The effect of therapeutic massage on H-reflex amplitude in persons with a spinal cord injury.

In a series of studies,[1-4] the effects of petrissage (a massage technique) on H-reflex amplitude in subjects without neurological impairment have been systematically studied. The findings generally reveal a transitory decrease in the H-reflex amplitude (an indirect measure of spinal motoneuron excitability and a measure of changes along the spinal reflex pathway) during massage.[1] This effect was specific to the muscle being massaged[3] and lasted only for the duration of the massage, with a return to premassage H-reflex values immediately upon cessation of the massage.[2] The decrease in the H-reflex amplitude was also found to vary directly with the amount of pressure applied.[4] Furthermore, these studies have attempted to identify the mechanisms involved in the recorded response. The contribution of antagonist reciprocal inhibition, changes in nerve conduction velocity, and changes in skin temperature[2] have all been excluded as possible mechanisms in the recorded decrease in the H-reflex. A recent study[5] examining the neuromuscular response to effleurage (another massage technique) has demonstrated that this manipulation also produced a decrease in the H-reflex amplitude during massage.

Having established that massage is effective in reducing the H-reflex amplitude in subjects without neurological impairment,[1-5] the logical research progression was to investigate the efficacy of massage in reducing the level of motoneuron excitability (H-reflex amplitude) in a neurologically impaired population. A systematic review of the literature located only one study that investigated the therapeutic effects of massage in a neurologically impaired population. Larue[6] examined the effects of electromyographic (EMG) biofeedback, an audio-relaxation tape, and massage on the reduction of tendon reflex activity in a group of subjects with hemiplegia and reported that postmassage reflex activity was decreased. (Latency and amplitude of the tendon tap reflex were quantified from the EMG signal.) Larue's experimental protocol, however, was designed primarily to compare the effects of the EMG biofeedback and the audio-relaxation tape. Massage was not expected to play a role in the decrease in hyperreflexia, and the protocol was therefore not designed specifically to investigate the effects of massage. Furthermore, the subjects in die study were required to change positions (from a sitting position to a prone position) during the course of the data collection. These postural changes may have influenced the results. Due to these methodological constraints, the reported decrease in hyperreflexia following massage must be cautiously interpreted. Consequently, the confirmation of the therapeutic effects of massage in a neurologically impaired sample still awaits scientific investigation.

Among the many neurologically impaired populations requiring treatment by physical therapists, patients with a spinal cord injury (SCI) represent a substantial proportion and generally exhibit a heightened state of motoneuron excitability in comparison with neurologically healthy subjects.[7-9] Furthermore, persons with SCI are usually young, and this factor is advantageous when conducting H-reflex studies, as it results in easily discernable H-reflexes.[10] Thus, persons with SCI provide a useful group with which to examine the effects of massage using the H-reflex as the dependant variable. In addition, changes in the amplitude of the H-reflex following SCI have been well documented.[7-9,11] Following an initial decrease in the H-reflex peak-to-peak amplitude during the phase of spinal shock, the H-reflex evolves to a heightened state of excitability within 3 to 6 months,[7-9,11] which then persists, This increased amplitude in the evoked H-reflex may be indicative of an increased state of excitability of the spinal motoneuron pool.[8] Although this state of motoneuron hyperexcitability also exists in the presence of hyperreflexia,[12-15] no clear correlation between the H-reflex amplitude and the degree of hyperreflexia has yet been established.[16-18]

Prior to the initiation of the study of the effects of massage in persons with SCI, one important methodological constraint needed to be addressed. In previous studies,[1-5] the subjects were tested while lying in a prone position. The prone position, however, can cause respiratory difficulties in persons with SCI[19] and is often difficult to assume on a narrow treatment table. These potential problems associated with the prone position rendered it inappropriate for data collection in our sample. The supine position, a variation of the standard semireclined sitting position recommended for H-reflex recordings,[20] has been suggested in the literature as an adequate subject position for H-reflex recordings.[21] However, as both neuromuscular status and the H-reflex amplitude can be influenced by such factors as body and limb position, temperature, and emotional stress,[21,22] we considered it important to verify whether massage, applied to subjects in the supine testing position, would produce effects similar to those recorded in the prone testing position.[1-5]

