‘A proposal for research in the epidemiology of psychiatric disorders’,by Alexander H Leighton (1950)

Abstract

The Classic Text is an outline of the Stirling County Study as conceptualized by Alexander H Leighton. It was first presented at a conference held in 1949 organized by the Milbank Memorial Fund, an American philanthropic foundation. The meeting brought together 30–40 experts from across North America. Leighton succinctly explained his frame of reference for the epidemiology of mental disorders and the methodology of the community-based study he conducted in Nova Scotia. The introduction to the text explains contextual points, certain specificities of Leighton’s framework, and the discussions that surrounded it, largely dominated by a group of Harvard professors, including Erich Lindemann and John E Gordon.

Introduction: Between observation and experimentation

The Classic Text ‘A proposal for research in the epidemiology of psychiatric disorders’ (Leighton, 1950) is an outline of the frame of reference for the Stirling County Study, published by psychiatrist and anthropologist Alexander H Leighton (1908–2007), during the preparatory phase of his study. To introduce it, I limit myself to a few contextual points, to certain specificities of Leighton’s epistemological choices, and to the discussions that surrounded it. The text is very short and is taken from the proceedings of a conference in 1949, organized by the Milbank Memorial Fund, ¹ an American philanthropic foundation. The importance of this conference has already been discussed by researchers from a variety of backgrounds interested in the history of psychiatric epidemiology, such as: Allan V Horwitz and Gerald N Grob (2011); Harry Yi-Jui Wu in a paper in this journal (Wu, 2015) and in his monograph (Wu, 2021); and the contributors to a book recently edited by Anne M Lovell and Gerald Oppenheimer (2022). However, Leighton’s outline was not reproduced in the 2005 Special Issue of the Milbank Memorial Fund’s journal, The Milbank Quarterly (Milbank Memorial Fund, 2005), which commemorated the centenary of the foundation by republishing landmark texts, notably on mental health, including a later text by Leighton and Jane M Hughes (1961).

In the early 1950s, Leighton succinctly outlined the methodology of the ‘community-based’ study he was then planning with his Cornell University team in rural Nova Scotia – in Canada (not Scotland, a mistake still made by authors today). The meeting brought together 30–40 experts from across North America. In his contribution, Leighton outlined his frame of reference for the epidemiology of mental disorders. As such it is valuable to analyse it from the point of view of the history of science, particularly concerning Leighton’s uses of observation and experimentation. Indeed, it is interesting to note that Leighton’s framework attempted to go beyond the level of observation that normally characterized epidemiology (see Giroux, 2011: 425) by integrating hypothetico-deductive reasoning and producing scientific data about the mental health of a population via a measurement tool in a controlled environment. This screening tool was a questionnaire, a type of assessment of mental disorders that could be used quantitatively and summarized by statistics. The questionnaire consisted of questions (stimuli) that elicited responses relating to the mental health of the community inhabitants under study.

However, this use of a quasi-experimental approach should be put into perspective. Firstly, hypothetico-deductive reasoning is a necessary but insufficient condition for an experiment. Secondly, in the design of the survey, the sociocultural characteristics were controlled, and not the setting in which the questionnaire was administered (which took place quite informally, with research staff who went ‘door to door’). In other words, Leighton did not conduct a study in experimental psychology, but a population-based study.

The purpose of this framework was to overcome and resolve the methodological biases of earlier studies, such as Robert EL Faris and HW Dunham’s Mental Disorders in Urban Areas (1939), based on hospital statistics that were not representative of the general population of the city of Chicago. Epidemiologists John E Gordon (Harvard) and Thomas Francis (University of Michigan) criticized this type of bias during the part of the Milbank Memorial Fund meeting entitled ‘Evaluation of this material’ (Milbank Memorial Fund, 1950: 73). Nevertheless, the merits of the Chicago study were recognized, for instance the identification of a geographical distribution of mental disorders. Moreover, Leighton’s research design was still conceived within an ‘ecological’ framework in the sense of the Chicago School (see Chapoulie, 2018). Interrelated variables, comparisons between populations, and the mapping of a sociocultural environment were all features that were part of the long history of sociology, while the questionnaire seemed to have been for Leighton and his contemporaries a kind of ‘experimental extension’ (Allamel-Raffin, Gangloff and Gingras, 2022: 201) ² to observation, without completely redefining psychiatric epidemiology as an experimental science.

