What are the ‘early warning signs’ of psychotic relapse?

Investigations (e.g. Reference Herz and MelvilleHerz & Melville, 1980; Reference Birchwood, Smith and MacmillanBirchwood et al, 1989; Reference JørgensenJørgensen, 1998) have consistently determined that subtle changes in thought, affect and behaviour precede development of frank psychosis.

‘Dysphoric’ symptoms (depressed mood, withdrawal, sleep and appetite problems) are most commonly reported, while psychotic-like symptoms (for example, a sense of being laughed at or talked about) are less frequent. Furthermore, these symptoms generally occur in a predictable order, with non-psychotic phenomena occurring early in the illness, followed by increasing levels of emotional disturbance and, finally, by the development of frankly psychotic symptoms (Reference Docherty, Van Kammen and SirisDocherty et al, 1978). The progression occurs, most frequently, over a period of less than four weeks (Reference Birchwood, Smith and MacmillanBirchwood et al, 1989; Reference JørgensenJørgensen, 1998).

Although these symptoms have sometimes been referred to as the psychotic ‘prodrome’, they are more accurately conceptualised as ‘early warning signs’ of psychotic relapse, since the concept of a ‘prodrome’ (a term derived from the medical literature) implies a disease progression that cannot be interrupted. However, investigators have found that people with psychosis actively use coping strategies to intervene in the onset of psychosis (Reference McCandless-Glimcher, McKnight and HameraMcCandless-Glimcher et al, 1986). Furthermore, strictly speaking, ‘prodromal’ symptoms of psychosis include only those non-specific symptoms that may signal the onset of a variety of illnesses. However, attempts to predict the onset of psychosis from non-specific or dysphoric prodromal symptoms alone have yielded poor sensitivies and/or specificities (e.g. Reference Jolley, Hirsch and MorrisonJolley et al, 1990), but results have been more promising when low-level psychotic symptoms are included in the predictor variables.

Can psychotic relapse be predicted accurately from these early warning signs?

Prospective studies (Reference Subotnik and NeuchterleinSubotnik & Neuchterlein, 1988; Reference Birchwood, Smith and MacmillanBirchwood et al, 1989; Reference JørgensenJørgensen, 1998) have shown that psychotic relapse can be predicted with a sensitivity of 50–79% and a specificity of 75–81% when standardised measures of ‘neurotic’ or ‘dysphoric’ symptoms are combined with those of low-level psychotic symptoms and ratings are conducted at least fortnightly.

However, there is considerable variability between individuals in the nature and timing of their early warning signs (Reference Birchwood, Smith and MacmillanBirchwood et al, 1989), and prediction of relapse is more accurate if changes in early warning scores are evaluated against individuals’ own baseline scores rather than compared with those of other patients (Reference Subotnik and NeuchterleinSubotnik & Neuchterlein, 1988; Reference JørgensenJørgensen, 1998). Thus, to be clinically useful, methods of identifying early warning signs of psychotic relapse must take into account this individual variation.

For these reasons, research attention has recently been directed towards identifying and managing each patient's ‘relapse signature’ (Reference BirchwoodBirchwood, 1995): his or her unique pattern of early warning signs most likely to indicate impending psychotic relapse. Later in this article, we will present a methodology used in our clinical practice for this purpose.

Can patients identify their own early warning signs?

A large percentage of people with schizophrenia and their relatives are aware of these early signs of impending relapse (Reference Herz and MelvilleHerz & Melville, 1980). One study found that 63% of patients maintained insight into their deteriorating mental state until the day of their relapse (Reference Heinrichs, Cohen and CarpenterHeinrichs et al, 1985). Jørgensen (1998) also found that patient self-reports of early warning signs predicted relapse with a sensitivity and a specificity almost equal to those derived from psychiatrists.

Cognitive–behavioural interventions

The ‘stress–vulnerability model’ (Reference Zubin and SpringZubin & Spring, 1977) views the symptoms of schizophrenia as the result of environmental stressors acting on the vulnerable individual, and predicts that a reduction in stress or the acquisition of stress management skills should decrease the chance of psychotic relapse. The association of stressful life events (Reference Hirsch, Bowen and EnamiHirsch et al, 1996) and stressful home environments (Reference Kuipers and BebbingtonKuipers & Bebbington, 1988) with relapse among people with schizophrenia adds weight to this model and its predictions.

