PT - JOURNAL ARTICLE AU - Joey A Robaina AU - Tracey P Bastrom AU - Andrew C Richardson AU - Eric W Edmonds TI - Predicting no-shows in paediatric orthopaedic clinics AID - 10.1136/bmjhci-2019-100047 DP - 2020 Mar 01 TA - BMJ Health & Care Informatics PG - e100047 VI - 27 IP - 1 4099 - http://informatics.bmj.com/content/27/1/e100047.short 4100 - http://informatics.bmj.com/content/27/1/e100047.full SO - BMJ Health Care Inform2020 Mar 01; 27 AB - Background Clinic ‘no shows’ (NS) can be a burden on the healthcare system, and efforts to minimise them can reduce lost revenue and improve patient care. Leveraging a large data set via the electronic health record (EHR) has not been previously attempted to identify ‘high risk’ groups in paediatric orthopaedics.Objective To use discrete data captured by the EHR system to identify predictors of non-attendance at paediatric orthopaedic outpatient appointments.Methods Appointments from January 2014 to March 2016 were included. Variables included appointment status, age, gender, type of visit, payor type (government vs private insurance), distance of residence to clinic, region of residence, clinic location, clinic type, and appointment day of the week, hour and month. Classification and regression trees (CART) were constructed to identify predictors of NS.Results 131 512 encounters were included, 15 543 of which were in the NS group (11.8%). CART identified three predictive covariates for NS: days in between scheduling and appointment, insurance type, and specific orthopaedic clinic type. The combination of covariates provided predictability of NS: if they had ≤38.5 days of waiting for appointment and had private insurance, the NS rate was 7.8% (the best result), compared with waiting >38.5 days for either a fracture or sports clinic, which had an NS rate of 29.3% (OR=4.9).Conclusion Payor type and duration between scheduling and appointment may predict non-attendance at outpatient paediatric orthopaedic appointments. Although these findings allow for predicting and interventions for at-risk groups, even the best performing NS group occurred 7.8% of the time, highlighting the complexity of the NS phenomenon.