EMG Biofeedback and Tension Headache: A Controlled … · EMG Biofeedback and Tension Headache: A Controlled Outcome Study THOMAS H. BUDZYNSKI PHD, JOHAN, N M . STOYVA PHD, CHARLE,

EMG Biofeedback and Tension Headache:A Controlled Outcome Study

THOMAS H. BUDZYNSKI, PHD, JOHANN M. STOYVA, PHD, CHARLES S. ADLER, MD andDANIEL J. MULLANEY, MA

A significant reduction in muscle contraction headache activity was observed in patientstrained in the relaxation of the forehead musculature through EMG biofeedback. Trainingconsisted of 16 semiweekly 20 min. EMG feedback sessions augmented by daily home practice.A pseudofeedback control group and a no-treatment control group failed to show significantreductions. A three-month follow-up questionnaire revealed a greatly decreased medicationusage in the experimental group.

In the late fifties, two British research-ers (1) employed a then unique electro-myographic (EMG) integration circuitto show that the resting level of frontalisEMG activity was higher in tensionheadache patients than normals. Since theimmediate cause of pain associated withthis common type of headache (moreproperly called muscle contraction head-ache) is usually due to a sustainedcontraction of the scalp and neck muscles(2,3,4), we hypothesized that if patientscould be taught to relax these muscles, thepain would be alleviated.

A previous study in our laboratory hasindicated that individuals can be trainedto lower frontalis tension levels throughEMG biofeedback. (5) Subjects in thisstudy reported that there was a generali-zation of the relaxation to other musclegroups especially in the head and neckarea. In view of these observations and the

From the University of Colorado Medical Center,Denver, Colorado 80220.

Received for publication October 2, 1972; revisionreceived March 16,1973.

Address for reprint request: Dr. T.H. Budzynski,University of Colorado Medical Center, 4200 E. 9thAve. m.s. 2621, Denver, Colorado 80220.

results of the British study, we decided toapply EMG feedback from the frontalis totension headache.

The results of a pilot study with fivepatients (6) revealed that the EMG feed-back training appeared to be effective inreducing the frequency and severity oftension headaches. However, to rule outthe possibility that these results weremainly attributable to either placebo orsuggestion effects, we initiated the pre-sent study which employed two controlgroups in addition to the experimentalgroup.

METHOD

PatientsAdvertisements placed in a local paper

asked for individuals afflicted with fre-quent tension headaches to participate ina study at the University of ColoradoMedical Center. The applicants wereoffered no pay. A 22-item telephonequestionnaire was used to screen outapplicants who appeared to have otherthan muscle contraction headaches.Those who passed the telephone inter-view next underwent a thorough medicaland psychiatric examination in order to

484

Copyrightc 1973 by the American Psychosomatic Society, I ncPublished by American Elsevier Publishing Company, Inc.

Psychosomatic Medicine Vol. 35, No. 6 (Nov.-Dec. 1973)

EMGBIOFEEDBACKANDTENSION: ACONTROLLED OUTCOME STUDY

rule out the possibility of neurologicaland other organic disorders and toconfirm the diagnosis of tension head-ache. Typically, this type of headache ischaracterized by a dull "band-like" painlocated bilaterally in the occipital region,although it is often felt in the foreheadregion as well. It is gradual in onset andmay last for hours, weeks, even months.

Following the examination patientswere asked to begin daily charting of theirheadache activity. The purpose of thischarting was to provide us with quantita-tive data on headache levels for the entirecourse of the study. As shown in Fig. 1 (ahypothetical patient), the charts were 3 by5 in. cards with the vertical scale repre-senting headache intensity from 0 to 5,with " 5 " indicating an intense, in-capacitating headache. A "4" representeda very severe headache which madeconcentration difficult, but the patientcould perform tasks of an undemandingnature. A " 3 " headache was painful, butthe patient would be able to continue athis job. The "2" level represented aheadache pain level that could be ignoredat times. A level " 1 " headache was a very

low level type which entered awarenessonly at times when attention was devotedto it.

