7.1. Introduction
The incentive to identify and develop effective early interventions for post-traumatic stress disorder comes from three sources. First, PTSD is a distressing and disabling condition from which a great number of sufferers do not spontaneously recover. Therefore, early and effective treatment might reduce the burden of PTSD on both the individual and society. Second, now that studies have identified the post-incident prevalence rates of PTSD from large-scale disasters and combat, there is concern to ameliorate the impact of PTSD by responding in the early days and weeks following such incidents. Third, occupational groups such as firefighters have campaigned to have the psychological impact of their work recognised and support services delivered as part of their conditions of employment. In addition, in military organisations, there exists a specific drive to early interventions – that of enabling traumatised combatants to return to front-line duties as soon as possible.
However, given that the prevalence of initial distress following a traumatic event is far greater than that of either acute stress disorder or PTSD, the potential exists to deliver interventions to people whose problems would spontaneously remit. As well as the time commitment required of the traumatised individual, interventions for traumatic stress generally involve confronting aspects of distressing experiences, the emotional cost of which might not warrant early intervention. This potential for diluting the cost-effectiveness of early interventions is a significant factor in service planning, particularly disaster planning and employee support. There is a vigorous debate between those who would provide some intervention for all victims and survivors of traumatic incidents, and those who advocate waiting and targeting interventions at people likely to develop the disabling symptoms of chronic PTSD (Litz et al, 2002).
7.2. Current practice
Several interventions often referred to generically as ‘debriefing’, such as crisis intervention (Raphael, 1986) and critical incident stress debriefing (CISD; Mitchell, 1983), have been developed since the 1980s to help deal with the immediate psychological aftermath of severe trauma. In particular, CISD – defined as a meeting of those involved in a traumatic event, which aims to diminish the impact of the event by promoting support and encouraging processing of traumatic experiences in a group setting (Richards, 2001) – gained widespread initial popularity. Subsequently, Mitchell & Everly (1997) coined the term ‘critical incident stress management’ (CISM) to differentiate the single-session, stand-alone debriefing meeting from a broader, multicomponent programme including pre-trauma training, CISD, follow-up and case management. Both CISD and CISM were designed to try to accelerate recovery before harmful stress reactions have had a chance to damage the performance, career, health and families of victims. However, there is no agreement on the best way to deliver early interventions or indeed whether it is possible to reduce the incidence of PTSD through this route (Litz et al, 2002). Indeed, the area is hotly contested. The efficacy of debriefing has been called into question in systematic reviews (e.g. van Emmerik et al, 2002; Rose et al, 2004), which have suggested that single-session CISD produced either no improvement compared with controls or had the potential to cause significant harm to those debriefed. It has been suggested that single-session debriefing might sensitise traumatised individuals further or might persuade people not to use the necessary natural social support networks likely to assist with recovery (van Emmerik et al, 2002). These studies led to claims that CISD was an ineffective technique and that it should not be routinely used in supporting people after traumatic incidents (Avery & Orner, 1998; Wessely et al, 1998). Other reviews, however, came to the conclusion that CISD is a useful technique as part of an overall CISM programme (Everly et al, 1999) and that the studies included in the negative reviews sacrifice internal validity for experimental control, use self-selected participants, misapply these techniques to individuals rather than to the groups for which they were originally designed, use CISD outside the time scale recommended and have debriefers who appear inadequately trained (Mitchell & Everly, 1997). Indeed, negative reviews such as that by van Emmerick et al (2002) do include the caveat that CISD was never designed to be a stand-alone intervention. Some have suggested that early intervention and debriefing should be directed at community or group support rather than individual treatment and have called for new research methods to investigate this approach (British Psychological Society, 2002).
Indeed, the lack of non-intervention controls in studies of ‘pure’ debriefing is a problem for clinicians and policy-makers alike, a problem compounded by the ethical difficulty of designing non-intervention conditions in sensitive post-incident or workplace environments where offering no support may be unacceptable to employees and employers alike. The provision of psychological debriefing as a community support and cohesion strategy (British Psychological Society, 2002) rather than a treatment intervention to prevent PTSD is beyond the scope of this guidance.
More recently, there has been significant interest in replicating some of the findings from the treatment of chronic PTSD in an early intervention format with populations identified as at risk of developing chronic PTSD. The belief that cognitive–behavioural therapy is effective for PTSD, the disquiet over debriefing and the desire referred to earlier to limit the duration of disability for sufferers has led to either the adaptation of routine CBT into shorter variants delivered close to the time of the incident or the application of more standard CBT within a few months of the incident. Given that the efficacy of CBT for PTSD was only established in the late 1990s, early interventions of this kind are a new development and have only recently been the subject of research.
