Evidence-Based Medicine and Electrophysiology in Schizophrenia

Abstract

In research on schizophrenia electrophysiological measures have been investigated to identify biomarkers of the disorder, indices enabling differential diagnosis among psychotic disorders, prognostic indicators or endophenotypes. The present systematic review will focus on the most largely studied electrophysiological indices, i.e., qualitative or quantitative (limited to spectral analysis) EEG and the P300 event-related potential. The PubMed clinical query was used with research methodology filters for each of the following categories: diagnosis/prognosis/aetiology and a broad sensitive search strategy. The key-words: SCHIZOPHRENIA AND EEG/P3/P300 were used. The search results were then narrowed by including the terms “human” and “English language”, and cross-referenced. Systematic reviews and meta-analyses, when available, were also used for cross-referencing. Case reports and studies irrelevant to the topics and methodologies under examination were excluded. The remaining papers were screened to verify the eligibility for this systematic review. Inclusion criteria were: a) a diagnosis of schizophrenia confirmed by DSM-III/ICD-9 criteria (or later editions of the same classification systems); b) the inclusion of both a schizophrenia study group and an healthy control group (when appropriate, i.e., for P300 and quantitative EEG); c) qualitative or spectral EEG findings and amplitude measures for P300. The included studies were then reviewed to verify homogeneity of the results, as well as the presence of the information needed for the present systematic review and meta-analysis. Previous reviews and studies meeting the above requirements (n=22 for qualitative EEG; n=45 for spectral EEG and n=132 for P300) were classified according to the Oxford Centre for Evidence-based Medicine (EBM) levels of evidence criteria. For qualitative EEG as a diagnostic test, the majority of studies predated the introduction of DSM-III and were excluded from the review. Few post DSM-III studies investigated the usefulness of qualitative EEG in the differential diagnosis between schizophrenia and psychosis due to general medical condition. None of them was Oxford CEBM level 3b (non-consecutive-study or cohort-study without consistently-applied reference standard) or better (exploratory or validating cohort-study). No meta-analysis could be conducted due to the lack of reliable quantification methods in the reviewed studies. For spectral EEG as a diagnostic test, most studies qualified as level 4 (case-control study with poor reference standard), and only 24% as level 3b or better. An increase of slow activity in patients is reported by most of these studies. As to meta-analyses examining 29 studies, with 32 independent samples for the delta band and 35 for the theta band, a moderate effect size was found and only 1 study yielded findings in the opposite direction for both measures. There was no identified source for the discrepancy. The analysis of moderator factors included medication, band frequency limits, spectral parameters and disease stage. The medication status was significant for the theta band but the effect was unclear as findings for drug-naïve and drug-free patients were in a different direction. Chronicity had a significant effect on both delta and theta bands, with slow activity increase larger in chronic than in first episode patients. For P3 amplitude reduction as a diagnostic index, 63% of the studies qualified as level 3b or better. Meta-analysis (52 studies, 60 independent samples) results demonstrated a large effect size. None of the studies reported opposite findings. The analysis of moderator factors, including medication status and disease stage, revealed no significant effect on data heterogeneity. In conclusion, the examined indices are good candidates but are not ready yet for clinical applications aimed to improve present diagnostic standards for schizophrenia. Further research carried out according to adequate methodological standards and based on large scale multi-center studies is mandatory.