Syncope or seizure? The diagnostic value of the EEG and hyperventilation test in transient loss of consciousness.

In a prospective study of consecutive patients (age 15 or over) with transient loss of consciousness 45 patients had a history of seizure and 74 patients had a history of syncope. All patients had an EEG, ECG, laboratory tests and a hyperventilation test and were followed for an average of 14.5 months. Epileptiform activity in the interictal EEG had a sensitivity of 0.40 and a specificity of 0.95 for the diagnosis of a seizure. Epileptiform activity nearly doubled the probability of a seizure in doubtful cases. If no epileptiform activity was found, this probability remained substantially the same. The hyperventilation test had a sensitivity of 0.57 and a specificity of 0.84 for the diagnosis of syncope. A positive test increased the probability of syncope half as much in doubtful cases. A negative test did not exclude syncope. Laboratory tests were not helpful except for an ECG which was helpful in elderly patients.

Accepted 1 February 1991 Syncope and related conditions where cerebral blood flow is impaired are usually distinguished from seizures on the history alone. The wide variety of epileptic manifestations and the clonic jerks that may be seen in fainting patients cause problems.' 2 When the diagnosis remains in doubt, other diagnostic tests are needed.
The finding of epileptiform activityspikes, spike-waves, sharp waves-in the EEG must reinforce the diagnosis of a seizure,3-5 though the extent remains uncertain, as some patients with this finding never have seizures. 6 Anxiety or hyperventilation attacks are mistaken for seizures.79 Many of the so-called pseudoseizures are actually hyperventilation attacks.'0 When associated with fainting the differentiation from epilepsy is particularly difficult." A hyperventilation test provoking characteristic symptoms (such as paresthesia, giddiness and dyspnoea) is used for the diagnosis. Does this test really help to distinguish seizure from (hyperventilation) syncope? Cardiac syncope may be mistaken for a seizure.'2 '3 An ECG cannot exclude cardiac syncope, but may suggest a cardiac disorder.
We carried out a prospective study of patients with transient loss of consciousness to assess the diagnostic value of a single interictal EEG, of the hyperventilation test, of a standard ECG and of the routine laboratory examination.

Patients
From March 1987 to March 1988 we included all consecutive patients (> 15 years of age) referred to the neurological department because of one or more episodes of transient loss of consciousness. Transient loss of consciousness was defined as an episode of less than one hour with inability to maintain posture and to recall events during the episode. We excluded patients with loss of consciousness due to trauma or subarachnoid haemorrhage and patients with epilepsy. We studied 119 patients. General practitioners referred 55 (46%) and other physicians 14 patients to our outpatient department. We interviewed in the emergency department 28 and on admission 22 patients. We followed all until December 1988. The mean period of follow up was 14-5 months (range: [8][9][10][11][12][13][14][15][16][17][18][19][20][21].
Diagnostic criteria A gold standard for the diagnosis of a seizure does not exist. Long-term EEG and video monitoring are unrealistic in patients with a single or rare event.'4 The international classification of seizures has no explicit criteria to distinguish seizures from syncope.'5 We classified a patient in the seizure group as follows: if an eyewitness observed more than a few movements during loss of consciousness and identified clonic movements from a range of movements imitated by the interviewer; if an eyewitness observed automatisms, such as chewing or lip smacking, during loss of consciousness; if the patient reported an unequivocal aura, such as a strange smell, preceding the event; if the patient felt confused immediately after the event (inability to recognise familiar persons or environment); if the patient had tongue biting. We classified all other patients in the syncope group. Patients younger than 65 years of age had an additional hyperventilation test.'7 After an initial spiro-and capnogram patients were asked to ventilate, in the sitting position, at a rate of 40 breaths per minute for at least three minutes. The end tidal CO2 percentage was monitored and had to be lower than 2-5% after hyperventilation. Blood gases were measured by fingerprick before and after the test and after the recovery phase. A pulmonary physiologist graded the test, using a predefined checklist'8 (table 1). If patients did not reach 90% of the baseline value of the end tidal CO2 percentage after three minutes recovery, the test was considered positive if two or more symptoms were provoked and were recognised by the patient as (part of) his complaint and as dubious if only one symptom was provoked or symptoms were not recognised. The test was considered negative if the end tidal CO2 restored to >90% of baseline value after three minutes recovery. A final diagnosis of hyperventilation syndrome was made if the patient responded to training in abdominal breathing and relaxation and no other cause was found during follow up.
A cardiologist assessed the standard ECG with computerised measurement of the QTinterval of all patients as normal or as abnormal.
The laboratory examination included serum sodium, potassium, calcium, phosphate, glucose, urea, sedimentation rate, liver func-tions, blood count. The attending physician decided if cerebral-CT scan or 24 hour cardiac monitoring were necessary. All patients gave informed consent to participate. The ethical committee of the hospital approved the study.

