ORIGINAL ARTICLE

A randomized controlled trial of hypnosis compared with biofeedback for adults with chronic low back pain G. Tan1, D.H. Rintala2, M.P. Jensen3, T. Fukui4, D. Smith4, W. Williams4,5 1 2 3 4 5

Department of Psychology, National University of Singapore, Singapore Department of Physical Medicine and Rehabilitation, Baylor College of Medicine, Houston, USA Department of Rehabilitation Medicine, University of Washington, Seattle, USA Michael E. DeBakey Veterans Affairs Medical Center, Houston, USA The Menninger Department of Psychiatry and Behavioral Sciences, Baylor College of Medicine, Houston, USA

Correspondence Diana H. Rintala E-mail: [email protected] Funding source Grant number D4421-I from the Veterans Health Administration Rehabilitation Research and Development Service. Conflict of interest Mark P. Jensen is an author of two books on the topic of this paper and receives royalties for the sales of those books. The other authors declare no conflict of interest. Accepted for publication 29 April 2014 doi:10.1002/ejp.545

Abstract Background: Chronic low back pain (CLBP) is common and results in significant costs to individuals, families and society. Although some research supports the efficacy of hypnosis for CLBP, we know little about the minimum dose needed to produce meaningful benefits, the roles of home practice and hypnotizability on outcome, or the maintenance of treatment benefits beyond 3 months. Methods: One hundred veterans with CLBP participated in a randomized, four-group design study. The groups were (1) an eight-session self-hypnosis training intervention without audio recordings for home practice; (2) an eight-session self-hypnosis training intervention with recordings; (3) a two-session self-hypnosis training intervention with recordings and brief weekly reminder telephone calls; and (4) an eight-session active (biofeedback) control intervention. Results: Participants in all four groups reported significant pre- to post-treatment improvements in pain intensity, pain interference and sleep quality. The hypnosis groups combined reported significantly more pain intensity reduction than the control group. There was no significant difference among the three hypnosis conditions. Over half of the participants who received hypnosis reported clinically meaningful (≥30%) reductions in pain intensity, and they maintained these benefits for at least 6 months after treatment. Neither hypnotizability nor amount of home practice was associated significantly with treatment outcome. Conclusions: The findings indicate that two sessions of self-hypnosis training with audio recordings for home practice may be as effective as eight sessions of hypnosis treatment. If replicated in other patient samples, the findings have important implications for the application of hypnosis treatment for chronic pain management.

1. Introduction Low back pain is one of the most common chronic pain problems, and results in significant costs to individuals, their families and society (Licciardone, 2008; Lew et al., 2009). Both case studies (King et al., 2001; Tan et al., 2010) and some controlled trials support the © 2014 European Pain Federation - EFIC®

efficacy of hypnosis for chronic low back pain (McCauley et al., 1983; Edelson and Fitzpatrick, 1989; Crawford et al., 1998). Control conditions for these studies have included treatments such as progressive relaxation, cognitive behaviour therapy and attention. Surface electromyography (sEMG) biofeedback is another reasonable active control for hypnosis Eur J Pain 19 (2015) 271–280

271

RCT of hypnosis versus biofeedback for CLBP

What’s already known about this topic? • Several studies support the efficacy of hypnosis for chronic low back pain but several questions remain. What does this study add? • This study assessed the value of self-hypnosis practice, the role of hypnotizability, the effect of number of sessions with a therapist, and the maintenance of benefits after treatment ends.

G. Tan et al.

treatment reductions in pain intensity than those in the biofeedback control condition. The secondary hypotheses were that hypnosis treatment would be more effective than the control condition for reducing pain interference and improving sleep quality. We also assessed (1) the role of the number of therapist-guided hypnosis treatments; (2) the maintenance of treatment effects for 6 months following treatment; (3) the role of hypnotizability; and (4) the number of home self-hypnosis practice sessions.

2. Methods because it involves face-to-face treatment sessions, includes relaxation and focused attention, and has been found to be somewhat effective in improving pain, pain interference and sleep quality (Flor and Birbaumer, 1993; Newton-John et al., 1995; Jensen et al., 2009b; Ehrenborg and Archenholtz, 2010; Yilmaz et al., 2010). Jensen (Jensen et al., 2009a) compared the effect of hypnosis and sEMG biofeedback on chronic pain in persons with spinal cord injury and found hypnosis to be more effective in reducing average daily pain. However, to our knowledge, there are no studies comparing hypnosis to sEMG biofeedback for chronic low back pain (CLBP). There also remain a number of unanswered questions regarding the effectiveness and mechanisms of hypnosis for CLBP management. First, what is the minimum number of therapist-guided sessions necessary to obtain the desired results? Most chronic pain studies have used 3–10 sessions and eight sessions have been found to be effective (McCauley et al., 1983) for chronic low back pain. Our pilot study (Tan et al., 2010) found four sessions to be effective for CLBP; can benefits be obtained in a very small number of sessions? Subjects find it increasingly expensive and difficult to attend numerous therapy sessions and to our knowledge, no previous chronic pain and hypnosis studies used only two sessions. Second, what is the role of home practice of self-hypnosis? Third, how long will the benefits last after treatment ends? Fourth, what is the effect of hypnotizability? Finally, what is the effect of the amount of home practice? This study was designed to address these questions in the context of a controlled trial where patients with CLBP were randomly assigned to one of four treatment conditions – three involving hypnosis and one involving sEMG biofeedback as an active control. Based on Jensen and colleagues’ (Jensen et al., 2009a) findings in persons with spinal cord injury, the primary study hypothesis was that participants in the hypnosis conditions would report greater pre- to post272 Eur J Pain 19 (2015) 271–280

2.1 Study design The study used a randomized, single-blind, four-group design. The four treatment conditions were (1) eight therapist-guided sessions of self-hypnosis training without recommendations for practice (HYP-8); (2) eight therapistguided sessions of self-hypnosis training with recommendations for practice (HYP-PRAC-8); (3) two therapist-guided sessions of self-hypnosis training with recommendations for practice (HYP-PRAC-2) plus six brief, weekly telephone calls; and (4) eight sessions of sEMG biofeedback-assisted relaxation training (BIO-8) that served as the control group.

