Effects of acute hypoxia on postural and kinetic tremor

Eur J Appl Physiol

ORIGINAL ARTICLE

EVects of acute hypoxia on postural and kinetic tremor

A. Legros · H. R. Marshall · A. Beuter · J. Gow · B. Cheung · A. W. Thomas · F. S. Prato · R. Z. Stodilka

Accepted: 1 April 2010© Springer-Verlag 2010

Abstract Human physiological tremor is a complexphenomenon that is modulated by numerous mechanical,neurophysiological, and environmental conditions. Research-ers investigating tremor have suggested that acute hypoxiaincreases tremor amplitude. Based on the results of priorstudies, we hypothesized that human participants exposedto a simulated altitude of 4,500 m would display anincreased tremor amplitude within the 6–12 Hz frequencyrange. Postural and kinetic tremors were recorded with a

laser system in 23 healthy male participants before, during,and after 1 h of altitude-induced hypoxia. A large panel oftremor characteristics was used to investigate the eVect ofhypoxia. Acute hypoxia increased tremor frequency contentbetween 6 and 12 Hz during both postural and kinetictremor tasks (P < 0.05, F = 6.142, Eta2 = 0.24 and P < 0.05,F = 3.767 Eta2 = 0.14, respectively). Although the physio-logical mechanisms underlying the observed changes intremor are not completely elucidated yet, this study con-Wrms that acute hypoxia increases tremor frequency in the6–12 Hz range. Furthermore, this study indicates thatchanges in physiological tremor can be detected at lowerhypoxemic levels than previously reported (blood satura-tion in oxygen = 80.9%). The eVects of hypoxia mainlyresult from a cascade of events starting with the activationof the hypothalamic–pituitary–adrenal axis causing in turnan increase in catecholamine release, leading to an augmen-tation of tremor amplitude in the 6- to12-Hz intervaland heart rate increase.

Keywords Acute hypoxia · Postural tremor · Kinetic tremor · Human simulated altitude

Introduction

Physiological tremor present in all humans can be deWnedas an involuntary, irregular, and continuous movement ofthe body’s extremities (Elble 1986). Three main mecha-nisms are reported to be involved in physiological tremorgeneration (Deuschl et al. 2001; McAuley and Marsden2000). First, mechanical properties of the consideredsegment tend to make it “vibrate” at its resonance fre-quency, which is speciWcally determined by its mass and itsstiVness level. For example, the unloaded index Wnger has a

Communicated by Susan Ward.

A. Legros (&) · H. R. Marshall · A. W. Thomas · F. S. Prato · R. Z. StodilkaImaging Program, Lawson Health Research Institute, St. Joseph’s Health Care, 268 Grosvenor St., London, Ontario N6A 4V2, Canadae-mail: [email protected]

A. Legros · H. R. Marshall · A. W. Thomas · F. S. Prato · R. Z. StodilkaDepartment of Medical Biophysics, University of Western Ontario, 1151 Richmond St., London, Ontario N6A 5C1, Canada

A. BeuterIMS Laboratory, Bordeaux Polytechnic Institute, Bordeaux University, 16 avenue Pey-Berland, 33607 Pessac Cedex, France

J. GowDepartment of kinesiology, University of Western Ontario, 1133 Sheppard Avenue, Toronto, Ontario M3M 3B9, Canada

B. CheungDefense Research and Development Canada, 1133 Sheppard Avenue, Toronto, Ontario M3M 3B9, Canada

123

Eur J Appl Physiol

resonance frequency between 25 and 27 Hz (Stiles andRandall 1967). A second mechanism, the stretch or myo-tatic reXex, works as a negative feedback loop. It involvesthe reXex contraction of a muscle in response to stretchingof an attached tendon or of the muscle itself. When thereXex gain and the conduction time for the aVerent andeVerent pathways are appropriate, an oscillation occurs(Stein et al. 1978). The loop time of this stretch reXexmechanism is 50 ms at the level of the Wnger and leads totremor generation at a frequency of 10 Hz (Lippold 1970).The third mechanism involves a rhythmical oscillation ofcentral origin inducing a synchronization of motor unitWring rates, at frequencies ranging between 8 and 12 Hz(Elble 1986; Goodman and Kelso 1983; Llinas and Volkind1973; Travis 1929). An argument often evoked to supportthis central mechanism is that postural tremor is modiWedbut preserved in deaVerented patients who no longer havefunctional feedback loops (Marsden et al. 1967). Althoughthis system has thus far proved too complex to understandfully, it is generally accepted that both the central andperipheral nervous systems contribute to the generation ofphysiological tremor in proportions that vary according tothe environmental conditions and the health of the subject(Elble 1986, 1996; Elble and Koller 1990; Vaillancourt andNewell 2000).

Postural tremor is present during the maintenance of aposition against gravity (Deuschl et al. 1998). It can berecorded with or without visual feedback and is sensitive toa variety of environmental conditions and physiologicalconditions. Postural tremor generation can be modulated byfactors acting on neural pathways such as caVeine, mer-cury, alcohol, temperature, limb position, and exposure tomagnetic Welds (Beuter and Edwards 1998; Koller et al.1987; Lakie et al. 1994a, 1994b; Legros et al. 2006;Mazzocchio et al. 2008). For example, a study listed morethan 30 factors inXuencing tremor characteristics (Wachsand Boshes 1966). These changes in tremor sometimesaVect its amplitude (Edwards and Beuter 2000) and some-times its frequency content (Beuter and Edwards 1999).Tremor is also often a predominant symptom of speciWcneurophysiological dysfunctions such as Parkinson’s dis-ease (Beuter and Edwards 1999, 2002; Edwards and Beuter2000) and other disorders (Britton and Gresty 1992; Brittonand Thompson 1995; Deuschl et al. 1998; Elble 1986). Ithas been reported that when tremor becomes abnormal, theobserved changes are a reduction in frequency, an increasein amplitude and amplitude Xuctuations, and a frequencydistribution more organized around dominant values(Beuter et al. 2000). In other words tremor slows down,becomes larger, and tends to Xuctuate in amplitude.

Kinetic tremor is deWned as involuntary oscillatorymovements occurring during voluntary movements of theextremities performed with or without visual feedback

(Deuschl et al. 1998). Typically, kinetic tremor is recordedin the index Wnger while the subject performs a compensa-tory tracking task. This provides the opportunity of measur-ing not only tremor during action but also to characterizethe performance both in time and space associated with thetask (Beuter and Edwards 2002). Kinetic tremor distur-bances observed during a tracking task suggests the pres-ence of a disturbance of the cerebellum or its pathways(Hallett 1991).

