Abstract
This article focuses on three key questions relevant to the concept of secondary prevention in psychosis:
For how long (and for whom) does early intervention need to be sustained for long-term impact on the course of psychosis?
When and for whom can the early intervention ‘grip’ be relaxed?
How does this interact with vulnerability?
What kinds of interventions are needed to achieve long-term change?
Vulnerability and early intervention
Follow back and birth cohort studies indicate that the onset of positive symptoms is not the beginning of psychosis, but the end of the beginning of psychosis. This alone confirms that there exists an underlying vulnerability of which the positive symptoms are one expression.
Much has been written about the liability to schizophrenia, whether it is specific to schizophrenia or a multifactorial trait distributed throughout the population [1]. The outcome studies of schizophrenia show that it is heterogeneous in outcome, underpinned in part by heterogeneity in vulnerability. The low vulnerability end of the spectrum includes those 20%% or so who have one psychotic episode followed by complete recovery [2] and those with ‘spectrum disorder’ that do not find their way into services. This is important because those of low vulnerability may require less intensive or sustained treatment and, in early detection programmes, run the risk of being absorbed inappropriately into services.
The concept of vulnerability itself shows that early intervention can never be a ‘magic bullet’. This is because all our treatments for psychosis are essentially symptomatic. This is certainly true for the neuroleptics and probably also for cognitive behavioural therapy (CBT) [3]. It seems unlikely, therefore, that reducing the duration of untreated psychosis (DUP) to 0 or to –1 will eliminate vulnerability, although it may place an early ceiling on it and will probably serve to delay the onset. However, this remains an interesting empirical question. Our task would be much easier if we had an objective metric of ‘vulnerability’ rather than the proxy measures that we commonly use at present. The implications for primary and secondary prevention of psychosis are different if we consider separately those of both high and low vulnerability.
Low vulnerability
It seems entirely plausible that individuals of low vulnerability (e.g. those who have no major prodrome, low genetic risk and rapid onset) have psychoses that are potentially transient and linked to major stressors: for example, linked to the difficult transitional stage of life in late adolescence. In these individuals, prompt treatment, including a focus on the stressor itself, might well eliminate psychosis for a lifetime. In such a scenario, early intervention could indeed be regarded as a ‘magic bullet’.
High vulnerability
Early detection and intervention in those of high vulnerability will, at best, prevent the onset of positive symptoms. In the context of high vulnerability, these individuals will always be at high risk of a first episode developing: the risk of a ‘DUP’ effect will be ever present, as well as later putative episodes. Indeed, the hypothesised neurotoxicity of psychosis may be cumulative across episodes and not simply act at the first episode. Either way, the vulnerability will require continued management; but for how long?
Critical period concept
The critical period concept, which we introduced some time ago in this Journal [4], derives from observations made by Manfred Bleuler about the ‘plateauing’ of symptoms and disability in the course of psychosis. It has been presented and discussed elsewhere at length [5] but its essential propositions are as follows:
1. The course of psychosis is most stormy at the onset and early in its manifest course, plateauing thereafter
Prospective studies have shown that, for many, there is a rapid period of progression of psychosis prior to and in the 3–5 years following the first presentation. There is a well-documented prodromal period of untreated psychosis of varying duration linked to underachievement in psychosocial domains [6]. The evidence for the risk of early progression is strong; the risk of relapse, for example, is high within 2 years and nearly three-quarters of patients can expect to relapse within 5 years [7]. Bleuler's early studies, taken together with the recent high-quality prospective studies in Nottingham, United Kingdom, provide compelling evidence that the long-term type is predictable by year three (including, on average, the 12 months of DUP), with a stabilisation in the absolute level of morbidity [8]. This fact is probably not lost on those who are affected; suicide risk is particularly high during this early phase, especially following a relapse.
The therapeutic implication of this lies in its clarification of a time scale for early intervention. The first 3 years of illness (treated and untreated) offer a window of opportunity to prevent or limit this potential early and long-term decline. Intervention efforts after the plateau of morbidity has been reached will face a greater challenge than those implemented after the first episode, however, the extent to which the damage is irreversible following the achievement of the plateau is of crucial significance and is an interesting research question.
If the therapeutic implications of this are to be realised, then it must first be demonstrated that intervention can change the early course of psychosis, leading to a lower plateau of morbidity and changed course type. An interesting research question then arises: if the early intervention ‘grip’ is then relaxed, will the individual deteriorate to another (natural) plateau or, conversely, will the improvements require relatively little maintenance? This, in part, will depend on whether the interventions inhibit the development of ‘toxic’ influences which, in their absence, will follow their natural course (e.g. neurological changes that are the consequence of relapse) or, conversely, introduce ‘healthy’ influences that introduce a virtuous circle. There is likely to be a combination of these two processes and, again, this poses an interesting research question.