This research was thus divided into two separate, but interrelated, studies. The purpose of study 1 was to investigate whether the recorded response (H-reflex amplitude) to massage with the subjects positioned supine was similar to that recorded in previous studies in which the subjects were tested in the prone position. It was expected that the H-reflex measurements recorded during massage in the supine position would follow the same trend as those previously recorded in the prone position (ie, a decrease in H-reflex amplitude during the massage). Study 2 was designed to investigate the effect of therapeutic massage on the H-reflex in a sample of persons with SCI. Based on our results with neurologically healthy individuals,[1-5] we hypothesized that therapeutic massage would cause a reduction in the state of excitability of the motoneuron pool as manifested by a decrease in the H-reflex peak-to-peak amplitude during the massage.

Methods

The two studies reported in this article shared many methodological features and therefore will be presented together. When there are obvious differences between the two studies, the information will be reported separately.

Subjects

Study 1. Seven neurologically healthy subjects (5 women, 2 men), with a mean age of 24.6 years (SD=2.6, range=20-28), volunteered to participate in this study. Subjects were recruited from the staff and student populations of both the Universite de Montreal and the Montreal Rehabilitation Institute (MRI). All subjects signed an institutionally approved (MRI) informed consent form prior to participation in the study.

Study 2. Ten subjects with traumatic SCIs (9 men, 1 woman), with a mean age of 26.5 years (SD=4.9, range=21-33), volunteered to participate in this study. Descriptive characteristics of the subjects, as recorded from their medical charts, are presented in Table 1. Participants were recruited from the inpatient and outpatient populations of the MRI. Subjects were informed of the project by their physiatrist, and their participation was contingent on voluntary acceptance and physician approval. All participants signed an institutionally approved (MRI) informed consent form prior to participation in this study. Criteria for admission to this study were (1) the presence of a traumatic lesion of the spinal cord above the lumbar segments, (2) hyperreflexia in the triceps surae muscle group as recorded from the subject's medical chart, (3) the ability to maintain a supine position for a period of at least 40 minutes, and (4) the absence of contractures or pressure sores that could prevent the subject from assuming the experimental position. Subjects with a history of metabolic problems (eg, diabetes), orthopedic problems (eg, hip dislocation), or neurological problems (eg, a previous peripheral nerve injury) that could have an effect on the H-reflex and subjects with any condition that could interfere with or impede adequate data collection (eg, urinary problems) were excluded from participation in the study.

Massage

The massage technique used in these, studies was one-handed petrissage applied to the belly of the triceps surae muscle group. The petrissage was performed in a standardized fashion in accordance with previously described guidelines[23,24] and as used in our earlier studies.[3,4] One-handed petrissage consists of alternately grasping and lifting the muscle away from the bone in a circular motion, releasing the tissues on the downward part of the motion.[23] The same therapist (JG) applied the massage to all the subjects to avoid intertherapist variation. The massage was applied at a standardized rate of 30 manipulations per minute and at a pressure of 1.75 kPa (7 in [H.sub.2.O]). The rate and pressure were based on the results of an earlier study conducted in our laboratory.[4]

The therapist was trained using a pressure quantification system (PQS) that has been described in detail elsewhere[4] and is reviewed only briefly here. The system consists of an air pump, an inflated Jobst upper-extremity sleeve,(*) and a calibrated pressure gauge.(dagger) The PQS provided quantitative feedback to the therapist regarding the amount of pressure applied to the sleeve. The therapist trained on the PQS until she could reproduce 10 petrissage manipulations at the preestablished pressure and rate without feedback (verbal or visual) within an accepted error margin of [+ or -]5%. The rate and pressure were monitored by an assistant using a stopwatch and the pressure gauge. The therapist retrained to these criteria prior to each data collection session to ensure accurate replication of the standardized rate and pressure.

H-Reflex and M-Response

Instrumentation. To ensure the standardization and reliability of the H-reflex during the data collection period, the stimulating and recording electrodes were positioned as shown in Figure 1. The anode, a 20-[cm.sup.2] pad, was positioned on the left quadriceps femoris muscle just proximal to the superior pole of the patella. The cathode, a 1-[cm.sup.2] Meditrace silver-silver chloride disposable electrocardiographic (ECG) surface electrode,(dagger) was positioned in the popliteal fossa over the tibial nerve following localization of the nerve using a hand-held electrode probe.

