Mp96-10 Initial Validation of Automated Data Extraction Methods in Urologic Oncology Practice

Abstract

A total of 2486 clinic appointments were compiled for four providers in the adult urology clinic over six months. Of the total, 408 were actual missed clinic visits at an overall no-show rate of 16.4%. The calculated number of patients missing their appointments was 488. Of the predicted 488 missed visits, the calculated number of patients was over by 130 with an average of 1.19 patients over per day, and under by 50 with an average of 0.46 patients under per day. The number of perfect days where the predicted number matched the actual number was 26/109 (23.9%), within +/-1 patients 61/109 (56.0%), and within +/-2 patients 87/109 (79.8%). Conversely, the model over predicted 4 or greater patient no-shows on 6/109 (5.5%) of days. Overpredicted patients per day ranged from 0.01-6.5 with a mean of 1.58.CONCLUSIONS: This review further characterizes the predictable patient characteristics associated with missed clinic visits for an under-served academic urology patient population. This model works well over a large number of patients with a 79.8% efficacy within 2 patients. Applying this to a clinical setting would be limited by overestimating the number of patients that would be scheduled. The model still will require validation when put to test on data from different practice settings and larger patient data sets. Additionally, we predict there may be confounding factors (type of insurance, distance to appointment, previous missed appointments) that we plan to study in order to add to the accuracy of the model.