The review concluded that electroencephalography may be useful as an adjunct in the diagnosis of dementia with Lewy bodies and Alzheimer’s disease and may help to confirm, refute or differentiate between these disorders and other dementias. In light of several review limitations, the authors' conclusions should not be considered as being reliable.

To evaluate the diagnostic utility of visual electroencephalography (EEG) in early-onset dementia focusing on Alzheimer’s disease, vascular dementia, dementia with Lewy bodies and frontotemporal dementia.

MEDLINE, EMBASE, Web of Knowledge and Scopus were search to 2012 (month not stated) for full-text studies; no search terms were reported. Google Scholar and reference lists of relevant articles were searched for further studies.

Studies on the use of visual EEG in patients below the age of 65 with suspected or confirmed Alzheimer’s disease, vascular dementia, dementia with Lewy bodies or frontotemporal dementia were eligible. Studies in which the populations straddled age 65 were included. Patients had to be clinically diagnosed with the most up to date diagnostic criteria. Studies had to report diagnostic accuracy or diagnostic utility outcomes.

Most studies focused on patients with Alzheimer's disease or dementia with Lewy bodies. Where reported, disease durations ranged from 1.7 to 4.9 years. In five studies the mean ages of patients were considerably higher than 65 years. One study included a healthy controls group.

The authors did not state how many reviewers selected studies.

Study quality was evaluated using the QUADAS tool. Poor quality studies were excluded but there were no details of what constituted a poor quality study.

The authors did not state how many reviewers performed the quality assessment.

The authors did not state how many reviewers extracted data.

There were clinical differences between studies so a narrative synthesis was performed, with studies grouped by diagnosis. Correlations between disease duration, disease severity, age and gender were noted. The authors stated that only statistically significant findings were included in the review.

Twelve studies (965 patients) were included in the review. Limited QUADAS results were reported. Not all patients had pathological confirmation of diagnosis. Many studies did not meet reproducibility criteria for EEG methodology and it was often unclear whether interpreters were blinded to diagnosis and clinical information. Confounding appeared to be an issue in some studies.

Alzheimer's disease: Three studies reported a correlation between EEG score and dementia severity and three studies found that EEG matched quantitative EEG in diagnostic utility. The most common EEG finding in early-onset Alzheimer's disease was an increase in slow wave activity (three studies).

Dementia with Lewy bodies: In two studies all patients had abnormal EEGs with loss of alpha dominance. For patients with vascular dementia, in one study patients with high Hachinski scores (>6) were significantly more likely to have focal abnormalities or lateralization compared with patients with a low score; diffuse abnormalities were equally common regardless of the ischaemic score.

Frontotemporal dementia: Results from two studies found that patients had abnormal EEGs but posterior dominant rhythm was preserved.

Results relating to individual studies were reported.

Electroencephalography may be useful as an adjunct in the diagnosis of dementia with Lewy bodies and Alzheimer’s disease and may help to confirm, refute or differentiate between these disorders and other dementias.

The review question and eligibility criteria were broad. In many studies the population age was considerably higher than 65, raising questions about the relevance of these studies to the early-onset population. Several relevant electronic databases were searched. The restriction to searching only for published studies meant that some relevant studies may have been missed. The authors did not report using methods such as independent duplicate study selection and data extraction to minimise reviewer error and bias during the review process.

A narrative synthesis approach appeared appropriate but was limited because full results for all studies were not reported. The authors stated that only statistically significant findings were included in the review. Quality assessment results were reported in a very limited way. It appeared that many studies may have been affected by important biases. The authors noted that most of the studies tended to be descriptive of the EEG patterns in disease.

Limitations in the review mean that the authors' conclusions should not be considered as being reliable.

Practice: The authors stated that EEG may provide a valuable contribution as part of a wider diagnostic evaluation panel.

Research: The authors stated that prospective well-powered studies were needed to investigate diagnostic utility more robustly (including use of serial EEG recordings). Studies should look at the ability of EEG to differentiate between the types of neurodegenerative dementia.

NHS Research Scotland.