Secondary Logo

Journal Logo

Original Article

Evoked potentials in the ICU

Amantini, A.*; Amadori, A.; Fossi, S.*

Author Information
European Journal of Anaesthesiology: February 2008 - Volume 25 - Issue - p 196-202
doi: 10.1017/S0265021507003183
  • Free


Definition and classification

The evoked potential (EP) is defined as a variation of electrical potential in the sensory organ, along the peripheral nerve and intracerebrally after a transient stimulus quantifiable in its physical parameters (intensity, duration and frequency). There are different EPs for the various sensory stimuli: somatosensory, auditory and visual. Generally EP amplitude is less than electroencephalographic (EEG) spontaneous activity and averaging techniques are necessary to extract such signals from the background biological and environmental noise and make them identifiable. Their waveform consists of several subsequent components with positive-negative polarity. The amount of averaging needed to obtain an identifiable, repeatable and measurable EP depends on the signal-to-noise ratio.

EPs can be classified as exogenous or obligatory when dependent mainly on the physical features of the stimuli and as endogenous or event-related when dependent on patient elaboration of the stimuli (categorization of the stimuli, discrimination of different stimuli, memory, decision making). The former originate mainly in specific primary and peri-primary cortical areas, the latter from poly-modal fronto-parietal associative areas [1].

EP classification based on latency is extremely useful since it defines the components according to a different anatomic topography (peripheral, subcortical and intracortical on primary and secondary areas) and to a different sensitivity to sedatives and anaesthetics (very low for short-latency and highest for long-latency-EPs).

Generally, EPs are differentiated by short-, middle- and long-latency components (within 10 s, 10-50 ms and longer than 100 ms), but such temporal limits can be slightly different according to the sensory mode.

Rational use in the ICU

International Federation of Clinical Neurophysiology Guidelines [2] define the standards for the use of EEG and EP in ICU. EEG and EP allow for a functional assessment of the central nervous system (CNS) and their aim is to support clinical examination and complement neuroimaging (computed tomography (CT) and magnetic resonance scans). If, on the one hand, a neurophysiological disturbance can occur in the absence of an anatomic lesion, conversely an anatomic lesion can occur outside the field of neurophysiological investigation. A more recent proposal of guidelines for the use of clinical neurophysiology in the ICU [3] states that the different tests (EEG, short- and long-latency EP, EMG) should not be used within a standard set but as a separate choice based on clinical evidence. Of the neurophysiological tests available in the ICU, we will discuss EPs and their use as diagnostic, prognostic and monitoring tools. The early components of EPs are used in the acute phase of cerebral damage when the patient, as a result of sedatives, neuromuscular blockade or the severity of coma, is difficult to assess on a clinical level. Indeed, short-latency EPs are largely resistant to sedation, which actually facilitates their recording by eliminating muscular and movement components. Long-latency and cognitive components of EPs are more applicable to a post-acute phase, since they are extremely sensitive to sedation and require clinical optimization to be recorded properly (remission of any pharmacological effects, resolution of metabolic and septic effects).

The most commonly used short-latency EPs in the ICU are somatosensory evoked potentials (SEPs) and brainstem auditory evoked potentials (BAEPs). The former have the advantage of having peripheral, spinal, brainstem and intracortical components, which are easily identifiable in all subjects and thus explore an extended CNS pathway. The latter, on the other hand, explore a limited tract of auditory transmission through the medulla and pons and the identification of the peripheral component (wave) is often difficult in the ICU. Furthermore, auditory EPs are difficult to identify in primary cortical areas; indeed the generators of such components lie in the depths of the Sylvian fissure (areas 41 and 42) and thus are difficult to record with surface scalp electrodes, unlike early cortical SEPs whose generators lie in the convexity of the post-central gyrus (particularly, area 2b). Furthermore, auditory myogenic reflexes, which are not eliminated with averaging, have a latency similar to early cortical acoustic components and hence can mask them. Taking problems such as these into account, the type of EP, which has been the most widely used in the ICU over the last few years and for which there is greater evidence regarding its clinical utility, is the short-latency SEP after median nerve stimulation.