The patient plotted one point for eachwaking hour, and the headache data wereaveraged to obtain a weekly score. Forexample, in Fig. 1, the hourly average forthis day would be computed in thisfashion:

HD = [(1 X 3) + (2 x 4) + (3 x 3) + (4 X4) + (5 X 2)] -H 24 = 1.92

The weekly score would be the simpleaverage of the seven H]j scores for thatpatient. An average of 1.92 for a weekwould indicate an extremely high level ofheadache activity.

In order to establish a baseline level ofheadache activity, all patients chartedheadaches for two weeks prior to anytraining. Data from the pilot study in-dicated that an average of 0.3 was amoderate level of headache activity.Those patients who scored below thisaverage (approximately 25%) for the twoweek baseline were not included in themain study but were assigned to a "casestudy" group. These "case study" pa-

INT

4

3

2

0

NAME: J—i •Ji~r**±

DATE: »-*-*/

/

/

/

/ \

\

\

\

\

Fig. 1. Headache rating chart.

Psychosomatic Medicine Vol. 35, No. 6 (Nov.-Dec. 1973) 485

THOMAS H. BUDZYNSKI, PhD, et al.

tients were not used in the main study,but they were given EMG feedback train-ing. Almost all of these individualsreported a sudden disappearance or de-cline of headaches soon after their accept-ance into the study. However, theirheadaches usually returned after five totwelve days. This placebo or suggestionresponse produced genuine wondermentin most of these patients.

All patients who passed the physicaland psychiatric interview screen weregiven the Minnesota Multiphasic Person-ality Inventory (MMPI). This test wasagain administered at the end of thetraining period. Those patients whoshowed evidence of severe psychologicalproblems as detected by the MMPI wereeliminated from the main study althoughthey were allowed to continue training.Dropouts were replaced with patientswho answered the second advertisementplaced in the paper. Of the 18 patientsselected for the study, 2 were male and 16were female. The mean age was 36 years,with a range of 22-44 years. The meanduration of severe headache activity forgroups A, B, and C was 9.6, 6.8, and 6.7years, respectively. Occupations includedsecretaries, teachers, housewives, gradu-ate students, nurses, and a writer.

Experimental DesignAfter the two-week baseline period

during which two no-feedback sessionswere used to assess pretraining EMGlevels, 18 patients were randomly as-signed to one of three groups for a total ofsix in each group. Group A patientsreceived the EMG biofeedback training(the experimental condition). Group Bpatients also received the "feedback"except that it was tape recorded fromGroup A (the "pseudofeedback" condi-tion), i.e., the feedback signal produced

and heard by the experimental patientswas tape recorded and then played backto the Group B patients. Thus, theyreceived noncontingent feedback. GroupC received no training but the patientswere asked to keep track of their head-aches on the daily charts (no-treatmentcondition).

After the two-week baseline, Groups Aand B received 16 sessions of training(ideally, two sessions per week) followedby a three-month follow-up period. Dur-ing this time the patients charted theirdaily headache activity. At the end of thethree-month follow-up, patients fromGroups A and B were brought back forthree no-feedback sessions to assess theirability to produce low EMG levels. Aquestionnaire was also administered toGroups A and B at the end of thethree-month follow-up. The questionnairewas particularly designed to assess evi-dence of symptom substitution and levelsof medication usage.

Upon completion of 16 sessions byGroups A and B, Group C patients wereallowed to begin feedback training. Simi-larly, after the three-month follow-up,Group B patients were told that theycould, if they so desired, receive addition-al training of a slightly different sort (realfeedback).

Instructions to the PatientsThe instructions to Group A patients

were as follows:"Tension headaches are primarily due

to sustained contraction or tightness inthe muscles of the scalp and neck.

The goal of this study is to learn to relaxyour muscles so that the tension levelnever gets too high, and you no longer getheadaches. This will involve a great dealof work on your part, both here in the lab,and also at home.

486 Psychosomatic Medicine Vol. 35, No.6(Nov.-Dec. 1973)

EMC BIOFEEDBACK AND TENSION: A CONTROLLED OUTCOME STUDY

In order to help you learn, we are goingto provide you with information as to thelevel of muscle tension in your foreheadregion. You will hear a series of clicks inthe headphones. The click rate will beproportional to your forehead tension;that is, the higher the tension, the fasterthe click rate. Your job will be to find outwhat makes the click rate slow down,because this means lower muscle tension.Try to eliminate those things that makethe click rate go faster. Do not try too hard,or this will defeat your goal of deeprelaxation. Remember to keep your atten-tion focused on the clicks—do not let yourmind wander.