Neurophysiological functions such as event-relatedpotentials, reaction time, psychomotor performances orcognitive processes have been reported to be aVectedunder exposure to high altitude (Virues-Ortega et al.2004; Virues-Ortega et al. 2006). For example, Li et al.(2000) investigated psychomotor performance and visualreaction time after acute exposure to hypoxia of 1 h dura-tion at altitudes of 300 m (control), 2,800, 3,600, and4,400 m. The results showed no diVerences in perfor-mance at 2,800 m, but tremor decreased signiWcantly at3,600 and 4,400 m. Consequently, the literature suggeststhat the critical altitude for changes in higher cognitivefunctioning appears to lie between 4,000 and 5,000 m. Ithas been shown that there is little variation at 3,800 m,but a marked deterioration in cognitive ability occursat 5,000 m (Nelson 1982). The group of Bartholomew(Bartholomew et al. 1999) showed that similar altitudes(3,810–4,572 m) had detrimental eVects on sustained cog-nitive performance such as short term memory. Pilotswere used as subjects in a hypobaric chamber with 30-minhypoxic exposures, and it was found that performances intasks involving high memory loads were signiWcantlydecreased at 3,810 and 4,572 m. Pavlicek et al. (2005)used a simulated altitude of 4,500 m, but did not Wnd anysigniWcant diVerences in word Xuency, word association,or lateralized lexical decision performance. However,subjects were exposed to hypoxia for only 30 min at eachof the following altitudes: 450, 1,500, 3,000 m and 450,1,500, and 4,500 m in a hypobaric chamber. Van der Postet al. (2002) conducted a randomized, single-blind, pla-cebo-controlled three-period cross-over trial on three airgas mixtures via mask breathing. There was a placebogroup with peripheral blood saturation in oxygen (SaO2)>97%, and one with 90 and 80%. Hypoxia exposure was130 min. It was found that at a SaO2 of 80%, most reac-tion times were increased. At this low level of SaO2,adverse events were also seen in the subjects, mainlyheadache and mild drowsiness. In their review paper,Virues-Ortega et al. (2004) found that at an extreme altitudeof 6,000 m, an impairment appears in codiWcation andshort-term memory, as evidenced by an increase in reac-tion time and latency of P300 (event related potential).However, at even lower altitudes, alterations in accuracyand motor speed have been identiWed, as well as deWcits in

123

Eur J Appl Physiol

verbal Xuency, language production, cognitive Xuency,and metamemory.

Although literature on tremor and hypoxia is scarce, itis admitted that hypoxia promotes cortical arousal throughthe activation of chemoreceptors of the ascending reticu-lar activating system (Bonvallet et al. 1959; Hugelin et al.1959). One might therefore expect, given physiologicaltremor’s responsiveness to neurologic changes, thattremor would be altered under hypoxic conditions.Indeed, some early studies have found that tremor wors-ens when the subject becomes hypoxic (Scow et al. 1950)but their analyses were not as sophisticated as those foundin current tremor literature. Only one recent study hasbeen published on the subject (Krause et al. 2000). Thisgroup explored the relationship between hypoxia, hypo-capnia, and the 8–12 Hz component of physiologicaltremor recorded via accelerometry at the index Wnger.Subjects were instructed to hold onto a bar with theirindex Wnger extended against the resistance of a tensiontransducer (pulling downwards) while breathing variouscombinations of O2, CO2, and N2 through a mask. FETO2

and FETCO2 (Fractional End Tidal concentrations in oxy-gen and carbon dioxide, respectively, i.e. fractional con-centration in oxygen and carbon dioxide in the airreleased at the end of the expiration) were measured toinfer the arterial concentrations of O2 and CO2 in eachsubject. Power spectral analysis revealed that hypoxiaampliWed the 8- to 12-Hz component of physiologicaltremor while hypocapnia further augmented the eVect inabout half the subjects.

The main aim of this study was to investigate if pos-tural tremor recorded in the unloaded Wnger respondedas reported by Krause et al. to an acute hypoxic stressand to complement postural tremor with tremor occur-ring during a voluntary movement (kinetic tremor). Thepresent study diVers in several ways from the work ofKrause et al. (2000). First, in this study a minor charge isused on the Wnger during tremor recording (less than2 g) as opposed to the 50 g upward tension used in theirstudy. Second, we not only recorded postural tremor butalso kinetic tremor during a compensatory tracking taskin 23 males (and analyzed a large panel of tremor char-acteristics at diVerent frequencies) while Krause et al.recorded postural tremor in only 12 males. Third, wemeasured SaO2 but not CO2 while they recorded FET O2

and FET CO2 and estimated the corresponding partialpressures. Finally, we measured tremor during and alsoafter the hypoxic stress which allowed a follow-up of theeVect over time. These diVerences provide a more com-prehensive basis of knowledge on the impact of acutehypoxia on physiological tremor and may lead toslightly diVerent results which will complement andreinforce the current literature.

Methods

Apparatus

Altitude stress was induced using two digitally controlled,high-performance air units (operated in parallel) that weredeveloped for simulating high-altitude athletic training(CAT-430, Colorado Altitude Training, Louisville, CO,USA). The air units simulated altitude inside a transparenttent where the subjects were positioned during testing. Thetent was equipped with a vestibule, acting as an “airlock”allowing entry and ingress without substantially aVectingthe atmospheric environment inside the tent. The simulatedaltitude was set to 4,500 m (normobaric condition with a14.8% O2 gas mixture) corresponding to an inspired partialpressure of oxygen of about 10.8 kPa (81 mmHg). This cor-responds to the altitude at which cognitive impairmentsbecome more prominent (Virues-Ortega et al. 2004). Apulse oximeter (Avant 2120 system, Nonin Medical, USA)was used to measure the subjects’ blood oxygen saturationlevels (SaO2) and heart rate (HR). HR was measured asbeats per minute (bpm).

