O2046: Predictive Scheduling in Pediatric Cardiology Clinics: Optimization for Both Patients and Physicians

Background: Missed clinic appointments, often called no-shows (NS), lead to clinic inefficiency, decreased access to care, and reduced physician productivity and satisfaction. Among our 21 clinics across the state, NS rates vary from 10% to 53%. To reach a target volume of 12 patients per day, the schedule is overbooked, which often causes wide swings in daily patient volumes. Our clinic physicians identify workflow inefficiency and high patient volumes as leading contributors to dissatisfaction. The purpose of this study is 1) to evaluate the effectiveness of a predictive scheduling model (PSM) in the optimization of the number of patients seen in a clinic day, and 2) to study physician satisfaction scores (PSS) after implementation of the PSM.

Methods: PSM was developed using over 400,000 appointment slots over 6 years (2014 to 2020). For follow-up visits, the NS prediction was based on patient’s NS history for the previous 5 years, and, for new visits, it was based on the clinic location and the time of year of the appointment. The predicted NS rates were used to calculate the total expected patient volume for the clinic day. Patients were scheduled for a clinic day until the expected daily volume reached the target daily volume of 12 patients. PSM was trialed in 4 clinics among 6 physicians. The appointment slots were changed from 15, 20-minute slots to 12, 30-minute slots. The effectiveness of PSM was assessed by tracking how closely it predicted the actual number of patients seen in a clinic day. PSS were measured on a scale of 1 (worst) to 10 (best) with clinic physicians rating via e-mail after each clinic. PSS for 4 months before and after implementation were compared.

Results: PSM was initially validated using 668 clinics with 7677 patients from Feb-March 2021. PSM correctly predicted within ±1 of the total patients actually seen in a clinic day 75% of the time and within ±2 patients 94% of the time (Figure 1). Physicians in trial clinics were overbooked, with more than 12 patients seen in a day, only 1% of the time, whereas this occurred 30% of the time before PSM implementation. Most trial clinics remained within 6% of their previous volumes from before implementing PSM. PSS increased from an average of 7.1 to 8.2 (16% increase) after implementation of PSM (Figure 2).

Conclusion: PSM was highly effective at predicting patient NS rates and achieving the target of 12 patients actually seen per clinic. Physicians in the trial clinics were able to see their target patient volume with minimal variability and more time with each patient, resulting in improved PSS. Future studies could assess the impact of PSM on improving patient satisfaction scores and access to care.