Factors impacting prazosin efficacy for nightmares and insomnia in PTSD patients - a systematic review and meta-regression analysis

Abstract

Posttraumatic stress disorder (PTSD) is a debilitating condition affecting 5.7 % of the global population in their lifetime. There is a strong association between trauma-related nightmares and insomnia with higher rates of physical illness, mental distress, and suicide among PTSD patients. Prazosin, an α1-adrenergic antagonist, has shown mixed results in treating these sleep disturbances. This study aims to evaluate the effect of prazosin compared to placebo on insomnia, nightmares, and global PTSD symptoms, and to examine variables that might influence this effect. We conducted a meta-analysis and a novel meta-regression analysis of randomized clinical trials (RCTs) comparing prazosin to placebo in samples of patients with PTSD. Data sources were MEDLINE, EMBASE, Scopus, ISI Web of Science, and PTSD Pubs. Examined variables were age, gender, military/civilian status, prazosin dose, treatment duration, baseline symptom severity, use of antidepressants, use of benzodiazepines (BDZ), prevalence of depression, and alcohol use disorder. Ten RCTs with 648 patients were included. Analysis revealed prazosin significantly improved insomnia (SMD = −0.654, p = 0.043) and nightmares (SMD = −0.641, p = 0.025), but not overall PTSD symptoms (SMD = −0.428, p = 0.077). Unexpectedly, higher BDZ use was associated with greater improvement in insomnia (β = −0.046; p = 0.01) and PTSD severity (β = −0.037; p = 0.004). These findings suggest that prazosin effectively reduces insomnia and nightmares in PTSD patients. Benzodiazepine co-administration seems to enhance prazosin’s efficacy, suggesting that the addition of prazosin might allow for a reduction of BDZ doses in these patients. Further research should empirically test the efficacy of prazosin alone versus prazosin combined with BDZ to confirm these findings.

1. Introduction

Posttraumatic stress disorder (PTSD) is a severe condition that affects up to 5.7 % of the population worldwide (Ferry et al., 2015; Kessler et al., 2012; Koenen et al., 2017). There is a strong correlation between PTSD and higher rates of physical illness, mental distress, and suicide (Sareen et al., 2007). A further implication of this disorder resides in its economic burden, which surpasses 230 billion dollars a year in the United States, accounting for healthcare expenditures and lost productivity (Davis et al., 2022). This debilitating condition is characterized by recurrent intrusive thoughts and memories, avoidant behavior, cognitive and mood disturbance, and hyperarousal (Greenberg et al., 2015).

Even during sleep, intrusive thoughts may cause additional distress by assuming the form of disturbing nightmares. Indeed, insomnia and nightmares have long been considered hallmarks of PTSD (Ross et al., 1989). It has been demonstrated that these symptoms alone can increase the risk of suicide (Weber et al., 2020), with nightmares presenting the strongest association, representing a fivefold elevation in suicidality risk (Sjöström et al., 2007).

Despite the high prevalence and acknowledged role of sleep disturbances in exacerbating daytime PTSD symptoms, effectively managing nightmares and insomnia in PTSD presents significant clinical challenges. For instance, current FDA-approved medications commonly prescribed for PTSD management, such as sertraline, might even worsen rather than alleviate these symptoms (Bernath and Ward, 2023; Wichniak et al., 2017).

Prazosin, an α₁-adrenergic antagonist, has been shown to reduce trauma-related nightmares and insomnia in individuals with PTSD through previous randomized clinical trials and meta-analyses (Reist et al., 2021; Skeie-Larsen et al., 2022; Zhang et al., 2020, 2022). However, there is a high heterogeneity among the results of these clinical trials. While some have demonstrated considerable efficacy of prazosin in alleviating the nighttime symptoms of PTSD (Ahmadpanah et al., 2014; M. A. Raskind et al., 2003, 2007, 2013; Taylor et al., 2008), others have shown no benefits when compared to placebo (Mahr et al., 2023; McCall et al., 2018; Petrakis et al., 2016; M. Raskind et al., 2018; Simpson et al., 2015). Notably, the populations in these studies varied substantially regarding gender proportion, civilian or military status, dose of prazosin, treatment duration, baseline severity of symptoms, and medications in use, among other factors. Whether any of these factors can interfere with prazosin efficacy is unknown.

