Development and external validation of a risk prediction model for early peripheral intravenous catheter failure in outpatients and emergency department patients receiving consecutive infusion therapy

BackgroundPeripheral intravenous catheter (PIVC) failure is a common complication during intravenous infusion therapy and may result in treatment interruption, repeated venipuncture, and increased healthcare burden. For outpatients and emergency department patients receiving consecutive infusion therapy, early identification of patients at risk of PIVC failure before discharge with a catheter in place remains challenging. This study aimed to develop and externally validate a risk prediction model for early PIVC failure.MethodsThis dual-center prospective cohort study included adult outpatients and emergency department patients receiving intravenous infusion therapy on consecutive days between January and June 2026. Patients from Shanghai General Hospital were included in the development cohort, and patients from Guangzhong Road Community Health Service Center, Hongkou District, Shanghai, were included in the external validation cohort. Eleven candidate predictors available during the initial visit and first infusion were incorporated into a multivariable logistic regression model. Model performance was assessed using discrimination, calibration, bootstrap validation, and decision curve analysis.ResultsA total of 485 patients were included, including 340 in the development cohort and 145 in the external validation cohort. Early PIVC failure occurred in 112 (32.9%) and 48 (33.1%) patients, respectively. The number of special intravenous medication categories (OR=2.533, 95% CI: 1.755–3.657), serum albumin (per 5-g/L increase, OR=0.311, 95% CI: 0.221–0.436), and PIVC placement in the dominant limb (OR=4.718, 95% CI: 2.244–9.920) were significant predictors. The model showed good discrimination, with an apparent AUC of 0.862 in the development cohort, an optimism-corrected AUC of 0.840 after 1,000 bootstrap resamples, and an AUC of 0.845 in the external validation cohort. Decision curve analysis suggested potential clinical utility.ConclusionsThis study developed and externally validated a risk prediction model for early PIVC failure among outpatients and emergency department patients receiving consecutive infusion therapy. The model may support early risk stratification, individualized catheter management, and clinical decision-making regarding PIVC retention after discharge.