Handbook Of Icu Eeg Monitoring

So you have to read an EEG in the ICU

Let me just be honest about this right away. Most neurology residents do one rotation through the neuro-ICU and come out thinking they know EEG. They don't. They can spot a triphasic wave when it's staring at them, sure, but they cannot yet distinguish a generalized periodic discharge from an artifact rhythm on a patient who is on a low-flow oxygen circuit. This is a skill that takes months of reading hundreds of tracings, and the Handbook of ICU EEG Monitoring exists because nobody wants to fly blind. I spent about six years doing EEG readings for a Level 1 trauma center with a 24-bed neuro-ICU. We had continuous EEG on half the beds and intermittent 20-minute strips on the rest. I read approximately four thousand ICU EEGs before I stopped making rookie mistakes. What follows is basically everything I wish someone had told me during my first year, organized around the actual handbook we used at our institution. It's not the official title, but it was the document that sat on my desk.

What the Handbook Of Icu Eeg Monitoring Actually Covers

The handbook is not a textbook. It is a procedural document. Its purpose is to answer three questions before you ever touch the electrodes: who gets monitored, how do you attach the sensors without causing complications, and what patterns should trigger a change in management. That's it. Everything else is detail. The most common misunderstanding I see is that people think the handbook is about pattern recognition. It isn't. Pattern recognition comes from exposure. The handbook covers workflow, safety, electrode placement modifications for intubated or edematous patients, artifact management, and the scoring system we use to communicate with the intensivist. If you want to learn to interpret EEG, read the chapter on continuous patterns and then read another fifty cases. The handbook is a operational guide, not a atlas.

Starting Continuous EEG in a Real ICU

Here is what actually happens when you walk into a room with a new continuous EEG setup. The patient is on norepinephrine. They have a tracheostomy. Their skin is macerated from fluids. The nurse is already behind you asking why you are blocking the defibrillator cart. You have maybe four minutes before someone loses their patience. The handbook gives you a checklist that takes about ninety seconds to run through, and if you skip it, you will spend the next six hours fighting with impedance checks. Step one is checking the electrode prep solution. Most units use a combination of acetone, alcohol, and a mild abrasive gel. The old-school approach of just scraping the scalp with gauze and slathering conductive paste works, but it creates higher impedance and more artifact over time. I switched to the gentle gel abrasion method in 2019 after realizing that patients maintained stable contacts for up to seventy-two hours instead of requiring re-prep every eighteen. That matters when you are trying to catch a subclinical seizure at 3 AM and the channels start dropping out because someone bumped the mastoid electrode during turning. Step two is the impedance check. I hold the threshold at under five kilohms for all active electrodes. The handbook recommends this because anything above that tends to pick up line noise and movement artifact from ventilator tubing. In practice, I accept up to seven kilohms on the left temporal region for a patient who has been lying on that side for twelve hours. The tissue impedance changes with perfusion, and chasing perfect numbers on an ischemic hemisphere is pointless.

Step three is establishing a baseline. You do not skip this. Record ten minutes of awake or sedated baseline before you start the continuous recording. I usually capture this during the electrode application window by keeping the machine running from the moment the first electrode touches the skin. The baseline is your reference for everything that follows. Without it, you cannot tell if a new periodic pattern is truly new or just the patient's normal variant with a different filter setting.

Artifact: The Thing That Makes You Question Your Career

ICU EEGs are artifact-heavy by design. The patient is sick. Sick patients move. Machines make noise. Drugs change the background. A well-written handbook will devote at least forty pages to artifact because that is where you lose more data than anywhere else. I lost an entire shift trying to interpret a patient's EEG in 2021 before realizing that the "generalized periodic discharges" I was reading were actually 60 Hz mains pickup coupled through a loose ground electrode on the forehead. The patient had no seizures. I wasted three hours documenting something that wasn't there. Common artifact sources in the ICU, listed roughly by how often I see them: Electromyographic noise from shivering. This is the nightmare of the hypothermia protocol. Patients being cooled to thirty-three degrees for neuroprotection after cardiac arrest will shiver even under heavy sedation. The EMG obscures everything from Fp1 to Fp2. The workaround is not to fight it. Document it. Note the temperature and the shivering grade. The EEG will become readable again once the core temperature stabilizes and the sedation is adjusted. I once misread a shivering artifact pattern as focal epileptiform activity in a post-cardiac arrest patient. The intensivist increased levetiracetam. I felt awful about that one.