Short-latency SEPs

Short-latency SEPs are sensitive both to traumatic structural damage and to hypoxic-ischaemic damage and are able to indicate a lesional topography. But, more than this, in the absence of a relevant lesion along afferent sensory pathways, a ‘global' index of brain function (reflected by brainstem, thalamo-cortical and intracortical transmission in both hemispheres) can be extrapolated from them. This latter information is very important in ICU patients who can be assessed only with difficulty because of sedation, neuromuscular blockade or the severity of coma. Furthermore, SEPs have the advantage of being resistant to sedatives and have a waveform, which is easily interpretable and comparable in subsequent recordings.

Since we believe that one of the reasons for the under-use of SEP in ICU is due to the technical difficulties of recording in such an environment and because the results are not perceived as very reliable, it is important to make some technical specifications regarding the recordings. Firstly, both the technician and the neurophysiologist need to be experts in EP recording in neurological diagnostic laboratories and also have specific ICU training. A correct approach to the neurophysiological test is vital. In addition to careful history-taking (coma aetiology, Glasgow Coma Score (GCS), neuroimaging, sedation) the technician must be aware of possible technical problems with the recording, such as obstacles to the stimulation at the conventional site, excessive artefacts due to the environment or patient and the need for sedation or neuromuscular blockade for a good recording.

Either subdermal needle electrodes or silver-silver chloride cup electrodes, possibly arranged in coloured plaits in order to reduce artefacts, are used.

The recording system should have at least two channels (we suggest four). For technical details on amplifiers, averaging and electrical safety, the reader is referred to appropriate guidelines [4].

It is important to set the recording system with the following parameters:

  • stimuli: monophasic rectangular electrical pulses of length 100-300 μs;
  • stimulus frequency: 3-5 Hz; time analysis: 50-100 ms; (with a lower frequency stimulation and a greater time analysis intracortical components with intermediate latency following N20-P25 are more easily identifiable);
  • filters: LF 2-5 Hz, HF 1.5-3.0 kHZ;
  • averaging: 200-400 pulses according to the recording conditions (signal/noise ratio) and to the identifiable components. Two series of reproducible responses must be obtained. A greater number of replicable series is recommended in the case of particularly clinically relevant pathological patterns (absence of cortical SEP - in such case a time analysis of 100 ms should be used to confirm the absence of any cortical response).

The intensity of the stimuli must be above the motor threshold (muscular twitch). If the patient undergoes treatment with neuromuscular blocking agents, it is necessary to verify that the stimuli are supra-maximal according to the amplitude on Erb's point, cervical or cortical. For SEP recording after median nerve stimulation, needles are located on Erb's point (referred to the contralateral Erb's point), spinous process C7 (referred to ante-collis) and C3′-C4′ (referred to Fz and ipsilateral mastoid); Fig. 1a. An extra-cephalic reference can be used, but it is exposed to a major number of artefacts. We also suggest using Fz as reference, because, even if it produces a greater inter-individual amplitude variability, it avoids the influence of subcortical potentials and reduces the noise level. Furthermore, it allows for the recognition of cortical responses even if their presence is doubtful compared with using only a mastoid reference.

Figure 1.
Figure 1.:
Example of somatosensory evoked potential (SEP) classification: (a) normal SEP (N) after median nerve stimulation; (b) pathological SEP (P) for central conduction time (CCT) increase; (c) absent SEP (A) considering both Fz and M reference; (d) and (e) pathological amplitude asymmetry (>50%) of cortical SEP between the two hemispheres (P) with normal CCT (calibration, time base, montages and peak identifications as in (a)) [5].

