The nature and purposes of the ICD and DSM

Comparing DSM and ICD, Tyrer (Reference Tyrer2014) refers to how ‘[w]ithout a classification system the necessary economical communication with colleagues to convey information becomes a lengthy description of clinical problems that is self-defeating’. Both are used for communication and have some similarities, but there are significant limitations, and also differences, that affect their utility in medico-legal reporting.

The differences partly reflect the different priorities of the World Health Organization (WHO) and the American Psychiatric Association (APA). The WHO’s ICD is a comprehensive classification of mental, behavioural and neurotypical disorders for use by a wide range of health professionals in countries of very varied sizes, cultures and resources, whereas the APA’s DSM is designed to meet the needs of one, or perhaps two, professions – psychiatrists and psychologists – in a single country (Kendell Reference Kendell1991).

Clark et al (Reference Clark, Cuthbert and Lewis-Fernández2017) note that they share ‘many features (e.g., a categorical structure) that respond to the purposes for which they were developed and primarily used, which include compilation of health statistics, allocation of mental-health resources, clinical communication, and decision making in regulatory, legal, and health-insurance systems, all ultimately in service of public mental-health-care needs’. Whether developed for use in legal systems is questionable but they are used.

The limitations of the ICD and DSM are that although both are regarded as categorical systems, this is questionable. Mental disorders are non-taxonomic. They are not unique, discrete entities. Most exist on dimensions. There are arbitrary boundaries on the ‘normality–pathology’ dimension, requiring a ‘threshold boundary’, and on the dimensions connecting overlapping disorders, between which there is no ‘zone of rarity’ and which therefore require a ‘disorder boundary’.

Using the example of depression, Tyrer (Reference Tyrer2014) illustrates the requirement for a threshold boundary, but notes that the threshold is not a clean one; many people just below threshold have depressive symptoms that do not qualify for a diagnosis even though research shows that they may be as unwell as others just over the threshold. Using the example of anaemia, Tyrer illustrates graphically the zone of rarity between iron-deficiency anaemia and pernicious anaemia. They do not overlap. As Kendell (Reference Kendell1991) says, about psychiatric syndromes, it has not been possible to identify points of rarity that separate individual syndromes from their neighbours.

There is no zone of rarity between DSM’s major depressive disorder and generalised anxiety disorder (Zhou Reference Zhou, Cao and Yang2017). Figure 1 illustrates how DSM categorisation, when applied rigidly and without the required flexibility and clinical judgement, misleads as to the nature of major depressive disorder and generalised anxiety disorder. As the introduction to ICD-11 (WHO 2024) states, both of these boundaries represent key issues in diagnosis, so unsurprisingly they also represent key issues when diagnoses are used in medico-legal reporting. Experts who rely on these boundaries to diagnose, or not diagnose, mental disorders or to distinguish one from another risk misleading the court.

FIG 1 How misapplication of DSM categorisation can mislead as to the nature of major depressive disorder and generalised anxiety disorder.

ICD and DSM are better regarded as nomenclatures than classifications.

The history of the ICD and DSM

In 1948, the WHO agreed to use its International Statistical Classification of Diseases, Injuries and Causes of Death, the first version of which originated in the first International Statistical Congress in Brussels in 1853, as ‘a global common language for defining and communicating about diseases and health conditions’ (Clark Reference Clark, Cuthbert and Lewis-Fernández2017), thus supporting the WHO’s objective of achieving the highest level of health for all people by pursuing universal healthcare. This public health focus was fundamental to ICD-6 (1949) and it has remained so. In ICD-8 (1966) the WHO made clear that the purpose of its glossary was to facilitate diagnosis in clinical settings as well as serve as a statistical classification system. When ICD-10 was issued in 1992, it was described as being ‘intended for general clinical, educational and service use’ (WHO 1992: p. 1).

