Diagnostic test accuracy of cellular analysis of bronchoalveolar lavage fluid in distinguishing pulmonary infectious and non-infectious diseases in patients with pulmonary shadow

Abstract

PurposeThis study aims to assess the diagnostic accuracy of cellular analysis of bronchoalveolar lavage fluid (BALF) in distinguishing between pulmonary infectious and non-infectious diseases in patients with pulmonary shadows. Additionally, it will develop and validate a novel scoring system based on a nomogram for the purpose of differential diagnosis.MethodsA retrospective analysis was conducted involving data from 222 patients with pulmonary shadows, whose etiological factors were determined at our institution. The cohort was randomly allocated into a training set comprising 155 patients and a validation set of 67 patients, (ratio of 7:3), the least absolute shrinkage and selection operator (LASSO) regression model was applied to optimize feature selection for the model. Multivariable logistic regression analysis was applied to construct a predictive model. The receiver operating characteristic curve (ROC) and calibration curve were utilized to assess the prediction accuracy of the model. Decision curve analysis (DCA) and clinical impact curve (CIC) were employed to evaluate the clinical applicability of the model. Moreover, model comparison was set to evaluate the discrimination and clinical usefulness between the nomogram and the risk factors.ResultsAmong the relevant predictors, the percentage of neutrophils in BALF (BALF NP) exhibited the most substantial differentiation, as evidenced by the largest area under the ROC curve (AUC = 0.783, 95% CI: 0.713–0.854). A BALF NP threshold of ≥16% yielded a sensitivity of 72%, specificity of 70%, a positive likelihood ratio of 2.07, and a negative likelihood ratio of 0.38. LASSO and multivariate regression analyses indicated that BALF NP (p < 0.001, OR = 1.04, 95% CI: 1.02–1.06) and procalcitonin (p < 0.021, OR = 52.60, 95% CI: 1.83–1510.06) serve as independent predictors of pulmonary infection. The AUCs for the training and validation sets were determined to be 0.853 (95% CI: 0.806–0.918) and 0.801 (95% CI: 0.697–0.904), respectively, with calibration curves demonstrating strong concordance. The DCA and CIC analyses indicated that the nomogram model possesses commendable clinical applicability. In models comparison, ROC analyses revealed that the nomogram exhibited superior discriminatory accuracy compared to alternative models, with DCA further identifying the nomogram as offering the highest net benefits across a broad spectrum of threshold probabilities.ConclusionBALF NP ≥16% serves as an effective discriminator between pulmonary infectious and non-infectious diseases in patients with pulmonary shadows. We have developed a nomogram model incorporating BALF NP and procalcitonin (PCT), which has proven to be a valuable tool for predicting the risk of pulmonary infections. This model holds significant potential to assist clinicians in making informed treatment decisions.