A multivariate approach to the prediction of no-show behavior in a primary care center

Abstract

To predict no-show behavior in a primary care center, we analyzed a wide range of factors in 376 patients. Of 1,181 appointments that were scheduled during a six-month follow-up period and that were not cancelled in advance, 968 (82%) were kept and 213 (18%) were no-shows. By multivariate logistic regression analysis based on two thirds of the patient sample, no-show behavior was independently correlated with the following four factors: the patient's age and race, the presence of any physician-identified psychosocial problems, and the percent of noncancelled and appointments that were kept during the prior 12 months. Neither patient satisfaction nor patient-physician concordance in problem identification were independent correlates of appointment keeping. When our four-factor logistic regression equation was independently tested on the other one third of the patients, it accurately predicted no-show behavior. We suggest that our predicted probability of no-show behavior can be used to guide changes in scheduling patterns or to recognize patients appropriate for interventions to change behavior.