This session will last about 30 minutes.Remember—do not go to sleep.Any questions?"The instructions to Group B patients

were as follows:"Tension headaches are primarily due

to sustained contraction or tightness inthe muscles of the scalp and neck.

The goal of this study is to learn to relaxyour muscles so that the tension levelnever gets too high, and you no longer getheadaches. This will involve a great dealof work on your part, both here in the lab,and also at home.

As you relax, it is important to keep outintruding thoughts. The varying click rateyou will hear in the headphones will helpyou to keep out these thoughts. It is veryimportant to keep your attention focusedon the varying rate of clicks. Do not letyour mind wander.

This session will last about 30 minutes.Remember—do not go to sleep.Any questions?"Patients in this control group were not

told that the feedback reflected tensionlevels in their forehead musculature be-cause they could easily have determinedthat this was not true.

Group C patients were told that theywere to chart their headache activity eachday and that training would begin after atwo-month base-line period. These pa-tients were brought to the laboratoryseveral times during this period for fullinstrumentation no-feedback sessions inorder to encourage them to remain in thestudy.

Home PracticeSince our pilot study results had in-

dicated the critical importance of dailypractice outside the laboratory setting,patients in Groups A and B were told topractice relaxation outside of the laborato-ry for two 15-20 minute periods everyday. No specific relaxation instructionswere given for the home practice exceptthat the patients were told to relax in thesame way they had in the laboratory—but,of course, without the aid of any instru-ments.

Instrumentation and LaboratoryProcedureThe "BIFS" EMG feedback system

(Bio-Feedback Systems, Inc., Boulder,Colorado) was designed to assist individ-uals in reaching a condition of thoroughmuscle relaxation by means of informa-tion feedback. The unit was able toprovide several types of auditory feedbackas well as visual feedback. However, forthis study, only auditory feedback in theform of a series of click sounds wasemployed. The frequency of the clickswas proportional to the integrated EMGlevel. A high EMG level produced a highclick rate. As the EMG level declined, theclick frequency decreased. The patient,who had EMG electrodes applied to theskin surface over the frontalis muscle,attempted to lower the click rate byprogressively relaxing the muscle.



The electrodes (one-half in. diam.;silver, silver-chloride) were placed 1 in.above each eyebrow and spaced 4 in.apart on the patient's forehead. Onereference electrode was located in thecenter of the forehead. Electrode resist-ances were less than 10,000 ohms. Thepatient reclined on a couch, in a dimly-lighted, electrically shielded room, andkept the eyes closed.

The EMG feedback unit functions asdiagrammed in Fig. 2. An a.c. differentialpreamplifier with a bandwidth of120-1000 Hz is used to amplify (gain =1000) the bioelectric signal generated bythe muscle. The amplified EMG signal isthen both quantified and converted into afeedback signal by the BIFS. The fluctuat-ing EMG level is changed into a varyingclick rate. Thus, the patient can "hear" hisown muscle activity. The quantification ofthe EMG is such that a digital readout,available each minute, represents theaverage level of EMG activity in mi-

crovolts (MV) peak-to-peak (p-p) for thatminute.

A cassette tape recorder was used topresent the feedback clicks, as recordedfrom experimental patients, to the pseu-dofeedback control patients.

RESULTS

EMG LevelsIn this carefully selected group of

tension headache patients, the level offrontalis EMG during the two baselineweeks averaged slightly over 10 V (p-p)for each group. These values are at leastdouble those shown by young normalsubjects in our laboratory. These readingsare also a considerable increase over the 6I V p-p baseline level for the five patientsin the pilot study (6) and probably reflectthe more stringent selection criteria usedin the present study. It is evident from

488

Fig. 2. Functional diagram of the EMG feedback system.