Tremor was recorded with a Class II laser diode (Microlaser sensor LM10, series ARN12821, Matsushita Elec-tronic Work, Ltd., Osaka, Japan) at a sampling rate of1000 Hz. The laser was located 8 cm above a 2-cm-diameterpiece of white cardboard (weighting 1.2 g) that was Wxed onthe tip of the dominant index Wnger (see Fig. 1, left side foran illustration of a postural tremor recording, right side foran illustration of a kinetic tremor recording). The beam wasdirected towards the cardboard allowing for the recording ofvertical Wnger displacements. The laser is an analog outputsensor using an optical triangulation with a §2 cm range ofmeasurement (corresponding to §5 Volts). It has a resolu-tion of §5 �m after Wltering out high frequencies. Positiondata were converted from Volts to millimeters using a cali-bration constant. A Multifunction Data Acquisition systemfrom National Instruments (National Instrument Corpora-tion, Austin, USA) with a LabView platform to program theinterfaces (LabView 8.0) was used to acquire data and con-nect the laser system to a dedicated laptop.

Subjects

Twenty-three healthy male volunteers (aged 23 § 5 years)recruited from The University of Western Ontario (UWO)and aYliate institutions completed this study. Subjects werepre-screened to ensure they were free from movement limi-tation, cardiac or cerebral pacemaker, history of headinjury, or epileptic seizure, that they did not suVer from dia-betes, radial keratotomy, anemia, or any cardiovascular,neurological or psychiatric diseases. It was conWrmed thatthey had no previous altitude training and were not taking

123

Eur J Appl Physiol

medications known to aVect blood oxygenation levels orthrombogenesis. Prior to the experimental sessions, partici-pants were instructed to refrain from alcohol, nicotine, andcaVeine for 24 h. The study was approved by the UWOHealth Sciences Research Ethics Board.

Procedure

Each subject was required to participate in the experimenton two non-consecutive days: on the baseline day theywere tested at sea level (deWned as our ambient level:251 m altitude in our laboratory) only, on the altitude daythey were tested at 4,500 m of simulated altitude and post-altitude (sea level) conditions. Baseline and altitude dayswere given in approximately a 2:1 cross-over design fol-lowing a single blinded procedure (15 subjects had thebaseline day Wrst and 8 subjects had the altitude day Wrst).On the Wrst day, the protocol was explained and the sub-ject provided written consent. He was then randomlyassigned to complete either the baseline, or the altitudeday. For the baseline and post-altitude conditions, the alti-tude simulator was set at sea level (21% of O2 in theambient air). For the altitude condition which lasted 1 haltogether, the altitude simulator was set to 4,500 m. Ineach testing condition, postural and kinetic tremor testingstarted 35 min after the subject entered the tent. After thecompletion of tremor tests, the subject stayed another22 min in the tent to complete other motor tests. In the

altitude condition, all subjects had their SaO2 stabilizedunder 87% within the Wrst 5 min in the tent (measured bythe pulse oximeter). During the altitude-day, between thealtitude and post altitude conditions, the subject wasinstructed to sit in a separated normoxic room for 10 min.This period of time was used to set the tent back to sealevel. The subject then returned to the tent to complete thepost-altitude condition.

Tremor tests

Postural tremor and kinetic tremor tests were recordedfor 1 min each with a 1-min rest period in between. Inboth conditions, the subject was sitting in an experimen-tal armchair and had his dominant arm resting on thearmrest with the palm of his hand supported by a mod-eled support. For the postural tremor test, the subjectwas instructed to relax and to point and maintain for1 min his dominant index Wnger forward without hyper-extending it. The subjects received a visual feedback ofthe position of their Wnger represented by a horizontalline displayed on a LCD screen in front of them: eachsubject had to maintain this line steady, aligned with the“zero” position displayed on the screen. The posturaltremor was recorded at the tip of his index Wnger for theduration of test (see Fig. 1). For the kinetic tremor test,the subject was assumed the exact same position. How-ever, instead of keeping his index Wnger steady during

Fig. 1 Position (top graphs, unWltered data) and velocity (middlegraphs, Wltered data) time series and velocity power spectra (bottomgraphs) for a postural (left) and kinetic (right) tremor recording. Thepower spectra illustrate the frequency content between 2 and 20 Hz forthe postural tremor condition and between 3 and 20 Hz for the kinetictremor condition. The thick gray square wave on the top right graph

represents the position of the target that the subject had to track overtime. Note that due to the tracking task, the amplitude of the kinetictremor time series is about one order of magnitude higher than theamplitude of postural tremor time series, which is reXected as well inthe corresponding power spectra

123

Eur J Appl Physiol

the test, he was asked to track as precisely as possiblewith his index Wnger a target which was moving up anddown. The target was represented by a horizontal linedisplayed on the screen in front of him that couldsequentially assume the “up” and the “down” positions.To fulWll the task, the subject was asked to track this linewith his feedback line (i.e. to try keeping these two linessuperimposed by moving his Wnger up and down accord-ing to the target line position). The target changed posi-tions at a mean frequency of 0.25 Hz, but the timebetween each change was pseudo-randomized (the targetassumed the same position and for the same durationbetween tests, but in a diVerent pre-programmed order).The amplitude needed at the Wngertip to follow the linewas 2 cm (see Fig. 1). Prior to starting the experiment,the subject was given a period of familiarization for eachtest.

Data analysis

Raw tremor data were converted from volts to millimetersand analyzed using Matlab (The MathWorks, Natick, MA,USA). Fingertip position time series were low-pass andhigh-pass Wltered in the frequency domain keeping frequen-cies between 2–20 Hz for postural tremor and 3–20 Hz forkinetic tremor (Fast Fourier–inverse Fast Fourier Trans-forms). Velocity time series (diVerentiated position timeseries) were used in the frequency domain analyses becausethey enhance the proportion of power in frequency range ofinterest (6–12 Hz). Indeed, while the proportion of lowfrequencies is prominent in postural tremor power spectra,the diVerentiation process [due to the simple frequency-dependent relationship linking the power spectrum of atime-dependent variable to its derivative: power invelocity = power in displacement £ (2 £ � £ frequency)2,see for example Norman et al. 1999] has the advantage ofamplifying the proportion of power at the frequencies cor-responding to physiological tremor. Postural tremor analy-ses were conducted on only 20 subjects since the data ofthree participants were partially lost because their Wngerdrifted out of the laser beam during recordings. Kinetictremor analyses were conducted on the 23 subjects. Groupaveraged power spectra were computed in the postural(Fig. 2, n = 20) and kinetic (Fig. 3, n = 23) tremor condi-tions. Interestingly, an increase of the frequency contentbetween 6 and 12 Hz was noticed in the altitude conditioncompared with the sea level conditions (baseline and post-altitude) for both postural and kinetic tremor tests, with amore pronounced eVect in the postural tremor condition. Itwas therefore decided to select the 6–12 Hz frequencyrange as a frequency range to focus on in our data analysis(instead of the 8–12 Hz range previously investigated in theliterature).

