Equine Sedation Electroencephalography.
Qualitative and Quantitative Characteristics of the Electroencephalogram in Normal Horses After Sedation
Article first published online: 10 APR 2012
Copyright © 2012 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 26, Issue 3, pages 645–653, May-June 2012
How to Cite
Williams, D.C., Aleman, M., Tharp, B., Fletcher, D.J., Kass, P.H., Steffey, E.P., LeCouteur, R.A. and Holliday, T.A. (2012), Qualitative and Quantitative Characteristics of the Electroencephalogram in Normal Horses After Sedation. Journal of Veterinary Internal Medicine, 26: 645–653. doi: 10.1111/j.1939-1676.2012.00921.x
This study was performed at the Center for Equine Health, School of Veterinary Medicine, University of California at Davis.
Poster presented at the 8th Annual World Congress of Veterinary Anesthesia held in Knoxville, Tennessee in September 2003. Abstract published in the WCVA Proceedings in Veterinary Anesthesia and Analgesia 2004, 31(4):40–41.
Submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Comparative Pathology in the Office of Graduate Studies of the University of California Davis.
- Issue published online: 2 MAY 2012
- Article first published online: 10 APR 2012
- Manuscript Accepted: 29 FEB 2012
- Manuscript Revised: 13 FEB 2012
- Manuscript Received: 7 OCT 2011
- Clinical Electrophysiology Laboratory
- Center for Equine Health
- Oak Tree Racing Association
- State of California
- Equine sleep;
The administration of certain sedatives has been shown to promote sleep in humans. Related agents induce sleep-like behavior when administered to horses. Interpretation of electroencephalograms (EEGs) obtained from sedated horses should take into account background activity, presence of sleep-related EEG events, and the animal's behavior.
Sedatives induce states of vigilance that are indistinguishable on EEGs from those that occur naturally.
Six healthy horses.
Digital EEG with video was recorded after administration of 1 of 4 sedatives (acepromazine, butorphanol, xylazine, or detomidine). Serum drug concentrations were measured. Recordings were reviewed, states were identified, and representative EEG samples were analysed. These data were compared with data previously obtained during a study of natural sleep.
Butorphanol was associated with brief episodes resembling slow wave sleep in 1 horse. Acepromazine led to SWS in 3 horses, including 1 that also exhibited rapid eye movement sleep. Periods of SWS were observed in all horses afer xylazine or detomidine administration. Normal sleep-related EEG events and heart block, occurred in association with SWS regardless of which sedative was used. Spectral data varied primarily by state, but some differences were observed between sedative and natural data.
Conclusions and Clinical Importance
Qualitatively, EEG findings appeared identical whether sedation-induced or naturally occurring. The startle response and heart block associated with some sedatives may be related to sleep. Alpha2 agonists can be used to obtain high quality EEGs in horses, but acepromazine does not promote a relaxed state in all animals.
delta frequency band (>0–<4 Hz)
theta frequency band (4–<8 Hz)
alpha frequency band (8–<13 Hz)
beta frequency band (13–<30 Hz)
gamma frequency band (>30 Hz)
early slow wave sleep
late slow wave sleep
rapid eye movement sleep
slow wave sleep
- V wave
vertex sharp wave
Electroencephalography in veterinary medicine sometimes is challenging. Electrical interference can be a problem and patient related artifacts can arise from muscle activity near the electrodes. Artifacts also are associated with movement. To optimize recording quality, animals often are sedated for EEG recording. Less commonly, general anesthesia may be used.[1, 2]
The ideal sedative for EEG purposes should induce a relaxed state in the animal. Alpha2 adrenergic agonists have been used in various species to record EEG signals. Medetomidine and dexmedetomidine have been used in dogs,[3, 4] cats,[5, 6] and sea lions. Recordings from normal and epileptic dogs under xylazine sedation were compared in 1 study. This drug is widely used for chemical restraint in the horse and its systemic effects have been well documented.[9, 10] Few reports describe the EEG characteristics associated with the administration of sedatives as single agents in this species.[11, 12]
Purohit et al recorded EEGs in 6 normal horses, before and after xylazine administration. The authors described a change in background activity from a predominately fast, low voltage pattern in the nonsedated horse to one characterized by slow waves of moderate amplitude after sedation. Occasional spindles were noted both before and after drug administration.
