Secondary Logo

Journal Logo

Original Article

Postoperative changes in the full-field electroretinogram following sevoflurane anaesthesia

Iohom, G.*; Whyte, A.; Flynn, T.; O'Connor, G.; Shorten, G.*

Author Information
European Journal of Anaesthesiology: April 2004 - Volume 21 - Issue 4 - p 272-278


Ambulatory surgery comprises 70% of all surgical procedures performed in the USA and up to 65% in Europe [1]. This proportion is likely to increase substantially during the next decade. If based solely on economic factors, this change in practice may be associated with increased risk to patients. These include those associated with residual effects of agents administered perioperatively.

Our previous study demonstrated that visual evoked potential latency was prolonged even 90 min postoperatively, when commonly used tests of psychomotor (Trieger dot test) and cognitive function (digit symbol substitution test) had returned to baseline levels [2]. This effect indicates that a disturbance of the visual pathway exists temporarily following general anaesthesia. This may have safety implications as some patients resume routine daily activity immediately after discharge, despite postoperative instructions.

The electroretinogram (ERG) is a summated potential generated by light-elicited electrical activity of retinal cells [3]. As the ERG can be recorded non-invasively and objectively, it is a useful tool for assessment of retinal function in basic research, animal models of various pathological states and clinical diagnosis. The ERG provides information about the function of specific retinal layers [3,4]. The a-wave is generated in the photoreceptor layer; the b-wave originates in the inner plexiform layer and the glial cells. Thus, the b-wave depends on the a-wave, and a defective photoreceptor layer would result in loss of the b-wave. Conversely, lesions of the neuronal layer have no influence on the a-wave. The ganglion layer is not involved in full-field flash ERG. The oscillatory potentials of the ERG consist of several rapid, low-amplitude potentials originating from depolarizing amacrine cells. Oscillatory potentials are used as a selective probe of neural circuits in the proximal retina and are sensitive to anoxia and drug effects [3,5]. The 30 Hz flicker ERG reflects cone activity exclusively. It is recorded using rod-suppressing background illumination at a flicker frequency of 30 Hz.

The objective of this study was to determine if abnormalities of the ERG exist after operation beyond the time at which currently used clinical discharge criteria for patients are met. Retinal abnormalities identifiable by ERG may signify diminished visual function such as contrast sensitivity. If such abnormalities persist into the postoperative period, they may justify changing our management and discharge criteria for patients undergoing ambulatory procedures.


Institutional Ethics Committee approval and written informed consent was obtained from 10 ASA I patients scheduled for non-neurological elective surgical procedures of approximately 1 h duration. Patients with decreased visual acuity or 'colour blindness' were excluded from the study. Patients received no preanaesthetic medication. They all had a full-field flash ERG recorded preoperatively and on two occasions postoperatively (i.e. at discharge from the recovery room and approximately 2 h after removal of the laryngeal mask airway).

Full-field flash ERGs were performed in accordance with the guidelines of the ERG Standardization Committee of the International Society for Clinical Electrophysiology of Vision (ISCEV) [6,7]. Each patient's right pupil was dilated approximately 20 min before the first preoperative measurement (with topical cyclopentolate hydrochloride 1% and phenylephrine 10%, one drop of each) and its diameter measured using a pupillometer (Essilor Digital Crp, Créteil, France). A corneal contact lens 'active' electrode (ERG-jet®; Universo, Switzerland), a self-stick reference electrode placed on the forehead and a ground electrode placed on the ear lobe were applied to each patient. The cornea was anaesthetized with topical proxymetacaine 1% and its surface protected during use with Methocel® 2% (Novartis Ophthalmics AG, Hettlingen, Germany), a non-irritating, non-allergenic ionic conductive solution (no more viscous than 0.5% methyl cellulose). The skin was cleaned and a water-soluble conducting electroencephalography paste (Ten 20®; D. O. Weaver and Co., Aurora, CO, USA) was used to achieve an impedance of <5 kΩ.

Full-field (Ganzfeld) stimulation was used, signals were amplified, filtered, on-line averaged, saved on disk and displayed using the Nicolet Bravo® ERG equipment (Nicolet Biomedical Inc, Madison, USA). Reference points were selected manually by an investigator unaware of the time of recording, and marked with cursors. The Nicolet® Bravo EP 3.2 (Nicolet Biomedia Inc, Madison, MI, USA) software package was subsequently used to measure the following phenomena.

