The use of benzodiazepines (BZDs) in the treatment of posttraumatic stress disorder (PTSD) is both common and controversial. Although BZDs are prescribed to 30% to 74% of patients with PTSD,1–2 there is little literature—and no reviews before this article—focusing exclusively on the use of BZDs to prevent or treat PTSD. Considering all the service members returning from Afghanistan and Iraq with combat-related PTSD, there is no better time to evaluate this topic than now.
Some argue that BZDs are effective symptomatic treatments for the anxiety, insomnia, and irritability associated with PTSD, and they defend the prescription of BZDs for PTSD as necessary for treatment-resistant patients with severe symptoms. Others contend that BZDs may diminish subjective anxiety in the short term at the cost of worsening other features of PTSD, such as promoting avoidance, in the long term. They explain the correlation between BZDs and increased symptom severity as the result of BZDs actually prolonging and worsening PTSD.
Most PTSD practice guidelines pay little attention to BZDs or caution against their use, citing weak evidence, risks that outweigh benefits, and contraindication for conditions that are commonly comorbid with PTSD such as traumatic brain injury (TBI) and substance use disorder (SUD). Some guidelines go further, declaring BZDs contraindicated for combat-related PTSD (Veterans Affairs and Department of Defense),3 traumatic grief (British National Formulary),4 and all PTSD (International Society for Traumatic Stress Studies).5
Despite the abundance of articles about PTSD and the frequent prescription of BZDs, little research has evaluated the use of BZDs for PTSD. In our literature review, we attempted to capture every available study about BZDs in PTSD to examine 3 questions:
What are the effects of BZDs on the development of PTSD in trauma patients?
What are the effects of BZDs on PTSD-associated outcomes in patients with PTSD?
What are the effects of BZDs on PTSD-associated outcomes in trauma patients with and without PTSD?
Studies were included for review using the following eligibility criteria:
Study design: clinical trials or observational studies.
Participants: any patient with a history of trauma assessed for PTSD.
Intervention: any dose, duration, or type of BZD.
Outcomes measured: PTSD-associated symptoms.
Studies were excluded if they were reviews or anecdotal or if BZDs were not distinguished from other medications.
The authors conducted electronic searches using PubMed, PsycINFO, MEDLINE, Cochrane Library, and Google Scholar (all the studies eventually selected are available in PubMed). Search parameters included all English-language articles published until June 30, 2014. Search terms included PTSD, stress disorder, benzodiazepine, and the generic names of the different BZDs. For example, the following search was used in PubMed:
((PTSD) OR (stress disorder)) AND ((benzodiazepine) OR (alprazolam) OR (chlordiazepoxide) OR (clonazepam) OR (clorazepate) OR (diazepam) OR (flurazepam) OR (lorazepam) OR (midazolam) OR (oxazepam) OR (temazepam) OR (triazolam)).
References in retrieved articles were further scanned for additional relevant articles. Duplicate articles were not counted in the total sample of identified records. Abstracts were screened for relevance. Full-text articles were retrieved to determine eligibility. Two authors independently determined eligibility, for which interobserver agreement was calculated using percent agreement and kappa statistics. Disagreements regarding eligibility were resolved by consensus among the authors. For each eligible study, 2 authors independently abstracted information concerning study characteristics.
The findings of the selected articles were categorized according to levels of scientific evidence based on clinical practice guidelines from the US Department of Health and Human Services6:
Multiple double-blind placebo-controlled trials and a confirmatory meta-analysis (in addition to level B of evidence).
At least 1 double-blind placebo-controlled trial (in addition to level C of evidence).
Anecdotal reports, case series, and open trials, in addition to expert endorsement or consensus.
Few case reports without any expert panel endorsement.
To evaluate evidence for an association between BZDs and PTSD, the following 3 null hypotheses were tested:
H1: BZDs are not associated with the development of PTSD in trauma patients.
H2: BZDs are not associated with PTSD-associated symptoms in patients with PTSD.
H3: BZDs are not associated with PTSD-associated symptoms in trauma patients with and without PTSD.
In each case, the hypothesis was tested using a meta-analysis carried out in the MetaEasy Excel add-in (http://www.statanalysis.co.uk/meta-analysis.html). Studies included in the meta-analysis were those that compared outcomes between a group of patients given BZDs and a control group. An estimate and 95% confidence interval (CI) for the standardized effect size (ES) was computed for each outcome in a study. To compute a single effect estimate for each study, MetaEasy uses the within-study median ES and confidence limits. Finally, a meta-analysis was used to test each hypothesis after pooling information over all of the relevant studies. Note that the meta-analysis for H3 is not simply pooling the other 2 meta-analyses, but includes all PTSD-related outcomes in addition to “PTSD diagnosis,” which was the only outcome measured in H1. Using a random-effects model (to account for the heterogeneity between studies), we estimated the ES and 95% CI associated with each hypothesis. All tests were 2 sided and at the 0.05 level of significance.