Psychosomatic Medicine Vol. 35, No. 6 (Nov.-Dec. 1973)

EMC BIOFEEDBACKANDTENSION: ACONTROLLEDOUTCOME STUDY

Fig. 3 that the mean EMG level for GroupA showed a considerable decrease fromthe baseline level in the first feedbacksessions. The mean EMG level of theGroup B patients also dropped somewhatafter the baseline sessions; however, themean value of Group B remained at ahigher level than the mean of Group A.The Group B curve also showed a greatdeal more variation than did that of thefeedback group, perhaps not a surprising

•- •• GROUP B

t

N = 6 , EACH GROUP

81 B2 I 2 3 4 5 6 7 8 9 10 II 12 8 14 15 16BASELINE FEEDBACKSESSIONS SESSIONS

MEAN FRONTALIS EMG LEVELSACROSS SESSIONS

Fig. 3. Mean frontalis EMG levels across sessions.Group A—true feedback. Group B—pseudofeedback.

result since it is characteristic of feedbackto decrease the variance of the response.

It was expected that the pseudofeed-back Group B patients would show some

decrease in EMG level as a result of thefocusing of attention on a meaninglessand comparatively monotonous stimulus(the "feedback" clicks). Furthermore, theshifting of attention from troublesome,anxiety-evoking thoughts to a relativelyneutral stimulus probably also contribut-ed to the lowered EMG level. Interesting-ly, it may be noted that the focusing ofattention on a neutral, meaninglessthought or word, to the exclusion of otherthoughts, is an essential characteristic ofmany meditative disciplines.

Although all three groups showed nodifferences in baseline EMG levels, therewas a significant difference (p < 0.05one-tailed) between Groups A and Bduring the last two weeks of training(Group C did not receive any training).

After the three-month follow-up period,the patients from Groups A and B weretested for three sessions with no feedback.The mean frontalis EMG levels were 3.92and 8.43 /uV p-p for A and B respectively,and again represented a significant differ-ence (p < 0.01 one-tailed) between thegroups. Apparently the trained group hadretained the learning over the three-month period.

Headache ActivityAs expected, the averaged headache

rating scores for both the A and B groupsdeclined over time (see Fig. 4). However,as Fig. 4 indicates, baseline levels ofheadache activity had been somewhatlower in Group A than in Group B or C.Therefore, a Kruskal-Wallis analysis ofvariance by ranks was first applied to thebaseline headache data. This test showedthat the starting levels of the three groupswere not significantly different from eachother.

Additionally, in order to eliminate thepossibility that different baseline levels



V r-\

V

N=6, EACH GROUP

GROUP

GROUP

GROUP

A ° •

B = »

C o « _

1-<*

\

\ A

Fig. 4. Headache activity during feedback training (all 3 groups) and duringthe three-month follow-up (A and B only). Group A—true feedback. GroupB—pseudofeedback. Group C—no treatment.

were contributing to a significant differ-ence between groups, we performed aslope analysis for each group. (7) OnlyGroup A produced a statistically signifi-cant decline (p < 0.001 when the regres-sion coefficient was tested against the nullhypothesis of zero slope).

The headache data for individual pa-tients in each group were also analyzed inthis fashion. The analysis revealed thatfour out of six patients in the A groupshowed significant declines (p < 0.05] inheadache activity, while in the pseudo-feedback control group, only one of sixshowed a significant decline. None of theC group patients showed a significantdecline below baseline levels.

Finally, the Kruskal-Wallis analysis ofvariance by ranks which had been appliedto the baseline headache activity was alsoused to test differences among the threegroups at the end of the training period(weeks eight and nine). At this time therewere significant differences in headacheactivity among the groups (p < 0.001).

Correlation Betwen EMG Levels andHeadache ActivityWhen weekly headache activity during

the baseline and training weeks wascorrelated with weekly frontalis EMGlevels, the A group data showed a + 0.90correlation while the B group showedonly - 0.05, or essentially no correlation.This result may be due to the fact that thepatients receiving the real feedback wereindeed learning to relax in the laboratoryand were able to apply this learningoutside the laboratory, whereas the pseu-dofeedback patients generally were una-ble to do so.

The three-month follow-up data (seeFig. 4) indicated that the B group patientsappeared to have stabilized at a meanheadache level of about 0.53, whereas theA group was producing very little in theway of headache activity during the lastmonth.

