LITERATURE REVIEW
2.1 The process of psychotic relapse
In this section, we will, by way of introduction, discuss the process of psychotic relapse and factors that influence it.
The vulnerability-stress-coping model is a commonly used theoretical framework for the research into, and the treatment of, schizophrenia (Zubin & Spring 1977; Zubin et al. 1992; Nuechterlein & Dawson 1984; Nuechterlein et al. 1992; Nuechterlein et al. 1994).
This model shows that various factors play a role in the development of a psychotic relapse. One factor is the vulnerability of the schizophrenic patient. This vulnerability is largely genetically determined and is present as a constant factor. Second, the amount of stress to which the patient is exposed is important. This stress can be caused by the patient himself or herself (for example, by means of all kinds of stress-inducing thoughts) or by environmental factors (for example, hostility on the part of people in his or her surroundings). A third factor are the coping and problem-solving skills of the patient. Most patients with schizophrenia do not submit to their situation passively but search for ways to deal with the aggravating circumstances. They also do this in the phase preceding the psychosis in which they perceive in themselves, disordering changes (Falloon 1987; Westacott 1995; Holmes et al. 1994; McCandless-Glimcher et al. 1986; Boker et al. 1989; Cohen & Berk 1985; Carr 1988; Baker 1995; Murphy & Moller 1993). The literature reports that patients have a very wide range of coping styles and that from them proceed several concrete coping strategies. The fourth factor in the vulnerability- stress-coping-model is the protection that the patient receives from his or her environment and that can serve as a buffer against stress. The protection can consist of practical, emotional, and social support to enable the patient to better handle the consequences of the illness (Wing 1978).
The complex interaction between vulnerability, stress, coping, and protection determines whether a patient psychotically decompensates at a given moment. It also determines the gravity of the relapse. When the first signs of a psychosis become apparent, one can act on the various factors of the model in order to counter the aggravation of the psychosis, such as by increasing medication, reducing stress, promoting coping, and taking protective measures in the environment.
Docherty et al. (1978), on the basis of a review of case studies and phenomenological studies, have tried to describe the process of psychotic relapse in six phases. In Phase 0, there is an equilibrium condition in which the patient functions well and succeeds in adapting relatively well to the demands the environment places on him or her. In Phase 1 (overextension), cognitive overload occurs accompanied by symptoms of anxiety, irritation, distractibility, and reduced performance. In the phase that follows, Phase 2, the patient experiences restricted consciousness. Apathy, listlessness, social withdrawal behaviour, feelings of hopelessness, loneliness, boredom, and dependence are some of the phenomena that often occur in this phase. In Phase 3, disinhibition, phenomena can occur that point to disturbed impulse regulation, such as sexual disinhibition, anger attacks, and excessive spending of money. In this phase, the mood may also be seen to improve. In Phase 4, psychotic disorganization occurs together with perceptual and cognitive deterioration, ultimately leading to total fragmentation of the consciousness and loss of self-control. In the last phase that Docherty et al. describe, reorganization of the psychotic consciousness occurs, for example in the form of a paranoid delusional system. This leads to the recovery of control and reduction of anxiety (psychotic resolution).
The description of Docherty et al. is interesting for our present purposes because they were the first to have tried to introduce some structure into the complex process of psychotic relapse. Many of the phenomena cited were also described as early warning signs in later research (see the next section). The authors describe how a number of phenomena can be seen as coping strategies for incipient psychotic symptoms.
This structuring also gave rise to criticism (see, for example, Vries & Delespaul 1988). The process description would not do justice to the dynamics between the various factors of the vulnerability-stress-coping model. According to the critics, the process is presented too deterministically and pessimistically, and justice is not done to the variability between patients. Such a prototypical course of psychotic relapse does not exist.
Birchwood and Spencer (2001) present a simpler model in which three phases are distinguished. In the first phase of relapse, primarily cognitive- perceptual changes occur, such as attentional dysfunction, derealisation, and racing thoughts. They progress to phenomena of dysphoria, accompanied by a depressive mood, pre-occupation with mental life, and loss of interest and self-care. In the third phase, the pre-psychotic or psychotic symptoms emerge to a greater extent, including becoming suspicious, ideas of reference and mild hallucinations.