Mysinger et al recorded EEG from 25 normal adult horses. Eight received xylazine and 17 received acepromazine. In the xylazine group, all demonstrated generalized slowing of background activity with an increase in amplitude. Acepromazine exerted little effect on the EEG, although increased spindle durations were noted. The authors described those animals that lacked spindle activity as being “apprehensive”.
Natural (nonsedated) states of vigilance with their corresponding ECoG/EEG features have been described in normal horses and ponies.[13-18] Four states are routinely recognized: (1) wakefulness (WK), (2) drowsiness (DR, relaxed wakefulness), (3) slow wave sleep (SWS), and (4) rapid eye movement (REM) sleep (paradoxical sleep[13-16]). Criteria for state designations vary slightly among authors. Wakefulness, drowsiness, and REM tend to be low voltage and high frequency, whereas SWS is high voltage and low frequency. State-related events such as sleep spindles, vertex sharp waves, and K-complexes also have been described. The animal's body position can vary for a given state and may reflect how well an animal has adapted to its surroundings.
To improve the likelihood of detecting abnormal activity in an EEG recorded from a human patient, a period of SWS should be included. Degen and Degen compared the use of promazine hydrochloride to that of sleep deprivation and found sedation to be nearly as effective for promoting SWS in human epileptics. Dexmedetomidine and clonidine, α2 agonists, also have been shown to induce SWS in humans.[21, 22]
The goal of this project was to determine the effects of commonly used sedatives on the EEG of normal horses with regard to state of vigilance. Because sedatives induce behavior that resembles sleep, the results of this study are applicable to the fields of veterinary anesthesiology and neurology. Furthermore, the knowledge gained may be used to optimize the recording and interpretation of equine EEGs in clinical and research settings.
Materials and Methods
Animals—Six adult horses of various breeds that were normal on complete neurologic examination (performed by MA) were selected from a research herd for use in this project. In addition to the 5 horses from a previous natural (nonsedated) study (4 mares [a Standardbred, a Paint, and 2 Thoroughbreds] and a Quarter Horse gelding, all between 4 and 13 years of age), a 6-year-old Quarter Horse gelding was included. Horses were kept in small groups on dry lots and provided care in accordance with an approved University of California, Davis Animal Use and Care Protocol.
Equipment—Recording conditions were identical to those previously described. All recordings took place in a foaling stall that had been adapted for equipment isolation while allowing the tethered horse to move about during the session. Video monitoring was used for assessing behavior. Before placement of the electrodes, a 6-inch catheter1 was positioned in the left jugular vein by aseptic technique. Lidocaine2 was used to block 2 sites where sutures were placed to hold the catheter in position.
Electrode Placement—After shaving, cleaning, and marking the sites (by means of individual templates designed for each horse), 16 surface electrodes3 for recording EEG (left and right frontopolars [FP1, FP2], 3 each for frontals [F3, Fz, F4], centrals [C3, Cz, C4], and parietals [P3, Pz, P4], 2 each for occipitals [O1, O2] and aurals [A1, A2]), 3 for EOG (left and right lateral canthi [OS, OD] and intercanthus [IC]), and a single ground electrode (Z), were positioned as described previously. Needle electrodes4 were used to record the ECG (base-apex configuration) and splenius muscle EMG activity. Respiration was monitored with a piezoelectric sensor5 placed over the caudal thorax.
Drug Administration—A Latin Square design was used to randomize drug administration. Acepromazine maleate6 (0.06 mg/kg), xylazine hydrochloride7 (1.0 mg/kg), detomidine hydrochloride8 (0.03 mg/kg), or butorphanol tartrate9 (0.05 mg/kg) was administered IV shortly after instrumentation. With the exceptions listed below, each horse received each of the 4 treatments. A minimum of 9 days elapsed between studies on any given horse to allow time for the previously administered agent to be cleared. For each sedative given, a blood sample was collected at the following times (in minutes) after drug administration: 0, 2, 5, 10, 15, 30, 60 for each study. Additional samples were collected at 120 and 180 minutes after acepromazine or detomidine administration. All samples were kept refrigerated or on ice, until the serum could be separated and stored at −80°C.
Drug Analysis—To determine if EEG findings were related to the amount of circulating sedative present, serum drug concentrations were measured as previously described.[23, 24]
Recording Data—EEG and video recordings were completed as previously described. Total recording period length varied with the duration of sedative effects. All EEGs were performed during the evening hours.
Data Analysis—Sleep staging criteria applied to these recordings were identical to those previously described.
Wakefulness was used to describe a horse in the alert state, actively engaged in exploring its environment.
