Early intervention in psychosis

Examining a revolution in psychiatry

What has been the biggest idea in psychiatry of the last 30 years? Since the development of SSRIs and atypical antipsychotics, there haven’t been any real pharmacological advances. However, there have been massive changes in the implementation of psychiatric services.

A major transformation has been early intervention in psychosis. This approach involves treating psychotic disorders at their earliest stages in an effort to mitigate or even prevent long-term negative outcomes.

The early intervention model is a sea change in the delivery of care for people with psychosis. How does this approach hold up? In this post, I’ll try to take a balanced but critical view.

Disclaimers

Before reading any piece, it can be useful to know the author’s stance, background and potential biases. Here are mine:

• Most of my research has been conducted within the early intervention in psychosis paradigm

• I have worked with some leading figures in the field, both clinically and in research

• My viewpoint is inevitably UK-centric, and more likely London-centric – my local NHS trust was one of the first to adopt the early intervention model

I will try to bear these potential biases in mind, as should you! I will try to adopt a ‘Scout Mindset’ rather than defending a particular standpoint.

Nothing here will be new, the debate over early intervention is longstanding - opposing viewpoints have been published in the British Journal of Psychiatry and the BMJ.

At the heart the matter, isn’t a question about whether we should give good quality clinical care to people early in a psychotic illness. I think everyone agrees these are serious disorders that merit effective, humane, well-funded services. Instead, the debate is whether we should ‘front-load’ funding for severe mental illness to those at the earliest stages.

The concept of early intervention

Early intervention as a general strategy is intuitively appealing. A stitch in time saves nine. Prevention is better than cure. Forewarned is forearmed.

There is precedent from other fields of medicine. Think primary prevention of cardiovascular disease by addressing obesity, cholesterol, and high blood pressure. Or catching cancer early through programmes like bowel screening.

Perhaps even more relevant to psychosis is secondary prevention. The best example I have is addressing risk factors to prevent an ischaemic stroke following a transient ischaemic attack.

Psychotic disorders, like schizophrenia, should be particularly suitable for intervening early. A well described aspect of schizophrenia is the prodromal stage, often characterised by subtle changes in the perception of the world and a decline in functioning. A preprint meta-analysis suggests 80% of patients with psychosis go through such a prodrome.

If you believe that early intervention is a good strategy for psychosis (which I do), how should this be implemented practically? Broadly, there are two distinct but complementary approaches: (i) identifying those at risk for psychosis, and (ii) intervening early after a first episode of psychosis.

Early identification before the onset of psychosis

Various terminologies are used for this approach – At Risk Mental State, Ultra High Risk, Clinical High Risk for Psychosis. They all refer to identifying the subtle signs that may represent early changes before the onset of psychosis. The term prodromal tends to be avoided, as most people deemed at risk do not go on to develop psychosis - the prodrome can only be applied in retrospect.

The original ‘at risk’ service, the PACE clinic, was set up in Melbourne in the mid-1990s by psychiatrists such as Patrick McGorry and Alison Yung. Their original paper is worth reading to understand the philosophy.

They make a comparison to measles - a disease with a non-specific prodromal phase (cold-like symptoms), which then goes own to manifest with a characteristic rash. Careful clinical observation may reveal ‘Koplik's spots’ – white lesions in the mouth which appear 2-3 days before the measles rash. They are pathognomonic of measles, meaning their presence is definitive of the diagnosis.

The PACE clinic, which became the model for UK services, assesses young people (age between 14-30) suspected of being at risk of psychosis. If judged to be ‘at risk’, they are offered social support, psychological therapy, and reviews by psychiatrists as required. Individuals are generally followed up for two years. This type of service is now part of NICE guidelines.

Young people can meet different criteria that are thought to confer increased risk of psychosis:

1.     Attenuated psychotic symptoms. These might be experiences such as voice-hearing that lack the intensity and vividness of true hallucinations, or unusual beliefs that can be questioned (unlike true delusions).

2.     Brief psychotic episodes. These are episodes of frank psychosis which spontaneously remit within one week without antipsychotic medication.

3.     Decline in functioning in people with a family history of psychosis.

Of the first 400 patients deemed ‘at risk’ from the PACE clinic, more than a third developed a psychotic disorder, over 10 years of follow-up. Most of the transitions to psychosis took place within the first two years of follow-up.

Since the inception of ‘at risk’ clinics, there have been two major goals in psychosis research: (i) identify which individuals at risk are going to be the ones who develop psychosis (find the equivalent of Koplik’s spots), (ii) develop preventative treatments for psychosis.

Early intervention following a first psychotic episode

What happens after a young person has a full-blown psychotic episode? The early intervention model proposes quick assessment and appropriate treatment with the goal of reducing the duration of untreated psychosis.

