By Louise M. Klebanoff, MD

Assistant Professor of Clinical Neurology, Weill Cornell Medical College

Dr. Klebanoff reports no financial relationships relevant to this field of study.

SYNOPSIS: Investigators found that adding witness-reported observations to patient demographics and patient-reported symptoms improved the diagnostic accuracy between epilepsy, syncope, and psychogenic nonepileptic seizures.

SOURCE: Chen M, Jamnadas-Khoda J, Broadhurst M, et al. Value of witness observations in the differential diagnosis of transient loss of consciousness. Neurology 2019;92:e895-e904.

The differential diagnosis of transient loss of consciousness (TLOC) includes epilepsy, syncope, and psychogenic nonepileptic seizures (PNES). The gold standard for confirming the diagnosis is simultaneously recording clinical events and physiological measures. In practice, this standard is reached rarely, and the diagnosis is made based on the patient’s history and witnessed descriptions. Misdiagnosis rates of 25% have been reported. Chen et al performed a retrospective study evaluating the contribution of additional witness observations in determining the etiology of transient loss of consciousness. Previously, the authors demonstrated that self-reportable associated symptoms collected using the Paroxysmal Event Profile (PEP), an 86-item symptom questionnaire, made an important diagnostic contribution in distinguishing between epilepsy, syncope, and PNES in laboratory-proven cases. When PEP data were added to basic patient information, 66% of patients with epilepsy, 91% with syncope, and 78% with PNES were classified correctly. The authors investigated to what extent a 31-item profile (Paroxysmal Event Observer [PEO]) of observer-reportable event manifestations improved the diagnostic differential.

The authors reviewed 249 patients from a total sample of 300 patients from three British medical centers who had completed both the PEP and PEO questionnaires (86 with epilepsy, 84 with syncope, and 79 with PNES). All diagnoses were confirmed by recordings of typical events with video electroencephalogram (EEG), ambulatory EEG, or tilt-table. Of 31 items collected, 24 differed significantly between the three groups. These factors were combined into four broader categories: unconsciousness, reduced self-control, excessive movement, and skin/face/recovery.

Observer-reported factors differentiated syncope and epilepsy better than patient-reported factors (accuracy: 96% vs. 85%; C-index P = 0.0004). When the analysis of the patient information was combined with that from observers, this distinction rose from 90% to 100% (C-index P = 0.005). In the differentiation of PNES and epilepsy, additional observer-reported factors improved the predictive accuracy from 76% to 83% (C-index P = 0.006) and from 93% to 95% (C-index P = 0.098) in PNES and syncope. When analyzed in isolation, more patients were classified correctly by the observer-derived data than by the patient-provided symptoms.

In this study, witness questionnaire responses correctly classified patients with a sensitivity of 84.4% and a specificity of 84.2%. The answers to the PEO factor “reduced self-control” distinguished most clearly among the three diagnostic groups, with low levels of TLOC-associated self-control more frequently observed in epilepsy than in syncope or PNES. Other distinguishing factors included that, unlike syncope and PNES, epileptic seizures “never” looked like normal sleep, and pale skin and limp collapse were more commonly seen in syncope than in epilepsy. PNES observers more commonly replied that they could “never” do something to make the attack pass more quickly. This study provides confirmatory data that witness-provided information contributes to the correct differentiation between epilepsy, syncope, and PNES. The witness-provided data were most useful in correctly distinguishing epilepsy from syncope or PNES. The differentiation of syncope from PNES was highly accurate without the addition of observation data. The poorest differentiation was between epilepsy and PNES. Although observer-provided data improved diagnostic accuracy, additional data (such as video EEG monitoring) likely will be required to optimize this differential.


This study provides support for the importance of witness observations in distinguishing common causes of transient loss of consciousness. By adding witness-reported observations to patient demographics and patient-reported symptoms, the diagnostic accuracy between epilepsy, syncope, and PNES improves. Although differentiating between epilepsy and PNES still may require additional data (such as a spell captured on video EEG), observer-reported data can improve the diagnostic accuracy of syncope and epilepsy significantly. Structured witness interviews that include observations of reduced self-control, sleep-like appearance, skin pallor, and/or limp limbs provide the most important distinguishing factors.