RESUMEN
A 1:1 matched case-control study was carried out to assess whether comments written about a candidate during a semi-structured interview can identify students likely to withdraw from a medical course better than global numerical scores. Fifty-nine students who withdrew from the undergraduate medical course at the University of Newcastle prior to completion were matched for sex, year and mode of entry into the course, prior academic qualifications and age at entry, with 59 students who had not discontinued or been excluded from the course. No significant differences were found between the numerical scores allocated for personal attributes to students who later withdrew from the course and those who continued. However, students who withdrew had statistically significantly more negative comments made about them at interview with respect to their supportive and encouraging behaviour (p = 0.04) and their motivation to become doctors (p = 0.05). Although no other comparisons reached statistical significance, students who withdrew consistently received more negative comments and fewer positive comments than their continuing peers. A stepwise logistic regression procedure identified the number of negative comments made by interviewers about a candidate on their supportive and encouraging behaviour as the only statistically significant predictor of later withdrawal (odds ratio 1.65 95% confidence interval 1.01-2.70).