Using academic admissions data to predict PA training attrition and academic difficulties: a survival analysis approach.
TL;DR
This work informs admissions policies by identifying students at-risk of academic probation and attrition who may need more academic support prior to program matriculation by employing a novel application of survival analysis.
Abstract
Background
Research studying health professions' student attrition and academic difficulties generally treat these outcomes as dichotomous variables that obscure important time-varying aspects. Scant attention has been paid to incorporating time-varying methods, such as survival analysis, to model attrition and academic difficulties in health professions trainees.
Objective
The objectives of this study are to (1) identify when students were most likely to attrit from the program and/or experience academic probation and (2) identify predictors of time to attrition and academic probation using undergraduate student performance metrics.
Design
Using a database of physician associate (PA) student admissions variables and academic performance, we examined the relationship between time-to-attrition and time-to-first-probation using semi-parametric multivariable Cox Proportional Hazards models. Results are presented in adjusted Hazards Ratios (HR) with 95% Confidence Intervals (CI).
Results
The rate at which students experienced at least one probation event was 131 per 19,563.5 person-weeks. The rate of students who experienced an attrition event was 32 per 22,210 person-weeks. Overall GPA (aHR = 0.16, 95% CI = 0.08-0.33), verbal GRE scores (aHR = 0.91, 95% CI = 0.87-0.94), having a science major in undergraduate studies (aHR = 0.43, 95% CI = 0.26-0.69) and number of undergraduate course withdrawals (aHR = 1.14, 95% CI = 1.05-1.24) were associated with time to first probation. Verbal GRE scores (aHR = 0.87, 95% CI = 0.80-0.94) and number of undergraduate course withdrawals (aHR = 1.24, 95% CI = 1.08-1.43) were associated with time to attrition.
Conclusions
We employed a novel application of survival analysis to predict PA school attrition and academic probation using an entire cohort of students. Our work informs admissions policies by identifying students at-risk of academic probation and attrition who may need more academic support prior to program matriculation.
