Abstract
The pursuit of prodromal diagnosis of psychosis is justified by the urgent needs of early intervention. Both prodromal diagnosis and early intervention dominate research and service agendas. We argue that prodromal diagnosis is a mistaken need and actually militates against early intervention. The idea of prodromal diagnosis misconstrues the conceptual nature and meaning of diagnosis. Early intervention demands that we should treat symptoms as they present rather than look for hidden illness processes.
Prodromal diagnosis and early intervention
For many clinicians, early intervention in psychosis appears to depend on the earliest diagnosis, if possible ‘prodromally’, before the process underlying psychosis is established [1]. Yet, for prodromal diagnosis, research optimism is subsiding under the weight of failure of prediction. ‘Transition to psychosis’ rates, which were as high as 40% in early samples, have fallen as low as 9% despite the refinement of detection techniques [2,3]. Prediction of transition seems to reflect the selection of the subjects more than the detection of a disease process [4].
Consequently, early intervention, despite its almost axiomatic appeal, has been given surprisingly harsh criticism for its failures [5,6] and its potential harms [7]. The slightly shocked response to this includes exhortations to ‘keep the faith’ [8] and to look at measures of social outcome, as well as demands for more research and expenditure to improve early detection and treatment, despite recent research outcomes suggesting that more research will simply spin the wheels deeper into the mud [4].
Surprisingly, if prodromal diagnosis and early intervention are scrutinised, they are not a pair of horses pulling the clinical cart in the same direction, but actually pull against each other to retard progress. This antagonism flows from misunderstandings in the concepts underpinning diagnosis and from misleading analogies with other branches of medicine [9].
Misguided concepts
Both prodromal diagnosis and early intervention appear to depend on the detection of the ‘signal’ of the disease process at the earliest point that it can be discriminated from the ‘noise’ of those symptoms that do not presage psychosis. Prodromal diagnosis and early intervention both see this at their core even though they differ how they deal with the ‘noise’. For prodromal diagnosis research, such non-predictive symptoms are to be discriminated and set to one side; for early intervention the non-predictive symptoms cannot be ignored because often these are the symptoms that have brought the person to treatment.
The assumption of a disease process of psychosis is where the difficulties begin. For this makes it seem that the problem is the validity of the diagnosis as an indicator of the essential disease process. However, the diagnosis of ‘psychosis’ started merely as the name of a collection of meaningfully connected human phenomena. Over generations of use it has become a label attached to a presumed but unidentified disease process. This attachment of a label feeds endless discussion of syndromal diagnosis and how it is to be advanced to diagnosis by aetiology or at least by pathological process. Why this construction leads us into trouble cannot be understood from within the medical framework because this sort of causal explanation is so embedded in medical theorizing. We are forced to look at what sort of concept ‘psychosis’ is.
Psychosis is a creation not a discovery
Prodromal diagnosis of psychosis, according to the most widely used criteria, is built on attenuated psychotic symptoms (APS), brief limited intermittent psychotic symptoms (BLIPS) and on trait and familial risks [10]. Look at APS and BLIPS. Both include ideas of reference, odd beliefs or magical thinking, perceptual disturbance, paranoid ideation, odd thinking and speech [10]. In prodromal states, the APS are less intense, and BLIPS are less continuous representations of the delusions, hallucinations and formal thought disorder found in the classical descriptions of schizophrenia, and are now clearly enunciated in the DSM and ICD. (It is as though, in its embryonic form, the psychotic process appears in the miniature or faint but accurate shape of the full blown condition and that it takes the form as it was described a century ago.) Historically, when schizophrenia was being described, the preludes to psychosis were seen to be quite different [11], and today clinicians outside institutions see all sorts of symptoms and signs that turn out to be precursors of psychosis.
However, the conceptual difficulties run deeper than the accuracy or validity of the prodromal description. Reviewing the history of the concept of schizophrenia, German Berrios et al. wrote ‘the current view of schizophrenia is not the result of one definition and one object of enquiry studied by various psychiatric teams, but a patchwork made of clinical features plucked from different definitions…It might simply be the result of historical ignorance or the application of some pedestrian operationalism’ [12].