Finally, it is important to establish normal criteria (Fig. 1):

  • absence of obligatory components (N9 in a derivation ipsilateral to Erb's point, N20 in a derivation with Fz reference, P14 in a derivation with an auricular or mastoid reference, N13 in a derivation with AC reference) (Fig. 1c). This criterion means that adequate technical conditions have to prevail to identify the components if they are present. Sometimes, the non-repeatability of some components (particularly spinal and subcortical) when many artefacts are present is a technical limitation rather than a patient abnormality;
  • prolongation of interpeak latency (N9-N13, N13-N20 and P14-N20). Generally, we refer to the limits of 2.5-3.0 SD about the mean values for each laboratory (Fig. 1b). However, we must keep in mind the problem due to the lack of normal values in ICU patients. Furthermore, several non-pathological factors can interfere with interpeak latencies (temperature, therapy, etc.). In this context, it is conceivable to use the limits of 3 SD, we considered pathological a central conduction time (CCT), N13-N20, >7.8 ms.
  • amplitudes and asymmetries. In the absence of a latency modification, an amplitude modification should be carefully evaluated, considering the high variability in the normal subject both for absolute values and for asymmetry between the two sides (Fig. 1d,e). For N20/P25 absolute amplitude (recorded with Fz reference) we used percentiles to determine the lowest limit of normality (5th percentile = 1.2 μV). Concerning asymmetry, we considered pathological an N20 amplitude reduction of more than 50% compared to contralateral.

EP can be used both for diagnostic/prognostic purposes and for the monitoring of the evolution of cerebral damage or the appearance of secondary damage. For the former, single recordings are enough, while for the latter, for greater clinical impact, continuous or close serial neurophysiological evaluation is necessary.

EP for diagnostic and prognostic purpose


The most useful neurophysiological test in the ICU from a diagnostic point of view is EEG (aetiological diagnosis when doubtful, non-convulsive seizures (NCS) and status epilepticus, differential diagnosis among motor manifestations, real conscious state beyond clinical evidence - locked-in syndrome, frontal syndromes, severe neuromyopathy in critically ill patient, akinetic mutism, etc. [6,7]).

Furthermore, EEG in Italy is a compulsory test to confirm the diagnosis of brain death. However, we must remember that short-latency EPs, SEP in particular, are very useful for the diagnosis of brain death in the presence of ‘confounding' factors, such as sedation, that would invalidate clinical and EEG criteria alone. SEP modification carefully reflects the rostro-caudal deterioration of brain function towards brain death, first with the loss of intracortical components and later that of the lemniscal caudal component (P14) that reflects medulla generator activity. The loss of all intracranial components (N20 and P14) (when previously recorded), associated with the preservation of extra-cranial components (N13, N10), is the SEP pattern of brain death [8].

As indicated by the recently proposed European Guidelines [3], short-latency EPs increase the timeliness, safety and reliability of brain death diagnosis, particularly when it cannot be performed with conventional criteria.


In this section, we will discuss the usefulness of short-latency EPs in the prognosis of most kinds of coma, particularly post-hypoxic-ischaemic and post-traumatic. It is useful to briefly remember the different epidemiology of the two types of coma. The former usually has an unfavourable prognosis regarding recovery of consciousness since 70-80% of the patients die or evolve towards a vegetative state. In the latter, on the other hand, about 60% of the patients recover consciousness, while about 30% die (usually in an early stage). Therefore, in post-hypoxic-ischaemic coma, the early prognosis of an unfavourable outcome is more important, while in post-traumatic coma, in addition to the diagnosis of recovery of consciousness, it is very useful to diagnose residual disability. SEPs, even if imperfectly recorded, can provide this prognostic information. We must also add that GCS and EEG are more useful for the prognosis of hypoxia-ischaemia rather than for traumatic coma.

Regarding an early prognosis of post-hypoxic-ischaemic and post-traumatic coma, we must remember at least six meta-analyses that have provided evidence proving the utility of short-latency EP and SEPs in particular [9-14].

In post-hypoxic-ischaemic coma, Bassetti and colleagues [15] performed a multivariate analysis on the prediction of unfavourable outcome, taking into account four prognostic factors (GCS < 5, absence of more than one brainstem reflex, severe EEG abnormalities and bilateral absence of SEP) with a positive predictive value of unfavourable outcome of 100% after 48 h. The association of clinical criteria to SEP led to the identification of 75% of the patients, while the further association of EEG led to a slight increase in the patients identified (77%). Furthermore, the most frequently observed negative pattern was the bilateral absence of SEP (50%), followed by GCS < 5 (40%), severe EEG abnormalities (40%) and absence of more than one brainstem reflex (6%).