In the literature, the question is raised about when one can still speak of early warning signs and when there are clear psychotic symptoms. According to Norman and Malla (1995), it is important to make a clear conceptual distinction between early signs and psychotic symptoms. However, this transition proceeds gradually and the introduction of a cut-off point is always arbitrary (Tarrier et al. 1991). For everyday practice, this distinction does not seem to be particularly important. Indeed, it is a matter primarily of reacting in as early a stage as possible to changes in the condition of the patient that point to an impending psychotic relapse (see also Bustillo et al. 1995). In the next section, we will go further into the nature of the early warning signs.
2.2 Early Warning Signs
Early warning signs of a psychosis can be defined as subjective experiences, thoughts, and behaviours of the patient that occur in the phase preceding a psychotic relapse (Heinrichs & Carpenter 1985; Herz & Melville 1980). The question is now which experiences, thoughts, and behaviours are characteristic for this phase.
Heinrichs and Carpenter (1985) conducted a prospective study of the early warning signs of a psychosis in 47 ambulatory patients with schizophrenia (n=38) or a schizoaffective disorder (n=9). During weekly appointments with the client, clinicians noted whether or not warning signs were present that indicated an impending psychotic relapse. On the basis of this, 32 early warning signs were identified. The ten most common are: hallucinations (53%), suspiciousness (43%), change in sleep (43%), anxiety (38%), cognitive inefficiency (26%), anger/hostility (23%), somatic symptoms or delusions (21%), thought disorder (17%) disruptive inappropriate behaviour (17%), and depression (17%).
When we look at the cumulative frequencies, then it is striking that all of the patients manifested at least one of these symptoms. The summary shows a mix of psychotic and non-psychotic symptoms, in which it is striking that the psychotic symptoms score high in comparison to the non-psychotic symptoms. This contrast with the retrospective study of Herz and Melville (1980) whereby the non-psychotic warning signs – observed by patients and family members – score much higher in the ranking. Apparently, it makes a difference whether the warning signs are registered prospectively or retrospectively and also who does the registration. It is possible that symptoms of dysphoria are indeed present, but are reported less by the patient and are also less directly observed by the care provider. They occur more in the experience of the patient and are not always expressed in concrete behavioural terms. When they are observed, then there is still the question of whether they are interpreted as warning signs of a psychosis. It could be that they are more apt to be attributed to the psychosis afterwards than at the time they occur.
Many of the early warning signs that are cited in other literature sources match those cited above or can be seen as explicit behavioural expressions related to these symptoms. The descriptions show that, even though certain early warning signs occur relatively frequently, individual variation is very great ((Heinrichs & Carpenter 1985; Herz & Melville 1980; O'Connor 1991; Tarrier et al. 1991; Kumar et al. 1989; Hamera et al. 1992; Subotnik & Nuechterlein 1988; Kennedy et al. 2000).
One much used scale for the measurement of early warning signs is the Early Signs Scale developed by Birchwood et al. (1989). This scale has satisfactory psychometric characteristics. The authors classify the most common early warning signs in four subscales: (1) anxiety / agitation, (2) depression / withdrawal, (3) disinhibition, and (4) incipient psychosis. The warning signs of the first two subscales occur primarily in the dysphoria phase. The warning signs of the last two subscales are visible primarily when the process of relapse has progressed further and the experiences and behaviours gradually take on a more psychotic character.
Alongside the more general early warning signs, as described above, the so-called idiosyncratic behaviours are also of importance for the early recognition of a psychosis (Norman & Malla 1995; Herz & Lamberti 1995;
Bustillo et al. 1995; Campo & Merckelbach 1996). These are very eccentric behaviours of the patient that often constitute an unmistakable signal for those in his or her environment that the condition of the patient is deteriorating. For example, the patient puts on striking clothing, dyes his or her hair in a striking colour, conducts certain magic rituals, or becomes engrossed in mathematical problems (without having a talent for it). A number of authors stress the importance of preparing a “relapse signature” for each patient that includes both the general early warning signs and the specific idiosyncratic behaviours (Birchwood 1992; Birchwood et al. 1992).
2.3 Recognition of early warning signs
For the use of early recognition and early intervention in care practice, two conditions need to be met: patients and the people in his or her environment must be able to recognize the early warning signs of a psychosis; and there must be sufficient time between the moment of the occurrence of the first warning signs and the time of serious psychotic relapse in order to make early intervention possible. We will now discuss these two conditions.