H-reflex recordings were taken from a distal recording site of the left triceps surae muscle group using a pair of pregelled surface electrodes (Meditrace silver-silver chloride disposable ECG surface electrodes). The placement of the recording electrodes was standardized, according to the procedure adapted by Morelli et al,[25] at 1/16th the distance between the left lateral malleolus and the distal crease of the popliteal fossa, with an interelectrode distance of 15 mm. The electrodes were positioned in parallel to the direction of the muscle fibers. The ground electrode (Meditrace silver-silver chloride disposable ECG surface electrode) was positioned over the left medial malleolus.

H-reflexes and M-responses were elicited from the tibial nerve every 10 seconds by a 1-millisecond square-wave impulse sent by a computer-driven Grass S88 stimulator(sections) and related stimulation isolation (Grass SIU5(sections)) and constant current (Grass CCU1(sections)) units. Electromyographic recordings were amplified (TECA model A6MK3(parallel)) and band-pass filtered (3 dB down at 8 and 1,600 Hz) prior to being digitized at 5,000 Hz. H-reflex and M-response peak-to-peak amplitudes and latencies (defined as the time from the application of the stimulus to the first positive deflection point that exceeds two standard deviations from the resting EMG baseline) were measured and recorded for analysis (for study 1, only H-reflex amplitudes were recorded). The M-response was monitored to ensure than any changes in H-reflex amplitude during or after the massage were not artifactual secondary to changes in the recording or stimulating conditions.

Experimental procedure. In study 1, the subjects independently adopted a supine position on a padded treatment table. Their left lower extremity was positioned in a foam block designed to maintain the limb in a standardized position of 40 degrees of hip flexion, 30 degrees of knee flexion, and 50 degrees of ankle plantar flexion. This position left the area between the insertion of the Achilles tendon and the popliteal fossa exposed so that the massage could be applied to the belly of the triceps surae muscle (Fig. 1). Before data collection, subjects were secured to the table using a Velcro(R)(#) strap to ensure their safety throughout the data collection period. Subjects were asked to maintain their head in the anatomical position throughout the entire data collection period, and to remain as still and as quiet as possible.

In study 2, all subjects were transferred from their wheelchairs onto a padded treatment table using a mechanical patient lift. Each subject's left lower extremity was placed in the same standardized position described for study 1. The approximate time from transfer to the beginning of data collection was 30 minutes in order to minimize any movement-induced changes in muscle activity.

Data Collection

In preparation for the placement of the electrodes, the areas of the popliteal fossa and the distal triceps surae muscle were shaved and then cleansed with a 70% isopropyl alcohol solution. The tibial nerve was localized in the popliteal fossa using the hand-held probe, following which a stationary surface electrode was positioned in its place. A 5-minute period of low-intensity stimulation was given to allow the subject to relax and to become familiar with the data collection procedure. During this period, the therapist retrained on the PQS in accordance with the criteria described earlier. An abbreviated H-reflex recruitment profile was then generated to establish the stimulation intensity (70% of the maximal H-reflex) to be used during the data collection session.

Data were collected every 10 seconds during the first 2 minutes of five sequential, 3-minute time periods (Fig. 2). The massage treatment (MASS) was given during the third time period, and the first two time periods (C1, C2) and the last two time periods (C3, C4) served as control conditions. A condition, for the purposes of our study, was defined as a series of 10 H-reflexes and M-responses recorded from the triceps surae muscle group at 10-second intervals at the intensity established during the recruitment profile. Although theoretically the 10 stimuli require only 100 seconds, a fixed (3-minute) time interval was chosen for each condition to permit the rejection of any signals containing an artifact and to permit the collection of replacement data. This situation accounted for approximately 1% of the trials. In addition, some time was also required to reinitialize the software to allow the following condition to begin at the designated time. The premassage control conditions (C1, C2) served to establish baseline values against which comparisons could be made with changes occurring during the massage condition. In addition, these premassage conditions assured the stability of the reflexes over time by providing H-reflex and M-response values for a 6-minute period. The postmassage control conditions (C3, C4) were initiated in order to verify whether any sustained H-reflex decrease existed due to the postion change or the sample group. In the 3-minute massage condition, data were collected only during the first 2 minutes. The remaining time (1 minute) was used to prepare the system so that C3 could begin immediately on cessation of the massage. Thus, recordings were obtained from the triceps surae muscle group during the massage.