Therefore, to examine some of the variables that could justify the heterogeneous findings in prazosin efficacy for nightmares, insomnia, and global PTSD symptoms, we’ve conducted the first meta-regression analysis on the topic.

2. Methods

2.1. Protocol and registration

This systematic review and meta-analysis was conducted in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021). Exploratory meta-regression analysis was performed to examine the impact of age, gender, military or civilian status, prazosin dose, treatment duration, use of antidepressants, use of benzodiazepines, and current alcohol use disorder on the efficacy of prazosin treatment for nightmares and insomnia in populations diagnosed with PTSD. The protocol of this review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) on December 16th, 2023, under the ID CRD42023488478.

2.2. Search strategy

We performed a comprehensive search of the literature on November 11th, 2023, across the databases MEDLINE, EMBASE, Scopus, ISI Web of Science, and PTSD Pubs. We utilized the following search strategy across all databases in the field “Topics” (Title, Abstract and Keywords) or its equivalent: (PTSD OR “posttraumatic stress” OR “post-traumatic stress”) AND (prazosin OR minipress). No filters were applied during the search process.

2.3. Eligibility criteria

We used the following PICOTT strategy to predefine our inclusion and exclusion criteria: (P) Participants: adults with PTSD; (I) Intervention: prazosin; (C) Control: placebo; (O) Outcomes: improvement of nightmares, insomnia, and PTSD symptoms; (T) Type of study: RCTs; (T) Time of follow-up: no restriction.

Included studies were RCTs investigating the efficacy of prazosin compared to placebo in treating at least one of the following: trauma-related nightmares; insomnia; and overall PTSD symptoms in civilian or military populations diagnosed with PTSD according to DSM-IV or DSM-IV TR criteria (American Psychiatric Association, 1994, 2000).

Exclusion criteria were studies performed on children or adolescents; Trials employing additional interventions alongside prazosin; Trials allowing change of current medication or introduction of new treatment during the study; Trials using an active comparator; Studies reporting no data for at least one of the outcomes of interest in the target population; Data generated from non-validated assessment tools for PTSD and its symptoms; and non-peer-reviewed manuscripts (i.e., gray literature).

2.4. Study selection

After removing duplicates, two authors (T.M. and B.G.) independently screened the remaining records for inclusion using Zotero (2024), a reference management software. Discrepancies in judgment were resolved through discussion or consulting a third author (W.B.). The screening process and reasons for exclusion are illustrated in Fig. 1.

Fig. 1..

PRISMA flow diagram of the study selection.

2.5. Data extraction and outcomes of interest

Data was blindly and independently extracted by two authors (T.M. and B.G.) for each selected study. Extracted data included population characteristics (civilian, military, or mixed), baseline and endpoint number of participants, mean age, gender distribution, prazosin dosage, treatment duration, baseline symptom severity, prevalence of depression, use of antidepressants, use of benzodiazepines, prevalence of current alcohol use disorder, and baseline and endpoint severity scores for nightmares, insomnia, and PTSD symptoms. When more than one tool was used for the same outcome, we chose the most frequent one among the studies for the analysis. To obtain missing data, we contacted the corresponding authors of the included studies by email. Of the five authors contacted, three responded, of which one provided the requested information.

2.6. Risk of bias assessment

The risk of bias in the included studies was assessed by two independent reviewers (M.M. and B.G.) using the ROB 2 (Risk of Bias 2) tool developed by the Cochrane Collaboration (Higgins et al., 2024). Any discrepancies in judgment were resolved through discussion or consulting a third reviewer (T.M.).