The DSM also has its roots in the 19th century. It grew out of the statistical manuals used to document diagnoses of patients in US psychiatric institutions. However, both classifications can be traced back to Brigadier General William Menninger, MD, who was Chief of the US Army Medical Corps’ Psychiatric Division during the Second World War and developed a mental disorder classification better suited to military medicine. After the war he recognised a need for a diagnostic manual suitable for civilian practice and developed, for use by the Veterans Health Administration, the Nomenclature of Psychiatric Disorders and Reactions (Office of the Surgeon General 1946). In 1948, the WHO drew heavily on this when it published ICD-6 and the APA did likewise when it published its first DSM in 1952 (DSM-I).

There is a misconception that the ICD is not used in the USA. Adapted ICD codes, known as ‘Clinical Modifications’, are required for Medicare billing and reimbursement, for coding morbidity and healthcare utilisation from patient records, and to provide statistical returns to the US National Center for Health Statistics.

The processes of revision

For more than 60 years countless committees have produced updated and revised versions of the ICD and DSM. Thus, they are based on consensus about clusters of clinical symptoms. They are not based on any objective laboratory measure or biological marker or on research that has established that a particular disorder can be sufficiently distinguished from other similar disorders, in terms of its epidemiology, response to treatment and prognosis, to justify its recognition as a discrete entity. Explaining why the National Institute of Mental Health (NIMH) would no longer fund research based on DSM-5, Thomas R. Insel, Director of NIMH, said:

Where expert witnesses rely on DSM, or ICD, diagnoses in court, they can expect to be challenged as to their lack of validity and the absence of research that supports their delineation.

In the case of ICD, there are committees of representatives from nearly 200 WHO member states, so the ICD represents international consensus as to the detailed descriptions of conditions that are used for a process mainly of prototypical diagnosis, matching a person’s psychopathology to a detailed description in the guidelines. The description requires a face validity acceptable in as many member states as possible. The classification has to accommodate cultural variations in psychopathology, clinical judgement and diagnostic practice (First Reference First, Gaebel and Maj2021). Global applicability is paramount. The prioritisation of public health needs, based on the priorities of sponsoring organisations, has been proposed as the reason why ICD-11 includes 11 disorders that are not in DSM-5 (Clark Reference Clark, Cuthbert and Lewis-Fernández2017).

DSM criteria are used for an algorithmic approach to diagnosis, where criteria are counted (‘≥5 symptoms’), time periods specified (‘during the same 2-week period’) and exclusion criteria specified. The basis for the algorithmic approach is questionable. Paris (Reference Paris2015: p. 9) has observed: ‘But if a typical DSM diagnosis requires, for example, five out of nine criteria, nobody knows whether four or six criteria would have been more or less valid’.

The over-rigid application of the DSM and ICD criteria is illustrated by the case of R (B) v Dr SS, Responsible Medical Officer [2005] (Box 1).

Agreement as to diagnosis can conceal significant disagreement, which can be difficult to explain in court, especially if the court wants to believe that these diagnostic systems are absolute and unshakeable:

When the nature of a person’s mental condition is at issue, the court is likely to need to know how the diagnosis is made and about the validity of the diagnostic category. This may require explanations about the roles of committees, political influences, pressure groups and the financial interests of pharmaceutical companies (McHugh Reference McHugh1999; Paris Reference Paris2015; Kapp Reference Kapp, Ne’eman and Kapp2020; Horwitz Reference Horwitz2021). There may be questions about the scientific basis for distinctions based on algorithmic criteria.

The limitations of the ICD and DSM diagnostic categories are illustrated in general clinical practice, where there are high rates of use of the categories ‘not otherwise specified’ (ICD) and ‘not elsewhere defined’ (DSM) (Clark 2017). These are the categories into which the baby gets thrown with the bathwater. In the everyday practice of the real world, where, to quote Kendell (Reference Kendell1975: p. 3), ‘the majority of patients do not conform to the tidy stereotyped descriptions found in textbooks’, there will be more babies thrown out with the bathwater than are left in the bath or, in the language of Clark et al (Reference Clark, Cuthbert and Lewis-Fernández2017), more in the ‘wastebasket categories’. The value of a diagnostic system that works in only a minority of patients is questionable. It calls into question how typical, for example, the defendant with ICD or DSM schizophrenia is of all people with schizophrenia.