7.3. Studies included
The review team conducted a new systematic search for RCTs that assessed the efficacy of treatments delivered in any of the two areas described above. From the main search for RCTs (see Appendix 6), 21 studies in all were identified that met the inclusion criteria. The retrieved studies were divided into three groups:
treatment for all – studies that investigated treatments delivered to all traumatic incident survivors, normally within the first month after the incident
early psychological interventions for acute PTSD and acute stress disorder – studies that investigated treatments delivered to people who were assessed as having a high risk of chronic PTSD, initiated within 3 months of the incident
early pharmacological interventions – studies using drug treatments for people in the acute phase of the disorder.
Ten studies were identified as falling within the ‘treatment for all’ category: BISSON 1997, BROM 1993, CAMPFIELD 2001, CONLON 1999, DOLAN, HOBBS 1996, LEE 1996, MAYOU 2000, ROSE 1999 and ZATZICK 2001. Nine studies, comprising five different types of intervention, were identified as falling within the category of early interventions for acute PTSD and acute stress disorder: BISSON 2004, BRYANT 1998, BRYANT 1999, BRYANT 2003, BRYANT A, BRYANT B, ECHEBURA 1996, EHLERS 2003A and OST 2003. (In EHLERS 2003A the self-monitoring period was taken to be part of the active intervention and as occurring within 3 months of the trauma.) Two studies were identified as falling in the pharmacological category: PITMAN 2002 and SCHELLING 2001. References given in shortened format and summary characteristics of individual included trials are given in Appendix 14.
7.4. Treatment for all
Four different types of early intervention for all were identified in the RCTs that met the inclusion criteria. These were education, collaborative care, trauma-focused counselling and debriefing.
7.4.1. Education
One study (ROSE 1999, n=157) included a 30 min educational intervention in its randomised design, comparing it with debriefing and an assessment-only control group. Individuals were assault victims and the intervention was delivered to them 9–31 days after the assault. Education consisted of a 30 min session related to the individual’s experiences and information on when and where to find help, and included a specially written leaflet. Education was delivered by a therapist whose qualifications were not described. Outcomes were reported at 6 months post-intervention.
7.4.1.1. Education versus control
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between education and control on reducing the likelihood of having a PTSD diagnosis at 6 months’ follow-up (k=1; n=103; RR=0.69, 95% CI 0.37 to 1.3).
[I]
There is evidence suggesting there is unlikely to be a clinically important difference between education and control on reducing the severity of PTSD symptoms (self-reported) at 6 months’ follow-up (k=1; n=91; SMD=−0.18, 95% CI −0.59 to 0.24).
[I]
7.4.2. Collaborative care
One study (ZATZICK 2001, n=34) compared a collaborative care programme delivered by a trauma support specialist for road traffic accident survivors with usual care. Collaborative care involved eliciting and monitoring patients’ post-traumatic concerns and joint provider–patient treatment planning. Monitoring was undertaken by consultation liaison psychiatrists and a trauma clinical nurse specialist. The intervention was delivered from 1 month after the accident for 4 months and involved trauma support specialists monitoring psychological health, reviewing the traumatic event, providing education on coping strategies, jointly developing problem definitions and plans with individuals, and liaison with a multidisciplinary trauma team. Outcomes were reported at 4 months post-intervention.
7.4.2.1. Collaborative care versus control
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between collaborative care and control on reducing the severity of PTSD symptoms (self-report measures) at 1 month’s follow-up (k=1; n=29; SMD=−0.5, 95% CI −1.24 to 0.24).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between collaborative care and control in severity of PTSD symptoms (self-report measures) at 4 months’ follow-up (k=1; n=26; SMD=0.4, 95% CI −0.38 to 1.18).
[I]
7.4.3. Trauma-focused counselling
One study (BROM 1993, n=151) compared a three- to six-session counselling programme with a monitoring control. Counselling was delivered 1–3 months after the traumatic incident to road traffic accident survivors. Counselling included practical help, education, support, reality testing and confrontation with the traumatic experience. Outcomes were reported at 6 months following the accident.
7.4.3.1. Trauma-focused counselling versus control
There is evidence suggesting there is unlikely to be a clinically important difference between trauma-focused counselling and control on reducing the severity of PTSD symptoms (self-report measures) at 6 months’ follow-up (k=1; n=151; SMD=0.17, 95% CI −0.15 to 0.49).
