The VA has committed extensive resources to ensuring that timely screening colonoscopy is readily available for Veterans. To minimize delays in care, the VA approved the use of contracted fee-for-service colonoscopy with non-VA providers at a significant cost. Yet, there has been little attention paid to colonoscopy absenteeism, which increases total colonoscopy demand as Veterans who do not show up ("no-show") for an exam are re-scheduled in the original queue. Preliminary pilot data from the West Los Angeles VA Hospital, the Loma Linda VA Hospital and the Durham VA Hospital in North Carolina report absenteeism to range between 20%-50%, suggesting that capacity could be significantly increased by addressing no-show rates. Racial disparities in colon cancer screening may also be addressed by this project.
We sought to develop and validate a predictive overbooking system that identifies patients at high risk for missing scheduled gastrointestinal endoscopy procedures ("no-shows" and cancellations), and offers their appointments to other patients on short notice. In the first phase, we retrospectively validated an algorithm that uses electronic health record (EHR) data to identify patients who do not show up to their appointments. In the second phase, we prospectively applied the algorithm in real-time scheduling and monitored outcomes.
We constructed a multivariable logistic regression model that assigned a no-show risk score optimized by receiver operating characteristic curve analysis. Based on these scores, we created a calendar of projected open slots to offer to patients and compared the daily performance of predictive overbooking with fixed overbooking and typical "1 patient, 1 slot" scheduling. We then prospectively tested the predictive overbooking system at the West Los Angeles VA endoscopy clinic over a 34-week period, alternating between traditional booking and predictive overbooking methods. For the latter, we assigned a no-show risk score to each scheduled patient, utilizing the previously developed logistic regression model built with EHR data. To compare booking methods, we measured service utilization-defined as the percentage of daily total clinic capacity occupied by patients-and length of clinic workday.
Data from 1392 patients in the retrospective analysis identified several predictors of no-show, including previous absenteeism, comorbid disease burden, and current diagnoses of mood and substance use disorders. The model correctly classified most patients during the development (area under the curve [AUC] = 0.80) and validation phases (AUC = 0.75). In the prospective study applying the prediction algorithm to actual clinic scheduling, we found that compared to typical booking, predictive overbooking resulted in nearly all appointment slots being filled - 2.5 slots available during control weeks vs. 0.35 slots during intervention weeks (P=0.0001). Service utilization increased from 86% during control weeks to 100% during intervention weeks, allowing 111 additional patients to undergo procedures during the prospective phase of the study. Physician and staff overages were more common during intervention weeks, but less than anticipated (workday length of 7.84 h (control) vs. 8.31 h (intervention), P=0.02). In a logistic regression model controlling for demographic and clinical characteristics, African Americans were twice as likely (adjusted OR, 1.99; 95% CI, 1.26-3.17) than Whites to participate in the fast-track option for recommended endoscopy.
Predictive overbooking may be used to maximize endoscopy scheduling and serve more Veterans within the constraints of existing resources. The study intervention increased the overall percentage of GI clinic patients undergoing endoscopy and disproportionately improved appointment uptake in African Americans. Future research should focus on adapting the model for use in primary care and specialty clinics.
We have successfully identified determinants of screening colonoscopy absenteeism and developed a predictive score with high accuracy. We implemented this model to increase clinic capacity and deliver more timely service to Veterans. Currently, we are trying to implement the predictive overbooking system permanently in our GI clinic, while also developing a general model for predicting appointment attendance across specialties. With future funding, we aim to deploy this model at other VA sites. We may also develop a targeted intervention to combat high rates of no-show in patients most at risk for colorectal cancer. We may also explore the reasons behind increased rates of booking "fast track" appointments in African Americans.
External Links for this Project
Grant Number: I01HX000878-01
- May FP, Reid MW, Cohen S, Dailey F, Spiegel BM. Predictive overbooking and active recruitment increases uptake of endoscopy appointments among African American patients. Gastrointestinal endoscopy. 2017 Apr 1; 85(4):700-705. [view]
- Reid MW, May FP, Martinez B, Cohen S, Wang H, Williams DL, Spiegel BM. Preventing Endoscopy Clinic No-Shows: Prospective Validation of a Predictive Overbooking Model. The American journal of gastroenterology. 2016 Sep 1; 111(9):1267-73. [view]
- Reid MW, Cohen S, Wang H, Kaung A, Patel A, Tashjian V, Williams DL, Martinez B, Spiegel BM. Preventing patient absenteeism: validation of a predictive overbooking model. The American journal of managed care. 2015 Dec 1; 21(12):902-10. [view]