There is evidence that even after the onset of early warning signs, stress management skills may be helpful in preventing psychotic relapse. For example, McCandless-Glimcher et al (1986) found that many patients with schizophrenia use cognitive–behavioural techniques to deal with the early warning signs of relapse without being formally instructed to do so. Furthermore, Hogarty et al (1997) found that treatment with an individualised and graded approach to stress management, particularly focusing on the identification and management of affective dysregulation preceding relapse (‘personal therapy’), was associated with a significant overall effect in delaying adverse events (including psychotic or affective relapse or treatment-related termination) relative to supportive therapy, for patients living with their families.

A second strand of cognitive therapy focuses on the meanings with which patients invest their symptoms and the evidence that they hold for their beliefs. Therapeutic techniques within this framework have been evaluated in frank psychosis and found to result in statistically significantly greater improvements in psychotic symptoms relative to control conditions (Reference Garety, Kuipers and FowlerGarety et al, 1994; Reference Drury, Birchwood and CochraneDrury et al, 1996a ). Furthermore, depending on the definition of recovery used, cognitive therapy has been found to reduce time of recovery from psychosis by 25–50% (Drury et al, 1996b). These successes have encouraged an extension of the theoretical concepts underlying these techniques to the early warning sign period. Birchwood (1995), for example, proposes that early warning signs may represent symptoms intrinsic to the illness combined with a psychological response that centres on a search for meaning and control, which may, in turn, contribute to whether the relapse is arrested or accelerated. Psychological responses involving denial or excessive fear of relapse are hypothesised to be internal stressors, which, in themselves, increase the probability of relapse. Cognitive therapy techniques that challenge these dysfunctional beliefs may thus prevent the escalation of early warning signs to frank psychosis.

Medication

Intermittent medication initiated only at detection of early warning signs has been shown to be inferior to continuous medication in preventing psychotic relapse and is not generally recommended (Reference Carpenter, Hanlon and SummerfeltCarpenter et al, 1990; Reference Gaebel, Frick and KöpckeGaebel et al, 1993). However, medication initiated on the development of early warning signs in combination with maintenance medication has been shown to reduce psychotic relapse rates to 12–23% over two years (Reference Marder, van Putten and MintzMarder et al, 1984 Reference Marder, van Putten and MintzMarder et al, 1987; Reference Jolley, Hirsch and MorrisonJolley et al, 1990; Reference Gaebel, Frick and KöpckeGaebel et al, 1993). Furthermore, in most cases this strategy has allowed a low maintenance dose of medication to be successfully used. Marder et al (1994), for example, demonstrated a significant reduction in the risk of relapse and time spent in psychosis from the second year of treatment with a combination of low-dose maintenance medication plus medication targeted at early warning signs, when compared with treatment with low-dose maintenance medication only.

Engagement and education

The identification of the relapse signature and drill forms an ideal medium through which to establish common ground with the patient and to acknowledge his or her point of view.

Initial sessions focus on understanding the individual's attitude towards his or her illness, especially his or her perception of the risk and controllability of relapse. Beliefs that may exacerbate the relapse process, for example, poor perceived control over the illness causing the individual to panic or fail to act on the occurrence of early signs, are identified.

It is suggested to the patient and his or her family that fear of relapse may be coped with through skill-learning, and the analogy of a ‘safety net’ is used to describe these skills. Such a discussion draws upon the positive steps already being taken by the individual to remain well, and occurs in the context of general psychoeducation concerning preventable risk factors of psychotic relapse (e.g. discontinuation of medication and illicit drug use). An example of a completed relapse prevention sheet is shown (see Boxes 2a , b ).

Identification of the relapse signature

The aim of the next part of the process is to construct a hypothesis about the individualised relapse signature: that is, a set of general and idiosyncratic symptoms, occurring in a specific order, over a particular time period, that serve as early warning signs of impending psychotic relapse.

Patients are first introduced to examples of early warning signs of psychotic relapse. They are then encouraged to review, either alone or with the support of their keyworker or family, any noticeable changes in their thoughts, perceptions, feelings and behaviours leading up to their most recent episode of illness, as well as any events that they think may have triggered these.

Two structured exercises are then used to expand and order this set of early warning signs.

Development of a relapse drill

Following the identification of the relapse signature, patients are supported in constructing a three-stage action plan known as a ‘relapse drill’.

Staging is an essential feature of the relapse drill. It follows directly from the early warning signs, which are stratified into three levels, from those occurring earliest in the relapse signature to those occurring immediately prior to the psychotic relapse. In general, the earliest early warning signs in the relapse signature tend to be non-specific symptoms, with low power to predict psychotic relapse. Interventions with potential risks (e.g. increases in antipsychotic medication) are generally used after the relapse signature has clearly progressed towards potential psychotic relapse.