DropoutsIt should be noted that there were four

490 Psychosomatic Medicine Vol. 35, No. 6 (Nov.-Dec. 1973)


dropouts from the original group B. Thesewere patients who felt that the trainingwas having no effect on their headaches.All of them were experiencing high levelsof headache activity when they retiredfrom the study. However, the Group Bpatients who remained felt that thetraining was helpful. There were also twodropouts in the C group, but none in the Agroup. All dropouts were replaced withnew patients.

Subjective ReportsWhile being instrumented prior to the

sessions as well as just after the sessions,the patients would often volunteer com-ments as to their success or lack of it withthe training. These comments were laterentered into a log book by the technicians.On the basis of these comments, it soonbecame apparent that in this study, aswell as the prior pilot study, the patientspassed through several discrete stages interms of their ability to use a "cultivated"relaxation response to reduce headacheactivity.Stage 1. Patient is unable to prevent orabort headaches.Stage 2. Patient becomes more aware ofthe tension preceding the headaches andcan relax to some degree with a consciouseffort. However, he cannot abort head-aches.Stage 3. Patient shows an increasingawareness of the tension, plus he is betterable to relax consciously and abort light-

to-moderate headaches. The frequencyand intensity of headaches is now dimin-ishing.Stage 4. Patient now seems to relaxautomatically in the face of stress anddoes not have to make a conscious effortto do so. The headache activity is nowappreciably reduced or even eliminated.The last stage would seem to indicate thatthe ability to relax in the face of stresseventually becomes an overlearned habitresulting in a change in life style.

MMPI ResultsAll the patients were given the Min-

nesota Multiphasic Personality Inventory(MMPI) before and after the trainingperiod. In general, the "before" profile ofscores showed that the Hs (hysteria), D(depression), and Hy (hypochondriasis)scales were somewhat elevated (themeans for all three groups were in the low60's). The "after" profiles of all threegroups showed reductions in these threescales; however, the only statisticallysignificant mean before-after change oc-curred in the Hy score of Group A (p <0.05 two-tailed).

Table 1 lists the number of patients ineach group who showed declines of 10points or more on these three scales. TheA group produced a total of 10 change-scores equal to or greater than 10 pointswhile Group B showed four, and group Ca total of three. A chi-square test per-formed with the data from this table

TABLE 1. Number of Patients in Each Group Showing Declines of Ten or More Points on Three MMPIScales

Croup ACroup BCroup C

Hs321

D311

Hy411

PsychosomaticMedicineVol.35,No.6(Nov.-Dec.1973) 491


showed that the three groups were signifi-cantly different (p <0.01) in the number ofnegative change-scores greater than ten.

The "before" profile showing the ele-vated triad of hysteria, depression andhypochondriasis is in general agreementwith Martin (4) who found the sameelevations in a large group of tensionheadache patients.

Follow-up DataA four-page post-training questionnaire

was used to assess drug usage, evidenceof symptom substitution, mood and be-havior changes, and interpersonal rela-tionships. All the patients in Groups Aand B received this questionnaire after thethree-month follow-up period. The pa-tients were asked to rate the severity orfrequency of symptoms on a scale of 0 to3. Ratings were made for four periods:before training, first half of training,second half of training, and after training.

In the A group decreasing severity orfrequency was seen in 27 of 28 items.Group B patients rated themselves asdecreasing on 23 of 28 items. In bothgroups the items showing the greatestdecreases were depression, tension, anx-iety, insomnia, fast heart beat, irritability,

persistent thoughts, sexual disinterest,and fear of driving. Lesser decreases inboth groups included chest pain, use ofalcohol, sweating, and sexual anxieties.The Group A patients in addition regis-tered large decreases in tiredness, apathy,fear of crowds, and compulsive behavior.Patients in both groups saw themselves asimproved in relationships with spousesand/or friends.

Although, generally, there was no evi-dence of symptom substitution, one pa-tient in Group A did report a certainamount of stomach distress as she pro-ceeded through training in deep relaxa-tion. This phenomenon appears to berelated to the sudden shift from a predom-inantly sympathetic autonomic state to-wards a parasympathetically dominantpattern. This transition does seem toproduce an increase in stomach acidity insome individuals. As the patient contin-ues the daily relaxation practice thisreaction tends to disappear.