Kraepelin's synthesis of co-occurring and often progressive phenomena into dementia praecox was his third attempt at a convergence of phenomena published in the last edition of his textbook. (Kraepelin later, in part, recanted this convergence [13].) Bleuler took the conceptual convergence and named it schizophrenia and this became the ‘real, recognisable unitary and stable object of enquiry’ [12] that has dominated western psychiatry since. The diagnosis identified phenomena that frequently co-occurred and drew them together. Although we see this as being like the discovery of a previously unknown continent, it was really more like the creation of a nation from different states. The phenomena were known before, even if they were combined into different diagnoses (for example ‘lypemania’ for certain delusional states [14]). Kraepelin and Bleuler probably thought that these manifold signs were drawn together by a single underlying process. (Bleuler, however, allowed some latitude from a categorical disease pathology saying ‘it is conceivable that the entire symptomatology may be psychically determined and that it may develop on the basis of slight quantitative deviations from the normal’[15].) Certainly, the profession has subsequently treated the diagnosis as an essential disease process and a foundation of its thinking.
There has never been an underlying disease process to be identified. The diagnosis is not a label of an object or process, it is a complex human concept which draws together a range of phenomena but is not a phenomenon itself. Such complex human concepts are common. The concepts of ‘weeds’ or ‘precious metals’ are examples of concepts that draw together various natural objects which are connected by networks of human needs and practices. (There is no characteristic of a plant that can be identified in a laboratory that makes it a weed; it is a plant growing in a place or manner that interferes with human needs or wants. Similarly, precious metals are defined only by human uses and valuations.) Concepts such as weeds and, we argue, psychoses, are not objects themselves and there is no reason to expect that they are connected by an essence, an underlying process or a cause. This is a subtle concept to grasp and it can be approached in a number of ways.
Absence of an underlying process
The cardinal features of psychosis, say delusions, hallucinations and thought disorder, occur in different neural and psychological realms. Why should they be joined by a single underlying process? The obvious answer to this is that, regularly, they co-occur and this reflects an underlying connection.
However, clinical experience and the ‘any two of five’ construction of the criteria in the ICD and DSM confirm that co-occurrence is variable and tenuous. In fact, two patients with the same diagnosis need not share a single cardinal symptom. While the idea of ‘fuzzy sets’ with ‘grades of membership’ is useful for conceptually modelling the disorder, it is useful only for understanding it as a complex concept (like weeds) that is meaningful to humans and does not necessarily point to an underlying neuropsychiatric process of psychosis.
Secondly, some phenomena, such as anxiety, co-occur more regularly in psychotic states than many of the cardinal symptoms of psychosis, yet they are ruled outside the central diagnostic concept. They are ‘co-morbid’, implying a different underlying process. In practice, co-occurrence is not the rule by which clinicians point to a unifying process or endophenotype. Clinicians use very obscure rules to identify one symptom as part of an endophenotype and another as ‘co-morbid’, and that identification does not seem to be based on evidence of co-occurrence.
Thirdly, most of the cardinal symptoms and signs occur commonly in the broader community without any apparent connection with an identifiable underlying disease [16,17]. With the diminishing rates of ‘transition to psychosis’, we can now see that, in many people, the cardinal signs of psychosis seem to be innocuous at least in terms of progression to illness.
Fourthly, recognition, for example of the anergia of ‘institutional neurosis’ [18] and the hallucinatory effects of sensory deprivation [19], seems to show that many signs are not necessarily internal to the person and their disease. We now understand that phenomena such as these, which were central to the diagnosis, need not flow directly from an internal disease process but come about by complex interaction with environment, interactions of which we remain only dimly aware. Similarly, worsening symptoms, signs and disability seem to indicate the progression of the underlying process. But worsening may also result from drug use, sleep disruption, affective change, isolation, perceived rejection, or the accumulation of domestic discord – factors that are outside the concept of an internal disease process of psychosis but inside the patients’ experience of illness. These complex human interactions can be observed by the experienced clinician but can be articulated only vaguely and certainly not in the criterion lists of the DSM, the ICD or the instruments of ‘ultra high risk’ research [10].