Context of the Milbank Memorial Fund and American philanthropic foundations

The classic text ‘A proposal for research in the epidemiology of psychiatric disorders’ (Leighton, 1950) was Leighton’s contribution to the annual meeting of the Milbank Memorial Fund, held on 16–17 November 1949. Less well known than major foundations such as the Rockefeller, Ford, Carnegie and others, this organization, founded in 1905, nonetheless funded landmark research projects in health after World War II. Most of the studies were on mental health, but it also funded the infamous Tuskegee syphilis study, a scandal which the foundation recently attempted to shed light on with the help of historian Susan M. Reverby. ³ Among the researchers often designated as pioneers in the literature devoted to the early epidemiology of mental disorders, we should add at least two specialists, belonging to distinct scientific orientations and born before Leighton: Erich Lindemann (1900–74), a physician and psychologist of German origin, who migrated to the USA in the 1920s (Stahnisch, 2020); and Robert EL Faris (1907–98), sociologist and emblematic figure of the Chicago School, already mentioned above.

The volume published by the Milbank Memorial Fund (1950) features short individual texts by Robert H Felix, Ernest M Gruenberg, John H Dingle, Thomas Francis Jr, Robert P Knight, and Leighton, and reproduces the discussions, largely dominated by a group of Harvard professors: Lindemann, John E Gordon, Warren T Vaughan and Johannes Ipsen. It is notable that the book gives ample space to exchanges, reproducing long discussions alongside short articles, which is unusual and makes the volume original. As a result, the meetings organized by the Milbank Memorial fund can be said to represent an important landmark in the history of psychiatric epidemiology regarding the openness, intensity and dissemination of the debate, and not just in terms of the data presented. In addition to the Harvard representatives who led or intervened most often in the discussions, John Romano (University of Rochester) and Leighton – two academics from New York State – were the most frequent speakers, alongside Robert Felix, who was the first Director of the National Institute of Mental Health (NIMH) and chairman of the debates.

The full list of 38 participants includes many more names of experts, ⁴ whose voices are silent or less audible in the proceedings, but who took part in the exchanges. Interestingly, this list gives an idea of the American academic institutions that dominated expertise in mental health epidemiology in the post-war era, such as Johns Hopkins, Yale, Harvard, and especially the universities and institutions of New York State (Cornell, Columbia, New York Academy of Medicine, New York University College of Medicine, New York State Department of Health, University of Rochester), and the federal capital (US Public Health Service, US Department of State, National Health Council). The only foreign professor present was a Briton, John R Rees, invited as President of the World Federation of Mental Health; Donald Ewen Cameron was the only professor from a Canadian university (McGill), but he was an American citizen. Experts from the East Coast of the USA predominated, but there were few from the rest of the American continent.

A bibliography of 362 references completed the volume published in 1950, revealing a desire to set a milestone at a time when psychiatric epidemiology was still in its infancy. This literature was largely international, including German, French, Dutch, Italian and Spanish publications. Nevertheless, it still reflected traditional data regarding asylums and armies, rather than a clear new step towards innovative epidemiological research.

In short, we must be careful to avoid believing that this conference represented all of the emerging scientific communities in psychiatric epidemiology. Nor should we assume that all the experts in psychiatric epidemiology supported by the Milbank Memorial Fund in the post-war era were present. For instance, Brian Murphy (HBM Murphy), a Scottish physician, received funding from the Milbank Memorial Fund in the 1950s, published in its journal (Murphy, 1961) and established himself as one of Canada’s leading epidemiologists of mental disorders. However, it is important to note that, at that time, he had not yet been recruited by McGill (Murphy moved to Montreal in the late 1950s). His interest in the statistical and population-based approach to mental health was then linked to his work with Holocaust survivors from all over Europe, brought together in displaced persons (DPs) camps despite their different sociocultural backgrounds. Murphy was able to study them as part of the United Nations Relief and Rehabilitation Administration (UNRRA) and International Refugee Organization (IRO, which succeeded UNRRA in 1948) administrations, and then when the DPs were living in their new country after their resettlement (Murphy, 1956). In fact, in 1950, Murphy was still training in public health at the London School of Hygiene and Tropical Medicine, before taking up a position in Singapore (still under British rule), where he would study sociocultural factors in mental health. I mention him here to emphasize that his training in London, certainly steeped in British colonial medicine, was quite different from that of the American psychiatrists, epidemiologists, sociologists and anthropologists gathered at this conference. The example is not insignificant, as my hypothesis is that Cameron was instrumental in recruiting Murphy to his university in Montreal, as Eric Wittkower’s collaborator in McGill’s Division of Social and Transcultural Psychiatry; thus he would be in competition with the ambitious Leighton, who like Cameron had been a student of Adolf Meyer.

