Psychosocial treatments for nightmares in adults and children: a systematic review

There is also extensive research on drug treatments for nightmares. According to Morgenthaler et al., [19] pharmacologic treatment may be slightly more effective than psychosocial treatments. They also suggest that there is conjecture over the lasting effects of these drugs once withdrawn, and that we are still unclear about the underlying pathophysiology of nightmares. There is also a need to better understand interaction effects between anti-depressants and nightmare specific drugs such as prazosin. The AASM currently suggest that the following drugs may be beneficial for treating nightmares; the atypical antipsychotics olanzapine, risperidone, and aripiprazole; clonidine; cyproheptadine; fluvoxamine; gabapentin; nabilone; phenelzine; prazosin; topiramate; trazodone; triazolam; nitazepam; and tricyclic antidepressants. One of the most widely studied, reviewed and utilized treatment drugs for nightmares is the alpha-1 adrenergic blocker prazosin [22]. The British Association for Psychopharmacology review of treatments stated that there was good evidence for the use of prazosin for reducing nightmares in adult and pediatric populations [20]. However, while still supporting its use, the AASM downgraded the effectiveness of prazosin for nightmares in 2018.

Suraev et al [23] and Betthauser et al [24] systematically reviewed cannabinoid therapies for managing sleep disorders and despite some promising preliminary evidence, both studies suggested that there is currently an absence of high quality clinical trials to support its use. Cowling and MacDougall [25] reported that the synthetic cannabinoid nabilone could reduce PTSD related nightmares. Dagan and Yager [26] argued that while medical cannabis could reduce nightmares, it may have negative effects on other PTSD symptoms such as dissociation and reckless behaviours. There is also some evidence for the use of anti-depressants in treating nightmares. A systematic review of the links between dreaming and anti-depressants found differing effects depending on the type of anti-depressant taken, including withdrawal effects, and concluded that more research is needed to draw conclusions [27]. In addition, side effects of some anti-depressents include an increase in nightmares. In summary, there is limited evidence supporting the use of cannabis or anti-depressants for nightmares. However, as an emerging area of investigation, an up to date review of research findings is needed.

In summary, treatment for nightmares and nightmare disorder include psychotherapeutic treatments such as image rehearsal therapy, pharmacological interventions such as prazosin and nitrazepam that affect the neurotransmitters, and atypical antipsychotics such as olanzapine. In recent times there have been numerous studies that have evaluated the effectiveness of treatments for nightmares, resulting in some changes to understanding of best practice. For example, some psychosocial treatments recently trialed include eye movement desensitization and reprocessing, and sleep dynamic therapy. In 2018, the AASM released a position paper (rather than a clinical practice guide) due to the limited number of high quality studies providing direct evidence of treatments for nightmares [19]. In addition, many studies in this area have evaluated treatments for sleep related outcomes in general, rather than treatments for nightmares specifically. There is also a lack of a universal outcome measure for nightmares which makes comparing study results difficult. As such there is a need to systematically review findings related to the effectiveness of treatment of nightmares to inform best practice. This current study reviewed psychosocial treatments for nightmares and adds to the review literature by including non-RCTs, studies on children, and a focus on newer (for nightmares) psychosocial treatments. We aimed to answer the following questions:

What is the effectiveness of psychosocial treatments for nightmares in adults and children?

What are the new promising psychosocial nightmare treatments for adults and/ or children requiring further investigation?

An initial search was conducted to find current meta-analytic, systematic review, and review articles using SCOPUS (2000-Present), PsychINFO (2000-Present), and MEDLINE (2000- Present) databases and the search terms “nightmare AND review OR analysis”. This informed the focus of the current review and the information presented in the introduction section of this report. A search was performed to collect relevant studies for the systematic review. The databases SCOPUS, PsychINFO, CINAHL, and MEDLINE were searched using the search terms “nightmare AND therapy OR treatment OR intervention”. Relevant database subject heading search terms were also included. In addition, the Cochrane data base was searched for trials (n = 582) and reviews (n = 65) using the search term “nightmare”.

Reference lists of more recent studies were screened for studies not picked up by the search. The final search was performed on August 30, 2020.

The following inclusion criteria was used to screen studies:

Excluded from the review were review articles, theoretical or commentary articles, books, audio documents, posters, symposiums, and classification manuals. For this manuscript, only studies evaluating psychosocial treatments were presented.

The first author reviewed all eligible studies. Studies were screened at two stages, title and abstract, and full text. Eligibility assessment was performed independently in a blind standardized manner by two reviewers for title and abstract, and one reviewer for full text using the software COVIDENCE. Overall agreement rates were 84.5%, for title and abstract screening with discrepancies solved through discussion in regular meetings between the two reviewers.

