The FIND survey study population

We selected a random sample of 3000 laypeople, 1500 physicians, 1500 nurses and all 200 members of the parliament (MPs) of Finland. We identified laypeople aged between 18 and 75 years from the Finnish Population Register Centre, and physicians and nurses aged less than 65 years from the registries of Finnish Medical Association and Finnish Nurses Association. A previous publication presents more detailed study procedures, including randomisation of surveys (three different survey versions), assessing comprehension and exclusion criteria.13 The reporting of the study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology statement.14

Survey

We selected 60 states of being that we considered to be familiar to the relevant stakeholders based on the existing literature, the International Classification of Diseases (ICD-10), and iterative discussion and consensus-building processes.13 We anticipated that some of these states would be considered as a disease by (almost) all, some by (almost) no one and that some states would elicit disagreement. We randomised the 60 states into three blocks (1, 2 and 3; each containing 20 states). We created three versions of the questionnaire: version A consisted of blocks in the order 1-2-3 (online supplementary material, pages 9–14), version B in the order 3-1-2 and version C in the order 2-3-1. Within each sample group, we randomised respondents to one of the three versions (figure 1). We mailed the questionnaires in June 2010 (first response round) and sent reminders in August and October 2010 (second and third response round; questionnaires were resent to those who had not previously responded). We made precontacts with MPs by email and telephone. We asked participants to respond to a claim using a five-point Likert scale ranging from strongly disagree to strongly agree: ‘(This state of being) is a disease’.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Study flowchart.

For the present study (FIND psychiatry), we selected all 20 states related to mental health. Participants represented five respondent groups: psychiatrists, non-psychiatrist physicians, nurses, MPs and laypeople. The psychiatrists included those physicians who self-reported as being consultants or residents of adult psychiatry, paediatric psychiatry, adolescent psychiatry or forensic psychiatry. The non-psychiatrist (other) physician group included all other medical specialties (online supplementary material, pages 5–6). We excluded participants who did not demonstrate comprehension of the questionnaire (n=23 of eligible sample, 0.4%)13 or who failed to respond to four or more mental health-related states of being (n=21 of eligible sample, 0.4%) (figure 1).

Statistical analysis

We summarised results separately for each of the five groups (psychiatrists, non-psychiatrist physicians, nurses, MPs and laypeople). We calculated which states of being were considered to be a disease (Likert scale responses 4 and 5) by ≥75% or ≥50% of all respondent groups, and which states of being were not considered as a disease (Likert scale responses 1 and 2) by ≥75% or ≥50% of all respondent groups. The states of being not included in the previously mentioned categories were labelled as those with a large degree of variation.

We created a composite score using individuals’ perceptions of each state of being. Responses ranged from ‘1’ (strongly disagree that the state is a disease) to ‘5’ (strongly agree that the state is a disease). As there were 20 states of being, the composite score had a potential range from 20 (participants reporting strongly disagree to all) to 100 (participants reporting strongly agree to all). For participants with missing responses between one to three states of being (n=219, 6.7%), data were imputed using the participants’ average score of the completed responses. We performed sensitivity analysis by imputing ‘neither agree nor disagree’ to missing variables but results were similar and we do not report them here.

We constructed multivariable linear regression models using the composite score as the dependent variable and the following independent variables: (1) respondent group (ie, psychiatrists, non-psychiatric physicians, nurses, MPs and laypeople); (2) gender and (3) age. Because almost all nurses were female, gender, age and respondent group could not be included in a single model. Therefore, we constructed two models both of which included age (reflecting 10-year increments). The first model also included the respondent group and the second model included gender. All analyses were performed using STATA V.13. The estimated coefficients were reported with 95% CIs and associated p values and R-squared values.

Patient involvement

We did not involve patients when designing the study. We pilot tested the questionnaire with 20 laypeople and 5 physicians, and made minor revisions on the basis of the feedback. Laypeople (as well as all physicians and nurses) were sampled at random, and we did not collect information regarding a history of psychiatric illness.

