REPLICATING ROSENHAN’S STUDY: A METHODOLOGICAL AND ETHICAL ARGUMENT AND PROPOSAL
Abraham Rudnick, MD, PhD, CPRP, FRCPC, CCPE,
Professor, Department of Psychiatry and Behavioural
Neurosciences, McMaster University
Email: [email protected]
Abstract: The quality of inpatient mental health care, particularly in psychiatric hospitals, has been criticized for many years to date. A famous critique of such care is related to Rosenhan’s study (published in 1973), which demonstrated unsound psychiatric diagnosis and suboptimal inpatient care. Attempting to replicate this study nowadays may be informative, yet current research standards would not allow conducting it without modification, as it would now be widely considered methodologically unsound and ethical unacceptable. In this paper I examine Rosenhan’s study in light of contemporary methodological and ethical standards and propose modifications to it for a sound replication attempt.
Keywords: Ethics, methodology, quality of mental health care, Rosenhan.
Acknowledgement: The inspiration for this article was provided by my correspondence with a patient advocate.
The quality of inpatient mental health care, particularly in psychiatric hospitals, has been criticized for many years to date. This has been the case at least since the mid-19th century when psychiatric asylums were overcrowded, to which a reaction starting in the mid-20th century was de-institutionalization and later provision of more and better community supports (Carling 1995). As part of this trend, a famous critique of such institutional mental health care is related to Rosenhan’s study, which demonstrated unsound psychiatric diagnosis and suboptimal inpatient care (Rosenhan 1973). Rosenhan studied the experience of 8 so-called sane people (some of whom were mental health care professionals) who falsely reported hearing voices, i.e., having auditory hallucinations, in order to be admitted to a psychiatric hospital; once admitted, they stopped reporting hearing voices but were all still provided care for, and eventually discharged with, a diagnosis of schizophrenia. The principal conclusions from this study were that psychiatric diagnosis is unsound and that inpatient care in psychiatric hospitals is suboptimal.
Attempting to replicate Rosenhan’s study these days may be informative, e.g., to examine psychiatric diagnosis and inpatient care in contemporary psychiatric hospitals, with the hope that due to arguable improvements in psychiatric diagnosis (American Psychiatric Association 2013) and quality of care (Jayaram 2015) there has been improvement in diagnosis and care in psychiatric hospitals. Yet current research standards would not allow conducting Rosenhan’s study without modification, as it would now be widely considered methodologically unsound and ethical unacceptable. For example, methodologically, although the study sampled various psychiatric hospitals, it did not specifically assess – and hence was not controlled for – clinical skills of the service providers. And ethically, most of the 8 participants were not identified to the participating hospitals, and were provided care for schizophrenia that they did not have, hence risking unnecessarily prolonged negative consequences such as serious adverse effects of medications and stigma. Of note is that there was no standardized regulation then of human research, as such regulation started in the late 1970s following the Belmont Report (http://www.hhs.gov/ohrp/humansubjects/guidance/belmont.html#).
To replicate this study nowadays, it would have to meet current methodological and ethical human research standards. I propose here a modified study design. The study would consist of a randomized controlled multi-site trial in psychiatric hospitals and in psychiatry units of general hospitals with a statistically powered (sufficiently sized) sample of trained actors simulating schizophrenia; half of the actors simulating auditory hallucinations with other schizophrenia-related symptoms and another half simulating auditory hallucinations without other symptoms. Mixed (quantitative and qualitative) evaluation methods would be used, such as reporting of admission and discharge diagnoses, care provided and length of admission, as well as related experiences of the actors, other psychiatric inpatients – who may sometimes identify simulators of their own mental illness better than staff do (Rosenhan 1973) – admitted to the study units, and involved staff. Informed consent of the involved patients (actors and others) and research ethics board approval, including for initial deceit (of staff), would be required. Other study designs may also merit consideration.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 5th ed. Washington, DC: American Psychiatric Publications, 2013.
Carling PJ. Return to Community: Building Support Systems for People with Psychiatric Disabilities. New York: Guilford, 1995.
Jayaram G (ed). Practicing Patient Safety in Psychiatry. Oxford: Oxford University Press, 2015.
Rosenhan DL. On being sane in insane places. Science 1973. 179(4070):250-8.