In the photopic ERG, the latencies (ms) of the a- and b-waves were measured from the onset of the flash to the trough and peak, respectively. The A-B amplitude (μV) of the photopic ERG was measured from the trough of the a-wave to the peak of the b-wave. The amplitude of the a-wave was measured from the baseline (onset of flash) to trough. In the case of the oscillatory potentials, the interval from the onset of flash to first and second peak were measured and averaged (oscillatory potentials latency). The oscillatory potential amplitude was defined as the average of the amplitudes of the first and second peak, measured from the respective peak to the following trough.

The 30 Hz flicker ERG amplitude was characterized as the amplitude of the third peak in relation to the preceding trough (Fig. 1).

Figure 1
Figure 1:
Characteristic ERG recordings from the right eye of a patient. (a) Photopic ERG. The arrow indicates the onset of the flash. A and B indicate the trough and peak, respectively. See text for definition of parameters. (b) Oscillatory potentials. The arrow indicates the onset of the flash. O1 and O2 are the peaks, o1 and o2 are the troughs of the two measured oscillatory potentials. See text for definition of oscillatory potentials latency and amplitude. (c) A 30 Hz flicker ERG. The arrow indicates the onset of the flash. A and B indicate the trough and peak, respectively.

Patients also completed visual analogue scales (VAS) for sedation, anxiety and pain immediately prior to each ERG measurement [8].

With standard monitoring in place, pulse oximetry, electrocardiography, capnography, non-invasive blood pressure, and inspired and end-tidal partial pressures of sevoflurane, N2O and O2 (Datex® AS/3 monitor; Datex Corp., Helsinki, Finland), induction of anaesthesia was performed by administering sevoflurane 8% initially in O2 100%, and maintained by clinically indicated concentrations of sevoflurane in a O2 33%/N2O 66% mixture. Intraoperative analgesia was provided by administering diclofenac 100 mg rectally, or parecoxib 40 mg intravenously (i.v.); a local/regional block with bupivacaine 0.5% in volumes appropriate to the block performed up to a maximum of 2 mg kg−1. All patients breathed spontaneously through a laryngeal mask airway.

The time of the earliest Aldrete recovery score allowing potential discharge from the recovery room was noted [9,10]. This did not always coincide with the actual discharge time. Time when the postanaesthesia discharge score (PADS) (Table 1) would have allowed discharge home was also recorded [11,12].

Table 1
Table 1:
PADS for determining home readiness.

Data were analysed using paired, one-tailed Student's t-test and χ2 or Fisher's exact test as appropriate. P < 0.05 was considered significant.


Ten ASA I patients aged 39.9 yr (range 31-48 yr) were studied. Four patients underwent excision of breast lumps, one vasectomy and five minor limb surgery. The duration of anaesthesia was 58.4 min (range 40-78 min); end-tidal (ET) CO2 and sevoflurane concentration at removal of the laryngeal mask airway was 5 ± 0.6% and 0.3 ± 0.1%, respectively. Postoperatively, patients met the discharge criteria from the recovery room (Aldrete score ≥ 9) at 11 ± 2.3 min after removal of the laryngeal mask airway, and the PADS (≥9) criteria at 33 ± 6.7 min (Table 2). The first postoperative ERG measurement was performed as soon as possible after this, at 46 ± 9.7 min after removal of the laryngeal mask, considered as time point T1. The second postoperative ERG measurement was performed at 129.7 ± 9.7 min after removal of the laryngeal mask, defined as time point T2. The pupil size was comparable between recordings in each patient (8 ± 0.6 mm before each ERG measurement).

Table 2
Table 2:
Patient characteristics, anaesthetic duration, end-tidal (ET) CO2 and sevoflurane concentration at removal of laryngeal mask airway, times at which Aldrete and PADS ≥ 9 were first achieved.

The latencies and amplitudes of the a-wave on the photopic ERG were similar throughout the studied period. The b-wave latency was prolonged compared with the preanaesthetic values at times T1 and T2 (30.5 ± 0.9 and 30 ± 1.3 ms vs. 29.2 ± 0.8 ms, P < 0.001 and P = 0.04, respectively) after removal of the laryngeal mask, as shown in Figure 2. The A-B amplitude of the photopic ERG was decreased compared to the baseline at postoperative times T1 and T2 (220.3 ± 52.7 and 210.3 ± 42.7 μV vs. 248.1 ± 57.6 μV, P = 0.04 and P = 0.01, respectively).

Figure 2
Figure 2:
Photopic ERG parameters: (a) a-wave latency; (b) b-wave latency and (c) A-B amplitude of photopic ERG. Data are mean (SD). *P < 0.001, #P < 0.05 compared to baseline. T0 preoperative time point, T1 46 ± 9.7 min and T2 129 ± 9.7 min after removal of the laryngeal mask.