Statement of principal findings

We identified mental health-related states of being that most respondents across all groups considered to be diseases (schizophrenia and autism), states that a majority considered to be diseases (depression, anorexia, panic disorder, generalised anxiety disorder, bulimia, attention deficit hyperactivity disorder and personality disorder), states that most respondents across all groups considered not to be diseases (grief and homosexuality), states that a majority of respondents considered not to be diseases (absence of sexual desire, premature ejaculation and transsexualism) and states in which there was wide divergence in views (alcoholism, work exhaustion, insomnia, drug addiction, gambling addiction and social anxiety disorder) (box 1 and figure 2). Perceptions differed by group: the inclination to medicalise states of being was highest among psychiatrists and other medical professionals, and lowest among MPs and other laypeople, suggesting considerable divergence between professional and lay conceptions of disease. The magnitude of the difference between psychiatrists and laypeople was substantial: an average difference of more than 10 points on an 80-point scale of inclination to consider a state as a disease. The largest differences between the two groups were seen for social and generalised anxiety disorders and gambling and drug addictions, with psychiatrists being far more likely to define these states as diseases than laypeople. Finally, we found that that younger people and women were modestly more inclined to consider states to be diseases than older people and men were.

Strengths and limitations of the study

The strengths of our study include a large sample of physicians, nurses, laypeople and legislators, an acceptable response rate with a high level of completion among those who responded, and inclusion of a substantial number—20—of states potentially related to mental health. In addition, the sample proved similar in its characteristics to the target population in terms of age and gender distribution, education, employment and marital status.13 Moreover, we found no significant differences in perceptions or participant characteristics by response round, reducing concern regarding selection bias.13

One limitation of our study is that it included only Finnish participants, and therefore, perceptions may differ in less affluent countries and those with different social and cultural values. Western, democratic and high-income states are, however, likely to have a similar distribution of perceptions.15 Second, although we addressed the potential impact of wording in a pilot study, individuals may react differently to a particular question depending on how questions were worded (‘framing effect’).13 16–18 Third, the response rate was substantially lower among legislators, raising issues of possible unrepresentativeness of those who did complete the survey. Finally, as this analysis is a part of larger scale FIND survey assessing not only mental health-related states of being but states from various other specialties, we did not survey whether these states were considered as ‘mental health-related disorders’ or ‘psychiatric conditions’ but as ‘diseases’. Furthermore, we did not define word disease in the survey but relied on each respondent’s own understandings of this term. These aspects can also be seen as limitations of our study.

Comparison with other studies

We used specific search criteria to search the Embase, Medline and PsycINFO databases (until 07 January 2019) to identify studies that assessed the concept of disease (online supplementary material, pages 7–8). Our search found as many as 14 931 articles. Although we found many studies that examined (1) the public’s and/or professionals’ attitudes toward people with mental illness, (2) the public’s and/or professionals’ attitudes about a single condition or (3) one stakeholder groups’ attitudes toward a range of condition, we found only one study that assessed the concept of disease over a wide range of conditions across several stakeholder groups. This survey, conducted in the 1970s, examined 38 conditions among non-medical academics, medical academics, general practitioners and secondary school students, and included only three mental health-related conditions (alcoholism, depression and schizophrenia), all of which were included in our survey.19

In the prior study, 53% of secondary school students, 67% of non-medical academics, 61% of medical academics and 86% of general practitioners considered alcoholism to be a disease. These numbers are similar to our study, in which 48% of laypeople, 64% of non-psychiatric physicians and 74% of psychiatrists considered alcoholism to be a disease. Although almost all people considered schizophrenia to be a disease in our study (eg, 90% of laypeople and 99% of both non-psychiatric physicians and psychiatrists), in the previous study, only 50% of secondary school students, 62% of non-medical academics, 77% of medical academics and 91% of general practitioners considered schizophrenia to be a disease. Differences in perceptions of depression proved even more striking: only 22% of secondary school students, 42% of non-medical academics, 50% of medical academics and 67% of general practitioners considered depression to be a disease in the earlier study, while in our research 75% of laypeople, 94% of non-psychiatric physicians and 97% of psychiatrists considered depression to be a disease.

Although these two studies were performed in different countries (the previous study in Canada and the UK, and our study in Finland), the most plausible explanation is that nowadays more people consider these mental-health related states of being as diseases than in the 1970s, and this is the case for both laypeople and healthcare providers. This pattern is consistent with the evidence of the widening definition of disease both within and beyond psychiatry, including the growing number of diagnoses in subsequent editions of the DSM and ICD, and the increasing number of people being diagnosed and treated for mild and subclinical psychiatric disorders.4 20–23

Meaning of the study: possible explanations and implications for clinicians and policymakers, and future research

Table 2 categorises states of being potentially related to mental health (addiction, diagnoses with uncertain biological basis, diminished function and behaviour patterns), revealing commonalities that may lead to characterisation as disease or non-disease, implications for action and consequences. These considerations highlight the importance of the issue of categorisation of disease and non-disease.

