One single-blinded study (n=16 patients), four open-label studies (n=63 patients) and three case reports (n=4 patients) were included. Overall studies 34% (28/83) dropped out. Reasons for dropouts included intolerable side-effects (n=14 patients) and non-compliance (n=4 patients). Studies scored between 22 and 25.5 out of 46 points for quality, indicating poor to moderate quality. Methodological problems included small sample size, absence of control groups, attrition bias, and limited generalisability of results.
Meta-analysis was based on four studies (one single-blinded trial and three open label trials, 63 patients total before treatment, 38 after eight weeks). Valproate was associated with a statistically significant reduction in PTSD symptoms at eight weeks compared to baseline, Hedge’s g -0.982 (95% CI: -1.499 to -0.465 using random-effects model), and a significant reduction in depression from baseline, Hedge’s g -0.775 (95% CI: -1.274 to -0.276 using random-effects model). No significant heterogeneity was found for either analysis (p=0.274 for PTSD and p=0.286 for depression). Results for fixed-effect models were similar.
All three case reports described improvements in hyper-arousal symptoms post-treatment with valproate.