Self-Efficacy as a Mechanism of Action of Imagery Rehearsal Therapy’s Efficacy : An
Exploratory Mediation Analysis
L'auto-efficacité comme mécanisme d'action de l'efficacité de la thérapie de révision et
répétition par imagerie mentale : une analyse de médiation exploratoire
Résumé
La Révision et Répétition par Imagerie Mentale (RRIM) est une thérapie cognitive- comportementale pour les cauchemars récurrents qui a été validée empiriquement, mais dont on ignore le mécanisme d'action. Certaines observations permettent de croire qu'une augmentation du sentiment de contrôle pourrait expliquer l'efficacité de la RRIM, mais cette hypothèse nécessite d'être testée. La Perception d'efficacité personnelle (PEP) est un concept connu et bien supporté empiriquement qui pourrait être utilisé pour opérationnaliser l'hypothèse du sentiment de contrôle. L'objectif de cette étude est de démontrer comment une mesure de PEP peut être un médiateur de l'effet de la RRIM sur les symptômes nocturnes. Trente-cinq victimes d'agression sexuelle souffrant d'un trouble de stress post-traumatique et de cauchemars récurrents souhaitant prendre part à une thérapie ont été assignées aléatoirement à suivre la RRIM ou être dans une condition d'attente. Les participants ont complété des questionnaires sur leurs symptômes nocturnes et leur PEP avant et après la thérapie ou la période d'attente. Des analyses de régression montrent qu'avoir suivi la RRIM prédit significativement l'amélioration de l'insomnie (β = -.549, t(34) = -4.230, p = .000) et de la qualité du sommeil (β = -.430, t(34) = -3.052, p = .005) à la fin de la thérapie, corroborant l'efficacité de ce traitement. La PEP serait un médiateur dans ces relations, puisqu'il a été démontré que la RRIM prédit une plus grande PEP liée aux rêves (β = .581, t(34) = 4.518, p = .000) et aux cauchemars (β = .384, t(34) = 2.744, p = .010), qui, à leur tour, seraient respectivement des prédicteurs de l'insomnie (β = -.365, t(34) = -2.388, p = .023) et de la qualité du sommeil (β = -.405, t(34) = -3.142, p = .004). Dans ces modèles de régression, les médiations sont partielles, puisque la RRIM continue de prédire directement la qualité du sommeil et l'insomnie, indépendamment de l'effet de la PEP. Ces résultats suggèrent que la RRIM améliore la qualité du sommeil et réduit l'insomnie en partie en augmentant la conviction des patients envers leur contrôle sur leurs rêves et cauchemars. Ces résultats montrent qu'une mesure de PEP, si elle est spécifique aux rêves et cauchemars, est pertinente pour rendre compte du mécanisme d'action de sentiment de contrôle de la RRIM. Ils soulèvent néanmoins la portée limitée de ce mécanisme pour expliquer toute l'efficacité de la thérapie.
Abstract
Imagery Rehearsal Therapy (IRT) is an empirically validated cognitive behavior therapy targeting recurring nightmares, for which the mechanisms of action (MA) remain poorly understood. Some observations suggest that an increase in sense of mastery could explain IRT’s efficacy, but this hypothesis remains to be empirically tested. Self-efficacy (SE) is a widely used and well-validated construct that could operationalize the more abstractly-defined sense of mastery concept. The objective of this study was to demonstrate how SE could mediate the effect of IRT on nocturnal symptoms. Thirty-five treatment-seeking victims of sexual assault with posttraumatic stress disorder (PTSD) and recurring nightmares were randomly assigned to either IRT or a waiting condition. Participants completed questionnaires about SE and nocturnal symptoms (insomnia, sleep quality, nocturnal symptoms associated with PTSD, nightmare distress and nightmare frequency), before and after either therapy or a waiting period. Regression analyses show that IRT significantly predicts improvements in insomnia (β = -.549, t(34) = -4.230, p = .000) and sleep quality (β = -.430, t(34) = -3.052, p = .005) at posttreatment, supporting IRT’s efficacy. SE mediates these relationships, as IRT was shown to predict greater SE about dreams (β = .581, t(34) = 4.518, p = .000), which in turn decreases insomnia (β = -.365, t(34) = -2.388, p = .023), as well as greater SE about nightmares (β = .384, t(34) = 2.744, p = .010), which accounts for variance in sleep quality (β = -.405, t(34) = -3.142, p = .004). In these regression models, mediation is partial, as IRT continues to directly improve sleep quality and insomnia independently of SE’s effect. Results suggest that IRT decreases insomnia and improves sleep quality in part through increasing the patients’ conviction that they have control over their dreams and nightmares. These results show that using a SE scale, albeit one specific to dreams and nightmares, is relevant to studying sense of mastery as a mechanism of action of IRT. However, they expose the limits of this MA’s ability to entirely explain IRT’s efficacy.
