The basics of EEG
First developed in the 1920s, EEG entails applying electrodes to the scalp’s surface to measure electricity generated by neural activity. The brain works primarily via action potentials that lead to the release of neurotransmitters into synaptic clefts. Although EEG can’t detect the tiny amount of electricity produced by the individual neurons, visible waves or spikes are generated when neuron clusters fire synchronously.
EEG waves are categorized according to their frequency, measured in hertz (hz), or waves per second. The faster-frequency bands are generally associated with wakefulness, and the slower-frequency bands are associated with relaxation or sleep. A good way to memorize the names of the waves is to use the mnemonic “BAT-D”—as in, “You’ll go ‘batty’ if you don’t get an EEG.” This acronym starts with beta waves, which are the fastest (most alert) and ends with delta waves, which are the slowest (most asleep). You can remember that beta waves are the fastest by picturing betta fish darting around very quickly, and that delta waves are slow by picturing the languid Mississippi delta.
Using EEGs to “rule out” organicity
The most common reason for psychiatrists to order EEGs is to assess whether a patient’s psychiatric symptoms might have a neurological cause. This is usually in the context of a confusing clinical picture, where something doesn’t quite fit what we would normally expect for a diagnosis. Such atypical situations might include the following:
• Atypical psychosis (eg, acute onset in someone who was completely normal, or unusual symptoms, such as olfactory hallucinations)
• Episodic anxiety or rage attacks
• Dementia that appears to wax and wane, in which case EEGs might help differentiate between delirium and dementia
The literature is full of case studies of patients with other atypical symptoms who turned out to have some abnormal aspects to their EEGs. But how often do such abnormalities actually change the diagnosis or affect treatment? To answer and determine their diagnostic yield, you have to look at epidemiologic studies that analyze all EEGs ordered over a period of time. Such studies tend to throw a wet blanket on our zeal for EEG.
For example, in one retrospective study, researchers analyzed all EEG referrals to a large tertiary care hospital in Ireland (O’Sullivan SS et al, Hum Psychopharmacol 2006;(4):265–271). Over a one-year period, 1,470 EEGs were reviewed, of which 91 (6.2%) were referred from psychiatrists. Most of these referrals were for ruling out seizure or some other organic brain problem as causes of psychiatric symptoms. Of the 91 EEGs, 16 (17.6%) were read as “abnormal,” but these abnormalities were generally due to artifacts or to already established epilepsy. For example, the most common abnormality, occurring in 8 patients, was “excessive theta slowing,” consistent with a “mild encephalopathy.” But most of these patients were taking antipsychotics and/or benzodiazepines, both of which are known to cause this EEG finding.
Did the abnormal findings alter clinical care? Apparently not. The researchers were able to locate follow-up records of 53 of the 91 patients. Over a mean follow-up period of 23 months, none of these patients’ charts documented any diagnostic or medication changes related to the EEG results. In addition, 5 patients were referred for evaluation of olfactory hallucinations, and all had normal EEGs.
The authors recommended EEG only when there is a very high index of suspicion of pathology, such as in cases of witnessed seizures.
EEGs as biomarkers of psychiatric disorders
The newer batch of EEG studies explores whether tracings can be used to diagnose specific psychiatric disorders. For example, for mood disorders, the largest trial to date was an international, prospective, randomized trial that compared baseline EEGs in 667 patients with major depressive disorder (MDD) and 336 healthy controls. While some differences between groups were noted, none significantly correlated with MDD (Arns M et al, Clin Neurophysiol 2016;127(1):509–519).
Using EEG to diagnose ADHD seems slightly more promising. In one large study, EEG did demonstrate differences between children with ADHD (n = 347) and controls (n = 619). However, these differences overlapped with autism spectrum disorder, limiting the utility of EEG to aid in ADHD diagnosis (Duffy FH et al, BMC Med 2017;15(1):51). The inconclusive nature of the data led the American Academy of Neurology to explicitly discourage the use of EEG for diagnosing ADHD.
EEG to predict antidepressant response It’s possible that EEG might be able to identify patients who are destined to respond to certain medications, making our jobs easier when it comes to choosing among the wide array of treatments. The most promising research has been done by a group at UCLA. Their method has been to put patients on escitalopram (Lexapro) for one week, then do an EEG, and see whether the tracings can predict which patients eventually respond. The most recent paper by the UCLA group, whose research has been ongoing for many years, compared 146 patients on Lexapro with 48 on placebo
(Leuchter AF et al, J Psychiatr Res 2017;84(1):174–183). The marker predicted remission with about a 60% sensitivity and 75% specificity (note that we’re estimating, since the paper didn’t clearly report these values for the entire sample). This means it correctly identified 6 out of 10 future remitters. Of every 10 future Lexapro responders, the test would identify 6 of them, but would miss 4. While intriguing, this EEG prediction technique is probably no improvement on clinical judgment, and so is not yet ready for prime time.
TCPR VERDICT: At this point, EEGs in psychiatry are mainly a research tool. There are very few clinical scenarios in which an EEG will actually affect your diagnosis or treatment. If your patient had an episode that looked like a seizure to an outside observer (as opposed to a subjective symptom), an EEG is indicated to help establish the presence of some type of epilepsy. But in others, the test will most likely end up as a piece of complicated, expensive, and ultimately ignored data.
EEG Waves, Fast to Slow
Beta (14–30 hz): Alert, intellectual activity
Alpha (8–13 hz): Relaxed, daydreaming
Theta (4–7 hz): Deep relaxation, meditation
Delta (1–3 hz): Deep sleep