Validated characteristics (Beuter and Edwards 1999;Edwards and Beuter 2000) were then calculated for eachdata set (amplitude, drift, median frequency, amplitudeXuctuations, frequency concentration, power in 6–12 Hzrange, and proportion of power in the following ranges:2–4, 4–6, 6–12, 12–20 Hz, deWned in the appendix). Twoadditional characteristics were added for kinetic tremoranalysis: mean tracking error and delay for movement initi-ation. Within-subjects ANOVAs were conducted usingSPSS (SPSS v16.0, USA) on each characteristic. Probabil-ity values were corrected for lack of sphericity using theGreenhouse-Geisser epsilon. A Bonferroni correction formultiple comparisons was used to safeguard against type Ierrors (false positives) in pair-wise comparisons. It was alsochosen to complete the statistical results with a measure ofthe eVect size: the partial eta squared (Eta2) which representthe proportion of the eVect which can be attributed to thediVerent conditions.

Fig. 2 20 subjects’ averaged power spectra for postural tremor in eachexperimental condition. This graph suggests that hypoxia induces anincrease of tremor power in the 6–12 Hz range: the power is increasedin the altitude condition and though decreasing, it does not return to itsbaseline level in the post-altitude condition

Frequency (Hz)

Pow

er(m

m2 .

s-2 .

Hz-

1 )

0

0.2

0.4

0.6

0.8

1

1.2

1.4

2 4 6 8 10 12 14 16 18 20

BaselineAltitude Post-altitude

Fig. 3 23 subjects’ averaged power spectra for kinetic tremor in eachexperimental condition. The proportion of low frequencies is highcompared with postural tremor due to the rapid tracking movements(illustrated in Fig. 1). This graph suggests that hypoxia induces anincrease of tremor power in the 6–12 Hz range. Again, the power isincreased in the altitude condition and though decreasing, it does notreturn to its baseline level in the post-altitude condition

Pow

er(m

m2 .

s-2 .

Hz-

1 )

Frequency (Hz)

0

10

20

30

40

50

60

3 5 7 9 11 13 15 17 19

BaselineAltitude Post-altitude

123

Eur J Appl Physiol

Results

It was of primary importance in this experiment to conWrmwhether the hypoxic environment actually induced hypox-emia in the subjects (i.e. reduced SaO2). All subjects had anormal SaO2 in the baseline condition (above 96% SaO2).After several minutes in the altitude condition, their SaO2

had stabilized below our predetermined threshold of 87% toan average value of 80%. Normalcy was achieved again inthe post-altitude condition for all subjects. DiVerences weresigniWcant for the main eVect (F = 962.15, P < 0.001,Eta2 = 0.98; Fig. 4a) and the decrease of SaO2 during thealtitude condition was conWrmed by pair-wise comparisonswith both pre- and post-altitude conditions (P < 0.001). Thesubjects partially compensated for the oxygen deWciency inthe altitude condition as evidenced by their increased HRfrom an average of 65.5 to 75.5 bpm (F = 25.81, P < 0.001,Eta2 = 0.56; Fig. 4b). Again, Bonferroni pair-wise compari-sons conWrmed the increased HR during the altitude condi-tion in comparison with the baseline and post-altitudeconditions (both with P < 0.001).

Within subject ANOVAs were then conducted on com-puted tremor characteristics and Bonferroni adjusted pair-wise comparisons were run when signiWcant main eVectswere found. Since they would give redundant information,we removed characteristics that were signiWcantly corre-lated between each other (Pearson correlations, thresholdWxed at r = 0.8 and P < 0.05). Therefore, only amplitude,drift, median frequency, mean power in 6–12 Hz range,and proportion of power in the following ranges: 2–4, 4–6,6–12 Hz were kept for postural tremor statistical analysis.For kinetic tremor analysis, amplitude, delay, error, medianfrequency, proportion of power in the 6–12 Hz range andpower in the 6–12 Hz range were kept for statistical analysis.

Postural tremor

Main eVects regarding postural tremor are reported inTable 1 and Bonferroni adjusted pair-wise comparisonswere conducted on characteristics showing a signiWcantmain eVect. Subjects exhibited a higher amplitude (F = 3.89,P < 0.05, Eta2 = 0.17) and mean power in the 6–12 Hz range

(F = 6.14, P < 0.01, Eta2 = 0.24; illustrated in Fig. 5) and asmaller proportion of power in the 2–4 Hz range (F = 4.44,P < 0.05, Eta2 = 0.19) in the altitude than in the baselineconditions. Bonferroni adjusted pair-wise comparisons con-Wrmed that amplitude and power in the mean 6–12 Hz fre-quency band were signiWcantly higher in the altitude than inthe post-altitude condition (P < 0.05 for both comparisons)and marginally higher in the altitude than in the baselinecondition (P = 0.1 and P = 0.06, respectively). The propor-tion of power in the 2–4 Hz range was signiWcantly smallerin the post-altitude than in the altitude condition (P < 0.01).

Kinetic tremor

Kinetic tremor was analyzed using the same statisticalprocedure. Thus, main eVects are reported, and Bonfer-roni adjusted pair-wise comparisons were conducted on

Fig. 4 a SaO2 was signiWcantly lower in the altitude than in the baseline and post-altitude conditions. Error bars represent the standard deviation (variability is too small to display standard error of the mean (SEM)). b HR was signiWcantly higher in the altitude than in the baseline and post-altitude conditions. Error bars represent SEM 70

80

90

100

Baseline Altitude Post-Altitude

SaO

2 (%

)

60

65

70

75

80

Baseline Altitude Post-Altitude

HR

(b

pm

)

a b

Table 1 Main eVects resulting from the within-subjects ANOVAs forpostural and kinetic tremor data

SigniWcant eVects (P < 0.05) are italicized and indicated with anasterisk (*)

EVect size (Eta2) and power are also reported

When signiWcant, the direction of the eVect is indicated by an arrow(% if increased and & if decreased during hypoxia)