Repetitive and frequent nightmares are a major complaint of patients with posttraumatic stress disorder (PTSD) (Spoormaker, Schredl, & van den Bout, 2006). A review showed that 50 to 70% of PTSD patients have nightmares (Spoormaker et al., 2006), and results from a treatment study of sexual assault victims (SAV) showed an average frequency of 6 nightmares/week, four nights/week (Krakow et al., 2000). Repeated nightmares have been associated with high levels of anxiety, depression and PTSD symptoms in SAV with PTSD (Krakow et al., 2002). Nightmare distress is higher among individuals with PTSD and is not explained by other PTSD symptoms (Davis, Byrd, Rhudy, & Wright, 2007), underlining the necessity for treatments which target nightmares specifically.
Imagery rehearsal therapy (IRT) is a recent cognitive behavior therapy (CBT) targeting recurring nightmares that combines rescripting a nightmare with visualising the new script every day (Krakow & Zadra, 2010). A meta-analysis of 11 studies, 5 of which were randomized clinical trials, a combined total of 511 participants, have shown large effect sizes for this technique’s efficacy at improving nightmare frequency, sleep quality, and PTSD symptoms among many different populations with nightmares; improvements were sustained for up to one year following end of treatment (Casement & Swanson, 2012). IRT is recommended as a first line of treatment for chronic nightmares (Aurora et al., 2010). It is of short duration and produces rapid results, namely decreased nightmare occurrence two to eight weeks following the beginning of treatment among SAV (Krakow et al., 2000). Significant clinical change in nightmare frequency is observed in about 70 to 90% of patients (Krakow & Zadra, 2010); in a sample of veterans reporting combat-related nightmares, a complete response to treatment (nightmare frequency becoming null) was observed in 23% of participants (Nappi, Drummond, Thorp, & McQuaid, 2010).
However, IRT lacks a supported theoretical model of its efficacy (Nappi, Drummond, & Hall, 2012). Several hypotheses explaining the efficacy of nightmare treatments have been put forth, but supporting empirical data is lacking (Nappi et al., 2012). Additionally, many IRT variants exist and, thus far, few dismantling studies have examined the contributions of various therapeutic components, rendering the choice of which approach to favour unclear (Harb et al., 2013). Thus, a common vision of IRT’s mechanisms of action (MAs) appears necessary to justify new treatment protocols, efficiently orient future research and help clinicians adapt treatment to their patient’s characteristics.
The first illustration of the different variables involved in the treatment of nightmares suggests that increased sense of mastery, modified beliefs, decreased arousal and prevention of avoidance all lead to activation and modification of the fear structure related with the nightmare phenomenon, allowing restoration of sleep functions, which consists in regulation of arousal,
integration of incompatible information in dreams and sleep continuity (Rousseau & Belleville, in press). This theory was developed through a systematic review of studies on nightmare treatments and a qualitative analysis of the explanations offered by authors for their results; nonetheless, the review highlighted the need for empirical data supporting these various explanations.
The most frequently cited MA for IRT's efficacy was an increased sense of mastery (Rousseau & Belleville, in press). Sense of mastery is defined as the conviction patients have that they can control their dreams and nightmares; this feeling would be gained through concrete experiences of control, mainly through application of therapeutic techniques. Germain and her colleagues (2004) have operationalized and measured sense of mastery, through a coding grid they developed. They compared verbatim transcripts of nightmare scenarios to the new scripts participants undergoing IRT created. They observed that new scripts had more mastery elements than original nightmares, even if incorporating elements of mastery wasn’t suggested in therapy. New scripts had a wider range of control modalities than nightmares as well as more positive and less negative elements (Germain et al., 2004). However, whether the observed increase in mastery was generalized beyond therapy remains unknown. Moreover, mastery was measured only once and its association with IRT's efficacy was not investigated, which limits its interpretation as a MA. Finally, Germain et al.’s (2004) measure is not easily replicable, as it involves qualitative coding of scenarios.
