Dr. Ronald Pies is, characteristically, correct in pointing out that language is important. Elegant language should be used to explain and contextualize the evidence of empirical research, not to obfuscate and not to serve as an alternative to scientific analysis. So I am sorry that Dr. Pies was confused, rather than inspired, by my essay. He raised several important questions that I had clearly failed to address adequately. In particular, Dr. Pies suggested my essay was extraordinarily broad and entirely unreferenced; if I had explained myself better, I’m hopeful that Dr. Pies might have found more with which to agree.


So… I highlighted what I termed “the poor reliability, validity, utility, epistemology and humanity of psychiatric diagnoses”.

These issues have been discussed… quite a lot recently, and we could end up repeating ourselves. So I won’t just refer to sources of those missing references (especially books by Richard Bentall, Jo Moncrieff and John Read and Jacqui Dillon), but also engage with Dr. Pies’ specific questions.

First, Dr. Pies asked what I meant by a “psychiatric diagnosis”? He asked whether that referred “to any diagnostic conclusion reached by an experienced psychiatrist? To any or all DSM-IV or DSM-5 criteria sets? If a social worker diagnoses “obsessive-compulsive disorder”, does that constitute a “psychiatric” diagnosis?” The answer, in brief, is “yes”, that sounds like a fair conclusion. I certainly don’t think that psychiatrists, as a profession, make flawed diagnoses and that other professions, such as social workers are somehow more (or less) skilled. I mean quite the reverse – that the theoretical approach, the carefully-crafted tools, and, specifically, the DSM-IV, DSM-5 and ICD criteria sets actively hinder caring and skilled clinicians as they try to understand and help their clients.


I’m genuinely appreciative that Dr. Pies set out his questions so clearly. Unfortunately – or perhaps rather positively – there are complexities in the answers that mean we have a lot to discuss. So Dr. Pies next asked whether I was aware “that the inter-rater reliability (kappa) of several psychiatric diagnoses has been shown to be comparable to, or better than, some diagnoses in pathology and internal medicine, when structured clinical interviews are used?” (and cited a specific paper).

So… I am aware that medical diagnoses are not perfect. I am, in fact, mildly concerned about the fact that I and my family put a great deal of trust in the judgment of medical professionals (and take medication as a consequence) without necessarily testing out possible alternatives. In general practice, for example, there have been a number of trials (with varying success) of the use of computerized ‘expert systems’ to take out the vagaries of human judgment.

In the field of psychiatry, my point was not to contrast the difficulties of the psychiatric diagnostic process with the supposed simplicities of ‘medical’ diagnosis – but to contrast the inadequacies of psychiatric diagnoses with a proposal for a problem-focused approach. Drawing this comparison is important and was the main point of my essay. Indeed, it might have been helpful – for me – if Dr. Pies had actually addressed this. Does he think, for example, that diagnosis adds clinical utility over and above the problem-focused approach I have described? And if so, what is this benefit?

But also… it’s interesting that Dr. Pies focused on four very high kappa coefficients from DSM-III. Because, of course, the kappa coefficients for ‘post-traumatic stress disorder’ (0.55) or ‘attention deficit disorder’ (0.50) or even ‘autistic spectrum disorder’ (0.01) are not so robust. And it’s interesting that we’re talking about DSM-III diagnoses, because the inter-rater reliability for the DSM franchise has been falling… as they refine the brand, its performance drops.

But even despite this cherry-picking, I think that Dr. Pies and colleagues are in danger of comparing apples with pears. One study that Dr. Pies refers to compares the ‘high’ reliability of psychiatric diagnoses with the poorer reliability of medical diagnoses by looking at inter-rater agreement in stroke patients. But the two kappa coefficients being discussed refer to different things. In relation to the DSM-III, the kappa coefficients refer to two clinicians agreeing (or not) whether a particular putative disorder is present – whether the criteria are met for its diagnosis. The kappa coefficients in the stroke paper refer to raters agreeing not on whether there has been a stroke or not, but on its aetilogy (comparing “atherosclerotic large-vessel disease, lacunar infarct, cardioembolic stroke, stroke of other known cause, and stroke of undetermined etiology”).

So I stand by my “psychiatric diagnoses are unreliable” claim (especially in routine clinical practice, when diagnostic manuals are more decorative than obligatory). And whilst I agree with Dr. Pies that validity is much more important, I worry about reliability when I see prominent psychiatrists suggest, for instance, that “Clinicians need to communicate to each other, and even a wrong diagnosis allows them to do so”.

