Development and assessment of a new nomogram for predicting tuberculosis risk in a Chinese inflammatory rheumatic disease population

Cheng Huan¹, Chao Lu^2,* 

¹Department of Intensive Care Unit, Shandong Provincial Hospital Affiliated to Shandong First Medical University, Jinan, P.R. China - ²Department of Clinical Laboratory, Shandong Provincial Hospital Affiliated to Shandong First Medical University, Jinan, P.R. China

Objective: To develop a predictive nomogram for tuberculosis risk in a Chinese population with inflammatory rheumatic disease (IRDs). 

Methods: All patients were divided either into IFN-γ negative group or positive group according to the results of T-Spot. This is a prospective study. Data from 276 patients collected between January 2018 and May 2018 at the Department of Rheumatology and Immunology, Provincial Hospital Affiliated to Shandong University were analyzed. The optimal predictive characteristics of risk factors were selected from patients with IRDs using the least absolute shrinkage and selection operator (LASSO) method for high-dimensional data reduction. We used multivariate logistic regression analysis to build a predictive model that incorporated the features selected by the LASSO regression model. Multivariate logistic regression analysis, compilation of the nomogram was done using R software (3.1.1). C-index, calibration plot and decision curve analysis were used to evaluate discrimination, calibration and clinical application of the model, respectively. The statistical tests were two-sided and P < 0.05 was considered statistically significant.

Results: Eleven potential predictors contained in the nomogram included age, HLA-B27, ESR, NLR, ANA, C4, marital status, smoking history, and use of glucocorticoids, NSAIDs, and immunosuppressive agents.The predictors were non-zero coefficients in the LASSO regression model. The C-index of the predicted nomogram was 0.775 (95% confidence interval: 0.645-0.777), which was confirmed to be 0.718 through bootstrapping validation. The decision curve showed that if the patient and doctor threshold probabilities were 3 and 87%, respectively, the use of the nomogram to predict the risk of tuberculosis infection increased the benefits more than the scheme.

Conclusions: This novel predictive nomogram established with age, HLA-B27, ESR, NLR, ANA, C4, marital status, smoking history, and use of glucocorticoids and NSAIDs and immunosuppressive agents could be easily used to predict the risk of TB in patients.