Two recent studies reported exploratory results linking a measure of mastery to a nightmare treatment outcome. Harb, Brownlow, and Ross (2016) showed that IRT allows patients to more frequently feel that they are in control of their dreams’ content (a lucid dreaming construct), which was associated with a decrease in nightmare distress. Miller, Davis, and Balliett (2014) observed that
Exposure, relaxation, and rescripting therapy (
ERRT) increased internal locus of control with regards to sleep and nightmares, rather than general locus of control, which is thought to be a more stable trait. This increase in internal locus of control was associated with a greater reduction in sleep latency. Although the correlational nature of the findings and the absence of a control group, amongst others, do not allow for conclusions about mediation to be drawn, there seems to be convergent results in favor of a sense of mastery MA, despite the fact that it is operationalized in several different ways. Miller et al. (2014) suggested that a measure of self-efficacy (SE) would allow assessment of whether patients attribute outcomes to their own capacity to achieve behaviors likely to impact nightmares.SE is a concept similar to sense of mastery, but it is more widely used in various domains of psychology; it is the perceived capacity to manage one’s own behaviors, cognitions and motivation to exert control on externals demands, and the perception that one’s actions will have the desired
effects (Albert Bandura, 1977; Maddux, 1995). Following a traumatic experience, a higher level of SE is correlated with less severe posttraumatic symptoms (arousal, avoidance behaviors, thoughts, etc.) (Benight & Bandura, 2004). Since nightmares can be a PTSD-related symptom, self-efficacy may be a valid and reliable construct to designate sense of mastery: treatment for nightmares may thus work via an increase in self-efficacy, namely the perceived capacity of patients to exert control over their dreams and nightmares. Using a SE measure in conjunction with a measure of nightmare symptoms could be a way to measure sense of mastery and gain insight into the IRT’s MA in PTSD patients.
The present study’s objective is to demonstrate how SE could mediate the effect of IRT on the nocturnal symptoms of PTSD patients experiencing recurring nightmares. Following the recommendations of Hayes (2013) on mediation analyses, it was hypothesized that receiving IRT would predict (1) a decrease in nocturnal symptoms and (2) higher SE over dreams and nightmares at posttreatment, and (3) that higher SE would account for the decrease in nocturnal symptoms.
Method
Participants and procedure.
SAV with a PTSD diagnosis and recurring nightmares were recruited through e-mail in the Laval University community, posters in waiting rooms of medical or psychological clinics, and referrals from a SAV help center in Quebec city, Canada. All participants completed a psychological evaluation assessing their eligibility. Inclusion criteria were: (a) being 18 years old or older; (b) being able to understand and communicate in French; (c) a PTSD according to the DSM- IV criteria (American Psychiatric Association, 2000)
(diagnosed at initial clinical assessment)
; (d) having been sexually abused; (e) having stable medication intake during the three months prior to beginning therapy; (f) reporting sleep difficulties (score ≥ 5 on the Pittsburgh Sleep Quality Index); (g) reporting at least one nightmare per week during the last month. A nightmare was defined according to DSM-5 criteria (American Psychiatric Association, 2013), i.e., an extremely dysphoric, and well-remembered dream that usually involves efforts to avoid threats to survival, security or physical integrity, and causing clinically significant distress or impairment. Exclusion criteria included (a) alcohol or substance abuse/dependence; (b) past or present psychotic episode(s), bipolar disorder or organic mental disorder; (c) suicidal ideas requiring immediate intervention; (d) having followed CBT in the last year; and (e) use of medication that affects nightmares (e.g. prazosin). Participants consented not to follow other forms of therapy during the study or to begin/modify medication intake.A total of 131 individuals contacted the research team and 123 were screened by telephone interview. Of these, 53 appeared eligible and were invited for a thorough initial clinical assessment. Of the 49 who underwent the interview, 43 met inclusion criteria. 22 were randomly assigned to the treatment condition and 21 to the control condition. 3 participants withdrew from the treatment condition because they were no longer available or interested in the study. In the control condition, 1 participant was excluded because of the apparition of psychotic symptoms, 3 for not answering SE questions and 1 for requesting his results be removed from the study. The final sample thus included 35 participants (treatment condition, n = 19; control condition, n = 16) (Figure 7).