Predictive (In)validity

In reference to my claim that psychiatric diagnoses “fail to predict the course of a person’s problems”, Dr. Pies pointed out that “several studies of DSM-defined schizophrenia have shown good predictive validity”. That’s a very fair point. There have been several studies along these lines, and it’s absolutely true to say that if you first receive a diagnosis of schizophrenia, then, 13 years later, you are much more likely still to meet that diagnosis (although there are some differences between the different diagnostic systems). But that wasn’t quite what I was, perhaps inelegantly, groping at.

I wasn’t suggesting that a person could experience a random collection of problems at one point in time and experience an equally random and unrelated variety of difficulties at a second time. I believe that a person’s difficulties will have structure and meaning, and have explicable aetiological causes. So if people experience hallucinations (let’s say) at one point in their lives, it is not particularly surprising that they experience hallucinations later. This merely speaks to the fact that human experiences are understandable and non-random, but does not necessarily speak to the existence of a valid syndrome called schizophrenia. Indeed it is very sad that the consistency across time is as high as is reported. If the ‘treatments’ for the diagnosed ‘disorders’ had been successful, then you would expect that the problems would be resolved, and you would hope that a few years later there would be relatively little concordance (if successfully treated, the symptoms would no longer exist).

But, as I say, I wasn’t suggesting – this is not our contention at all – that people’s experiences are random. The point is that diagnoses such as ‘schizophrenia’ do not efficiently predict the course or outcome for individuals – whether they will remain in hospital, remain reliant on medication, work, marry, raise children (Ciompi, 1984; Sartorius et al, 1987).

(Non)indicator of Treatment Success

And that leads on to treatment, of course. And I’m afraid, here, Dr. Pies and I may disagree a little more. Dr. Pies asked whether I “believe that a DSM diagnosis of bipolar I disorder does not predict, in general, a favorable response to lithium carbonate?” First, and I’m sorry if I’m repeating myself, the DSM-5 field trials revealed a kappa coefficient of 0.54 for the diagnosis of bipolar I disorder… so I wouldn’t place too much faith in the reliability of the diagnosis.

Many of my medical colleagues have some faith in the effectiveness of lithium carbonate in stabilizing moods although (as I suspect readers will realize) I would prefer to use other effective, and less toxic, methods myself. But my argument – our argument – isn’t to try to undermine and criticize established research. I don’t think it’s remotely sensible to claim that medication has no effect on our mood, or that those effects cannot be helpful. But, and probably partly because their reliability is so low, diagnoses are not as effective in predicting responses to drugs as are approaches based on specific problems.

To try to be as clear as possible: medication can be helpful and medical psychiatrists play a vital role. But they would be much more effective in helping people if they responded to specific problems than if they attempted to give the treatments predicated by diagnoses.

This is clearer still in respect to Dr. Pies’ next question [would I believe that] “a diagnosis of major depressive disorder fails to predict a good response to cognitive-behavioral therapy?”. Well… technically, yes… and no. It’s pretty clear (although some of my colleagues seem to be more skeptical) that CBT is helpful for people with low mood… but also with a wide range of other problems. That’s understandable. People make sense of their world. These frameworks of understanding are, in large part, consequent on the experiences they’ve had, especially in childhood and adolescence, and how they’ve learned to respond. And, of course, the way in which people make sense of their world has profound effects on their emotions and behavior; their “mental health”. So, therefore, it’s entirely unsurprising that CBT – a pretty straightforward, but highly skillful approach to helping people examine and change their framework of understanding of the world – is effective. But, again for entirely unsurprising reasons, CBT is helpful across a wide range of problems; any problem where a person’s cognitive schema plays a role (which is, of course, true in many cases).

So CBT is very helpful… but it’s not a specific treatment for ‘major depressive disorder’. And this, technically means that while somebody with a diagnosis of major depressive disorder may well respond well to cognitive-behavioral therapy, so would someone without that diagnosis. Once again… it’s the identification of problems, not the diagnosis of illnesses, that’s useful.