[I]
7.4.4. Debriefing
Seven RCTs of individual psychological debriefing were identified: BISSON 1997, CONLON 1999, DOLAN, HOBBS 1996, LEE 1996, MAYOU 2000 and ROSE 1999; n=629. Studies involved individuals who had experienced a range of traumatic events including road traffic incidents, assaults, miscarriages, fires and unspecified other incidents. Psychological debriefing was delivered between 10 hours and 31 days after the incident, with a duration of 30–120 min. Five studies were of individual treatment only, one study included some debriefing of groups of two to five PTSD sufferers and another included family members in some debriefing sessions. All debriefing interventions were single sessions and included education about traumatic stress, expression of emotions and planning for the future. Debriefing was delivered by a range of professionals, including nurses, mental health nurses, psychiatrists and psychologists. The training and qualifications of the debriefers was not comprehensively described in any of the studies. Five studies reported post-intervention outcomes up to 4 months; three studies reported outcomes from 6 months to 13 months and one study reported outcomes to 3 years. There was no randomised study of critical incident debriefing, the group-focused approach advocated by Mitchell & Everly (1997), in contrast to the single-session, typically individually focused debriefing interventions considered in this section.
7.4.4.1. Delayed versus immediate debriefing
One study of delayed versus immediate debriefing for victims of robbery was included (CAMPFIELD 2001). Debriefing lasted 1–2 hours and was conducted individually or in small groups. There was evidence favouring immediate debriefing (occurring within 10 hours of the trauma) to delayed debriefing (occurring within 48 hours of the trauma) for reducing PTSD severity at 2 weeks post-trauma. However, the study provided no data to indicate how sustained this relative improvement was, and in the absence of a control group it is not possible to determine whether both treatments led to an improvement relative to natural recovery at 2 weeks.
7.4.4.2. Debriefing versus control
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between debriefing and control on reducing the likelihood of having a PTSD diagnosis at 3–6 months’ follow-up (k=2; n=238; RR=1.2, 95% CI 0.84 to 1.71).
[I]
There is limited evidence suggesting a difference favouring control over debriefing on reducing the likelihood of having a PTSD diagnosis at 13 months’ follow-up (k=1; n=133; RR=1.87, 95% CI 1.12 to 3.12).
[I]
There is evidence suggesting there is unlikely to be a clinically important difference between debriefing and control on reducing the severity of PTSD symptoms (self-report measures) at 1–4 months’ follow-up (k= 5; n=356; SMD=0.11, 95% CI −0.1 to 0.32).
[I]
There is evidence suggesting there is unlikely to be a clinically important difference between debriefing and control on reducing depression symptoms at 1–4 months’ follow-up (k= 3; n=225; SMD=0, 95% CI −0.27 to 0.26).
[I]
7.5. Treatment for all – clinical summary
When brief, single-session interventions of debriefing or education are offered as an individually structured intervention to any person involved in a traumatic incident, there is evidence suggesting that there is unlikely to be a clinically important effect on subsequent PTSD and across a range of self-report measures. However, one study (BISSON 1997) suggested that there is limited evidence of harmful effects of debriefing at 13 months’ post-injury for PTSD diagnosis. On current evidence, therefore, single-session debriefing may be at best ineffective.
An important reservation in interpreting the evidence for early interventions for all is that all the studies were of survivors who had experienced individual traumas and who in the main received an individual intervention. No trial on critical incident stress debriefing as it was originally conceived by Mitchell and colleagues (i.e. as a group intervention for teams of emergency workers, military personnel or others who are used to working together) or critical incident stress management (i.e. a multicomponent programme of debriefing, follow-up and case management) met our methodological inclusion criteria. As a consequence we have a lack of evidence for practice in these situations. Furthermore, there is a paucity of methodologically sound early intervention studies, containing detailed descriptions of training and fidelity checks on interventions used.
Notwithstanding these methodological reservations, given the evidence that there is unlikely to be a clinically important effect on subsequent PTSD, we do not recommend that systematic, brief, single-session interventions focusing on the traumatic incident are provided individually to everyone who has been exposed to such an incident. However, we do recommend the good practice of providing general practical and social support and guidance to anyone following a traumatic incident. Acknowledgement of the psychological impact of traumatic incidents should be part of healthcare and social service workers’ responses to incidents. Support and guidance are likely to cover reassurance about immediate distress, information about the likely course of symptoms, and practical and emotional support in the first month after the incident.
7.6. Early psychological interventions for acute PTSD and acute stress disorder
Five different types of early intervention for all were identified in the RCTs that met the inclusion criteria: trauma-focused cognitive–behavioural therapy (as defined in Chapter 5), trauma-focused CBT supplemented with hypnosis or anxiety management, relaxation techniques and a self-help booklet.
