Differential Effects of Hypnotic Suggestion on Multiple Dimensions of, Pain Lisa A DahJgren, PhD, Richard Marguerite D. Malone, PhD
M. Kuru,
bparhnmt
Uiry
ofPsyAo@
W**
PhD, Michael J. Strube, PhD, and tn St. L0ui.s. Mrxrouti
Within theof multidimtnsiannl @in awwwnt, this .study extewded nn ear&r finding that h@wtk ana&& and wlawtion qgestbns have kntial effects OR #win wduction b evaluating these stmbgk in .subjectsundkping I c&d @s.uw @kwol. TAirtptum high3 srucgribdc subjtxts um mndhl9 as igw& to an analgesM or n vlbdion suggstioft tmatmmt grmrP. Six @in n+rts we Men at l&x intervat~ fbr wch~CORdiliOR.~bcuclirvmrorunsvrwdaccovatiola.AtX2x2x
6 n?@te&nt&asurts aaa@is gcovtnitancc (ANWVA)
tnrcobd a signifiaant
gmup
(anal&s& rvbakn) 4 @in dimension (intekty, unp&asant~ss), by corniition (satgpstian alow h@otic induction flus sum) inierextiou. Analysis o/the simw main effks, iwkiing bolh group and amdition um.stant, mm.&d that cs#hdOn oJ+otic an+ia ve&taii rrpopt o/@-sin ititiy sigkjkant~ mow than n@wt of pain unp&usantness. Gmwmty, hy#notic n&axation ndwxd pain uaplmrantness nwn than inlensi$ nK clinical implications oJ the study are di.uus.wti J
Pain Symptom Manage 199S;Nk46&470.
lijpwsir,
muttidinunsioiuat pain,
cotd pmsor
During the past 20 years, the experimental literature has shown that hypnosis can be effective in pain control and that successful outcome is generally related to hypnotizability.‘.’ It has also been demonstrated that hyp notic analgesia, with or without induction, can reduce clinical pain:%” While most of the recent- experimental litc+uqe on hypnotic analgesia has fqcused on conflicting theoreti‘. reprint requesfs lo: Richard Katz. Department 6f Psychology, i:anlpus Box
Address
Washington .U.niwrpi,ty, Oric Rrookinlp Louis. MO 69130, i’SA. ArcgM for puMicdlon: February 21, 1995.
YhD,
1125. Drive, St.
cal expllnadons about its core determinants. some recent work has begun to explore the differential effectiveness of various hypnotic suggestions on multiple pain dimensions.rG In 1989, Malone rt al., working within Ctaccly et al.‘s multidimensional lijagnitude estimation model, examined the influence of hypnotic analgesia and relaxation suggestions on two dimensions of pain: intensity and unpleasantness.‘.‘“” FortyGe highly sutiep tihle subjects were assigned to one of three groups: analgesia suggestion, relaxation +ggestion, anti a no su&estion control group. Subjects were administered electrical shocks, ranging from 0.5 milliamperes to 4.5 milliamperes, in both the waking and hypnotic states. 0153924/95/$950 s!mI cN853924 (95)00?5~
Analgesia suggestion
wirhin groups Intensi
(X=
16)
‘Benvcen-groupsretaxatkm suggestion (N= 16) SD M=
Mmn
SD
1.98 1.90 1.45
0.06 0.10 0.35
1.98 ::ii
1.82 1.77 1.63
0.12 0.19 0.25
1.87 1.77 1.36
taring
i. F&zadon
dip trial (baseline)
2. Dip-wake 3. Dip-hypnosis (2 and 3 counrcrbalanced acre Sn)
Un lcawntnere rating’ I. rpamiliarization dip trial (baseline) 2. Dip-wake 9. Dip-4lypnosis (2 and 3 counterbalanced
acres Ss)
All subjects rated the shocks using the 16 painintensity descriptors and the 16 pain unplca+ antness descriptors developed by Gracely et al.‘” Descriptors were assigned numerical values using magnitude estimation procedures described by Lodge. I4 They found that hyp notic analgesia altered subjects’ perceptions of pain intensity without changing perceptions of pain unpleasantness, whereas hypnotic relaxation reduced perceptions of unpleasantness but not of intensity. The Malone et al.’ study is limited, however, in that it used electric shock as the pain modality. Electric shock is a poor analog to clinical pain. Additionally, part of the study’s findings were based on post-hoc analyses. The purpose of the current study is IO extend the Malone et al.’ design from elccttical stimulation to a more clinically relevant pain technique. In this study, subjects who were highly susceptible to hypnosis received either an analgesia suggestion or a relaxation suggestion. These subjects were asked to rate both the intensity and the unpleasantness of their pain during a cold pressor protocol in both a waking and hypnotized condition. It wa~hypothesized that the two different sugges lions would differentially~reduce ihe intensity . and unpleasantness dimenstons of pain. Spe cifically, the condition of hypnotic induction plus relaxation. suggestion was expected to reduce 1eport.s of unpleasantness significantly more than intensiry;whereas the condition of hypnotic induc:ion plus analgesia was