Drowsiness (relaxed wakefulness) was defined as a period of quiescence, standing or lying still, that followed wakefulness.
Slow Wave Sleep was identified by the appearance of a sleep spindle with a duration of at least 0.5 seconds (Stage 2 in humans). A gradual increase in background activity amplitude with a concurrent frequency decrease followed (Stages 3 and 4).
Rapid Eye Movement Sleep was identified as return to a low amplitude pattern (that was not associated with arousal) and diminished EMG activity in the splenius muscle.
For each EEG, durations were calculated for each state of vigilance. For acepromazine, detomidine, and xylazine, all 6 horses were included. Butorphanol administration was associated with excitement in 2 horses, and tremors and aggression in 2 others. For safety reasons, the last 2 recording sessions planned for this drug were canceled. Data regarding state of vigilance were compared with that obtained during a previously reported natural study (n = 5).
Using a transverse montage for review, quantitative electroencephalographic (qEEG) analyses consisted of the selection of 4 relatively artifact-free 10-second epochs, representative of each vigilance state for each horse. As was the case in the previous report, the values were obtained from the C4-A2 channel. Absolute power for each frequency band was obtained via Fast Fourier Transform (FFT) analysis of each epoch. Frequency band definitions were identical to those used in the previous study. For the additional calculation of relative power, the gamma (γ) band (>30 Hz) was included in this study. Relative power has been utilized in previous veterinary qEEG publications,[3, 5] but comparisons between absolute and relative power have not been described in this species. To address this, means were obtained for each horse and combined means, both for absolute (Abs) and relative (Rel) power, were calculated for each frequency band. For the latter, absolute power in each frequency band was divided by the total power (power in all bands combined) and converted to a percentage.
Because SWS was plentiful in all horses during xylazine and detomidine, 2 sections of EEG were analyzed for this state in both studies. Early SWS (E-SWS) was obtained between 10 and 15 minutes after drug administration, and late SWS (L-SWS) after a minimum average elapsed time between sets of 23.5 minutes for xylazine and 60.3 minutes for detomidine. Mean cardiac and respiratory values were obtained from the 4, 10-second segments.
Statistical analyses, evaluating the effects of (1) drug and (2) state of vigilance on combined mean data for each frequency band (δ, θ, α, β, γ), for each study (natural, acepromazine, xylazine, and detomidine [because of the problems listed above, butophanol was excluded]) were performed by 2-way repeated measures analysis of variance (RMANOVA), with significance set at P < .05. Post hoc tests were performing utilizing the sequentially rejective method of multiple comparison adjustment (Holm–Bonferroni).
Totals for mean percentage time spent in each state of vigilance were variable among sedatives and when compared with data from the previously reported natural study (Table 1). For qEEG, the mean total power for all frequency bands was 190.5 μV2 for SWS, compared with DR at 94.5 μV2 and REM at 99.3 μV2 (data combined across all studies). Comparisons of statistical data (P-values) are displayed for Abs and Rel power by frequency band for the main effects of drug (Table 2a) and state of vigilance (Table 2b). Relative power results are displayed in Fig 1.
|Drug Effects Analysis||δ||θ||α||β||γ|
|A vs N||.707||.821||.148||.437||.601||.915||.402||.284||.659||.498|
|X vs N||.051||.665||.008a||.423||.001a||.015||.022||.437||.640||.030|
|D vs N||.163||.141||.008a||.926||.002a||.051||.022||.102||.707||.898|
|X vs D||.135||.378||.007a||.588||.082||.850||.116||.018||.176||.025|
|State Effects Analysis||δ||θ||α||β||γ|
|A vs N||.005a||.003a||.016a||.718||.008a||.864||.023||<.001a||.013a||<.001a|
|X vs N||.005a||<.001a||.015a||.334||.004a||.009a||.049||<.001a||.019a||<.001a|
|D vs N||.003a||.001a||.002a||.560||.001a||<.001a||.057||<.001a||.018a||<.001a|
|X vs D||.002a||.013a||.015a||.060||.003a||<.001a||.021||<.001a||.009a||<.001a|
|X E-SWS vs L-SWS||.213||.066||.132||.951||.080||.062||.025a||.037a||.024a||.038a|
|D E-SWS vs L-SWS||.430||.924||.896||.478||.030a||.041a||.497||.850||.284||.301|
Mean serum drug concentrations are shown in Figure 2 with the approximate times corresponding to E-SWS and L-SWS epoch selection for xylazine and detomidine.