Data were extracted by two reviewers, and included the following:

A process for assessing bias within RCT studies was formulated according to Cochrane recommendations [29] and using the COVIDENCE software template. For RCTs an assessment for each study was made and reported as either low risk, high risk or unclear risk;

For non-RCTs, several tools were considered including the Risk of Bias in Non-Randomized Studies—of Interventions (ROBINS-I) [30], and the Joanna Briggs Institute (JBI) critical appraisal checklist for case reports [31]. However, due to the variability of study designs included, a selection of relevant criteria were assessed for each study and reported as either low risk, high risk, or unclear risk. These assessments were made based on a comparison to the other non-RCTs in this collection rather than in comparison to the stringent RCT expectations.

The percentage of published articles with significant findings were also considered along with collective estimates of conflicts of interest and funding sources.

Data synthesis was performed narratively. Due to the diverse outcome measures, treatments, and samples evident in the literature on nightmares, a focused narrative approach allowed for the best synthesis of the data. Studies were grouped into main categories in relation to the type of treatment, including group or individual, type of psychosocial treatment, and also by age, adults and children. Further sub-categories were devised for grouping treatment types; CBT based, Exposure based, Image rehearsal/rescripting based. Once all articles were grouped into categories, the collective efficacy was assessed. This process was completed in discussions between two of the researchers and resulted in therapies being classified at low, medium, or high on quantity of evidence, quality of evidence, and support for use. This process was informed by the effects reported and the risk of bias. A conclusion was generated for all groups and sub-groups and then reported.

The primary change to this study was in study selection. Full text reviewing was conducted by one reviewer rather than two reviewers. This occurred due to the withdrawal of one of the reviewers from the research. It was decided to proceed with the primary reviewer alone rather than include a new reviewer who was less familiar with the studies. Due to the large number of studies included, the decision was made to report the study in two separate papers (psychosocial and pharmacological). We acknowledge that comparisons between the two broad treatment modalities are important, and will be discussed in a subsequent manuscript.

The strategy for literature search and selection is outlined in Fig. 1. Title and abstract review resulted in 454 eligible studies. After excluding 389 studies for the reasons listed in Fig. 1, there remained 65 studies for the qualitative synthesis. The 35 RCT studies comprised a combined 3048 participants (F = 1940, M = 1108), with mean ages ranging from 10 to 59 years. The 34 non-RCT studies comprised a combined 500 participants (F = 170, M = 327, Transgender = 3) with ages ranging from 4 to 71 years. Only two RCTs and 7 non-RCTs included children and/or adolescents. Most RCT studies originated from either the USA (N=19) or the Netherlands (N=7), and 12 of the studies had military or military veteran samples. For the non-RCT studies, most studies originated from the USA (N = 27) and had civilians with PTSD (N = 11) or military or military veteran sample with PTSD (N = 13). Participants were most commonly suffering from PTSD (N = approximately 1880) or idiopathic (N = approximately 450) nightmares. Female participants were commonly victims of sexual assault, while males commonly suffered from war related trauma. The most common therapies evaluated were image rehearsal/rescripting-based therapies (N= 34), exposure-based therapies (N= 8), mixed exposure and image rehearsal based therapies (N=11), and CBT based therapies (N=6). The interventions consisted of individual, group, and self- help/directed modalities, ranging from 1 to 20 sessions in total, with each session lasting between 60 and 180 minutes. Attrition rates for the 35 RCT studies ranged from zero to 46% from baseline to final measurement. For the non-RCT studies, attrition rates ranged from zero to 64.7%. Most studies included follow up measurement/s of between 4 weeks and 12 months. The nightmare related outcomes measured included frequency (total number, number of nights with), intensity, and related distress, measured by self-report or a standardized instrument. The majority of RCT studies assessed both within and between group effects using analysis of variance, whereas most of the non-RCTs did not use a comparison group, assessing within-subjects effects only.

One study (see Table 1) assessed the efficacy of treatment for children [32] and found that a self-help form of IRT significantly reduced nightmare frequency compared to a passive control (sustained at 9 month follow up). Another study assessed IRT with an adolescent (13 years to 18 years) female only cohort and found significant reductions in nights and nightmares per month compared to no significant changes in the control group [33].

Of the remaining studies that assessed forms of image rehearsal or rescripting therapies, all but one showed significant reductions in nightmare symptomatology. All of the exposure and exposure plus image rehearsal studies found significant symptom reduction. While the CBT and CBT-I interventions were found to be effective, two of the five studies found exposure or IRT treatments to be superior. The two studies of cognitive processing therapy (CPT) showed significant symptom reduction, while the two studies of lucid dreaming therapy (LDT) showed inferior results to exposure therapy. Psychosocial therapies outperformed passive controls in most studies and were at least equivalent to pharmacological interventions. Group and individual therapies were shown to be equally effective across a range of treatment types, and all but one self-guided therapy study produced significant symptom reduction. Treatment effects were sustained or increased at follow-up in most studies. Effects appeared independent of the sample presentation (idiopathic or part of a more complex presentation).

