Meta-analysis in 2004 of the pooled results was limited because of the following problems.
In Porsteinsson 2001, although the physicians having direct responsibility for patient care were blinded, a non-blinded physician, who had no direct contact with these physicians, adjusted divalproex sodium dosage on the basis of reports from blinded raters and from confidential laboratory reports. Therefore, because the physician who controlled therapy knew which patients were receiving divalproex, the trial did not satisfy the criterion of concealed allocation.
In Tariot 2001, 54% of the treated patients dropped out compared with 29% of control patients. Of all treated patients, 22% dropped out because of adverse effects, and the study had to be discontinued prematurely.
The third trial (Sival 2002) had a cross-over design. No results from the first phase of the study were available, and although the statistical section stated, "the t-test for independent samples is used to analyse the two-period cross-over trial", because the samples were not independent - they are the same patients in the treatment and placebo groups - a question must be raised about the correctness of the analyses.
The valproate preparation used in the trials varied - one used short-acting sodium valproate, one long-acting divalproex sodium, and the third early-onset acting divalproex sodium. Average doses differed (480 mg/d - 1000 mg/d), as did duration of therapy (3 weeks - 6 weeks), and ways of evaluating patients and their response to therapy.
A limited meta-analysis, pooling the results concerning adverse effects (Porsteinsson 2001, Tariot 2001) revealed the following: sedation occurred more frequently in patients treated with valproic acid than in controls. Urinary tract infection was more common among patients treated with valproic acid than controls.
An updated systematic review (October 2008) of two new studies (Tariot 2005, Herrmann 2007) applied meta-analysis to the effect of valproate on agitation in demented patients and also combined these studies with the earlier reports to examine adverse effects among valproate treated patients. Because the study of Herrmann et al involved a cross-over design, only those results from the first part of this study were included in the updated review.
The new meta-analysis of pooled results showed no improvement of agitation among valproate treated patients, compared with controls, and showed an increase in adverse events (falls, infection, gastrointestinal disorders) among valproate treated patients.
