DSM-III prompted a series of research studies which identified problems with some of the new categories and criteria. This prompted Spitzer to revise the manual, leading to the publication of DSM-III-R in 1987. By this time, administering structured interviews to service users was generally considered to increase the reliability of diagnosis, therefore Spitzer, together with some colleagues, developed one for DSM-III-R. There has been a structured clinical interview for each subsequent edition of the DSM.
Ultimately, DSM-III marked a radical diversion from previous DSMs, with its increased concern about the reliability of diagnosis and its move away from psychodynamic concepts. The claim that it was atheoretical was somewhat undermined by the introduction of the PTSD category and of symptoms more consistent with the medical model, as well as its retention of personality disorder subcategories with a psychodynamic heritage.
4 DSM-IV to DSM-5: the end of an era?
DSM-IV was much less radical in scope than DSM-III, but DSM-5 rejected many of the principles enshrined by Spitzer in DSM-III.
4.1 DSM-IV and DSM-IV-TR
DSM-IV was published in 1994 (APA, 1994). Its goals, as described by its chair, psychiatrist Allen Frances, were modest: they were ‘to introduce rigour, objectivity, and transparency in how decisions were made’ (Frances, 2013, p. 70). Conceptually and structurally, DSM-IV largely followed DSM-III, though it attempted to address the issue of cultural variation in mental health problems. Although the DSM is a US system, it is used in many countries and, of course, the US population itself is culturally varied. Culture researchers therefore argued that the manual needed to acknowledge that its construction of mental health was culturally shaped, but DSM-IV took a more minimalist approach. It included an appendix with an outline of a cultural formulation and a glossary of so-called ‘culture-bound syndromes’. These could refer to the way in which a DSM disorder was expressed differently in different cultures, or to an indigenous ‘folk’ category of distress. An example given by DSM-IV was ‘amok’, which referred to ‘a dissociative episode characterized by a period of brooding followed by an outburst of violent, aggressive, or homicidal behavior directed at people and objects’ (APA, 1994, p. 845). This had apparently been reported primarily (although not exclusively) in Southeast Asia.
A further edition (DSM-IV-TR) was published six years later, in 2000, though the revisions were primarily textual rather than conceptual.
4.2 DSM-5: the end of Spitzer’s revolution?
Those involved in planning DSM-5 challenged key tenets of Spitzer’s approach. They identified a number of problems stemming from the changes implemented in DSM-III. These problems included the narrow and overly rigid categories introduced in DSM-III. As noted in the introduction to DSM-5: ‘the once plausible goal of identifying homogenous populations for treatment and research resulted in narrow diagnostic categories that did not capture clinical reality, symptom heterogeneity within disorders, and significant sharing of symptoms across multiple disorders’ (APA, 2013, p. 12). DSM-5 argued that apparently high rates of comorbidity were a result of overly narrow categories that required clinicians to make more than one diagnosis, something that could be circumvented by using broader categories.
However, during the DSM-5 planning process, the government-funded National Institute for Mental Health (NIMH) had proposed an entirely different way forward: the Research Domain Criteria (RDoC) project. Thomas Insel (the NIMH director), and his colleagues, proposed that the answer lay not in the conceptualisation of clinical categories but in the identification of problems in the brain itself, in ‘neural circuitry’ (Insel et al., 2010). However, the team acknowledged this was a high risk, long-term option with no guarantee of a successful outcome. The competing approaches of the DSM-5 planners and the RDoC project suggested that there was a growing divide between practitioners and biomedically oriented psychiatric researchers.
DSM-5 was published in 2013 (APA, 2013) and led to widespread debate among academics, professionals, service users and in the media. DSM-5’s feedback website received over 13,000 comments and thousands of organised petitions (APA, 2013). DSM-5 argued for the need to adopt a more dimensional system of diagnosis rather than a categorical approach. In a categorical approach you either meet criteria for a category or you do not, whereas in a dimensional approach, problems are viewed as lying along a spectrum and are thus a matter of degree. A dimensional system for personality disorders had originally been proposed by the authors of DSM-5 but, since this proposal had ‘not been widely accepted’ (APA, 2013, p. xliii), the personality disorder categories in Section II were largely unchanged; the proposed dimensional approach to personality disorder was, instead, included in Section III, ‘emerging measures and models’. This section also included much more extensive detail on cultural formulation, and one of the appendices included a glossary of cultural concepts of distress. DSM-5 aimed for a simplified system of diagnosis and it abandoned the multi-axial system introduced in DSM-III. Axes I–III were now combined and axis IV was dropped (in favour of ICD codes for environmental factors). Axis V (the rating system to assess a person’s everyday functioning) was abandoned due to its lack of conceptual clarity and questionable psychometric characteristics. A final change was that DSM-5 was referred to by Arabic rather than Roman numerals (e.g. ‘DSM-III’). This was to enable incremental updates in the future; thus, if and when DSM-5.1 is published, it may include updates only on some categories and will not be a wholesale revision as previous editions have been.
Dimensional system of diagnosis A system of organising diagnoses that assumes that mental health problems are best understood as existing along a spectrum or dimension (e.g. how much they disrupt one’s life). Different types of dimensions have been proposed for different forms of distress.
In summary, DSM-IV largely followed the approach of DSM-III, focusing more on the process of decision-making. DSM-5, on the other hand, challenged the conceptual framework of the third edition, although its more radical proposal to move to a dimensional system was thwarted.
5 Debates about DSM-5 and psychiatric diagnosis
The publication of DSM-5 led to widespread debate. This section will discuss some of the issues raised, issues which have been encountered throughout the history of the DSM.
5.1 Reliability and validity redux: the appropriateness of a medical lens and the role of social norms
DSM-III had aimed to improve the reliability of diagnosis and its field trials suggested that, through using structured clinical interviews, levels of reliability ranging from fair to satisfactory could be achieved between raters. However, this meant that symptoms had to be defined that were simple and obvious, with ‘personal and contextual factors’ omitted (Frances, 2013, p. 25). This raised the risk that the reliability of diagnosis had been prioritised over the validity of the categories. Some have argued that the use of the structured clinical interview for DSM-IV (commonly referred to as the SCID) has not improved reliability (Kutchins and Kirk, 1999). However, even if structured interviews did improve reliability, they are only used by researchers; they are not in everyday use by psychiatrists. Moreover, reliability is no guide to validity. We can all agree on the characteristics of a unicorn or of Santa Claus but that does not mean that either exists in the real world.
The problem of reliability reared its head again when the DSM-5 field trials were published by Freedman et al. (2013). Figure 4.1 shows their reliability results for categories for adults. DSM-5 aimed to develop categories that could be applied by ordinary clinicians with relatively little training and using unstructured clinical interviews. Reliability is measured by the kappa (κ) coefficient, where a score of −1 indicates complete disagreement and a score of 1 indicates complete agreement. Spitzer and Fleiss’s (1974) article implies that a kappa value of 0.4–0.7 is no better than fair agreement,