Herz and Melville (1980) conducted a retrospective study in which they interviewed 145 patients and 80 family members. The patient group consisted of stabilized patients who were being treated on an ambulatory basis (n=99) and partially of clinical patients who were recovering from a psychosis (n=46). The respondents were asked if they could indicate changes in experiences, thoughts, and/or behaviours that preceded the most recent psychotic episode. The conclusion was that 70% of the patients and 93% of the family members could name one or more specific changes. There was great agreement between patients and family members about the most important warning signs (Spearman rank-order correlation = .78 / p < .001). The least agreement was about the pre-psychotic or psychotic symptoms, such as incoherent speech and visual hallucinations. The family members scored them higher than did the patients. The authors state that this was probably due to denial on the part of the patients of psychotic symptoms.
In a retrospective study by Kumar et al. (1989) on the recognition of early warning signs and the coping strategies that followed, 86% of the patients could name one or more signs. Heinrichs et al. (1985) found a lower percentage of 63%.
A limitation of retrospective studies is that patients may well be able to name the warning signs of a psychosis when looking back, but this does not guarantee recognition and acknowledgement when they actually occur.
Patients can play down the changes in their condition at the time they occur because of their desire to function healthily. The ability to recognize symptoms (and to attribute to them their correct significance) can also diminish as the process of relapse progresses because the degree of illness insight decreases (Birchwood et al. 1992; David 1990; McEvoy et al. 1989; Van Meijel & Lendemeijer 1997). The literature is not clear about the number of patients in whom this problem occurs or the degree of gravity with which it occurs.
For the time being, it may be concluded that adequate recognition of early warning signs is improved by involving several parties: the patient himself or herself, the family or other members of the social network, and the care providers. When the recognition capacity of the patient diminishes, others can take over this task. Family members in particular are well able to notice subtle changes in the patient, so they can make an essential contribution to the monitoring of the condition of the patient (Carpenter & Heinrichs 1983). The question of the period between the first warning signs and the psychotic relapse is particularly important from a clinical perspective. One condition for effective early intervention is that this period be of sufficient length. The research results are hopeful in this regard. The study of Herz and Melville (1980) shows that this period was reported by patients and family members to be less than one day in only a few cases (7 - 11%), in which there was hardly time to intervene. A larger percentage (16-24%) said that this period lasted from one to seven days, but a majority of the patients and family members (50 - 68%) reported a period of more than 7 days extending to longer than a month.
Other retrospective and prospective studies confirm that an increase of early warning signs and/or symptoms is apparent for several weeks before a psychotic relapse in a majority of cases (Subotnik & Nuechterlein 1988; Tarrier et al. 1991; Birchwood et al. 1989; Henmi 1993; Birchwood et al. 1992).
The conclusion is that, in a great majority of cases, the time interval between the first early warning signs and serious psychosis is long enough to give sufficient opportunity to apply early intervention strategies. With a small minority; probably less than 10% the psychosis occurs so rapidly that preventive measures have little or no chance of succeeding.
2.4 Predictive value
The significance of early recognition and early intervention increases with the increase of the predictive value of the warning signs for the occurrence of a psychotic relapse. There is still relatively little knowledge about the question of which early warning signs are good predictors of a psychotic relapse. Malla and Norman (1994), Tarrier et al. (1991) and Hirsch and Jolley (1989) consider the occurrence of depressive feelings to be a good predictor. Other authors point primarily to the high predictive value of mild psychotic phenomena (Birchwood 1992; O'Connor 1991; Tarrier et al. 1991; Jorgensen 1998).
For the evaluation of the clinical and conceptual significance of early warning signs, one needs a prospective analysis of the relationship between warning signs and psychotic relapse (Malla & Norman 1994). The positive predictive value (PPV) can then be measured, that is, the probability that observed early warning signs will actually be followed by a psychotic relapse. One problem here, however, is that, in clinical practice, one generally intervenes therapeutically upon the occurrence of warning signs of a psychosis, so the natural course of symptom development is interrupted. The chance of a psychotic relapse is reduced upon successful intervention. When a psychotic relapse would have occurred under natural circumstances, it is prevented by therapeutic intervention. With this, the PPV of the early warning signs declines. No alternatives are conceivable for the determination of this predictive measure because, for ethical reasons, therapeutic intervention is, in principle, always indicated when it is established that the condition of the patient deteriorates.
