Determining outcome in FEP

Older studies about the course and outcome in FEP are dominated by the Kraepelinean view that non-affective psychosis is a progressive chronic condition with enduring deficits whereas affective/bipolar psychosis is episodic. However, in reality, there is considerable overlap between the two and the initial years of non-affective psychosis are characterised by remissions and relapses. ^{Reference Craddock and Owen3–Reference Robinson, Woerner, Alvir, Bilder, Goldman and Geisler5} Furthermore, in recent years, this simple dichotomous view of outcome in FEP is being challenged on several fronts.

First, the clinical staging model of psychosis adopts the approach taken in cancer staging. ^{Reference McGorry, Hickie, Yung, Pantelis, Henry and Jackson6} Illness progression is measured in steps or stages. Each stage has its own relative risk of progressing to the next stage within a defined period. Outcome is measured by (a) the speed and (b) likelihood one stage will progress to the next. Stage one is the pre-psychotic clinical stage of ultra-high risk. Stage two is the first acute episode, stage three is the phase of relapses with remissions, and stage four is the chronic treatment-refractory unremitting stage. Studies suggest that 18% of individuals progress from stage one to stage two within 6 months (increasing to 36% within 3 years). ^{Reference Fusar-Poli, Bonoldi, Yung, Borgwardt, Kempton and Valmaggia7} Likewise, only 6% progress from stage two to stage four within the first year of treatment (increasing to 23% within 10 years). ^{Reference Edwards, Maude, McGorry, Harrigan and Cocks8,Reference Demjaha, Lappin, Stahl, Patel, MacCabe and Howes9} Progression at each stage may not necessarily depend on the same factors and the best treatments may be stage specific. ^{Reference Fusar-Poli, McGorry and Kane10}

Second, rather than viewing each psychotic disorder as distinctly separate conditions with their own individual trajectories and outcomes, psychosis is instead seen as a far more dynamic multidimensional process, the end result of an accumulation of multiple interacting risks and protective factors over time. It is analogous to the build-up of factors with age that ultimately lead to an event such as a myocardial infarction, such as the cumulative effects of genetics, diet, weight, cholesterol, smoking, exercise and stress. Simply categorising a psychotic episode by one of these factors alone, such as ‘cannabis psychosis’, would be like categorising patients who have heart attacks into ‘smoking myocardial infarctions’ or ‘non-smoking myocardial infarctions’, treating them differently, and assuming very different outcomes. There are many risk factors for psychosis and each patient is likely to have their own unique constellation. It is this unique combination that is likely to determine individuals' outcomes and should be the focus of planning when designing individual care plans.

Third, psychotic symptoms occur in many mental illnesses not just schizophrenia and bipolar disorders. Kelleher et al highlight this very effectively in a large study revealing the high prevalence of hallucinations in as many as 24% of patients with common mental disorders such as anxiety disorders. ^{Reference Kelleher and DeVylder11} Are psychotic symptoms therefore an integral part of these conditions, or an indicator of severity – occurring only at the extremes of these acute episodes in the same way as seen in manic or depressive episodes of bipolar disorder? What relevance have psychotic symptoms then to outcomes in these conditions? Are they only seen in individuals with additional risk factors for psychosis? Do they warrant the same interventions as traditional episodes of psychosis? Are they followed by the same deficits and outcomes? Are they a marker of poor prognosis?

Finally, there is no clear boundary between more normative anomalous psychotic-like experiences (PLEs) and the clinical syndrome of psychotic disorder. Indeed, PLEs are common in the general population affecting 10–17% of adults. ^{Reference Johns and Van Os12} This raises the question of at what stage do PLEs become a psychotic disorder – when they become distressing or if they persist beyond a set period of time? The prevalence of PLEs poses a challenge to how we define remission in psychosis. One can have recovery without remission. Indeed, a certain degree of paranoia and inner ‘voices’ may be an indication of healthy functioning and creativity. It becomes rather subjective whether they are defined as appropriate or pathological to a person's social environment.