The risk of bias assessment for each domain was categorized as ‘low risk’, ‘some concerns’, or ‘high risk’. Justifications for the risk of bias judgments were provided, considering specific criteria outlined in the ROB 2 tool. A summary of the risk of bias assessments for each included study is presented in Fig. 2.

Fig. 2..

Risk of bias of the included studies.

2.7. Statistical analysis

Standardized mean difference (SMD) was calculated by the method of Cohen to obtain the pooled endpoint differences between prazosin and placebo scores for insomnia, nightmare, and PTSD symptoms, separately. The more negative the SMD, the greater the efficacy of prazosin. This calculation used the random-effects inverse-variance model with DerSimonian-Laird estimate of tau² (Borenstein et al., 2009). This method is indicated to deal with heterogeneous findings between the individual studies. Sensitivity analysis was carried out to investigate the impact of studies with outlier findings on the pooled estimate. The magnitude of heterogeneity was evaluated by the I² statistic, which informs the proportion of inter-studies variance (heterogeneity) relative to the total variance (Higgins et al., 2003). Meta-regression models were fitted to investigate possible causes of heterogeneity. All analyses were performed using Stata 16 software. (Stata Statistical Software: Release 16, 2019).

3. Results

3.1. Study selection and characteristics

A total of 2074 records were identified through our initial search. After removing duplicates and screening by title and abstract, 26 articles were selected for full-text review. Out of these, 13 studies initially appeared to fit the inclusion criteria. However, two studies lacked data for the target population, and one study reported as a randomized controlled trial described no data for a control group in its methods or results. Finally, 10 RCTs were included in our meta-regression analysis, comprising 648 patients. The screening process and reasons for exclusion are illustrated in Fig. 1.

The number of participants per study varied between 10 and 304 and their mean age ranged between 30 and 56 years. We ensured that there was no sample overlap between studies by reviewing the protocols on clinicaltrials.org and contacting the authors when necessary. All papers utilized the DSM-IV diagnostic criteria for PTSD. The mean duration of PTSD was between 96 (Ahmadpanah et al., 2014) and 199 (Mahr et al., 2023) months, but it was reported by only two studies. Six studies reported the percentage of benzodiazepine users, which varied between 11.8 % and 60.6 %. There were no changes in treatment regimens during the study, except in Raskind et al. (2018), in which there was an increase of benzodiazepine users of 1 % in the prazosin group and 2 % in the placebo group. The mean daily dose of benzodiazepines is only reported by one study (Taylor et al., 2008). None of the studies specify the frequency of use of this medication. Further information about study characteristics can be found in Table 1.

Table 1.

Baseline characteristics of included studies.