The PPV has been determined in three studies, and also in these studies pharmacological and psychosocial interventions affected the natural course of the symptom development resulting in low or very low PPVs ranging between 15% to 43% (Marder et al. 1994; Gaebel et al. 1993; Jolley et al. 1990). These studies confirm that the most active interventions do indeed lead to the lowest PPV.
Most researchers concentrate on establishing the sensitivity and the specificity of early warning signs (Subotnik & Nuechterlein 1988; Hirsch & Jolley 1989; Jolley et al. 1990; Birchwood et al. 1989; Tarrier et al. 1991; Marder et al. 1994; Marder et al. 1991; Gaebel et al. 1993; Jorgensen 1998; Malla & Norman 1994).
Sensitivity refers to the proportion of relapsing patients for whom there was an earlier increase in early warning signs (true positive rate), whereas specificity refers to the proportion of nonrelapsing patients for whom there was no increase in early warning signs (true negative rate). Here, too, one must note that therapeutic interventions affected the research results.
As regards the sensitivity of early warning signs, no consistent picture has emerged: the sensitivity varies in the various studies between 8% (Gaebel et al. 1993) and 81% (Jorgensen 1998). Most of the studies score well above 50%. For the specificity of the early warning signs, a somewhat more univocal picture emerges: the values lie between 60 and 93%.
Further examination of these studies shows that the degree of comparability is very small. Differences in practical and methodological choices account for a part of the variable outcomes. We list the most important of them.
The first important factor is the selection of the early warning signs of which the predictive value is established. For example, Gaebel et al. 1993 selected six common early warning signs and conducted a prospective study of their predictive value within different treatment conditions. The pre- selection, based on the research of Herz and Melville (1980), consists of a number of non-specific symptoms and symptoms of dysphoria. The sensitivity did not exceed 14%. Jorgensen (1998), however, used the 34-item Early Signs Scale (Birchwood et al. 1989) in which pre-psychotic symptoms are also included that are known to have a greater predictive capacity. This scale was filled in by the patient himself or herself. The authors arrived at a sensitivity of 74% in their study. One may conclude in this regard that differences in the pre-selection of the early warning signs that are used in the study contribute to the differences in the predictive outcome measures found. Therefore, the conclusion of Gaebel et al. (1993) that the clinical significance of early warning signs is limited because of the low predictive value is, in our opinion, very premature.
A second problem is that the early warning signs are scored with varying frequencies in the studies, sometimes once in four weeks. This last frequency is certainly too low, since a large portion of the psychotic relapses arises within a few days or weeks. With low frequency scoring, early warning signs that are indeed present will not be noticed in some of the patients, which leads to the incorrect conclusion that no warning signs preceded the psychosis. The possibility of preventive intervention is reduced sharply with such low frequency monitoring.
Third, the studies apply different operationalisations of psychotic relapse, which affects the predictive values that are found.
Fourth, the follow-up-periods vary considerably in the studies, which makes it difficult to compare the outcomes.
Because of these problems and particularly because of their combination the study of the predictive value of early warning signs provides little clarity. It is important to conduct further research into the conditions under which satisfactory predictive values can be achieved. Low sensitivity of early warning signs is certainly a problem for clinical practice, for it means that a large portion of the psychotic relapses are not preceded by early warning signs, which correspondingly reduces the importance of early recognition. A low specificity score would also be a problem for it would mean that many patients who do not have a psychosis but who do show early warning signs are given unnecessary treatment and undergo the accompanying stress.
The question is now how can the predictive value of early warning signs be optimised within clinical practice. It is important to prepare an individual profile, a “relapse signature”, for each patient on the basis of previous psychotic relapses of early warning signs in which the pre-psychotic warning signs, along with the more aspecific symptoms have a place (Birchwood 1992; Birchwood et al. 1992). Using standard instruments can be helpful, but it must not be forgotten that certain warning signs are highly idiosyncratic and therefore do not appear on the standard lists. An open clinical interview with the patient and with family members is indicated to identify these idiosyncratic symptoms.