2.2 Participants Participants were recruited from the Michael E. DeBakey VA Medical Center (MEDVAMC) Pain Management Program via posters, the staff of the programme and advertisement in a veterans’ newsletter. Recruitment began in December 2006 and ended in July 2009. The follow-up was completed by March 2010. Our sampling frame comprised 888 names of potential participants (Fig. 1). We included subjects with (1) CLBP of at least 6 months’ duration; (2) average pain rating ≥ 5 on a 0–10 scale during the past week; (3) pain with a primarily musculoskeletal or mechanical aetiology; and (4) agreement to adhere to study protocol. We excluded subjects with (1) acute pain due to recent injury or pain due to cancer; (2) primarily neuropathic pain aetiology; (3) severe psychopathology, significant cognitive deficits or severe hearing loss that would interfere with participation; (4) an active substance abuse problem; and (5) participation in our previous pilot study on hypnosis and CLBP (Tan et al., 2010). See Fig. 1 for a detailed study flow diagram. One hundred participants completed the full 8-week treatment protocol and 59 withdrew after randomization – 33 (56%) before treatment began and 26 (44%) after participating in at least one session. The average age of the participants who completed the study was 55 years (range = 25–83 years); 79% were male; 47% described their race/ethnicity as Caucasian, 32% as African-American, 15% as Hispanic and 6%, as other or mixed. The 59 participants who withdrew were not significantly different than the completers in age (56 years, t = 0.62; p = 0.534), sex (86% male;

© 2014 European Pain Federation - EFIC®

G. Tan et al.

RCT of hypnosis versus biofeedback for CLBP

Did not meet inclusion criteria (n = 384)

Figure 1 Study flow diagram. Note: B, number who withdrew before the first session and reasons they withdrew. Some participants had more than one reason for withdrawing. Ninety percent of the sampling frame (n = 797) was contacted at least once. Of those contacted, 48% did not meet the inclusion criteria, 28% declined participation, 4% met the criteria but were put on a waiting list and never randomized and 20% were randomized to a treatment group. Pain of a primarily neuropathic nature was the most common reason for exclusion (58%) and lack of interest, lack of follow through and living too far from the medical centre were the most common reasons for declining participation (20%, 18% and 16%, respectively).

Neuropathic pain (n = 224, 58.3%) Cogni ve impairment (n = 48, 12.5%) Pain intensity ra ng < 5 (n = 38, 9.9%) Substance abuse (n = 33, 8.6%) Psycho c (n = 15, 3.9%) Cancer pain (n = 10, 2.6%) Suicide idea on (n = 5, 1.3%) Disability claim pending (n = 5, 1.3%) Pain not chronic, < 6 months (n = 2, 0.5%) Severely hearing impaired (n = 2, 0.5%) In previous hypnosis study (n = 2, 0.5%)

Unable to contact (n = 90) Deceased (n = 1) Assessed for eligibility (n = 797)

Wait listed, never randomized due to limita ons of therapist’s schedule (n = 28) Randomized (n = 159)

Declined to par cipate (n = 226) Not interested (n = 45, 19.9%) Did not follow through (n = 41, 18.1%) Lived too far away (n = 35, 15.5%) No reason recorded (n = 21, 9.3%) Scheduling problems (n = 17, 7.5%) Desires or receiving different treatment (n = 12, 5.3%) Li ga on pending (n = 10, 4.4%) Lacks transporta on (n = 9, 4.0%) Wary of hypnosis or advised not to par cipate (n = 8, 3.5%) Mental health problems (n = 7, 3.1%) Has moved or soon moving away (n = 6, 2.7%) High cost of travel to medical center (n = 4, 1.8%) Insufficient compensa on (n = 3, 1.3%) Desired delay, then unable to contact (n = 3, 1.3%) Didn’t come to first appointment (n = 2, 0.9%) Misc. (n = 3, 1.3%)

HYP-8 (n = 47) Completed interven on (n = 25) Withdrew (n = 22; B = 12) 6 lost contact (B = 5) 5 back pain intensity < 5 (B = 5) 4 me/scheduling constraints 3 wants a different treatment 2 transporta on problems (B = 1) 1 each – substance abuse (B), advised by pastor not to par cipate (B), unrelated legal problems (B), can’t sit s ll long enough, treatment isn’t helping, home prac ce causes too much stress, hearing impaired

HYP-PRAC-8 (n = 39) Completed interven on (n = 25) Withdrew (n = 14; B = 9) 3 me/scheduling constraints (B = 1) 3 wary of hypnosis (B = 2) 2 back pain < 5 (B = 2) 2 pa ern of missed appointments (B = 1) 1 each – lives too far away (B), transporta on problem (B), lost contact (B), unrelated health problem (B), in jail (B), can’t sit s ll long enough, had disturbing images with hypnosis

HYP-PRAC-2 (n = 35) Completed interven on (n = 25) Withdrew (n = 10; (B = 6) 4 back pain < 5 (B = 2) 2 can’t sit s ll long enough (B = 1) 1 each – lives too far away (B), having back surgery (B), advised by surgeon not to par cipate (B), substance abuse, had bad dreams a er hypnosis.

