Wednesday, February 08, 2023

Truth about psychiatric diagnosis

The study by Rosenhan (1973) published in Science has probably always been difficult to believe. I’ve said before, though, that psychiatry doesn’t need to be so defensive about it (see eg. previous post). DSM-III was motivated to improve the reliability of psychiatric diagnosis by introducing operational criteria, at least partly due to the challenge of the Rosenhan study, because Robert Spitzer thought unreliable diagnoses must be invalid. 

The Rosenhan study was actually more designed to challenge the validity rather than reliability of psychiatric diagnosis. Andrew Scull (who I’ve mentioned before eg. see previous post) has recently published an article in History of Psychiatry summarising the evidence that the study was fraudulent. Six of the seeming nine original participants are said to have never been traced. Data from one of the participants was not included in the Science report and this psychology student at the time of the study, who became an academic psychologist, published his own more positive account of his experience (see article). 

At the very least, Rosenhan (1973) is biased, inaccurate, dishonest and exaggerated its findings. As Andrew says, there is a "possibility that they [ie. the six missing participants] may still surface" but he thinks it is more likely they "never existed at all". 

I know of two reports of modified repeats of the Rosenhan study. Because they are later, both were done in the context of the rundown of the traditional psychiatric hospital. Although all Rosenhan's pseudopatients were said to have been admitted to hospital, both these reports suggest that feigning an auditory hallucination does not now generally lead to admission, maybe because of the pressure on beds. Scribner (2001) used 7 volunteers with long well documented histories of chronic schizophrenia, six of whom were actually denied treatment and turned away. The baseline histories were therefore very different from Rosenhan (1973), whose pseudopatients were said not to have had a history of mental disorder. 

The other report was from the book Opening Skinner's Box (2004) by Lauren Slater. Interestingly, she too has been accused of never conducting her study (see article). As she says in the book in her chapter on Rosenhan:

Psychiatry as a field is, of course, predicated on the belief that its own professionals know how to reliably diagnose aberrant mental conditions and to make judgments based on those diagnoses about a person’s social suitability

Interestingly again, she seems to suggest that Martin Seligman, an eminent psychologist, was one of Rosenhan's pseudopatients, which is not mentioned by Scull (2023). As far as I know, Seligman is still alive, so it may be possible to check this.

Slater herself has a "formidable psychiatric history" and was admitted to a psychiatric hospital aged 14. She does not deny the reality of mental illness. Slater says she used someone else's name, so that she wasn't recognised, and denied any psychiatric involvement in the past. She relates that she presented herself nine times saying she was hearing a voice, and that, although she was treated kindly and was not admitted, she was prescribed a total of 25 antipsychotics and 60 antidepressants. Almost every time she says she was given a diagnosis of psychotic depression.

I suppose Rosenhan could be said to have had more impact on psychiatry as a social scientist than Andrew (except maybe Andrew’s influential dismissal of Foucault in the literature - see eg. previous post)! Not excusing Rosenhan's behaviour, but I think the scientific literature is plagued by such dishonesty as Rosenhan's. There is evidence, though, that at least aspects of his Science paper are correct. Certainly it was possible for a person who is not mentally ill to obtain admission to psychiatric hospital and mislead psychiatrists into diagnosing schizophrenia. Maybe this has always been the main message that people have taken from the study. Rosenhan does seem to have elaborated the details to reinforce his conclusion that psychiatric diagnosis is subjective and does not reflect inherent patient characteristics.

What worries me is that Andrew’s complete dismissal of Rosenhan’s study as fraudulent may reinforce the case that psychiatric diagnosis is objective, which it isn’t in any absolute sense. The limitations of psychiatric diagnosis do need to be acknowledged (see eg. previous post). If psychiatric diagnosis is meaningful, there will be inevitable inconsistencies.


Mark Needham said...

Scull relies on Cahalan's work to claim fraud, but her account of Rosenhan is... odd. I've done some investigation of my own, and it's clear that a major influence on Rosenhan was the work of Thomas Scheff and the labelling theory of mental illness, which had considerable interest among psychiatry's critics at the time. Less than a year before Rosenhan entered Haverford State in 1969 he wrote a review of Scheff's Being Mentally Ill in Contemporary Psychology, and the final paper makes several references to work in the labelling tradition. But Cahalan gives only a couple sentences to labelling theory and never mentions Scheff — despite some of her sources making clear that he was an influence. To add to the puzzle, Scheff is still alive, and at least as of 2016 he was still healthy enough for an interview. It's a strange, completely unaccountable omission, and the more I look, the more inclined I am to call Cahalan a fraud than Rosenhan. And by extension, I think Scull's judgment of the study is questionable as well.