Figure 3 displays the characteristics of the oscillatory potentials. The oscillatory potential latency increased at T1 and T2 compared with preanaesthetic values (21.4 ± 0.5 and 20.8 ± 0.6 ms vs. 20.4 ± 0.4 ms, P < 0.001 and P = 0.03, respectively). Although oscillatory potential latency is still increased at T2, a tendency to revert towards baseline values was observed with a significant decrease between T2 and T1 (P = 0.002). The oscillatory potential amplitude was decreased at T1 compared with preanaesthetic values (17.5 ± 6.1 μV vs. 22 ± 6.4 μV, P = 0.04) and returned to baseline by T2.

Figure 3
Figure 3:
Oscillatory potentials. (a) Latency (ms) and (b) amplitude (μV). Data are mean (SD). *P ≤ 0.05, comparison to baseline; #P < 0.01, comparison between the two postoperative time points. T0 preoperative time point, T1 46 ± 9.7 min and T2 129 ± 9.7 min after removal of the laryngeal mask.

The 30 Hz flicker amplitudes were similar throughout the study period, although the tendency was towards a decreased amplitude at T1 (200.7 ± 17.1 μV vs. 220.6 ± 44.9 μV, P = 0.08).

Figure 4 shows the VAS scores for anxiety, sedation and pain. Anxiety scores were decreased at time T2 compared to baseline (16 ± 31 mm vs. 40 ± 38 mm, P = 0.04). Patients felt drowsy at time points T1 and T2 compared to the preanaesthetic scores (48 ± 33 and 22 ± 23 mm vs. 0 mm, P < 0.001 and P < 0.01, respectively). The degree of sedation decreased at time point T2 compared to T1 (P = 0.04). The VAS scores for pain show a great spread around the mean and they were similar throughout the study period. However, in all cases pain VAS scores were less than 40 at time point T2, allowing for potential home discharge.

Figure 4
Figure 4:
(a) Anxiety, (b) sedation and (c) pain VAS scores. Data are mean (SD). *P < 0.05, comparison to baseline; #P < 0.05, comparison between the two postoperative time points. T0 preoperative time point, T1 46 ± 9.7 min and T2 129 ± 9.7 min after removal of the laryngeal mask.


The main finding of this study is that the latency of the b-wave and that of the oscillatory potentials was prolonged and the A-B amplitude of the photopic ERG was diminished at 2 h following N2O/sevoflurane anaesthesia, after home discharge criteria had been met. Thus it is likely that abnormalities of the ERG continue to persist, and are likely to be due to the effects of residual sevoflurane.

The b-wave is the most prominent component of the ERG. Although conflicting evidence exists, one possibility is that b-wave parameters are influenced by GABAminergic activity modulated through both GABAA and GABAC receptors. GABA is the most important inhibitory neurotransmitter in the mammalian central nervous system. Traditionally, GABA-gated neurotransmitter receptors were classified in two structurally and pharmacologically distinct subclasses: GABAA and GABAB. Recently a third type of distinct GABA receptor, termed GABAC, has been proposed. GABAC receptors gate Cl currents in various parts of the vertebrate brain and were first described in interneurons of the spinal cord [13]. GABAC receptors are highly enriched in the vertebrate retina. Compared with GABAA receptors, GABAC receptors have a higher sensitivity for GABA, their currents are smaller and do not desensitize. On the single-channel level, these receptors are characterized by longer mean-open times and smaller Cl conductance [14].

Recently, Dong and colleagues demonstrated that GABAC feedback is an important mechanism through which third-order retinal neurons contribute to b-wave generation [15]. In the isolated superfused rabbit retina, GABA decreases the b-wave amplitude and GABA antagonists increase it [16]. More recent work has demonstrated that although a GABAA antagonist may increase the b-wave amplitude, a GABAC antagonist may decrease it [17]. In vivo, in rabbits, both selective GABAC receptor blocking agents and non-selective GABA receptor blocking agents reduced b-wave amplitude significantly and rendered b-wave kinetics slower [15]. Thus the current state of knowledge suggests that b-waves are cornea-positive waves that originate in the inner plexiform layer and may be modulated by a GABAminergic feedback mechanism from amacrine cells to bipolar cells.

Superimposed on the b-wave are short, fast and rhythmic components, the oscillatory potentials. These reflect the inner retinal activity, probably deriving from the amacrine cells. It has been suggested that these oscillatory potentials are generated as a chain of events involving an inhibitory mechanism within the retina. An increase or decrease of this inhibitory feedback mechanism would have an opposite effect on the response from the cells involved in generating the oscillatory potentials [18].