Zandbergen and colleagues [16] have suggested clinical ‘guidelines' for the diagnosis of unfavourable outcome in post-hypoxic-ischaemic coma, postponing to 72 h the prognostic assessment and limiting SEP recording to those patients with an absent photo-motor reflex or motor GCS score 1-3. Recent proposals for European Guidelines on the use of neurophysiological tests in ICU involve assessment at 24 h [3].

A recent report from the quality standard subcommittee of The American Academy of Neurology [17] indicates SEPs as the only instrumental test, associated with clinical examination, for accurately predicting poor outcome in comatose patients after cardiac arrest.

Taking indications like these on board, we have adopted, in our hospital, an EEG-SEP protocol, studied using the cardiac and general ICUs, for the early prognosis of post-hypoxic-ischaemic coma. Such protocols involve an EEG-SEP recording after 24 h in patients with motor GCS score ≤3, and in the case of an unfavourable pattern for the recovery of consciousness (bilateral absence of SEP, severe EEG abnormalities), considering the clinical relevance of such a pattern, we confirm the result after 48-72 h.

The clinical relevance, in the case of an unfavourable pattern, is to discourage aggressive intervention in the haemodynamics of severely unstable patients and to inform relatives of the outcome early, directing such patients towards low intensity rehabilitation recovery once discharged from ICU.

Robinson and colleagues' [13] meta-analysis not only confirms the prognostic certainty and safety of SEPs for predicting an unfavourable outcome in post-hypoxic-ischaemic coma (100% of non-consciousness recovery in the case of bilateral absence of SEP) and a similarly unfavourable prognostic value of such a pattern in post-traumatic coma (90-95% of ‘unawakening' in the case of bilateral absence of SEP) but also indicates the favourable prognostic value of a normal SEP in post-traumatic coma (more than 90% of consciousness recovery in case of bilateral normal SEP) unlike that of post-hypoxic-ischaemic coma (50-60% of consciousness recovery).

Logi and colleagues [18] point out that in post-hypoxic-ischaemic coma not only is the absence of cortical responses associated with an unfavourable prognosis for the recovery of consciousness but also report an SEP amplitude value, under which no patient has recovered consciousness (1.2 μV of peak-to-peak N20 amplitude and 0.6 μV for onset-peak). Such an approach is more applicable to post-hypoxic-ischaemic damage, which results in neural cortical depletion, rather than to post-traumatic coma. In the latter, since there is no selective vulnerability but rather several pathogenetic mechanisms, it is preferable to look for more predictive patterns rather than indicating absolute values of amplitude and latency (CCT) associated with prognosis [19]. For example, classifying SEPs on both hemispheres as in Figure 1, it is possible to aggregate the modifications in three categories with a different prognostic meaning: Grade 1 (NN, NP: Normal) with a favourable prognosis, Grade 3 (AA: Absent) with an unfavourable prognosis and Grade 2 (NA, PP, AA: Preserved) with a doubtful prognosis. It should be noted that Grades 1 and 3 allow for the correct classification of about 65-70% of severe post-traumatic comatose patients. Furthermore, Grade 1 in our study has a positive predictive value of ‘awakening' of more than 90% and of good functional recovery according to the Glasgow Outcome Scale (GOS) of more than 80% [5]. SEPs allow for a correct prediction of outcome in almost 70% of severely injured trauma patients, predicting not only ‘awakening' but also good recovery and severe disability, thus helping to identify the right rehabilitation on discharge from ICU.

Finally, a recent meta-analysis set out to assess the best early indicator of prognosis out of SEP, CT scan, EEG, GCS, motor responses and photo-motor reflexes in post-traumatic and post-hypoxic-ischaemic coma [14]. The study selected 26 comparable papers including more than 800 patients and concluded that SEP is the best single prognostic indicator. According to such evidence, SEPs should always be associated with clinical examination for early prognosis in acute cerebral damage.

Continuous neurophysiological monitoring with EP

The aim of continuous neurophysiological monitoring is to help to evaluate the evolution of primary cerebral damage and to identify secondary damage in time.