The authors reviewed 8422 citations, 249 abstracts, and 109 full-text articles. The selection process is illustrated in Figure 1 using the PRISMA flow diagram7 with reasons for exclusion. The percent agreement (kappa statistic) for eligibility was as follows: full-text review, 89.9% (κ=0.63); randomized-controlled trials (RCTs), 99.1% (κ=0.89); nonrandomized clinical trials, 100% (κ=1.00); observational studies, 97.7% (κ=0.63). The percent agreement for data abstraction was 76.1%.
After excluding ineligible articles, 18 studies were selected for review and are summarized in Table 1. Sample sizes varied from 6 to 2931, with a total of 5236 completers. Because not every study reported the same characteristics, the following numbers were calculated using available data for participants as summarized in Table 2. Participants survived 1 or more of the following: physical injuries (n=2979), life-threatening medical conditions (n=841), combat-related trauma (n=431), sexual trauma (n=277), disaster exposure (n=203), and other traumas (n=47). The majority of the participants (approximately 67%) survived a physical injury. The mean age was approximately 44 years and approximately 38% were women.
Table 3 compares study designs, outcomes, and levels of evidence.
The studies supporting BZD efficacy for PTSD demonstrate short-term improvement in sleep9 and anxiety.10
The studies demonstrating BZD inefficacy for PTSD demonstrate no significant improvement compared with controls for overall severity of PTSD symptoms,8–10,12,14,16–24 startle reflex,8,13,25 psychotherapy outcomes,8,17,23 depression,10,12 overall well-being,10 sleep,9,11 nightmares,11 anxiety,12,16 aggression,15–16 substance use,16 and social functioning.16
The studies showing BZDs being associated with worsened PTSD outcomes demonstrate worsened overall severity of PTSD symptoms,8–9,12,14,16–23 psychotherapy outcomes,8,17,23 depression,12 aggression,15–16 substance use,16 and anxiety.16
Twelve studies (4 RCTs, 2 nonrandomized, 6 observational) obtained data sufficient for estimating ESs for PTSD-associated symptoms. Individual outcome measures and their associated ESs and 95% CIs are shown in Table 4. Figure 2 summarizes the ESs and CIs in forest plots by study for hypotheses H1, H2, and H3. Results are presented so that a positive ES corresponds to BZDs improving PTSD-associated outcomes and a negative ES corresponds to BZDs worsening PTSD-associated outcomes. The row labeled “POOLED” contains the estimated ES and 95% CI for the meta-analysis. The row labeled “RCTs only” illustrates the results when restricted to only the 4 RCTs (denoted by “RCT” to the right of the corresponding CI).
The estimated ES of BZDs on the development of PTSD in trauma patients was −0.3974, with a 95% CI of (−0.6057, −0.1891). Thus, we reject H1 and conclude that BZDs increase the likelihood of developing PTSD when taken by trauma patients.
The estimated ES of BZDs on PTSD-associated symptoms in PTSD patients was −0.0839, with a 95% CI of (−0.3544, 0.1866). Thus, we have insufficient evidence to reject H2. That is, we have insufficient evidence to conclude that BZDs alleviate PTSD-associated symptoms when taken by patients who already have PTSD.
The estimated ES of BZDs on PTSD-associated symptoms in trauma patients with and without PTSD (here, “symptoms” includes all PTSD-related outcomes including PTSD diagnosis for trauma patients who were not previously diagnosed with PTSD) was −0.2798, with a 95% CI of (−0.3981, −0.1616). Thus, we reject H3 and conclude that BZDs have an overall adverse impact in the prevention and treatment of PTSD. When the analysis was restricted only to the 4 RCTs, the estimated ES of BZDs on PTSD-associated symptoms in trauma patients with and without PTSD was −0.0422, with a 95% CI of (−0.4505, 0.3661). Thus, the RCTs alone do not provide sufficient evidence to reject H3. On the basis of the RCTs alone, we conclude that there is no evidence that BZDs alleviate PTSD-associated symptoms in PTSD patients or prevent the development of PTSD in trauma patients.
Inefficacy (Level of Evidence A)
Before our study, the ceiling for the level of evidence for inefficacy was at B due to the lack of a confirmatory meta-analysis. However, this meta-analysis and at least 1 measure in every study that was reviewed, including all 4 RCTs of BZDs in PTSD,8–11 suggest a lack of efficacy of BZDs for PTSD. All PTSD-specific measures that were used, such as the Clinician-administered PTSD Scale (CAPS) and the PTSD Checklist, demonstrated that BZDs are, at best, not significantly different from placebo or no BZD for PTSD. BZD inefficacy is also endorsed by every available PTSD practice guideline. These findings are likely explained in part by the tolerance and cognitive effects associated with BZDs and also indicate that BZDs appear to inadequately target PTSD pathophysiology.