BIO-8 (n = 38) Completed interven on (n = 25) Withdrew (n = 13; B = 6) 5 me/scheduling constraints (B = 2) 2 high cost of travel to medical center (B = 1) 2 deceased (B = 2) 1 each – back pain < 5 (B), lives too far away, treatment isn’t helping, pa ern of missed appointments, believed biofeedback was the control group

Analysed for pre-post change (n = 25)

Analysed for pre-post change (n = 25)

Analysed for pre-post change (n = 25)

Analysed for pre-post change (n = 25)

Lost to 6-month follow up (n = 10) 10 unable to contact

Lost to 6-month follow up (n = 1) 1 unable to contact

Lost to 6-month follow up (n = 3) 2 mailed packet not returned 1 unable to contact

Analysed for follow-up (n = 15)

Analysed for follow-up (n = 24)

Analysed for follow-up (n = 22)

χ2 = 1.38; p = 0.291) or race/ethnicity (51% Caucasian, 40% African-American, 7% Hispanic and 2% other or mixed race/ethnicity; χ2 = 4.57; p = 0.206).

2.3 Procedure Potentially eligible participants met with the coordinator at the MEDVAMC. Eligible participants read and signed a consent form approved by the Institutional Review Board for Human Subject Research for Baylor College of Medicine and Affiliated Hospitals. Research staff then collected information about pain, medication, health history and demographics. Participants were then tested for hypnotizability and completed the baseline outcome measures (described in the Measures section). Participants were randomly assigned by the therapist to one of the four conditions using a table of random numbers. The study coordinator did not know to which group the participants were assigned. Participants assigned to the HYP-8, HYP-PRAC-8 and BIO-8 groups participated in a series of eight weekly face-toface sessions. Participants assigned to the HYP-PRAC-2 group participated in two face-to-face training sessions, followed by six weekly brief (average = 10 min) telephone interviews to problem-solve and encourage practice. Each participant in the three hypnosis groups was administered seven hypnotic suggestions (described below) during the first two treatment sessions. They selected the two suggestions to which they responded the most after sessions 1 and 2. These two suggestions were used in the remaining face-to-face sessions for the HYP-8 and HYP-PRAC-8 groups, and were included in audio recordings made for the participants in the HYPPRAC-8 and HYP-PRAC-2 conditions. Participants in the HYP-PRAC-8 and HYP-PRAC-2 groups were asked to practice both with and without the audio recordings throughout the 8-week treatment period. All participants in the HYP-

© 2014 European Pain Federation - EFIC®

Sampling frame (n = 888)

Lost to 6-month follow up (n = 7) 4 unable to contact 2 mailed packet not returned 1 unrelated health problem Analysed for follow-up (n = 18)

PRAC-8 and HYP-PRAC-2 groups were asked to complete daily practice diaries to assess compliance with home practice instructions. Participants assigned to the BIO-8 control group received eight sessions of sEMG-assessed relaxation training in which sEMG electrodes were placed on the frontalis muscles on the forehead and muscle activity was assessed and fed back to the participant via a computer screen. Participants were instructed to watch the computer screen and use the visual and auditory feedback to help them learn how to decrease their frontalis muscle tension. They were told that an ability to decrease frontalis muscle tension would generalize to their back muscles, so that they would experience a sense of global relaxation and pain relief. At 1 week and 6 months after treatment ended, the coordinator again administered the outcome measures to participants who completed the training.

2.4 Measures Screening measures assessed demographics, average pain intensity, previous and current pain treatments (including hypnosis), confirmation diagnosis of CLBP and other related medical history. The recruitment interview also included the Short Blessed Test (Katzman et al., 1983) to screen for significant cognitive impairment, the Mini-International Neuropsychiatric Interview (Sheehan et al., 1998) to screen for significant psychopathology and substance abuse problem(s), and The Self-Report Leeds Assessment of Neuropathic Symptoms and Signs scale (Bennett et al., 2005) to screen for pain primarily neuropathic in nature. Outcome measures completed at pre- and post-treatment and at 6-month follow-up included a modified version of the Brief Pain Inventory (Tyler et al., 2002) to assess pain intensity and pain interference and the Pittsburgh Sleep Quality

Eur J Pain 19 (2015) 271–280

273

RCT of hypnosis versus biofeedback for CLBP

Index (Buysse et al., 1989). Outcome predictors included global hypnotizability assessed at pre-treatment using the 5-item Stanford Clinical Hypnotizability Scale (Hilgard and Hilgard, 1994) and amount of self-hypnosis practice assessed by participant diaries. Descriptions of the standardized measures are available only online as Supporting Information Method S1.

2.5 Hypnotic suggestions Participants assigned to the hypnosis treatment groups were administered a standard relaxation induction followed by seven suggestions in the first two sessions of treatment. The hypnotic suggestions focused on deep relaxation, sensory substitution, pain intensity reduction, imagined anesthesia, decreased pain unpleasantness, managing breakthrough pain and post-hypnotic suggestions for effective selfhypnosis (Jensen, 2011). Each subject was allowed to pick two favourite suggestions to individualize their hypnotic scripts which were repeated to them for sessions 2–8.

2.6 Statistical analysis Sample size was determined based on the results of two previous studies that included the outcome measures for pain intensity (Tan et al., 2010), pain interference (Tan et al., 2010) and sleep quality (Buysse et al., 1989). The outcome measure that required the largest sample size (20 per group) was sleep quality. Allowing for an estimated 20% attrition rate and several secondary hypotheses, a recruitment sample size of 40 participants for each of the four groups was selected.

2.6.1 Pre- to post-treatment effects of hypnosis versus biofeedback For testing the primary and secondary hypotheses, participants in the three hypnosis treatment conditions were combined into one hypnosis group (ALL-HYP). We calculated means and standard deviations for pre- and post-treatment. We assessed pretreatment group equivalency with t-tests. We performed repeated-measures analyses of variance (ANOVAs) to test the primary and secondary hypotheses comparing the effects of hypnosis and biofeedback on pain intensity, pain interference and sleep quality. We performed a responder analysis (Dworkin et al., 2008) by computing the percentage of participants in each treatment group who reported at least a 30% decrease in pain intensity. We calculated chi-square and risk estimates for this binary outcome.