2.5 Research on the effects
Research on the effects of early recognition and early intervention is scarce. Most of the research has been done on the application of early recognition methods combined with divergent medication strategies (Gaebel et al. 1993; Gaebel et al. 2002; Herz et al. 1991; Herz et al. 1989; Herz et al. 1982; Marder et al. 1994; Carpenter et al. 1990; Schooler et al. 1997; Schooler et al. 1995; Inderbitzin et al. 1994; Pietzcker et al. 1993; Jolley et al. 1990).
Generally, it concerns the comparison of maintenance medication in variable dosages with intermittent strategies with which anti-psychotic medication is administered upon the first signs of an approaching relapse. Systematic monitoring of early warning signs is part of the intermittent medication strategy. The occurrence of psychotic relapse is the primary outcome measure in these studies.
A detailed description of the results of these studies goes beyond the scope of this article. The data relevant to our purpose can be summarized as follows.
Intermittent medication strategies appear in general to be inferior to the strategies of offering maintenance medication, also when the intermittent medication policy is combined with intensive monitoring of early warning signs. Intermittent medication strategies lead to less medication consumption and fewer side effects such as tardive dyskinesia, extra-pyramidal symptoms, and negative symptoms caused by neuroleptica. These side effects are often very burdensome for the patient and hinder personal and social functioning. They have a negative impact on medication compliance. The benefits from reduced side effects achieved by the intermittent medication strategy, however, is overshadowed by a higher risk of relapse, so most authors advise against it or propose stringent selection of the patients for whom this strategy could be indicated. Maintenance medication still seems to be the best way to prevent psychoses.
The conclusion is that early recognition and early intervention are recommended mainly in combination with medication maintenance treatment and cannot serve to replace this treatment. Early recognition and early intervention of themselves have an added value with respect to maintenance medication. Adequate early recognition can lead to an early increase of the antipsychotic medication and thus have a preventive effect.
The studies cited above always combined different medication interventions with the application of early recognition and early intervention. In order to establish the effectiveness of early recognition and early intervention as such, it is necessary to standardize the medication treatment. Two studies are important here.
Herz et al. (2000) conducted a controlled study on the effects of a program for relapse prevention in schizophrenia. The experimental group received a combination of (1) psychoeducation about psychotic relapse and early warning signs, (2) active monitoring of early signs, (3) early intervention upon the occurrence of early signs, (4) supportive group or individual meetings directed to improving coping skills, and (5) multifamily psychoeducation groups. The patients in the control group were offered care as usual consisting of individual supportive therapy and medication management. The patients of both groups were prescribed standard doses of maintenance medication. During the follow-up period of 18 months, the patients in the experimental group had significantly fewer psychoses than did the patients in the control group (17% vs. 34%) and were readmitted less often (22% vs. 39%).
Stenberg et al. (1998) studied the effects of the Liberman Module ‘Symptom Management’ with readmission as the outcome measure. This is a group-education program focusing on early recognition and early intervention. The control group received care as usual. The follow-up period was 2 years. The researchers concluded that the number of readmissions did not differ between the experimental and the control condition but that large differences did occur in the duration of the hospital admissions. The experimental patients stayed an average of 2.6 weeks in the hospital while the control patients stayed 20 weeks. This indicates that the patients who had participated in the training program had had much less serious psychoses than did the patients in the control group.
Indicative of possible effects is also the descriptive correlational study of Novacek and Raskin (1998) of 370 ambulatory patients with a serious psychiatric disorder. A good half of the patients suffered from schizophrenia. Data on a large number of areas – including the area of early recognition and illness course – were compiled with the aid of score lists filled in by case managers, interviews with patients, and client-service records. The research results indicate that poor recognition of warning signs was related to poorer treatment outcomes and greater use of expensive types of services. Improvement of recognition was related to better treatment outcomes and lower costs.
The effects of early recognition and early intervention can also be defined in terms of changed experience of the illness. Research into illness experience and the way in which this is influenced by self-management strategies – has hardly been developed. The hypothetical statements of various authors arise mostly from experience. The application of early recognition and early intervention would lead to increased self-efficacy of the patient as regards the self-management of the illness and the exercise of control over his or her own life. This would lead to enhanced self-respect of the patient (see, for example, Carpenter & Heinrichs 1983; Breier & Strauss 1983). The research of Boevink et al. (1995) into the quality of life of chronic psychiatric patients demonstrates the importance for psychiatric patients to get hold of their own life, to have a perspective to the future, and to learn to deal with themselves and their mental problems adequately. Further research is desirable in this area.