Tremor Characteristics Direction of changes

F P Eta2 Power

Postural Amplitude % 3.89 0.05* 0.17 0.56

Drift n/a 0.25 0.72 0.01 0.08

MedFreq n/a 2.71 0.09 0.12 0.46

Pow 2–4 & 4.44 0.02* 0.19 0.70

Pow 4–6 n/a 1.97 0.16 0.09 0.34

Pow 6–12 n/a 2.34 0.13 0.11 0.37

Meanpow 6–12 % 6.14 0.01* 0.24 0.76

Pow12-20 n/a 0.09 0.90 0.00 0.06

Kinetic Amplitude n/a 1.49 0.24 0.06 0.29

Delay n/a 1.76 0.19 0.07 0.33

Error n/a 2.33 0.13 0.10 0.36

MedFreq % 6.87 0.01* 0.24 0.84

Pow 6–12 % 6.43 0.00* 0.23 0.87

Meanpow 6–12 % 3.77 0.03* 0.15 0.63

123

Eur J Appl Physiol

characteristics showing a signiWcant main eVect (seeTable 1). Once again, a main eVect was found on the meanpower in the 6–12 Hz range showing higher values in thealtitude condition (F = 6.43, P < 0.05, Eta2 = 0.15; Fig. 6).The proportion of power in the 6–12 Hz range and themedian frequency also exhibited higher values in thealtitude condition (F = 6.97, P < 0.01, Eta2 = 0.24 andF = 6.87, P < 0.01, Eta2 = 0.24, respectively). No signiW-cant eVects were observed on the tracking performancecharacteristics, i.e., delay (F = 1.76, P > 0.05, Eta2 = 0.07)and error (F = 2.33, P > 0.05, Eta2 = 0.10). Bonferroniadjusted pair-wise comparisons indicated that the meanpower in the 6–12 Hz frequency range was marginallyhigher in the altitude than in the post-altitude condition(P = 0.06) and the baseline condition (P = 0.08); that pro-portion of power in the 6–12 Hz range was signiWcantlyhigher in the altitude condition than in the two other condi-tions (P < 0.05); and that median frequency was signiW-

cantly higher in the altitude than in the post-altitudecondition (P < 0.05) and marginally higher compared withthe baseline condition (P = 0.07).

Discussion

In the present study, the hypothesis suggesting an increaseof tremor intensity in the 6- to 12-Hz frequency range withaltitude has been conWrmed statistically in a sample of 23subjects. This is in line with the unique study found in theliterature reporting an activation of tremor in the 8–12 Hzfrequency range with acute hypoxia (Krause et al. 2000).This frequency range is believed to be associated with acentral nervous system autonomous oscillator. This sug-gests that acute hypoxia may interact with the physiologicalrhythm of a central pacemaker involved in tremor genera-tion (Elble 1986; Goodman and Kelso 1983; Llinas andVolkind 1973; Travis 1929). One of the main interests inadding kinetic tremor was to access tremor produced duringa goal-directed movement known to involve slightly diVer-ent control mechanisms than a postural task (Gross et al.2002). However, it has been argued that the central pulsa-tile control occurring in this frequency range during move-ment is rather a biphasic motor output reXecting thedescending motor command than tremor per se (Vallbo andWessberg 1993; Wessberg and Vallbo 1996). Interestingly,in the present study, the increase in tremor intensity in the6- to 12-Hz frequency range observed in postural tremor isalso observed in the kinetic tremor data, which reinforcesthe reliability of this eVect and supports an eVect at the cen-tral level, but this perspective is weakened by the absenceof eVect on tracking the performance (which is centrallydetermined).

However, our approach diVered in several ways fromthat used by Krause et al. (2000). The Wrst diVerence con-cerned the way tremor was measured: we measured dis-placement rather than acceleration of the index Wnger,which may have led to slightly diVerent results since thetwo methods do not carry exactly the same information(Norman et al. 1999). In addition, we measured two typesof tremor instead of one to get a more comprehensive viewof the eVect of hypoxia on motor and cognitive abilities.Another diVerence in our study involved the absence ofloading applied to the index Wnger. Krause et al. had thesubjects extend their Wnger and apply a force of 50 gagainst a spring. This may have altered the characteristicsof the tremor making some components more prominent,and others less so (Arihara and Sakamoto 1999; Lippold1981). In addition, we did not induce as much hypoxia asdid Krause et al. Indeed, they set PETO2 = 45 mmHg (EndTidal partial Pressure in oxygen) implying that PaO2 wasabout 33 mmHg (Cotes et al. 2006), and therefore SaO2 is

Fig. 5 Power in the 6- to 12-Hz range for postural tremor in baseline,altitude and post-altitude conditions. Error is reported as the SEM. Asillustrated in Fig. 2, the power is increased in the altitude condition andthough decreasing, it does not return to its baseline level in the post-altitude condition (not statistically signiWcant)

Mea

n P

ower

6-1

2Hz

(mm

2 .s-

2 .H

z-1 )

Baseline Altitude Post-Altitude

0

0.2

0.4

0.6

0.8

1

1.2

Fig. 6 Power in the 6- to 12-Hz range for kinetic tremor in baseline,altitude, and post-altitude conditions. Error is reported as the SEM. Asillustrated in Fig. 3, the power is increased in the altitude condition andthough decreasing, it does not return to its baseline level in the post-altitude condition (not statistically signiWcant)

0

10

20

30

Mea

n P

ower

6-1

2Hz

(mm

2 .s-

2 .H

z-1 )

Baseline Altitude Post-Altitude

123

Eur J Appl Physiol

approximately 62% (Ganong 2001) whereas our subjects’average SaO2 was about 80%. This was done to help assessthe sensitivity of tremor to mild or moderate changes inoxygen status. Furthermore, although we have kept thecharacteristics of our subjects similar to Krause et al. wehave increased the sample size from 12 to 23. Finally interms of analysis, Krause et al. considered the power of the8- to 12-Hz frequency component of physiological tremorwhile our data tend to show that frequencies within a largerfrequency band (6–12 Hz) are also enhanced by an acutehypoxia. Additional tremor characteristics known to be sen-sitive to environmental perturbations have also been inves-tigated in the present study (Beuter and Edwards 1999;Edwards and Beuter 2000; Legros and Beuter 2005).