Measuring outcome in psychosis

Are we overly reliant on psychosis measures in determining outcome? Such measures as ‘response’, ‘remission’, ‘relapse’, ‘refractory psychosis’, although helpful, only portray part of the picture. Equally important is monitoring the course of comorbidities that may not necessarily follow the same trajectory of psychotic symptoms, for example post-psychotic depression, suicidality, cognitive deficits and negative symptoms, unwanted treatment effects, incapacitating psychological reactions such as self-stigmatisation and, finally, complicating physical illnesses/mortality. ‘Functional recovery’ may give a broader measure of outcome but it tends to lag behind symptomatic remission by months if not years. This may be a reflection of secondary comorbidities experienced after a psychosis and the practical time it takes to establish one's role in society (after months/years of dislocation) rather than one's true functional capacity.

The internationally agreed definitions of ‘remission’ and ‘recovery’ may be very applicable to chronic illnesses but not after the first episode. ^{Reference Bebbington, Craig, Garety, Fowler, Dunn and Colbert13} Relying on achieving a minimum of 6 months of remission and 2 years of recovery will not only fail to capture the episodic cyclical nature of these illnesses but also how each repeated episode/relapse reduces the likelihood of remission and recovery. A shorter minimum duration, such as 1 month of remission and 6 months of recovery, is far more likely to capture the true extent of remission and recovery from the first episode.

Predicting outcomes and prognosis

Are we any closer to predicting outcome in FEP? The meta-analysis by Lally et al ^{Reference Lally, Ajnakina, Stubbs, Cullinane, Murphy and Gaughran1} in this issue provides a robust evidence base of overall outcomes for patients with FEP. It also highlights very intriguing differences in outcomes in different continents and changes over the past few decades. But it cannot make distinctions at an individual level. Ideally, what we and our patients need now is an evidence base that can inform us better at an individual level about how different treatment and lifestyle choices (e.g. whether to stop using cannabis or take medication for another year) have an impact on outcomes. A ‘decision tree analysis’ is one way of distinguishing the different outcomes depending on which ‘branch’ one choses. It has been successfully modelled in economic evaluations of early intervention services. ^{Reference McCrone, Dhanasiri, Patel, Knapp and Lawton-Smith14} However, to capture the complexity and multiplicity of factors that determine diagnosis and outcome in FEP one might resort to a ‘decision tree forest’ model with each tree representing a particular domain of influence such as genetic, drugs, traumas, treatments, physical health. The recent availability of ‘big data’ such as centralised electronic health records makes the populating and testing of such complex models feasible. These mathematical modelling techniques could be used not only to map the factors involved in the evolution of a psychosis (i.e. the roots and origins of the psychosis) but also to predict a variety of outcomes depending on which option/branch is chosen.

Up to now, one of the problems in making these predictions is that we do not have a baseline for the natural history and course of untreated psychotic disorders. There are few outcome studies of untreated patients and even these have had some form of intervention without medication. What some of these studies show is that a surprisingly large minority of patients with FEP who consistently refuse antipsychotic medication can still achieve remission (41%) and a functional recovery (33%). ^{Reference Conus, Cotton, Francey, Donoghue, Schimmelmann and McGorry15}

Recovery and outcome in FEP

Recovery is more than simple remission of psychotic symptoms and return to premorbid functioning. It is not just a passive response to treatment. Nor is it just understanding one's prognosis and engaging with the treatments recommended. Recovery is a far more complex process. It demands change and adaptation. It requires tackling the antecedents that triggered the psychosis in the first place. If psychosis is the natural end result of an accumulating array of risk factors over time (genetics, early neurodevelopmental insults, childhood traumas, drugs in adolescence and stress), then addressing these issues is an important part of long-term recovery and relapse prevention. Returning to the same set of circumstances and lifestyle that originally precipitated the psychotic episode may actually be the least healthy outcome.

Recovery includes emotional and psychological dimensions, coming to terms with the shock of losing one's mind, the embarrassment, reconnecting with reality, letting go the fantasies, making sense of what happened, getting one's life back on track, one's routine, re-establishing one's goals, reputation, self-esteem and social network. It may not necessarily result in one returning to one's previous developmental trajectory as what is required is adaptation to a new set of circumstances. A more complete recovery requires understanding, learning from, integrating, assimilating and coming to terms with the whole experience in a way that hopefully resolves some of the predisposing factors that may have contributed to the episode in the first place and reduces the risk of future episodes. Influencing and measuring these outcomes is the real challenge.