Whether one conceives a state of being as a disease—one might call this medicalisation—or not has important implications. On one hand, medicalisation increases public awareness of mental illness and is likely to increase the social acceptability of suffering from mental illness, may increase public funding devoted to the management of mental problems and to research addressing those problems, may increase rates of help-seeking for mental health problems and may facilitate access to care (eg, insurance reimbursement). These consequences of medicalisation have undoubtedly contributed to improvements in quality of life for many people.

In addition, medicalisation may lead to a more compassionate and less judgmental view of the afflicted, and influence views on how best to help people with their problems. For instance, Canada has recently witnessed a debate regarding harm reduction and safe injection sites as a way of dealing with drug addiction.24 Those inclined to view drug addiction as a disease have tended to support these strategies. Those who view addiction as a moral failure have been included to oppose these management approaches.

On the other hand, medicalisation generally promotes the minimisation of social and cultural explanations for human suffering (eg, social inequality).10 25–27 Medicalisation is likely to lead to pharmacological interventions, even when psychotherapeutic treatment or lifestyle modifications (eg, regular exercise) or even reassurance are equally or more effective.4 20 28–30 For example, physicians who consider social anxiety disorder to be a disease (over 80% of psychiatrists and two-thirds of other physicians—figure 2) may be more prone to offer medical treatment; those who do not may normalise the experience and offer reassurance (perhaps in keeping with their patients’ views—less than 40% were inclined to the disease label). Thus, general practitioners holding medicalised conceptions of mental health conditions may be driving increasing diagnosis and treatment of mental illness by clinicians with limited formal training in psychiatry, contributing to overutilisation of psychotropic medications.4 20 28–30

Similarly, labelling work exhaustion as a disease may lead clinicians to focus on the individual and not to work-related factors, such as bullying, management problems and work overload. In this kind of situations, the problem is again not only potential overuse of pharmacological interventions but also underuse of interventions that could lead to better outcomes, such as occupational rehabilitation.4 20 29–31 Finally, with healthcare consuming an ever-increasing proportion of the gross domestic product of developed countries, medicalisation may target resources toward increasingly minor conditions and away from other threats to population health.32

An important contribution of this study is our finding regarding the gradient of medicalisation across groups, and particularly, the much greater inclination for psychiatrists to label states of being as diseases. For some states (eg, approximately two-thirds of psychiatrists see drug and gambling addictions as diseases; this is true of only one-third of laypeople—figure 2), one might view this as an enlightened view by psychiatrists associated with a reluctance to blame the victim, and a desire to provide help rather than censure. For other states, in which there is a large discrepancy (social anxiety and generalised anxiety disorder), one might see the nefarious influence of the pharmaceutical industry and specialty self-interest leading to over-medicalisation resulting in approaches that do more harm than good.21 33

The view of the psychiatrists who participated in our study is reflected in the expansion of disease definitions that took place in the new edition of the diagnostic manual, DSM-5, which has raised criticism of overdiagnosis and treatment in the field of psychiatry.34–36 Other observers have suggested that the revised DSM is a work in progress that serves as a better platform for developing neuroscience and brings psychiatry closer to rest of the medicine.37 38 The contrasting views of professionals and laypeople provide a possible justification for lay input into medical practice, including lay participation in clinical practice guidelines and diagnostic manuals.39

Our results, therefore, highlight the desirability of a more inclusive conversation regarding what states of being should be considered diseases, and the implications of these decisions, particularly, around issues of medicalisation. Our findings reflect on the profound implications of attitudes regarding which health states should be seen as diseases; one could argue that this paper should be required reading for those categorising health states, and making recommendations regarding their management, both in the clinical and health policy spheres.

Laypeople are now routinely included in clinical practice guideline panels.40 Including them in groups that produce diagnostic manuals would also be highly desirable. In terms of future research, qualitative studies exploring the reasons for peoples’ decisions regarding what states constitute diseases, and the implications they see for these decisions would likely provide additional important insights into this crucial issue.