As indicated above, there remains uncertainty regardingthe origin of tremor. However, we note that the 8- to 12-Hzcomponent of physiological tremor is usually reported asbeing mainly caused by a central pacemaker. That is, it issuspected that a population of neurons Wres rhythmically insuch a way that the signals generated ultimately lead totremor in the extremities. Numerous suggestions regardingthe location of the central oscillator have been made, andsome possibilities include the thalamus, the inferior olive,and the basal ganglia (see Deushl et al. 2001 for a review).It does seem likely, however, that the oscillator involvestranscortical communication (Koster et al. 1998). A fewstudies have capitalized on magnetoencephalographyrecordings and advanced directed coherence analysis tech-niques attempting to localize the central oscillator. Onegroup found that the 8- to 12-Hz component of tremorduring slow voluntary movement (Vallbo and Wessberg1993) originates at the premotor cortex and is modulatedby a cerebello-thalamo-cortical network (Gross et al.2002). Another group reported similar results when askinghealthy volunteers to emulate Parkinsonian tremor (Polloket al. 2004). Since the brain is exquisitely sensitive tohypoxia, it is conceivable that the network producingtremor is disrupted in some way by hypoxic conditions.

What is not clear, however, is whether hypoxia actsdirectly on tremor via the somatomotor system or indirectlythrough other physiological systems (e.g. respiratory andcardiovascular systems). The link between hypoxia, cate-cholamine, and tremor has been investigated using a varietyof research approaches including cellular and animal stud-ies (Chalmers et al. 1966; Hiramatsu et al. 1971). Resultsfrom these studies are converging toward a commondescription regarding the cascade of events in process.Hypoxic stress induces a release of catecholamine (sympat-homimetic “Wght-or-Xight” hormones that are released bythe adrenal glands in response to stress) which in turninduces an increase in HR and an increase in oxygen uptakeeYciency (Peterson et al. 1999). Though tremor is usuallynot aVected by local sympathetic nervous system activity,

an adrenaline increase leads to an increase of muscularoscillations in the 10-Hz range as a consequence of adecrease in tetanic tension (Marsden and Meadows 1970).For instance, isoproterenol can mimic the action of cate-cholamine, which is thought to be mediated via O2-recep-tors on extrafusal muscle Wbers and increases physiologicaltremor amplitude (Pickles et al. 1981). It has also beenreported that oral bronchodilator drugs increase physiologi-cal tremor and HR (Watson and Richens 1974).

Since both tremor amplitude and HR are concomitantlyaVected by catecholamine, one might ask the question ifin fact the increase in tremor amplitude is not mainly themechanical consequence of the increase in HR. This is,however, an unlikely scenario since it is reported thatthough ballistocardiographic impulses evidently contrib-ute to the generation of resting tremor, they have onlyminor inXuence on physiological tremor recorded in theoutstretched Wnger (Marsden et al. 1969). Moreover,according to these authors adrenaline induced tremor maynot be due to the associated cardiac modulations. It seemstherefore reasonable to advance that at least part of theincrease of tremor amplitude in the 6–12 Hz rangereported here (as consequence of acute hypoxia) is dueto an increase of circulating adrenaline release. Peripheral�-adrenoreceptors would then stimulate peripheral vaso-dilation reducing in turn the visco-elasticity properties ofthe muscle and thus its Wltering properties (Fellows et al.1986).

Interestingly, the eVect of hypoxia also bears some simi-larities with that of fatigue. According to the literature, theeVect of fatigue on postural tremor in neurologically nor-mal subjects is conWned to an increase in the peak power ofthe neurally generated 8- to 12-Hz tremor component(Lippold 1981; Morrison et al. 2005; Vaillancourt and Newell2000). It has for example been reported that exhaustion ischaracterized by a concomitant increase of tremor ampli-tude and of HR (Vetter and Horvath 1961). However, thiswork also shows that tremor amplitude is back to its base-line values 7 min after the cessation of the work whereasthe HR was still elevated. Our current results show thatwhile tremor is not completely reaching its baseline level1 h post exposure, HR is. Again, this tends to support theidea that the increase in tremor size is not a direct conse-quence of an increase in HR. Our results, therefore suggestthat hypoxia and fatigue may share some common mecha-nisms; nevertheless, diVerent processes seem to be involvedpost-stress.

Although the most consistent change was located in the6 to 12-Hz range for both postural and kinetic tremor, itwas not the only characteristic that achieved statisticalsigniWcance. Indeed, the amplitude of postural tremorwas increased and its proportion of low-frequency range(2–4 Hz) was decreased with altitude. However, these two

123

Eur J Appl Physiol

observations are in fact direct consequences of the increaseof the power of tremor in the 6–12 Hz range: this led to anincrease of the overall tremor amplitude and to a decreaseof the proportion of power in the other frequency ranges,including the 2–4 Hz range (illustrated by Fig. 2). Regard-ing kinetic tremor, the median frequency slid to higher fre-quencies, but again it was a direct consequence of theincrease of tremor power in the 6–12 Hz frequency range(illustrated by Fig. 3).

Often, in abnormal circumstances, the power spectrumnarrows and frequencies shift to lower values (Beuter andEdwards 1999). No pathology was expected to be inducedby this study, and indeed, none apparently was. The local-ized increase in power at 6–12 Hz was not large enough tobring the increase of the average amplitude of the wholespectrum to statistical signiWcance. Finally, neither of thecharacteristics of performance for the tracking task (i.e.delay and error) demonstrated an eVect with respect to oxy-gen availability. Thus, the level of hypoxia induced in thisstudy may not have been long enough to alter reaction timeor voluntary movements as reported elsewhere (Virues-Ortega et al. 2004).

In conclusion, this study was designed to describe theeVect of hypoxia on physiological tremor both in pos-tural and kinetic conditions. It is shown that the tech-nique of evaluation and analysis of human physiologicaltremor used here is adapted to detect subtle altitude/hypoxia-induced changes. However, an experiment uti-lizing a brain activity monitoring technique such asfMRI, or EEG, while simultaneously collecting tremordata under hypoxic conditions, is warranted to furtherexplore the relationship with central processes. The mostimportant eVect observed was an increase in the spectralcontent from 6 to 12 Hz of physiological tremor (pos-tural and kinetic). This eVect has been observed previ-ously using a diVerent protocol indicating that the eVectis robust and reproducible. Though the speciWc mecha-nisms involved are still not known, and beyond the con-tribution of a central pacemaker that remains to beelucidated, the main mechanism in process here seemsto be associated with an hypoxia-induced catecholaminerelease which would in turn modulate the stiVness prop-erties of the muscles and lead to an increase of tremoramplitude in the 6–12 Hz range. Although furtherresearch needs to be conducted to elucidate the preciserelationship between acute hypoxia and physiologicaltremor, the present work has begun to explore the natureof the connection between the two.