This heterogeneity should keep us from the misconception of a uniform reading of early psychiatric epidemiology. Leighton was teaching applied anthropology (not psychiatry) at the time and had not previously undertaken any work in the field of epidemiology. It is therefore preferable to present him as a sort of adventurer or entrepreneur in psychiatric epidemiology in 1949, and not as a leader. The time was still ripe for tinkering and interdisciplinarity, and discussions at the Milbank Memorial Fund show that there was no agreement among stakeholders as to whether anthropologists or sociologists were best equipped to participate in mental health epidemiological studies. It is also for this historical reason that I have chosen a scientific outline as a Classic Text.

‘The study of disease in nature is combined with the experimental method’

The two main discussions presented in the Milbank Memorial Fund volume are entitled ‘Minor disorders’ (there is only one presentation on ‘Major disorders’) and ‘An epidemiologic analysis of suicide’. In the first, Lindemann began by recalling the definition of health promoted by the World Health Organization (WHO, founded in 1948), which included mental health: ‘a state of complete physical, mental and social well-being’. Then, Gordon, Head of the Department of Epidemiology at the Harvard School of Public Health, took the floor to remind psychiatrists, who constituted the majority of the participants, that epidemiology was a study of population health, and that its methods were not reducible to medical statistics:

I choose to distinguish this activity as medical statistics in contrast to epidemiology. Epidemiology deals primarily with the study of whole populations, the sick and the well, by establishing categories of persons and separating environmental factors to permit comparison and quantification of fractions of those populations with and without the particular disease. (Milbank Memorial Fund, 1950: 13)

Next there was an original and unusual move: in a succinct presentation on the methods of epidemiology, he placed it between the sciences of observation (like astronomy) and experimentation (laboratory style of scientific thinking ⁵ since Claude Bernard in medicine), and not just as a study of diseases based on observation and comparison, and ancient perspectives inherited from the Hippocratic corpus (see Grmek, 1983). He could not have been more explicit:

There is the third alternative, where comparative study, the study of disease in nature is combined with the experimental method. That method of approach is as typical of the study of group phenomena among medical problems as it is of disease in the individual person. The epidemiologic method by reason of the uses to which it is put and the circumstances under which it is applied, is heavily weighted with the comparative study of disease in nature; it also makes use, under modern conditions, of the experimental approach. (Milbank Memorial Fund, 1950: 13)

Finally, Gordon sent the ball back to Lindemann and to his definition of epidemiology as ‘medical ecology’, which echoed both a well-known paradigm in sociology (Chicago urban ecology; see Chapoulie, 2018) and possibly, but in a very different way, a research programme in biology, notably at the WHO (International Biological Program) at the same time (see Collins and Weiner, 1977; Kwa, 1987; Schleper, 2017). Gordon went on to explain the methods – variables, indicators and significant relationships (correlation coefficients) – whose calculation enables us to return to the theoretical level and question the factors behind mental disorders. Again, distinguishing several approaches, he returned to experimentation, which he conceived in two modalities, either in the laboratory or in the form of fieldwork:

The third general feature of the epidemiologic approach is to test, by controlled experiment, the theories derived from analysis of these relationships. The controlled experiment is sometimes by laboratory methods and sometimes by field procedures. Each has its individual usefulness. (Milbank Memorial Fund, 1950: 14)

