Influencing Factors and Risk Prediction Model Construction of Medication Adherence in Pediatric Patients with Inflammatory Bowel Disease

OBJECTIVE To analyze the influencing factors of medication adherence in pediatric patients with inflammatory bowel disease(IBD), and to construct a risk prediction model to explore its predictive value.

METHODS  Children with IBD who admitted to Children's Hospital, Zhejiang University School of Medicine from January 2022 to January 2024 were selected as the research object, and the relevant data were collected through questionnaire survey. According to the medication adherence score, the pediatric patients were divided into the non-adherence group(score≤16) and the adherence group(score≥17). Then the demographic and clinical characteristics between the 2 groups were compared, the influencing factors of medication adherence were confirmed by multivariate binary Logistic regression analysis, which were used to construct a risk nomogram prediction model for poor medication adherence in pediatric patients with IBD.

RESULTS A total of the 152 pediatric patients, 69 cases(45.4%) had poor adherence. Multivariate binary Logistic regression analysis showed that pediatric patients’s age(OR=1.13, 95%CI 1.02−1.25, P=0.024), frequency of administration(OR=2.00, 95%CI 1.22−3.27, P=0.006), mother’s educational level(OR=0.59, 95%CI 0.38−0.92, P=0.020) and parents’ score of Crohn’s and Colitis Knowledge Score(CCKNOW)(OR=0.90, 95%CI 0.83−0.99, P=0.023) were significant influencing factors of medication adherence in pediatric patients with IBD. The nomogram prediction model of medication adherence was constructed based on this 4 indicators: Logit(P)=0.117×age+0.692×frequency of administration−0.533×mother’s educational level−0.102×parents’ CCKONW score. The receiver operating characteristic curve showed that the area under the curve of the nomogram prediction model was 0.759(95%CI 0.683−0.835). The Hosmer-Lemeshow fitting test showed that a good accuracy of model fitting(χ²=5.983, P=0.650). The calibration curve showed that the predicted probability was close to the actual probability, indicating good distinguishing, calibration and forecasting ability of this model.

CONCLUSION The influencing factors including pediatric patients’s age, frequency of administration, mother’s educational level and parents’ disease-related knowledge level. The relevant nomogram model had good predictive value, and is helpful for early identification of high-risk populations with poor medication adherence.