7.6.1. Cognitive–behavioural therapy
All nine studies identified for ‘early intervention for acute PTSD’ had one treatment group that underwent some form of CBT (see Chapter 5 for descriptions of the treatments that fall within this category): BISSON 2004, BRYANT 1998, BRYANT 1999, BRYANT 2003, BRYANT A, BRYANT B, ECHEBURA 1996, EHLERS 2003A and OST 2003; n=491 (one of these studies, BRYANT 2003, was a 4-year follow-up to BRYANT 1998 and BRYANT 1999). These studies involved individuals who had experienced accidents or physical and sexual assaults. In six studies, individuals were identified within 1 month of the trauma occurring and treatment was continued into the period 1–6 months after the trauma. In the other three studies, PTSD sufferers were identified within 3 months and treatment was completed within 6 months. All individuals were included in the studies on the basis of symptomatic criteria, but these varied. Four studies required survivors to have a diagnosis of acute stress disorder; two others required survivors to meet symptomatic diagnostic criteria for PTSD; another two studies required individuals to meet PTSD symptomatic diagnostic criteria and, in addition, to exceed cut-off scores on screening tools; and a final study included individuals on the basis of exceeding cut-off scores on screening tools only. Treatment was delivered to individuals rather than in groups and ranged from 4 sessions to 16 sessions of 1–2 hours’ duration, a total time ranging from 4 hours to almost 17 hours of therapy. Treatment also included varied combinations of education, relaxation, imaginal exposure, image habituation training, thought stopping, distraction, cognitive restructuring and in vivo exposure. Post-intervention outcomes were reported in four studies at 6 months and in another five studies at 9 –13 months. One study reported a further follow-up at 4 years.
7.6.1.1. Trauma-focused CBT versus control
There is limited evidence suggesting a difference favouring trauma-focused CBT over waiting list (random effects) on reducing the likelihood of having a PTSD diagnosis post-treatment (k=3; n=252; RR=0.4, 95% CI 0.16 to 1.02).
[I]
There is limited evidence suggesting a difference favouring trauma-focused CBT over waiting list (random effects) on reducing the likelihood of having a PTSD diagnosis at 9–13 months’ follow-up (k=2; n=209; RR=0.41, 95% CI 0.11 to 1.45).
[I]
There is limited evidence suggesting a difference favouring trauma-focused CBT over waiting list (random effects) on reducing the severity of PTSD symptoms (self-report measures) (k=3; n=224; SMD=−0.98, 95% CI −1.81 to −0.14).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between trauma-focused CBT and waiting list (random effects) on reducing the severity of PTSD symptoms (self-report measures) at 9–13 months’ follow-up (k=2; n=171; SMD=−0.68, 95% CI −1.23 to −0.12).
[I]
There is limited evidence suggesting a difference favouring trauma-focused CBT over waiting list (random effects) on reducing the severity of PTSD symptoms (clinician-rated measures) (k=3; n=224; SMD=−0.88, 95% CI −1.72 to −0.04).
[I]
There is evidence suggesting there is unlikely to be a clinically important difference between trauma-focused CBT and waiting list (fixed effects) on reducing the severity of PTSD symptoms (clinician-rated measures) at 9–13 months’ follow-up (k=2; n=171; SMD=−0.45, 95% CI −0.75 to − 0.14).
[I]
7.6.2. Prolonged exposure with anxiety management
One study (BRYANT 1999, n=36) compared the effectiveness of prolonged exposure and anxiety management techniques against prolonged exposure. Prolonged exposure entailed a minimum of four 50 min sessions of imaginal exposure to the traumatic memories as part of the five 90 min treatment sessions. Anxiety management included breathing retraining, self-talk and progressive muscular relaxation exercises. Individuals had all experienced road traffic accidents or non-sexual assaults and outcomes were reported at 6 months post-intervention.
There is limited evidence suggesting a difference favouring prolonged exposure over prolonged exposure with anxiety management on reducing the likelihood of having a PTSD diagnosis post-treatment (k=1; n=38; RR=0.58, 95% CI 0.3 to 1.15).
[I]
There is limited evidence suggesting a difference favouring prolonged exposure over prolonged exposure with anxiety management on reducing the likelihood of having a PTSD diagnosis as observed at 6 months’ follow-up (k=1; n=38; RR=0.64, 95% CI 0.37 to 1.11).