2.6.2 Effects of hypnosis dose and practice recommendations We performed a series of repeated-measures ANOVAs with hypnosis treatment condition (HYP-8, HYP-PRAC-8, HYP-

274 Eur J Pain 19 (2015) 271–280

G. Tan et al.

PRAC-2) as the independent variable and the three outcome measures as the dependent variables. In the event of significant between-group differences, we performed a series of paired t-tests for the relevant outcome variable(s). We also computed the percentage of participants in each hypnosis group who reported at least a 30% decrease in pain intensity. We calculated chi-square and risk estimates comparing responder rates for the HYP-8 group with those for a combined ‘practice’ group (HYP-PRAC-8 and HYP-PRAC-2).

2.6.3 Maintenance of treatment effects We performed a series of repeated-measures ANOVAs for the three outcome measures, with both group (hypnosis vs. control) and time (pretreatment, post-treatment, 6-month follow-up) as the independent variables. We also performed paired t-tests for each outcome within each group to assess change from pre- to post-treatment and, separately, from post-treatment to the 6-month follow-up point. We performed similar analyses comparing maintenance effects for the three hypnosis groups. Effect sizes were computed for all analyses using Cohen’s d for t-tests and partial eta squared (ηp2) for ANOVAs. Description of the analyses for hypnotizability and number of home practice sessions are available only online as Supporting Information Method S2.

3. Results 3.1 Effects of hypnosis on treatment outcomes Means and standard deviations for the outcome measures at pre- and post-treatment are presented in Table 1 for the ALL-HYP and BIO-8 groups. There was no significant baseline difference between the two groups on any outcome measure. The paired t-tests for within-group change indicated that both groups had significant improvements in all three outcome measures from pre- to post-treatment. There were large effect sizes (Cohen’s ds) for pain intensity and interference for the ALL-HYP group and medium effect sizes for the BIO-8 group (as per Cohen, 1988). For sleep quality, the effect sizes were medium for both groups. The repeated-measures ANOVA indicated there were large (as per Cohen, 1992) and statistically significant time main effects for all three outcomes, indicating that both groups improved from pre- to post-treatment. For pain intensity, there was a small, statistically significant time × group interaction effect. There was greater reduction in pain intensity in the ALL-HYP group than in the BIO-8 group. For pain interference the interaction effect yielded a nonsignificant trend (p = 0.050). For sleep quality, the interaction effect was not statistically significant. Clinically meaningful reductions (≥30%) in pain intensity © 2014 European Pain Federation - EFIC®

0.086

0.179

0.37 (1, 98)

0.77 (1, 98)

© 2014 European Pain Federation - EFIC®

Pain interference (BPI) Sleep quality (PSQI)

were reported by 39 (52%) of the 75 hypnosis participants and 9 (36%) of the 25 BIO-8 participants. Although participants in the ALL-HYP group, were 1.44 times more likely to have at least a 30% reduction in pain intensity, this difference in responder rates was not statistically significant [χ2 = 1.92; p(1-sided) = 124; odds ratio = 1.93 (CI = 0.76–4.90); relative risk = 1.44 (CI = 0.82–2.54); absolute risk = −0.16]. ALL-HYP, participants who received hypnosis treatment; BIO-8, participants who received biofeedback; BPI, Brief Pain Inventory; PSQI, Pittsburgh Sleep Quality Index. *p < 0.10. **p < 0.05. ***p < 0.001.

0.009 0.87 (1, 98) 0.157 18.31*** (1, 98)

0.039 3.94* (1, 98) 0.268 35.93*** (1, 98)

0.042 4.29** (1, 98) 0.306 0.394

5.52 6.05 19.70 18.31 3.99 4.02 75 25 75 25 75 25 ALL-HYP BIO-8 ALL-HYP BIO-8 ALL-HYP BIO-8 Pain intensity (BPI)

23.29 25.50 54.29 55.94 10.95 10.24

1.69* (1, 98)

14.99 21.18 33.32 45.40 8.83 8.88

9.43 9.22 27.02 23.67 4.57 3.99

8.46*** (1, 74) 2.79** (1, 24) 7.78*** (1, 74) 2.52** (1, 24) 5.08*** (1, 74) 2.11** (1, 24)

1.074 0.554 0.887 0.498 0.494 0.340

43.15*** (1, 98)

ηp2 F (df) ηp2 Cohen’s d t (df) SD Mean t (df) SD Mean n Group Outcome measure

Pretreatment

Cohen’s d

Time effect Post-treatment

Paired t-test for within-group change Effect size for baseline differences t-test for between-group baseline differences

Table 1 Comparison of the effect of hypnosis versus biofeedback from pre- to post-treatment.

F (df)

Effect size for time × group effect Effect size for time effect

Time × group effect

RCT of hypnosis versus biofeedback for CLBP

Effect size for paired t-test

Repeated-measures ANOVA

G. Tan et al.

3.2 Effects of hypnosis dose and recommendations for practice The means and standard deviations for each of the three hypnosis groups are displayed in Table 2. The results of a one-way ANOVA for between-group differences indicated that there was a significant baseline difference among the three groups for pain intensity. There was no significant baseline difference among the three groups for pain interference or sleep quality. The paired t-tests, which examined changes within each hypnosis group from pre- to post-treatment, showed that all three groups had significant improvements in all three outcome measures. Repeated-measures ANOVAs revealed a large time effect of hypnosis for all three outcome measures but there was no significant time × group interaction effect indicating that the improvement in these three variables did not differ significantly among the three hypnosis groups. Clinically meaningful reductions (at least 30%) in pain intensity was reported by 10 (40%) in the HYP-8 group, 13 (52%) in the HYP-PRAC-8 group and 16 (64%) in the HYP-PRAC-2 group. Fifty-eight percent of the participants in the combined practice groups (HYP-PRAC-8 and HYP-PRAC-2) were responders compared with 40% for the HYP-8 group. [χ2 = 2.163; p(1-sided) = 0.110; odds ratio = 2.07 (CI = 0.78–5.51); relative risk = 1.45 (CI = 0.85–2.48); absolute risk = −0.18]. Although these results were not statistically significant, participants in the combined practice groups were 1.45 times more likely to have a 30% or greater improvement in pain intensity than those in the HYP-8 group.