That, then, is what I mean by “psychiatric disorders have poor ‘utility’”. They aren’t useful over and above simply focusing on the problems people experience. Two people with the same diagnosis may have two largely (sometimes completely) non-overlapping sets of problems, with no established underlying pathology. Proponents of psychiatric diagnosis argue that a diagnosis is useful for communication, but since the information doesn’t tell you what problems a person is actually experiencing, doesn’t specify what treatment will be effective and doesn’t point towards a specific set of causal agents, I find this claim hard to understand. I worry that what’s actually being communicated is what Dawkins calls a ‘meme’ – a belief system (this person is suffering from schizophrenia, and we all know what that means, don’t we?).


“What studies can Prof. Kinderman cite showing that psychiatric diagnoses lack “humanity”? In what sense is this true?”

Dr. Pies is absolutely correct – when I refer to ‘humanity’, I’m making a moral judgment. I fear that the ‘diagnosis-treat’ model leads to inhumane treatment. I absolutely do not believe that my traditional psychiatric colleagues are themselves inhumane, but I believe that, with great sympathy and with the best interests of their clients at heart, inhumane practices abound.

So… a study… well, rather sadly the United Nations recently published a report into psychiatric care suggesting that some practices were “akin to torture”. Not “inhumane”; “akin to torture”.

Sadly, these experiences are not limited to UN reports, nor to the stranger shores of experimental therapy.

I receive a fair number of unsolicited letters. One summed up the situation nicely:

Rather than engaging with the patients on the ward, the staff instead shepherded them around like sheep with bullying commands, threats of ‘jabs’ (injections), and removal to an acute ward elsewhere in the hospital, if they did not co-operate. The staff also stressed medication rather than engagement as a way of controlling the patients. And the staff closeted themselves in the ward office, instead of being out and about on the corridors and in the vestibule where they should have been. The staff wrote daily reports on each patient on the hospital’s Intranet system; these reports were depended upon by the consultant psychiatrists for their diagnoses and medication prescriptions, but were patently fabricated and false, because the staff had never engaged or observed properly the patient they were writing about in their reports. The psychiatrists themselves were rarely seen on the ward, and only consulted with their patients once a week.

Is diagnosis part of this picture? Unfortunately yes. A few years ago, Erika Setzu and I, together with colleagues at the University of Liverpool, conducted a simple research study into serious mental health problems. We asked people who either had been, or were currently, in great distress and admitted to acute psychiatric units, about their own understanding of the circumstances of being admitted to hospital. One interview (with, it’s perhaps important to note, a male participant) went as follows:

Erika: “Do you know why [the problems began]?”

Interviewee: “No, I don’t know for certain. I had taken some pot before my first admission and I thought somebody might have dropped some acid on it. I also had a difficult childhood. I was physically and emotionally abused, sexually abused as well. This is very difficult for me to get my head round.”

A little later, the interview continued:

Interviewee: “I started to hear voices, but they were not nice voices, they were horrible.”

Erika: “Did you recognise them?”

Interviewee: “It was the man that abused me …I met this man that was a builder, in construction, you know? And he said that he wanted to give me a job, but they were all lies, he was trying to con me. He took me back to his house, he locked the door and he had sex with me. … And then other voices as well. I went to … hospital and the nurses were very good to me.”

Erika: “When you went to hospital what did they say it was wrong with you?”

Interviewee: “Schizophrenia, paranoid schizophrenia.”

Erika:  “What do you think personally?”

Interviewee: “What do you mean?”

Erika: “Do you think it is what you’ve got?”

Interviewee: “Oh yes, that’s what I have got.”

To my mind, the reduction of this man’s narrative from the understandable consequences of rape to the symptoms of schizophrenia is inhumane. It is also, as I’ll suggest in a moment, an epistemological point, too. But I am concerned at how we become uninterested in what has happened to people and become interested in what they’ve “got” and what drugs we can give them. The point is that such diagnoses expunge our lived experiences from the narrative, rendering our stories pointless, and reducing our human responses to symptoms of illness.


Dr. Pies asked; “What studies can Prof. Kinderman cite showing that psychiatric disorders have poor “epistemology”? What can this claim even mean, since epistemology is the branch of philosophy concerned with the nature and scope of knowledge? How has Prof. Kinderman measured the quality of the “epistemology” of psychiatric diagnosis?