A major disadvantage of BZDs is that tolerance develops to hypnotic and myorelexant effects within days to weeks, and to anticonvulsant and anxiolytic effects within weeks to months.26–28 Therefore, BZDs are unlikely to be effective long-term hypnotics or anxiolytics, which is confirmed by several general studies of BZDs for sleep and anxiety.27,28 Tolerance to BZDs is a distinct problem in PTSD because most patients have symptoms that persist for longer than 3 months.29
BZDs may be ineffective for PTSD because of amnestic effects that unintentionally target learning how to cope with PTSD symptoms rather than traumatic memories. Although therapeutic effects decrease with tolerance, cognitive effects (ie, BZD-induced neurocognitive disorder) usually persist for attention, memory, and learning.28 Cognitive impairments are more common with long-term use and high doses, but they can also occur with short-term use and low doses.28,29 Unfortunately, PTSD is a risk factor for BZD-induced neurocognitive disorder, as are conditions that are often comorbid with PTSD such as SUD, neurocognitive disorders (including TBI), and psychotic, bipolar, and depressive disorders.29
BZDs may be ineffective for PTSD because the pathophysiology of PTSD differs from that of the anxiety disorders for which BZDs have some efficacy. Studies of flumazenil, which have demonstrated that GABA-receptor antagonism induces panic in patients with panic disorder but not in healthy controls or patients with PTSD,30,31 suggest that the pathophysiology underlying anxiety in PTSD is different from that in panic disorder despite experiential similarities. Researchers in 2 of the studies that were reviewed13,25 concluded that, while locus ceruleus dysregulation is implicated in both panic disorder and PTSD, the amygdala and hippocampus are also implicated in PTSD anxiety. Shalev et al13 speculated that these structures may be less responsive to BZDs than the locus ceruleus. In addition, rather than targeting specific implicated structures, BZDs indiscriminately depress global brain function (including structures such as the prefrontal cortex that are already hypoactive in PTSD and which, when functioning adequately, allow for various cognitive processes and modulation of the amygdala). Therefore, anxiety in PTSD may be different than anxiety in other disorders and may require different treatments.
Worsened Outcomes (Level of Evidence B)
Thirteen of the studies that were reviewed (including 2 RCTs), several practice guidelines, and some case reports suggest that BZDs have the risk of worsening the severity and prognosis of PTSD. All but 210,24 of the 13 studies that used PTSD-specific measures (eg, CAPS, PTSD Checklist) demonstrated that BZDs are associated with worse overall severity of symptoms when compared with placebo or no BZD. Potential biopsychosocial explanations for BZDs worsening PTSD outcomes include discontinuation symptoms, disruption of normal stress responses, avoidance of cognitive and emotional processing of trauma, and worsening of underlying PTSD pathophysiology (eg, effects on the hypothalamic-pituitary-adrenal [HPA] axis and on gamma-aminobutyric acid [GABA], glutamate, and serotonin systems).
Discontinuation symptoms provide a model for how BZDs may worsen PTSD. Chronic BZD use leads to GABA-receptor desensitization and glutamate receptor sensitization.26,32 When BZDs are suddenly discontinued in tolerant patients, the patients experience decreased inhibition from GABA and hyperactive excitation from glutamate, causing withdrawal symptoms that can mimic and worsen PTSD symptoms (eg, anxiety, insomnia, agitation, autonomic hyperactivity, perceptual disturbances). Although less severe than withdrawal, rebound symptoms, which are the inverse of the therapeutic effects of BZDs and include worsened anxiety, insomnia, and irritability, can occur shortly after discontinuation, including between doses (especially with BZDs that have a short half-life). Discontinuation symptoms are commonly misinterpreted as a worsening of underlying conditions while the iatrogenic contribution of BZDs is overlooked.27,28,33 Although anxiety, insomnia, and irritability may be temporarily exacerbated during withdrawal, general studies of BZDs have demonstrated that these symptoms are usually less severe after discontinuation than while taking BZDs.27,28 Both PTSD and BZD use have been associated with decreased GABA-receptor sensitivity and hyperactive glutamatergic activity.34 Because BZDs can synergistically worsen underlying PTSD pathophysiology, BZDs may actually exacerbate PTSD symptoms rather than improve them.
One of the most consistent findings in this review, which was supported by an RCT,9 a nonrandomized-controlled trial,12 6 observational studies, and a systematic review of PTSD risk factors in patients on an intensive care unit,35 is that BZD use after trauma increases the risk of developing PTSD. Only 2 studies of trauma patients receiving BZDs20,24 did not find an increased risk for PTSD, although both suggested inefficacy for PTSD prevention. Those studies providing sufficient data9,12,19 suggest that the risk of developing PTSD is 2 to 5 times higher in groups receiving BZDs than in control groups. BZDs likely disrupt normal HPA axis stress responses and memory-related processes. Interfering with normal evolutionarily advantageous physiological responses seems to increase vulnerability to subsequent stress and worsen outcomes in PTSD.36 Three animal studies4,33,37 have demonstrated that BZDs increase posttraumatic behaviors upon subsequent exposure to stress, suggesting that the fear-sensitizing effects of BZDs may act synergistically with trauma-related fear, creating a generalized fear response to subsequent stressors (eg, trauma-related cues). Despite theoretical predictions that BZDs might prevent the development of PTSD after trauma (eg, by inhibiting memory consolidation and preventing stress-induced changes in the noradrenergic system),12 no studies support BZDs for PTSD prevention, and this review suggests that the short-term antistress effects of BZDs may actually increase the long-term risk of PTSD. In hindsight, Gelpin et al12 acknowledged:
The inhibitory effect of benzodiazepines on memory acquisition is mostly anterograde. Hence, benzodiazepines do not alter memory for prior episodes and, therefore, should not have affected traumatic memories when administered several days after the trauma. Moreover, recovery from trauma should not be equated with forgetting, but rather adaptation, reappraisal, and learning. Administered during the recovery phase, benzodiazepines may, in fact, interfere with such relearning … it may be argued that early treatment with benzodiazepines negatively affected survivors who might have otherwise recovered (p. 393).