There are various reasons as to why neurophysiological monitoring needs to be more widely used: the increased neurological interest and involvement in neurointensive care units; the widespread application of neurophysiological techniques in intraoperative monitoring; technological innovation and the advantages of digital EEG (analysis of the signals and elaboration of quantitative trends, facility of storage and net transmission). As far as the usefulness of continuous EEG in the ICU is concerned (conventional and quantitative EEG), there is enough evidence for the diagnosis and treatment of NCS and non-convulsive status epilepticus (NCSE) and pilot studies for the diagnosis of acute ischaemia, but there are very little data regarding its utility for the prognosis and no data for the monitoring of the evolution of acute cerebral damage [20]. Despite the large use of SEPs in intraoperative monitoring and the proven prognostic utility of SEP in acute cerebral damage (close relationship between SEP modification in the early stage of coma and cerebral damage that then determines the prognosis), there are, in the literature, very few studies on continuous SEP monitoring in the ICU [21,22].

This is surprising, especially if we consider that for intracranial pressure, the most widespread monitoring device for acute cerebral damage, there is insufficient evidence of a close causal relationship between ICP increase and cerebral damage, at least for a certain range of values [23]. In addition to the clinical rationale, the ‘technical' rationale for continuous SEP monitoring in the ICU is due to the possibility of associating the high variability of the EEG with the stability of SEPs, with the added advantage of their easy display of latency and amplitude trends, which are easily interpretable, at first glance, by non-trained personnel. All this would facilitate early clinical identification of patterns, which indicate a high risk of neurological acute or progressive deterioration.

Considering such a consistent rationale, we decided to perform a pilot study of continuous EEG-SEP monitoring in comatose patients in the neurointensive care unit [24]. We included patients with head trauma and intracerebral haemorrhage with GCS < 9 undergoing ICP monitoring. The recording protocol specified alternating cycles of EEG and SEPs after stimulation of right and left median nerves, thus monitoring both hemispheres (amplitude and latency trends with settable alarms). In more than 50 patients continuously monitored for an average of 8 days, we observed that in all the patients who were stable from a clinical point of view (GCS and serial CT scan), SEPs never showed significant modification in either latency or amplitude except those due to sedation. On the contrary, whenever a neurological deterioration occurred (23%), SEP always showed significant latency and amplitude modifications.

We also assessed the correlation between SEP modification and the course of ICP. We noticed that SEP modifications can precede or follow ICP increases, probably according to a different pathogenetic mechanism of neurological deterioration (earlier SEP modifications in the case of ischaemia, later in the case of increasing brain oedema without hypoxic damage). Furthermore, SEP trends (stable or deteriorated) had a greater predictive value than the course of ICP course (calculated as mean maximum daily values) for outcome in the acute stage.

These preliminary results lead us to believe that neurophysiological monitoring is an ideal complement to the other parameters monitored in NICU while ICP is a mere pressure index (from which we extrapolate a haemodynamic clue as to cerebral perfusion), SEP monitoring fundamentally reflects neuronal capacity to extract oxygen and thus to what extent cerebral parenchyma remains metabolically active during acute cerebral damage. Neurophysiological monitoring can help in the management of ICP trend and its modifications and contribute to optimizing the choice and timing of clinical strategies for the individual patient.


