What's Wrong With EEG Interpretation

Andrew N. Wilner, MD; Selim Benbadis, MD


March 02, 2022

Andrew N. Wilner, MD: Welcome to Medscape. I'm Dr Andrew Wilner, reporting on the 75th annual American Epilepsy Society meeting.

With me today is my old friend and colleague, Dr Selim Benbadis, professor of neurology and director of the Comprehensive Epilepsy Center at the University of South Florida in Tampa, Florida. Welcome, Selim.

Selim Benbadis, MD: Thank you, Andrew.

Wilner: You led a special interest group at this year's meeting on the persistent problem of EEG overinterpretation. How common is it?

Benbadis: It's very common. And you're right to say it's a persistent problem, because you, me, and others have been talking about this for many, many years, and yet it doesn't go away.

As you know, epilepsy centers have the same experience in that about 30% of patients that come for intractable seizures turn out to not have seizures. Part, if not all, of the reason they've been misdiagnosed as having epilepsy is because of an overread EEG.

Wilner: I read EEGs pretty much every day, sometimes for the same patient. Just the other day I was looking at an EEG and said, "Oh my goodness, this guy's having subclinical seizures." And of course, that's why we ordered it. He was comatose. Then I looked at the filter settings and someone had changed them. They weren't seizures at all, just muscle activity that was read as seizures because of the filter settings.

There's a lot of technical aspects that people have to know to not make mistakes. I've been doing this for about 30-40 years, and even I'm still subject to almost making an error. How much training do you need before you can get this right?

Benbadis: That's a great question, and one that I'm not sure I know how to answer. But what we do know is that right now there is zero mandatory training. That's really amazing and something that the non-neurology medical community doesn't know and would think is odd. To me, that would be like cardiologists not being trained on ECG.

As you know, the reality is not as bad because most residency programs make it mandatory locally in the program. But it's not mandated by the Accreditation Council for Graduate Medical Education, which I think is amazing. So I don't know the answer to your question about how much EEG reading training is needed, but it is some and it should be mandatory for at least, let's say, 2 months.

With ECGs, you can argue that internists, pediatricians, and others read those. But neurologists are not a subspecialty of medicine, so we're the only people who read EEGs.

Unfortunately, because of the absence of mandatory training, many neurologists in the community who want to read EEGs have limited or no training. And what happens when you haven't had enough training and experience is that you'll get anxious; you don't underread but rather overread.

It's like if someone asks you and me to look at an echocardiogram or a CT of the abdomen. We haven't looked at one of those in years and would be very anxious trying to figure out what is this or what is that. It's normal human nature. But there should be some mandatory training in neurology residency.

Wilner: During this symposium, I was surprised to learn that only about one third of neurologists actually read EEGs in regular practice, or at least had training to do so.

Do you think there should be a certification that's required by some official body that says, "I know how to read EEGs?"

Benbadis: Well as you know, we have certifications, they're just not required and certainly not to read routine EEGs in the real world. Now, if you go to a large hospital or to an academic center, they are required. We have several EEG boards to provide that certification.  

So I don't know if we need to go that far, but at the least during training there should be documentation that you've looked at a minimum of 500 or 1000 routine EEGs, unless you've done an EEG rotation for 2 or 3 months. Something like that may work. But right now, there is none, which is absolutely amazing. And try saying that to colleagues, such as to internists, and you'll see that they're shocked and will ask, "How is that possible? Who is going to read the EEG if it's not the neurologist?"

Best Way to Read an EEG: Blind or With Context

Wilner: You and I are definitely on the same page with this. I'm still trying to improve my own reading, so I was curious, when you read an EEG, do you read it blind without the clinical history or do you like to have the clinical history?

Benbadis: That's a great question. In fact, a colleague of mine is doing a survey on that very question to neurologists.

The strategy should be to read the EEG blind the first time without being biased by the history. You decide whether the EEG is normal or abnormal, and if it's abnormal, you identify in writing why that's so. Once that's done, then you can go back and look at the history. And you are allowed to look at the EEG again. That's when you're going to formulate the part that we call the clinical interpretation, which explains what the findings mean; for that, you absolutely need the history.

That's what we do and what we train residents to do. You should make up your mind about the EEG without being blinded by the history. That's the healthy way to do it.

Now we both know that in private practice, you're in a rush. Oftentimes, you're reading EEGs on patients you know, who you saw in clinic. This makes it hard. But at least in theory, that's the way it should be done.

Wilner: I like the idea of taking two passes at it. Because, for example, if you know the patient has had a hemispherectomy, you're probably not going to say it's an artifact when there's a difference between the two hemispheres. But if you didn't know that, you might read it a little differently.

We live in a modern digital age. How much help do you think computer software programs like spike and seizure detection offer? When I was a resident, Jean Gotman worked next to me. He was a young man developing spike and seizure detection, and that was a long time ago. Should we use that stuff or not?

Benbadis: It's interesting. One of my faculty raised the topic of whether computers can save us from this problem of overinterpretation? And the answer is no. They help for a lot of things, and they definitely help for detecting trends in the intensive care unit. And spike and seizure detections are good, but they do not replace clinical judgment and a good, experienced neurologist. Instead, they complement them and are useful.

Why EEG Is Still a Relevant Technology

Wilner: Let me ask a slightly provocative question. EEG is a very old test. We're talking about a technology that was initially developed by Hans Berger in 1924. That's almost 100 years ago, and we still use this technology in every hospital, every day. So, is there a future for this?

Benbadis: There is. We use EEG, but we don't use it like it was used 100 years ago. For most things, imaging has supplanted EEG. There's no question that structural imaging; MRI; even functional imaging, PET, and single-photon emission CT (SPECT), and so forth; and new nuclear scans can do better than EEG.

But there are definitely a few areas where EEG will not be replaced by imaging, and epilepsy is one of them because it's a disorder of electricity. You can never diagnose epilepsy on the basis of structural imaging or even functional imaging. You need electricity. So there are certain things that nothing but EEG can do.

That being said, I do agree that it's overused and often used for the wrong reasons, and that's led to problems, including misinterpretation.

Wilner: Dr Benbadis, this has been a great discussion. Are there any last thoughts you'd like to share with the audience?

Benbadis: No, I think you've asked all of the right questions. I would just reiterate that we do need more education and some sort of mandatory or minimum training for future neurologists.

Wilner: Thank you for speaking with Medscape.

Benbadis: Thank you for having me.

Wilner: I'm Dr Andrew Wilner, reporting from the annual American Epilepsy Society meeting.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.