In epistemological terms, diagnoses convey the idea that people’s difficulties can be understood in the same way as bodily diseases, which excludes the possibility of finding meaning in people’s ‘disordered’ responses and experiences, and thus prevents people from understanding how they might use their own resources to address their difficulties. Worse still, diagnoses are used as pseudo-explanations for troubling behaviours (he did this because he has schizophrenia) without consideration of the circularity of that argument, and the broader context (e.g., whether a paranoid person has actually been victimised). And as a result of all these failings, the diagnostic tools that we are currently living with mean a person’s social and interpersonal difficulties are often ignored in the hope that the right medication regimen will achieve the desired return to normal functioning.


Dr. Pies was confused by my use of the term “mapping”; and asked a few very pertinent questions:

What does Prof. Kinderman mean when he says that psychiatric disorders “fail to map onto an entity discernible in the real world?” What sort of “entity” does he have in mind? A lesion? A bacterium? A damaged section of the brain? Can Prof. Kinderman identify the “entity” in “the real world” onto which the diagnosis of migraine headache “maps”? How about chronic facial pain? What is the “entity” onto which the diagnosis of fibromyalgia “maps”? Do any of these three recognized disorders/diseases “map neatly onto biological findings”? Or are these not “real diseases” or disorders in the sense Prof. Kinderman would define these terms? What does it mean, “to map onto”, anyway? Is this a scientifically-defined term or merely a metaphor derived from the now largely discredited school of logical positivism?

Well, first, I can’t comment on the validity of “migrane”, or “firbrobmyalgia”. But if these disorders do not have identified biological underpinnings I would start to be skeptical of their validity. As a scientist, I recognize we might not yet know all there is to know, but I’d start to be worried if there were no biological abnormality, disease vector… as indeed we find with many psychiatric disorders, of course. I’m not a medical doctor, and I trust my medical colleagues are scientific and skeptical in this regard, but if “migraine” or “firbrobmyalgia” have no discernable ‘biomarker’, then yes, I’d be skeptical of their validity, too.

The mention of ‘chronic facial pain’ is very interesting, as that strikes me as exactly the kind of problem-focused identification of a difficulty that we’re recommending. The person is experiencing pain, it’s in the face, and it’s chronic. That implies nothing but identifies the issue. It’s clear, descriptive and makes no implications as to the underlying nature of the problem. ‘Migraine’ is different, and begins to make assumptions that need to be justified.

So “mapping onto real world entities” does include the notion of biomarkers. It also speaks to the construct validity of psychiatric disorders. That is, when we expose psychiatric diagnoses to factor analysis or cluster analysis or other statistical techniques, the wise and careful judgments of committees of experts turn out to be … wrong. If these diagnoses were valid, we’d see large numbers of people meeting the criteria for no disorder at all, some people meeting the criteria for one disorder, but not others, and relatively few meeting the criteria for more than one. The diagnoses should – in Plato’s phrase – “carve nature at the joints”. But I think it is hard to argue that they  DSM and ICD achieve this. The reason – as Thomas Insel recently implied, but others such as Mary Boyle and Richard Bentall have been saying for years – is that few of the current diagnostic categories were “discovered” from the bottom-up. Rather, for a variety of reasons that can be traced historically, they were largely determined a priori, by committee and not on the strength of evidence. A recent example of this process in action is the inclusion of Disruptive Mood Dysregulation Disorder in the DSM 5. It is acknowledged that this diagnosis lacks validity and specificity and has a weak basis in empirical research. It has not been “discovered” by application of the scientific method but conjured into existence for non-scientific reasons (to solve an embarrassing problem for the APA, namely the enormous rise in diagnoses of childhood ‘bipolar disorder’ and consequent dangerous proliferation of medication).

What we see (and just one piece of research below) is a pattern of human experiences which has meaning and structure… but not support for DSM and ICD diagnoses, no matter how much the expert committees think constructs such as ‘schizophrenia’ must exist.

Diagnoses don’t “map onto” real entities. They don’t reflect biomarkers and they don’t reflect the “joints” in nature.

Dr. Pies expands these ideas by asking, “Could it be said that one very relevant real-world “entity” is the suffering and incapacitated patient, and that at least some psychiatric disorders actually “map” quite well onto the self-generated description of that patient’s suffering and incapacity?