Author (year)	Country	N (baseline/ endpoint)	Veterans, %	Female, %	Age, years (mean ± SD)	Prazosin dose, mg (mean ± SD)	Treatment duration, weeks	Baseline insomnia score (prazosin / placebo)	Baseline nightmares score (prazosin / placebo)	Baseline PTSD score (prazosin / placebo)	Use of BZD, %	Use of AD, %	Assessment tools (Insomnia / Nightmares / PTSD)	Reported findings
Ahmadpanah et al. (2014)	Iran	66 / 66	NA	28.8	35.2 ± NA	15 ± 0	8	15.52 ± 2.17 / 15.48 ± 1.97	-	7 ± 0.04 / 7 ± 0.4	60.6	45.5	PSQI / - a / MINI	Favors prazosin for sleep quality and PTSD symptoms
Mahr et al. (2023)	USA	8 / 8 (crossover)	0	100	30.12 ± NA	3 ± NA	3	-	3.62± 0.52 / 3.62 ± 0.52	-	NA	NA	PSQIb / Sleep- 50c/ -	Favors prazosin for nightmares in a sample of women with bulimia nervosad
McCall et al. (2018)	USA	20 / 6	0	85	39.8 ± 14.5	5.5 ± 3.5	8	22 ± 4.2 / 18.7 ± 3.6	23.4 ± 5.9 / 20.9 ± 4.3	62.4 ± 8 / 60.8 ± 11.7	16.6	33.3	ISI / DDNSI / PCL	Favors placebo for insomnia, nightmares, and PTSD symptoms
Petrakis et al. (2016)	USA	96 / 75	100	6.25	44 ± NA	14.5 ± 3.14	12	21.47 ±0.94 / 22.8 ± 0.97	5.92 ± 0.32 / 5.44 ± 0.34	71.86 ± 24.65 / 75.71 ± 26.36	NA	NA	PSQI / CAPS item B2 / CAPS	There was no difference between prazosin and placebo for PTSD symptoms or sleep symptoms in a sample of veterans with AUD
Raskind et al. (2003) e	USA	10 / 10 (cross-over)	100	0	53 ± 3	9.5 ± 0.5	9	7.4 ± 1.3 / 7.3 ± 0.9	6.9 ± 0.9 / 7.1 ± 0.9	79.1 ± 17 / 83.6 ± 17.6	40	70	CAPS item D1 / CAPS item B2 / CAPS	Favors prazosin for nightmares, insomnia, and PTSD symptoms
Raskind et al. (2007)	USA	34 / 29	100	5 f	56 ± 9	13.3 ± 3	9 – 12 g	13.5 ± 4.2 / 13.4 ± 2.7	6.5 ± 1 / 6.1 ± 1	76 ± 22 / 78 ± 18	11.8	61.8	PSQI / CAPS item B2 / CAPS	Favors prazosin for nightmares, insomnia, and PTSD symptoms
Raskind et al. (2013)	USA	67 / 67	100	14.9	30.4 ± NA	17.56 ± NA	15	14.1 ± 3.9 / 14.5 ± 3.77	6 ± 1.73 / 6.6 ± 1.81	77.3 ± 23.67 / 85.7 ± 23.1	NA	35.8	PSQI / CAPS item B2 / CAPS	Favors prazosin for nightmares, insomnia, and PTSD symptoms
Raskind et al. (2018)	USA	304 / 245	100	2.3	51.9 ± NA	14.8 ± 6.1	26	14.4 ± 3.3 / 14.7 ± 3.5	6.3 ± 0.9 / 6.3 ± 0.9	80.7 ± 15.5 / 81.9 ± 17.1	12.5	77.6	PSQI / CAPS item B2 / CAPS	There was no difference between prazosin and placebo in alleviating nightmares, insomnia, or PTSD symptoms
Simpson et al. (2015)	USA	30 / 30 h	30	36.6	43.3 ± 11.7	16 ± 0	6	-	-	3.7 ± 2.17 / 2.7 ± 2.17	0	23.3	- / - / PSS-I	There was no difference between prazosin and placebo for PTSD symptoms
Taylor et al. (2008)	USA	13 / 13 (cross-over)	0	84.6	49 ± 10	3.1 ± 1.3	3	6 ± 1 / 6.1 ± 0.8	4.8 ± 1.7 / 3.9 ± 2.3	58 ± 13 / 56 ± 16	23	76.9	CAPS item D1 / CAPS item B2 / PCL	Favors prazosin for TRN and PCL score

AD: antidepressants; AUD: Alcohol Use Disorder; BZD: benzodiazepines; CAPS: Clinician-Administered PTSD Scale for DSM-IV; CAPS item D1: “difficulty falling or staying asleep”; CAPS item B2: “recurrent distressing dreams of the event”; DDNSI: Disturbing Dreams and Nightmares Severity Index; ISI: Insomnia Severity Index; MINI: Mini International Neuropsychiatric Interview; NA: not available; PSQI: Pittsburgh Sleep Quality Index; PCL: PTSD Checklist; PSS-I: PTSD Symptom Scale – Interview; SD: standard deviation; TRN: trauma-related nightmares.

^aAhmadpanah et al. (2014) used a non-validated measurement scale for nightmares, which did not meet our inclusion criteria.

^bMahr et al. (2023) used only 4 PSQI items, each one with a different SD, therefore we decided not to include it in our analysis.