Data Analysis

Study 1. Descriptive statistics (means, standard deviations, standard errors of the mean, and coefficients of variation) were calculated for both the H-reflex and M-response peak-to-peak amplitudes for the 10 trials in each condition. The means of the H-reflex and M-response peak-to-peak amplitudes for the 10 trials for each condition for each subject were entered into a one-way repeated-measures analysis of variance (ANOVA). Newman-Keuls post hoc procedures were used to locate any statistical differences between the means. Statistical analyses were performed with the SYSTAT analysis program (Version 2).(**) All tests for significance were performed at the probability level of P<.01 for study 1.

Study 2. The same statistical analyses described for study 1 were also performed with the data obtained from study 2. Two slight differences, however, existed: (1) In study 2, the H-reflex latencies were also analyzed (in the same manner as the H-reflex amplitudes in study 1), and (2) the probability level for significance for study 2 was set at P<.05. The reduced value for study 2 reflects the explorative nature of the study.

Results

Study 1

The results of this preliminary study were similar to the results of earlier studies[1-4] in that there was a decrease in the amplitude of the H-reflex during the massage. The values for the H-reflex peak-to-peak amplitudes recorded during the control conditions ranged from 3.48 to 3.70 mV, whereas the mean H-reflex amplitude recorded during the massage condition was 1.59 mV. Values for the mean M-responses ranged from 0.42 to 0.44 mV. The descriptive statistics for the H-reflex and M-response peak-to-peak amplitudes for each condition are presented in Tables 2 and 3. The ANOVA for the H-reflex amplitudes detected a significant difference between the resulting means (F=21.54; df=4,24; P=.000). The ANOVA table for the H-reflex amplitudes is presented in Table 4. No significant difference was found between the mean M-responses over the period of data collection (F=2.28; df=4,24; P=.09). The stability of the M-response throughout the period of data collection indicated that no change occurred in the recording conditions during this time interval.

[TABULAR DATA OMITTED]

Subsequent post hoc analysis of the H-reflex means, using the Newman-Keuls procedure, determined that the H-reflex amplitude recorded during the massage condition was significantly different from the values recorded during the four control conditions. No other comparisons were found to be significant.

Study 2

M-response. The mean group peak-to-peak amplitudes for the M-responses elicited during each condition varied between 0.16 and 0.19 mV (Tab. 5). The ANOVA detected no significant difference (F=1.92; df=4,36; P=.13) between these means (Tab. 6). This finding is indicative of the stability of the recording and stimulating conditions throughout the period of data collection.

[TABULAR DATA OMITTED]

H-reflex. Descriptive statistics for the H-reflex peak-to-peak amplitudes and latencies are presented in Table 5. The means of the peak-to-peak amplitudes varied from 2.01 to 2.81 mV. The latencies for all conditions ranged from 32-36 to 33.77 milliseconds. The resulting ANOVAs (Tabs. 7, 8) revealed a significant conditions effect for the mean H-reflex peak-to-peak amplitudes (F=4.06; df=4,36; P=.008) but not for the corresponding H-reflex latencies (F= 1.52; df=4,36; P=.22). Post hoc analyses determined that the mean H-reflex amplitude during the massage condition was significantly reduced compared with the means of C1, C2, and C4. No statistically significant difference was found between the mean of the massage condition and the mean of C3. The mean of C3 was increased compared with the mean of the massage condition but was decreased compared with the means of the other control conditions, although this difference was not statistically significant.

[TABULAR DATA OMITTED]

Eight of the 10 subjects showed a decrease in H-reflex amplitude (range=3%-84%). The other 2 subjects (subjects 3 and 6) showed increases of 20% and 4%, respectively. Inspection of the data revealed no obvious association between the descriptive characteristics of those subjects who responded with a decrease in H-reflex amplitude ("inhibition") due to the massage and those who did not (Tabs. 1, 9).