As shown in Table 2, all studies randomly assigned participants to treatment groups, although nearly half of the studies failed to report the randomization technique in sufficient detail for replication. The majority of studies blinded participants, however, several studies reported problems with the blinding process, which may have resulted in participants being aware of their group allocation. All studies were single blinded based on the difficulty of blinding clinicians to the treatment they deliver. There was little evidence of incomplete outcome data and selective reporting, and the majority of studies blinded the outcome assessment. It was noted that many studies failed to report between group effects in the abstract and in concluding, reporting just the within group changes. In a minority of studies, this gave a false impression that reported effects were based on comparisons to the control group. A minority of studies failed to report attrition rates.

The studies reviewed (see Table 3) included 7 case studies, 10 case series, and 16 before and after studies, of which one utilised a control group for part of the study [68]. The case studies were mostly medium (5 – 10 sessions) to long (15+ sessions) in length of treatment, whilst the case series were mostly brief (1- 2 sessions) or short (3 - 4 sessions). The before and after studies included a mixture of different lengths. Effectiveness comparisons by length of treatment did not reveal notable findings. All 34 studies reported an improvement in one or more nightmare symptom, with half of these studies conducting analyses of significance.

The image rehearsal or rescripting therapies were the most common form of intervention used. Half of these rescripting studies also included CBT-I or select features (e.g., stimulus control, sleep restriction, and/or sleep hygiene) in their treatment delivery. Significant reductions in various nightmare symptomatology were reported in half of the rescripting studies. Six of the remaining studies reported decreases in symptomology however did not conduct analyses of significance. One of these also reported a slight increase in mean frequency of nightmares whilst observing decreases in nightmare distress, intensity, and vividness [69]. The final study found non-significant improvements, however measured dreams rather than nightmares specifically [81]. The only study delivering CBT-I alone provided weak positive results  [79].

Two of the six studies employing mixed exposure and image rehearsal-based therapies through ERRT reported significant improvements in nightmare symptomology. The remaining four studies reported improvements, without measuring statistical significance. Kovacevic and Davis [86] used CPT following ERRT and found nightmare frequency and severity reduced following ERRT and then further decreased until cessation during CPT. Of the two exposure-only studies, the self-directed study found that significant reductions were maintained over 4 years of follow-ups [82] and the other study reported an extinction of nightmares maintained at 1 and 3month follow-ups but did not conduct analyses of significance [72].

The two EMDR studies [98, 101] reported improvements in nightmare symptomology, however, both were third person accounts (mother and doctor), with no formal statistics or analyses. The two hypnotherapy studies reported improvements, however neither used analyses of significance. Furthermore, Hauri et al [85] post assessment did not separate nightmares participants from the other parasomnia participants. The embodied imagination case report [100] declared an extinction of nightmares however did not use statistics or follow up. The lucid dreaming therapy study reported a decrease in mean nightmare frequency, however, the number of participants who were successful in lucidly changing their dreams was the same as the number of participants whose dreams changed without lucidity.

Seven of the 34 studies explored treatments (3 IRT, 1 ERRT, 2 hypnotherapy, and 1 EMDR) in adolescents and children (aged 4 – 16 years). Nightmare symptoms improved in each study, however, Davis et al [75] and Fernandez et al [78] noted the need to rewrite the child’s script before improvements occurred. Interestingly, this coincided with the end of therapy; which meant no further therapist driven exposure work. Also noteworthy was that Hauri et al [85] used a mixed sample of children and adults and did not comment on child outcomes specifically.

Regarding sample presentations, three studies clearly stated the use of civilians with idiopathic nightmares, nine studies included posttraumatic civilian cohorts, and 12 studies included posttraumatic military cohorts. The remaining nine studies consisted of cohorts which were either mixed (idiopathic and posttraumatic), unclear (did not state civilian/military and/or idiopathic/posttraumatic), or complex (e.g., psychiatric) in their presentations. There was a mixture of therapies amongst the different cohorts however, the posttraumatic cohorts (civilian and military) mostly received IRT with CBT-I or its sleep features.

Finally, many studies collected formal follow up data (n = 16), whilst others used informal follow up through third parties (e.g., doctor/parent report; n = 2). In most of these studies, treatment gains were sustained or increased. This left 15 studies that did not employ any follow up.

As shown in Table 4, the majority of studies included participants with the common baseline frequency of nightmares of at least one per week. Those rated as high either used vague terms such as “frequent” or “recurrent” to describe baseline nightmare frequency or included participants with less than 1 nightmare per week, or without nightmares at all. Most studies did not comment on whether participants were receiving concurrent psychological or pharmacological support or only addressed one of these elements. Most studies did not explicitly report on blinding of outcome assessors, or reported that the outcome assessor was also involved in the delivery of the treatment and the in-session assessments. No other blinding was reported. Regarding selective outcome reporting, studies were mixed in their attention to reporting on all outcomes and dropouts. Finally, in considering data sampling in the context of this group of non-RCTs, there was a mixed response. Data collection and reporting methods were considered, with several studies failing to use formal statistics or collection methods. The frequency of data collection (e.g., daily/each session/follow-up), and sample size were considered as part of risk.