^cMahr et al. (2023) reported the score of the B2 nightmare item of CAPS divided into frequency and intensity, each one with a different SD, so we decided to use the reported Sleep-50 questionnaire score.

^dAlthough Mahr et al. (2023) report significant improvements in nightmares with prazosin, this trial shows no statistical significance for this outcome in our meta-analysis.

^eFour patients from the second placebo group dropped out, but intent-to-treat analysis was used to compensate for this loss.

^fTwo out of 40 participants (5%) on the trial by Raskind et al. (2007) were women, however, we don’t have the data for the 34 participants who finished the trial.

^gRaskind et al. (2007) had a one-to-four-week titration period followed by eight weeks of treatment. To ensure consistency with the other trials, we standardized the treatment duration to 12 weeks in our meta-regression analysis.

^hAlthough only nine individuals in the prazosin group and 11 in the placebo group completed six weeks on the trial by Simpson et al. (2015), analysis was performed on all 30 assigned participants.

3.2. Risk of bias

Regarding the methodological quality of the ten selected studies, two had a high risk of bias (McCall et al., 2018; Simpson et al., 2015), four had some concerns (Mahr et al., 2023; M. A. Raskind et al., 2003, 2007; Taylor et al., 2008), and four had low risk (Ahmadpanah et al., 2014; Petrakis et al., 2016; M. Raskind et al., 2018; M. A. Raskind et al., 2013). Seven studies adequately described the randomization process (D1). Two studies had deviations from the intended intervention, due to the lack of appropriate analysis to estimate the effect of the low participant adherence (D2). Six studies included sufficient outcome data in the final analysis (D3). Except for one study, there were no errors in measuring participants’ outcome variables (D4). However, the high risk of bias in the outcome measurement in Simpson et al.’s (2015) study suggests that the scales used in this study were inadequate. All studies adequately selected the reported results (D5). Additional details can be seen in Fig. 2.

Tests for publication bias were not conducted, as funnel plot asymmetry tests lack power with fewer than 10 studies, making it difficult to identify true asymmetry over random variation. Moreover, in the presence of significant heterogeneity, the minimum number of studies required might be considerably more than 10. In our case, all three meta-analyses exhibited substantial heterogeneity, with I-squared statistics showing values greater than 80 % (Sterne et al., 2011).

3.3. Insomnia

Eight articles were included in the insomnia analysis, comprising 515 patients at the endpoint. The pooled standard mean difference (SMD) for prazosin versus placebo on insomnia outcomes was −0.654 (95 % CI: −1.288 to −0.020), indicating a statistically significant improvement in insomnia symptoms with prazosin (p = 0.043). However, substantial heterogeneity was observed (I² = 89.2 %, p < 0.001). (Fig. 3).

Fig. 3..

Forest plot for the difference of endpoint means of insomnia symptoms (standardized mean difference-SMD) and respective 95% confidence intervals between prazosin and placebo groups. Mahr et al. (2023) used only 4 PSQI items, each one with a different SD, therefore we decided not to include it in our analysis. Simpson et al. (2015) did not include insomnia measures, therefore it also couldn’t be included.

Sensitivity analysis was conducted with the study from Ahmadpanah et al. (2014), which presents a point estimate substantially different from those reported by the other studies. We observed a reduction in the magnitude of the combined SMD when it was removed from the analysis (SMD = −0.38; 95 % CI: −0.81 to 0.05), with a marginally significant p-value of 0.08.

Meta-regression analysis revealed a statistically significant association between treatment effect and percentage of patients using benzodiazepines (β = −0.047; p = 0.01). The efficacy increased as the percentage of benzodiazepine users in the studies increased. The same association was not found for mean baseline insomnia severity, percentage of women, percentage of veterans, mean age, percentage of patients with depression, and percentage of users of antidepressants (all p-values >0.10). (Fig. 4).

Fig. 4..

Association between the effect size of prazosin on insomnia and percentage of users of benzodiazepines. The size of the circles increases as the sample size increases (precision of the estimates).