1. Regan D. Human brain electrophysiology: evoked potentials and evoked magnetic fields. In: Science and Medicine. New York: University of Toronto, Elsevier, 1989.
2. Guérit JM. Medical technology assessment EEG and evoked potentials in the intensive care unit. Neurophysiol Clin 1999; 29: 301-317.
3. Guèrit JM, Amantini A, Amodio P et al. Proposal of Guidelines on the Use of Neurophysiological Tests in the Intensive Care Unit (ICU). XXVII International Congress of Clinical Neurophysiology, Stockholm, 8-12 May 2005.
4. Kraft GH, Litchy WJ, Aminoff MJ, Baran EM, Eisen AA, Stolov WC. American Association of Electrodiagnostic Medicine (AAEM) somatosensory evoked potentials subcommittee. Guidelines for somatosensory evoked potentials.. Muscle Nerve Suppl 1999; 8: S123-S138.
5. Amantini A, Grippo A, Fossi S et al.. Prediction of “awakening” and outcome in prolonged acute coma from severe traumatic head injury: evidence for validity of short latency SEPs. Clin Neurophysiol 2004; 116: 229-235.
6. Guerit JM, Fischer C, Facco E et al.. Standards of clinical practice of EEG and EPs in comatose and other unresponsive states. The International Federation of Clinical Neurophysiology. Electroencephalogr Clin Neurophysiol Suppl 1999; 52: 117-131.
7. Fischer C, Mutschler V. Traumatic brain injuries in adults: from coma to wakefulness. Neurophysiological data. Ann Readapt Med Phys 2002; 45: 448-455.
8. Facco E, Munari M, Gallo F et al.. Role of short latency evoked potentials in the diagnosis of brain death. Clin Neurophysiol 2002; 113: 1855-1866.
9. Zandbergen EG, de Haan RJ, Stoutenbeek CP, Koelman JH, Hijdra A. Systematic review of early prediction of poor outcome in anoxic-ischaemic coma. Lancet 1998; 352: 1808-1812.
10. Attia J, Cook DJ. Prognosis in anoxic and traumatic coma. Crit Care Clin 1998; 14: 497-511.
11. Rothstein TL. The role of evoked potentials in anoxic-ischemic coma and severe brain trauma. J Clin Neurophysiol 2000; 17: 486-497.
12. Carter BG, Butt W. Review of the use of somatosensory evoked potentials in the prediction of outcome after severe brain injury. Crit Care Med 2001; 29: 178-186.
13. Robinson LR, Micklesen PJ, Tirschwell DL, Lew HL. Predictive value of somatosensory evoked potentials for awakening from coma. Crit Care Med 2003; 31: 960-967.
14. Carter BG, Butt W. Are somatosensory evoked potentials the best predictor of outcome after severe brain injury: a systematic review. Intensive Care Med 2005; 31: 765-775.
15. Bassetti C, Bomio F, Mathis J, Hess CW. Early prognosis in coma after cardiac arrest: a prospective clinical, electrophysiological, and biochemical study of 60 patients. J Neurol Neurosurg Psychiatry 1996.; 61: 610-615.
16. Zandbergen EG, de Haan RJ, Koelman JH, Hijdra A. Prediction of poor outcome in anoxic-ischemic coma. J Clin Neurophysiol 2000; 17: 498-501.
17. Wijdicks EF, Hijdra A, Young GB, Bassetti CL, Wiebe S. Quality Standards Subcommittee of the American Academy of Neurology. Practice parameter: prediction of outcome in comatose survivors after cardiopulmonary resuscitation (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2006; 67: 203-210.
18. Logi F, Fischer C, Murri L, Mauguiere F. The prognostic value of evoked responses from primary somatosensory and auditory cortex in comatose patients. Clin Neurophysiol 2003; 114: 1615-1627.
19. Facco E, Munari M. The role of evoked potentials in severe head injury. Intensive Care Med 2000; 26: 998-1005.
20. Claassen J, Mayer SA, Hirsch LJ. Continuous EEG monitoring in patients with subarachnoid hemorrhage. J Clin Neurophysiol 2005; 22: 92-98.
21. Konasiewicz SJ, Moulton RJ, Shedden PM. Somatosensory evoked potentials and intracranial pressure in severe head injury. Can J Neurol Sci 1994; 21: 219-226.
22. Moulton RJ, Brown JI, Konasiewicz SJ. Monitoring severe head injury: a comparison of EEG and somatosensory evoked potentials. Can J Neurol Sci 1998; 25: S7-S11.
23. Forsyth R, Baxter P, Elliot T. Routine intracranial pressure monitoring in acute coma. Cochrane Database Syst Rev 2001; 3: CD002043.
24. Fossi S, Amantini A, Grippo A et al.. Continuous EEG-SEP monitoring of severely brain injured patients in NICU: methods and feasibility. Neurophysiol Clin 2006; 36: 195-205.


© 2008 European Society of Anaesthesiology