Psychiatric Classification: A ‘Hard’ Problem PDF

Psychiatric Classification: A ‘hard’ problem Richard Bentall Aims and learning objectives: Learning objectives: The understand why psychiatric classification is a hard problem, standard approaches to diagnosis, and new emerging alternatives. Specific aims: To understand the his...

Psychiatric Classification: A ‘hard’ problem Richard Bentall Aims and learning objectives: Learning objectives: The understand why psychiatric classification is a hard problem, standard approaches to diagnosis, and new emerging alternatives. Specific aims: To understand the history of diagnostic concepts To understand the categorical model of diagnosis and its problems To be briefly acquainted with four new approaches to classification (quantitative approaches, research domains criteria, symptom approaches and network models). Conventional approaches to psychosis At both the clinical and the population level, this approach has had almost no impact on mental health: Long term outcomes for patients with psychosis have not improved since the end of WW2 (Jääskeläinen et al. 2013) Developing countries with poorly resourced I will arguehealth mental that we need to services think have of psychosis better outcomes(and mental illness more (Sartorious generally) as a public health et al. 1997) challenge. The role of psychiatric diagnoses Psychiatric diagnoses have many functions: Communication between clinicians Determining access to services Public health surveillance Selection of participants for research into mechanisms and aetiology Prediction of clinical outcome Prediction of treatment response For more than a century psychiatry and related disciplines have employed a categorical diagnostic system which has its origins in the work of Kraepelin and his contemporaries.  I A brief history of the categorical system Psychiatry in Germany The term ‘psychiatry’ (from the Greek psych = soul and iatros = doctor) was coined by Johann Christian Reil (1759-1813). Teaching in psychiatry first began in In Leipzig in 1811.Griesinger 1865, Wilhelm established the first modern academic psychiatry department in Berlin. Two years later, he founded the Archives for Psychiatry and Nervous Disease. In his opening editorial, he said: “Psychiatry has undergone a transformation in its relation to the rest of medicine. This transformation rests principally on the realization Griesinger that patients with so-called ‘mental Emil Kraepelin (1856- 1926) Wrote Compendium of Psychiatry, 1883. Used wide ranging research methods – inc clinical observation, psychological tests, psychopharmacology, cross- cultural studies Saw course of illness (after Kahlbam) as clue to classification. Kraepelin Collected detailed case studies) Revised his ideas in successive editions of the Compendium – eventually a Textbook of Psychiatry Emil Kraepelin (1856- 1926) “Judging from our experience in internal medicine it is a fair assumption that similar disease processes will produce identical symptom pictures, identical pathological anatomy and identical aetiology. If, therefore, we possessed a comprehensive knowledge of any one of these three fields - pathological anatomy, Kraepelin symptomatology, or aetiology - we would at once have a uniform and standard classification of mental diseases.” Emil Kraepelin (1856- 1926) Kraepelin believed that diagnosis by symptoms would be a Rosseta stone that would lead to an understanding of aetiology: Emil Kraepelin (1856- 1926) First described schizophrenia (which he called ‘dementia praecox’, because it occurred in the young and had a poor outcome) and differentiated it from manic-depression (good outcome). Later saw paranoia as a separate illness. Assumed that schizophrenia was a Kraepelin product gross deficits in cognitive processes (attention, memory, etc.) caused by underlying neuropathology. Consensus view of classification circa 1970s Subtypes: paranoid, hebephrenic, catatonic, simple Schizophrenia Bipolar disorder Psychoses Affective psychoses Psychiatric disorders Unipolar depression Paranoia/delusional disorder Generalized anxiety Neuroses Panic disorder Obsessive-compulsive disorder Personality disorders (psychopathy)? Minor depression Two systems of classification The American Psychiatric Association in 1948 formed a task force to create a new standardized system. The Diagnostic and Statistical Manual of Mental Disorders was published in 1952. Currently in its 5th edition (2013) In 1948, World Health Organization assumed responsibility for an International List of Causes of Death, first compiled in 1853 and revised 4x by the French government. When creating ICD-6 in 1951, WHO added nonfatal diseases, including psychiatric disorders. TheseCurrently are categorical systems in its 11 th (they edition divide people (2018) into categories)  II Problems of diagnosis and the neoKraepelinians Early concerns: US-UK diagnostic project Venn diagram showing relationships between US and UK diagnostic concepts as revealed in the US-UK Project (from Kendell, R. E., Cooper, J. E., Gourlay, A. J., Copeland, J. R. M., Sharpe, L., & Gurland, B. J. (1971). Diagnostic criteria of American and British psychiatrists. Archives of General Psychiatry, 25, 123-130.) Early concerns: The problem of reliability Reliability refers to the consistency of diagnosis; validity to its usefulness (scientific value). Diagnoses can be reliable without