3.3 Maintenance effects Fig. 2A presents the means for pain intensity at pretreatment, post-treatment and the 6-month follow-up for the 79 participants (61 in the ALL-HYP group and 18 in the BIO-8 group) who provided data at all three time points. The ANOVA indicated a large, significant time main effect (F = 24.50; p < 0.001; ηp2 = 0.241) and a non-significant time × group interaction effect (F = 2.18; p = 0.116; ηp2 = 0.028). Fig. 2B presents Eur J Pain 19 (2015) 271–280

275

Group

276 Eur J Pain 19 (2015) 271–280

25 25 25 25 25 25 25 25 25

n

F (df)

24.28 5.44 3.18* (2, 72) 24.50 5.52 21.08 5.14 61.20 17.55 2.50 (2, 72) 52.06 19.62 49.60 20.65 11.36 3.55 2.02 (2, 72) 11.80 4.08 9.68 4.15

Mean SD

Pretreatment

0.053

0.065

0.081

ηp2 Cohen’s d 0.810 1.267 1.179 0.911 1.012 0.889 0.425 0.953 0.466

t (df) 3.64** (1, 24) 6.02** (1, 24) 5.23** (1, 24) 4.20** (1, 24) 4.89** (1, 24) 4.36** (1, 24) 2.07* (1, 24) 4.75** (1, 24) 2.30* (1, 24)

SD 10.24 8.78 9.04 29.39 25.60 25.99 4.37 4.31 4.92

Mean 17.52 14.16 13.28 39.08 31.88 29.00 9.96 8.72 7.80

0.05 (2, 72)

1.45 (2, 72)

58.98** (1, 72) 0.450

26.14** (1, 72) 0.266

F (df) 1.18 (2, 72)

ηp2

0.039

0.001

0.032

ηp2

Effect size for Effect Size for Time × group time × group Time effect effect effect

71.92** (1, 72) 0.500

F (df)

Paired t-test for within-group Effect size for change paired t-test Time effect

Repeated-measures ANOVA

Sleep quality index (mean ± SE) BPI pain interference (mean ± SE)

BPI pain intensity (mean ± SE)

BPI, Brief Pain Inventory; HYP-8, participants who received eight face-to-face hypnosis treatments; HYP-PRAC-8, participants who received eight face-to-face hypnosis treatments and recommendations for home practice; HYP-PRAC-2, participants who received two face-to-face hypnosis treatments and recommendations for home practice; PSQI, Pittsburgh Sleep Quality Index. *p < 0.05. **p < 0.001.

HYP-8 HYP-PRAC-8 HYP-PRAC-2 Pain interference HYP-8 (BPI) HYP-PRAC-8 HYP-PRAC-2 Sleep quality HYP-8 (PSQI) HYP-PRAC-8 HYP-PRAC-2

Pain intensity (BPI)

Outcome measure

ANOVA for between-group Effect size for baseline baseline differences differences Post-treatment

Table 2 Comparison of the effect of hypnosis on the three hypnosis groups from pre- to post-treatment.

RCT of hypnosis versus biofeedback for CLBP G. Tan et al.

similar results for pain interference. The ANOVA indicated a large and significant time main effect (F = 16.68; p < 0.001; ηp2 = 0.178) and a small, significant time × group interaction effect (F = 3.24; p = 0.042; ηp2 = 0.040). Fig. 2C presents the results for sleep quality. The ANOVA revealed a large and statistically significant time main effect (F = 11.22; p < 0.001; ηp2 = 0.127) and a non-significant time × group interaction (F = 1.67; p = 0.192; ηp2 = 0.021). The results of paired t-tests for pretreatment to post-

A

B

C

40 Pain intensity

30 23.5

26.0

20

50

Pretreatment

21.6

10

56.8

25

10.6

7

ALL-HYP

20.8

14.7 16.3

0

100 Pain interference

75

52.8 49.6 46.0

31.0 34.9

0

21 Sleep quality

14 11.2

9.4 8.4 10.7

8.1

0

Post-treatment

BIO-8

6-Month follow-up

Figure 2 Comparison of ALL-HYP (n = 61) and BIO-8 (n = 18) groups for maintenance of treatment effects on (A) pain intensity, (B) pain interference and (C) sleep quality over 6 months. Note: ALL-HYP, participants who received hypnosis treatment; BIO-8, participants who received biofeedback; BPI, Brief Pain Inventory.

© 2014 European Pain Federation - EFIC®

G. Tan et al.

RCT of hypnosis versus biofeedback for CLBP

A 40

BPI pain intensity (mean ± SE)

Pain intensity 30 25.3 24.3 20

18.4

21.5

13.9 12.9

10

19.3 17.3 13.3

0

B BPI pain interference (mean ± SE)

100 Pain interference 75

50

62.5 51.1 48.2

37.9 31.6

25

42.9 38.6 25.3

25.6

4. Discussion

0

Sleep quality Index (mean ± SE)

C

three time points. The ANOVA indicated a large and significant time main effect (F = 34.91; p < 0.001; ηp2 = 0.376) and a non-significant time × group interaction effect (F = 0.64; p = 0.640; ηp2 = 0.022). Fig. 3B presents similar results for pain interference. The ANOVA indicated a large, significant time main effect (F = 33.54; p < 0.001; ηp2 = 0.366) and a nonsignificant time × group interaction effect (F = 0.76; p = 0.556; ηp2 = 0.025). Fig. 3C presents the means for sleep quality. The ANOVA indicated a large and statistically significant time main effect (F = 17.51; p < 0.001; ηp2 = 0.232) and a non-significant time × group interaction effect (F = 0.49; p = 0.747; ηp2 = 0.016). The results of paired t-tests from pretreatment to post-treatment and post-treatment to follow-up for Fig. 3A–C are available only online as Supporting Information Results S1. The results for hypnotizability and number of home practice sessions are available only online as Supporting Information Results S2 and Supporting Information Table S1.