Horses were reactive to stimuli during all studies. Loud sounds, the sight of researchers moving about (during blood collection periods), or touch (occasional electrode replacement) at times resulted in abrupt transitions in mentation (startle response).
Butorphanol was administered to 4 horses. It was associated with 3 brief periods of SWS in horse #1. Drowsiness was noted in this horse and in horse #6. Only WK was recorded from the other 2 horses, both of which appeared agitated. The mean duration of recording (in hours : minutes : seconds) was 0 : 32 : 17. Median heart rates were 53.5 for DR (compared with 33 for the natural study) and 56 during SWS (31.5) with respiratory rates of 15 (12) and 12 (15). Because of the paucity of readable butorphanol data, statistical analyses were not done for this study. However, the limited qEEG data may be seen in Figure 1b.
Acepromazine findings varied. Half of the horses had periods of SWS during this drug study. Horse #3 had several brief periods of DR, but horses #1 and #2 were awake throughout the recording period. Drug effects were not immediate, with DR occurring 10 minutes postadministration in horse #4, 21 minutes postadministration in horse #5, and 17 minutes postadministration in horse #6. One hour after drug administration, horse #4 adopted a recumbent posture for 17 minutes, nearly all of which was lateral. A brief rolling episode caused the loss of all ECG, EOG, and EMG data. While recumbent, this horse alternated among all 4 states of vigilance. Rapid eye movements during REM were apparent in the rostral (frontopolar) channel. It was the only example of REM sleep outside of the natural study with the exception of 1, 10-second epoch in horse #5, after acepromazine administration. Previously described events associated with SWS (vertex sharp waves [V waves], sleep spindles, and K-complexes) were seen in these EEG recordings. Numerous DR-associated transient discharges (TD) were noted in horse #5′s EEG recording (Fig 3), and occasional ones also were seen in the EEG recordings from horses #3 and #4. As in the natural study, they demonstrated maximal voltage at the central vertex and their morphology resembled spike-and-wave or multiple spike/sharp wave complexes. With the exception of lateral recumbency (not observed during the natural study) postures were identical to those previously reported. Statistical analyses of qEEG data revealed no significant differences in Abs or Rel power between acepromazine and the natural study across all frequency bands (Table 2a). However, there were multiple differences by state (Table 2b) with 7 of 10 analyses (5 Abs, 5 Rel) attaining significance. Second degree AV block was a common finding during DR and SWS. Median heart rates (and respiratory rates) were 46.5 (14) and 39 (12) for DR and SWS, respectively. Mean recording duration was 1:00:46.
Xylazine administration resulted in a dramatic increase in the amount of DR and SWS in all horses. A base-wide stance, with the head held just above the floor, was adopted by all animals for the duration of the studies (mean, 0:53:26). In DR, TDs were a common finding in the EEG recordings of horses #4, #5, and #6, and were seen occasionally in #3. Normal sleep events were present. Intermittent bursts of fast β/slow γ activity (28–32 Hz) were noted in the EEG recordings from 4 horses (Fig 4). These events were voltage maximal (30 μV) in the frontopolar region and also were apparent in the EOG. Sweat artifact was common, causing the baseline to drift in some channels (Fig 4). Statistical differences (2/10) between xylazine and the natural study were observed (Table 2a). Differences were greater by state (8/10, Table 2b). Second degree AV block was associated with DR and SWS (Fig 4). Median heart rates (and respiratory rates) were 30.75 (9.75), 26.25 (9), and 34.5(7.5) for DR, E-SWS, and L-SWS, respectively. REM was not present during the recording period.
Detomidine effects were identical to those of xylazine, but persisted over twice as long (mean, 2:13:45). TDs were present during DR. Fast β/slow γ bursts were seen in 3 of the horses. Sweat artifact was common, as were normal sleep events. Some differences (2/10) between detomidine and the natural study were observed (Table 2a). Differences were greater by state (8/10, Table 2b). Second degree AV block was present in DR and SWS. Median heart rates (and respiratory rates) were 37 (9), 24 (13), and 36 (10 .5) for DR, E-SWS, and L-SWS, respectively.
Comparisons between the 2 α2 agonists showed 1/10 analyses were significant by drug main effect and 8/10 by state main effect (Table 2).
Significant differences were found between E-SWS and L-SWS for xylazine in the β and γ bands, 4/10, and for detomidine in the α band only, 2/10 (Table 2b). For all studies, normal sleep events were included in qEEG analyses, but TDs were not.