being valid, but not valid without being reliable. Spitzer & Fliess (1974) introduced the kappa statistic (varying between 0 and 1) to correct for the base-rate problem: k= Po - Pc 1-Pc where Po is the proportion of observed agreement between clinicians and Pc is the level of agreement expected by chance. Early concerns: Spitzer & Fliess’ (1974) review of reliability The studies from which the data were derived were I Schmidt and Fonda (1956); II Krietman (1961); III Beck et al. (1962); IV Sandifer et al. (1964); V Cooper et al. (1972); VI Spitzer et al. (1974). The data from V (the US-UK Diagnostic Study) are analysed separately for the New York and London samples. The neoKraepelinian movement Partly because of these problems, in the US in the 1970s, a number of psychiatrists tried to return psychiatry to its Kraepelinian roots. In her book, The broken brain (1990), Nancy Andreasen predicted that, in the future, psychiatric interviews would be 15 minutes long, and that psychotherapy would only have a marginal role in the treatment of mental illness. Klerman’s (1978) neoKraepelinian manifesto 1. 4. Psychiatry There is a branch of medicine. is a boundary 2. Psychiatry should use modern scientific between methodologies andthe normal base its practice onand the sick. scientific knowledge. 5. There are discrete mental 3. Psychiatry treats people who are sick and who require treatment for mental illness. illnesses. There is not one, but 4. There is a boundary between the normal and the many mental illnesses. sick. 5. There are discrete mental illnesses. There is not one, 6. The focus of psychiatric but many mental illnesses. physicians should be particularly 6. The focus of psychiatric physicians should be particularly on the biological aspects of mental illness. 7. on the There biological should be an explicit aspects and intentionalof concern mental illness. with diagnosis and classification. 8. Diagnostic criteria should be codified, and a legitimate and valued are of research should be to validate such criteria by various techniques. 9. In research efforts directed at improving the reliability and validity of diagnosis and classification, statistical techniques should be utilized. DSM-III Arguably the neoKraepelinian’s greatest achievement was the introduction of a new diagnostic manual, the third edition of the DSM (DSM-III) in 1980. Committee headed by Robert Spitzer Diagnoses based on consensus agreement Political considerations affected some of the diagnoses (homosexuality – eventually excluded – and PTSD – included) Diagnoses designed using ‘Chinese menu’ system of operational criteria with the aim of solving the reliability problem. DSM-III criteria for schizophrenia Revisions of DSM-III DSM-III has since been revised three times, each time generating a lot of income for the American Psychiatric Association. DSM-III DSM-IIIR DSM-IV DSM-5 Year of 1980 1987 1994 2013 publication Pages 500 570 883 947 Words 324k 436k Diagnoses 163 174 201 163 with criteria Criteria 962 1392 1500 1452 Price $31 $38 $49 $210 Revenue $9.8 $16 million $20 million $120 million million predicted Has the reliability problem been solved?: The DSM-5 field trials Results of the DSM-5 field trials, reprinted from Freedman et al. American Journal of Psychiatry, 170, 1-5, 2013. Note that, the convention of describing kappa values above 0.7 as satisfactory has been replaced by the assumption that any The problem of ‘comorbidity’ Psychiatric diagnoses should be jointly exhaustive and mutually exclusive. Soon after DSM-III was published, it was noticed that the exclusion criteria in the definitions led to underestimation of the ‘comorbidity’ between symptoms. Robbins et al. (1981) suspended these rules on data from the Epidemiological Catchment Area Study: Given schizophrenia, the odds ratio for mania was 46 Given schizophrenia, the odds ratio for depression was 14. Amazingly, they concluded: “The most likely explanation for co-occurrence is that having one disorder puts the affected person at risk of developing other disorders” There have been many comorbidity studies since – it remains a persisting problem with conventional psychiatric diagnoses. Clinical utility There is also very little evidence that categorical diagnoses have clinical utility. For example, there has been very little research on specificity of treatment response but: Antidepressant drugs are recommended for a wide range of affective spectrum disorders (Hudson & Pope, 1990). Antipsychotics have been indicated for both schizophrenia and bipolar disorder (Tamminga & Davis, 2007) and, as also seems to be the case for lithium, response appears to be predicted by severity of symptoms rather than (Johnstone et al. 1988).  