21

4.1 Effects of hypnosis on treatment outcomes

Sleep quality

14 11.7 10.7

8.8

9.4

8.7

7

7.0

9.6 9.1 6.6

0 Pretreatment HYP-8

Post-treatment HYP-PRAC-8

6-Month follow-up HYP-PRAC-2

Figure 3 Comparison of HYP-8 (n = 15), HYP-PRAC-8 (n = 24) and HYPPRAC-2 (n = 22) groups for maintenance of treatment effects on (A) pain intensity, (B) pain interference and (C) sleep quality over 6 months. Note: BPI, Brief Pain Inventory; HYP-8, participants who received eight face-toface hypnosis treatments; HYP-PRAC-8, participants who received eight face-to-face hypnosis treatments and recommendations for home practice; HYP-PRAC-2, participants who received two face-to-face hypnosis treatments followed by six weekly brief telephone calls and recommendations for home practice.

treatment and post-treatment to follow-up for Fig. 2A–C are available only online as Supporting Information Results S1. Fig. 3A presents the means for pain intensity at pretreatment, post-treatment and the 6-month follow-up for the three groups of hypnosis participants (n = 15 in the HYP-8 group; 24 in the HYP-PRAC-8 group and 22 in the HYP-PRAC-2 group) who provided data at all © 2014 European Pain Federation - EFIC®

We found support for the study hypothesis that participation in self-hypnosis training would result in greater reductions in pain intensity than participation in a biofeedback intervention. We did not find support for the hypotheses that hypnosis would be more effective than biofeedback for reducing pain interference and improving sleep quality. The results are generally consistent with the literature and provide further albeit limited support to the efficacy of hypnotic analgesia (e.g. Montgomery et al., 2000; Patterson and Jensen, 2003). When interpreting the findings, it is important to keep in mind that we used an active control condition. Compared with wait-listed or usual treatment control groups, the use of active controls makes it more difficult to find significant treatment effects (Williams et al., 2012; web reference). Treatment involving sEMG biofeedback has been shown to be efficacious for pain management (e.g. Tan et al., 2003). Moreover, Patterson and Jensen (2003) have argued that relaxation training contains many ‘hypnotic-like’ components and noted that response to relaxation training has been found to be associated with hypnotizability. Our biofeedback intervention contained two of the basic components of hypnotic treatment (focused awareness, plus suggestions), making it arguably a type of hypnotic intervention. The lack of ongoing additional suggestions during sEMG biofeedback may explain, at least in part, why Eur J Pain 19 (2015) 271–280

277

RCT of hypnosis versus biofeedback for CLBP

the biofeedback treatment was somewhat less effective than the hypnosis treatments. This small but consistent difference in efficacy between biofeedback and hypnosis has also been shown by others (Jensen et al., 2009a). Future investigators might consider alternative control conditions for hypnosis trials, such as patient education or cognitive therapy, both of which tend to have minimal impact on pain intensity (Jensen et al., 2011).

4.2 Clinical improvement in outcomes To address questions regarding clinical improvement, we performed responder analyses as recommended by the IMMPACT group (Dworkin et al., 2008). We found that clinically meaningful reductions (at least 30%) in pain intensity were reported by 52% of the hypnosis participants and 36% of the biofeedback participants. Although this difference was not statistically significance, the direction lends support to the ANOVA analysis.

4.3 Effects of hypnosis dose and practice effects We also explored the dose and practice effects of hypnosis. Improvement in pain intensity, pain interference and sleep quality did not differ significantly among the three hypnosis groups. This finding suggests that effective self-hypnosis training can be delivered at a very low cost (e.g. as little as two face-to-face sessions with a therapist) along with hypnosis audio recordings for home practice. We know of no other psychosocial pain intervention that can be effectively provided in as few as two sessions. Although the results for clinically meaningful pain intensity reduction (i.e. ≥30%) were not statistically significant in this sample, the direction of the results suggests that home practice may improve the results. Further research is needed.

4.4 Maintenance of treatment effects Our study also assessed the longer term (up to 6 months) maintenance of the treatment effects. In general, we found that the beneficial effects of hypnosis (both in terms of reductions in pain intensity and pain interference) were maintained for at least 6 months post-treatment for all three hypnosis groups. Moreover, there was no difference in maintenance between the three hypnosis groups. The finding of maintenance of benefits for at least 6 months posttreatment is consistent with one previous study 278 Eur J Pain 19 (2015) 271–280

G. Tan et al.

(Jensen et al., 2009a) showing long-term maintenance of hypnotic analgesic benefits in a sample of individuals with chronic pain and disability. Discussion of the results for hypnotizability and frequency of practice is available online only as Supporting Information Discussion S1.