In a recent article, Hubbell et al describe the results of a survey sent to members of the American Association of Equine Practitioners (AAEP) regarding their drug preferences for analgesia, anesthesia, and chemical restraint in the horse. For the latter purpose, detomidine, either alone (18%) or in combination (40%) was the first choice for 58% of the respondents. Thirty-four percent preferred xylazine alone (10%) or in combination (24%). Butorphanol was the drug typically added to either α2 agonist, when multiple drugs were utilized. The study found that acepromazine (alone or in combination) was the drug of choice for <1% of those who responded to the survey.
The findings in this study support the use of α2 agonists as chemical restraint for EEG recording in horses. In all horses, administration of xylazine or detomidine resulted in acquisition of high quality EEG recordings as demonstrated by the presence of relatively little artifact. The doses employed for this study were those typically used for analgesia or as premedication for general anesthesia.[9, 10] They were considerably higher than those routinely used clinically for EEG recording (personal observations, DCW and MA [acepromazine and butorphanol are not typically used for this purpose]). However, the qEEG data showed no significant differences between E-SWS and L-SWS for either drug in the δ and θ frequency bands (Table 2), where the majority of power is during SWS (Fig 1), suggesting that sleep state may be more critical than drug concentration. The startle response, attributed to α2 agonists, seems likely to be nothing more than a sudden arousal from SWS, generally in response to an external stimulus. Acepromazine administration resulted in complete EEG recordings (including SWS) in only 50% of the horses. Because of the observed excitement, butorphanol alone is not recommended for this purpose.
In equine medicine, anesthesiologists have published extensively on the topic of qEEG. Unfortunately, the raw EEG used to create the data is not available for review. Even under ideal circumstances, artifact can alter the EEG signal (Fig 4), thus qEEG results should always be interpreted with caution. The American Academy of Neurology and the American Clinical Neurophysiology Society recommend that qEEG in the clinical setting only be used as an accompaniment to traditional EEG interpretation.
On statistical analysis, the effects of the independent variables of drug and state on the dependent variable of frequency band power tended to be nonsignificant among drugs, whereas significant differences generally were found by state for all drugs (Table 2). This supports the findings that DR and SWS are different states (WK was primarily artifact and REM data were basically limited to 1 horse, thus statistical analyses could not be performed on these states), but that the sedative used to attain those states has little impact on the EEG. As was the case in a study of cats, results were in conflict when comparing Abs and Rel power (Table 2). This can be explained by differences in the total power, by age, for the previous study and by state, for this study (SWS has double the power of DR [and REM]). By converting power to percentages (Rel), total power information is lost resulting in SWS and DR being treated equally. The θ results demonstrate this phenomenon, whereby with the exception of acepromazine all Abs findings were significant, but all Rel results were not (for both drug and state main effects). Intermittent θ activity was common in DR (and REM) during the natural and acepromazine studies, but less so with xylazine and detomidine, although this may be a function of sample selection. Statistical differences in β and γ may be a matter of the amount of muscle artifact, common during DR as well as the intermittent bursts of activity in this range noted in some xylazine and detomidine recordings. Similar bursts have been observed in human and foal EEGs, particularly when sedation has been used to induce sleep.[29, 30] Other examples were published in Lewin's dissertation and the article by Mysinger et al. Sweat artifact likely had some influence on qEEG values in the δ band. Variability in α data may be related to the number, duration, or both of spindles in each epoch.
Confusion exists in the equine literature regarding the α rhythm. Mysinger et al used the term α range activity synonymously with the presence of spindles. The alpha rhythm is a specific pattern recognized in the EEGs of humans, defined by its distribution (caudal) and reactivity (it disappears upon opening of the eyes). If the horse did exhibit an alpha rhythm, it would be a prominent feature of the EEG during DR (relaxed WK). However, the qEEG α data show that there actually was less relative power in DR than in SWS for acepromazine, xylazine, and detomidine, and the 2 states were virtually identical in the natural study (Fig 1). Previous ECoG recordings from the occipital region do not support the presence of an α rhythm in this species.[13-16] In contrast, spindles are brief rhythmic bursts that, once they reach a minimum of 0.5 seconds in duration, are used to identify the onset of Stage 2 sleep in humans and SWS in horses. These were numerous in this study and as was the case in Mysinger et al were associated with a relaxed state, and thus were more consistent when α2 agonists were administered.
