In the wake of higher education reform: The admission criterion mean baccalaureate grade revisited

Dear Sir

A Lancet editorial in 1948 (Lancet 1948) commented that school leaving grades ‘may sometimes operate unfairly against the brilliant but erratic’. Nevertheless, school leaving grades still serve as the primary instrument for admission of medical students in most countries and the literature attributes to them better prognostic validity than other admission instruments. Yet, undergraduate medical curricula have undergone major reforms in recent years. Learner-centred, problem-oriented and extensive practical training have replaced the traditional chalk and talk lectures. It seems important, therefore, to reassess the prognostic value of school graduation grades for performance in modern medical education.

We therefore monitored the in-course performance of students of three consecutive cohorts during the pre-clinical 2 years and compared them to their mean baccalaureate grades (bGrades). Students with best bGrades (1.0–1.3) had been admitted directly; those with worse bGrades had been admitted by the constitutionally guaranteed latency quota after a waiting time of several years. Admission of the first two cohorts had been based almost exclusively on bGrades, that of the third cohort also on the newly integrated Test for Medical Studies (TMS).

The average correlation between the mean semester grades and bGrades was r = 0.445 in the first two cohorts and only r = 0.198 in the third cohort. The maximal difference between the mean semester grades of the students with best and worst bGrades was 12 points out of 100. However, the distribution of grades among students with any given bGrade was larger than that (mean range 27 points), showing that individual prognosis of academic performance by bGrades is difficult. Moreover, regardless of their bGrades students younger than 22 at admission on average performed better than older students, and more young than older students completed the pre-clinical course on time (mean 76% vs. 51%). In addition, the partial correlation between bGrades and mean semester grades under control of age was significantly lower than the observed correlation between bGrades and semester grades.

These observations indicate that the relative weight of bGrades in the admission decision should be limited and young applicants with inferior bGrades should be given an equal chance to compete for an offer. We have implemented this in Heidelberg using the TMS as a second measure of cognitive intelligence, with the result that attrition has been reduced by up to 75% and academic performance has become good and uniform over the range of qualifying bGrades (1.0–2.3; data submitted to publication).