Analysis of risk factors for recurrence of pediatric immune thrombocytopenia and construction of a risk prediction model

Lai Siyan; He Rongxiang; Li Wei; Liang Daoqiong; Qian Min; Tan Wei; An Shichun

doi:10.11986/j.issn.1673-873X.2025.06.16

Abstract：Objective To analyze risk factors for recurrence of pediatric immune thrombocytopenia(ITP)and to construct a risk prediction model.Methods A total of 189 pediatric ITP patients admitted to the Second People's Hospital of Panzhihua from May 2020 to May 2023 were enrolled and followed for one year after treatment for recurrence.Clinical data,including sex,age,blood type,and onset season,were collected.Multivariate logistic stepwise regression analysis was used to identi-fy risk factors for recurrence and to construct a risk prediction model.Model calibration was evaluated with the Hosmer-Lemeshow test,and predictive performance was assessed by receiver operating characteristic(ROC)curve analysis.Results During one-year follow-up,35 patients experienced recurrence(18.52％)and 154 patients(81.48％)did not.Significant differences between the recurrent and non-recurrent groups were observed in blood type,disease course before treatment,presence of prodromal infection,onset season,antinuclear antibody status,pre-treatment peripheral blood lymphocyte count(PBL),pre-treatment platelet count(PLT),pre-treatment platelet distribution width(PDW),pre-treatment mean platelet volume(MPV),and time from treatment to PLT rising to an effective value(P＜0.05).Multivariate logistic stepwise re-gression analysis identified the following independent risk factors for recurrence:longer disease course before treatment(OR=4.627,95％CI:1.225-17.477),a history of prodromal infection(OR=3.180,95％CI:1.306-7.743),higher pre-treatment PLT(OR=2.683,95％CI:1.436-5.014),lower pre-treatment PBL(OR=3.589,95％CI:1.632-7.893),longer time for PLT to rise to an effective value after treatment(OR=3.815,95％CI:1.858-7.833),bloodtype O as compared with type A(OR=4.233,95％CI:1.374-13.040),and onset in autumn as compared with spring(OR=2.732,95％CI:1.339-5.576;P＜0.05).The resulting risk prediction model yielded an area under the ROC curve(AUC)of 0.872(95％CI:0.816-0.916),with a sensitivity of 86.11％and specificity of 88.27％,indicating good predic-tive performance.Hosmer-Lemeshow test indicated good agreement between predicted and observed outcomes(x2=0.834,P=0.116).Conclusion Longer pre-treatment disease course,a history of prodromal infection,higher pre-treatment PLT,lower pre-treatment PBL,longer time for PLT to rise to an effective value after treatment,blood type O,and autumn onset are risk fac-tors for recurrence of pediatric ITP.The constructed risk prediction model demonstrates good predictive performance.