if! 0:11
O..ll 0.1s 038
expected to reduce reports of intensity more than unpleasantness.
Subjeds Thirty-two high susceptible unrlergzaduates ( I7 females and 15 males) were randomly assigned to either the relaxation or analgesia treatment condition. individuals scoring 23 or above out of a possible 40 points on the Creative Imagination Scale (ClS) were also administered the Stanford Hypnotic Clinical Scale (SHCS).“.‘” Only subjects who scored 4 tar 5 on the fivepoint SHCS were used. The CUIIbincd cut-off scores were identical to thos,* used by Malone et al.’ FhC@&l?
.Subjccts met with the experimenter on three separate occasions: one screening se+ sion and two separate experimental sessions. The design is summarized in Table 1.
During the screening session, both the ClS and the SHCS were administered via audiotape. Subjects who received 25 or more on the ClS and 4 or 5 on the SHCS were given a Calibration and Magnitude Estintation Booklet developed by Lodge to take home and complete.‘j The self-instructional ‘booklet described magnitude estimation procedures
and provided examties and exercises for subjects to complete.
In experimental session one, subjects were asked to complete a beef questionnaire; those who reported current use of pain medication were dropped. Subjects were then randomly assigned to either analgesia or relaxation sue gestion condition. All of the following factors were cciunterbalanced: (a) order of hypnotic induction. (b) use of right or I& hand for immersions, and (c) order of pain dimension ratings. In addition, hypnotic depth was asses& using the Long Stanford Scale.” To ensure that subjects did not enter a hypnotic state during the nonhypnotic portion of the study, &xi that compamble depths of hypnosis were attained during the hypnotic conditions, subjects gave .hypnotic depth ratings twice, immediately before and immediately after each experimental condition. Both experimental sessions began with a familiarization (baseline) trial. For this trial. each subject was seated with eyes closed in a comfortable chair with a towel on the lap and arms on the chair’s arm re+s. The word list was taped to a clean blackboard 137 cm from the chair, with the bottom edge of the list 81 cm from the floor. Half of the subjects received the list of words that pertained to intensity, and half of the subjects received the list of words that pertained 10 unpleasantness. In a pilot study, we found that by the time the subject had nted ,the pain eperience on the intensity dimension, a variable number of seconds had already elapsed before they could rate the unplcasantnesa dimension. This made the ratings on the unpleasanure:.s and intensity dimension nonequivalent. For this reason. we decided to collect the data for unpleasantness and intensity dimensions on separate days. This, ,of course, represents a methodological compromise. CollecGng different dimension I;rtings on different days raises the possibility of rating differences arising from unknown- betwcen4ay variation. Our use of counterbalancing, and subsequent finding of nonsignificant order efrccts, argues against any s$ematic between-day. confourids. The subject was then asked to open his or her yes, to report the hypnotic depth level after the experi&nier spoke the zue word
“state,” and then to place hi or her hand in the cold wter. The hand stayed in the water for 80 sec. Every 10 sec. the subject was asked to rate either the intensity or the unpIeasantY ness of the pain experience after the cue word “report” was spoken by the experimenter. The subject rated the intensity or the unpleasantness by choosing a word from the appropriate word list The words were listed in order of the group means of magnitude found by CraceIy et al. ’ and were identical to the order and list used by lHalone et al.” After the six pain reports (one every 10 set) were collected, the subject lifted his or her baud out of the water and pIaced it on the towel. The experimenter then asked