It is not possible to summarize the rest of the presentations and discussions in this short introduction (the relationship between epidemiology and the social sciences, ⁶ the types of psychopathological behaviour that could be the subject of epidemiological studies, etc.), but I would like to quote a remark by Lindemann, at once ironic and humbling, at the end of the first discussion. It sounds like an appeal to Leighton, who had already developed the frame of reference for his study in Nova Scotia:

Looking back over this material, we feel that it is far from being integrated, that it is a mass of observations and that they are colored by our diverse special interests. There is still an emphasis on the clinical approach. We haven’t quite mastered the epidemiological approach. We are quite amateurish in the area of the social sciences, but it may well be that in our further work in this particular field we will take along from this meeting a sense of new integration and a new awareness of possible modes of joint operation in this difficult field. (p. 34)

The integration of social science methods into the design of epidemiological studies is indeed what characterized Leighton’s frame of reference. Since 1948, he and his students had begun to investigate a rural area in Canada. The description of the sociocultural environment by a team of researchers experienced in the social sciences (‘a careful descriptive study of the main features of social organization and cultural patterning in the community’; Leighton, 1950: 132) would take place mainly in 1950–1. Leighton would be able to put his hypotheses to the test in the field in 1952.

Moreover, he pursued his academic career 20 years later at Harvard (he and Jane Murphy moved there in 1965–6), where Lindemann and Gordon held academic positions in 1949. Among the participants at the meeting, Faris represented Chicago’s approach in social sciences. As Faris and Dunham’s study was rooted in the 1930s, the historian is tempted to think that Leighton ensured a generational transition by making a significant contribution to the professionalization of psychiatric epidemiology as an academic discipline between the 1940s and 1960s.

But to return to the conference proceedings: the second major discussion was less innovative (fairly classic exchanges on the epidemiology of suicide) from the point of view of the history of science. Nevertheless, it is interesting to note that, at the end of their debates, the participants returned to the frame of reference presented by Leighton, whose scientific ambition and specific methodology had visibly surprised the speakers. In his response, Leighton emphasized both the uniqueness of his study’s frame of reference, bringing together social scientists and health professionals, and its experimental aspect:

That would be one frame of reference. To use this map analogy, we would plot on our map where the broken families in the community were; but we would also plot other factors, and then we would test whether those theories were correct where you get a cluster of these stress-producing factors that we have identified. (Milbank Memorial Fund, 1950: 174)

In addition to the map metaphor, typical of the techniques used in so-called ecological studies inherited from the Chicago School, Leighton introduced the idea of a controlled environment, with communities under greater stress than others being ‘plotted’ prior to the study.

‘A proposal’: outline of Leighton’s frame of reference

The Classic Text, ‘A proposal for research in the epidemiology of psychiatric disorders’, is composed of rather uneven parts, including some lists and descriptions of criteria (hypotheses, general concepts, values, community size), the layout of which was inconsistent in the original report (Leighton, 1950). This information was given in notes, or in the text, and sometimes in two or three places. To improve their readability, I have chosen to include these in the text.

At the outset, Leighton simply began by outlining his aims, premises, hypotheses and variables. There are no biomarkers in psychiatry as in other branches of medicine (‘In all other branches of medicine, pathology is paired with a corresponding division of physiology’; Leighton, 1950: 129–30), so the test technique (‘psychological test, sensitive to personality disorder’; p. 133) had a prominent place in his frame of reference, which borrowed from psychology, not just anthropology. From the outset, the stated aim was to study the impact of environmental factors on mental disorders.

The framework was designed to reveal the distribution of mental disorders in small communities. He also outlined the tools (‘existing technical tools’; p. 129) he envisaged to assess the symptoms of mental disorders in the chosen population, i.e. a test. This measurement tool was not developed by Leighton – contrary to what we often read in the secondary literature – but by psychologist Allister M Macmillan, a close collaborator of Leighton who joined his team in the late 1940s and died prematurely at the start of his academic career. The questionnaire was called the Health Opinion Survey (HOS; see Macmillan, 1957). A priori, the statistical calculations applied to the data collected by this screening tool were of the same type as those made in psychology and sociology at that time.