Drug usage decreased dramatically inGroup A patients. As seen in Table 2, 4 of6 went from a rate of 3 to 4 capsules ofprescription tranquilizers and painkillers(typically valium, librium, fiorinal, anddarvon) per day to only occasional use of

TABLE 2. Drug Usage in Group A (Experimental)

Patient

1

2

3

4

5

6

Before Study

Fiorinal, Valium3-4 daily

Darvon, Equagesic(all day)Valium

4 per dayAnacin—10 per

dayDarvon—4 per

daynone

First Halfof Training

Fiorinal, Valium3-4 daily

Darvon, Equagesic(all day)Valium1 x day

Anacin—4 perday

Darvon—4 perday

none

Second HalfofTraining

Librium3-4 daily

Darvon, Equagesicseldomnone

Anacin—2 perweek

Darvon—2 perday

none

AfterTraining

Librium3 per week

Darvonseldomnone

Anacin—2per week

none

none



the tranquilizer. Another patient in GroupA decreased his intake of aspirin from upto 10 per day to 2 per week. The finalpatient in this group took no medicationfor his headaches (and did not show asignificant decrease in headache activity).

Two of the Group B patients (see Table3) reported decreases in medication,while three did not change in their usage.One patient switched from a fiorinal-by-day, librium-by-night schedule to libriumday and night. Interestingly, one patientfrom this group who showed no decreasein drug usage had reported a decrease inheadache activity.

The questionnaire also required pa-tients to rate their level of headacheactivity before, during, and after thetraining. In Group A, five of six rated theirheadaches as decreasing. One patientindicated no change. Three of the Group Bpatients rated themselves as decreasing inheadache activity, while three others saw

no change. One of those who rated herheadaches as decreasing did not show adecrease in her daily charting of theheadache activity.

Eighteen-month Follow-up. Approxi-mately 1V£ years after the completion offeedback training, four of the six Group Apatients (two had left Colorado) werecontacted. Three of the four previouslyhad shown significant declines in head-ache activity during training. The threereported that their headaches remained ata very low level (roughly one or two mildheadaches a month). Because they nowfelt more relaxed generally, they nolonger engaged in a daily period of deeprelaxation, using this approach only whenfeeling particularly tense. The fourthpatient (who had not shown a significantreduction during training) reported thathis headaches continued, though at areduced rate.

"Real" Training for Group B and C

TABLE 3. Drug Usage in Group B (Pseudofeedback Control)

Patient

1

2

3

4

5

6

Before Study

Fiorinal—2 perday

Librium—2 atnight

Librium—2 perday

Aspirin—4-5day

Anacin—4-6per day

Meprobamate—3per day

Elavil—2 perday

WigraineValium

Equagesic—3per day


First Halfof Training

Fiorinal—2 perday

Librium—2 atnight

Librium—1 perday

Aspirin—2 perday

same

same

same

Equagesic—2per day


Second Halfof Training

Fiorinal—2 perday

Librium—1 atnight

Aspirin—1 perday

same

same

same

none

AfterTraining

Librium—4day and night

Aspirin—1per day

same

same

same

none



Patients. After the three-month follow-up,three patients from the pseudofeedbackgroup (B) decided to try "another type" oftraining. Two of these people showedsignificant decreases in their headachescores through the training period. Thethird individual did not improve signifi-cantly.

Five of the Group C patients alsoreceived training consisting of 16 sessionsof EMG feedback. Their training wasinitiated after their nine-week "baseline"period was completed. Four of the fiveshowed significant declines in headacheactivity.

The biofeedback training of the eightformer control group patients along witha number of "pilot" headache patientswas augmented with cassette tape record-ings for home practice. In several in-stances portable EMG feedback units wereused at home as well.

