This paper describes and evaluates a method of predicting referral appointment-keeping behavior for psychiatric emergency room patients. Studies have shown that referral systems which provide appointments and reminders achieve higher appointment adherence. However, studies have not been conducted of those experiencing a good referral system to predict adherence behavior and indicate further areas for improvement. Discriminant analysis is applied to patient's psychiatric signs and symptoms for selecting variables that predict referral appointment keeping behavior. All patients seen during a 6 week period were selected from a 1,000 bed hospital's Psychiatric Emergency Room. The data were collected during the patient's visit to the psychiatric emergency room and prior to aftercare treatment. The subset of non-psychotic patients requiring outpatient psychiatric care was the focus of this analysis. The model indicates that patient's adherence with referral appointment can be predicted with 76% accuracy using the best three discriminating variables. This prediction per cent increases to 97% with all contributing variables. The non-adherers are identified with 66% and 100% accuracy, respectively. Distressingly, "Suicidal Behavior" and "Potentially Harming Others" relates to non-adherence. If the non-compliant emergency patient, especially those dangerous to themselves and others, can be identified in advance, the care systems can be changed to increase the number of patients following referral recommendations. This reduces both initial treatment cost and avoids cost resulting from more severe illnesses arising for lack of initial treatment.
ASJC Scopus subject areas
- General Computer Science
- Modeling and Simulation
- Management Science and Operations Research