To develop a nomogram model for guiding clinical accurate evaluation on the severity of adenovirus pneumonia in children, and carry out the internal validation.

Total of 128 children with adenoviral pneumonia diagnosis in Xuzhou Children’s Hospital Affiliated to Xuzhou Medical University from August 2019 to August 2021 were included, retrospectively, who all met the criteria of the guideline for the diagnosis and treatment of adenoviral pneumonia in children (2019 Edition). During hospitalization, according to the diagnostic criteria of community-acquired pneumonia, 128 cases were divided into severe group (50 cases) and non-severe group (78 cases). The gender, age, weight, premature delivery, duration of fever, peak body temperature, basic diseases, serum leukocyte count, percentage of neutrophils and lymphocytes, C-reactive protein (CRP), lactic acid, interleukin (IL)-6, lactate dehydrogenase (LDH), procalcitonin (PCT), percentage of CD4⁺ T and CD8⁺ T lymphocytes, CD4⁺/CD8⁺, CD16⁺CD56⁺ and CD19⁺ T lymphocytes, and mixed infection were compared between the two groups. The dimension of risk factors was reduced by LASSO regression, then the independent risk factors were screened by multivariate Logistic regression analysis, nomogram predictive model was drawn according to the regression coefficient (β). The area under the curve (AUC) of model for severe adenovirus pneumonia (SAP) prediction was calculated by receiver operating curve (ROC), and the goodness of fit of the model was evaluated by Hosmer-Lemeshow test, consistency and benefit of the model were evaluated by calibration curve and decision curve.

Compared with non-severe group, Univariate comparison showed that the duration of fever was longer [5.2 (3.0, 8.5) d vs. 2.9 (1.0, 5.0) d; Z = 8.326, P < 0.001], concentrations of IL-6 [45.6 (35.4, 56.9) pg/ml vs. 30.2 (25.2, 38.6) pg/ml; Z = 15.326, P < 0.001] and LDH were higher [452.6 (385.6, 523.4) U/L vs. 365.9 (302.1, 445.2) U/L; Z = 9.625, P < 0.001], levels of CD4⁺ T [31.2 (27.8, 34.2)% vs. 35.5 (33.2, 38.9)%; Z = 7.526, P < 0.001] and CD4⁺/CD8⁺ were lower [1.2 (1.0, 1.4) vs. 1.4 (1.1, 1.6); Z = 5.230, P = 0.004], and the rate of mixed infection was higher in severe group [46.0% (23/50) vs. 19.2% (15/78); χ² = 10.460, P = 0.001]. LASSO regression screened four variables with non-zero coefficients, namely duration of fever, IL-6 concentration, CD4⁺ T lymphocyte percentage and mixed infection. Multivariate Logistic regression analysis showed that duration of fever (OR = 3.125, 95%CI: 2.565-3.896, P < 0.001), IL-6 concentration (OR = 2.012, 95%CI: 1.428-2.639, P < 0.001), CD4⁺ T lymphocyte percentage (OR = 0.369, 95%CI: 0.124-0.678, P = 0.009) and mixed infection (OR = 1.457, 95%CI: 1.124-1.895, P = 0.001) were the independent risk factors for SAP. The total score of nomogram model was 160. ROC showed that the AUC value of nomogram for predicting SAP was 0.852 (95%CI: 0.779-0.901, P < 0.001). The Hosmer-Lemeshow test value was 0.786, suggesting that the goodness of fit of the model was high. The calibration curve and decision curve showed that the consistency and benefit of the model were acceptable.

The main factors affecting the severity of adenovirus pneumonia in children are the duration of fever, the concentration of IL-6, the percentage of CD4⁺ T lymphocytes and mixed infection. The developed nomogram prediction model to evaluate SAP has simple operation and strong visualization effect. It has high efficacy and goodness of fit, good consistency and benefit, and has good clinical application value.