Three studies examined the effects of BZDs in patients receiving psychotherapy: Van Minnen et al23 found that daily BZD use was associated with worse outcomes, and Rosen et al17 and Rothbaum et al8 had mixed results (ie, inefficacy or worsening, depending on whether measures were rated by observers or patients). Rather than augmenting psychotherapy, BZDs seem to do nothing or to inhibit recovery. Evidence-based trauma-focused psychotherapies (eg, prolonged exposure, cognitive processing therapy) require that patients experience and then master anxiety. BZDs can impair that experience by numbing emotions, decreasing learning efficiency, and inhibiting memory processing of material learned in therapy.17,38 BZD-induced “emotional anesthesia”26 directly interferes with the therapeutic effects of exposure to anxiety-provoking stimuli (in psychotherapy or the natural environment) by inhibiting fear activation, a “necessary condition for effective exposure therapy.”23 Several animal and human studies have demonstrated that BZDs interfere with fear extinction, which is critical to exposure therapy.2,39,40 For fear extinction to occur, patients must emotionally and cognitively process the experience of anxiety, but BZDs allow patients to avoid these processes. Some patients with PTSD use distraction techniques to avoid internal reminders of trauma, some rarely leave places of comfort to avoid external reminders, and others engage in reckless behaviors to “escape.”29 BZDs may provide another form of avoidance, an attempt to self-medicate hyperarousal, numb feelings, suppress memories, and escape thoughts. Overcoming avoidance behaviors is essential for successful treatment, but it is often the patient’s largest obstacle for recovery. As Herman41 explains:
The helpless person escapes from her situation not by action in the real world but rather by altering her state of consciousness …. Traumatized people who cannot spontaneously dissociate may attempt to produce similar numbing effects by using alcohol or narcotics …. Although dissociative alterations in consciousness, or even intoxication, may be adaptive at the moment of total helplessness, they become maladaptive once the danger is past. Because these altered states keep the traumatic experience walled off from ordinary consciousness, they prevent the integration necessary for healing …. They narrow and deplete the quality of life and ultimately perpetuate the effects of the traumatic event (p. 44).
Because an avoidant coping style is a poor prognostic factor for trauma-related disorders29 and BZDs are inherently avoidant (eg, they inhibit cognitive processing and induce emotional numbing), BZDs may prolong and worsen PTSD.
Two studies that were reviewed measured aggression, both of which found that BZDs were associated with aggression in some patients with PTSD.15,16 In general, BZDs have been known to cause “paradoxical reactions” (eg, behavioral disinhibition, impulsivity, irritability, aggression) in which patients may engage in uncharacteristic behaviors such as assaults, theft, or sexual indiscretions without any history of similar behaviors before use or after discontinuation.26,32,42,43 This is troublesome for patients with PTSD who often already display irritability, aggression, and reckless behavior. Proposed mechanisms for paradoxical reactions include rebound irritability, inhibition of serotonin regulation, inhibition of emotional reactivity to aversive events that deters behavioral activation, and inhibition of cognitive processing in which causal associations are formed between behaviors and their consequences.32,42 Risk factors for paradoxical reactions include several conditions common in PTSD: SUD (especially alcohol use disorder), neurocognitive disorders (especially TBI), anxiety disorders (comorbid to and including PTSD), previous impulsivity, and previous aggression.2,15,16,40 Unfortunately, factors such as SUD, TBI, and comorbid anxiety disorders are also correlated with increased prescription of BZDs for PTSD.1,2,39
Two of the studies that were reviewed measured depression: Braun et al10 found that BZDs were ineffective for depression in PTSD and Gelpin et al12 found that BZD use after trauma increased the risk of developing major depressive disorder. In general, BZDs have been known to cause or worsen dysphoria and suicidality (ie, BZD-induced depressive disorder) even in individuals without a history of depression.5,39,43,44 Unfortunately, although the therapeutic effects of BZDs decrease with tolerance, depression and impulsivity with high suicidal risk commonly persist.27 The mechanisms responsible for BZD-induced depression are a matter of speculation but they may be similar to those causing paradoxical reactions (eg, inhibition of serotonin regulation, impulsivity). Regardless of the explanation, the prospect of BZDs worsening depression is of concern for patients with PTSD, who commonly have negative moods and cognitions, anhedonia, suicidality, and comorbid depressive disorders.