Acknowledgments We would like to thank Mr. John Patrick for hishelp in conducting the experiment, Mr. Michael Corbacio for hisassistance with LabView and Matlab programming, and Mr. LynnKeenliside for his technical contributions and expertise. This work wasfunded in-part by Defence Research and Development Canada, the

Canadian Institutes of Health Research, the Natural Sciences andEngineering Research Council of Canada, the Ontario Research Fundand the Canadian Foundation for Innovation.

Appendix

Characteristics computed on tremor time series1

Amplitude Root Mean Square of the Wlteredposition time series centered ontheir mean. Computed on positiondata Wltered between 2 (3 forkinetic tremor) and 20 Hz. Largervalues correspond to worse perfor-mance

Drift QuantiWes the amplitude of thedrift of the Wnger (slow move-ments with frequencies <0.1 Hz).Computed on position data only inthe postural tremor condition.Larger values correspond to worseperformance

Median frequency Computed on tremor velocitypower spectrum between 2 (3 forkinetic tremor) and 20 Hz. Deter-mines the value at which 50% ofthe power is below this frequencyand 50% is above. Smaller valuesusually correspond to worse per-formance

Amplitude Xuctuations Characterizes the variability oftremor amplitude over time. It isthe standard deviation of the enve-lope around tremor oscillations.Computed on position data. Largervalues correspond to worse perfor-mance

Frequency Computed on tremor velocity powerconcentration spectrum between 2 and 20 Hz.

QuantiWes the degree of organiza-tion of tremor by computing thewidth of the interval containing68% of the power of the spectrumbetween 2 (3 for kinetic tremor)

1 Please note that the description of the quantiWcation of the changes interms of ‘better or worse performance’ regarding tremor characteristicsis stated here as an indication only. Though they could be adapted indescribing pathological tremor such characterizations would be toospeculative to be used to fully describe tremor staying within the phys-iological range

123

Eur J Appl Physiol

and 20 Hz. Smaller values corre-spond to worse performance

Proportional power Computed on tremor velocity powerin 2–4 Hz range spectrum between 2 and 20 Hz.

Proportion of the power containedin this range compared with thespectrum between 2 (3 for kinetictremor) and 20 Hz. Larger valuesusually correspond to worse per-formance

Proportional power Computed on tremor velocity powerin 4–6 Hz range spectrum between 2 and 20 Hz.

Proportion of the power containedin this range compared with thespectrum between 2 (3 for kinetictremor) and 20 Hz. Larger valuesusually correspond to worse per-formance

Proportional power Computed on tremor velocity powerin 6–12 Hz range spectrum between 2 (3 for kinetic

tremor) and 20 Hz. Proportion ofthe power contained in this rangecompared with the spectrumbetween 2 (3 for kinetic tremor)and 20 Hz. Smaller values usuallycorrespond to worse performance

Mean Power in Computed on tremor velocity powerthe 6–12 Hz range spectrum between 2 (3 for kinetic

tremor) and 20 Hz. Average of thepower contained in this frequencyrange. It is the range containingthe physiological tremor compo-nents. Computed on velocity dataWltered between 2 (3 for kinetictremor) and 20 Hz. Larger valuescorrespond to worse performance

Proportional power Computed on tremor velocity powerin 12–20 Hz range spectrum between 2 and 20 Hz.

Proportion of the power containedin this range compared with thespectrum between 2 (3 for kinetictremor) and 20 Hz. Larger valuesusually correspond to worse per-formance

Mean tracking error Mean of the absolute diVerencebetween the Wnger position and thereference position during a record-ing. Computed on position dataWltered between 3 and 20 Hz.Larger values correspond to worseperformance

Delay Average delay between the dis-placement of the target on the

screen and the displacement of thesubject’s index Wnger. Computedon position data Wltered between 3and 20 Hz. Larger values corre-spond to worse performance.

References

Arihara M, Sakamoto K (1999) Contribution of motor unit activityenhanced by acute fatigue to physiological tremor of Wnger.Electromyogr Clin Neurophysiol 39:235–247

Bartholomew CJ, Jensen W, Petros TV et al (1999) The eVect of mod-erate levels of simulated altitude on sustained cognitive perfor-mance. Int J Aviat Psychol 9:351–359

Beuter A, Edwards R (1998) Tremor in Cree subjects exposedto methylmercury: a preliminary study. Neurotoxicol Teratol20:581–589

Beuter A, Edwards R (1999) Using frequency domain characteristicsto discriminate physiologic and parkinsonian tremors. J Clin Neu-rophysiol 16:484–494

Beuter A, Edwards R (2002) Characterization and discrimination ofkinetic tremor in Parkinson’s disease. Rev Neurol (Paris)158:338–340

Beuter A, Edwards R, Lamoureux D (2000) Neuromotor proWles:what are they and what can we learn from them? Brain Cogn43:39–44

Bonvallet M, Audisio M, Hugelin A (1959) Reticular and cortical fac-tors in the changes in motoneuron excitability seen during acutehypoxia. J Physiol (Paris) 51:411–412

Britton TC, Gresty MA (1992) Mechanisms for essential tremor. Lan-cet 340:610

Britton TC, Thompson PD (1995) Primary orthostatic tremor. BMJ310:143–144

Chalmers JP, Korner PI, White SW (1966) The control of the circula-tion in skeletal muscle during arterial hypoxia in the rabbit.J Physiol 184:698–716

Cotes JE, Chinn DJ, Miller MR (2006) Lung function: physiology,measurement and application in medicine. Blackwell, Victoria

Deuschl G, Bain P, Brin M (1998) Consensus statement of the move-ment disorder society on tremor. Ad Hoc ScientiWc Committee.Mov Disord 13(Suppl 3):2–23

Deuschl G, Raethjen J, Lindemann M et al (2001) The pathophysiol-ogy of tremor. Muscle Nerve 24:716–735

Edwards R, Beuter A (2000) Using time domain characteristics todiscriminate physiologic and parkinsonian tremors. J Clin Neuro-physiol 17:87–100

Elble RJ (1986) Physiologic and essential tremor. Neurology 36:225–231

Elble RJ (1996) Central mechanisms of tremor. J Clin Neurophysiol13:133–144

Elble RJ, Koller WC (1990) Tremor. The John Hopkins Universitypress, London

Fellows IW, Macdonald IA, Wharrad HJ et al (1986) Low plasmaconcentrations of adrenaline and physiological tremor in man.J Neurol Neurosurg Psychiatry 49:396–399

