BackgroundMRCGP and MRCP(UK) are the main entry qualifications for UK doctors entering general [family] practice or hospital [internal] medicine. The performance of MRCP(UK) candidates who subsequently take MRCGP allows validation of each assessment.In the UK, underperformance of ethnic minority doctors taking MRCGP has had a high political profile, with a Judicial Review in the High Court in April 2014 for alleged racial discrimination. Although the legal challenge was dismissed, substantial performance differences between white and BME (Black and Minority Ethnic) doctors undoubtedly exist. Understanding ethnic differences can be helped by comparing the performance of doctors who take both MRCGP and MRCP(UK).MethodsWe identified 2,284 candidates who had taken one or more parts of both assessments, MRCP(UK) typically being taken 3.7 years before MRCGP. We analyzed performance on knowledge-based MCQs (MRCP(UK) Parts 1 and 2 and MRCGP Applied Knowledge Test (AKT)) and clinical examinations (MRCGP Clinical Skills Assessment (CSA) and MRCP(UK) Practical Assessment of Clinical Skills (PACES)).ResultsCorrelations between MRCGP and MRCP(UK) were high, disattenuated correlations for MRCGP AKT with MRCP(UK) Parts 1 and 2 being 0.748 and 0.698, and for CSA and PACES being 0.636.BME candidates performed less well on all five assessments (P < .001). Correlations disaggregated by ethnicity were complex, MRCGP AKT showing similar correlations with Part1/Part2/PACES in White and BME candidates, but CSA showing stronger correlations with Part1/Part2/PACES in BME candidates than in White candidates.CSA changed its scoring method during the study; multiple regression showed the newer CSA was better predicted by PACES than the previous CSA.ConclusionsHigh correlations between MRCGP and MRCP(UK) support the validity of each, suggesting they assess knowledge cognate to both assessments.Detailed analyses by candidate ethnicity show that although White candidates out-perform BME candidates, the differences are largely mirrored across the two examinations. Whilst the reason for the differential performance is unclear, the similarity of the effects in independent knowledge and clinical examinations suggests the differences are unlikely to result from specific features of either assessment and most likely represent true differences in ability.

No abstract

Context High‐stakes medical examinations seek to be fair to all candidates, including an increasing proportion of trainee doctors with specific learning differences. We aimed to investigate the performance of doctors declaring dyslexia in the clinical skills assessment (CSA), an objective structured clinical examination for licensing UK general practitioners. Methods We employed a cross‐sectional design using performance and attribute data from candidates taking the CSA between 2010 and 2017. We compared candidates who declared dyslexia (‘early’ before their first attempt or ‘late’ after failing at least once) with those who did not, using multivariable negative binomial regression investigating the effect of declaring dyslexia on passing the CSA, accounting for relevant factors previously associated with performance, including number of attempts, initial score, sex, place of primary medical qualification and ethnicity. Results Of 20 879 CSA candidates, 598 (2.9%) declared that they had dyslexia. Candidates declaring dyslexia were more likely to be male (47.3% versus 37.8%; p < 0.001) and to have a non‐UK primary medical qualification (26.9% versus 22.4%; p < 0.01), but were no different in ethnicity compared with those who never declared dyslexia. Candidates who declared dyslexia late were significantly more likely to fail compared with those candidates who declared dyslexia early (40.6% versus 9.2%; p < 0.001) and were more likely to have a non‐UK medical qualification (79.3% versus 15.6%; p < 0.001) or come from a minority ethnic group (84.9% versus 39.2%; p < 0.001). The chance of passing was lower for candidates declaring dyslexia compared to those who never declared dyslexia and lower in those declaring late (incident rate ratio [IRR], 0.82; 95% confidence interval [CI], 0.70–0.96) compared with those declaring early (IRR, 0.95; 95% CI, 0.93–0.97). Conclusions A small proportion of candidates declaring dyslexia were less likely to pass the CSA, particularly if dyslexia was declared late. Further investigation of potential causes and solutions is needed.

The Recorded Consultation Assessment (RCA) was rapidly developed to replace the Clinical Skills Assessment (CSA) for UK general practice licencing during COVID-19. We aimed to evaluate candidate perceptions of the RCA and relationships with performance. We conducted a crosssectional survey of RCA candidates with attitudinal, demographic, and free text response options, undertaking descriptive and factor analysis of quantitative data with qualitative thematic analysis of free text. Binomial regression was used to estimate associations between RCA pass, candidate characteristics and questionnaire responses.645 of 1551 (41.6%) candidates completed a questionnaire; 364 (56.4%) responders permitted linkage with performance and demographic data. Responders and non-responders were similar in exam performance, gender and declared disability but were significantly more likely to be UK graduates (UKG) or white compared with international medical (IMG) or ethnic minority graduates. Responders were positive about the digital platform and support resources. A small overall majority regarded the RCA as a fair assessment; a larger majority reported difficulty collecting, selecting, and submitting cases or felt rushed during recording.Logistic regression showed that ethnicity (white vs minority ethnic: odds ratio [OR] 2.99,95% confidence interval [CI] 1.23, 7.30, p = 0.016), training (UK vs IMG: OR 6.88, 95% CI 2.79, 16.95, p < 0.001), and English as first language (OR 5.11, 0% CI 2.08, 12.56, p < 0.001) were associated with exam success but questionnaire subscales, consultation type submitted, or extent of trainer review were not. The RCA was broadly acceptable but experiences were variable. Candidates experienced challenges and suggested areas for improvement.

Seeking to counter this systematic unintended discrimination could be the single most important way of ensuring the highest standards of training."