4.5 Limitations The study is limited by the fact that it is primarily targeted on male veterans with a long-standing history of generally severe chronic musculoskeletal low back pain. Thus, the findings do not necessarily generalize to non-veterans or to individuals with low back pain of shorter duration. Second, we excluded individuals with evidence for neuropathic pain, further limiting the generalizability of the findings. Recent research suggests that hypnosis may be even more effective for neuropathic pain than it is for non-neuropathic pain (Jensen et al., 2009a). Future researchers should examine the potential moderating influence of pain type on outcome. Third, there was a high dropout rate (37%) among the 159 participants who were randomized to a treatment group. More than half (56%) of the 59 persons who dropped out did so before attending any treatment sessions. Although the withdrawal rate in our study is high, it is comparable to another randomized controlled study at another Veterans Affairs medical centre (VAMC) examining the relative effectiveness of behavioral therapy for chronic heart failure (Chang et al., 2004; web reference). After consenting to participate, 29% withdrew before beginning treatment. Many veterans live a fairly long distance from VAMCs. Common reasons for withdrawing from research studies that require many visits include lack of transportation, time and effort needed to attend, and financial costs of travel. Thus, higher dropout rates may occur more frequently in VAMC research studies with high participant burden conducted at veterans’ medical centres than in research conducted at sites where most of the participants live nearby.

4.6 Strengths In spite of the limitations, this study makes important contributions to the field of pain management for persons with CLBP. First, this study is one of very few randomized controlled trials of self-hypnosis for this population. Second, we examined the efficacy of a very brief (two-session) treatment, thus addressing the issue of the cost of treatment. Third, we assessed long-term (6 months) outcomes. © 2014 European Pain Federation - EFIC®

G. Tan et al.

Conclusions This study demonstrated that individuals with CLBP could benefit from learning self-hypnosis. Specifically, significant and substantial reductions in pain intensity, pain interference and improvement in sleep quality were found for those assigned to each of the three hypnosis treatment conditions. It is also noteworthy that this is the first published study examining the dose effect of self-hypnosis training, and we found that this treatment could be effectively delivered with very limited resources; specifically, with as few as two faceto-face sessions. This study also replicated the finding that it is possible to carry out a standardized self-hypnosis training protocol that allows for treatment tailoring, in this case with individualized hypnotic suggestions (Jensen et al., 2009a,b; Tan et al., 2010). Chronic pain remains a major issue related to significant financial costs and suffering (Institute of Medicine, 2011); consequently, this finding is important for the applicability of hypnosis to clinical populations. Given the minimal cost, the lack of significant adverse effects of this noninvasive treatment, as well as its demonstrated efficacy, our findings suggest that self-hypnosis might be considered as a viable first-line treatment for the management of CLBP.

Author contributions Gabriel Tan: PI of the research project. Took primary responsibility for the design and implementation of the study and for the design and interpretation of the data analyses. Wrote the first draft of the Introduction, Methods and Discussion sections, and provided significant editorial input into the Results section; Diana H. Rintala: Co-PI of the research project. Provided significant input into study design and performed the bulk of the data analyses. Provided significant input into the interpretation of the results and provided significant editorial input into the entire paper; Mark P. Jensen: Provided significant input into the study design, data analysis plan, and interpretation of the results, and provided significant editorial input into the entire paper; Tenley Fukui: Provided significant input into the study design, wrote the treatment manuals in consultation with co-authors, was the study clinician, and provided significant editorial input into the entire paper; Donna Smith: Coordinated the research project and provided significant input into the development and implementation of the protocols; Wright Williams: Provided significant editorial input into the entire paper, and served as site PI for the completion of the study. All authors have reviewed and are in agreement with the submission of this article for publication in the European Journal of Pain.

