Day-of-surgery (DOS) cancellations in elective urologic procedures
Key Mechanisms
Multifactorial causes including hospital-related factors (e.g., operating room availability) and patient-related factors (e.g., absenteeism), influenced by social determinants of health (SDOH)
Target Population
Adult patients (≥18 years) scheduled for elective urologic surgeries at a safety-net hospital
Care Setting
Elective urologic surgery in hospital operating rooms
Key Highlights
DOS cancellations reduce hospital productivity, increase costs, and delay patient treatment.
Approximately 86% of elective surgery cancellations are preventable with targeted interventions.
Incorporating SDOH data into predictive models improves accuracy in forecasting DOS cancellations.
Guideline-Based Recommendations
Diagnosis
Define DOS cancellation as any scheduled procedure canceled on the day of surgery.
Use comprehensive EHR data including medical, procedural, and SDOH factors to identify patients at risk.
Management
Develop predictive models using logistic regression with lasso regularization to select key predictors.
Apply targeted, cost-effective interventions for patients identified at high risk of DOS cancellation.
Monitoring & Follow-up
Evaluate model performance using AUC, kappa statistic, sensitivity, specificity, PPV, and NPV.
Perform internal validation with training and test data splits and cross-validation to ensure model reliability.
Risks
Unavailability of operating room time and patient no-shows are primary risk factors for DOS cancellations.
Ignoring SDOH factors may reduce predictive accuracy and limit intervention effectiveness.
Patient & Prescribing Data
Adults undergoing elective urologic surgery at a safety-net hospital
Predictive modeling incorporating holistic EHR and SDOH data can identify patients at risk for DOS cancellations, enabling early intervention to improve surgical productivity and reduce cancellations.
Clinical Best Practices
Exclude emergent procedures to focus on elective surgery cancellation predictors.
Use census-tract linked area deprivation index (ADI) as a measure of neighborhood socioeconomic disadvantage.
Apply random oversampling to balance datasets when modeling imbalanced outcomes like DOS cancellations.
Utilize targeted search terms and clinical expertise to extract SDOH data from EHR notes and social history.
Optimize model tuning parameters to maximize prediction accuracy and reduce collinearity.
A hallmark of Fox Chase leadership, fellows, residents and faculty is their ability to persevere—to keep climbing the often-difficult ascent of cancer discovery and delivery—no matter what. Meet some of the exceptional researchers, scientists, and clinicians determined to make decisive inroads in cancer care.
