Assessment of malondialdehyde levels in schizophrenia: a meta-analysis and some methodological considerations.

Oxidative stress has long been suggested to participate in the pathophysiology of schizophrenia. Most of the published data on the subject rely on malondialdehyde levels assessment through thiobarbituric acid reactive substances (TBARS), despite significant methodological concerns. The present paper reports on a meta-analysis of studies published up to July 2006. Relevant studies were identified through PubMed search and the References section of experimental and review papers on the subject. Studies were retained for analysis if they provided adequate methods description, controls and subjects numbers, means and standard deviations. Using a random effect model, an effect size of 1.22 (CI: 0.64-1.80) was found, with significant bias and high heterogeneity. Inspection of the bias assessment plot suggested a lower value in the order of d approximately 0.5, in line with the results of the larger studies. The heterogeneity of the results appeared to be unrelated to assay type or biological sample. However a meta-regression analysis suggested an association, at trend level, between the precision-weighted effect size and the proportion of drug-free patients, which might therefore account for some of the heterogeneity.