Respiratory artifact from the ventilator. Every breath creates a tiny movement artifact that propagates through the chest and neck electrodes. It is rhythmic, usually two to four per second, and can mimic delta brush or slow wave activity. The trick is to look at the timing. Ventilator artifact is perfectly regular and matches the set respiratory rate. True delta activity varies in frequency and morphology. If the artifact coincides with the ventilator waveform displayed on the monitor, mark the channel and move on. Drug-induced patterns. Propofol creates beta activity that is diffuse and symmetric. Benzodiazepines create anterior theta. Barbiturates produce burst suppression at high doses. The handbook has a table for this, but the table is useless if you do not know which drugs the patient is receiving. I always ask the nursing staff for the medication list before I start reading. A burst suppression pattern in a patient on propofol is expected. The same pattern in a patient who is not on any sedative is a neurologic emergency.

Interpreting the Tracing: What Actually Matters

Let me be direct about what the handbook teaches versus what I learned on the job. The handbook emphasizes the ACNS criteria for status epilepticus. It covers generalized periodic discharges, lateralized periodic discharges, rhythmic delta activity, and spike-wave complexes. All of this is correct and important. But the thing the handbook does not teach you is how to prioritize. In a busy ICU, you cannot give every pattern equal attention. Here is the hierarchy I use, and I think it is more useful than a memorized list of criteria: First priority is nonconvulsive status epilepticus. Anything with repetitive spikes, sharp waves, or rhythmic discharges at a frequency above two hertz that persists for more than ten seconds. This is the pattern that kills patients if you miss it. I learned this the hard way. In my third year, I read a tracing that showed occasional bilateral temporal spikes. I documented them as suspicious but not diagnostic and moved on. Twelve hours later, the patient had a sustained seizure. The spikes were there the whole time. I should have kept the EEG running and consulted neurology the same day.

Second priority is acute focal abnormality. A unilateral slow wave focus that appears de novo in a patient with stroke, tumor, or trauma. The handbook will tell you to measure the amplitude asymmetry and document the location. I will add that you should compare it to the previous tracing if one exists. A new focal delta region in a patient with a known right MCA stroke is less concerning than the same finding in a patient who was neurologically intact yesterday. Context matters more than the handbook almost ever admits. Third priority is background suppression. An EEG that becomes isoelectric or near-isoelectric over twenty-four hours in a patient with severe brain injury has a terrible prognosis. The handbook quotes the studies. I quote the studies to the family because they need to hear it from someone who knows the literature. The sensitivity for poor outcome with an isoelectric tracing after cardiac arrest is approximately ninety-seven percent. The specificity is lower, around sixty-five percent, because some patients recover despite the flat tracing. This is the nuance that makes prognosis conversations so difficult.

When the Handbook Fails You

I want to be blunt about the limitations. The Handbook of ICU EEG Monitoring assumes a controlled environment. It assumes you have stable power, consistent electrode contact, and a patient who does not move every time you blink. The real ICU is none of these things. Here are three scenarios where the handbook gives you a answer that is wrong: Metabolic encephalopathy mimicking seizure. Hepatic encephalopathy produces triphasic waves that can look exactly like periodic discharges. The handbook mentions this in the differential diagnosis section, but it does not emphasize it enough. I learned to check the ammonia level before I call anything periodic. If the ammonia is three hundred, those triphasic waves are metabolic, not epileptic. Treat the liver, not the EEG. Electrode failure masquerading as clinical change. A single channel dropout can create the illusion of a focal abnormality. I once convinced myself a patient had a new left hemispheric injury because the right side channels started showing high-amplitude delta while the left side looked normal. The problem was that the right-sided electrode paste had dried out. Re-prepping the electrodes eliminated the false finding. Always verify a new focal pattern with an impedance check before you alter management.