for a report of the hypnotic depth using the cue word “state,” and instructed the subject to close his or her eyes. At leasi 8 min were allowed to pass between any condition (base line, waking, hypnotic), during which time the subject &VJ asked lo leave the room and 4k up and down a hallway. When the experimenter was satisfied that the subject understood the cold pressor prow col, two experimental trials were administered. One of these trials followed the appropriate suggestion given without a hypnotic induction and the other followed the hypnotic induction plus suggestion. Instructions for hypnotic induction were adopted from Weitzenhoffer and Hilgard,‘n and analgesia and relaxation suggeslidns were taken from Bassman and Wesrer,“’ These instructions were ide.ntical to those :used by Malone et al.’ Both sets of instructions are available from second author. The relaxation suggestion was constructed to provide multisensory grounding for the subjects. St;ltemencs such as, “Each time as you inhale and exhale you might be surprised to find yourself more and more comfortable, more and more relaxed than you have felt in a long, long time.” provided rhythm. A, brief description of various muscle groups preceded suggested image’ y for a beach, complete’with various sensory images such as warm sand, white clouds, birds, sah’air, and haves.. The analgesia suggestion was similarly constructed but avoided mentioning relaxation. The imagery used was of a large jar of fngrant, analgesic hand cream. The order of these two conditions was counterbalanced between subjects. Except for, the adminisu;l-
tion of suggestion or hypnotic induction plus suggestion, the experimental trials did not differ fhn the familiatization (baseline) trial. Hypnotic depth was not manipulated during any of fht trials in the first experimental se+ sion. Experimental session two began by reviewing the experimental protocol. The subject then went through the familiarization trial and the two experimental trials, just as had been done in the first experimental session. In this session, however, the subject was presented with the word list not used in the firer’ expcimental session. Hypnotic depth was made comparable ,between both experimental sessions by using deepening instructions when necessary.
No significant .differences were found between the analgesia and relaxation groups on any of the demographic variables (age, sex, education), susceptibility measures (CIS. SHCS). or hypnotic depth. At baseline, differences between the analgesia and relaxation groups were nonsignificant for, both intensity and unpleasantness descriptors. There were also no order.effects for condition (hypnotic, waking) or pamdimension (intensity, unpleasantness). A Pearson correlation was performed between intensity and unpleasantness for each experimental unit to determine the relative independence of these two pain dimensions. The average correlation was r = 0.45. Although correlations of this magnitude account for less than 20% of the variance, it should be noted. ‘that these two pain dimensions are trot su&titally indepcndcnt. This finding is consistent with an earlier reanalysis of the data of Malone et al.’ by Dahlgren,‘W which also found a similar pattern of intercorrelation between intensity and unpleasantness ratings. In a previous study by Gtacely et al.,” which used healthy pain-free volunteers, the test-retest reliability for tne pain dimension of intensity wm 0.96, and test-retest reliability for uupleasantness was 0.89.” Thus, our estimate of the correlation between unpleasantness and intensiv is not likely to be substantially altered by measurement error. ,. ‘,
