This study (from the Auckland med school) might have something to do with it:
https://www.springerlink.com/content/h16121u1p0k58411/
Fhttps://www.springerlink.com/content/h16121u1p0k58411/rom the abstract: Prior academic achievement remains the best measure of subsequent student achievement within a medical programme. Interview scores have little predictive value. Future directions include longer term studies of what UMAT predicts, and of novel ways to combine selection tools to achieve the optimum student cohort.
https://www.springerlink.com/content/h16121u1p0k58411/
Fhttps://www.springerlink.com/content/h16121u1p0k58411/rom the abstract: Prior academic achievement remains the best measure of subsequent student achievement within a medical programme. Interview scores have little predictive value. Future directions include longer term studies of what UMAT predicts, and of novel ways to combine selection tools to achieve the optimum student cohort.