[I]
7.6.3. Trauma-focused CBT and hypnotherapy
One study (BRYANT B, n=63) compared the effectiveness of trauma-focused CBT versus trauma-focused CBT with an additional element of hypnotherapy in the form of a 15 min hypnotic induction audiotape recording, for individuals who had experienced road traffic accidents or non-sexual assaults. Outcomes were reported to 6 months post-intervention.
7.6.3.1. Trauma-focused CBT versus trauma-focused CBT and hypnotherapy
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between trauma-focused CBT and trauma-focused CBT with hypnotherapy on reducing the likelihood of having a PTSD diagnosis post-treatment (k=1; n=63; RR=1.21, 95% CI 0.6 to 2.46).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between trauma-focused CBT and trauma-focused CBT with hypnotherapy on reducing the likelihood of having a PTSD diagnosis as observed at 6 months’ follow-up (k=1; n=63; RR=1.06, 95% CI 0.59 to 1.92).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between trauma-focused CBT and trauma-focused CBT with hypnotherapy on reducing the severity of PTSD symptoms (self-report measures) (k=1; n=47; SMD=0.13, 95% CI −0.45 to 0.7).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between trauma-focused CBT and trauma-focused CBT with hypnotherapy on reducing the severity of PTSD symptoms (self-report measures) as observed at 6 months’ follow-up (k=1; n=47; SMD=0.07, 95% CI −0.5 to 0.64).
[I]
7.6.4. Relaxation
One study of trauma-focused CBT (ECHEBURA 1996, n=20) used relaxation alone as a comparator condition with a group of female survivors of sexual assault. Relaxation was progressive muscle relaxation training and was delivered in five hour-long sessions. Post-intervention outcomes were reported to 12 months.
7.6.4.1. Trauma-focused CBT versus relaxation
There is limited evidence suggesting a difference favouring trauma-focused CBT over progressive muscular relaxation training on reducing the likelihood of having a PTSD diagnosis post-treatment (k=1; n=20; RR=0.4, 95% CI 0.1 to 1.6).
[I]
There is limited evidence suggesting a difference favouring trauma-focused CBT over progressive muscular relaxation training on reducing the likelihood of having a PTSD diagnosis at 12 months’ follow-up (k=1; n=20; RR=0.2, 95% CI 0.01 to 3.7).
[I]
7.6.5. Supportive psychotherapy
Five studies (BRYANT 1998, BRYANT 1999, BRYANT 2003, BRYANT A and BRYANT B; n=191) involved a comparison of ‘supportive psychotherapy’ against other treatments. Across the studies, the researchers defined supportive psychotherapy to include active listening, education, problem-solving and unconditional support to individuals.
7.6.5.1. Trauma-focused CBT versus supportive psychotherapy
There is evidence suggesting a difference favouring trauma-focused CBT over supportive psychotherapy on reducing the likelihood of having a PTSD diagnosis at 6 months’ follow-up (k=3; n=105; RR=0.51, 95% CI 0.32 to 0.8).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between trauma-focused CBT and supportive psychotherapy on reducing the likelihood of having a PTSD diagnosis at 4 years’ follow-up (k=1; n=80; RR=0.9, 95% CI 0.61 to 1.33).
[I]
There is evidence suggesting a difference favouring trauma-focused CBT over supportive psychotherapy on reducing the severity of PTSD symptoms (self-report measures) (k=3; n=94; SMD=−1.11, 95% CI −1.55 to −0.67).
[I]
There is limited evidence suggesting a difference favouring trauma-focused CBT over supportive psychotherapy on reducing the severity of PTSD symptoms (self-report measures) at 6 months’ follow-up (k=3; n=94; SMD=−0.8, 95% CI −1.22 to −0.37).
[I]
7.6.6. Self-help
One study of trauma-focused CBT (EHLERS 2003A, n=85) used self-help as a comparator condition for individuals who had experienced road traffic accidents or physical assault. Self-help patients were given a 64-page booklet based on CBT principles, accompanied by one 40 min session with a therapist at the beginning of treatment to explain the book and its content. Post-intervention outcomes were reported to 9 months.
For self-help booklet intervention delivered 1–6 months after the incident, the evidence varies from suggesting that there is not a clinically important difference between the intervention and control, to being inconclusive across the different measures of outcome.
7.6.6.1. Self-help booklet versus control
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between self-help booklet and waiting list on reducing the likelihood of having a PTSD diagnosis post-treatment (k=1; n=57; RR=1.09, 95% CI 0.81 to 1.46).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between self-help booklet and waiting list on reducing the likelihood of having a PTSD diagnosis at 9 months’ follow-up (k=1; n=57; RR=1.1, 95% CI 0.71 to 1.71).