III Project(s) for a new approach to classification Four new approaches There is now widespread agreement that categorical diagnoses are not fit for most scientific and clinical purposes, but there is no agreement about alternatives. Four approaches are current: Research Domain Criteria (NIMH) Symptom-orientated research Network models Quantitative classification approach Quantitative Classification Quantitative classification The aim is to develop a scientifically valid method of classification by using advanced statistical techniques (typically forms of confirmatory factor analysis) to identify syndromes (clusters of symptoms) and test models that explain the covariation between symptoms. Assumes a latent variable approach: The five dimensions model of psychosis Factor analytic studies of symptoms suggest that all psychotic disorders can be explained by five independent dimensions: positive symptoms, negative symptoms, cognitive disorganization, depression and mania. From van Os & Kapur (2009) Diagnostic spectra Derived using factor analyses of diagnoses (i.e. based on examining patterns of comorbidity). Krueger (1999) conducted a meta-analysis of adult comorbidity studies and applied confirmatory factor analysis to pooled data on 11 common mental disorders. Results suggested internalizing (mood disorder) and externalizing (distress expressed behaviourally) spectra. Supported by analysis of World Mental Health Surveys of non- psychotic diagnoses in > 21,000 people in 14 countries (Kessler et al. 2011) Kotov et a. (2011) carried out the first comparable study including psychotic diagnoses – a third spectrum. Since found in other studies (e.g. Wright et al 2013). Hence, three spectra: internalizing (mood), externalizing (behaviour) and psychosis Hierarchical models Alternatives to quantitative classification he symptom approach An alternative approach has been to carry out research focusing on specific symptoms (Bentall, 2003): Hallucinations Delusions Thought disorder (disorganized speech) Negative symptoms Manic symptoms Once these symptoms have been explained, perhaps there will be no ‘disorders’ left to explain? Symptom approach The idea is to take each individual symptom - for example, hallucinations, delusions and thought disorder - and explain it in turn. Clinical implications: Clinicians simply write down a list of symptoms. Caveat: CBT with patients should be problem-based. The first task of the therapist is to get the patient to describe the problem, which may not be symptoms. esearch Domain Criteria Proposed by the US National Institute of Mental Health (Insel et al. 2010). The idea is to research transdiagnostic processes linked to psychopathology. esearch Domain Criteria Proposed by the US National Institute of Mental Health (Insel et al. 2010). The idea is to research transdiagnostic processes linked to psychopathology. Clinical implications: RDoC has important implications for research but, as yet, there are few clinical implications. In the future, it may be possible to identify new targets for intervention. Network models It has recently been pointed out that syndromes (clusters of symptoms) may occur, not because of common underlying disease processes, but because one symptom are causally connected to others in networks (Borsboom & Cramer, 2013; McNally, 2016) The network approach to comorbidity for two fictitious persons, Alice (left) and Bob (right) (from Boorsbom et al. 2011). etwork model of psychosis A network model of 79 symptoms measured in 408 male psychotic patients from van Rooijen, G., et al. (2017). Positive associations are colored green and negative associations are colored red. The thickness of the connecting ‘edges’ shows the strength of the association. Note anhedonia/asociality closer to depression than positive symptoms; depression also feeds into psychosis via delusions. etwork model of psychosis Network models can be used to examine transdiagnostic issues. For example, given evidence that trauma is highly associated with psychosis (Varese et al. 2012) and that many patients meet the diagnostic criteria for both schizophrenia and PTSD, Hardy et al. (2021) used data from 216 schizophrenia + PTSD patients to examine the relationships between PTSD and positive symptoms (largely connected by negative beliefs about the self and the world). Network models There are some important limitations of network models: 1. There is (as yet) no statistical test to establish whether a network or a latent variable model best fits the data. 2. There latent structure implies causal relationships between symptoms but, in most cases, what we really have is correlations. There are two potential solutions: Directed acyclic graphs sometimes enable causal factors to be identified in cross-sectional data. Network models There are some important limitations of network models: There are as yet no clinical implications of network models. Initially network models hoped that by identifying which symptoms were most central, they would be able to find the most effective targets for intervention. However, that promise has so far not been fulfilled. Fitting the new approaches together There is no contradiction between the four new approaches to classification. Classificationi Syndromes sts should be (Quantitati Processes Symptoms promiscuous Networks ve (RDoC) realists Classificati on) But which is the right level of explanation? Philosopher John Dupré argues that classificationists should not expect nature to yield a one-size fits all classification that works for all purposes. But, for the purposes of research into John Dupré aetiology and mechanisms, symptoms  Summary 1. For much of the history of psychiatry, categorical diagnoses have been used to guide research and treatment. 2. Although still widely represented in psychiatry and clinical psychology textbooks, these diagnoses have limited scientific validity or clinical utility. 3. Vigorous efforts are under way to develop alternative classification systems, but there is as yet no consensus about the best approach. 4. Different systems may suit different purposes.