One study16 that was reviewed measured substance use, and the findings suggested that BZDs are associated with substance use in some patients with PTSD. Although BZDs are some of the more commonly misused substances following trauma, data about BZDs and SUD specific to patients with PTSD are limited. In general, 58% to 100% of those prescribed chronic BZDs become physically dependent (especially with high doses and short-acting BZDs).28,45 Risk factors for developing BZD use disorder include preexisting or active SUD, family history, early onset of use, medical availability, chronic medical conditions, chronic pain, chronic anxiety, chronic insomnia, chronic dysphoria, previous impulsivity, and personality disorders.2,27–29,45 Unfortunately, SUD and chronic anxiety are also correlated with increased prescriptions for BZDs for PTSD.1,2,39 Because BZD use disorder develops in at least 50% of patients with a history of SUD who are prescribed BZDs, many authors and organizations have declared BZDs contraindicated in all patients with histories of SUD, except during withdrawal.26,40 Although a previous SUD may be the predominant risk factor, when BZDs are continuously available, drug reinforcement can lead to misuse by patients without any history of substance misuse.46 “Their greatest asset is also their greatest liability: drugs that work immediately tend to be addictive.”26 SUD occurs in 21% to 43% of patients with PTSD47 and in as many as 50% of veterans with PTSD.3 This high comorbidity suggests that PTSD and SUD are functionally related, a concept supported by several studies that indicate a pathway related to corticotropin-releasing hormone and norepinephrine whereby PTSD precedes SUD.47 The high risk of SUD in patients with PTSD is one reason why so many authors and organizations recommend against treating PTSD with BZDs.
Efficacy (Level of Evidence D)
A few anecdotal reports and parts of 2 RCTs support short-term symptomatic treatment, but there is no available expert consensus endorsing BZDs for PTSD treatment, so the ceiling for the level of evidence is D. Although both the studies supporting efficacy are RCTs,9,10 they had mixed findings (ie, they also demonstrated inefficacy or worsening of PTSD) and, at best, they only supported short-term use for some PTSD-associated symptoms. For example, Mellman et al9 found that temazepam initially improved sleep but that it was not significantly different from placebo after the first night and worsened overall PTSD severity in the long term. Braun et al10 found short-term improvement in anxiety (described as a “slight reduction,” “modest,” and “disappointing”), but no significant difference from placebo in any other measure (overall severity of PTSD symptoms, depression, overall well-being). Other studies that were reviewed demonstrated inefficacy for sleep11 and anxiety.12,16 Only 1 other nonanecdotal study supported efficacy: Lee et al48 found that lorazepam improved intrusive symptoms, but the RCT was excluded from this review because the participants experienced an artificial “trauma” by video and were assessed only 1 day later (less than the 1 month threshold for PTSD). The study by Lee and colleagues also found no significant improvement in anxiety, depression, or arousal. The authors suggested that lorazepam is “atypical” and differs from other BZDs such as diazepam that can trigger intrusion. Even if BZDs improve PTSD-associated symptoms on a short-term basis, the benefits are unlikely to last due to tolerance.
There is no evidence besides anecdotal reports that supports the use of BZDs for the treatment of PTSD core symptoms (ie, intrusion, avoidance, hyperarousal) or for long-term symptomatic treatment of PTSD. Many researchers have criticized the frequent citation of case reports to justify the use of BZDs to treat patients with PTSD, “despite risks and lack of studies.”11 These case reports are mostly retrospective and based on subjective reports. Patients’ reports of their experiences while taking BZDs are inherently unreliable, as 1 case series49 concedes:
It is possible that patients’ memories of subjective sensations while intoxicated do not correspond to their actual affective state. For instance, many people report euphoria after the fact with alcohol intoxication, even though at the time of intoxication they were tearful and agitated (p. 374).
In the case of sleep, BZDs are often credited (like alcohol) for improving sleep quality, but they actually promote sleep induction while inhibiting the deepest, most restorative stages of sleep.28,45 At times, subjective reports of improvement with BZDs may reflect distortions due to cognitive impairments or they may be due to patients mistaking the temporary relief of discontinuation symptoms for improvement of baseline symptoms or mistaking sedation for genuine improvement of their condition.
The findings of Mellman et al50 highlight the importance of caution when extrapolating the results of anecdotal evidence to clinical practice. In this prospective case series of 4 recent trauma patients with insomnia, the researchers found that short-term temazepam was associated with improved PTSD symptoms; however, this was a pilot study for Mellman et al,9 the RCT that found that short-term temazepam increases the risk of developing PTSD. In addition, “because benzodiazepines reduce anxiety without addressing the underlying PTSD, clinicians may incorrectly believe the patient has improved, thus delaying definitive PTSD care.”40 BZDs “need to be carefully considered, taking into account their potential harm to the spontaneous recovery process, and the trajectory of PTSD, and not only judging them according to their immediate (comforting) effects.”36
There was little consistency in participants, diagnostic method, trauma type, recency, severity, intervention, follow-up, or outcome measures among the studies selected for review. For example, the studies conducted in intensive care units examined only life-threatening medical conditions,18–20,22 whereas McGhee et al24 examined only combat-related PTSD. Nine articles studied the use of any BZD, whereas others studied specific agents. Follow-up ranged from 2 weeks to 4 years. Seven studies used multiple assessment instruments and no instrument was used in more than 4 studies. These inconsistencies resulted in heterogeneity among the studies. However, random-effects models were used in the meta-analyses to account for this heterogeneity, resulting in wider CIs for ESs than would have resulted if a fixed-effects approach had been used.