Raw values for both line ptoduc&n (LP) aad numeric estimation (NE) were transformed into geometric means using the eqtation recommended by Lodge:‘* (Log LP Log NE)-5. This transformation was identical to that used by Malone et aIs Thedatawereanalyzedasa2X2X2X6 repeated-measures analysis of covariance (ANCOVA) with the fitst factor a betweengroup factor (analgesia versus relaxation), and three withln+&&ct factors (pain dimension, hypnotic or nonhypnotic condition, and time). Because large, individual differences exist in pain response, the baseline (&miii&on) condition was covarled from the ueatment effects (waking and hypnosis). Table 1 presents the means and standard deviations of experimental cells summed over pain reports (time). Initial analysis revealed a significant group (analgesia, relaxation), by pain dimension (intensity, unpleasantness), by condition (suggestion alone, hypnotic induction plus suggestion) interaction (F= 21.93, PC 0.001). Further analysis, holding group constant, revealed a signifiant condition by dimension intetaction in both the analgesia (F= 9.75, PC 0.01 ), and the relaxation groups (1; = 11.15, P k 0.01). Therefore, simple-simple main effects were performed. In these analyses, both group and condition were held constant, so that the differences between pain dimensions were analyzed only in the hypnotic induction plus suggestion condition. Results demonstrates that the application of hypnosis plus analgesia suggestion reduced reports of pain intensity signi6cantly more than reports of pain unpleasantness (F = 11.22, P < 0.01). Convet&y, the application of hypnosis plus relaxation suggestion reduced reports of pain unpleasantness significantly more than inten sity (F= 15.18, PC 0.01). On the other hand, there were uo significant effects for either of the wdking suggestions on the intensity. or’ unpleasantness dimensions. Thus, the major hypothesis of differen’tial effects between hyp notic suggestion in two different Pain dimensions wds supported. l
Consistent with the Malone et al.’ study, we found that different hypnotic suggestions differ-
enMy aft& the two dimensions of pain. Spe ci6dly. we found that Rlrpnotic induction plus ana@esia suggestion r&h.tced the intensity clikxxkrr of pain sign&a&y more than it m&teed the unpleasantness dimension ConVelxeIy, hypnolic induclion plus refaxadon sug gestion reduced the UnpleasaMnerrs dimension ofpaintai@kantiymorechanicreducedthe incenslty dimension. Thiu delnonsuation of d& ferenc pain interventions affecting dinerent dinteaions of pain is consistent with a growing bodyofiicemtureinwhichpainisscudiedasa multidimensional experbce. ‘O-tBs’St Gracely and his coIIeagues found differentiai reductions in pain dimensions with different drugs.*’ Although timccly et al.” found these differemU effects in two different studies, the curtint study a able co compare two different treatments within the same sitidy. Doing this greatly enhanced the validity of the treatment effects. as well as the validity of using independent dimensions of pain in pain rzsearch. If this study had not included two dimensions of pain, it would have appeared that the analgesia suggestion, but ‘not the relaxation suggestion was the most effective pain intervention. This is exactly what has been found by previous researchen who have not used a multidimensional pain assessment technique. Stacher and his colleagues~ found that hyp notic induction plus analgesia suggestion, but not hypnotic induction plus relaxation suggesuon, was effective in reducitqpain. Their pain measure, however, was onJy pain intensity. HilgardyJ used two dimensions of pain, “sensation” and “distress,” but only one kind of intervention, analgesia suggestion; he found that thiu one intervention reduced both dimensions of pain, but that the “sensation” dimension wb9 reduced more than the “dis tress” dimension. It is clear that incomplete exploration of either the multiple ,dimensions ~of pain, or the use of only one type of hyp notic suggestioti, may lead to conflicting and confusing results in laboratory studies. .. Like Malone .tt al.” and Stacher et al.,” we counterbalanced the order of waking and hyp notic conditions and found no significant order effects. like them, we conclude.that the administration of hypnotic induction was responsible for greater pain reduction. Othen have suggestedthat Ihighly susceptible subjects