An essential aspect of the text is the way in which the author presented his study as community-based, with the main sociocultural variables under control, in small communities linked to each other in the same ecological matrix: ‘One step in the direction of reducing the number of variables and of increasing the opportunities for verification, is to limit all the personality studies, both pathological and normal to one community or a set of closely interrelated communities’ (Leighton, 1950: 130). Then, part of the proposal explicitly presents hypothetico-deductive reasoning:

If the community is one in which the cultural patterns and social organization (e.g., values, beliefs, customs, leadership and class structures, status positions, sources of psychological stress) have been ascertained by the techniques of the other social sciences, then the chances of uncovering significant leads concerning the relationship of social environment and personality discover will be greatly increased. It should be possible to evaluate the established theories in this field by seeing whether or not there is a higher frequency of mental ill-health where theory suggests it should be; and correspondingly, whether or not there is a lower incidence of ill-health where theory says it should be low. (p. 131)

The proposal was structured according to a theoretical framework (‘theoretical expectations’), which would become the subject of Leighton’s best-known book: My Name is Legion (Leighton, 1959). In addition to the books, in the following years he would publish articles on his frame of reference, sometimes more explicit and narrative than the 1949 proposal, revealing the features of his study as it progressed, but without changing the research design. For example, a long article in the American journal Psychiatry (Leighton, 1955) also dealt with his frame of reference. This article contained lengthy developments articulating dynamic psychiatry with Leighton’s more environmental orientation. However, this type of psychoanalytic or depth-psychological compromise quickly disappeared from his scientific articles, reports and books. These concessions were certainly due to the Freudian context of the time, in which he lost interest as his epidemiological studies progressed, while at the same time articulating public health and social science issues. The dynamic psychiatry of Meyer, with whom Leighton trained, also left its mark on his frame of reference, through, for example, the notions of disorganization, circumstance and reaction (‘circumstances that are important in the development and reactions of personality’ (Leighton, 1950: 130).

The list near the beginning of the text provided examples of the study’s basic variables and hypotheses. A comparison with the final variables ⁷ formulated in Leighton’s books shows that he refined his framework over the course of the 1950s.

The ‘Orientation’ section also indicated that the frame of reference was not focused on pathology, but on ‘stress factors’. If the framework was valid, the variables should have enabled epidemiologists to identify both risk factors for mental disorders and positive factors for well-being. He conceived the sociocultural environment to include both danger and protection. The reference to Faris and the Chicago School’s ecological framework did not prevent Leighton from imposing his mark, clearly indicating his preference for a small, rural sample rather than an urban sample.

The short section entitled ‘The research problem’ was nothing more than what he would call the basic hypothesis in his books, articles and reports. The ‘Outline of a research plan’ assigned as the first step the descriptive study of the main aspects of the social organization and cultural values of the study communities. This part of the survey took place mainly between 1950 and 1952, led by a team of anthropologists, including a photographer. The results were the subject of the team’s second book on the Stirling County Study: People of Cove and Woodlot (Hughes, Tremblay, Rapoport and Leighton, 1960).

Observation and/or experimentation: what framework for the Stirling County Study in the history of science?

In the history of medicine and health sciences, as I have already mentioned, epidemiology is presented as being observational (Giroux, 2011). A recent collective work (Allamel-Raffin et al., 2022) explores the varied forms of experimentation in different scientific fields. Several contributions to this book distinguish between observation and experimentation in disciplines as varied as physics, chemistry, biology, medicine, psychology and so on. What emerges is that there is no single form of observation, just as there is no single form of experimentation. Some disciplines use both observation and experimentation, depending on the topic, the level of scale, the level of depth of knowledge, whether at an exploratory stage or at a level of stabilized knowledge, among other considerations. There are several stages in the production of scientific facts.

In the case of the Stirling County Study, observation was clearly intended to test the validity of a scientific prediction, both as a premise and as a basic hypothesis: the rate of mental disorders varies according to socio-economic and cultural variables. The study had to establish relationships between two phenomena. This kind of statistical relationship was influenced by certain independent variables (see the list in my paper in this Special Issue; Delille, 2023). The system also had secondary exploratory dimensions, because the scientific team did not know in advance which variables would be most likely to affect mental disorders, and which could be the subject of in-depth investigations at a later date. Observation thus enabled certain properties to be better identified, and the basic hypothesis and certain secondary hypotheses to be confirmed or rejected, depending on the variables involved.