Absolutely. I can’t agree more. That’s exactly and precisely it. A “self-generated” description. That’s precisely our point. But Dr. Pies spoils the moment. His comment goes on to describe:

For example, when a patient complains of having had, for the past 12 years, command auditory hallucinations telling him to kill himself; the feeling that his arms are melting into the chair; the belief that a radio transmitter has been implanted into his brain by the KGB; that everyone around him can overhear his thoughts, and plant dangerous thoughts in his mind…”

And to that point I’m completely in agreement. But Dr. Pies– and this is, in a sense, the problem the point of the DSM – doesn’t follow his own good advice. He doesn’t stop at the client’s “self-generated” list of difficulties but insists on applying the (flawed) logic of a committee of experts… “is there a legitimate and clinically useful sense in which the DSM-IV or DSM-5 category of schizophrenia “maps” onto this particular patient’s experience and world-view?” To which the answer clearly is NO!

We can, we should, work with self-generated problem-lists. The difficulties arise when, completely unnecessarily, scientifically unwisely, we insist upon adding meaningless and misleading committee-generated labels to this useful and valid description of a person’s problems. The labels necessarily obscure the real nature of the person’s difficulties. My question to Dr. Pies would be: since clearly describing a person’s problems seems so helpful, what additional benefit is offered by a diagnosis?

Dr. Pies points out that I have, myself, used psychiatric diagnoses. Of course. I would like to see a system whereby we all study, care for and talk about specific problems. But I don’t yet live in that world. In the meantime I have to do two things; live in the reality of the academic and clinical world in which I find myself, and work to change it.


Dr. Pies was intrigued by my suggestion that “psychiatric diagnoses have ‘ethical failings’” – meaning that diagnoses are “used as pseudo-explanations for troubling behaviours.”

I think this does have huge ethical implications. The best (or perhaps worst) example of this is “coercive paraphilic disorder”. In any ethical world, if we come across a man who repeatedly rapes women (or men), I would hope we would try to investigate and understand all aspects of why this happens. I can conceive of circumstances where this would lessen the culpability of the individual. But I think there are huge ethical implications of saying that “the rapes are in fact symptoms of his ‘paraphilic coercive disorder” or that “he raped because he is suffering from coercive paraphilic disorder”.

Framing the situation this way implies that his behaviours were CAUSED by the “disorder” to which the label refers. But, the idea that there might exist such a disorder did not come about – as we might reasonably expect – as the consequence of a process of scientific discovery. The disorder was brought into being by a committee of (as has been widely noted, financially compromised) experts who share a particular perspective. For me, this is ethically troubling.

Dr. Pies also was perplexed by the view that people claim that “psychiatric diagnoses imply that people’s difficulties can be understood, or ought to be understood, “in the same way as bodily diseases”.

He’s right of course. This claim does not square with the “biopsychosocial model”, which has been the predominant paradigm in American academic psychiatry since George Engel introduced the term. But [URL] the principles of the biopsychosocial model are often rather undermined by the tendency to see biological factors as dominant or ‘primary’. So social factors are often regarded as moderating underlying genetic vulnerabilities… I think people often find it difficult to think of genetic factors mediating the effects of social factors (such as abuse) on outcome… although, in scientific terms, that is of course precisely what’s happening.

And is it true that people say that psychiatric problems are essentially the same as bodily illnesses?

Well… I invite readers to explore this question for themselves. Our beloved Google reveals that the search “depression is a brain disease” textbook gives us several (actually thousands of) hits. I’ll offer a prize (really, and it’ll be fantastic) for the best example. My favorites include Charles Nemeroff making a typical robust offering… although here he does reveal some of his stock options. Or “San Diego’s Voice on Mental Illness” which tells the citizens that depression is “an Illness Like Any Other: Mental Illnesses are Brain Disorders”. This is important, of course, because the “an illness like any other” … approach actually increases stigma.

Am I aware that well-known psychiatric textbooks promote a much more humane and scientifically valid approach? Of course. But I would maintain that – as my correspondent above suggested – that’s what’s SAID. What’s done in practice is often quite different and, in some cases, quite brutal.

Peter Kinderman

About Peter Kinderman

Peter Kinderman is Professor of Clinical Psychology at the University of Liverpool, and an honorary Consultant Clinical Psychologist with Mersey Care NHS Trust and has twice been elected Chair of the British Psychological Society’s Division of Clinical Psychology. Professor Kinderman is Head of the Institute of Psychology, Health and Society at the University of Liverpool, comprising psychiatrists, GPs, clinical and other applied psychologists, sociologists, public health physicians, nurses, sociologists and academics.