[TABULAR DATA OMITTED]

A continued reduction (albeit not statistically significant) in the H-reflex amplitude was seen in subsequent conditions following the cessation of the massage. This carryover effect was noted to be present in 6 of the 8 subjects who showed a decrease in the H-reflex amplitude during the massage (Tab. 9).

Discussion

Study 1

The postural change brought about by adapting to the supine position did not affect the pattern of the H-reflex recordings during massage because there was the expected decrease in the H-reflex peak-to-peak amplitude during the massage condition. The decrease in the H-reflex amplitude recorded was of a magnitude of 56%, which is similar to that recorded in earlier studies.[1-4] As a result, the supine position was determined to be appropriate and of potential usefulness for data collection involving a, subject group for which the prone position would be contraindicated. Based on the findings of this study, the supine position was chosen as the experimental position for subjects participating in study 2.

Study 2

Our investigation studied the effect of one-handed petrissage on the H-reflex peak-to-peak amplitude in a sample of persons with SCI. There was a 27% mean group decrease in the H-reflex peak-to-peak amplitude during the massage. This decrease was less than that seen in earlier studies involving neurologically healthy subjects.[1-4] In those studies, the mean decrease in H-reflex amplitude produced by massage varied from 39%4 to 71%.[1, 2] the decrease in H-reflex amplitude seen in study 2, although not as great as the decreases seen previously, was statistically significant and confirms the hypothesis that massage is effective in reducing the excitability of the spinal motoneuron pool in individuals with SCI, as measured by changes in the H-reflex amplitude.

The effect produced by massage in persons with SCI was not as uniform as the effects seen in our earlier studies involving subjects without neurological impairment.[1-4] In those studies, all participants responded to the massage with a significant decrease in the H-reflex amplitude. Although this was not the case in study 2, the degree of "inhibition" (reduction in H-reflex amplitude) produced in those subjects who responded to the massage was of a sufficient magnitude to result in an overall average decrease ([+ or -]SD) in the H-reflex amplitude of 27%[+ or -]35% during the massage, compared with premassage control conditions 1 and 2.

The variability, as well as the decreased magnitude, of the response seen in our study compared with that of our earlier work[1-4] was not unforeseen. Levin and Chapman[26] investigated the effect of stimulation of the common and superficial peroneal nerves on the H-reflex in both subjects without neurological impairment and subjects with SCI. They found that the response to common peroneal nerve stimulation in subjects with SCI was not uniform and that the magnitude of the response was greatly reduced compared with that of subjects without neurological impairment. More importantly, they related the variability in the response to the severity of the hyperreflexia in their subjects. Although the severity of the hyperreflexia was not measured in our study, this factor could explain the variability in the recorded response.

Although for the group there was no decrease in amplitude during C3 compared with the other control conditions (C1, C2, C4), there was a tendency toward a carryover effect (a sustained decrease in the H-reflex amplitude following cessation of the massage). This lack of significance was marginal and can probably be attributed to the lack of statistical power due to the small sample size used in this study and to the large variability in the response to massage recorded in these subjects.

A carryover effect was observed in 6 of the 8 subjects who responded to the massage with a decrease in H-reflex amplitude. This is the first time such a carryover effect has been observed with the application of massage[1-4] and is a very encouraging development, which may have clinical implications. A similar tendency has been demonstrated in relation to circumferential pressure applied to the leg.[27] Robichaud et al[27] found no persistence of the decreased H-reflex (recorded from the triceps surae muscle) in a control group (a facilitation was noted following the cessation of pressure), but a 1-minute carryover was seen in a group of subjects with hemiplegia. It is difficult at this time, however, to hypothesize as to the mechanism involved in this carryover effect.

We have previously demonstrated[4] that the "inhibitory" effect produced by massage is directly related to the degree of pressure applied during the massage. The pressures used in that study corresponded to 2.50 kPa and 1.25 kPa as measured on the PQS developed for the study. Both pressures were shown to have an "inhibitory" effect on the H-reflex amplitude, with the greater pressure resulting in the greater decrease in the H-reflex amplitude. With the deeper pressure (2.50 kPa), however, some participants complained of discomfort. Therefore, as a precautionary measure in testing subjects with decreased sensation and a fragile muscle mass, a mean pressure of 1.75 kPa was selected as the criterion pressure. It is impossible at this point to state whether a greater pressure would have resulted in a greater response in the group of subjects with SCI. This issue awaits further investigation.

