Outpatient Skull Base/Brain Tumor Clinic Visit No-Show Prediction

Introduction: Outpatient patient visit no-shows result in clinic workflow inefficiencies, delayed access to care, and opportunity costs for providers and staff. Early identification of patients likely to no-show could allow for targeted patient outreach or strategic clinic scheduling/overbooking. Predictors of clinic no-show among the skull base/brain tumor patient population have not been previously characterized.

Methods: All outpatient skull base/brain tumor clinic appointment records between 11/1/2017 and 5/1/2022 were extracted from the electronic health record (EHR) at a single institution. Appointments which were cancelled prior to the date of the visit were excluded. The primary outcome was a clinic visit to which a patient did not check in. Predictive covariates included patient demographics and appointment factors. A multivariable logistic regression model was created using a randomly selected training set of 66% of clinic visits. The model was applied to the remaining 33% validation set and predictive performance was assessed using receiver operating characteristic (ROC) analysis.

Results: Among 5,916 total clinic visits, 494 no-shows were identified. Odds ratios with 95% confidence intervals for the predictive value of variables included in the model are depicted in [Fig. 1]. Increased odds of no-show were seen with older age (OR = 1.02, p < 0.01, [Fig. 2]), greater number of prior no-shows (OR = 3.64, p < 0.01 for three or more prior no-shows) and longer lead time (number of days between appointment scheduling and appointment date, OR = 1.0002 per day of lead time, p = 0.04; [Fig. 3]). Categorical variables for marital status, tobacco use, insurance type, EHR mobile app enrollment, and appointment type each had statistically significant associations with no-show. No-show rates were similar across all providers; telehealth visits were not associated with increased rate of no-shows. When applied to the validation set, the model correctly classified 92.1% of visits, including 82.4% of no-shows. The area under the ROC curve was 0.76 ([Fig. 4]).

Conclusion: The predictive model derived from structured data within the EHR accurately predicted clinic no-shows. This model may be used for targeted patient outreach to improve clinic attendance as well as strategic overbooking to improve clinic operational efficiency.

No conflict of interest has been declared by the author(s).

Publication History

Article published online:
15 February 2022