Untreated psychosis was once thought to represent a neurotoxic state. This hypothesis is not supported by evidence, but a more general association of untreated psychosis with poor outcome remains. Perhaps by letting psychotic beliefs take hold within a person’s mind, or by wrecking their support networks. Reducing the duration of untreated psychosis is a priority.

Psychiatric services within the UK have been restructured to meet this priority with the development of early intervention in psychosis community teams. The NHS waiting time standard states that those experiencing a first psychotic episode should have initiated treatment within two weeks of referral. Following this, patients will be supported for a set period, usually three years. After that, the patient may be discharged back to their GP or referred onward to longer-term community mental health teams.

Restructuring services for early intervention was a massive undertaking but seemed like a good investment. Intervening at a stage when many patients still have a cohesive family, social and occupation structure might return extra gains. In a classic paper, Max Birchwood argued that there is a ‘critical period’ in the first 2-3 years following psychosis onset in which interventions would have a disproportionate benefit on illness course.

To lay my cards on the table, I think early intervention in psychosis is one of the biggest ideas in modern psychiatry. I am broadly in agreement that intervening early might be the best way to prevent the negative consequences of chronic illness. My hunch is that any new effective treatment for psychosis is likely to be most beneficial at the earliest stage of illness. Finally, preventing onset of frank psychosis at a pre-clinical stage would be a transformative achievement.

However some issues have come to light since the widespread adoption of early intervention, which challenge the model to varying degrees.

Challenges to the early intervention model

1. Non-specificity of ‘at risk’ symptoms

As I discussed in a previous post, psychotic symptoms are often found in other diagnoses or even in healthy individuals. The inclusion criteria for someone to be ‘at risk’ of psychosis may instead represent the expression of another mental disorder, or physiological process.

None of the symptoms representing the ‘at risk state’ have specificity for psychosis. Almost 3/4 of those meeting at risk criteria have another co-morbid psychiatric diagnosis, most commonly depression. It is possible that by focusing solely on symptoms as a pathway to developing psychosis, we neglect more treatable common mental disorders which are presenting with some psychotic-like experiences.

2. Differences in transition to psychosis between ‘at risk’ groups

There are marked differences in rates of developing psychosis among the groups deemed ‘at risk’. The highest rate is in those who have transient, self-limiting psychotic episodes – more than half will have converted to a psychotic disorder within five years. Those with disorganised or dangerous features, have an even higher rate of conversion to psychosis, almost 90%.

But is having a brief psychotic episode really a prodrome, or just the first episode of psychosis? Is a psychotic episode lasting less than one week a separate category from one lasting a month?

Attenuated psychotic symptoms (which most resemble a psychosis prodrome) have a lower conversion rate. In general population samples reporting attenuated symptoms, the five year rate of transition to psychosis is 1%, which is roughly the same as the lifetime prevalence rate for psychosis.

3. Declining transition rates

Over time, the rate of transition to psychosis has declined from 40% in initial samples down to 15% at one year in more recent samples. One of the largest cohort studies in North America has a two year transition rate of 12%.

The reasons behind this decline are not fully understood, but one theory is a dilution effect, with a larger proportion of individuals at less risk being included in these clinics.

Lower transition rates threaten the validity of the ‘at risk’ concept. Although the incidence of new onset psychosis is still much higher than the general population (roughly 0.05% per year), the vast majority of those categorised as ‘at risk’ for psychosis, will not develop a psychotic disorder (approaching 90% in some samples).

What’s more, lower transition rates make identifying biomarkers of psychosis and treatments to prevent this incredibly difficult. As I discussed in a previous post about suicide prediction, the predictive value of a test is exquisitely sensitive to the baseline rate of the outcome. Finding a treatment that reduces the rate of conversion to psychosis has to similarly contend with preventing an outcome that is relatively uncommon.

Think of the effective preventive treatments we have for cardiovascular disease, such as statins. These are not generally recommended unless the risk of a heart attack or stroke is more than 10% in the next 10 years. For a treatment to be given to prevent an uncommon outcome, the treatment must be well-tolerated, acceptable to patients, and have a completely benign side-effect profile.

Such treatments for psychosis are simply not available at present. Even if they were, could you convince a young person to take it over a period of years to prevent an outcome that 80-90% of them won’t experience in any case?

4. Failure to find Koplik’s spots for psychosis

A major goal in psychosis research has been to find a biomarker to predict psychosis onset in those at clinical high risk. Much of this has used neuroimaging. (I have previously discussed the lack of any robust neuroimaging markers for psychosis.)

The initial conceptualisation of identifying individuals at a putatively prodromal stage of psychosis, was that we would find the equivalent of Koplik’s spots – markers of impending psychosis in those at elevated risk.

Despite intensive research efforts and sophisticated analyses, we are still as far as ever from using a brain scan or any biological measure to predict the risk of transition to psychosis in individuals presenting with increased risk.

