Preventing patient absenteeism: validation of a predictive overbooking model

Area of Science:

• Health Informatics
• Healthcare Operations Research
• Predictive Analytics in Medicine

Background:

• High rates of patient no-shows and cancellations in resource-intensive clinics, such as gastrointestinal endoscopy, lead to significant underutilization of services.
• Traditional scheduling methods often fail to account for individual patient risks of missing appointments, resulting in wasted slots and reduced access to care.
• Developing accurate predictive models is crucial for optimizing appointment scheduling and improving healthcare operational efficiency.

Purpose of the Study:

• To develop and validate a predictive model for identifying patients at high risk of appointment no-shows using electronic health record (EHR) data.
• To assess the impact of predictive overbooking, informed by this model, on service utilization in a gastrointestinal endoscopy clinic.
• To compare the effectiveness of predictive overbooking against traditional scheduling and fixed overbooking strategies.

Main Methods:

• A multivariable logistic regression model was developed retrospectively using EHR data to identify no-show predictors.
• The model was prospectively validated in a Veterans Administration healthcare network clinic, assigning a no-show risk score.
• A scheduling system based on predicted open slots was implemented to test predictive overbooking against standard scheduling.

Main Results:

• Key predictors of no-show included prior absenteeism, comorbid disease burden, and diagnoses of mood and substance use disorders.
• The predictive model demonstrated strong performance in both development (AUC = 0.80) and validation (AUC = 0.75) phases.
• Predictive overbooking reduced unused appointments from 6.18 to 0.51 per day, increasing service utilization from 62% to 97% of capacity.

Conclusions:

• EHR data can be effectively utilized to accurately predict patient no-shows.
• Implementing predictive overbooking based on these EHR-derived risk scores significantly enhances clinic service utilization.
• This strategy allows for maximized operational capacity while maintaining control over clinic overflows.