D v The Bishop’s Conference of Scotland [2022] (Box 2) is a case in which the court found psychiatric injury even though D’s condition did not fall within DSM-5 (American Psychiatric Association 2013).

Autism spectrum disorder illustrates the difference between the algorithmic and prototypical approaches (First Reference First, Gaebel and Maj2021). For persistent deficits in social communication/social interaction, DSM-5-TR (American Psychiatric Association 2022) requires all three specific manifestations, whereas ICD-11 (WHO 2024) states that the ‘manifestations may include the following’ (emphasis added) and lists seven manifestations, three of which correspond to the DSM-5 requirements. For restrictive, repetitive patterns of behaviour, ICD-11 provides a list of seven items as examples, but DSM-5 is more prescriptive as it requires two out of a list of four items.

These are matters that need to be taken into consideration when deciding what weight to attach to ICD or DSM diagnoses and to the absence of a diagnosis where criteria are not met.

Proceed with caution

There is some recognition of these limitations in the introductions to both the ICD (Box 3) and DSM (Box 4).

These cautionary statements make clear that the use of the ICD and DSM is to be informed by clinical judgement and they call for flexibility. These publications are neither comprehensive statements about the current knowledge of the disorders nor comprehensive textbooks of psychiatry.

Given these cautionary notes it seems somewhat surprising that a psychiatric expert in Ludlow v National Power Plc [2000] should have been critical of the DSM as being too formulaic to be discriminating and to have said that it marginalised clinical judgement and common sense.

Few courts and tribunals seem to be aware, or advised by experts, of such warnings, notwithstanding the fact that in DSM-5 they appear under a bold heading ‘Cautionary Statement for Forensic Use of DSM-5’. In McTear v Imperial Tobacco Ltd [2005] an expert agreed that the passage in DSM-IV (Box 4) was, in general terms, the correct approach and said that it urged the use of caution in the drawing of conclusions from the fact that an individual’s presentation met the criteria for a DSM-IV diagnosis.

Having regard to these warnings, any reliance on the ICD or DSM in making diagnoses in evidence for courts and tribunals has to be viewed with caution.

Such caution is also necessary where there has been authoritative criticism of the ICD or DSM, such as the criticism by Kopelman & Fleminger (Reference Kopelman and Fleminger2002) of the organisation and definitions of neuropsychiatric conditions in DSM-IV and ICD-10, many of which criticisms still hold, and Kopelman’s (Reference Kopelman2022) criticism of ICD-11’s definitions of Korsakoff syndrome and alcohol-related brain damage.

Personal injury litigation

Rorrison v West Lothian College [1999] (Box 5) illustrates how it is common practice in personal injury pleadings in Scotland to aver that a pursuer was or is suffering from a specified condition recognised in the DSM or ICD. When a similar issue arose in Laudanska v University Abertay [2003] the court referred to the ‘hypothesis’ that the decision in Rorrison was correct and how it was not convinced that it should be automatically followed; but it was. In Mather v British Telecommunications Plc [2000], the pursuer averred that she had developed and suffered from a panic disorder with agoraphobia in DSM-IV criteria. The court found that she had plainly averred that she had suffered from a recognised psychiatric illness.

In Hussain v The Chief Constable of West Mercia Constabulary [2008], the court said: ‘A recognised psychiatric illness is one which has been recognised by the psychiatric profession. In general, they are illnesses that are within the ICD […]’.

Although the Judicial College Guidelines for the Assessment of General Damages in Personal Injury Cases (Judicial College 2024) does not advise the use of the ICD or DSM, its guidance on awards for post-traumatic stress disorder refers to how it relies on ‘cases which variously reflect the criteria established’ (p. 14) in DSM-IV and DSM-5.