Our meta-analytic approach also had some limitations. In particular, publication bias, if present, would result in an underreporting of nonsignificant studies. However, in the present context, there is less reason for studies with nonsignificant results not to have been published, as they would have supported the view that BZDs are not harmful. An additional potential limitation is that the meta-analytic method chosen implicitly assumes that there is no systematic bias across these studies favoring positive associations.
Although there is evidence that BZDs can worsen PTSD-associated symptoms, the authors cannot exclude the likelihood that those patients who were treated with BZDs in the observational studies that were reviewed were more severely affected and had worse prognoses. For example, the results of the intensive care unit studies were likely confounded by indication (eg, patients who are more delirious, agitated, or anxious in the ICU may be more likely to receive higher BZD doses). Therefore, BZD use may be an indicator, rather than a cause, of poorer prognosis. However, such confounding factors were eliminated in a study by Treggiari et al,51 a RCT that found that lower sedation in critically ill patients is associated with fewer PTSD symptoms (this study was excluded from this review because it did not distinguish BZDs from other sedatives). Likewise, similar confounding factors were eliminated in the reviewed RCTs that demonstrated worsening of PTSD.8,9
The greatest limitation of this review was the limited number of RCTs available. Of the 4 placebo-controlled trials, only 2 were double-blind,8,10 whereas 111 was single-blind and another was open-label.9 Nevertheless, the authors believed it was worthwhile to compile the data from all of the available studies given the widespread use of BZDs for PTSD and disagreements and misconceptions among clinicians about this practice. When the meta-analysis was limited only to the RCTs, the results were inconclusive due to the small sample sizes of those RCTs and the great amount of heterogeneity among the studies with 2 showing nonsignificant negative effects and 2 showing nonsignificant positive effects. Although one might argue that only RCTs should be considered for inclusion in a rigorous review, in light of the limited number of studies of BZDs in PTSD, we elected to use more expansive inclusion criteria to create a comprehensive review of the available literature, and to stimulate clinical thought and further research. Further studies are recommended, especially randomized placebo-controlled trials with extended follow-up.
Although BZDs have been in use since 1960 and trauma survivors have always existed, hard knowledge is scanty. Nevertheless, based on our meta-analysis and qualitative synthesis, we can conclude that BZDs are more likely to be ineffective than effective for the treatment or prevention of PTSD and that risks tend to outweigh potential short-term benefits. Consistent evidence supports a lack of efficacy, especially for PTSD core symptoms, psychotherapy augmentation, and depression. There is also suggestive evidence that BZDs may worsen outcomes, with BZDs being correlated with worse overall severity of PTSD symptoms, increased risk of trauma patients developing PTSD, and worse psychotherapy outcomes. However, more double-blind placebo-controlled trials are needed before it can be concluded that BZDs consistently worsen PTSD. There is little evidence for anything except the most transient efficacy, which is limited to a few symptoms, and this is outweighed by better evidence for inefficacy and potential risks. For these reasons and others, BZDs should be considered relatively contraindicated in trauma patients.
Most studies specific to BZDs in PTSD are small and few are RCTs. However, taken together and in combination with general BZD studies, they raise enough questions about potential harms that providers should use considerable caution when continuing BZD prescriptions and would be safer to avoid starting them altogether in PTSD patients. Some of these potential problems are general concerns about the medication class (eg, cognitive effects, dependence, misuse), and others are specific to the diagnosis (eg, adverse effects synergistically worsening PTSD symptoms, inhibiting psychotherapy, promoting avoidance). Although BZDs might be effective if they were to selectively inhibit the stress and anxiety centers of the brain that are often hyperactive in PTSD (eg, amygdala, HPA axis), they indiscriminately target the entire brain, including those areas that are already hypoactive in PTSD, including the cognitive and memory centers (eg, prefrontal cortex, hippocampus), and serotonergic circuits (implicated in PTSD, anxiety, depression, suicidality, impulsivity, aggression). Although it may be tempting to treat PTSD-associated symptoms with BZDs, they are best avoided due to evidence of long-term risks outweighing evidence of any short-term benefits, and the difficulty of discontinuing BZDs once started. When patients with PTSD are already taking BZDs, providers should evaluate whether the treatment is actually improving the patients’ functioning or if there are any—often subtle—iatrogenic effects on the course of their condition. After weighing risks and benefits, some providers will choose to continue BZDs, some will unilaterally discontinue or change medications, and some will work through the stages of change to help patients transition toward evidence-based treatments. Regardless, recovery from PTSD should denote improved functioning (eg, healthy relationships, employment), not simply sedation.