Misleading analogies
Psychiatrists have been well trained in the medical tradition that bends toward aetiological explanation and we grasp our subject matter by creating analogies with disorders in other branches of medicine. Infectious disease and certain cancers are the most obvious and powerful analogies for they bring with them fully developed ways of thinking about diseases and their treatment. Hence psychiatrists not only fashion the name ‘antipsychotics’ by analogy with ‘antibiotics’ but also take concepts of disease process, progress and treatment that are analogies of other disorders [20]. Through these analogies, the identification of schizophrenia, which historically was the meandering and uncertain recognition of co-occurring phenomena (traced by Berrios et al. [12]), has been retold as the discovery of an underlying ‘disease process’ that connects them. That ‘discovery’ was created by analogies with other diseases for which a single factor or process was a reasonable premise. It was not discovered by evidence that was before us. The evidence from the prodromal diagnosis project as well as the continuing frustrating absence of any consistent pathology actually seems to point away from a discoverable disease process.
The origin of misguided concepts
This ‘identification’ of underlying connection is an ordinary human habit of mind that Ludwig Wittgenstein called ‘a craving for generality…the tendency to look for something in common to all the entities which we commonly subsume under a general term’ [21]. We subsume various phenomena under a diagnosis and assume they are connected by an essence such as a common underlying pathological process. They may be connected by an underlying pathology, but also they may be connected by any number of human practices including the way we are educated to see connection. Wittgenstein gives the example of ‘games’ [22]. We identify a collection of activities as ‘games’ because we are game-playing animals raised in a culture that includes games. There is no essential characteristic that defines games or allows us to recognize that an activity is a game. We can neither set up criteria to decide what is a game, nor create instruments to detect that it is a game. We cannot know why some games cease being played nor predict what activity will be seen as a game in the future.
Wittgenstein extends this argument to mental states [22]. We think of ‘understanding’ as a particular mental state with particular processes. However, we recognize ‘understanding’ when we see someone doing algebra or empathizing with grief or following a map. These are various states that are brought under a collective term but they are not aspects of a single underlying mental process. The concept of ‘understanding’ is meaningful and useful but it is neither a phenomenon itself nor a brain process.
If the diagnosis of psychosis is analogous, and we believe it is, then we cannot expect an underlying process for which a prodrome can be identified. There are phenomena, some of which are mental processes, but they are collated into a diagnosis by virtue of the intricate network of human relationships between them. Some phenomena will be related by physiology but others will be related by habits, cultural meanings, learned behavioural responses, the reactions of others and so on. That collation will reflect the training and culture of the person making the diagnosis, not the ‘true identity’ of an underlying process.
If this is the case for psychiatric diagnosis, then the problem of prodromal diagnosis is not due to the insensitivity of our techniques or the variability of disease expression. The problem is that what we are looking for is not a thing but a complex human judgement made on the behaviour of the person in the world.
If this is so, the prodromal diagnosis project will fail, and it cannot be made to succeed by eliminating the subtle and complicated cultural activity of making a diagnosis or by constricting the concept to a phenomenological simplicity. In this situation, simple phenomena are ambiguous in terms of being ‘pathology’. In some situations a phenomenon will be pathology, in another it will be innocuous. Phenomena are given their pathological meaning by their complex relationships with other phenomena within and around the patient.
These then are some of the conceptual confusions that contribute to the failure of prodromal diagnosis and make it a vain future project.
What about early intervention?
The problem of prodromal diagnosis is not a problem for the diagnostic project alone. If early intervention depends on prodromal diagnosis then it is a problem for early intervention. It is more than a problem. Because of what the drive to prodromal diagnosis does to the clinical situation, it is an obstruction to early intervention.
In their conceptual history of schizophrenia, Berrios et al. [12 p.135] conclude ‘[It] may be that there is no such thing as a unitary disease of schizophrenia but only a collection of mental symptoms, some congenital, some relics from evolution and others acquired. Be that as it may, some of the symptoms [by which we define the disease] are blighting people's lives…’.
People come with symptoms and when we start with the question ‘Are these indicators of the covert prodromal illness?’ problems begin for both the clinician and the patient.
