Abstract
Affective and non-affective psychotic disorders carry a particularly high lifetime risk of suicide [1]. Suicide is five times more likely in those with psychotic depression than with non-psychotic depressive disorders [2]. Bipolar disorder carries an even higher risk than unipolar depression [3]. Schizophrenia carries a 10% lifetime risk of suicide [4, 5] with the highest risk in those who developed their illness during adolescence i.e., 21.5% of males with adolescent onset [6]. In schizophrenia suicide is much more likely during the early course of illness [7], particularly within the first couple of years of diagnosis [4, 8]. Even the first prodromal phase of psychosis carries a high risk. In a sample of 25 ‘prodromal patients’ 91.5% had experienced suicide ideation and 24% had actually made a suicide attempt [9].
Though a high proportion of patients with schizophrenia experience suicide ideation, only a small number commit suicide. What places this subgroup at high risk may have little to do with the primary symptoms of the illness but more with the person's emotional and cognitive responses to the impact of the illness and the availability of social support [10]. Contrary to traditional expectations, psychotic symptoms such as command hallucinations may only contribute to about 10% of suicides in schizophrenia [11]. The early recovery period of remission is the time of maximum risk [12], especially for men [11]. The same appears to hold true for the first episode psychosis population [12]. In these patients, hopelessness [14] and depression [15, 16] are particularly common in the months after the first episode. This is not surprising, given that this is when patients face some of the most significant adjustments to the illness while they struggle to regain their pre-morbid level of cognitive and emotional functioning.
Higher premorbid IQ, higher socioeconomic background, higher expectations, relatively greater degree of deterioration, loss, stigma, poorer social support and social exclusion/rejection all increase the risk of suicide for those with schizophrenia [5]. Indeed the capacity for ‘insight’ has been associated with higher risk of suicide in schizophrenia [17] while cognitive deficits and negative features actually appear to protect against suicide in schizophrenia [18]. In patients with certain cognitive styles perhaps one of the most important mediating risk factors is the precipitous development of ‘insight’ when faced with the consequences of their illness, thereby inducing a state of ‘hopelessness’ and depression in the post-psychotic phase. The evidence for this seems to be supported by a recent study of depression and suicidality in schizophrenia [19, 20].
In addressing suicide prevention, significant benefits may accrue just from simply influencing better access to more effective treatment of the underlying mental illness. Patients who drop out prematurely or experience an inadequate response to treatment are at significantly greater risk of committing suicide than those who remain engaged in treatment [21]. However, even when treatments are effectively applied, only a few specific pharmacological treatments and cognitive interventions have been shown to reduce the rate of suicide in controlled trials and their effect appears to be small [22, 23]. It is possible that even some standard cognitive interventions for psychosis may have little effect on or even aggravate suicidality in suicide-prone patients [24]. Enhancing a person's insight into psychosis may carry risks if it is not coupled with recovery-enhancing adaptation and protection against relapse, hopelessness and depression.
There are surprisingly few studies of the effect of psychosocial interventions in suicide prevention [23]. In their systematic review of randomised controlled studies (n = 15) of psychosocial interventions following suicide attempts, van der Sande et al. [23] reported that only two of the studies reported significant reductions in the frequency of subsequent suicide attempts and this was only with the application of cognitive-orientated therapies [25, 26]. The rest of the studies revealed mixed results [23, 27] including their own subsequent study [27]. None of these studies have involved patients with psychotic disorders despite this group being at particularly high risk of suicide. Also, as yet there are no reports of any of these therapies being adapted for patients with psychotic disorders. It is unclear to what extent these counselling techniques need to be modified in psychosis and whether this form of counselling reduces the risk of suicide in the subgroup of psychosis patients at particularly high risk.