Clinical Implications

The effectiveness of massage in decreasing motoneuron excitability in a sample of persons with SCI is comparable to that seen with similar modalities commonly used in the clinical setting. The effects of tendon pressure, circumferential pressure, and stretching on the H-reflex have been investigated in both subjects without neurological impairment[27-31] and subjects with neurological impairment.[13,27,32,33] Tendon pressure has been shown to produce a decrease in the H-reflex amplitude of 29% in 7 out of 8 subjects with hemiplegia,[32] which is similar to the average decrease in the H-reflex noted in our study. In contrast to massage, the effect of intermittent tendon pressure in subjects without neurological impairment is not uniform.[29] Kukulka et al[29] reported that 28 of the 32 control group subjects they tested responded to intermittent tendon pressure with a decrease in H-reflex amplitude.

Recently, Robichaud et al[27] investigated the effect of circumferential pressure (using an air splint) on the H-reflex in both subjects with and without hemiplegia. The pressure application resulted in a decrease in the H-reflex in both groups (40% in the nonhemiplegic group and 52% in the hemiplegic group) for recordings taken at 5 minutes into the pressure application. The inhibitory effect of stretching on the H-reflex in individuals without neurological impairment[30,31] also appears to be consistent. The percentage of decrease in H-reflex amplitude noted with stretching, however, falls between 21% and 47%.[30,31] As was observed with massage, the effect of stretching in a neurologically impaired sample is also inconsistent.[32] Although all these associated H-reflex studies[13,27-33] Were oriented toward the clinical setting, only one study[27] demonstrated a carryover effect following the application of the modality. Thus, it is not possible to compare the carryover effectiveness of massage to these other physical modalities.

The fact that no currently used physical modality appears to be consistently effective in reducing motoneuron excitability (as evidenced by changes in the H-reflex) in a neurologically aired sample makes the option of a massage application interesting, particularly in a population of persons with SCI. In our study, the application of massage resulted in an increase in motoneuron excitability in only two subjects; all other subjects showed a decrease in the H-reflex amplitude. The application of massage in an SCI population offers a unique advantage that the other modalities do not, in that massage is known to increase local circulation in subjects without neurological impairment.[34] Thus, in the event that the massage is ineffective in reducing the hyperexcitable state of the motoneuron pool in a particular subject, it may increase the local circulation in a region where circulation is often deficient secondary to muscle paralysis or loss of vasomotor adjustment resulting from the injury. Thus, based on the findings of these H-reflex studies,[13,27,32,33] the application of massage in a population of persons with SCI is a potentially viable alternative to modalities that are currently used in the clinical setting to control conditions associated with increased motoneuron excitability. The efficacy of massage in reducing the hyperexcitable state of the motoneuron pool in other neurologically impaired populations is still unknown. This question awaits further investigation.

Conclusion

The results of this study indicate that massage produces a reduction in H-reflex amplitude during massage in 8 out of 10 persons with SCI. This reduction is comparable to that seen in earlier studies with neurologically healthy subjects.[1-4] the reduced level of motoneuron excitability is seen to carry over in some individuals for periods that may be of clinical usefulness. In the grouped data of the subjects with SCI, however, no significant sustained decrease in the H-reflex amplitude was observed. The use of massage is therefore not supported if a long-term carryover effect is the main treatment goal.

Acknowledgments

We thank V Welch, BSc, for her valuable assistance during the data collection. We also thank Dr CE Chapman and C Goulet, MSc, for their constructive comments concerning an earlier version of this manuscript. Finally, we acknowledge the patients of the MRI who participated in this study.

(*) The Jobst Institute Inc, 653 Miami St, PO Box 653, Toledo, OH 43694. (dagger) Model 661.10, Wika Instruments Canada Ltd, 9783045 Ave, Edmonton, Alberta, Canada T6E 5V8. (double dagger) Graphic Controls Canada Ltd, 215 Herbert, Gananoque, Canada K7G 2Y7. (sections) Grass Medical Instruments Co, 101 Colony Ave, Quincy, MA 02169. (parallel) TECA Corp, 220 Ferns Ave, White Plains, NY 10603. (#) Velcro USA Inc, 406 Brown Ave, Manchester, NH 03108. (**) SYSTAT Inc, 1800 Sherman Ave, Evanston, IL 60201.