Statistical methods
The sensitivity is the true-positive rate and the specificity is the true-negative rate of a test. The positive likelihood ratio is the ratio of the probability that a test is positive in diseased persons to the probability that a test is positive in non-diseased persons (sensitivity/ 1-specificity). The negative likelihood ratio is the ratio of the probability that a test is negative in diseased persons to the probability that a test is negative in non-diseased persons (1-sensitivity/specificity EEG One patient refused an EEG after receiving a pacemaker. Table 3 presents the results of the EEG in the remaining 118 patients. No patient was receiving antiepileptic drugs before the EEG was taken. If the pre-test probability of a seizure is 50% and epileptiform activity is found the post-test probability of a seizure becomes 88%. If no epileptiform activity is found the post-test probability of a seizure in this case becomes 38%. The predictive value of the EEG is shown in the figure for different pretest probabilities. Epileptiform activity was found in 12 out of 23 patients with recurrent seizures and in six out of 22 patients with a  presumed, but not confirmed as loss of consciousness did not recur. No case of prolonged QT-interval syndrome was found.
Laboratory examination Apart from abnormal liver functions in all four patients with alcohol related seizures, none of the other abnormalities found had diagnostic or therapeutic consequences.

Discussion
Epileptiform activity in the EEG is specific, but not sensitive for the diagnosis ofa seizure as the cause of transient loss of consciousness. It follows that an interictal EEG can be used to confirm, but not to dismiss the clinical diagnosis of a seizure. Unfortunately, the EEG cannot predict recurrence of seizures, which confirms the study of Hopkins et al in a much greater group of patients. 22 We do not agree with their conclusion that the EEG is not necessary, because of the often weak nature of a clinical diagnosis of a seizure and the value of the EEG to diminish diagnostic uncertainty. This is important because if the first event is considered a seizure the next recurrence will result in a diagnosis of epilepsy. A hyperventilation test should only be considered as positive if the patient recognises more than one of the symptoms that preceded or followed loss of consciousness. A negative test does not exclude (hyperventilation) syncope. We cannot recommend the hyperventilation test as a routine diagnostic procedure because the diagnostic gain is marginally better than questioning the patient about episodes of breathlessness. The patient may, nevertheless, be reassured by the test.23 Routine laboratory examination is not necessary; an ECG is only indicated in elderly patients.
Some aspects of this study deserve further comment. First, the study has been conducted in an academic neurological setting, which biased the selection of patients. In a similar study from an emergency room, however, a seizure was diagnosed in 58 out of 198 patients (29%).24 The difference with the proportion of seizures (38%) in our study is not significant (chi square: P > 0 1).
Second, there is doubt about the final classification of patients. In four patients of the syncope group (6%, CI from 2% to 13%25)  Negative likelihood ratio (CI) 0-7 (0-6-0-9) CI = 95% confidence interval epileptiform activity was found. This activity consisted of sharp waves meeting most of the described epileptiform criteria.26 Their history did not suggest epilepsy and follow up seemed long enough to exclude epilepsy in these patients. The confidence interval shows the range of this false-positive percentage. If in reality the percentage is 2%, it is the same as the percentage found by Zivin and Ajmone Marsan.6 We believe that 6% or an even higher percentage is closer to reality as their study was retrospective and carried the risk of "circular reasoning": the finding of epileptiform activity may have been used to diagnose epilepsy in patients with obscure symptoms. Our study confirms that epileptiform activity in the EEG is not synonymous with epilepsy.27 The falsenegative rate of 60% (CI from 44% to 74%) of patients in the seizure group is comparable with the false-negative rate of 45% (CI from 40% to 50%) in another study. 2 Third, an important limitation of our study is the relatively small number of patients. Nevertheless, the specificity of the EEG in our study is the same as in the larger study of Goodin and Aminoff, who reported a higher sensitivity of 0-52.5 Their retrospective study carried the same risk mentioned above of circular reasoning which may lead to a higher true-positive rate.
The starting point of a diagnosis of epilepsy is often the evaluation of transient loss of consciousness. If the clinical diagnosis of a seizure is in doubt, the interictal EEG is a very useful diagnostic tool. Our findings apply to normal clinical practice as the spectrum of transient loss of consciousness is well represented. 29 We thank Dr PJ Sterk (Department of Chest Medicine) and the staffof the department of cardiology for their cooperation in this study.