BACKGROUND AND OBJECTIVES: Missed appointments represent a significant challenge to the efficient and effective provision of care in the outpatient setting. High no-show rates result in ineffective use of human resources and contribute to loss of follow-up. Shade Tree Clinic (STC) is a student-run, comprehensive primary care clinic that serves more than 350 Middle Tennessee residents. This study aimed to use available data to predict no-shows to improve clinic efficiency and service quality. METHODS: Data were pulled from clinic scheduling software for all appointments at STC between January 1, 2010 and December 31, 2015. Weather data were added for each appointment date using an online database. Multivariable logistic regression was used to create models from these historical data. RESULTS: A total of 13,499 appointments were included with an overall show rate of 69.2%. The final model contained previous show rate (OR 1.063; P<.001), day of the week (OR 1.20; P<.001), automated reminder (OR 1.40; P<.001), snow in inches (OR .33; P<.001), and high ambient temperature in degrees (OR 1.01; P<.001). Using a cutoff probability of the 25th percentile, the model had a negative predictive value of 61.0%. CONCLUSIONS: Based on readily available data and a novel conceptual framework, we can identify the quarter of patients least likely to present for scheduled appointments and target them for interventions, allowing care providers to more effectively address community health care disparities through the clinic. This analysis is replicable at any clinic using an electronic medical record.
BACKGROUND AND OBJECTIVES: Missed appointments represent a significant challenge to the efficient and effective provision of care in the outpatient setting. High no-show rates result in ineffective use of human resources and contribute to loss of follow-up. Shade Tree Clinic (STC) is a student-run, comprehensive primary care clinic that serves more than 350 Middle Tennessee residents. This study aimed to use available data to predict no-shows to improve clinic efficiency and service quality. METHODS: Data were pulled from clinic scheduling software for all appointments at STC between January 1, 2010 and December 31, 2015. Weather data were added for each appointment date using an online database. Multivariable logistic regression was used to create models from these historical data. RESULTS: A total of 13,499 appointments were included with an overall show rate of 69.2%. The final model contained previous show rate (OR 1.063; P<.001), day of the week (OR 1.20; P<.001), automated reminder (OR 1.40; P<.001), snow in inches (OR .33; P<.001), and high ambient temperature in degrees (OR 1.01; P<.001). Using a cutoff probability of the 25th percentile, the model had a negative predictive value of 61.0%. CONCLUSIONS: Based on readily available data and a novel conceptual framework, we can identify the quarter of patients least likely to present for scheduled appointments and target them for interventions, allowing care providers to more effectively address community health care disparities through the clinic. This analysis is replicable at any clinic using an electronic medical record.