Construction and Verification of Risk Predictive Nomogram in Patients of Connective Tissue Disease with Severe Pneumonia.

PURPOSE

To construct Risk Predictive Nomogram in patients of connective tissue disease (CTD) with severe pneumonia.

METHODS

Eighty CID patients with severe pneumonia in rheumatology and respiratory department of chongqing University Three Gorges hospital from January 2020 to December 2022 were retrospectively reviewed and analyzed. Independent risk factors for severe pneumonia in CTD were screened by univariate and binomial logistic regression analysis. The nomogram was constructed by R software. Area under the curve (AUC) of receiver operating characteristic (ROC) was used to evaluate the nomogram's discrimination, and the calibration curve and Hosmer-Lemeshow test were used to reflect the nomogram's calibration.

RESULTS

The study cohort was including 48 patients in the general pneumonia group and 32 patients in the severe pneumonia group. The model variables included Ln CD4/CD8, Ln CRP, Ln PCT and Ln IFN-γ. Hosmer-lemeshow test P value less than 0.05 (χ2 = 7.753, = 0.458), the area under ROC curve of nomogram was 0.9084 (95% CI: 0.8461-0.9707), and the optimal cutoff value of nomogram was 0.490, the sensitivity was 0.872, the specificity was 0.848. In a retrospective study design, 50 patients with CTD complicated with pneumonia admitted to the same hospital from January to June 2023 were selected to verify the model. The nomogram verification results showed Hosmer-Lemeshow test (χ2 = 7.1171, = 0.5241), AUC value was 0.8958 (95% CI: 0.808-0.9837), and optimal cutoff value was 0.664, the sensitivity was 0.988, the specificity was 0.812.

CONCLUSION

The prediction nomogram in this study is helpful for clinical staffs to screen high-risk patients with severe pneumonia in CTD, and has high clinical application value.