© 2014 European Pain Federation - EFIC®

RCT of hypnosis versus biofeedback for CLBP

References Bennett, M.I., Smith, B.H., Torrance, N., Potter, J. (2005). The S-LANSS score for identifying pain of predominantly neuropathic origin: Validation for use in clinical and postal research. J Pain 6, 149–158. Buysse, D.J., Reynolds, C.F. III, Monk, T.H., Berman, S.R., Kupfer, D.J. (1989). The Pittsburgh Sleep Quality Index: A new instrument for psychiatric practice and research. Psychiatry Res 28, 193–213. Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences, 2nd edn (Hillsdale, NJ: Lawrence Earlbaum Associates). Cohen, J. (1992). A power primer. Psychol Bull 112, 155–159. Crawford, H.J., Knebel, T., Kaplan, L., Vendemia, J.M., Xie, M., Jamison, S., Pribram, K.H. (1998). Hypnotic analgesia: 1. Somatosensory event-related potential changes to noxious stimuli and 2. Transfer learning to reduce chronic low back pain. Int J Clin Exp Hypn 46, 92–132. Dworkin, R.H., Turk, D.C., Wyrwich, K.W., Beaton, D., Cleeland, C.S., Farrar, J.T., Haythornthwaite, J.A., Jensen, M.P., Kerns, R.D., Ader, D.N., Brandenburg, N., Burke, L.B., Cella, D., Chandler, J., Cowan, P., Dimitrova, R., Dionne, R., Hertz, S., Jadad, A.R., Katz, N.P., Kehlet, H., Kramer, L.D., Manning, D.C., McCormick, C., McDermott, M.P., McQuay, H.J., Patel, S., Porter, L., Quessy, S., Rappaport, B.A., Rauschkolb, C., Revicki, D.A., Rothman, M., Schmader, K.E., Stacey, B.R., Stauffer, J.W., von Stein, T., White, R.E., Witter, J., Zavisic, S. (2008). Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. J Pain 9, 105– 121. Edelson, J., Fitzpatrick, J.L. (1989). A comparison of cognitive-behavioral and hypnotic treatments of chronic pain. J Clin Psychol 45, 316–323. Ehrenborg, C., Archenholtz, B. (2010). Is surface EMG biofeedback an effective training method for persons with neck and shoulder complaints after whiplash-associated disorders concerning activities of daily living and pain – A randomized controlled trial. Clin Rehabil 24, 715– 726. Flor, H., Birbaumer, N. (1993). Comparison of the efficacy of electromyographic biofeedback, cognitive-behavioral therapy, and conservative medical interventions in the treatment of chronic musculoskeletal pain. J Consult Clin Psychol 61, 653–658. Hilgard, E.R., Hilgard, J.R. (1994). Hypnosis in the Relief of Pain (New York: Brunner/Mazel). Institute of Medicine (2011). Relieving Pain in America: A Blueprint for Transforming Prevention, Care, Education, and Research (Washington, DC: The National Academies Press). Jensen, M.P. (2011). Hypnosis for Chronic Pain Management: Therapist Guide (Oxford: Oxford University Press). Jensen, M.P., Barber, J., Romano, J.M., Hanley, M.A., Raichle, K.A., Molton, I.R., Engel, J.M., Osborne, T.L., Stoelb, B.L., Cardenas, D.D., Patterson, D.R. (2009a). Effects of self-hypnosis training and EMG biofeedback relaxation training on chronic pain in persons with spinalcord injury. Int J Clin Exp Hypn 57, 239–268. Jensen, M.P., Barber, J., Romano, J.M., Molton, I.R., Raichle, K.A., Osborne, T.L., Engel, J.M., Stoelb, B.L., Kraft, G.H., Patterson, D.R. (2009b). A comparison of self-hypnosis versus progressive muscle relaxation in patients with multiple sclerosis and chronic pain. Int J Clin Exp Hypn 57, 198–221. Jensen, M.P., Ehde, D.M., Gertz, K.J., Stoelb, B.L., Dillworth, T.M., Hirsh, A.T., Molton, I.R., Kraft, G.H. (2011). Effects of self-hypnosis training and cognitive restructuring on daily pain intensity and catastrophizing in individuals with multiple sclerosis and chronic pain. Int J Clin Exp Hypn 59, 45–63. Katzman, R., Brown, T., Fuld, P., Peck, A., Schechter, R., Schimmel, H. (1983). Validation of a short Orientation-Memory-Concentration Test of cognitive impairment. Am J Psychiatry 140, 734–739. King, B.J., Nash, M.R., Spiegel, D., Jobson, K. (2001). Hypnosis as an intervention in pain management: A brief review. Int J Psychiatry Clin Prac 5, 97–101. Lew, H.L., Otis, J.D., Tun, C., Kerns, R.D., Clark, M.E., Cifu, D.X. (2009). Prevalence of chronic pain, posttraumatic stress disorder, and persistent postconcussive symptoms in OIF/OEF veterans: Polytrauma clinical triad. J Rehabil Res Dev 46, 697–702.

Eur J Pain 19 (2015) 271–280

279

RCT of hypnosis versus biofeedback for CLBP

Licciardone, J.C. (2008). The epidemiology and medical management of low back pain during ambulatory medical care visits in the United States. Osteopath Med Prim Care 2, 11. McCauley, J.D., Thelen, M.H., Frank, R.G., Willard, R.R., Callen, K.E. (1983). Hypnosis compared to relaxation in the outpatient management of chronic low back pain. Arch Phys Med Rehabil 64, 548–552. Montgomery, G.H., DuHamel, K.N., Redd, W.H. (2000). A meta-analysis of hypnotically induced analgesia: How effective is hypnosis? Int J Clin Exp Hypn 48, 138–153. Newton-John, T.R., Spence, S.H., Schotte, D. (1995). Cognitivebehavioural therapy versus EMG biofeedback in the treatment of chronic low back pain. Behav Res Ther 33, 691–697. Patterson, D.R., Jensen, M.P. (2003). Hypnosis and clinical pain. Psychol Bull 129, 495–521. Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Hergueta, T., Baker, R., Dunbar, G.C. (1998). The MiniInternational Neuropsychiatric Interview (M.I.N.I.): The development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry 59 (Suppl. 20), 22–33. Tan, G., Fukui, T., Jensen, M.P., Thornby, J., Waldman, K.L. (2010). Hypnosis treatment for chronic low back pain. Int J Clin Exp Hypn 58, 53–68. Tan, G., Sherman, R.A., Shanti, B.F. (2003). Biofeedback pain interventions. Pract Pain Manag 3, 12–18. Tyler, E.J., Jensen, M.P., Engel, J.M., Schwartz, L. (2002). The reliability and validity of pain interference measures in persons with cerebral palsy. Arch Phys Med Rehabil 83, 236–239. Yilmaz, O.O., Senocak, O., Sahin, E., Baydar, M., Gulbahar, S., Bircan, C., Alper, S. (2010). Efficacy of EMG-biofeedback in knee osteoarthritis. Rheumatol Int 30, 887–892.

280 Eur J Pain 19 (2015) 271–280

G. Tan et al.

Web references Chang, B.H., Hendricks, A.M., Slawsky, M.T., Locastro, J.S. (2004). Patient recruitment to a randomized clinical trial of behavioral therapy for chronic heart failure. BMC Med Res Methodol 4, 8. Retrieved from: http://www.biomedcentral.com/1471-2288/4/8 (accessed on 6 March 2014). Williams, A.C., Eccleston, C., Morley, S. (2012). Psychological therapies for the management of chronic pain (excluding headache) in adults. Cochrane Database Syst Rev (11), doi: 10.1002/14651858.CD007407 .pub3. Retrieved from: http://ca3cx5qj7w.search.serialssolutions.com/ OpenURL_local?sid=Entrez:PubMed&id=pmid:23152245 (accessed on 6 March 2014).

Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Discussion S1. Hypnotizability and frequency of practice. Method S1. Measures. Method S2. Statistical analysis. Results S1. Maintenance effects within groups. Results S2. Hypnotizability and frequency of practice. Table S1. Spearman rho correlation coefficients of the relationship between number of practice sessions and change in the outcome variables during the 8-week treatment period.

© 2014 European Pain Federation - EFIC®