Upon confirmation of their eligibility, participants received a participant number and were randomly assigned to one of the two treatment conditions, either a 5-week waiting period or IRT, after which all participants were offered 15 weeks of PTSD CBT (data not reported here). Randomization was carried out by a research assistant not associated with the project: she produced sealed and numbered envelopes containing assignment to a condition and opened them upon confirmation of a new participant number assignation. A series of questionnaires, including questions on SE and nocturnal symptoms, was completed at the eligibility assessment (pre- treatment) and one week after the 5-week treatment/waiting period (posttreatment).
IRT was
offered to participants in the control condition at the end of the study.
Measures.
Clinician-Administered PTSD Scale.
The Clinician-administered PTSD scale (CAPS; Blake et al., 1995) is a semi-structured interview allowing diagnosis and evaluation of PTSD’s severity in about 45 minutes. It assesses 17 PTSD symptoms as presented in the DSM-IV-TR by measuring the frequency and severity of each on a 4-point Likert scale. The CAPS has good test-retest reliability (.77 to .96), internal consistency (.85 to .87), sensitivity (.84), specificity (.95) and interrater agreement (.78) (Blake et al., 1995).
Structured Clinical Interview for DSM-IV.
The structured clinical interview for DSM-IV (SCID-I; First, Spitzer, Gibbon, & Williams, 1996) is a semi-structured interview allowing diagnosis of disorders of the DSM-IV-TR (mood disorders, psychotic disorders, substance use disorders, anxiety disorders, somatic and eating disorders). The SCID-I has excellent interrater agreement (First & Gibbon, 2004), specificity and sensitivity (Basco et al., 2000). In the present study, this interview was used to verify the presence
of comorbid disorders and possible exclusion criteria; the section on PTSD was not administered, as it was already covered by the CAPS. A non-validated French version was used.
The Pittsburgh Sleep Quality Index.
The Pittsburgh sleep quality index (PSQI) (Buysse, Reynolds, Monk, Berman, & et al., 1989) is a 19-item self-report questionnaire that assesses seven components of sleep quality in the last month: (a) subjective sleep quality; (b) sleep latency; (c) sleep duration; (d) sleep efficiency; (e) sleep disturbances; (f) use of sleep medication; and (g) impairment of daytime functioning. Each component or subscale is rated on a scale from 0 to 3, with higher scores indicating greater problem severity. A global sleep quality score ranging from 0 to 21 is calculated by adding the seven component scores. The PSQI has demonstrated good internal consistency, test-retest reliability, and overall validity in validation studies with individuals suffering from insomnia or other psychiatric and medical disorders (Buysse et al., 1989). Validation of the French version has yielded satisfactory results (Blais, Gendron, Mimeault, & Morin, 1997).
The PSQI Addendum for PTSD.
The PSQI addendum for PTSD (PSQI-A) (Germain, Hall, Krakow, Shear, & Buysse, 2005) is an extension of the PSQI designed to assess the frequency of seven disruptive nocturnal behaviors: (a) hot flashes; (b) general nervousness; (c) memories or nightmares of a traumatic experience; (d) severe anxiety or panic not related to traumatic memories; (e) bad dreams unrelated to traumatic memories; (f) episodes of terror or screaming during sleep without fully awakening; and (g) episodes of “ acting out ” dreams, such as kicking, punching, running, or screaming. Each item is rated on a 0 - 3 scale referring to the frequency of each disturbance, where 0 = not in the past month, 1 = less than once a week, 2 = once or twice a week, and 3 = three or more times a week. A global score of 0 to 21 can be obtained from the sum of all seven items. PSQI-A has good internal consistency, sensitivity and specificity. Validation of the French version has shown similar psychometrics results (Ait-Aoudia et al., 2013).
Nightmare Distress Questionnaire and Nightmare Frequency.