As an aside, I don't think Seligman was one of the original pseudopatients. If I remember correctly, Cahalan investigated, and determined that he and Rosenhan did another covert tour of a mental hospital sometime after the original study.

DBDouble said...

Thanks, Mark.I agree Rosenhan’s study needs to be seen in the context of labelling theory. I also now see that Cahalan spoke to Seligman and he denied being a pseudopatient for the study. He did, however, say that he went undercover with Rosenhan for two days in 1973 after the publication of “On Being Sane in Insane Places” to help Rosenhan gather more colour for his book. She says she tracked down medical records, I think from Rosenhan’s papers, to confirm this. It’s not clear from Seligman’s quotes in Slater’s book that Slater realised this, calling Seligman a pseudopatient and describing his account of learning how to ‘cheek’ medication.

DBDouble said...

Correspondence with Andrew Scull

Mark Needham said...

It's a shame we can't examine all the material ourselves. If I put my defense attorney's hat on for a moment, Cahalan and Scull's evidence for fraud looks less than conclusive. One example, apparently an earlier draft of the paper from before Harry Lando was excluded contained the same numbers as the final version — as if they'd been fabricated. But Rosenhan writes in response to the letters in Science,

"On another matter, a writer questions whether the pseudopatients, several of whom had no previous experience in psychiatric hospitals, were capable of making the "complex" interactional observations reported in the article. In fact, those data were obtained in four hospitals by experience researchers who had worked in similar settings." (pg. 368)

Only four hospitals with quantitative data. Since some participants visited more than one hospital, it's possible the numbers were collected by as few as two: Rosenhan himself and someone other than Lando.

On another point, I have questions about Cahalan's characterization of Rosenhan's first hospitalization at Haverford State. In a lecture she gave to Weill Cornell, at around 46:00 she puts on the screen an excerpt from a letter from Rosenhan to Spitzer:

"On page 12, the material cited from Dr. Bartlett [the admitting doc] continues to be offensive and misleading. You now have it from myself and the superintendent of the hospital (who arranged my hospitalization) that my stay there was part of a teaching exercise, and had nothing directly to do with research. Moreover, Dr. Bartlett now confirms these facts. He initially wrote his letter, a copy of which was sent to you, because he had been needlessly humiliated by the hospital superintendent at a conference--an unfortunate matter that already has had unfortunate consequences. Even a cursory investigation on your part, directly to the hospital involved, or to Swarthmore College where I taught at the time, would have early [sic] indicated to you that your information was hearsay at best. (I do not care to think what that information is at worst.) The use to which that information already has been put by Psychiatric News should give you pause."

Cahalan says the claim that his stay was part of a teaching exercise was a lie. Was it? I think it's possible that his first hospitalization was exploratory and predated any clear idea of the final study. He may have felt the need to exaggerate his symptoms to be admitted. In another letter Rosenhan writes concerning, I think, Spitzer's request for the pseudopatient records,

"I had not thought much about trust until you mentioned it in your letter. You may have something there. Neither the way in which Bartlett says you obtained some of your information, nor the Psychiatric News episode, nor the hassle about the name of the hospital, have contributed much to my trust. Between those incidents and an unfortunate experience I had a few years back with these materials, I'm possibly being too cautious."

Is it really so suspicious that we don't know all the participants? Rosenhan seems to have had reason to protect his records, and after fifty years any older pseudopatients are almost certainly deceased.

(cont. in next comment)

Mark Needham said...

As a final note, it's not quite right for Scull to say that he never revisited the topic. In 1975 Spitzer solicited several responses to Rosenhan's study for the Journal of Abnormal Psychology. I believe he intended it to affirm the value of diagnosis and justify the then-in-development DSM-III. (One of Spitzer's colleagues on the DSM committee, Theodore Millon, wrote a paper for it.) Rosenhan provided "The Contextual Nature of Psychiatric Diagnosis" which elaborates his thoughts on the way the hospital setting and preconceptions from labels can bias clinical perception. (Last year I reached out to someone involved with this paper, but unfortunately they declined to go on the record.) Cahalan barely mentions this work, and Scull not at all.

Maybe I'm being too defensive of Rosenhan here, but I'm concerned about the narrative that's forming. I'm open to the possibility of fraud and I don't particularly care about Rosenhan's reputation itself, but like you I'm concerned it gives the impression that DSM diagnoses are valid — and even more I'm troubled at how labelling theory is being written out of history.

DBDouble said...

Thanks for your further comments, Mark.