DISCUSSION

The results of an earlier pilot study withfive patients (6) had suggested that train-ing in relaxation of the forehead muscleswith EMG feedback might be effective ineliminating muscle contraction or tensionheadaches. That conclusion was furtherstrengthened by the analysis of the datafrom this second experiment which em-ployed two control groups in addition tothe experimental group. It now seemsapparent that chronic tension headachepatients can learn to decrease their restingforehead EMG levels by 50 to 70% inthree to six 20-minute feedback sessions.When they subsequently engage in regu-lar, daily relaxation, the headache activitydiminishes considerably. Recently, otherlaboratories (8,9) have also reported thatEMG feedback training is useful in thealleviation of tension headache.

These results are in keeping with asuggestion made independently by Mal-mo (10) of McGill University MedicalSchool, who has worked extensively withelectromyographic recording since theearly 1950's. Malmo proposed that sys-tematic muscle relaxation training mightwell be useful for treating tension head-ache.

Stages of ProgressAs the patients progressed through

training, their verbal reports suggestedthat they first developed a heightenedawareness of maladaptive tension levels.This was followed by an increasing abilityto remove the tension (and slight-to-mod-erate headaches) through relaxation. If thepatients then applied this new learning toevery day stress situations, a change inlife style frequently seemed to occur. Atthis stage, patients typically reported thatthey no longer overreacted to stress. This"automatic" moderation of arousal levelin the face of stress has also been reportedby anxiety patients who have receivedEMG feedback-assisted relaxation train-ing in our laboratory.

Transfer to Real LifeThe no-feedback sessions at the end of

the three-month follow-up revealed thatthose patients who had received feedbacktraining still retained the ability to pro-duce low forehead EMG levels. Theeighteen-month follow-up interview in-dicated that most experimental grouppatients had also managed to keep them-selves relatively free of headaches eventhough they had been chronic headachesufferers for years prior to the training. Inthis group also, the use of powerfulprescription drugs decreased dramati-cally.

Only one of the pseudofeedback con-


EMC BIOFEEDBACK ANDTENSION: A CONTROLLED OUTCOME STUDY

trols produced a significant decline inheadache activity over time. This patientwas the youngest member of that group.She regularly performed the daily homerelaxation practice. Although she was notgiven specific relaxation instructions, thispatient learned to discriminate the inter-nal cues of thorough relaxation such asheaviness and warmth in the arms.

Importance of Daily PracticeThis study, as did the pilot study,

pointed up the importance of daily homepractice. The two experimental grouppatients who did not show significantdeclines had found it difficult to carry outthe home relaxation assignments. Typi-cally they reported that the hectic state ofaffairs at home did not permit quietperiods of relaxation. Other patients stat-ed that they would have preferred moreexplicit relaxation instructions for homeuse. A few found the daily home practiceto be somewhat boring.

Addition to the Basic TechniqueIn the present study, only a minimal

sort of training was employed—EMGfeedback from the frontalis muscle. Proba-bly this training could be strengthenedconsiderably. For example, we have re-cently begun experimenting with twohome practice techniques which shouldadd both structure and novelty to thehome training. One technique makes useof a 30-minute cassette tape containingrelaxation instructions on either side. Theother technique utilizes a battery-pow-ered portable EMG feedback unit (Bio-Feedback Systems, Inc.). The tape andportable unit can be employed singly,sequentially, or simultaneously.

Preliminary results indicate that each ofthese two supplementary methods will bea valuable addition to the minimum

procedure used for Groups A, B and C. Bynow, tapes and portable equipment havebeen used in the training of some of theapproximately 30 tension headache pa-tients (including those from the B and Cgroups who later received the "real"feedback) who have been trained inmuscle relaxation with EMG biofeedbacksince the completion of this second study.The overall results indicate that roughly75% showed significant declines in head-ache activity.

It is possible that some of those who didnot show decreases may have been unwil-ling to give up their headaches. Theheadaches may have allowed those pa-tients to avoid certain anxiety-arousingsituations, or to manipulate others in theirfamily or at work. In these instances,psychotherapy or behavior therapy isrequired. (11)

Even though EMG feedback trainingalone is not effective in all cases, thetechnique would seem to be of considera-ble value for a substantial proportion oftension headache cases. The training doesnot involve drugs or other kinds oftherapy and can be accomplished withrelatively inexpensive portable equip-ment. Training can be carried out by atechnician (or perhaps by the patienthimself) under professional supervision.In most instances, beneficial results canbe achieved in four to eight weeks. Sincemany tension headache patients experi-ence pain in the back of the neck andshoulders, it is possible that faster resultscould be obtained with some of thesepatients through feedback from thesemuscle sites. These locations were notused in this study because it is moredifficult to obtain precise electrode loca-tion here than on the forehead.