3.4. Nightmares

On the analysis of nightmare severity, combining frequency and intensity, the pooled SMD for prazosin versus placebo was −0.641 (95 % CI: −1.200 to −0.082), also showing significant improvement with prazosin (p = 0.025), with substantial heterogeneity (I² = 83.2 %, p < 0.001). (Fig. 5).

Fig. 5..

Forest plot for the difference of endpoint means of nightmare severity (standardized mean difference-SMD) and respective 95% confidence intervals between prazosin and placebo groups.

The variables investigated by meta-regression were mean baseline nightmare severity, percentage of women, percentage of veterans, mean age, percentage of users of benzodiazepines, and percentage of patients with depression, but the findings were not statistically significant (all p-values >0.10). Nevertheless, a borderline statistical significance was observed for the association with the use of antidepressants, suggesting that the efficacy of prazosin might be less pronounced in studies with a higher proportion of antidepressant users (β = 0.028; p = 0.066).

3.5. PTSD severity

For PTSD severity, the pooled SMD for prazosin versus placebo was −0.428 (95 % CI: −0.902 to 0.046), which was not statistically significant (p = 0.077), although this borderline p-value is suggestive of association. There was considerable heterogeneity among the studies (I² = 82.8 %, p < 0.001). (Fig. 6).

Fig. 6..

Forest plot for the difference of endpoint means of PTSD severity (standardized mean difference-SMD) and respective 95% confidence intervals between prazosin and placebo groups.

We conducted a sensitivity analysis removing the study from Ahmadpanah et al. (2014). As observed in the insomnia analysis, the removal of this study showed a reduction in the magnitude of the combined SMD (SMD = 0.23; 95 % CI: −0.53 a 0.09), with a p-value of 0.16.

Meta-regression analysis was carried out using the variables mean baseline PTSD severity, percentage of women, percentage of veterans, mean age, percentage of users of benzodiazepines, percentage of users of antidepressants, and percentage of patients with depression. As in the case of insomnia, the efficacy of prazosin increased as the percentage of benzodiazepine users in the studies increased (β = −0.037; p = 0.002). (Fig. 7). The same association was not found for the other variables (all p-values >0.10).

Fig. 7..

Association between the effect size of prazosin on PTSD severity and percentage of users of benzodiazepines. The size of the circles increases as the sample size increases (precision of the estimates).

4. Discussion

Our meta-analysis shows that prazosin has a statistically significant effect in reducing insomnia and nightmares in patients with PTSD, but only a borderline significant effect on overall PTSD symptoms. Furthermore, our meta-regression analysis indicates that studies with a higher percentage of users of benzodiazepines alongside prazosin were associated with a higher improvement of insomnia and PTSD severity in these patients.

Notably, our meta-regression analysis is the first to investigate variables that might explain the heterogeneity in prazosin efficacy observed across clinical trials. Previously, several hypotheses have been proposed to account for this variability. For instance, Hoskins et al. (2021) suggested that prazosin dose could be a determining factor. However, our analysis did not find a significant association between prazosin dose or treatment duration and its efficacy. This might imply that prazosin’s therapeutic effects are achieved relatively quickly, indicating a possible ceiling effect and that dosing can be individualized without affecting the overall outcome. However, it is important to point out that these findings are exploratory as the number of studies was small, the distribution of dosages across the combined sample indicates a tendency towards 15 mg, and meta-regression models investigated one variable at a time.

In 2018, a position paper stated that, despite the observed benefits of prazosin for many patients, the Task Force unanimously decided to downgrade its level of evidence following the PACT trial (Raskind et al., 2018), which demonstrated no improvement in nightmares or PTSD symptoms (Morgenthaler et al., 2018). The paper also suggested a possible interaction between prazosin and antidepressant medications. While our findings did not show a significant association between the use of antidepressants and prazosin’s efficacy for insomnia and PTSD symptoms, prazosin’s effectiveness on nightmares may be reduced in studies with a higher percentage of users of antidepressants. This reduction could be attributed to the known side effects of many selective serotonin reuptake inhibitors (SSRIs), which can cause more vivid dreams and potentially worsen nightmares (Tribl et al., 2013). With a larger sample size, a significant association might become apparent.