Although there is little evidence of benefits associated with BZDs in PTSD, substantial evidence supports the benefits associated with other treatments. A myriad of evidence-based treatments for PTSD exist (eg, psychotherapy, serotonergic antidepressants, adrenergic inhibitors),3,38,40 all of which should be exhausted before BZDs are considered. For years, sedatives were the only thing we had in our armamentarium for PTSD. Now, we have many more tools and our patients—whether survivors of assault, combat, or any other trauma—deserve those treatments that have been proven to be safer and more effective than BZDs.
1. Harpaz-Rotem I, Rosenheck RA, Mohamed S, et al.. Pharmacologic treatment of posttraumatic stress disorder among privately insured Americans. Psychiatr Serv. 2008;58:1184–1190.
2. Lund BC, Bernardy NC, Vaughan-Sarrazin M, et al.. Patient and facility characteristics associated with benzodiazepine prescribing for veterans with PTSD. Psychiatr Serv. 2013;64:149–155.
4. Matar MA, Zohar J, Kaplan Z, et al.. Alprazolam treatment immediately after stress exposure interferes with the normal HPA-stress response and increases vulnerability to subsequent stress in an animal model of PTSD. Eur Neuropsychopharmacol. 2009;19:283–295.
5. Foa EB, Keane TM, Friedman MJ, et al.. Effective Treatments for PTSD: Practice Guidelines From the International Society for Traumatic Stress Studies, 2nd ed. New York, NY: Guilford; 2009:566.
6. US Department of Health and Human Services. Clinical Practice Guideline No 5: Depression in Primary Care, Vol 2: Treatment of Major Depression. Rockville, MD: AHCPR Publications; 1993:71–123.
7. Moher D, Liberati A, Tetzlaff J, et al.. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.
8. Rothbaum OR, Price M, Jovanovic T, et al.. A randomized, double-blind evaluation of D-cycloserine or alprazolam combined with virtual reality exposure therapy for posttraumatic stress disorder in Iraq and Afghanistan war veterans. Am J Psychiatry. 2014;171:640–648.
9. Mellman TA, Bustamante V, David D, et al.. Hypnotic medication in the aftermath of trauma. J Clin Psychiatry. 2002;63:1183–1184.
10. Braun P, Greenberg D, Dasberg H, et al.. Core symptoms of posttraumatic stress disorder unimproved by alprazolam treatment. J Clin Psychiatry. 1990;51:236–238.
11. Cates ME, Bishop MH, Davis LL, et al.. Clonazepam for treatment of sleep disturbances associated with combat-related posttraumatic stress disorder. Ann Pharmacother. 2004;38:1395–1399.
12. Gelpin E, Bonne O, Peri T, et al.. Treatment of recent trauma survivors with benzodiazepines: a prospective study. J Clin Psychiatry. 1996;57:390–394.
13. Shalev AY, Bloch M, Peri T, et al.. Alprazolam reduces response to loud tones in panic disorder but not in posttraumatic stress disorder. Biol Psychiatry. 1998;44:64–68.
14. Zatzick DF, Rivara FP, Nathens AB, et al.. A nationwide US study of post-traumatic stress after hospitalization for physical injury. Psychol Med. 2007;37:1469–1480.
15. Shin HJ, Rosen CS, Greenbaum MA, et al.. Longitudinal correlates of aggressive behavior in help-seeking US veterans with PTSD. J Trauma Stress. 2012;25:649–656.
16. Kosten TR, Fontana A, Sernyak MJ, et al.. Benzodiazepine use in posttraumatic stress disorder among veterans with substance abuse. J Nerv Ment Dis. 2000;188:454–459.
17. Rosen CS, Greenbaum MA, Schnurr PP, et al.. Do benzodiazepines reduce the effectiveness of exposure therapy for posttraumatic stress disorder? J Clin Psychiatry. 2013;74:1241–1248.
18. Jones C, Backman C, Capuzzo M, et al.. Precipitants of post-traumatic stress disorder following intensive care: a hypothesis generating study of diversity in care. Intensive Care Med. 2007;33:978–985.
19. Samuelson KAM, Lundberg D, Fridlund B. Stressful memories and psychological distress in adult mechanically ventilated intensive care patients—a 2 month follow-up study. Acta Anaesthesiol Scand. 2007;51:671–678.
20. Bienvenu OJ, Gellar J, Althouse BM, et al.. Post-traumatic stress disorders symptoms after acute lung injury: a 2-year prospective longitudinal study. Psychol Med. 2013;43:2657–2671.
21. Baranyi A, Krauseneck T, Rothenhausler HB. Posttraumatic stress symptoms after solid-organ transplantation: preoperative risk factors and the impact on health-related quality of life and life satisfaction. Health Qual Life Outcomes. 2013;11:111.
22. Girard TD, Shintani AK, Jackson JC, et al.. Risk factors for post-traumatic stress disorder symptoms following critical illness requiring mechanical ventilation: a prospective cohort study. Crit Care. 2007;11:R28.
23. Van Minnen A, Arntz A, Keijsers GPJ. Prolonged exposure in patients with chronic PTSD: predictors of treatment outcome and dropout. Behav Res Ther. 2002;40:439–457.
24. McGhee LL, Maani CV, Garza TH, et al.. The relationship of intravenous midazolam and posttraumatic stress disorder development in burned soldiers. J Trauma. 2009;66(suppl):S186–S190.
