Authors :
Doyle Yuan, MD – University of Texas Southwestern Medical Center
Irfan Sheikh, MD – University of Texas Southwestern Medical Center at Dallas
Roohi Katyal, MD – Louisiana State University Health Sciences Center
Wei Zhao, MD – Washington University in St. Louis
Kevin Hochstrasser, MD – Washington University in St. Louis
John McLaren, MD – Boston Children's Hospital
Peter Kaplan, MD – Johns Hopkins
Ioannis Karakis, MD, PhD – Emory University
Stephan Schuele, MD, MPH – Northwestern University
Luisa Londono, MD – Epilepsy Center, Neurological Institute, University Hospitals Cleveland Medical Center, Cleveland, OH, USA
Renée Shellhaas, MD, MS – Washington University in St. Louis
Elza Yacubian, MD – Universidade Federal de São Paulo
Adriana Bermeo-Ovalle, MD – Rush University Medical Center
Rebecca Matthews, MD – Emory University
Selim Benbadis, MD – University of South Florida
Jay Gavvala, MD – UTHealth Houston
Katia Lin, MD, PhD – Universidade Federal de Santa Catarina - UFSC
R. Edward Hogan, MD – Washington University in St. Louis
Adam Greenblatt, MD – Washington University School of Medicine
M. Brandon Westover, MD, PhD – Beth Israel Deaconess Medical Center
Sándor Beniczky, MD, PhD – Aarhus University Hospital, Member of European Reference Network EpiCARE and Danish Epilepsy Centre
Presenting Author: Fábio Nascimento, MD – Washington University School of Medicine
Rationale:
The American Clinical Neurophysiology Society recommends that reporting of routine EEG (rEEG) should consist of an objective description of the findings and a subsequent interpretation that correlates the findings with the clinical indication for the EEG. This recommendation is founded upon the importance of describing rEEG findings without clinical information to prevent bias (“history bias”). However, prior survey-based data have shown that this is not consistently followed by EEG readers. This study examined whether presenting clinical information alongside EEGs alters the interpretation of these studies by epileptologists.
Methods: Experts independently rated the same 25 full-length, outpatient, adult (52%) and pediatric (48%) rEEGs (duration of 20-62 minutes) on two separate occasions (Parts I and II) ≥60 days apart. In Part I, rEEGs were presented without additional clinical information (except for age and sex). In Part II, the same rEEGs were presented but this time accompanied by a clinical vignette prepared by the authors. Experts were not informed about the design of the study nor were they told the rEEGs would be the same in both Parts. The 25 clinical vignettes were rated by four authors as to the likelihood of epilepsy, yielding a consensus score from +4 to -4. Cases with a score of +2, +3, or +4 were considered “likely epilepsy” (n=14), while those with a score of -2, -3, or -4 were considered “unlikely epilepsy” (n=6). EEGs were presented to experts using our state-of-the-art online platform (“EEGHub”). For each recording, experts were asked to indicate the presence and type of abnormalities, specifically focal epileptiform; generalized epileptiform; focal non-epileptiform; generalized non-epileptiform; focal seizures; and generalized seizures. We used McNemar’s test to assess whether clinical history changed EEG interpretations in one direction more than the other.
Results: Twelve experts (mean of 22 ± 9.5 years reading EEG) completed both Parts. Among “likely epilepsy” cases, 30 responses (18%) changed from negative to positive for the presence of any abnormality, while 9 responses (5%) changed from positive to negative (p=0.001). Subgroup analyses of flipped decisions on the presence of epileptiform (including ictal) or non-epileptiform abnormalities did show more negative-to-positive changes than the opposite for both categories, but the differences did not reach statistical significance (p=0.09 and p=0.14, respectively). Among “unlikely epilepsy” cases, more responses changed from positive to negative for the presence of any abnormality than the opposite direction, but the difference was not statistically significant (p=0.79) (
Figure 1).
Conclusions: We found that a clinical history suggestive of epilepsy presented with an EEG may bias experts toward calling the EEG abnormal. Based on our findings, we recommend that rEEGs be fully described before reviewing clinical data, to ensure reliable and unbiased reads. Future studies with more cases and/or raters may further delineate the impact of history bias in the interpretation of EEG.
Funding: None