A variety of evidence suggests thatbiofeedback techniques may have ap-


THOMAS H.BUDZYNSKI, PhD, etal.

plications to stress-related disorders otherthan tension headache. For example,researchers at the Menninger Foundation(12) have explored the use of skintemperature feedback with migraine pa-tients. In our own laboratory, we have forseveral years regularly employed EMGfeedback techniques in the systematicdesensitization of phobias. (13) Insomniamay be another potential application.Drowsiness is a frequent accompanimentof profound muscle relaxation; perhapsEMG (and related) feedback techniqueswould be useful in some instances ofsleep-onset insomnia. Observations insupport of such a surmise may be found in

Jacobson's writings (14) on progressiverelaxation and in the autogenic trainingliterature. (15) It may be noted that boththese older approaches systematicallytrain patients in the ability to shift readilyinto a relaxed, low arousal condition.

This research was supported by theNational Institutes of Mental Health,Grant Number MH-15596, and ResearchScientist Development Award, GrantNumber K01-MH-43361.

We are most grateful to Susan Biomand John Nagel, M.D. for their technicalassistance.

REFERENCES

1. Sainsbury P, Gibson JF: Symptoms of anxiety and tension and accompanying physiological changes inthe muscular system. J Neufol Neurosurg Psychiat 17:216-224,1954

2. Ostfeld AM: The common headache syndromes: Biochemistry, pathophysiology, therapy. Springfield,Illinois, Thomas, 1962, p. 19

3. Wolff HG: Headache and other pain. New York, Oxford University Press, 1963, pp. 582-6164. Martin MJ: Tension headache, a psychiatric study. Headache 6:47-54,19665. Budzynski TH, Stoyva JM: An instrument for producing deep relaxation by means of analog information

feedback. J Appl Behav Anal 2:231-237,19696. Budzynski TH, Stoyva JM, Adler CS: Feedback-induced muscle relaxation: Application to tension

headache. Behav TherExp Psychiat 1:205-211,19707. Snedecor GW, Cochran WG: Statistical methods, 6th ed. Ames, Iowa, Iowa State University Press, 1967,

152-1538. Wickramasekera I: Electromyographic feedback training and tension headache: Preliminary observa-

tions. Amer JCHnHyp 15:83-85, 19729. Raskin M, Johnson G, Rondestvedt JW: Chronic anxiety treated by feedback-induced muscle relaxation.

Arch Gen Psychiat 28:263-267,197310. Malmo RB: Emotions and muscle tension: The story of Anne. Psychol Today 3:64,197011. DengroveE: Behavior therapy of headache. J Amer Soc Psychosom Dent Med 15:41-48,196812. Sargent JD, Green EE, Walters ED: Preliminary report on the use of autogenic feedback techniques in the

treatment of migraine and tension headache. Psychosom Med, (in press).13. Budzynski TH, Stoyva JM: Biofeedback techniques in behavior therapy, in Beitrage der Neuropsycholo-

gie zur Angstforschung. Reihe Fortschritte der Klinischen Psychologie, BD 4. Edited by N. Birbaumer,Munchen, Wien, Verlag Urban Schwarzenberg (in press). English version republished in Biofeedbackand self-control: 1972. Chicago, Aldine-Atherton (in press) (Edited by D. Shapiro et al.)

14. JacobsonE: Progressive relaxation, 2nded. Chicago, University of Chicago Press, 1938, pp. 418-41915. Schultz JH, Luthe W: Autogenic training: A psychophysiological approach in psychotherapy. New York,

Grune&Stratton,1959

496 Psychosomatic Medicine Vol.35, No. 6 (Nov.-Dec. 1973)

Documents

EMG Biofeedback and Tension Headache: A Controlled … · EMG Biofeedback and Tension Headache: A Controlled Outcome Study THOMAS H. BUDZYNSKI PHD, JOHAN, N M . STOYVA PHD, CHARLE,