bring an expectation of greater e&acy for hypnotic pain reduction to the experimental situation. For example, Stam and SpanosY t>redicced that subjects exposed to hypnotic, waking, and analgesia suggestions would refrain from maximally responding during the waking trial so that they could experience and reporc less pain during~ hypnosis,*” the “hold-, back effect.” They believe that the implicit expectation. conveyed to subjects by a repeatek’-measures design is that they shouid reduce pain more when hypnotized than when awake. Stam and Spano8’ found that highly susceptible subjects who were tested for waking analgesia while anticipating a later hyp notic trial reported smaller pain decrements during the waking trial than during the hypnotic trial, in contrast to subjects tested for waking analgesia followed by a second waking analgesia trial. They concluded that the superiority of hypnotic to waking analgesia sugges tiotis found in within-subjects designs is due to the subject’s implicit belief about the comparative effectiveness of the two treatments. Jacobs et al.’ reexamined Scam and Spanos’sy5 notion of a “holdback effect” using more-stringent controls, which minimized experimenter-induced expectations, co examine better the impact of design-induced expectations. In their study, 36 highlv susceotible subjects were assigned to one of ‘three experimental groups ‘(waking analgesia followed by hypnotic analgesia. waking analgesia followed by waking analgesia, a hypnotic analgesia followed by waking analgesia). Each group received three 60-set immersions of cold pressor pain stimulation using a magnitude estimation and a category rating scale. This study. which included a direct comparison of repeated .measures and independent groups, did not support the hypothesis of a “holdback effect.‘* Consistent with previous reports. within-subjects treatment effects were demonstrated for both the, magnitude estimation daLa and the category nting scale data. Irrespective of order, hypnotic analgesia was more effective than waking analgesia, and as expected, no differences were found in subjects’ pain reports on two successive waking analgesia trials. There were also no betweensubject differences in either magnitude estimation or the category rating scale, reflecting
the negative results typically reported for between-subjects design. The existence of a “hoklback effect” would ra%z serious doubts about the interpretation of both the present study and the study bf MaL one et aLs Such design-genitrated “holdback effects” seem unlikely on the basis of. the Jacobs et al.“study. The finding of differenw effectiveness for different hypnotic ‘suggestions has important clinical implications. Price and his colleagues” explored the role of Gutiety in six diiferent categories of pain patients: low back pain, upper back pain, myofascial pain dysfunction, causalgia, cancer pain, and labor pain. Subjects were asked to rate their level of pain intensity and degree of unpleasantness. Their .results indicated that the affective dimension of different types of clinical pain is differentially influenced by psychological context. These observations suggest that a variety of severe pathological and dinical pain may be reduced through the use of hypnotic intervention aimed specifically at the most salient dimension for that pain experience. In addidon, they suggest that the scope of intervention may be extend&d to any pain that either induces’anxiety i! a patient or is influenced by an anxievproducing context. For example, the use of hypnotic induction plus relaxation suggestion may be the most effective intervention for reducing pain in patienlc admitted to an emergency room, when the anxiety from recent trauma is heightened. Alternatively. for patients who express little anxiety over a pr* cedure, such as in some dental procedures, the most effective intervention may he hyp natic induction plus analgesia suggestion. Which intervention is most effective for which type of pain and in which context is an empirical question. Results such as these, which have found that dfierent hypnotic interventions produce difTerential effectiveness for two dimensions of pain, suggest that +udies exploding reduction :of specific pin dimensibns will yield important
clinical information.
w erences 1. Brown D. UinicaI clypnosisresearch since 1986.
in: Fromm E, Nash M. eds. Contemporary hypno& research. New York CuiIford, 1992327458. 2. Climes J. Recent advances in the application of hypnosis ?o pain management Am J Clin Hypn 1994;37:117-129.
3. Malone M, Kurrz R, Strube M. The effects of hypnotic suggestion on pain Hypn 1989,31:2?1-230.
report.