Professional regulation

The General Medical Council and the Nursing and Midwifery Council use the ICD when referring to psychiatric diagnoses (e.g. Stefan v General Medical Council (Health Committee of the General Medical Council) [2001]).

Employment litigation

The Employment Tribunal, which is concerned with ‘mental impairment’ resulting from, or consisting of, a mental illness only if the illness is ‘clinically well recognised’, applies guidance issued by the Department for Education and Employment which states that it is very likely that a respected body of medical opinion would include as clinically well-recognised those illnesses specifically mentioned in publications such as the ICD. However, the use of ‘include’ and ‘such as’ does allow for reliance on diagnoses that are not in the ICD. This was recognised in Morgan v The Staffordshire University [2002], where, in addition to proof of a mental illness specifically mentioned in the ICD, the tribunal recognised at least two other routes to establishing ‘mental impairment’: (a) proof of a mental illness specifically mentioned in a publication, ‘such as’ the ICD, of very wide professional acceptance; (b) proof by other means of a medical illness recognised by a respected body of medical opinion.

Failure by a psychiatrist to make what the court called ‘a statement to the effect of whether the recurrent depressive illness comprised (in his view) a recurrent depressive disorder (or some other illness) for the purposes of ICD-10’ in Woods v Royal College of Nursing [2003] resulted in the applicant’s failure to succeed on the issue of disability. The tribunal did state that it might have been that, if the applicant had produced an expert’s report guiding the tribunal as to the type of illness from which she suffered (and whether it was contained within the ICD and if so where), the tribunal would have made a different decision. Likewise in Wilson v Southern Counties Fuels Ltd [2004], the tribunal referred to the failure to comply with guidelines as a bad start. This was a case in which there was a diagnosis of ‘clinical depression’ but the tribunal said that what was required was ‘to bring someone who is, or has been, suffering from depression within one of the clinically well recognised illnesses’. The term ‘clinical depression’ is for many psychiatrists synonymous with a depressive illness or episode with biological or somatic symptoms, and very much a clinically well-recognised illness. The outcome of both of these cases might have been different if reference had been made to ICD-10 or, in the case of Wilson, a statement that a depressive illness or episode with biological or somatic symptoms is clinically well-recognised.

Immigration and asylum cases

Another example of a court or tribunal regarding it as a failure not to use the DSM or ICD is in the Asylum and Immigration Tribunal (AIT). In 2005 a judge in an AIT referred to failure to apply DSM-IV or ICD-10 as indicative of a lack of rigour on the part of the expert (DE (suicide, psychiatric treatment, J applied) Turkey [2005]).

Health, Education and Social Care Chamber (Mental Health)

The Tribunals Judiciary (2019) Practice Direction on statements and reports in mental health cases before the First-tier Tribunal (Mental Health) (Health, Education and Social Care Chamber) states that the report must include ‘whether the patient is now suffering from a mental disorder and, if so, whether a diagnosis has been made, what the diagnosis is, and why. There is no requirement to use any particular classification or nomenclature’ (para. 12(f)).

A comment in the Upper Tribunal (Mental Health) (Health, Education and Social Care Chamber) gives an indication of the position that might be taken when the issue is one of classification for the purposes of the Mental Health Act 1983. In DL-H v Devon Partnership NHS Trust v Secretary of State for Justice [2010], the tribunal made this comment for any value it might have in the future, pending consideration by the Upper Tribunal in an appropriate case:

The case of B outlined in Box 1 illustrates what happens when the diagnostic criteria are used ‘over-prescriptively’.

It is doubtful that if the Mental Health Bill 2025 (for England and Wales) is enacted without changes to Clause 1(3), the position will be any different. This clause includes a reference to ‘psychiatric disorder’, which ‘means mental disorder other than autism or learning disability’ (the latter being a term which it is reasonable to regard as applying to ICD-11’s ‘disorders of intellectual ability’ and DSM-5’s ‘intellectual disability (intellectual development disorder)’. It does not appear that this new form of categorisation of mental disorders will affect any reliance on the ICD or DSM.