The aim, setting and context of the LifeSPAN program
Aim
The LifeSPAN project was developed by a group of senior clinicians at EPPIC (and the Mental Health Services for Kids and Youth, MHSKY) some of whom had already been involved in studies of a number of psychological therapies in first episode psychosis. Local clinical audits of suicides and a number of surveys prompted the development of a range of suicide prevention strategies at EPPIC. These strategies aimed at improving mechanisms (both clinical and administrative) to enhance the detection and monitoring of these high suicide risk patients. In addition a treatment intervention was designed (LifeSPAN therapy) specifically for acutely suicidal young patients with mental illnesses. This was to be evaluated in a randomised controlled trial to see whether it might provide additional benefits over and above the other interventions being introduced.
Setting for the LifeSPAN program
The LifeSPAN program was funded by the Government of Australia as part of the Youth Suicide Prevention Strategy [28] and operated between 1st January 1997 and 31st March, 1999. The program employed 2.5 staff (including two full-time clinical psychologists) at MHSKY, which provides a catchment area-based youth mental health service in the Western Region (population 800 000) of Melbourne. The service includes the EPPIC program: an early intervention program for young people (aged 15–29) presenting with first episode psychosis. The Early Psychosis Prevention and Intervention Centre's services include an early detection and crisis assessment team, an acute inpatient unit, an outpatient group program, assertive follow-up teams and an intensive outreach mobile support team. Approximately 250 new patients are accepted into the EPPIC service each year. Treatments are provided via an integrated bio-psychosocial model with a strong emphasis on low-dose medication, cognitive-orientated individual, group and family therapies. Follow-up is provided for 18 months. The Early Psychosis Prevention and Intervention Centre has an established track record in the early engagement and optimal treatment of young people with first episode psychosis [29].
Preparatory phase of the LifeSPAN program
This preparatory phase aimed at providing a baseline of the prevalence of suicidality among patients attending EPPIC and of the standard suicide prevention strategies provided by staff at the centre. This included: (i) suicide audits; (ii) surveys of patients' suicidality; (iii) surveys of clinicians' routine clinical practice in suicide assessment and risk management using a concept mapping exercise [30] with 40 of the 60 staff in EPPIC; (iv) a reevaluation of data already available from other studies of cognitive therapies within EPPIC; and finally (v) the introduction of a suicide risk monitoring system into the service.
(a) An audit of suicides at EPPIC
Data are available for all suicides among patients who had attended EPPIC since its inception at the end of 1992 and for patients who had attended the preceding first episode research unit (the Pre-EPPIC service based in Royal Park Hospital) between 1991 and 1992. Data are also available for all new cases of patients accepted into the EPPIC service for each of these years. An analysis was done of the number of suicides occurring among these patients within their first 2 years of followup. Although the numbers are not large there is a clear drop in the rate of suicide between the first episode samples attending the pre-EPPIC service (e.g., suicide rate of 4% in 1991) and more recent cohorts attending the EPPIC service (e.g., suicide rate of 0.4% or less for each year from 1995 to 2000). A more detailed analysis of the characteristics of the patients who committed suicide before 1998 is reported by McGorry et al. [13].
(b) Clinical audit of suicidality
A clinical audit was conducted involving all patients (n = 312) attending the EPPIC service (up to 18 months after their first contact with EPPIC) using data from Health of the Nation Outcome Scale (HoNOS) ratings [31] (in the MHCaSC study at the Centre in 1996 [32]). patients' HoNOS measures were rated by EPPIC case managers on a fortnightly basis during a 3-month study period. This data not only provided a more accurate cross sectional measure of suicidality in those attending the service but identified the phase of illness when EPPIC patients were most at risk.
