Transient Loss of Consciousness
Virtual Special Editions are collections of targeted papers curated by a Guest Editor. Here Dr. Alistair Wardrope, Sheffield Teaching Hospitals NHS Foundation Trust, United Kingdom (author of the Editor’s Choice article in Volume 61) talks about the differential diagnosis of transient loss of consciousness.
The differential diagnosis of transient loss of consciousness
The differential diagnosis of transient loss of consciousness (TLOC) poses a challenge for specialist and generalist clinician alike. The articles in this Virtual Special Edition explore the nature of this challenge, some of the reasons it proves so persistent, and directions for future research.
TLOC is a common condition, with a lifetime prevalence of 50%; over 90% is due to syncope, epilepsy, or psychogenic non-epileptic seizures (PNES). However misdiagnosis is a common and persistent problem.
Maria Oto’s review explores the extent of misdiagnosis in TLOC and discusses some of its drivers.1 Oto also highlights the differing rates of misdiagnosis between generalist and specialist, a point borne out by Heather Angus-Leppan’s prospective study of the ultimate fates of patients referred to a specialist epilepsy service with a presumptive diagnosis of epilepsy. In this patient group, only 43% of the epilepsy diagnoses were ultimately confirmed by the assessing neurologist.2 Angus-Leppan’s paper is fascinating moreover for demonstrating how specialists arrived at these diagnoses. In the best Oslerian tradition, the patient and collateral history proves vital, with physical examination not changing the diagnosis in a single patient, and investigations leading to a change of diagnosis in just one.
Sundarajan et al.’s systematic review of the role of biomarkers in diagnosing PNES3 helps to explain the low yield of investigations in Angus-Leppan’s study. The general picture that emerges from their review is that the evidence base for such biomarkers is equivocal, conflicting, and at high risk of bias. Even tests that are highly sensitive and specific for a given diagnosis in ideal settings prove to be of limited use in practice, chiefly due to post-ictal time-sensitivity.
Confronted by these challenges, the intent of our systematic review was to identify candidate features of the initial clinical consultation that might lead practitioners toward a particular diagnosis in patients presenting with TLOC.4 Two main points emerge: more research exists on distinguishing epilepsy and PNES, despite the first question in most generalists’ minds being whether TLOC represents ‘fit or faint’; and no single historical feature identifies a given diagnosis with high sensitivity and specificity. We identified some candidate clinical decision rules (CDRs) that calculate likely diagnoses from combinations of these features; while potentially useful, no such tool has yet attempted the tripartite classification problem of epilepsy vs. syncope vs. PNES, or has been validated against gold-standard diagnostic criteria.
Having explored the challenge of diagnosis the cause of TLOC, the last three articles in this Virtual Special Edition address current best practice, and promising directions for future research. Nowacki and Jirsch walk the reader through the clinical assessment of a first TLOC presentation, drawing together many threads from the above articles.5 Robson et al.’s study on catastrophising or normalising descriptions of seizure experiences6 shows how we can enhance the diagnostic value of the history by attending not just to what the patient is saying, but how they say it. While such subtle conversational clues might seem the preserve of the specialist, Patterson et al.’s validation of a diagnostic scoring application7 highlights the potential of smart phones to revolutionise evidence-based practice for generalists.
Taking account of the limited computing capacity of clinicians employing CDRs in routine practice, most such measures have traditionally been limited to simple additive scores, but in an era where most carry powerful computers in their pockets, more complex algorithms are accessible. Developing such tools, however, depends on clinicians’ identification of the most salient features upon which to train them; all the articles in this Virtual Special Edition can contribute to this task.