The doctor's quandary
Prodromal diagnosis moves the focus of the assessment from what the presenting symptoms are doing to the patient, to what they mean as indicators of an incipient covert process. What the symptoms might be takes precedence over what they are now and we are distracted from the careful examination of the present state and what has led to it. Current professional rhetoric makes early and prodromal diagnosis paramount and this biases the assessment of the patient. The phenomena are evaluated by their importance to an imagined diagnosis. We may simply look past phenomena that do not fit with prodromal diagnosis even though they may be woes that are lethal to the patient.
For the conscientious clinician, prodromal diagnosis will induce a hesitation in intervention. There are many states–suspicious sensitivity, subjective clouding of thought, roiling affect–that warrant treatment even though the processes that underlie them and what they will become are quite unknown. They may warrant treatment not only with medication but also with psychological and psycho-educational approaches usually reserved for ‘psychosis’; but those treatments have come loaded with the malignant expectations and meanings associated with ‘psychosis’ [23]. Similarly, the treatment for the covert ‘prodromal’ illness might appear inappropriate or unacceptable for the symptoms as they are at the time of presentation. The clinician apprehensive about diagnosing ‘psychosis’, should wait for ‘grounds more relative than this’ (Hamlet, Act II, sc. ii) before acting and is left, Hamlet-like, delaying action until crisis.
With these uncertainties, the clinician may turn to ‘diagnostic instruments’ believing that they are microscopes that allow us to see what is otherwise not visible to the naked clinical eye. The instruments are no more than lists of decontextualized phenomena compiled by experts, and the lists are proven failures as detectors of progressive illness. The instruments more resemble microscopes looked through from the wrong end, distancing clinicians from what is in front of them and making clinically vital connections in the history harder to see. Most phenomena are part of a particular pathological concept by virtue of their histories and their connections to other phenomena, not by their particular natures. ‘Nothing is hidden’ [22, para 435] and if it cannot be seen by skilled examination then, perhaps, it should not enter consideration. The question ‘Could it be…?’ should be no more than a prompt for careful examination giving due attention to the phenomena including phenomena such as the subtle alterations of tone and demeanour that the clinician can clearly sense but cannot articulate (like a musical instrument's sound that can be known but not said [22 para 78]). These clinical data are knowable but not sayable in a check-list of phenomena. (And this underlines the harmful futility of looking for simple rules or algorithms.)
If clinicians start treatment for a covert condition, they cannot know when to stop for they cannot know whether the non-appearance of disease is due to treatment or that it was never there. This should induce a further pause in intervention because all aspects of treatment carry their own significant morbidities (not only medicine side-effects but also demoralization by the baleful expectations of the diagnosis itself).
The dilatory patient
The treatment is often taken out of the clinician's hands by a patient who does not accept the treatment from which he can see no benefit. When the clinician looks past the patient's complaints to find the covert condition, the patient, feeling unheard and unconvinced, is likely to depart perplexed and angry. This is all the more so when the covert condition is associated with the history and meanings accumulated by ‘schizophrenia’ over a century, and in the present with street madmen and with hopelessness, violence and exclusion. If this does not discourage people from coming in the first place, then they are likely to leave when they see the clinician forging those links at interview.
Effects
So, prodromal diagnosis is inimical to early intervention by what it does to the fine minds of clinicians and to the fears of patients. In fact it may be inimical to any intervention at all until something in the patient's behaviour forces the sort of oppressive and painful intervention that the champions of early intervention so disparage [8]. (And do not presume that that behavioural change necessarily is the consequence of the progression of untreated underlying disease. It is as likely the consequence of the accumulation of domestic discord or social isolation.)
‘Centres of excellence’ and special services for younger and less disabled people mitigate many of these effects. There, the rhetoric is changing [24,25]. Symptoms are more the focus, because that is more acceptable to the patient, and the clinicians’ diagnostic ideas are less stated even though they remain wedded to a disease concept borrowed from other areas of medicine [26,27]. Outside these centres, the explicit prodromal illness model and the pursuit of the hidden unknowable disease holds sway.
Conclusion
The conflict between prodromal diagnosis and early intervention continues while psychiatrists remain shackled to disease models taken selectively from other branches of medicine. While we look past symptoms to the imagined covert disease rather than describing and treating what is before our eyes, delayed and inappropriate treatment will continue.
Footnotes
Acknowledgements