Forty-six percent of all patients at EPPIC were rated as suicidal using the HoNOS suicidality subscale: 34% rated as mildly suicidal (score of 1 or 2) and 12% moderate/severely suicidal (score of 3 or 4) within the previous 2 weeks. When the time after first contact with EPPIC was divided into 30-day (i.e., monthly) intervals and the proportion of patients rated as moderately to severely suicidal within each interval were plotted, an interesting pattern emerged indicating when patients might be at highest risk of being moderately/severely suicidal. The proportion of patients rated as moderately/severely suicidal was highest in the first month of contact with EPPIC (17% moderately/severely suicidal), dropping to less than 3% for the second and third months, rising to 10% by the fifth month and remaining high until the sixteenth before gradually dropping again to low levels by the time of discharge from EPPIC between months 18 and 20. Positive correlations were found between patients' suicidality ratings and their concurrent ratings on HoNOS of psychosis, depression and substance abuse. The associations were strongest for depression and suicidality scores and weakest for substance use and suicidality [16].
(c) Staff surveys
The next step was to look at gaps in EPPIC clinicians' knowledge and practice in suicide prevention. A concept mapping exercise [30] suggested that EPPIC clinicians perceived that they were well able to diagnose and assess when patients are at risk of suicide. However, the clinicians perceived themselves as less able to address the psychological factors that contribute to a person's suicidality. Clinicians rated social support as an important protective factor against suicide; however, they believed that the service did not adequately address this particular need [33].
(d) Re-evaluation of the effect of existing CBT studies at EPPIC
A preliminary review of the 6-month outcome data in a sample of first episode psychosis patients randomised to COPE (a cognitive orientated psychotherapy to assist adjustment during the recovery phase of first episode psychosis) actually revealed less improvement in suicidality at 6 months follow-up in the COPE treatment group compared to the control group [24]. This therapy was designed to promote recovery by enhancing psychological adaptation and adjustment [34].
(e) The introduction of a routine suicide risk monitoring system
This strategy included the introduction of a standardized suicide risk assessment (completed on every patient at entry into the program) and routine monthly rating of suicidality by the patient's case managers using the Brief Psychiatric Rating Scale (BPRS) suicidality subscore [35]. Ratings were also done at any time if a patient indicated an immediate risk of suicide behaviour. High risk patients identified by this system were referred and considered for LifeSPAN therapy (see below).
Trial phase of the LifeSPAN program: randomised controlled trial of LifeSPAN therapy
Method
The LifeSPAN project's staff developed a brief individual cognitively, oriented therapy (LifeSPAN Therapy) specifically designed for acutely suicidal youth with severe mental illness. All patients at EPPIC who were rated between 4 and 7 on the Expanded Version 4 of the BPRS suicidality subscore [35] were referred for consideration of LifeSPAN Therapy. A score of 4 equates to ‘suicidal thoughts frequent, without intent or plan’ and 7 equated to a ‘specific suicidal plan and intent or suicide attempt’. If patients agreed to participate they were included in a randomised controlled trial of LifeSPAN Therapy. Patients were excluded if they had attended the service for more than a year. This was to ensure that patients in the trial received standard services at EPPIC during the 6 month follow-up period.
The treatment group received LifeSPAN therapy in addition to standard clinical care at EPPIC, while the control group only received standard clinical care. Ratings for both groups were conducted at three time points: entry into the trial; at 10 weeks (after the final session); and a 6-months follow-up. Ratings were conducted blind to the therapy (as far as was possible). Patients were deemed to have completed the LifeSPAN therapy if they had received eight sessions or more.
Ratings included standardized measures of suicidality, e.g., Suicide Ideation Questionnaire (SIQ) [36], Suicide Intent Scale [37] and the Reasons for Living Inventory (RFL-24) [38]. Measures of psychopathology included the Brief Psychiatric Rating Scale (BPRS) (Expanded Version 4) [35], Scale for the Assessment of Negative Symptoms (SANS) [39], Quality of Life Scale [40] and the Global Assessment of Function (GAF) Scale. Measures of cognitive style included the Beck Hopelessness Scale (BHS) [41], the Self Esteem Scale (SES) [42] and the Self Report Problem Solving Rating Scale (SRPSRS) [43].
