and 1166 patients' data from January 2019 to December 2020 served as the testing set. Demographics, whether the primary cancer was lung cancer, lung metastasis, the pathological classification of lung cancer patients, the number of metastases, history of surgery, history of chemotherapy, history of radiotherapy, history of central venous catheterization, history of hypertension, diabetes, and whether with myelosuppression were recorded. The presence of concurrent pulmonary infection was recorded and defined as the primary outcome variable. Stepwise forward algorithms were applied to informative predictors based on Akaike's information criterion. Multivariable logistic regression analysis was used to develop the nomogram. An independent testing dataset was used to validate the nomogram. Receiver-operating characteristic curves and the Hosmer-Lemeshow test were used to assess model performance.

RESULTS

The sample included 2755 patients with advanced cancer. An independently validated dataset included 1166 patients with advanced cancer. In the training dataset, gender, age, lung cancer as primary cancer, the pathological classification of lung cancer patients, history of chemotherapy, history of radiation therapy, history of surgery, the number of metastases, presence of central venous catheterization, and myelosuppression were identified as predictors and assembled into the nomogram. The area under curve demonstrated adequate discrimination in the validation dataset (0.77; 95% confidence interval, 0.74 to 0.79). The nomogram was well calibrated, with a Hosmer-Lemeshow statistic of 12.4 ( = 0.26) in the testing dataset.

CONCLUSIONS

The present study has proposed an effective nomogram with potential application in facilitating the individualized prediction of risk of pulmonary infection in patients with advanced cancer.