The drill is developed collaboratively and focuses on patient strengths, carers and service resources. Past coping strategies and therapeutic interventions that have been found to be helpful in preventing relapse are reviewed collaboratively with the individual and incorporated into the drill. Specific early warning signs may suggest new approaches to offer further protection against relapse. For example, anxiety, dysphoria and other affective changes may respond to techniques incorporating stress management. Similarly, patients suffering from low-level psychotic phenomena may benefit from techniques designed to challenge delusional and dysfunctional thinking drawn from the cognitive therapy literature.

At each stage, the relapse drill considers three areas for intervention.

Rehearsal and monitoring

Having identified an individual's relapse signature and drill, the patient and relevant involved carers are provided with their own copies of the relapse prevention sheet and monitoring is outlined as a shared responsibility between the individual, carers and mental health services. To enhance effective use, the relapse drill is rehearsed using personalised scenarios and role-plays concerning the patient's response if he or she should detect early warning signs. Hypothetical situations are used to discuss any difficulties that might arise (for example, denial or panic responses) and how to deal with these.

Clarification of the relapse signature and relapse drill

Clarifying the relapse signature and refining the relapse drill are other important areas of monitoring. Individuals are encouraged to replace existing coping strategies, forms of support and service interventions with more effective ones learned from ongoing therapy or experience. In this way, impending or actual relapse is used as a positive opportunity to refine the relapse signature and improve the relapse drill, thus increasing control over the illness.

Lack of ‘insight’

It may still be possible to construct a relapse signature among individuals who are unable to conceptualise their past psychotic experiences as unreal. Such was the case with a 24-year-old man, who, following his psychotic episode continued to believe that during his illness his musical compositions had been stolen by a famous rock band. He was, nevertheless, able to construct a relapse signature identifying problems of increasing dysphoria and self-neglect leading up to the development of the delusion without conceding the belief itself to be delusional. On some occasions, individuals may themselves refuse involvement in relapse work but consent to involvement of their family.

Similarly, other patients may concede that their psychotic experiences are not real, but may attribute them to factors other than ‘illness’. This is especially the case among individuals suffering from their first episode of psychosis, where denial of illness may serve the function of preserving self-esteem and should not be overzealously challenged. Individuals may still be able to identify early warning signs while attributing their problems to a multiplicity of causes other than illness – such as alcohol, interpersonal conflicts or spiritual experiences. Acceptable interventions might include increased support from family and services and, with the goal of normalising sleep or preventing re-hospitalisation, temporary increases in medication may also be accepted.

More difficult to solve is the problem of early loss of insight. In our clinical experience there is a subgroup of people who, while having ‘past insight’ (Reference McGorry and McConvilleMcGorry & McConville, 1999) and an ability to construct a relapse signature retrospectively, lose ‘present insight’ early in the relapse process. Although families may be involved in the relapse drill in this group, we have also employed prospective monitoring using a standardised measure of early warning signs (Reference Birchwood, Smith and MacmillanBirchwood et al, 1989). Completed fortnightly by the patient and a family member, it has been used to help teach the individual to discriminate the changed perceptual, cognitive or affective processes that constitute the relapse signature.

‘Sealing over’

A concept related to, but separate from, insight is that of recovery style. A recovery style characterised by ‘integration’ is one in which the individual is aware of the continuity of his or her mental activity before, during and after the psychotic experience, assumes responsibility for his or her psychotic productions, is curious about the experience and has flexible ideas about recovery (Reference McGlashan, Levy and CarpenterMcGlashan et al, 1975). On the other hand, an individual who ‘seals over’ tends to isolate the psychotic experience, views it as alien and seeks to encapsulate it. Individuals whose recovery style is that of sealing over may find the ‘early warning signs’ approach, with its focus on the close examination of the illness, anxiety-provoking. For them, it may be necessary to temporarily suspend the process of constructing a relapse signature and to attend to the establishment of a secure therapeutic alliance with mental health services through working on shared goals such as vocational and social aspirations.

Lack of syndrome stability

Particularly in the early phases of psychotic illness, there is evidence of a considerable lack of diagnostic clarity and stability, which may be increased by factors such as comorbid substance misuse and the individual's psychological reaction to the illness (Reference McGorryMcGorry, 1994). Thus, a similar instability in the clinical presentation of the early warning signs of psychotic relapse may be expected. Fortunately, each relapse may be used to clarify the relapse signature and to refine the drill. However, there is an unavoidable paradox inherent in this work: increasing clarity of the relapse signature and, therefore, potential increased control of the process of relapse, only emerges with the additional information gained through each psychotic relapse. This may be resolved by considering such information as crucial to reducing the duration of relapse, since untreated psychotic symptoms are linked to speed of recovery and subsequent relapse (Reference Drury, Birchwood and CochraneDrury et al, 1996b ).