Interobserver variation in the interpretation of SSEPs in anoxic–ischaemic coma

https://doi.org/10.1016/j.clinph.2006.03.018Get rights and content

Abstract

Objective

To study interobserver variation in the interpretation of median nerve SSEPs in patients with anoxic–ischaemic coma.

Methods

SSEPs of 56 consecutive patients with anoxic–ischaemic coma were interpreted independently by 5 experienced clinical neurophysiologists using guidelines derived from a pilot study. Interobserver agreement was expressed as kappa coefficients.

Results

Kappa ranged from 0.20 to 0.65 (mean 0.52, SD 0.14). Disagreement was related with noise level and failure to adhere strictly to the guidelines in 15 cases. The presence or absence of N13 and cortical peaks caused disagreement in 5 cases each. For recordings with a noise level of 0.25 μV or more, mean kappa was 0.34; for recordings with a noise level below 0.25 μV mean kappa was 0.74.

Conclusions

Interobserver agreement for SSEPs in anoxic–ischaemic coma was only moderate. Since the noise level strongly influenced interobserver variation, utmost attention should be given to its reduction. If an artefact level over 0.25 μV remains, absence of N20 cannot be judged with sufficient certainty and the SSEP should be repeated at a later stage.

Significance

Because of its moderate interobserver agreement, great care has to be given to accurate recording and interpretation of SSEPs before using the recordings for non-treatment decisions.

Introduction

Bilateral absence of the N20 cortical potential of median nerve somatosensory evoked potentials (SSEP) has been identified as the most valid predictor of poor outcome in patients who are unconscious after circulatory arrest (Robinson et al., 2003, Zandbergen et al., 1998). Its prognostic value has been studied more extensively than that of any other factor, and it appears less susceptible to sedatives and metabolic changes than the clinical neurological examination and the EEG (Coulthard and Rood, 1993, Koht et al., 1988, Sloan et al., 1990). Since absence of the N20 is often used as a basis for non-treatment decisions in these patients, it is important to know whether these SSEPs can be interpreted reliably.

In earlier studies of the prognostic value of SSEPs in anoxic–ischaemic and nontraumatic coma Madl et al. mentioned good reproducibility of the SSEPs, tested by repeated measurements (Madl et al., 1993, Madl et al., 1996). In another study two physicians reviewed recordings independently (Madl et al., 2000), but interobserver variation was not mentioned. We therefore, studied the interobserver variation in the interpretation of SSEP-recordings in patients with anoxic–ischaemic coma.

Section snippets

Study design

The study consisted of two phases: a pilot phase and the actual interobserver study. The pilot phase aimed to define guidelines for the interpretation of SSEPs in patients with anoxic–ischaemic coma. In this phase, a convenient sample of SSEPs from 20 patients with anoxic–ischaemic coma of 3 days' duration was collected. Inclusion criteria were: registration after median nerve stimulation on both sides; registration over Erb's point, the cervical spine and the cerebral cortex, and absence of

Pilot phase

Absolute agreement among all 5 observers for the primary question (bilateral absence of N20, or not) was found for 17 of the 20 SSEPs. A separate consideration of left- and right-sided tests yielded 40 recordings (20 from each side). Agreement among all 5 observers on the question ‘N20 absent or not?’ was reached in 32. The most important sources of disagreement among observers in the remaining 8 recordings were related to confusion concerning variant stimulation sites (in a few cases with

Discussion

The average interobserver agreement of the interpretation of SSEP-recordings was only moderate (K=0.52).

The main source of disagreement was related to the noise level. In spite of the guidelines agreed upon, some observers felt they could reliably conclude the N20 to be absent in some recordings with a considerable noise level, consequently increasing the interobserver variation for such recordings. In cases with low noise levels (noise <0.25 μV), the observer agreement was substantial (K=0.74).

References (11)

There are more references available in the full text version of this article.

Cited by (78)

  • SSEP amplitudes add information for prognostication in postanoxic coma

    2021, Resuscitation
    Citation Excerpt :

    To minimize muscle artifacts, neuromuscular blocking medication was administered if necessary. All SSEP recordings were assessed in the clinical setting by trained clinical neurophysiologists who scored, based on criteria of Zandbergen et al., whether the N20 was ‘bilaterally absent’.3 When peripheral responses at Erb's point or cervical level were absent, the recording was excluded from the current analysis.

  • Does a combination of ≥2 abnormal tests vs. the ERC-ESICM stepwise algorithm improve prediction of poor neurological outcome after cardiac arrest? A post-hoc analysis of the ProNeCA multicentre study

    2021, Resuscitation
    Citation Excerpt :

    On post-hoc assessment, the cause of these three false positive results was clearly identified as an incorrect reading of the SSEP record due to excessive noise. This is in line with results of other studies showing that the risk of SSEP misclassification increases when signal/noise ratio is low.31–34 As far as bilateral absence of PLR is concerned, two multicentre studies 5,6 and a large single-centre study35 published after the 2015 guidelines reported a 6–7% FPR at 72 h or later for this sign, which is in line with the rate observed in our study.

View all citing articles on Scopus
View full text