References

[1] Morelli M, Seaborne DE, Sullivan S). Changes in H-reflex amplitude during massage of the triceps surae in healthy subjects. J Orthop Sports Phys Ther. 1990;12:55-59. [2] Morelli M, Seaborne DE, Sullivan SJ. H-reflex modulation during manual muscle massage of human triceps surae. Arch Phys Med Rehabil. 1991;72:915-919. [3] Sullivan SJ, Williams LRT, Seaborne DE, Morelli M. Effects of massage on alpha motoneuron excitability. Phys Ther. 1991;71:555-560. [4] Goldberg J, Sullivan SJ, Seaborne DE. The effect of two intensities of massage on H-reflex amplitude. Phys Ther. 1992;72:449-457. [5] Sullivan SJ, Seguin S, Seaborne DE, Goldberg J. Reduction of H-reflex amplitude during the application of effleurage to the triceps surae in neurologically healthy subjects. Physiotheraphy Theory and Practice. 1993;9:25-31. [6] Larue J. Les Effects i Court Terme du Biofeedback, EMG, de la Relaxation Sophronique et du Massage Relaxant sur le Niveau de Spasticite et de la Vitesse D'extention Volontaire du Membre Superieur Chez Six Hemiplegiques. Quebec City, Quebec, Canada: Laval University; 1984. Thesis. [7] Little JW, Halar EM. H-reflex changes following spinal cord injury. Arch Phys Med Rehabil. 1985;68:18-22. [8] Herman RM. Reflex activity in spinal cord lesions. Am j Phys Med. 1964;43:252-259. [9] Diamantopoloulos E, Zander Olsen P. Excitability of motor neurones in spinal shock in man. j Neurol Neurosurg Psychiatry. 1967;30: 427-431. [10] Sabbahi M, Sedgewick EM. Age-related changes in monosynaptic reflex excitability. J Gerontol. 1982;37:24-32. [11] Cadilhac J, Georgesco M, Benezech J, et al. Potentiel evoque cerebral somesthesique et le reflexe d'Hoffmann dans les Iesions medullaires aigus interet physiopathologique et prognostic. Electroencephalogr Clin Neurophysiol. 1977;43:160-167. [12] Bour LJ, Ongerboer de Visser JH, Koelman TM, et al. Soleus H-reflex tests in spasticity and dystonia: a computerized analysis. Journal of Electromyography and Kinesiology. 1991;1:9-19. [13] Angel RW, Hofmann WW. The H-reflex in normal, spastic and rigid subjects. Arch Neurol. 1963;8:591-596. [14] Taylor S, Ashby P, Verrier M. Neurophysiological changes following traumatic spinal lesions in man. j Neurol Neurosurg Psychiatry. 1984;47:1102-1108. [15] Milanov IG. A comparison of methods to assess the excitabilty of lower motoneurons. Can J Neurol Sci. 1992;19:64-68. [16] Shemesh Y, Rozin P, Ohry A Electrodiagnostic investigation of the motor and spinal reflex arch (H-reflex) in spinal cord injury. Paraplegia. 1977-1978;15:238-244. [17] Matthews WB. Ratio of maximum H-reflex to maximum M-response as a measure of spasticity. J Neurol Neurosurg Psychiatry. 1966;29: 201-204. [18] Katz RT, Rovari GP, Brait C, Rymer Z. Objective quantification of spastic hypertonia: correlation with clinical findings. Arch Phys Med Rehabil. 1992;73:339-347. [19] Buchanan LE, Nawoczenski DA. Spinal Cord Injury Concepts and Management Approaches. Baltimore, Md: Williams & Wilkins; 1987:50. [20] Hugon M. Methodology of the Hoffmann reflexes in man. In: Desmedt JE, ed. New Developments in Electromyography and Clinical Neurophysiology, Volume 3: Human Reflexes, Pathophysiology of Motor Systems, and Methodology of Human Reflexes London, England: S Karger; 1973:277-293. [21] Brunia CHM, Burke D, Delwaide PJ, et al. A dicussion of the methodology of the triceps surae T and H reflexes. In: Desmedt JE, ed. New Developments in Electromyography and Clinical Neurophysiology, Volume 3: Human Reflexes, Pathophysiology of Motor Systems, and Methodology of Human Reflexes. London, England: S Karger; 1973:773-780. [22] Chapman CE, Sullivan SJ, Pompura J, Arsenault AB. Changes in hip position modulate soleus H-reflex excitability in man. Electromyogr Clin Neurophysiol. 1991;31:131-143. [23] Tappan FM. Healing Massage Techniques: Holistic, Classic and Emerging Methods 2nd ed. East Norwalk, Conn: Appleton & Lange; 1988. [24] Hollis M. Massage for Therapists. Oxford, England: Blackwell Scientific Publications Ltd; 1987. [25] Morelli M, Sullivan SJ, Seaborne DE. Comparison of human triceps surae H-reflexes obtained from mid and distal recording sites. Electromyogr Clin Neurophysiol. 1990;30:181-186. [26] Levin M, Chapman CE. Inhibitory and facilitatory effects from the peroneal nerve onto soleus H-reflex in normal and spinal man. Electroencephalogr Clin Neurophysiol 1987;67: 468-478. [27] Robichaud JA, Agostinucci J, Vander Linden DW. Effect of air-splint application on soleus muscle motoneuron reflex excitability in nondisabled subjects and subjects with cerebrovascular accidents. Phys Ther. 1992;72:176-185. [28] Kukulka CG, Fellows WA, Oehlertz JE, Vanderwilt SG. Effect of tendon pressure on alpha motoneuron excitability. Phys Ther. 1985; 65:595-M. [29] Kukulka CG, Beckman SM, Holte JB, Hoppenworth PK. Effects of intermittant tendon pressure on alpha motoneuron excitability. Phys Ther. 1986;66:1091-1094. [30] Robinson KL, McComas AJ, Belanger AY. Control of the soleus motoneuron excitability during muscle stretch in man. j Neurol Neurosurgery Psychiatry. 1982;45:699-704. [31] Etnyre BR, Abraham ID. H-reflex changes during stretching and two variations of proprioceptive neuromuscular facilitation techniques. Electroencephalogr Clin Neurophysiol. 1986;63: 174-179. [32] Leone JA, Kukulka CG. Effects of tendon pressure on alpha motoneuron excitability in patients with stroke. Phys Ther. 1988;68:475-480. [33] Burke D, Andrews C, Ashby P. Autogenic effects of static muscle stretch in spastic man. Arch Neurol. 1971;25:367-372. [34] Hovind H, Neilsen SL. Effect of massage on blood flow in skeletal muscle. Scand J Rehabil Med. 1974;6:74-77.