One study bearing upon this issue is reported by Linszen et al. in which a family intervention program was evaluated in first-episode psychosis with the aim of controlling early relapse [9]. Family intervention was compared with an individual intervention that included, in both groups, neuroleptic treatment, case management and psychoeducation. Family intervention was offered, therefore, in the context of a high quality of care. The rate of relapse over 15 months was contained in both groups to 16%%; however, when these patients left the trial (and the adolescent service programme) and were followed up between 17 and 55 months later, the rate of relapse escalated to 64%% in both groups. The median survival time (when 50%% of the cohort had relapsed) was 19 months. Linszen et al. noted that, after the study, all the people were referred to other services and the favourable early impact of their intensive programme, in spite of the short DUP of the group, disappeared rapidly. The sample was very young (mean age of 19.0 years) and mostly male (70%%), confirming that young males are a particularly high-risk group but who nevertheless had a short DUP. The ‘grip’ was relaxed early in this group and relapse proliferated. The ‘grip’ clearly needs to be maintained longer and, according to the authors, the only way to prevent poor outcome in schizophrenia seems to be ‘continuation of medication and case management, medication and stress management for a minimum period of 5 years, a period which approaches the “critical phase”’ [9].
This impressive study suggests that early outcome can be improved, but the issue of sustaining change is the major challenge. The critical period concept argues that at least 3 years will be needed before ‘relaxing the grip’ is even considered. Tom McGlashan has argued that interventions may only be effective as long as they are active [10]; the critical period concept proposes that if sufficient change can be achieved in all dimensions (clinical, social and occupational) then relaxing the ‘grip’ may, for some groups, be attempted. This is an important research question, which needs to be addressed.
2. The critical period witnesses the ontogeny of significant variables
Here, we argue that the biological, psychosocial and cognitive changes that are influential in the course of psychosis are not ‘givens’ but actively develop during this period. The possibility of biological toxicity was first raised by Wyatt [11]. Pretreatment exposure to psychosis has been suggested by Wyatt as a key factor, but also cumulative exposure to psychosis through relapse, treatment resistance and episodes of untreated psychosis following the first treatment may also contribute to this theoretical toxicity. The prevention and minimisation of relapse and the focus on early treatment resistance will be key therapeutic objectives here.
There is evidence that family relationships are also developing during this period. The metamorphosis of components of expressed emotion (EE) occurs in one direction (emotional over-involvement to criticism) and suggests the operation of a process consistent with recent concepts that EE should be viewed as a state not a trait characteristic [12]. The results of the Amsterdam study failed to find any additional benefits of behavioural family intervention in first episode psychosis in the context of an individual case management approach [9]. Therapeutically, the content of family intervention may need to change to reflect this flux of family relationships, with a key therapeutic objective being the prevention of negative attitudes in relationships. In this context, it is perhaps not surprising that the Amsterdam First Episode Family Intervention study failed to find any additional benefits above standard family intervention [9]. In fact, this study demonstrated negative effects of intervention in some families, particularly families characterised as low EE, underlying the operation of a different process; for example, those involving early adjustment patterns centred around loss and mourning [12].
3. The desynchrony between clinical and social functioning begins in early psychosis
While symptoms and social functioning are clearly not orthogonal, the prospective studies, including those in early psychosis, emphasise their desynchrony on cross-sectional evaluation. The studies of Biehl et al. [13] and Schubart et al. [14] emphasise continuity within each domain, such that the best predictor of social functioning after 5 years is an earlier measure of the same characteristic. From a therapeutic point of view, it should not be assumed that improvements in social and community functioning can be ‘bought’ by a focus on symptoms alone. Early intervention may best be conceived as a process involving three domains: symptoms, psychological and psychosocial functioning. For example, improvements in vocational functioning may enhance self-esteem and promote social engagement, which may reduce vulnerability to relapse.
Critical period and early intervention
The critical period concept presents clear, testable hypotheses about the impact of sustained early intervention in those of high vulnerability. These are:
That early intervention can improve the early course of psychosis and lead to a new plateau. That early intervention can prevent mental and social decline. That early intervention leads to better outcome than the same ‘quantum’ of intervention provided beyond the critical period. That following sustained intervention throughout the critical period, there will be no return to baseline if the intensity of intervention is relaxed.
Conclusion
Interventions that alter the trajectory of psychosis are at hand and improving the early outcome should not be too difficult, indeed the study of Linszen et al. shows that this can be achieved [9]. Sustaining change, however, is the main challenge. Relaxing the ‘grip’ too early, as Don Linszen suggested in the Amsterdam study, may lead to a proliferation of relapse. The critical period concept suggests that 3 years is the minimum period necessary for most patients of high vulnerability to sustain change before the ‘grip’ can, if at all, be relaxed. Whether, by effective early detection and intervention, we are preventing potential long-term difficulty or merely delaying it remains a fascinating question and one that is a core but, as yet, unanswered question.