[I]
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between self-help booklet and waiting list on reducing the severity of PTSD symptoms (self-report measures) (k=1; n=52; SMD=−0.27, 95% CI −0.81 to 0.28).
[I]
There is evidence suggesting there is unlikely to be a clinically important difference between self-help booklet and waiting list on reducing the severity of PTSD symptoms (clinician-rated measures) at 9 months’ follow-up (k=1; n=52; SMD=0.07, 95% CI −0.47 to 0.62).
[I]
7.7. Clinical summary of early psychological interventions
When trauma-focused CBT is delivered between 1 month and 6 months after the incident, there is evidence suggesting that it is effective for people at risk of developing chronic PTSD, compared with the effect of being on a waiting list, for PTSD diagnosis post-treatment and at 9–13 months’ follow-up, as well as a number of other outcomes assessed post-treatment, which included self-report measures of PTSD severity, anxiety and quality of life and clinician-assessed PTSD severity. However, the evidence is inconclusive for a number of outcomes assessed at 9–13 months’ follow-up (self-report measures of PTSD severity, anxiety and quality of life) and the evidence suggests that there is no clinically important difference for clinician-assessed PTSD severity at 9–13 months.
Trauma-focused CBT delivered between 1 month and 6 months after the incident is also more effective for people at risk of developing chronic PTSD compared with being on a waiting list or receiving non-trauma-focused interventions such as self-help booklets, relaxation or general supportive counselling.
Although trauma-focused CBT is effective for people at risk of developing chronic PTSD, there is great variation in the dimensions of delivery. The variable response rates in different studies are unexplained and may be due to differences in the PTSD sufferer intake variables (for example, symptomatic PTSD criteria versus diagnoses of acute stress disorder), number of treatment sessions, the expertise of the therapists or the length of individual therapy sessions. The interaction and predictive effects of symptom severity and the duration and number of sessions in trauma-focused CBT are likely to be highly important but have not been systematically varied in controlled trials.
7.8. Early intervention drug treatments for PTSD
There are few trials of early intervention drug treatments and only two studies met the inclusion criteria. The results of the review of these studies are summarised below. For further information on the differences between drug trials and trials of other interventions, see Chapter 6.
7.8.1. Propranolol versus placebo
One study (PITMAN 2002) compared propranolol and placebo. Propranolol is a beta-adrenoceptor blocker and crosses the blood–brain barrier. This trial was based on a priori hypotheses about the role of the amygdala in the development of PTSD. Participants were administered propranolol (40 mg four times a day) or placebo, beginning within 6 hours of the traumatic event.
The evidence is inconclusive and so it is not possible to determine if there is a clinically important difference between propranolol and placebo on reducing the likelihood of having a PTSD diagnosis at 1 month (k=1; n=41; RR=1.14, 95% CI 0.55 to 2.35).
[I]
There is limited evidence suggesting a difference favouring placebo over propranolol on reducing the likelihood of having a PTSD diagnosis at 3 months’ follow-up (k=1; n=41; RR=1.28, 95% CI 0.69 to 2.38).
[I]
7.8.2. Hydrocortisone versus placebo
One study (SCHELLING 2001, n=20) explored the effect of hydrocortisone (a corticosteroid) and placebo on the reactions to the intense physical and psychological stress during septic shock in the intensive care environment. Studies of hydrocortisone are of particular interest given the evidence of disturbance in the HPA axis in PTSD.
There is limited evidence suggesting a difference favouring hydrocortisone over placebo on reducing the likelihood of having a PTSD diagnosis at approximately 31 months after treatment (k=1; n=20; RR=0.17, 95% CI 0.03 to 1.17).
[I]
7.9. Clinical summary of early intervention drug treatments
Given the small number and scale of studies of early intervention drug treatments, it is not possible to draw strong conclusions. At present there is no conclusive evidence that any drug treatment helps as an early intervention for the treatment of PTSD-specific symptoms. However, for sufferers who are acutely distressed, and may in particular be experiencing significant sleep problems, consideration may be given to the use of medication.
7.10. Economic evaluation of early versus later delivery of psychological treatment
7.10.1. Introduction
The phenomenon of spontaneous or natural remission has both health and economic consequences in PTSD as in other conditions. The proposition is that where treatments are given to patients who would otherwise naturally recover, resources could be better spent on patients who need them. The difference between natural remission and treatment-related recovery is critical, and this difference may change with time elapsed since the traumatic event. A large number of mental health economic studies have been conducted generally (see McCrone & Weich, 2001), but none has addressed the incremental costs of alternative interventions, nor the cost-effectiveness of early versus late delivery of cognitive–behavioural therapy, in particular for PTSD. Moreover, few studies have presented a decay curve showing the changing slope of natural remission over time. For example, to examine this phenomenon, Richards (2005) presented ‘caseness’ data for PTSD, using a General Health Questionnaire (GHQ). At 3 days post-trauma, 60% of the patients who were directly involved in a raid suffered from PTSD symptoms. However, this figure nearly halved to 31% by 2 weeks post-trauma, and again to 17% by 1 month post-trauma (, ).