Additionally, Raskind et al. (2016) proposed that baseline blood pressure (BP) could be a predictive biomarker of prazosin efficacy. Indeed, their secondary analysis of a trial with 67 participants (Raskind et al., 2013) found an association between higher standing systolic BP at baseline and greater improvement in PTSD symptoms with prazosin treatment. However, Manhapra et al.’s (2019) subsequent secondary analysis of a trial with 96 participants (Petrakis et al., 2016), which replicated Raskind’s methodology, did not find baseline BP to be a predictor of improvements in PTSD symptoms or sleep parameters for individuals on prazosin. Although we aimed to include baseline standing BP in our meta-regression, there was no significant difference in baseline BP values among the four RCTs that reported this data in both the prazosin and placebo groups (McCall et al., 2018; M. Raskind et al., 2018; M. A. Raskind et al., 2007, 2013). Hence, due to the lack of heterogeneity between BP values among studies, we weren’t able to assess whether this variable could influence prazosin efficacy.

Manhapra et al. (2019) suggested three hypotheses for the discrepancy between these two studies: (1) differences in age and time since PTSD diagnosis between participants from both studies; (2) variations in the percentage of participants using antidepressants; and (3) a higher prevalence of active AUD in Manhapra et al.’s study. In our analysis, age and prevalence of AUD were not associated with prazosin’s efficacy. While we intended to include the variable of time since PTSD diagnosis, it was only reported in two of the included studies, limiting our ability to assess its impact on prazosin efficacy.

One of the key findings from our meta-regression analysis was the significant association between the use of benzodiazepines and prazosin efficacy for insomnia and PTSD severity. Studies with a higher percentage of benzodiazepine users, such as Raskind et al. (2003) and Ahmadpanah et al. (2014), reported the most substantial improvements in insomnia and PTSD symptoms. Conversely, the PACT trial, in which only 11.8 % of patients were benzodiazepine users in the prazosin group, did not show significant benefits. Our findings suggest a potential synergistic effect between the alpha-1 antagonism of prazosin and the GABAergic action of benzodiazepines, possibly addressing insomnia from multiple neurobiological pathways (Broese et al., 2012; Lie et al., 2015). PTSD patients show heightened noradrenergic activity, contributing to hyperarousal and trauma-related nightmares, leading to sleep disturbances. Benzodiazepines may help ameliorate insomnia by enhancing the inhibitory effects of GABA, thereby reducing hyperarousal (Miao et al., 2014). However, PTSD patients may exhibit altered sensitivity to GABA-A receptor, the main inhibitory receptor in the central nervous system (CNS), which may render these medications less effective (Möller et al., 2016). Therefore, they might benefit from a synergistic effect from additional neurotransmitter pathways. Prazosin, by blocking alpha-1 adrenergic receptors, reduces noradrenergic activity, thereby decreasing hyperarousal and improving sleep quality.

Unexpectedly, baseline symptom severity did not significantly influence prazosin’s efficacy across the three outcomes, suggesting that while benzodiazepine co-administration may enhance prazosin’s efficacy, this effect is independent of the severity of symptoms at baseline. Besides the lack of statistical power due to the small number of studies, this finding may be attributed to the low variability in baseline scores, particularly for nightmares and PTSD. While a similar trend might apply to insomnia symptoms, the use of different assessment tools for insomnia across studies hinders accurate comparison. In addition, while the overall PTSD severity reduction with prazosin was not statistically significant, the borderline p-value suggests that with a larger number of studies, this association might reach significance. Conversely, the limited effect of prazosin on global PTSD symptoms could be explained by the symptoms network approach, which views PTSD as a system of interconnected symptoms (Borsboom, 2017; McElroy et al., 2019; Melani et al., 2024). While prazosin significantly reduces individual symptoms such as insomnia and nightmares (two interconnected symptoms), its overall impact on PTSD severity may be limited by the persistent interaction of other interconnected symptoms.

