Trends in the Results of Part II and III of the German Medical Licensing Examinations

Objective Trend analysis of the results of the second clinical (written, "M2") and third clinical (oral, "M3") part of the German medical licensing examination. Methods Publicly available data were analyzed (2014-2020) using meta-regression models (mixed-effects) for evaluation of correlations between results and the time point of the examinations. Results A total of 63,811 and 32,852 students participated in the M2 and M3 examinations, respectively. The average percentage of correctly answered questions across all M2 examinations was 77.7 % (minimum: 73 %; maximum: 82.7 %). Generally, there was great variation: The percentage of participants with "very good" results varied between 0.1 % (spring 2020) and 17 % (fall 2015) as did the percentage of participants with "good" results (between 21 % (spring 2020) and 52 % (fall 2015)). Analysis of M2 examinations showed an overall negative trend. Over time, for example, the percentage of correctly answered questions (slope beta = -0.82 %; p < 0.01) and the percentage of participants with "very good" (beta = -1.17; p = 0.04) and "good" (beta = - 2.58; p = 0.02) results decreased significantly. The time point of the examinations explained between 20 and 34 % of heterogeneity between examination results. With regard to the M3 examinations, there was only little variation over time, and no negative trends were observed (p = 0.08-0.60). Conclusion Results of the M2 examinations varied greatly over the last years, and there was a significant negative trend. In contrast, results of the M3 examinations remained constant with little variation. Possible explanations such as worsening written skills or higher requirements in the M2 examinations need to be investigated. The observed great variation of average results questions the reliability and comparability of the M2 examinations in Germany.