The complex neuronal structure of the retina renders it sensitive to the influence of drugs. Drugs can affect both phototransduction and neuronal transmission. Drug-induced ocular effects in human beings, which can be detected using the ERG, are well known. Examples of ERG effects induced by anaesthetic agents are summarized in Table 3.

Table 3
Table 3:
Known ERG effects of anaesthetic agents.

The mechanism by which sevoflurane produces its anaesthetic effects has not been completely defined. Sevoflurane is known to enhance GABAminergic transmission by (a) increasing the affinity of GABA for the GABAA receptor and (b) inducing a picrotoxin-like open-channel block at the GABAA receptor [27]. All patients in this study also received N2O that is known to potentiate GABAA receptor-mediated effects (to a much lesser effect than volatile anaesthetic agents) and to inhibit GABAC effects [28].

Our study findings demonstrate for the first time, in a clinical setting, the effects of sevoflurane, a non-selective GABA agonist, on the ERG. This ERG dysfunction is present after discontinuation of sevoflurane (administered in clinically relevant concentrations). This effect is likely to be true for other inhalational anaesthetic agents, especially when considering the low blood:gas partition coefficient (0.63-0.69) of sevoflurane. Owing to its low solubility in blood, alveolar concentrations of sevoflurane should decrease rapidly upon cessation of the agent (washout) [29].

A limitation of our study is the relatively small number of patients (10) studied. Using 'within patient' comparisons, and anaesthesia with a single agent, sevoflurane, reduced possible confounding variables. ERG changes were consistent and varied little between patients. Averaging, filtering and artefact rejection employed ensured reproducibility of the recorded waveforms by decreasing test-retest variability to a minimum. Stimulus and amplifier characteristics that may have substantial effect on the latency and amplitude of the measurement have been standardized and automated. Thus detection of subtle changes compared to the baseline preoperative measurement was possible. However, further work and a larger study need to be done.

What is the clinical significance of these findings? An abnormal ERG does not necessarily indicate that a visual deficit is present. However, patients taking vigabatrine (an anti-epileptic drug) have been shown to have a similar pattern of ERG changes (increased latency, decreased amplitude) and this strongly correlated with visual field defects [30]. Furthermore, one may speculate that it is the receptive field properties of ganglion cells - e.g. their centre-surround organization, which contributes to contrast sensitivity of the retina - that may be affected [14]. Abnormal contrast sensitivity following anaesthesia has obvious safety implications including the additional danger of driving motor vehicles especially under conditions of poor visibility. Each of these possibilities warrants further investigation.

In conclusion we have demonstrated that ERG abnormalities are consistently present after operation in patients following sevoflurane anaesthesia. Although these abnormalities tend to revert towards normal (preoperative) values 2 h postoperatively, they are certainly detectable after standard clinical discharge criteria have been met. The clinical significance of these reversible effects is uncertain. It is conceivable that the effects observed are mediated through GABAC receptors, which are present in abundance in the retina and may offer an accessible and sensitive measure of residual sevoflurane effects.


The authors wish to thank Peter Good, PhD, for his help in interpreting the oscillatory potential recordings.


