3 Problems with diagnostic classifications
3.1 Validity and reliability
Fish’s classification seems simple enough. However, an indication of the fragile and arbitrary character of diagnosis can be seen if we look at the ways in which the boundaries between ‘variations in mental life’ and ‘mental illnesses’ become readily permeable, as do the divisions within these headings. If ‘abnormal variations’ become chronic, then they are reclassified as ‘personality disorders’ or ‘paranoid schizophrenia’. Also, many doctors consider neuroses to be minor ‘mental illnesses’. Thus there is only a precarious agreement on divisions of labelling by those adopting a diagnostic approach.
There are other fundamental problems about diagnosis today regarding validity and reliability (see box for definitions).
Box 2 Validity and reliability
Validity refers to whether there is objective evidence to support a diagnosis. Conceptual validity is whether a diagnostic category is conceptually separate from others (see Case Study 2, below). Predictive validity is whether diagnosis can predict the outcome of an illness. Reliability relates to whether it is a stable diagnosis (would different people reach the same diagnosis, and would the patient be diagnosed the same at different points in time?).
Case Study 2: Different diagnoses
Julie has been hospitalised because she experiences frightening hallucinations following her cannabis smoking (which she has been doing daily for several years). One psychiatrist argues that she has ‘drug-induced psychosis’, whilst another says that she has developed ‘schizophrenia’. Another suggests the diagnosis of ‘substance misuse’. When in hospital without access to her preferred drug, Julie becomes depressively withdrawn and is deemed to be suffering from ‘major depression’ and is treated with ‘anti-depressants’. Between hospital admissions, she gets involved in petty crime and moves from one difficult sexual relationship to another, which at times spills over into domestic violence, with police involvement. This culminates in another psychiatrist recording a diagnosis of ‘dual diagnosis’, which implicates a ‘personality disorder’. A few years down the line all of these diagnoses appear cumulatively in Julie’s case notes.
Conceptual validity problems are also seen within specific diagnoses. For example, schizophrenia is a disjunctive concept (Bannister, 1968): two patients with the diagnosis may have no symptoms in common. Whilst reliability can be improved by psychiatrists being trained carefully in the use of common symptom checklists (such as those in the DSM), reliability is not the same as validity and it is possible to consistently use a label which is still not valid. Predictive validity is particularly imperfect in psychiatry, because human behaviour (of any sort) is difficult to predict accurately. These problems with the validity and reliability of psychiatric diagnoses have led some to argue that mental disorder is very difficult to measure and that the dividing line between the normal and abnormal is fuzzy (Wakefield, 1992).