Brief description of the LifeSPAN therapy
LifeSPAN therapy comprises 8–10 individual sessions. It draws on the experience at EPPIC with Cognitive Oriented Therapy for Early Psychosis (COPE) [34] and suicide prevention manuals such as Choosing to live [44] and Cognitive therapy of suicide behaviour: A manual for treatment [45]. The therapy was provided by one of two psychologists independent of the EPPIC service. There were four phases: initial engagement; suicide risk assessment/formulation; cognitive modules; and a final closure/handover. The initial two phases aim at a detailed, collaborative risk assessment and formulation of suicidality, identifying the foci to target in the next phase of the therapy. The third phase combines eight modules. The core module addresses further a functional analysis of the patient's suicidality, the rationale for suicide, the quality of hopelessness, and the reasons for living. The additional modules included: problem-solving training, psychoeducation for psychosis, emotional pain tolerance, stress management, self esteem, help-seeking and social skills training. In the Closure Phase at the end of the LifeSPAN therapy, the patient and therapist confirm their formulation of the patient's suicidality and identify the early warning signs or triggers for suicide ideation/behaviours. They reaffirm and negotiate suicide protective strategies, e.g., further counselling for self esteem issues, and reinforce help-seeking options with supportive relationships such as family or case manager. The patient's case manager is then invited to the final LifeSPAN session and in this ‘handover’ session they finalize a Care Plan to address suicide risk and prevention [46].
Results
Of the 92 patients referred and meeting the criteria for the trial, 36 refused to participate. The other 56 were randomised to either the treatment group (n = 31) (LifeSPAN Therapy plus standard clinical care) or control group (n = 25) (standard clinical care). Thirteen of the 56 dropped out during the next 10 weeks and one before the 6 month follow-up. This left 21 patients in each group (treatment and controls) at the end of 10 weeks (Time 2). There were no significant differences in the demographic characteristics of the groups.
Both treatment and control patients improved progressively (initial rating, 10 weeks, and 6 month follow-up ratings) on ratings of suicide ideation (SIQ) and the number of suicide attempts (in the 10 weeks prior to each time point). However, there were no significant differences between the groups in the relative change in ratings between each time point, though the treatment group did show a larger average drop in the Suicide Ideation Questionnaire (SIQ) score. The LifeSPAN group (compared to the control group) demonstrated better improvement (using Analysis of Covariance) with the Beck Hopelessness Scale at 10 weeks (p = 0.032) and 6 months (p = 0.014) and in Quality of Life Scores at 10 weeks (p = 0.001) and 6 months follow-up (p = 0.014). A positive (Pearson correlation 0.5–0.7) association was found between the Beck Hopelessness Score and the Suicide Ideation Questionnaire (SIQ) score at each time point. There was only a weak association between the SIQ score and the number of suicide attempts (in the previous 10 weeks).
Overall, both the LifeSPAN and control groups improved at each point (10 weeks and 6 months follow-up) in their mean ratings of the BPRS total score, BPRS depression subscore, and other measures of psychopathology.
Tragically, two patients committed suicide, one from each group (no other patients committed suicide at EPPIC during the study). Both were outpatients and had been in treatment for 29 and 35 weeks, respectively. They had made numerous and serious attempts before referral to the study and committed suicide several weeks before the 6 month follow-up.
Discussion
The results of the LifeSPAN audit and study confirm the high level of transient suicidality in first episode psychosis patients, particularly at first contact during the acute psychotic phase and then several months later during the early recovery phase. This is despite the provision of intensive psychological and social programs in EPPIC at the latter stage.