Our finding that nightmares significantly improve with prazosin corroborates those of previous meta-analysis comparing prazosin to placebo following the PACT trial (Reist et al., 2021; Zhang et al., 2020). Nonetheless, Zhang et al.’s meta-analysis found no benefit over placebo for overall PTSD symptoms and sleep quality. In contrast, Reist et al.’s study reported positive results for all three outcomes. These discrepancies may be due to differences in the studies included in each meta-analysis. Additionally, while previous meta-analyses might have used the number of participants at baseline for the analysis, we applied more adequate methods by analyzing only the data of patients who completed the studies (endpoint number of participants), when appropriate analysis wasn’t performed to compensate for the loss of participants, such as intent-to-treat analysis.

The interpretation of our findings must account for some limitations. The small number of studies included may affect the robustness and generalizability of our results. The small number of studies limits statistical power, increasing the potential influence of outliers or study-specific characteristics on the results. As such, our sensitivity analysis removing the study from Ahmadpanah et al. (2014) showed a reduction in the magnitude of the effect of prazosin in treating insomnia. Furthermore, inconsistent reporting of variables, such as baseline blood pressure and time since PTSD diagnosis, limits our ability to perform detailed analyses and may obscure factors influencing prazosin’s efficacy. Another point that requires caution is that we compared means between placebo and prazosin at study endpoints instead of ideally comparing the mean changes (deltas) from baseline in each group. This was because the standard deviation of the delta differences couldn’t be calculated from the available data. Finally, while our meta-regression suggests an association between benzodiazepine use and prazosin efficacy, the lack of detailed dosage data across studies hinders a comprehensive assessment.

Despite these limitations, we were able to identify a potential synergistic effect between prazosin and benzodiazepines, which, if confirmed in a controlled trial, could significantly enhance treatment options for patients with PTSD. Combining prazosin with benzodiazepines might offer a more effective therapeutic strategy for insomnia in PTSD. Further, the addition of prazosin might allow for a reduction of benzodiazepine doses in these patients. This approach would be clinically beneficial by minimizing the risks associated with long-term benzodiazepine use, such as sedation, cognitive impairment, and dependence, especially in patients with PTSD, who have a higher predisposition to substance abuse than the general population (Ashton, 2005; Barker et al., 2004; Dell’Aquila and Berle, 2023).

5. Conclusion

While prazosin has demonstrated efficacy in treating trauma-related nightmares and insomnia in PTSD, some patients remain non-respondent. Understanding the factors that predict higher prazosin efficacy is crucial for individualizing treatment and improving outcomes. This meta-analysis and meta-regression analysis present two main findings. First, it corroborates previous meta-analyses indicating that prazosin is effective in reducing nightmares and insomnia in PTSD. Second, it suggests that benzodiazepine co-administration may enhance prazosin’s efficacy for insomnia and global PTSD symptoms, even when considering baseline symptom severity. Although other authors have suggested various factors that could contribute to prazosin’s mixed efficacy, none of the other analyzed variables in this study demonstrated a significant association. Notably, reduced use of antidepressants has shown a marginally significant improvement in prazosin’s efficacy for nightmares. However, the limited number of studies included in this review constrains the generalizability of our findings.

Further research is needed to explore the potential synergistic effect between prazosin and benzodiazepines and to determine optimal dosing strategies for treating insomnia in PTSD patients. Additionally, studies should investigate the efficacy of this combination in patients suffering from insomnia without PTSD. Given the marginally significant improvement of nightmares with reduced antidepressant co-administration, future research should also examine the association role of antidepressants in diminishing prazosin’s efficacy for nightmares. Ultimately, understanding these interactions could lead to more personalized and effective treatment plans, improving patient outcomes and quality of life.