5. No specific intervention for preventing psychosis

Relating to declining transition rates, we have no preventative treatment for psychosis (or at best, no evidence that any treatment is any worse or better than any others). This includes both pharmacology, psychological therapy, or general support.

Leaving aside a lack of preventative treatment (as discussed, this is a difficult goal), there isn’t even a specific treatment that is effective for reducing attenuated symptoms of psychosis. Thus, at present young people are being identified at high risk of psychosis when no clear preventative or effective symptomatic treatment is available.

6. Small proportion of psychosis patients who come though ‘at risk’ clinics

Imagine the holy grail of early intervention has been obtained – we are able to prevent psychosis onset. For this, you need to imagine there is a high transition rate to psychosis in at risk samples – say the 30-40% identified in original studies. Imagine we have a highly predictive biomarker (maybe a brain scan) of psychosis onset. Also imagine we have a treatment that is acceptable, free of side-effects, and effective in preventing development of psychosis.

Would this scenario reduce the overall burden of psychotic illness? Probably not by much. In South London, which has some of the most established ‘at risk’ clinics in the UK, a study led by Olesya Ajnakina determined the proportion of people with first episode psychosis who had previously been in contact with ‘at risk’ services. This was less than 5%, meaning more than 95% of first episode cases were not detected prior to the first episode.

Although the majority of people with psychotic illnesses have a prodromal phase, the proportion who can be detected by mental health services during this prodromal phase is much lower. And in increasing the number of people taken on by at risk clinics in an attempt to find more cases prior to their first episode will inevitably cause a dilution effect, lowering the rates of transition further.

7. The negative effect of duration of untreated psychosis may be a measurement artefact

The notion that active psychosis is inherently neurotoxic lacks credible evidence. However, it still makes sense that the duration of untreated psychosis is important and treatment should be commenced quickly, to avoid the social and functional negative consequence of someone being acutely psychotic.

Recent research though, casts doubt on the association between duration of untreated psychosis and negative outcomes. A study by Katherine Jonas and colleagues suggests that patients with both short and long durations of untreated psychosis had similar declines in functioning. Crucially though, those with longer duration of psychosis had more deterioration prior to their first admission, whereas those with shorter duration of psychosis saw these declines after their first admission.

In other words, those with longer duration of psychosis were further along their illness course than those with shorter duration - the supposed negative effect of untreated psychosis was instead a measurement artefact.

If duration of psychosis is not as crucial as previously argued, we need to question whether this metric should be centred in guidelines.

8. Lack of long-term evidence for early intervention versus treatment as usual

Ultimately, whether early intervention services are effective should be an empirical question, answerable by a randomised controlled trial. A meta-analysis of ten such trials found early intervention was superior to ‘treatment as usual’ across a variety of outcomes in the short-term.

Longer-term evidence of benefit is scarce though. A Cochrane review found evidence for early intervention during the treatment phase but not after treatment had ended. Some of the initial trials are now reporting long-term outcomes. One, based in Denmark, found no difference between early intervention and treatment as usual at 20 years (long after the treatment period had ended). In fact, gains from early intervention were fading at five year follow-up and there were no significant differences from ten years.

Obviously, sustained effects many years after a complex intervention has ended is a high bar. However, a lack of superiority over treatment as usual does contradict the ‘critical period hypothesis’ put forward by some proponents of early intervention.

9. Allocation of resources away from chronic severe mental illness

Universal health systems like the NHS have limited, finite resources. Inevitably the funding for early intervention has to be a reallocation of other resources. In this case, from generic community mental health teams whose caseload is made up largely of patients with chronic psychotic disorders. By definition, those who require ongoing longer-term community psychiatric care are those with illnesses at the more severe end of the spectrum.

Is early intervention diverting funding away from these patients towards those at the earliest stages of illness?

10. Continuity of care

One common complaint about the structure of UK psychiatric services is a lack of continuity of care. It is plausible for patients to initially be seen by crisis services (such as home treatment team), be transferred to an early intervention in psychosis team for three years (with a different inpatient team for each hospital admission) and then to be transferred to a generic community team.

Early intervention is, by definition, time-limited, while the majority of those suffering a first episode of psychosis will go on to have a relapsing, remitting illness. Psychotic disorders, like schizophrenia, sadly tend to be life-long conditions. By time-limiting the first phase of care in patient’s illness, is early intervention contributing to a lack of continuity of care?

What does the future hold?

Obviously I’m not the first person to notice these challenges, many of the above points have been made by leading researchers in the field. The original pioneers of early intervention in Melbourne have adapted into a more comprehensive ‘youth mental health service’, not restricted to those at risk of psychosis.

Psychiatric services will continue to evolve over time. Whether early intervention in psychosis will answer the challenges above, evolve into something else, or revert back to generic mental health services - time will tell.