1. Troy AM, Cunningham AJ. Ambulatory surgery: an overview. Curr Opin Anaesthesiol 2002; 15: 647-657.
2. Iohom G, Collins I, Murphy D, et al. Postoperative changes in visual evoked potentials and cognitive function tests following sevoflurane anaesthesia. Br J Anaesth 2001; 87: 855-859.
3. Karwosky C. Introduction to the origins of electroretinographic components. In: Hackenlively JR, Arden GB, eds. Principle and Practice of Clinical Electrophysiology of Vision. St Louis, USA: Mosby-Year Book, 1991: 87-91.
4. Carr RE, Siegel IM. Electrodiagnostic Testing of the Visual System: A Clinical Guide. Philadelphia, USA: F. A. Davis Company, 1990: 3-8, 12-15, 143-144.
5. Ikeda H. Clinical electroretinography. In: Halliday AM, ed. Evoked Potentials in Clinical Testing. Bath, UK: Churchill Livingstone, 1982: 121-148.
6. Marmor MF, Arden GB, Nilsson SE, Zrenner E. Standard for clinical electroretinography. International Standardization Committee. Arch Ophthalmol 1989; 107: 816-819.
7. Marmor MF, Zrenner E. Standard for clinical electroretinography (1999 update). International Society for Clinical Electrophysiology of Vision. Doc Ophthalmol 1999; 97: 143-156.
8. Bond A, Lader M. The use of analogue scales in rating subjective feelings. Br J Med Psychol 1974; 47: 211-218.
9. Aldrete JA, Kroulik D. A postanaesthetic recovery score. Anesth Analg 1970; 49: 924-934.
10. Aldrete JA. The postanaesthesia recovery score revisited. J Clin Anesth 1995; 7: 89-91.
11. Chung F. Recovery pattern and home-readiness after ambulatory surgery. Anesth Analg 1995; 80: 896-902.
12. Marshall S, Chung F. Assessment of 'home readiness': discharge criteria and postdischarge complications. Curr Opin Anaesthesiol 1997; 10: 445-450.
13. Sivilotti L, Nistri A. GABA receptor mechanism in the central nervous system. Prog Neurobiol 1991; 36: 35-92.
14. Enz R, Cutting GR. Molecular composition of GABAC receptors. Vision Res 1998; 38: 1431-1441.
15. Dong CJ, Hare WA. GABAC feedback pathway modulates the amplitude and kinetics of ERG b-wave in a mammalian retina in vivo. Vision Res 2002; 42: 1081-1087.
16. Gottlob I, Wundsch L, Tuppy FK. The rabbit electro-retinogram: effect of GABA and its antagonists. Vision Res 1988; 28: 203-210.
17. Kapousta-Bruneau NV. Opposite effects of GABA(A) and GABA(C) receptor antagonists on the b-wave of ERG recorded from the isolated rat retina. Vision Res 2000; 40: 1653-1665.
18. Blain L, Lachapelle P, Molotchnikoff S. The effect of acute trichloroethylene exposure on electroretinogram components. Neurotoxicol Teratol 1990; 12: 633-636.
19. Satoh H, Fukuda N, Kuriki H, et al. A procedure for recording electroretinogram (ERG) in conscious monkeys, and effects of some drugs. Nippon Yakurigaku Zasshi 1980; 76: 581-594.
20. Wioland N, Bonaventure N. Photopic c-wave in the chicken ERG: sensitivity to sodium azide, epinephrine, sodium iodate, barbiturates, and other general anesthetics. Doc Ophthalmol 1985; 60: 407-412.
21. Chynoransky M. The effect of diazepam on the electro-retinogram. Bratisl Lek Listy 1990; 91: 527-532.
22. Andreasson S, Tomqvist K, Ehinger B. Full-field electro-retinograms during general anesthesia in normal children compared to examination with topical anesthesia. Acta Ophthalmol (Copenh) 1993; 71: 491-495.
    23. Yagi M, Tashiro C, Yoshiya I. Changes in the electro-retinogram during enflurane anesthesia. Masui 1989; 38: 1438-1443.
    24. Raitta C, Karhunen U, Seppälainen AM. Changes in the electroretinogram and visual evoked potentials during general anaesthesia using enflurane. Graefes Arch Clin Exp Ophthalmol 1982; 218: 294-296.
    25. Wongpichedchai S, Hansen RM, Koka B, Gudas VM, Fulton AB. Effects of halothane on children's electroretinograms. Ophthalmology 1992; 99: 1309-1312.
    26. Wasserschaff M, Schmidt JG. Electroretinographic responses to the addition of nitrous oxide to halothane in rats. Doc Ophthalmol 1986; 64: 347-354.
    27. Hapfelmeier G, Schneck H, Kochs E. Sevoflurane potentiates and blocks GABA-induced currents through recombinant α1β2γ2 GABAA receptors: implications for an enhanced GABAergic transmission. Eur J Anaesthesiol 2001; 18: 377-383.
    28. Yamakura T, Harris RA. Effects of gaseous anesthetics nitrous oxide and xenon on ligand-gated ion channels. Comparison with isoflurane and ethanol. Anesthesiology 2000; 93: 1095-1101.
    29. Eger II EI, Gong D, Koblin DD, et al. The effect of anesthetic duration on kinetic and recovery characteristics of desflurane versus sevoflurane, and on the kinetic characteristics of compound A, in volunteers. Anesth Analg 1998; 86: 414-421.
    30. Besch D, Kurtenbach A, Apfelstedt-Sylla E, et al. Visual field constriction and electrophysiological changes associated with vigabatrin. Doc Ophthalmol 2002; 104: 151-170.

    ANAESTHESIA; ANAESTHETICS INHALATIONAL, sevoflurane, nitrous oxide; ANAESTHESIA RECOVERY PERIOD; DIAGNOSTIC TECHNIQUES, ophthalmological, electroretinography

    © 2004 European Academy of Anaesthesiology