Abstract
The Editor has asked me to provide a commentary on the editorials by Rosenman & Anderson [1] and McGorry [2]. In some ways, I could be seen as an unusually unqualified commentator, being neither a clinician providing early intervention services nor a psychosis researcher. However, I can at least claim impartiality, having no strong commitment to any position on this topic, and I have been an interested observer of the debates about early intervention for psychosis that have appeared in the psychiatric literature in recent years.
Despite their apparently opposing positions, there appear to be a number of areas of agreement between Rosenman & Anderson and McGorry. I begin by summarizing these, then move on to where there is disagreement. Finally, I come to my own conclusion about what role, if any, ‘prodromal diagnosis’ should have in clinical practice.
Areas of agreement
We cannot identify a prodrome, but we can predict higher risk
Rosenman & Anderson object to ‘prodromal diagnosis’ on the basis that an underlying process of psychosis cannot be shown to exist. McGorry also rejects ‘prodromal diagnosis’, although on a different basis. He sees this as a deterministic term that implies that the person in the prodrome will inevitably progress. Both parties agree that prediction of transition to psychosis is probabilistic and that most people in the psychosis-risk group will not progress.
Indeed, the broader literature on psychosis risk generally rejects the use of the term ‘prodrome’ [3,4]. This term is really only applicable retrospectively. It is possible to look back on the earliest symptoms of people with schizophrenia before positive psychotic symptoms develop and see patterns of difference from controls. However, this does not mean that these symptoms can be applied prospectively to identify individuals who will develop psychosis or schizophrenia. For this reason, most people researching the area use terms like ‘psychosis risk syndrome’ or ‘ultra-high risk’, implying increased risk, but not inevitability.
High risk patients have serious problems that require intervention
Rosenman & Anderson argue that clinicians should treat the patient's presenting problems, rather than try to treat some underlying disease process that they argue does not exist. McGorry also believes that high-risk individuals need treatment and he has been a prominent advocate in this area.
Again, there is broad agreement in the literature about treatment of high-risk patients. The Royal Australian and New Zealand College of Psychiatrists and several other professional organizations have produced relevant clinical practice guidelines [5,6]. Elements of treatment commonly advised are: monitoring, supportive therapy, symptom treatment, psycho-education, and cognitive behaviour therapy. None of the guidelines recommend use of antipsychotic medications for most patients. Such recommendations do not appear to be based on any assumption of an underlying disease process.
The focus of intervention should not be confined to the psychosis-risk symptoms
Rosenman & Anderson are concerned that a focus on psychosis risk will distract clinicians from the symptoms that brought the person to care. They point out that many of these patients will be concerned about symptoms other than those in psychosis-risk diagnostic criteria.
McGorry has recognized this issue with the clinical staging model that he and others have developed [7]. This model starts with a general stage of disabling distress (which they label as Stage 1a), with some people also developing additional symptoms indicating high risk of transition to psychosis (which they label as Stage 1b). If the person does make a transition to psychosis, this is labelled as Stage 2. It is notable that in this staging model, Stages 1a and 1b have a blurred boundary and they share a common level of psychological distress that requires intervention in its own right. Indeed, people in Stage 1 would commonly have a high-prevalence disorder by DSM or ICD criteria.
Progression to psychosis is far from inevitable
Rosenman & Anderson point out most people deemed to be at high risk for psychosis do not become psychotic and that the rates of transition have become lower in more recent years. McGorry agrees and argues that there is nothing deterministic about development of a psychotic disorder.
There is no known biological substrate for psychosis risk
Rosenman & Anderson argue that there is no underlying disease process in psychosis. They go further and argue that no such disease process will ever be found because psychosis is not a natural category waiting to be discovered, but rather a construction based on human needs and practices. They see psychosis as more analogous to the concept of ‘weed’ than to a classification of a plant species, or to the concept of ‘precious metal’ than to a chemical element.
McGorry agrees that no underlying disease process has been found. However, he does not dismiss this notion altogether and argues that we should investigate possible biomarkers and biological substrates of psychosis risk.
Areas of dispute
Could there be an underlying biological process?
Both parties agree that no underlying biological process has been discovered. Rosenman & Anderson dismiss this as a possibility, while McGorry seeks to investigate the issue further. His clinical staging model certainly posits that biological and endophenotypic markers exist [7]. For the psychosis high risk state, the model mentions the following as possibilities: niacin sensitivity, folate status, MRI and MRS changes, and HPA axis dysregulation. On this issue, only future data will tell who is right. A recent review of psychosis risk research concluded that the most promising findings on underlying mechanisms are in the areas of neuroimaging and neurocognition, but these are still inconclusive [3].
Does high risk diagnosis encourage or discourage early intervention?
The central disagreement is whether focusing on high risk status paradoxically delays early intervention. Rosenman & Anderson argue that a psychosis risk focus will lead the clinician away from intervening with the patients’ presenting problem, may lead to hesitation on the clinician's part because they are uncertain about the nature of the underlying disease, and it may convey negative expectations and demoralize the patient.