25. Shalev AY, Rogel-Fuchs Y. Auditory startle reflex in post-traumatic stress disorder patients treated with clonazepam. Isr J Psychiatry Relat Sci. 1992;29:1–6.
26. Longo LP, Johnson B. Addiction: part I. Benzodiazepines—side effects, abuse risk and alternatives. Am Fam Physician. 2000;61:2121–2128.
27. Michelini S, Cassano GB, Frare F, et al.. Long-term use of benzodiazepines: tolerance, dependence and clinical problems in anxiety and mood disorders. Pharmacopsychiatry. 1996;29:127–134.
28. Ashton H. The diagnosis and management of benzodiazepine dependence. Curr Opin Psychiatry. 2005;18:249–255.
29. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Fifth ed. Arlington, VA: American Psychiatric Association; 2013:265–290, 550–560.
30. Coupland NJ, Lillywhite A, Bell CE, et al.. A pilot controlled study of the effects of flumazenil in posttraumatic stress disorder. Biol Psychiatry. 1997;41:988–990.
31. Randall PK, Bremner JD, Krystal JH, et al.. Effects of the benzodiazepine antagonist flumazenil in PTSD. Biol Psychiatry. 1995;38:319–324.
32. Tasman A, Kay J, Lieberman JA. Psychiatry, 3rd ed, Vol 1. Chichester, UK: John Wiley & Sons; 2008:1186–1200, 2603–2615.
33. Li S, Murakami Y, Wing M, et al.. The effects of chronic valproate and diazepam in a mouse model of posttraumatic stress disorder. Pharmacol Biochem Behav. 2006;85:324–331.
34. Geuze E, van Berckel BNM, Lammertsma AA, et al.. Reduced GABAA
benzodiazepine receptor binding in veterans with post-traumatic stress disorder. Mol Psychiatry. 2008;13:74–83.
35. Davydow DS, Gifford JM, Desai SV, et al.. Posttraumatic stress disorder in general intensive care unit survivors: a systematic review. Gen Hosp Psychiatry. 2008;30:421–434.
36. Zohar J, Juven-Wetzler A, Sonnino R, et al.. New insights into secondary prevention in post-traumatic stress disorder. Dialogues Clin Neurosci. 2011;13:301–309.
37. Hebert MA, Potegal M, Moore T, et al.. Diazepam enhances conditioned defeat in hamsters. Pharmacol Biochem Behav. 1996;55:405–413.
38. Berger W, Mendlowicz MV, Marques-Portella C, et al.. Pharmacologic alternatives to antidepressants in posttraumatic stress disorder: a systematic review. Prog Neuropsychopharmacol Biol Psychiatry. 2009;33:169–180.
39. Hawkins EJ, Malte CA, Imel ZE, et al.. Prevalence and trends of benzodiazepine use among Veterans Affairs patients with posttraumatic stress disorder, 2003–2010. Drug Alcohol Depend. 2012;124:154–161.
40. Jeffreys M, Capehart B, Friedman MJ. Pharmacotherapy for posttraumatic stress disorder: review with clinical applications. J Rehabil Res Dev. 2012;49:703–715.
41. Herman J. Trauma and Recovery: The Aftermath of Violence—From Domestic Abuse to Political Terror. New York, NY: Basic Books; 1992:44–45.
42. Bond AJ. Drug-induced behavioural disinhibition: incidence, mechanisms and therapeutic implications. CNS Drugs. 1998;9:41–57.
43. PDR Network. Physicians’ Desk Reference, 2014. Available at: http://www.pdr.net
. Accessed March 5, 2014).
44. Baldwin DS, Anderson IM, Nutt DJ, et al.. Evidence-based guidelines for the pharmacological treatment of anxiety disorders: recommendations from the British Association for Psychopharmacology. J Psychopharmacol. 2005;19:567–596.
45. Pary R, Lewis S. Prescribing benzodiazepines in clinical practice. Resid Staff Physician. 2008;54:8–17.
46. Griffiths RR, Weerts EM. Benzodiazepine self-administration in humans and laboratory animals—implications for problems of long-term use and abuse. Psychopharmacology. 1997;134:1–37.
47. Jacobsen LK, Southwick SM, Kosten TR. Substance use disorders in patients with posttraumatic stress disorder: a review of the literature. Am J Psychiatry. 2001;158:1184–1190.
48. Lee HS, Lee HP, Lee SK, et al.. Anti-intrusion effect of lorazepam: an experimental study. Psychiatry Investig. 2013;10:273–280.
49. Bremner JD, Southwick SM, Darnell A, et al.. Chronic PTSD in Vietnam combat veterans: course of illness and substance abuse. Am J Psychiatry. 1996;153:369–375.
50. Mellman TA, Byers PM, Augenstein JS. Pilot evaluation of hypnotic medication during acute traumatic stress response. J Trauma Stress. 1998;11:563–569.
51. Treggiari MM, Romand JA, Yanez ND, et al.. Randomized trial of light versus deep sedation on mental health after critical illness. Crit Care Med. 2009;37:2427–2534.