Provided that, if there is reliance on the DSM for a diagnosis of autism, the requirement to satisfy all three criteria for persistent deficits in social communication and social interaction or at least two out of four manifestations of restricted repetitive patterns of behaviour, interests or activities, is not interpreted over-prescriptively, there should be no problem. It is doubtful that the tribunal will decide that someone does not have autism on the basis that only two of the three social communication criteria are satisfied or just one manifestation of restrictive, repetitive patterns of behaviour, interests or activities.

There are minor definitional differences between ICD-11’s ‘disorders of intellectual disability’ and DSM-5’s ‘intellectual disability (intellectual development disorder)’. However, as both include what amounts to significant impairment of intelligence, which is the only specified feature of ‘learning disability’ in Clause 1(3), these minor definitional differences should be of no consequence.

Social security cases

In a social security case in England and Wales, DE v Secretary of State for Work and Pensions (PIP) (Personal independence payment) [2021], it was held that a tribunal ‘should refer to the DSM-5 if considering a case involving Alcohol Use Disorder’. The Social Security Appeals Tribunal Northern Ireland does not regard reference to the ICD as essential. However, in (Unnamed) [2003], an unsuccessful appeal to the Social Security and Child Support Commissioner, it was said that it would have helped. The claimant had relied on a letter from a general practitioner and it was noted by the tribunal that, although it referred to the claimant being depressed, there was no ‘clinical diagnosis’ made by the general practitioner ‘in accordance with’ the ICD.

Police pension cases

It is common to see the ICD used in cases considered under the Police Pensions Regulations 1987 and the Police (Injury Benefit) Regulations 2006. This is because Home Office guidance on medical appeals under these regulations (now archived and being updated) refers to how ‘The police authority should require the SMP [selected medical practitioner] to describe wherever possible any disease or medical condition causing disablement by reference to internationally authoritative guides available to doctors such as ICD 10 […] and DSM IV’ (Home Office 2007). However, the words ‘wherever possible’ and ‘such as’ should not be overlooked.

Gender issue cases

In many cases where experts rely on, or refer to, both the DSM and ICD, the courts have been uncertain as to their relative status. This happened in JR111, Re Application for Judicial Review [2021], where the parties were unable to provide the court with much assistance as to which of the two taxonomies was in more prevalent use in clinical practice in the UK or when and how a psychiatrist or psychologist would use one rather than the other, although it was recognised that previous UK authorities in the particular field of gender issues seemed to place greater emphasis on the ICD classification.

The issue was whether, in order to secure a gender recognition certificate under the Gender Recognition Act 2004, a transgender person was still required to show, among other things, that they had ‘gender dysphoria’, defined by the Act as ‘the disorder variously referred to as gender dysphoria, gender identity disorder and transsexualism’ (para. 25). The claimant objected to this on both principled and practical grounds. It was submitted on her behalf that the requirement for a diagnosis of a gender identity disorder ‘irrationally requires transgender people to say that their understanding of their gender is caused by a mental disorder rather than a normal function of human variation’. It was also averred that the ‘pathologisation’ of transgender people was now out of line with international best practice, including by reference to ICD-11. This was because, although DSM-5 emphasises that incongruence is the ‘core component’ of a diagnosis of gender dysphoria, there remains an emphasis on clinically significant distress or impairment that does not find similar expression in the ICD’s description of gender incongruence. The applicant succeeded because the court found that, notwithstanding the uncertainty as to the relative status of the ICD and DSM, the requirement to provide a specific diagnosis which was defined as a ‘disorder’ in order to secure a gender recognition certificate failed to strike a fair balance between the interests of the applicant and those of the community generally.

Criminal cases

One of the most critical considerations of the ICD and DSM in any proceedings has been that of Lord Justice Hughes in R v Dowds [2012], where, quoting from the introductions to the ICD and DSM, including the cautionary section in the latter, he applied the concept of the ‘imperfect fit’ between questions of ultimate concern to the law and the information contained in a clinical diagnosis to the case in point.