However, observation does not capture all of the ambitions of the Stirling County Study, as formulated in Leighton’s text, for there was clearly a hypothetico-deductive rationale in the design of this study, and a technique for the empirical identification of mental disorders. In fact, for Leighton at that time, the questionnaire technique used in the field to assess mental disorders in the target group consisted of an experimental framework, in which individuals from the target population responded to provoked stimuli (the questions); the administration of the questionnaire modified the observation framework and revealed new phenomena: the presence of mental disorders. Answers to the epidemiological study questionnaire produced behavioural changes – facts, not merely observations. Whatever we think of it today, this ambition to produce scientific facts must be taken from Leighton’s 1949 frame of reference.

Another aspect of the study must also be considered, to avoid over-simplification of the research design. As I have already mentioned, the characteristics of the sociocultural environment were controlled in this study, in a first stage that took place early in the study, and not in the individual questionnaire administration, which took place later. These assessments did not take place in an experimental psychology laboratory, but in situ, essentially in the population’s living environment (through house-to-house interviews). Control was also exercised through the choice of communities, as not all the villages and towns in the county were included in the epidemiological study; but in the end, only six were chosen for their prosperity or socio-economic disintegration. The choice of communities and the control of socio-economic characteristics were part of the research design (‘The areas in between will for the time being be dropped from considerations’; Leighton, 1950: 133).

To check the robustness of the results, another technique was used, called ‘the key informant technique’, which involved cross-referencing results with doctors, social workers, priests, teachers and other local notables (p. 133) in the small communities studied. This was a common technique at the time, directly derived from social anthropology and the population control techniques of colonial administrations (Lovell and Oppenheimer, 2022: 28). In a second phase, the team’s doctors conducted diagnostic interviews with part of the sample, after the epidemiological study, again for the sake of scientific rigour, to check that the results of the screening tool (HOS) were consistent with psychiatrists’ diagnoses.

Finally, can we really speak of scientific experimentation here, or is it simply an analogy to the experimental devices of laboratory science? In other words, did Leighton’s framework simply ‘pretend’ to be an experiment or, as mentioned at the start, did it include a kind of ‘experimental extension’ to observation? I would like to keep the question open for further studies on the history of psychiatric epidemiology. With regard to the scientific controversies surrounding the distinct contributions of observation in epidemiology, experimentation in randomized clinical trials (RCTs), and the more radically empirical approach of ‘evidence-based medicine’, philosopher of science Élodie Giroux reminds us that the term ‘quasi-experimentation’ has been used in relation to the methods of many human and social sciences (Giroux, 2011), in much the same way as Durkheim did in the early days of academic sociology (see Dominique Raynaud’s contribution in: Allamel-Raffin et al., 2022: 87–102). Yet it was with the help of the social sciences that Leighton developed his framework.

Conclusion

In Leighton’s frame of reference, as in others, it is important to articulate the successive scientific procedures to contextualize the facts produced by applying the questionnaire technique to epidemiology, which surpassed observation. There is indeed a theoretical construct (basic hypothesis, variables) behind this questionnaire (HOS), and then, after it has been administered, empirical results. One of Leighton’s aims was to demonstrate that he had mastered a kind of hypothetico-deductive approach to studying mental health pathologies, a challenge since the beginnings of psychiatry. He pursued this line of reasoning by mobilizing quantitative measurement techniques to operationalize what was observed and, indeed, to repeat the measurements to demonstrate reliability. In this way, the scientific team subjected its ideas to the experience of measured data. Leighton thus presented a frame of reference that included a kind of experimentation for theoretical reasons. The complexity of the research design was one of Leighton’s great scientific innovations compared with previous epidemiological studies in mental health.

I do not wish to draw a final conclusion on the placement of Leighton’s frame of reference, which I situate between observation and experimentation depending on the different stages in his epidemiological study. Rather, I would like to specify two areas that I intend to pursue in the future, from the point of view of the history of science. I think we need to examine two distinct issues: on the one hand, the relationship of psychiatric epidemiology research designs to observation and experimentation; and on the other, the history of questionnaires and other screening instruments (see Cassin, 2014) in science and health. These are indeed different aspects that merit historical attention beyond Leighton’s short proposal in 1950.