Davis 2008.

Study characteristics
Methods	NEW TRIAL ADDED TO UPDATED REVIEW Design: multi‐centre trial, randomised, placebo‐controlled, parallel arm, flexible dose, double‐blind Duration of intervention: 8 weeks Placebo run‐in: Not specified Post‐treatment: Not specified
Participants	Setting: Quote: “Tuscaloosa Veterans Affairs Medical Center (TVAMC) and Birmingham VA Medical Center" Sample size: 85 randomised to divalproex and placebo Mean age: 55.2 years (SD, 6.8) Sex: 83 men and 2 women, 4 combat‐related trauma (5%) Diagnostic measure: DSM‐IV, CAPS Inclusion criteria: Quote: “Eligibility criteria included the following: age 19 to 70 years; stable medical condition; no substance use disorder (other than caffeine and nicotine) in the previous 2 months; no use of psychotropic medications for the previous 2 weeks (6 weeks for fluoxetine); no lifetime history of bipolar, psychotic, or cognitive disorders; no history of seizure disorder; no history of sensitivity to divalproex; and no current suicidal ideation, homicidal ideation, or psychotic symptoms that might interfere with the patient’s ability to give informed consent or preclude safe maintenance on divalproex monotherapy for the duration of the study. Women of childbearing potential had to use a medically approved method of contraception and could not be pregnant or breastfeeding during the course of the study" Exclusion criteria: See inclusion criteria Comorbidity: Not specified Dropout rates: 5/82 (2/41 in the divalproex and 3/41 in the placebo group)
Interventions	Pharmacological intervention: Quote: “Divalproex was started at a 500‐mg tablet twice daily and increased in 500‐mg increments, as tolerated, until maximum therapeutic benefit was achieved, but not to exceed 3000 mg/d. No other psychotropic medications were allowed during the study, except for low‐dose trazodone (50 mg/d), if needed, for insomnia"
Outcomes	Primary outcomes: CAPS (i.e. subscale: CAPS‐D) Secondary outcomes: CAPS (i.e. subscales: CAPS‐B, CAPS‐C), HAM‐A, TOP‐8, MADRS, CGI‐S, CGI‐I, DTS Time points: Quote: "Patients returned for assessment of symptoms and side effects at weeks 2, 4, 6, and 8 by a trained assessor" Data estimation: Quote: "Analyses of differences between groups in terms of response (≥30% improvement on CAPS and CAPS subscales) and the completers‐only groups were conducted (i.e. LOCF)"
Notes	Dates of trial: Not stated Industry‐funded: Yes. Quote: "This study was supported by a VA Research and Development Merit Award and an investigator‐initiated grant from Abbott Laboratories, Chicago, IL" Medication provided by industry: Yes Any of the authors work for industry: No
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization was determined by a random‐number generator with allocation sequence that was concealed in the pharmacy until the entire study was concluded"
Allocation concealment (selection bias)	Low risk	Quote: "Patients meeting the entrance criteria were randomized 1:1 to divalproex sodium (enteric‐coated, delayed‐release; Depakote) or visually identical placebo tablets, which were both supplied by Abbott Laboratories. Randomization was determined by a random‐number generator with allocation sequence that was concealed in the pharmacy until the entire study was concluded. Patients were assigned their randomization number at the time the study medication prescription was written"
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	The study was described as "double‐blind", although no information was provided on which parties were blinded and how blinding was achieved Quote: "This 8‐week, randomized, double‐blind, placebo‐controlled study was conducted at Tuscaloosa Veterans Affairs Medical Center (TVAMC) and Birmingham VA Medical Center, approved by the TVAMC and Birmingham VA Medical Center institutional review boards, and conducted in accordance with the Declaration of Helsinki and Good Clinical Practice Guidelines"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Trained assessor who was blind to medication assignment"
Incomplete outcome data (attrition bias) All outcomes	Low risk	A similar proportion of participants withdrew from the divalproex group (2/41, 5%) compared to the placebo group (3/42; 7%) Quote: "One hundred one (N = 101) US military veterans signed informed consent, and 85 were randomized. Of those not randomized (n = 16), 9 subjects were screen failures (4 for positive urine drug screen, 3 for unstable medical condition, 1 for CAPS <45, and 1 for current alcohol abuse), 4 subjects withdrew consent, and 3 were lost to follow‐up. Of the 85 randomized patients, 2 subjects failed to return, and 1 subject had an adverse event before receiving 1 postbaseline assessment, and they were considered nonevaluable. Thus, 82 evaluable subjects were considered in the efficacy analysis (n = 41 placebo, and n = 41 divalproex). Sixty‐eight subjects completed the entire 8‐week study (9 were lost to follow‐up or withdrew consent; 3 were withdrawn for adverse events [2 divalproex, 1 placebo]; one for a serious adverse event unrelated to study medication or procedures, and one was withdrawn for lack of efficacy [placebo]). All but 2 of the 82 randomized evaluable subjects were male (98%), and all but 4 had a combat‐related trauma (95%). There were no significant intergroup differences in age, sex, trauma type, or duration of illness. All randomized subjects who took at least 1 dose of study medication and were assessed for at least 1 postbaseline visit were included in the intent‐to‐treat population for efficacy (a priori definition). The primary method of analysis for efficacy comparisons between divalproex and placebo groups was a repeated‐measures analysis of variance model using the last observation carried forward, with treatment as qualitative factor"
Selective reporting (reporting bias)	High risk	Information was not provided for the GAF (Global Assessment of Functioning) outcome
Other bias	Unclear risk	Funding for study provided by industry Quote: "This study was supported by a VA Research and Development Merit Award and an investigator‐initiated grant from Abbott Laboratories, Chicago, IL" Unequal distribution of genders