The underlying suicidogenic process may differ for the acute psychotic phase and the early recovery period. Suicidality during the acute psychotic phase may in part be driven by a reaction to overwhelming psychotic phenomena, impaired functioning, and the distressed responses of significant others and clinicians. This may account for the rapid reduction in suicidality once services have engaged with patients and their psychosis begins to resolve. This occurs at a time when most have been discharged home and are being closely monitored [47]. That suicidality scores rise again around the fourth month onwards and persist for the next year is surprising. For most of our patients this is at a time when negative symptoms remain prominent and psychosocial functioning is still impaired. Emerging insight, persistent dysfunction, enduring losses and reduced quality of life may all contribute to a sense of hopelessness and depression. This pattern is also reflected in the work of Nordentoft et al. [48] who reported that on first clinical contact, hallucinations and hopelessness predicted suicide ideation and planning whereas the converse was true for thought disorder and delusions. However, at 1 year follow-up only hopelessness was predictive.
Our work provides very encouraging leads. First, the audit suggests that the EPPIC model of early intervention may reduce suicide rates in first episode psychosis comparison. This is despite the high rates of youth suicide in Australia during the same decade. A suicide rate of 0.2% per year during the first 2 years of followup with EPPIC is substantially less than that reported in the literature, i.e. 1–2% within the first year [49].
The mechanism by which early intervention may impact on suicide is unclear. Reducing the duration of illness, improving remission rates, attending to psychosocial issues that augment functional recovery and quality of life may all play a role. Nordentoft et al. [48] is one of the few teams that have evaluated the impact of early intervention service on suicidality. In a randomised controlled trial comparing the one year outcome of patients receiving early intervention assertive outreach and standard community care, hopelessness diminished significantly more in the former than the latter. Surprisingly, no association was found between duration of untreated psychosis and suicidal behaviour.
The LifeSPAN project suggests that it is possible to incorporate an effective system into a multi-layered program to detect and monitor at-risk patients. Ideally this should offer ‘real time’ feedback to alert clinicians of their progress. One option is periodic ‘zoning’ of patients into three risk categories (low = green, moderate = amber, high = red) [50]. Our study also demonstrates that it is feasible to conduct randomised controlled trials in this area. However, the high measures of suicidality and the death of two participants highlight the difficulty of conducting these investigations. Nonetheless, as no other deaths occurred in our service at the time and suicide rates remained unchanged compared with historical data, it suggests that the highest risk group was identified by the monitoring system.
Our most pivotal findings are that standard care reduces suicidality and adding LifeSPAN therapy leads to an improvement in cognitive factors linked to suicide risk, e.g., hopelessness [46]. Further analysis of our data, we hope, will reveal which aspects of the intervention are salient. We should note however, that cognitive therapy per se may not necessarily reduce the risk of suicide and that any effort should be done with the utmost care.
Our study has a number of limitations. LifeSPAN therapy is possibly too brief given the complexity of issues it addressed. The findings need to be interpreted in the context of a small sample size, the high quality of standard care at EPPIC, the contamination effect of other suicide prevention strategies introduced during the GORRY study, the absence of a control treatment, and the partial analysis of the data. It also is unclear how effective the intervention might be if delivered by clinicians with limited training and supervision in cognitive therapy.
An aspect of the LifeSPAN data not yet analysed is the effect of medication on suicidality in patients attending EPPIC. Since the initial analysis, a large multicentre randomised controlled trial of clozapine and olanzapine has been reported that may have a specific effect in reducing suicidality in schizophrenia [51]. This begs the question of whether a combination of clozapine and a form of therapy like LifeSPAN might not have a synergistic effect.
In conclusion, we have confirmed the high prevalence of suicidality among young people with first episode psychosis, particularly in the recovery phase. Specific strategies to identify those at risk and to engage them in early intervention is obviously indicated. Standard treatment may yield some benefit but a specific initiative like LifeSPAN magnifies this. Specific medications may also play a role, offering complementary benefits, but this aspect needs to be examined further.
Footnotes
Acknowledgements
The LifeSPAN project was funded by the Commonwealth Department of Health & Family Services. We thank Keith Hawton, Mark Williams, Conrad Hauser, Andrew Fuller, Roy Butterham and clinical and administrative staff at MHSKY (Jane Edwards, Kerryn Pennell, Heather Manning and Angela White).