A Multivariate Model to Predict Student Physician Assistant National Certification Exam Performance.

Abstract

Introduction: The Physician Assistant National Certification Exam (PANCE) is the standard for assessing the medical knowledge of the Physician Assistant graduate. Performance on this high-stakes examination is often the culmination of countless hours of preparation. A tool to predict PANCE scores empowers faculty to identify and prepare students who may be at risk.

Methods: This retrospective, single-institution study examined scores from 10 standardized PA Education Association examinations for their combined accuracy in predicting student first-time numeric PANCE scores. Individual scores from 4 consecutive Physician Assistant program cohorts (n = 91) were analyzed using a multiple regression model to obtain a coefficient of multiple correlation (R) with ANOVA (analysis of variance) statistical testing for significance. A predictive equation was then developed to predict first-time PANCE scores of the fifth cohort (n = 31). A simple linear regression was used to correlate the predicted PANCE score from the model with the actual PANCE score.

Results: The multiple regression model was statistically significant, evidenced by the ANOVA results (F = 22.53, P < 0.0005). The multiple regression model shows a strong multiple correlation (R = 0.86), demonstrating the effectiveness of this combination of standardized exams in predicting PANCE results.

Discussion: The multiple regression model reliably predicts first-time PANCE scores, thus providing validity evidence for the use of these standardized PA Education Association examinations in assessing content/task areas. Applying this model can identify students in our program at risk for PANCE failure and improve success as evidenced by a first-time pass rate in the most recent graduating cohort of 100%.