Am J’ Clin
4. Tenenbaum S, Kurtz R, Bienias J. Hypnotic sus-
ceptibility and hypnotic and cognitive shategies for pain control. Am J Clin Hypn 1980;35:40+. 5. Halroyd J. Hypnosis as a methodology in psychological research. IX Fromm E, Nash M, eds. Contemporary hypnosis reselrch. New York: Cuilford, 1992:201-226. 6. Spanos N, MacDonald D. Cwynn M. Instructional set and the relative eflkacy of hypnotic and waking analgesia. Can J Behav .!Zci1988;50:64-72. 7. Miller M, Bowers K HypnoiK: analgesia: dissociated c+erience or dissociated control? J Abnorm Psycho11993;102:29-38. 8. Baker S, Kirxher I. Hypnotic and placebo analgesia: order effects and the placebo label. Contemp Hypn 1993;10:117-126. 9. Jacobs A, Kurtz R, Strube M. Hypnotic ana& sia, expectancy effects, and choice of design: a reexamination. Int J Clin Exp Hypn 1995;42:5!&9. IO. Gracely R. Psychophysical assessmentof human pain. In: Bonica J. ed. Advances in pain research and therapy, ~013. New York: Raven, 1979:805- R24. 11. Chapman C, Casey K, Dubner R, Foley K, Cracrly R. Reading A. Pain mcasurcment: an overview. Pain 1985;22:1-31. 12. Jammer 1, Turslcy B. Discrimination between intensity and afiective pain descripton. a psychological evaluation. Pain 1987;30:271-285. 13. Gracely R McGrath P, Dubner R Ratio scales aud affective verbal pain descriptors. Pain 197Sa;5: .5-18.
14. Lodge M. Magnitude scaling qualitative meawrement &opinions. Beverly IIills, CA: Sage, 1981. 15. Wilson S, B?rber T. The Creatiw Imagination Scale as a measure of hypnotic responsiveriess: applications to experimenial and clinical tiypnosis.’ Am J Clin Hypn 1978;20:235-249. 16. Morgdn A, Hilgard J. The Stanford Hypnotic, Clinical Scale for Adults. Am J Clin Hypn 1979;2I: W-197.
The authors wish to thank Carolyn Letz, Laura Stallings, and Karen Kaminsky for their help a.,research assistants. Thanks also to Dr. Barbara Bremer foi critical cornmen& and Barb Dcnham for teihnicaj assistance.
17. Tart C. Self-report scalesof, hypnotic depth. I~U J Clin Exp CIypn 1970;18:105-125. 18. Weitzenlroffer k, Ililgard E. Sanford Iiypnotic Susceptibility Scale. ,Forms A and B. Palo Alto. CA: Consultinn PsvcholoPistsPress. 1959.
19.&aunanS.\W,11.Hypnosisandpaia contra!. In: Water W II, Smith A Jr, eds. Clinical hypnosis: a multidiscipiinar’y approach. Philadelphix fippincotc, 19g4z23x&3. 2O.DaQrenLAtestofindependencebetween two dimensions of pain. Unpublished masters the!i&L!3t. Louis, MO: wimhington university, 1989. 21. Gracely R, McGrath P, Dubner R Ikdidity and sedtivityofratioscalesofscnscnyaridaffective~r‘bel pin deseriprols: manipulation of affect by diazepam. P&n 197&5:1%Z9. 22. Clark W, Carroll J, Yang J, Jarhl M. Multiditnrmiod scaling reveals two dim,nsions of ther naal pain. f Exp Psychol [Hum Pmep] 1986;12: lo!blo?.
23. Stacher G, Schuster P, Bavet P, Schulze 0. iffects of suggestion of relaxation. or ana@ia on pain threshold and pain tolerance in the waking and hypnotic state. J Psychosom Res 1975;19%9265. 24. Hilgard E. The attexiuation of pain on hypnosis. Pain 1975;1:21%231. 25. Stam H, Spanos N. Experimental designs. expectancy effects. and hypnotic analgesia. J Abnorm Psycho11980;89:751-762. 26. Pri-e D, Hawkins S. Baker C. Sensory-affective relations+8 among different type of clinical and experimental pain. Pain 1987;2%:297-307.