Post-traumatic stress disorder (PTSD) ‘caseness’ of patients directly involved in a raid (x, weeks post-trauma; y, PTSD % caseness). Data from Richards (1997).
Data corresponding to decay curve in Figure 7.1.
7.10.2. Method
For this guideline, all psychological interventions and different service provision options for the treatment of PTSD were briefly reviewed from a health economics perspective. The Guideline Development Group decided to focus on the question of the appropriate time at which to initiate treatment: that is, were there significant additional costs associated with intervening early or later in the course of PTSD? An economic evaluation was therefore undertaken using data from published sources (Bryant et al, 1998; Ehlers et al, 2003), along with a cost-effectiveness analysis in accordance with the NICE guideline development recommendations (National Institute for Clinical Excellence, 2004). The cost-effectiveness was evaluated to determine the consequences of moving patients from one treatment category to an incrementally earlier treatment category. The NHS perspective was adopted, with only direct staffing costs included in the analysis.
Component costs were measured from the health services perspective based upon 2002–3 prices, and estimated as hours of treatment multiplied by the hourly wage of a clinical psychologist according to the 2002–3 Unit Costs of Health and Social Care manual (Netten & Curtis, 2003). According to these estimates, the average annual wage of a clinical psychologist is £33 193 per year and requires £3775 in on-costs, £4230 in direct revenue overheads and £1713 in capital overheads. This translates into £66 per hour of client contact, including these additional expenses.
Using data from Bryant et al (1998), total cost was estimated in the instance of early intervention from five 1.5-hour sessions. These earlier timed interventions were evaluated alongside longer-term data from Ehlers et al (2003), where total cost was estimated from [(5 × 1.5 h) + (5 × 1 h)] sessions, using the above per-hour costing estimates. Total cost reflected treatment-related recoveries as well as recoveries that might otherwise have occurred naturally in patients who received CBT.
Types of trauma comprised motor vehicle accidents (Ehlers et al, 2003) and motor vehicle accidents or industrial accidents (Bryant et al, 1998). Treatment-related recoveries were calculated separately for each study by calculating total number of recoveries in the CBT group minus the natural recoveries estimated to arise from the supportive counselling or control groups. Cost-effectiveness ratios comparing treatments of various lengths at 2 weeks versus 12 weeks were estimated and presented as incremental cost per additional treatment-related recovery. Follow-up times ranged from 3 months to 12 months post-trauma.
In this analysis, cost was calculated as cost-of-treatment multiplied by the number of sessions, and effectiveness the number of patients who would not otherwise naturally recover (e.g. chronic PTSD sufferers). Uncertainty was estimated in a sensitivity analysis in which the stochastic and deterministic data were varied over plausible ranges.
Incremental cost-effectiveness was determined by dividing the difference between the total costs (TC) of programme 1 and programme 2 by the difference in numbers of treatment-related recoveries (TR) for each programme, to give the incremental cost-effectiveness ratio (ICER):
(TC2 – TC1)/(TR2 – TR1) = ICER
This ratio indicates the cost-effectiveness of each timing of programme. In comparing two points in time post-trauma, the ICER indicates the marginal cost necessary to achieve an additional treatment-related recovery. Where the cost is negative, this indicates a reduction in cost per additional treatment-related recovery, relative to the comparator.
Costs and health outcomes were discounted at 3.5%. Discounting is a technique that assumes individuals prefer to delay costs rather than incur them in the present, and forego future benefits in exchange for gaining some benefit in the present. The choice of a discount rate is a disputed issue, however. Although the most common recommendation in health economics literature is that costs and health outcomes should be discounted at the same rate, there is considerable variation in practice, with some studies failing to discount health effects at all, and others citing that different age-groups will discount costs and benefits differently. For instance, younger individuals may be at a disadvantage with respect to the terms on which they can borrow money (i.e. money allocated today is worth far more than it is anticipated to be in the future), and older people or individuals with comorbid conditions that impinge on their quality of life are more aware of their own mortality and therefore may discount benefits more heavily (i.e. an intervention today is worth more than in the future).