The Nightmare distress questionnaire (NDQ; Belicki, 1992a) is a self-report questionnaire that assesses 13 indices of waking suffering associated with nightmares in the last month (e.g., being afraid to fall asleep for fear of having a nightmare, considering that nightmares interfere with one's well-being). Each item is rated on a 0 - 4 frequency scale (from 0 = never, to 4 = always). The last two items measure interest in nightmare therapy, and thus were not included in the total. A global score of 0 to 44 was calculated, with higher scores indicating more distress. NDQ has
excellent internal consistency (Belicki, 1992a), reliability and validity (Belicki, 1992b). A non- validated French version was used. In order to measure nightmare frequency, an additional question enquiring about number of nightmares in the past week was included.
Insomnia Severity Index.
The Insomnia severity index (ISI; Blais et al., 1997) is a self-report questionnaire that assesses 7 aspects of sleep difficulties: (a) failing asleep, nocturnal awakenings and morning awakenings; (b) satisfaction with sleep; (c) interference with daily functioning; (d) noticeability; and (e) worries about sleep. Each item is rated on a 0 - 4 scale (0 = no problem, 4 = extreme difficulty). A global score of 0 to 28 can be obtained, with higher scores indicating more severe insomnia symptoms. The French version of the ISI has good internal consistency and test-retest reliability (Blais et al., 1997).
Self-Efficacy.
The Self-efficacy (SE) scale of nightmares was created following the recommendations of Bandura (2006). The SE questions are answered on a scale from 0 to 100 (0 = cannot do at all, 50 = moderately certain can do, 100 = highly certain can do). Three questions were created; two about dreams and nightmares, and one about other PTSD symptoms. The latter was created to verify the specificity of the two former questions, as few or no changes were expected during the specific treatment targeting nightmares. The two questions on dreams and nightmares were: To what extent do you feel able to control the content of your dreams? and To what extent do you feel able to overcome your nightmares? The question on PTSD symptoms was: To what extent do you feel able to overcome the symptoms associated with the traumatic event?
Treatment.
Trained graduate psychology students delivered individual treatment for nightmares. Treatment delivery was supervised by a psychologist specialized in CBT for nightmares (G. Belleville). A standardized treatment manual outlined the material to be covered in each session and guided therapists in the application of various therapeutic strategies. The treatment manual developed by Nappi, Drummond, Thorp, and McQuaid (2010) was translated to French for this study (Belleville and Dube-Frenette, not published). Therapists were instructed to give 5 50-minute sessions of IRT. The IRT protocol included: (1) psychoeducation about nightmares, (2) practice of pleasant mental imagery, (3) rescripting a nightmare, (4) mental imagery of the new script and (5) relapse prevention. Participants received a manual that covered the content of each session. Each
therapy session further included several homework assignments. Participants were encouraged to observe and record nightmare frequency and distress level each morning in a nightmare diary, as well as sleep quality and possible events of the previous day that could have increased the probability of nightmare occurrence.
Participants randomly assigned to the control condition received a phone call from their therapist once a week to begin forming a therapeutic alliance and help manage their distress through empathic listening. Therapists were instructed not to discuss nightmares and participants did not keep nightmare diaries.
Treatment integrity was assessed by independent judges
(undergraduate psychology
students)
on a sample of 4 participants’ session recordings (22% of participants in the experimental condition). All (100%) of the strategies used by therapists (techniques used, reading and exercises asked) corresponded to those in the treatment manual. No allusion to SE was made during treatment.Data analyses.
Statistical analyses were performed with SPSS 19 for Windows. In order to ascertain whether SE was a mediating variable, we conducted a series of regression analyses for each dependent variable, following Hayes’ (2013) recommendations. Experimental condition was defined as the independent variable (0 = control condition; 1 = treatment condition), and the dependent variables were nocturnal symptoms at post-treatment (PSQI, PSQI-A, ISI, NDQ, nightmare frequency). The mediating variable could be the score on one of the SE questions at post- treatment: SE to control dreams, SE to overcome nightmares and SE to overcome PTSD symptoms. To control for interindividual variability, SE and nocturnal symptom scores at pre-treatment were included as covariates.
To assess a mediation effect, a series of regressions were performed. Following an enter