The Editorial by Benning & Broadhurst (Reference Benning and Broadhurst2007, this issue) is an impassioned cri du coeur bemoaning the abandonment of the long case examination in future MRCPsych examinations. In Spring 2008 the clinical examination will consist of an objective structured clinical examination (OSCE) in two parts and both the patient management problems and the individual patient assessment (the long case) will be discontinued; this is a substantial change in emphasis.

The authors correctly point out that the long case examination has been used for over 150 years in final medical examinations and believe that the cessation of this test will lead to a failure to test ‘the ability to integrate and synthesise all of the information obtained from an interview [with a patient]’. This part of the MRCPsych examination was until a few years ago considered to be the most important component of both the MRCPsych Part I and Part II examinations, and failure in this section of the examination in either part meant an irretrievable fail whatever the results in the other components. Candidates who took the MRCPsych examinations in the late 1980s and 1990s will be aware of the importance of an adequate formulation of each long case that they saw. It is this skill that the authors are concerned will be lost if it is no longer necessary to examine a long case in the examination.

I believe too that the ability to take a history and perform a mental state and appropriate physical examination in a patient with a psychiatric illness is an essential skill for a psychiatrist. The issue is how this skill should be assessed. The authors state that the paramount need to ensure standardisation is the main reason why the long case has lost its attraction. However, there are more convincing explanations why the long case is considered by many to have served its purpose in testing doctors in their final summative examinations.

The prime reason why the long case is facing demise in the MRCPsych examination is that it is an extremely unreliable test. Studies have shown that the reproducibility coefficient of marks achieved in a long case are as low as 0.24 (Reference Meskauskas, Lloyd and LangsleyMeskauskas, 1983). This means that 76% of the variability in scores is due to errors of measurement, and little credence can be attached to this one result as a measure of competence. These results are not comparable with results from OSCEs, which achieve reproducibility coefficients of over 0.72 or better (Reference Wass, Jones and van der VleutenWass et al, 2001), 0.73 in other forms of clinical examinations and 0.88 for multiple choice question written papers (Reference NorciniNorcini, 2002).

The results illustrated above were obtained in traditional long case examinations in which the candidate interviews a patient alone and is then questioned by the examiners on the history and examination of the patient. Although there is a substantial increase in reliability if the interview is observed directly by examiners (Reference Wass and JollyWass & Jolly, 2001), time constraints make this procedure difficult to carry out in examination practice.

It is therefore clear that the long case has not passed ‘the rigorous scrutiny of modern medical education’ as the authors assume many of us believe. It is primarily because of the poor reliability of candidates’ scores in this type of test that the Royal College of Physicians discontinued the long case in the Part II MRCP examination in 2001 and replaced this with a form of OSCE entitled practical assessment of clinical examination skills (PACES). This assesses the clinical skills of history-taking and examination, the interpretation of physical signs, development of management plans, communication of clinical information and appreciation of ethical issues.

The main reason for the considerable disparity of intercase scores in the long case is because of the degree of complexity of different cases that are selected in examinations (Reference Elstein, Shulman and SprafkaElstein et al, 1978). Examination of a patient with a bipolar mood disorder who at examination has only a few residual symptoms of affective illness is a radically less difficult proposition than the assessment of the essential features of a dementing illness in a patient who is accompanied by an informant with rudimentary knowledge of the patient. Candidates may be lucky or unlucky in the selection of patients they are asked to see in an examination and will perform above or below their general ability depending on the nature of the illness that the patient has.

In addition, because of the subjective nature of assessment in a long case examination, examiner unreliability is high. For scores to be reproducible, examiners must apply the same standards. It has been shown that examiners differ considerably when assessing long case encounters even when assessing the same event (Reference Noel, Herbers and CaplowNoel et al, 1992).

I agree with Benning & Broadhurst that the standardisation of examinations involves assessment of objective data and neglect of subjective information. Subjective judgements by definition involve individual bias and should not be assessed positively in an examination. The skill of making an accurate formulation is based on weighing up all the information obtained and making an accurate appraisal of the patient from this. This relies on identification of salient features from the history and examination of the patient and determining which are of most importance in contributing to the presentation of the patient at interview. This requires judgement on the part of the enquirer; this can be assessed objectively.

It is possible to overcome these difficulties by examining each candidate on a number of long cases (Reference McKinley, Fraser and van der VleutenMcKinley et al, 2000; Reference NormanNorman, 2002). Wass et al (Reference Wass, Jones and van der Vleuten2001) showed that the reliability of long case assessments could be increased to a figure of 0.84 if 10 cases were seen by each candidate, and this compared very favourably with a reliability coefficient of 0.72 of a greater number of OSCEs carried out concurrently. Unfortunately it is not feasible to test candidates with such a high degree of rigour in the MRCPsych examination or any other postgraduate medical examination because of the vast degree of resources that would be required.

The authors assume that because the OSCE is standardised it cannot measure the skill of taking a psychiatric history successfully. It is true that a checklist marking process in an OSCE examination is not suitable for the assessment of more advanced psychiatric skills (Reference Wilkinson, Frampton and Thompson-FawcettWilkinson et al, 2003; Reference Tyrer and RaoTyrer, 2005). However, when marked according to more global judgements better discrimination is obtained (Reference Regehr, Freeman and RobbRegehr et al, 1999). Furthermore, the essential elements of a full history can be assessed in an OSCE by assessing different aspects of the history in a longitudinal format. This can be carried out by having two or more stations in the OSCE concerned with different aspects of the history of the same patient. An appropriate examination station can also be included if necessary. Although not comparable entirely with the same assessment in a long case, this scenario enables an assessment to be made of more aspects of a single clinical case than can be identified in one OSCE station. It is proposed that part of the OSCE in the MRCPsych examinations next year should consist of five pairs of linked stations, which should allow for the assessment of more complex competences.

Benning & Broadhurst do not mention an advantage of the long case in the assessment of a true patient. The ability of a psychiatrist to evaluate a patient in the flesh is important in determining competence in practice. It may be possible to use real patients in an OSCE in the future but this may present difficulties in ensuring standardisation of the encounter. Ethical problems may affect the choice of patients (Reference Sayer, Bowman and EvansSayer et al, 2002).

Although this commentary makes it clear that the disadvantages of a long case assessment are sufficient to preclude its use in an MRCPsych Part II examination, I agree fully that the ability to take an accurate history is essential for any psychiatrist. This ability should be assessed but it is impractical to do this within a formal examination setting. Such an assessment should be carried out during training as part of what is described as a formative assessment. Assessments of interviews with patients and relatives in a variety of clinical situations should be carried out at regular intervals during senior house officer training. The new regulations for the MRCPsych examinations attempt to include such assessments. Candidates will be expected to complete a minimum of eight assessments of clinical expertise (ACEs) as well as a number of other workplace-based assessments. Three ACEs will be assessed by a validated College-approved assessor, with these marks counting towards the final clinical mark of the OSCE as part of a summative assessment.

It should be a requirement that trainees pass such assessments before progressing further in training. Pavlakis & Laurent (Reference Pavlakis and Laurent2001) have shown that candidates who have training in interviewing techniques perform better than naïve trainees in obtaining salient information when taking a history from a patient.

In this way the long case will not die but will be successfully resuscitated. It should live again.

Declaration of interest

None.