Psychiatric epidemiology: dimensions and categories

In their article ‘Social disorder and diagnostic order: the US Mental Hygiene Movement, the Midtown Manhattan study and the development of psychiatric epidemiology in the 20th century,’ March and Oppenheimer describe a stark division between early and recent investigations. 1 In their view, both of the early studies (Stirling County and Midtown Manhattan) employed a non-diagnostic unidimensional approach to assessing mental health. 2,3 In contrast, the recent studies (the Epidemiological Catchment Area Program and the National Comorbidity Surveys) assumed discrete diagnostic categories of psychiatric disorders. 4–6 Being closely associated with the Stirling County Study and having considerable knowledge of the Midtown Manhattan Study, we give a different view of this history. We emphasize important distinctions between the Stirling and Midtown studies. In addition, we highlight the fact that tensions between dimensions and categories relevant to psychiatric phenomena remain strong. The Stirling County Study is an ongoing investigation of psychiatric epidemiology, instigated in 1948 by Alexander Leighton. A few years later, the Midtown Manhattan Study was started by Thomas A. C. Rennie who died in 1956. At that time, Leighton assumed some administrative responsibility for Midtown but did not try to change its psychiatric approach. This approach was markedly different from that of Stirling and already reflected the fact that Rennie had developed a strong interest in psychoanalysis and had given his support for a unidimensional assessment of mental health. Leo Srole took charge of carrying Midtown forward. Although different in psychiatric orientation, the two studies shared the characteristics of: being interdisciplinary; relying on survey methodology for sampling procedures; and using questions from existing psychiatric interviews. The only structured psychiatric interviews available at the time were those that had been prepared for use in the military or in psychological clinics. The resultant schedules for Stirling and Midtown differed considerably, but both reflected the symptom-based orientation of the parent interviews. Also, in both studies, the plans for how the responses should be assessed were worked out after the data had been collected. The 1950s work of Stirling was described in thee books; the subtitle of each was The Stirling County Study of Psychiatric Disorder and Sociocultural Environment.2,7,8 In the first volume, Leighton devoted many pages to illustrating the different categories of psychiatric disorders and indicating that all of them needed to be assessed in a study such as this book had introduced. 2 He avoided using the word ‘diagnosis’ because at the time it usually implied that aetiology was known. Believing that aetiology should be investigated rather than assumed, he used the term ‘patterns’ to describe the different categories of psychiatric disorders. In the third volume, which reported the study's psychiatric findings, the different patterns of psychiatric disorders were presented in terms of prevalence rates. 8 Depression was described as a disorder involving ‘prevailing unhappiness’, ‘sadness’ or ‘blue mood’ and its prevalence was 7%. Schizophrenia was described as involving a ‘twist or defect in the processes of perception and thinking’ and its prevalence was 1%. 2,8 Throughout the Stirling volumes, the language of the first American Diagnostic and Statistical Manual (DSM–I) was used. 9 There were, however, repeated references to the inadequacies of the manual for psychiatric epidemiological research, such as assuming aetiology, lack of symptom specification and absence of durational requirements. Illustrative of the problems posed by DSM–I is its definition of depression which incorporates a presumed causal mechanism: ‘The anxiety in this reaction is allayed, and hence partially relieved, by depression and self-depreciation. The reaction is precipitated by a current situation, frequently by some loss sustained by the patient, and is often associated with a feeling of guilt for past failures or deeds’. 9 Many of the inadequacies of DSM–I were resolved in DSM–III, as explained in an article introducing the longitudinal work of the Stirling County Study. 10,11 In contrast to the emphasis on psychiatric disorders in Stirling, Midtown focused on mental health as conveyed in the titles of its two volumes: Mental Health in the Metropolis and Life Stress and Mental Health.3,12 The approach utilized in Midtown involved a six-category gradient representing a symptom formation classification of mental health which ‘places the stricken people along a single heuristic dimension according to severity of their symptoms and the disability they entail’. 3 Srole indicated that this approach was influenced not only by Rennie’s interest in psychoanalysis but also by Menninger's emphasis on a unitary concept of mental illness 13 and by the review of psychiatric literature by Felix and Bowers which suggested a trend toward viewing mental illness as differing more in degree than in kind. 14 Whereas the Stirling Study focused on the discrete categories or patterns of psychiatric disorders, it also presented findings in terms of two dimensional measures. One was called ‘Caseness’. A central feature of the procedures referred to the ‘personal confidence’ which the psychiatric evaluators held that the survey methods indicated a ‘case’ of mental illness. ‘Caseness’ was rated as ‘definite’, ‘probable’, ‘possible’ and ‘doubtful’. The other continuous measure was titled ‘Typology of need for psychiatric attention’. It had five categories ranging from the ‘most abnormal’ to ‘psychiatrically well’. This approach anticipated decades of struggles with the problem of diagnostic uncertainty and of the correspondence between diagnosis and need for treatment, issues that remain unsolved. 15 Aside from relatively small studies that use clinically intensive interviews, diagnostic certainty/uncertainty has rarely been measured. 16 Finally, we note recent empirical support for dimensional approaches to psychiatric nosology that seem somewhat consistent with the Midtown rejection of diagnostic categories. This support derives from evidence about comorbidity and from factor analytical studies. 17,18 The Stirling Study and recent epidemiological investigations have kept track of each discrete category represented in a given interview. They have consistently reported that it is almost as common for a person to report two or more diagnoses as to report a single diagnosis. The extent of comorbidity between disorders has, through latent variable approaches, given further weight to the idea of dimensions of psychopathology. However, the number of underlying dimensions that would be needed to explain the universe of psychiatric illness is a matter of active research. 19 To our knowledge, however, none of the current endeavours assumes that mental health-mental illness is a unidimensional phenomenon. We are appreciative towards March and Oppenheimer for spurring our review of psychiatric epidemiology. Their focus on dimensions vs categories provided us the stimulus to suggest that, regarding those concepts, there have been both continuities and discontinuities that stretch over the past 65 years. A remarkable feature of these years has been the steady growth of the field. Despite the methodological differences noted here, the findings for North America consistently indicate that about 20% or slightly more of the general population suffers from a psychiatric disorder. 3,4,5,6,8,20,21 Many of these studies also indicate that prevalence is higher among those living at the low end of the socioeconomic hierarchy. Further, because the aetiology of most psychiatric disorders remains unknown, empirical evidence from different perspectives, both categorical and dimensional, is extremely important. This work was supported by grant 126099 from the Canadian Institutes of Health Research.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.