Guidance for Treatment of Epileptiform Activity in Critically Ill Patients

Epileptiform activity is detected in more than half of critically ill patients who undergo EEG. When prolonged, it is associated with in-hospital mortality, and many survivors have long-term functional and cognitive disabilities.

Subscribe to the latest updates from Neuroscience Advances in Motion

Thank you for subscribing!

Error: Please enter a valid email address.

Email Address Submit

For years some neurologists have held that epileptiform activity has a causal role in worsening outcomes and thus requires aggressive treatment. Others attribute the worse outcomes to other factors, such as medication side effects or the underlying illness.

Now, in a uniquely designed observational study, researchers in the Clinical Data Animation Center at Massachusetts General Hospital have detected a high likelihood of a causal relationship. Haoqi Sun, PhD, an instructor in Neurology at Beth Israel Deaconess Medical Center, Sahar F. Zafar, MD, MBBS, associate professor in the Department of Neurology at Mass General, M. Brandon Westover, MD, PhD, neurologist–investigator at the Mass General Research Institute and director of the Clinical Data Animation Center, and colleagues give treatment recommendations in The Lancet Digital Health.

Rationale for the Study Design

In a usual retrospective study, the effect of epileptiform activity on neurological outcome is confounded by a feedback loop: physicians administer antiseizure medications based on the patient's epileptiform activity, and epileptiform activity is affected by the medications.

To disentangle the two, the Mass General team studied a "counterfactual scenario" and estimated the degree to which untreated epileptiform activity might worsen neurological outcomes.

Methods

The retrospective, cross-sectional study included 995 adults admitted to the Mass General ICU between December 1, 2011, and October 14, 2017, and had epileptiform activity identified by a clinical neurophysiologist or epileptologist. Epileptiform activity could be seizures, lateralized periodic discharges, generalized periodic discharges, or lateralized rhythmic delta activity.

For this analysis a deep neural network and automated algorithm classified every two-second EEG segment as containing epileptiform activity or not. A timeseries was generated as the fraction of two-second EEG segments containing epileptiform activity over a six-hour window, chosen to allow observation of the effects of antiseizure medications and adjustment of medication.

Epileptiform activity burden was measured across all six-hour windows within the first 24 hours of EEG as the mean and maximum epileptiform activity fraction (EA_mean and EA_max, respectively).

The two measures separated potentially different effects of prolonged periods of less intense epileptiform activity (EA_mean) from intense but brief epileptiform activity (EA_max).

Neurological Outcome

The researchers combined pharmacological modeling with a method of matching each patient to others with similar demographics, clinical factors, admission diagnosis, and drug responsiveness. The outcome was the modified Rankin Scale score at hospital discharge, dichotomized as poor (≥4) or favorable (≤3).

It was estimated that in an untreated cohort:

• Patients with higher levels of EA_max were at higher risk of poor neurological outcome
• The risk of a poor outcome increased monotonically as EA_max increased, culminating in a mean increase of 22% as a patient's untreated epileptiform activity increased from mild (0% to <25%) to very severe (75%–100%)
• Patients with higher EA_mean were also at higher risk of a poor outcome, and similar to EA_max, the risk increased monotonically with an increase in epileptiform activity burden
• Severe or very severe epileptiform activity burden longer than 24 hours increased the risk of poor outcome by a mean of 18% compared with mild prolonged epileptiform activity burden
• Moderate but long-lasting epileptiform activity (EA_mean of 2% to <10%) increased the risk of poor outcome by a mean of 14%

Effect Heterogeneity

Patients with hypoxic–ischemic encephalopathy or acquired brain injury were at the highest risk of a worse outcome in response to a large EA_max, possibly because inflammation exacerbated the harm.

Implications for Better Treatment

Current approaches to treating epileptiform activity tend to be generic and based solely on the duration of activity. This study has two principal clinical implications:

• Interventions should be prioritized for patients with a mean epileptiform activity burden of ≥10%, and treatment should be more conservative when the maximum epileptiform activity burden is lower
• Treatment should be tailored to individual preadmission profiles because the potential for epileptiform activity to cause harm depends on medical history and reason for admission

The research team is organizing an international data-sharing collaboration in hopes of developing an optimal treatment protocol.

view original journal article Subscription may be required

Learn more about the Clinical Data Animation Center

Visit the Department of Neurology