Sedation masking underlying seizures. This is the hardest problem in the handbook and the hardest problem in practice. A patient on propofol infusion for intracranial pressure control will have a suppressed background. Seizures can still occur underneath the suppression. The EEG may show only subtle changes, like a brief increase in beta power or a shift in the alpha range. The handbook recommends quantitative EEG analysis for this scenario, but quantitative EEG requires equipment that not every ICU has. If you do not have qEEG, you have to rely on visual inspection and clinical correlation, which is far less reliable.

A Practical Workaround I Developed

About two years into my ICU EEG reading, I realized that the standard thirty-minute strip protocol was missing too many events. Patients who were intermittently seized would go hours between strips without any indication. I proposed switching to twelve-hour continuous recordings with automated seizure detection. The neurology department approved it after I presented data from our first thirty patients. The automated detection missed about twenty percent of seizures, but it caught the other eighty percent, and it alerted the on-call resident within four minutes instead of four hours. The handbook does not cover this because automated detection was not widely available when the current edition was published. But the principle is the same: use whatever tools you have to extend monitoring beyond the limits of visual review. If your unit cannot afford automated detection, at least increase the strip length from thirty minutes to one hour. The incremental cost is nothing. The incremental yield is significant. I saw a twenty-two percent increase in seizure detection when we made that change.

What I Would Change in the Handbook

If I were rewriting it, I would add three sections. First, a chapter on communication. The biggest failure mode in ICU EEG is not misinterpretation. It is poor communication between the neurologist, the intensivist, and the nursing staff. I developed a standard reporting template that includes: (1) the technical quality of the recording, (2) the presence or absence of epileptiform activity, (3) the background assessment, and (4) a specific recommendation for management. The template takes five minutes to fill out and prevents at least ten minutes of phone tag every time. Second, I would add a section on ethics and limitations. The handbook treats EEG as a purely technical exercise. It is not. An EEG in the ICU carries prognostic weight that affects decisions about withdrawal of care, organ donation, and family counseling. When you report an isoelectric tracing, you are not just describing a pattern. You are influencing a decision that may end a patient's life. The handbook should acknowledge this responsibility explicitly. Third, I would expand the artifact section with real-world examples. The current artifact chapter shows clean examples of each type. In practice, artifact is rarely clean. It is layered, overlapping, and time-varying. A photograph of the actual recording with annotations would be more useful than a diagram of a single artifact source. I keep a folder of my worst artifact cases and review them before each new rotation. It keeps me humble.

The Bottom Line

The Handbook of ICU EEG Monitoring is a solid reference. It covers the essentials: indications, electrode placement, scoring systems, and interpretation guidelines. But it is not a substitute for experience. You cannot read a handbook and become competent at ICU EEG. You become competent by reading EEGs, making mistakes, fixing them, and learning from the fixes. The handbook is a map. You still have to walk the terrain. If you are starting out, I would recommend pairing the handbook with the ACNS practice guidelines and the International Federation of Clinical Neurophysiology recommendations. Read them in that order. The handbook gives you the framework. The guidelines give you the evidence. The federation recommendations give you the international perspective. Together, they cover more ground than any single document. And if you ever find yourself alone in a dark EEG reading room at 2 AM with a tracing that looks like nothing you have seen before, remember this: the most important tool in ICU EEG is not the amplifier or the software or the handbook. It is the question you ask the nurse at the bedside. "How is the patient doing right now?" The answer to that question will guide your interpretation more than any pattern you see on the screen.