McGorry responds that the ultimate test of early intervention is whether it works. He points to research on duration of untreated psychosis, showing a correlation between a long duration and poorer outcome. He also cites evidence that reducing the delay to treatment of psychosis improves outcomes and that randomized trials of intervention with high risk patients show positive effects. He puts the onus on critics to justify late or delayed intervention as an alternative.
It is important to note, however, that Rosenman & Anderson are not questioning the treatment of psychosis or arguing for delayed intervention. I would characterize their position as advocating ‘on-time intervention’ (my term, not theirs). They are saying that the patient's problem should be treated when it occurs, but not pre-empted on the basis of a prediction that it may occur.
Other comments
Having compared and contrasted the respective views of Rosenman & Anderson and McGorry, I would like to add a few comments of my own.
Psychosis risk syndrome should be a research diagnosis not a clinical diagnosis
There has been a proposal to add an Attenuated Psychotic Symptoms Syndrome to DSM-5 [6]. On the other hand, there has been significant opposition to this proposal, with the chair of the DSM-IV Task Force describing it as ‘the most worrisome of all the suggestions made for DSM5’ [8]. The concern is that there is no specific diagnostic test and no safe treatment.
On the issue of treatment, McGorry points to the positive findings of randomized controlled trials of interventions to reduce risk of conversion to psychosis. A meta-analysis of five trials found an 11% conversion to psychosis after 1 year in the intervention groups, compared to 32% in control groups [9]. However, these gains were not maintained over 2–3 years of follow- up. It appears that intervention can delay, but perhaps not reduce the cumulative risk of psychosis. One solution would be to extend intervention. However, intervention is not without risk to the patients. Two of the trials used antipsychotic drugs, which are well known for the metabolic risks, but may also lead to brain tissue loss [10].
My conclusion is that this is a promising area of research, but is far from the point where it can be part of routine clinical practice. Research requires standard diagnostic criteria to ensure consistency across studies, but there is currently no justification for a new clinical diagnosis, particularly where patients already have high prevalence disorders meriting intervention.
The ultimate goal should be to give the patient a better life, not to prevent psychosis
More broadly, it could be argued that the real issue is not whether psychosis risk intervention prevents transition to psychosis, but whether it helps patients to have a better life. Aiming for a ‘better life’, involves assessing outcomes like functioning, disability, quality of life and recovery. I believe the most important question that needs to be answered is whether people who participate in psychosis-risk interventions have a better life over a long follow-up period compared to those who receive standard treatment, rather than what proportion of them make a transition to psychosis.
If effective low-risk interventions are found, they may be better implemented universally
An implicit assumption of psychosis risk research appears to be that if a successful intervention can be found to reduce transition to psychosis, then this will be implemented clinically, perhaps in specialist high risk clinics. However, I would argue that this depends on the nature of the intervention that is found to be successful and its costs and risks. If the successful intervention was cheap and very low risk, it might be best implemented universally. For example, there are two rare neurological disorders, Wernicke-Korsakoff syndrome and neural tube defects, which are prevented in Australia through supplementation of food staples with thiamine and folate respectively. Thus, universal intervention may even be applicable for low incidence disorders.
Perhaps the most promising intervention for reducing psychosis risk examined so far is omega-3 polyunsaturated fatty acid supplementation [11]. Let us imagine that the initial positive findings are replicated, that the side effects are miniscule and the cost of supplementation is low. The best strategy might be to try to increase omega-3 intake in the population as a whole or to increase it during the period of life when psychosis risk is greatest, rather than to have clinicians try to identify and treat psychosis risk syndrome.
Another reason to consider universal approaches is the ‘prevention paradox’, which was pointed out many years ago in relation to cardiovascular disease by the epidemiologist Geoffrey Rose [12]. This states that the majority of cases of a disease may come from the low and moderate risk section of the population rather than from the high risk section. Where the prevention paradox applies, applying preventive interventions only to the high risk section of the population will fail to prevent most cases. In the psychosis risk area, the concern has been principally with the rate of transition to psychosis in those that meet the high risk criteria. What is not known is what proportion of psychosis in the population will arise from the section of the population that fails to meet these criteria.
My conclusion is that investigating potential interventions is best carried out with high risk patients, ensuring an ‘enriched’ sample and better chances of being able to detect an effect. However, the method of researching the topic (a high risk sample) may not necessarily be the same as the best method of putting it into practice.
The potential role of psychosocial factors has been neglected
Rosenman & Anderson allude to the potential role of psychosocial factors with their comment that any behavioural change in the patient may be due to ‘accumulation of domestic discord or social isolation’ rather than to progression of an underlying disease. I believe they are right in pointing to the neglect by psychosis risk researchers of psychosocial risk factors (as distinct from psychosocial consequences of psychosis). It is striking that in a recent 42-page review of the area, there is no mention of psychosocial factors in psychosis risk [3]. This is in contrast to the growing literature on psychosocial risk factors for schizophrenia [13].
Conclusion
Psychosis risk is an important but controversial area for research and practice, which needs to be widely discussed. I would encourage readers to continue the discussion in the Letters section of the journal.
Footnotes
Acknowledgements