Acepromazine findings shared some features between those of Mysinger et al and this study. However, half of the horses in this study did exhibit patterns characteristic of SWS, not just spindles superimposed on a low amplitude, high frequency background. Mysinger et al do not mention the duration of the EEG recordings, it may be that recording periods were too short for the background to develop fully. Neither Mysinger et al or Purohit et al attempted to record natural sleep with which to compare with their findings.
Heart and respiratory rate comparisons could not be made between this study and the Mysinger et al and Purohit et al studies (the former did not report them and the latter described the change in percentages). Except for an increase in heart rate after butorphanol administration, the values here were similar across all sedatives and when compared with the natural study. For both xylazine and detomidine, minor heart rate decreases were noted in E-SWS, but L-SWS values were close to those of DR. Although present during the natural and acepromazine studies, second degree AV block was more frequent in the xylazine and detomidine studies, especially immediately after drug administration. This suggests that the α2 agonists have an additive effect to that of the sleep-induced heart block. Respiratory rates tended to be lower with xylazine and detomidine, consistent with the respiratory depression associated with a2 agonist administration. Acepromazine and butorphanol related respiratory rates were similar to the natural study results.
By applying the natural study EEG/behavior criteria, states of vigilance were readily identifiable, regardless of the sedation used to induce them. Sleep was consistently triggered by use of xylazine and detomidine, but was limited to SWS. This state is considered an activation technique for promoting epileptiform activity in the EEGs of human epileptics, and it appears to be beneficial in veterinary patients as well.[7, 30] In some cases, acepromazine may be a good alternative, but its effects are typically not immediate and patience is required. Acepromazine does not appear to suppress REM, thus it may be well suited for performing EEGs during sleep studies. At the dose used in the current study, butorphanol was not compatible with obtaining high quality EEGs, but it has been used successfully in combination with other drugs in some species, including the horse.[7, 28, 32] Butorphanol may play a minor role in achieving the desired level of sedation. Second degree AV block was associated with sleep in all studies (except butorphanol), but was exacerbated by α2 agonist administration. This study validates the use of xylazine and detomidine for recording EEGs in horses. It is the first to describe the association between sedative administration and changes in states of vigilance in the horse. Additional EEG studies utilizing combinations of sedatives are indicated.
The authors thank Mr John Doval, Dr Cynthia Baker, and Dr Scott Stanley for technical assistance.
This research was funded by the Center for Equine Health with funds provided by the Oak Tree Racing Association, the State of California pari-mutuel fund, and contributions by private donors. It was also supported by the Clinical Electrophysiology Laboratory at the William R. Pritchard Veterinary Medical Teaching Hospital.
Angiocath, Becton-Dickinson, Franklin Lakes, NJ
Lidocaine, Hospira, Inc., Lake Forest, IL
F-E5GH-120, Grass/Astro-Med, West Warwick, RI
1497, Nihon Kohden America, Inc., Foothill Ranch, CA
PromAce, Fort Dodge Animal Health, Overland Park, KS
Rompun, Bayer HealthCare, Monheim, Germany
Dormosedan, Pfizer Animal Health, Exton, PA
Torbugesic, Fort Dodge Animal Health
- 6Electroencephalography. In: Vaden SL, Knoll JS, Smith FWK, Tilley FP, eds. 5-Minute Veterinary Consult: Laboratory Tests and Diagnostic Procedures: Canine and Feline. Hoboken, NJ: Wiley-Blackwell Publishing, 2009..
- 17Eine Methode zur nichtinvasiven EEG-Ableitung am wachen, stehenden Pferd. Berlin; 1998. PhD dissertation. Freien Universität Berlin publisher..
- 19Effects of sleep and sleep deprivation on interictal epileptiform discharges. In: Bazil CW, Malow BA, Sammaritano MR, eds. Sleep and Epilepsy: The Clinical Spectrum. Amsterdam: Elsevier Science BV, 2002;157–164., .
- 23Pharmacokinetics and adverse effects of butorphanol administered by single intravenous injection or continuous intravenous infusion in horses. AJVR 2001;62:182–189., , , et al.
- 25A simple sequentially rejective multiple test procedure. Scand J Stat 1979;6:65–70..
- 29The normal sleep of adults over 20. In: Fisch BJ, ed. Fisch and Spehlmann's EEG Primer. Amsterdam: Elsevier, 2007:99–206..
- 31Rechtschaffen A, Kales A, eds. A Manual of Standardized Terminology: Techniques and Scoring System for Sleep Stages of Human Subjects. Washington, DC: Public Health Service, U.S. Printing Office; 1968.