Ganong WF (2001) Review of medical physiology. McGraw-Hill,New York

Goodman D, Kelso JA (1983) Exploring the functional signiWcance ofphysiological tremor: a biospectroscopic approach. Exp BrainRes 49:419–431

Gross J, Timmermann L, Kujala J et al (2002) The neural basis of inter-mittent motor control in humans. Proc Natl Acad Sci USA99:2299–2302

123

Eur J Appl Physiol

Hallett M (1991) ClassiWcation and treatment of tremor. JAMA266:1115–1117

Hiramatsu Y, Takahashi H, Izumi A (1971) EVect of inosine on adren-aline and vasopressin induced myocardial hypoxia. Jpn J Pharma-col 21:355–360

Hugelin A, Bonvallet M, Dell P (1959) Reticular & cortical activationof chemoreceptor origin during hypoxia. Electroencephalogr ClinNeurophysiol Suppl 11:325–340

Koller W, Cone S, Herbster G (1987) CaVeine and tremor. Neurology37:169–172

Koster B, Lauk M, Timmer J et al (1998) Central mechanisms inhuman enhanced physiological tremor. Neurosci Lett 241:135–138

Krause WL, Leiter JC, Tenney MS et al (2000) Acute hypoxia acti-vates human 8–12 Hz physiological tremor. Respir Physiol123:131–141

Lakie M, Frymann K, Villagra F et al (1994a) The eVect of alcohol onphysiological tremor. Exp Physiol 79:273–276

Lakie M, Walsh EG, Arblaster LA et al (1994b) Limb temperature andhuman tremors. J Neurol Neurosurg Psychiatry 57:35–42

Legros A, Beuter A (2005) EVect of a low intensity magnetic Weld onhuman motor behavior. Bioelectromagnetics 26:657–669

Legros A, Gaillot P, Beuter A (2006) Transient eVect of low-intensitymagnetic Weld on human motor control. Med Eng Phys 28:827–836

Li XY, Wu XY, Fu C et al (2000) EVects of acute mild and moderatehypoxia on human mood state. Space Med Med Eng (Beijing)13:1–5

Lippold OC (1970) Oscillation in the stretch reXex arc and the originof the rhythmical, 8–12 C-S component of physiological tremor.J Physiol 206:359–382

Lippold O (1981) The tremor in fatigue. Ciba Found Symp 82:234–248Llinas R, Volkind RA (1973) The olivo-cerebellar system: functional

properties as revealed by harmaline-induced tremor. Exp BrainRes 18:69–87

Marsden CD, Meadows JC (1970) The eVect of adrenaline on thecontraction of human muscle. J Physiol 207:429–448

Marsden CD, Meadows JC, Lange GW et al (1967) EVect of deaVer-entation on human physiological tremor. Lancet 2:700–702

Marsden CD, Meadows JC, Lange GW et al (1969) The role of theballistocardiac impulse in the genesis of physiological tremor.Brain 92:647–662

Mazzocchio R, Gelli F, Del Santo F et al (2008) EVects of posture-related changes in motor cortical output on central oscillatoryactivity of pathological origin in humans. Brain Res 1223:65–72

McAuley JH, Marsden CD (2000) Physiological and pathologicaltremors and rhythmic central motor control. Brain 123(Pt8):1545–1567

Morrison S, Kavanagh J, Obst SJ et al (2005) The eVects of unilateralmuscle fatigue on bilateral physiological tremor. Exp Brain Res167:609–621

Nelson M (1982) Psychological testing at high altitudes. Aviat SpaceEnviron Med 53:122–126

Norman KE, Edwards R, Beuter A (1999) The measurement of tremorusing a velocity transducer: comparison to simultaneous record-ings using transducers of displacement, acceleration and muscleactivity. J Neurosci Methods 92:41–54

Pavlicek V, Schirlo C, Nebel A et al (2005) Cognitive and emotionalprocessing at high altitude. Aviat Space Environ Med 76:28–33

Peterson BS, Skudlarski P, Gatenby JC et al (1999) An fMRI study ofStroop word-color interference: evidence for cingulate subregionssubserving multiple distributed attentional systems. Biol Psychi-atry 45:1237–1258

Pickles H, Perucca E, Fish A et al (1981) Propranolol and sotalol asantagonists of isoproterenol-enhanced physiologic tremor. ClinPharmacol Ther 30:303–310

Pollok B, Gross J, Dirks M et al (2004) The cerebral oscillatorynetwork of voluntary tremor. J Physiol 554:871–878

Scow J, Krasno LR, Ivy AC (1950) The immediate and accumulativeeVect on psychomotor performance of exposure to hypoxia, highaltitude and hyper-ventilation. J Aviat Med 21:79–81

Stein RB, Lee RG, Nichols TR (1978) ModiWcations of ongoing trem-ors and locomotion by sensory feedback. Electroencephalogr ClinNeurophysiol, suppl 5, pp 12–519

Stiles RN, Randall JE (1967) Mechanical factors in human tremorfrequency. J Appl Physiol 23:324–330

Travis LE (1929) The relation of voluntary movement to tremors.J Exp Psychol 12:515–524

Vaillancourt DE, Newell KM (2000) Amplitude changes in the 8–12,20–25, and 40 Hz oscillations in Wnger tremor. Clin Neurophysiol111:1792–1801

Vallbo AB, Wessberg J (1993) Organization of motor output in slowWnger movements in man. J Physiol 469:673–691

van der Post J, Noordzij LA, de Kam ML et al (2002) Evaluation of testsof central nervous system performance after hypoxemia for a modelfor cognitive impairment. J Psychopharmacol 16:337–343

Vetter K, Horvath SM (1961) Analysis of physiological tremor duringrest and exhaustion. J Appl Physiol 16:994–996

Virues-Ortega J, Buela-Casal G, Garrido E et al (2004) Neuropsycho-logical functioning associated with high-altitude exposure.Neuropsychol Rev 14:197–224

Virues-Ortega J, Garrido E, Javierre C et al (2006) Human behaviourand development under high-altitude conditions. Dev Sci 9:400–410

Wachs H, Boshes B (1966) Tremor studies in normals and parkinson-ism. Arch Neurol pp 66–82

Watson JM, Richens A (1974) The eVects of salbutamol and terbuta-line on physiological tremor, bronchial tone and heart rate. BrJ Pharmacol 1:223–227

Wessberg J, Vallbo AB (1996) Pulsatile motor output in human Wngermovements is not dependent on the stretch reXex. J Physiol 493(Pt3):895–908

123