J Goldberg, MScPT, is Physical Therapist, Hopital Marie-Enfant, Montreal, Quebec, Canada H1T 1C9. She was a graduate student in the Master of Science Program in Sciences Biomedicales (Readadaption), L'Ecole de Readaption, Faculte de Medecine, Universite de Montreal, Montreal, Quebec, Canada H3C 3J7, at the time of this study.

DE Seaborne, MSc, is Professor, Department of Physiotheraphy, L'Ecole de Readaption, Faculte de Medecine, Universite de Montreal.

SJ Sullivan, PhD, is Associate Professor, Department of Exercise Science, Concordia University, Montreal, Quebec, Canda H4B 1R6, and is affiliated with the Centre de Recherche, Institut de Readaption de Montreal, 6300 Darlington Ave, Montreal, Quebec, canada H3S 2J4, and L'Ecole de Readaption, Faculte de Medecine, Universite de Montreal. Address all correspondence to Dr Sullivan at the Centre de Recherche, Institut de Readaption de Montreal.

BE Leduc, MD, is Staff Physiatrist, Department de Physiatre, Institut de Readaption de Montreal.

This study was approved by the Ethics Committee of the Institut de Readaption de Montreal.

This study was supported in part by a grant from the Reseau de Recherche en Readaption de Montreal et del l'Quest du Quebec and the fonds de Recherche en Sante du Quebec awarded to Ms Goldberg.

This articlee was submitted April 22, 1993, and was accepted April 11, 1994.