Although the effect of discounting was small in this study, the choice to include this method followed NICE recommendations (NICE, 2004) and is common practice in health economic evaluations. Discounting reduced the total cost by £231 in the case of treatment at 12 weeks and by £8 in the case of treatment at 2 weeks post-trauma. Discounting the benefits had no effect on the whole number of treatment-related recoveries on which the cost-effectiveness analyses were based.
7.10.3. Results
The results of this illustrative analysis are presented in and .
Frequency of recovery and costs at various treatment times.
Results of the cost-effectiveness analysis of early versus late intervention.
7.10.4. Summary
Assuming a remission is worth more than £2420, then cognitive–behavioural therapy at 12 weeks is the most cost-effective option. Achieving faster recoveries by treating early, however, may provide intangible benefits to those who suffer severe initial PTSD symptoms, particularly by preventing the conditions from becoming chronic. Future early versus late intervention studies should include a waiting list control in order to reduce the uncertainty associated with similar treatments (e.g. psychological debriefing, cognitive–behavioural therapies, self-help booklets and repeated assessments).
7.10.5. Conclusions
The findings of this study strengthen the impression that treating patients with cognitive–behavioural therapy at 12 weeks after the traumatic event is cost-effective, assuming a willngness-to-pay threshold of £2420 per additional treatment-related remission. Of course, caution must be exercised, because the estimates are based on small sample sizes and a select range of traumas. Indeed, the indirect and social costs of treating patients later may present hidden costs that are not included in this analysis (see section 2.7). Further research is needed to evaluate the cost of investing in early PTSD screening and prevention methods compared with the cost of treating PTSD at a later stage. This should be trial-based, where all costing and effectiveness parameters are estimated from data prospectively collected within an RCT. Rigorous health economic evaluations are needed to ensure interventions are cost-effective and equitable in preventing relapses across a wide range of the population.
This analysis does not account for fluctuations in treatment and/or behavioural effects between the two referenced studies. A sensitivity analysis demonstrated that a 10% increase or decrease in caseness, treatment-related recoveries or cost of treatment did not alter the conclusions. Nevertheless, there are certain confounding variables that may limit economic evaluations of the existing samples. These include the type and subtype of trauma; gender, age and other socio-demographic variables; differential study retention and withdrawal rates; differential inclusion criteria; comorbidity factors; time from trauma to presentation and treatment; type of treatment and skill of the person delivering treatment; number and length of sessions; the possibility that some or all of the chronic cases in the 12 week cohort might have recovered equally if given only the number of treatments as in the 2 week cohort; and the fact that patients needed to agree to randomisation. More importantly, this economic evaluation included neither the intangible cost of suffering nor the added financial cost to the patient in terms of work absence and lost opportunities as a consequence.
7.11. Clinical practice recommendations
7.11.1. Immediate interventions for all survivors of traumatic incidents
- 7.11.1.1.
All health and social care workers should be aware of the psychological impact of traumatic incidents in their immediate post-incident care of survivors and offer practical, social and emotional support to those involved.
[GPP]
- 7.11.1.2.
For individuals who have experienced a traumatic event, the systematic provision to that individual alone of brief, single-session interventions (often referred to as debriefing) that focus on the traumatic incident should not be routine practice when delivering services.
[A]
- 7.11.1.3.
Drug treatment may be considered in the acute phase of PTSD for the management of sleep disturbance. In this case, hypnotic medication may be appropriate for short-term use but if longer-term drug treatment is required, consideration should also be given to the use of suitable antidepressants at an early stage in order to reduce the later risk of dependence.
[C]
7.11.2. Early interventions for acute PTSD
- 7.11.2.1.
Trauma-focused cognitive–behavioural therapy should be offered to those with severe post-traumatic symptoms or with severe PTSD in the first month after the traumatic event. These treatments should normally be provided on an individual out-patient basis.
[B]
- 7.11.2.2.
Trauma-focused cognitive–behavioural therapy should be offered to people who present with PTSD within 3 months of a traumatic event.
[A]
- 7.11.2.3.
The duration of the trauma-focused cognitive–behavioural therapy should normally be 8–12 sessions, but if the treatment starts in the first month after the event, fewer sessions (about 5) may be sufficient. When the trauma is discussed in the treatment session, longer sessions (for example, 90 min) are usually necessary. Treatment should be regular and continuous (usually at least once a week) and should be delivered by the same person.
[B]
- 7.11.2.4.
Non-trauma-focused interventions such as relaxation or non-directive therapy, which do not address traumatic memories, should not routinely be offered to people who present with PTSD symptoms within 3 months of a traumatic event.
[B]
