Authors: Douglas L. Weeks, PhD Anthony A. Whitney, MS, LMHC, BCB Angelique G. Tindall, PhD Gregory T. Carter, MD, MS
Pain Intervention
Affiliations: From the Rehabilitation Research Department (DLW), Rehabilitation Psychology Clinic (AAW, AGT), and Physical Medicine and Rehabilitation Department (GTC), St. Luke’s Rehabilitation Institute, Spokane, Washington.
Correspondence: All correspondence and requests for reprints should be addressed to: Douglas L. Weeks, PhD, Rehabilitation Research Department, St. Luke’s Rehabilitation Institute, 711 S Cowley St, Spokane, WA 99202.
No funding, grants, or equipment from any source supported this work. The authors received no financial benefits from the study. No part of this work has been published or presented. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
DOI: 10.1097/PHM.0000000000000285
Pilot Randomized Trial Comparing Intersession Scheduling of Biofeedback Results to Individuals with Chronic Pain Influence on Psychologic Function and Pain Intensity ABSTRACT
Disclosures:
0894-9115/15/9410-0869 American Journal of Physical Medicine & Rehabilitation Copyright * 2015 Wolters Kluwer Health, Inc. All rights reserved.
ORIGINAL RESEARCH ARTICLE
Weeks DL, Whitney AA, Tindall AG, Carter GT: Pilot randomized trial comparing intersession scheduling of biofeedback results to individuals with chronic pain: influence on psychologic function and pain intensity. Am J Phys Med Rehabil 2015;94:869Y878.
Objective: The objective of this study was to compare the effectiveness of two biofeedback schedules on long-term improvement in physical and psychologic reactivity to chronic nonmalignant pain.
Design: This study is a prospective, randomized pilot trial. Methods: Twenty adults with chronic pain engaged in heart rate variability (HRV) biofeedback training for nine sessions with HRV presented visually. Two groups, formed by random assignment, were compared: The faded feedback group received concurrent visual HRV biofeedback in session 1, with the amount of biofeedback systematically reduced for ensuing sessions so that, by session 9, the participants were controlling HRV without external feedback. The full feedback group received visual HRV biofeedback continuously across all sessions. Outcome measures assessed at baseline, immediately after the program, and 3 mos after the program included pain intensity, fear-avoidance beliefs, and self-report physical functioning. Use of biofeedback skills was also assessed 3 mos after the program. Nominal variables were analyzed with W2. Continuous measures were analyzed with repeated-measures analyses of variance.
Results: The faded feedback schedule resulted in greater use of biofeedback skills at 3 mos and improved pain intensity and fear-avoidance beliefs after the program and at 3 mos. Physical functioning did not differ between groups.
Conclusions: Systematically reducing the frequency of external visual feedback during HRV biofeedback training was associated with reduced reactivity to chronic pain. Results of this pilot study should be confirmed with a larger randomized study. Key Words:
www.ajpmr.com
Biofeedback, Chronic Pain, Motor Learning, Psychologic Feedback
Biofeedback Scheduling in Chronic Pain Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
869
B
iofeedback therapy is a common component of multidisciplinary intervention for chronic pain.1 Biofeedback techniques provide patients with realtime visual or auditory information about a biologic signal that is theoretically reactive to pain but otherwise unknown to the patient. Typical biologic signals fed back to patients are muscle contraction activity of a target muscle, heart rate, heart rate variability (HRV), galvanic skin response, temperature, or respiratory rate. The patient’s goal in biofeedback therapy is to bring the biologic signal under intentional control in hopes of disrupting reactivity to pain. Comprehensive systematic reviews have concluded that there is low-quality evidence that biofeedback therapy is more effective for reducing pain or psychologic symptoms at the conclusion of a series of training sessions (e.g., in the short term) than a wait list control condition.2,3 However, these reviews have failed to find long-term benefit for pain relief or psychologic symptoms. In fact, most studies failed to reassess study groups in follow-up tests that are separated from original training, leading to an absence of evidence for long-term benefit. Thus, further well-designed trials are needed to establish whether progress in biofeedback training is sustained in the long term. The acquisition of pain-modifying strategies through biofeedback training results from learning to volitionally control a physiologic signal, that is, psychomotor learning. Laboratory research on variables that influence psychomotor learning has examined the schedule for providing external feedback within a training session that is optimal for long-term retention of skill. In most of these feedbackscheduling studies, a paradoxical relationship has been established as optimal for long-term retention: scheduling methods that reduce the frequency at which external feedback is provided to a learner during training (acquisition phase) result in greater levels of skill in retention tests separated from acquisition.4Y6 In these retention tests, presentation of external feedback is extinguished to assess the learner’s performance of the criterion skill based on intrinsic capability. This same research has shown that feedback schedules providing frequent feedback during acquisition encourage near-perfect performance while the feedback is present but result in poorer performance in retention tests. A theoretical explanation for this paradoxical relationship between reduced provision of feedback and enhanced learning is the guidance hypothesis.7Y9 The hypothesis suggests that provision of external
870
Weeks et al.
feedback at high frequencies relative to skill attempts during acquisition aids in guiding the learner to the correct response. However, external feedback becomes so guiding that the learner gains a dependence on the feedback to support performance, which is revealed as poor performance in retention testing. Thus, a high frequency of feedback (e.g., feedback given on 100% of skill attempts) provides learners (and clinicians) with the false sense that they have learned the skill well enough to perform in the absence of external feedback. In contrast, a feedback schedule that reduces the frequency with which the learner is provided with feedback reduces the opportunity for dependence on feedback. Instead, the learner must engage in more self-regulation of performance during acquisition attempts when external feedback is not present. Practice with self-regulation of performance enables the learner to perform at a higher level of skill in retention testing when external feedback is absent. One scheduling method to reduce feedback frequency is to fade the presentation of feedback across training so that the relative frequency is systematically reduced as experience accrues.4 On the basis of the guidance hypothesis, continuous visual feedback offered in biofeedback training should be reduced as early as possible in acquisition to reduce the likelihood for dependence on external feedback to control the biologic process in the long term. To assess this premise, the authors conducted a randomized pilot study comparing two biofeedback schedules for encouraging retention of pain modification ability and improvement in physical and psychologic functioning in people with chronic pain. Participants engaged in HRV biofeedback training for 3 wks and then were contacted after 3 mos to collect follow-up information. Two groups were compared: The faded feedback group received a high relative frequency of HRV biofeedback early in acquisition, with the amount of biofeedback systematically reduced across sessions. In contrast, the full feedback group received HRV biofeedback continuously in all sessions. The authors hypothesized that the groups would not differ in ability to control HRV at the conclusion of biofeedback training. They also hypothesized that, at 3 mos, the faded feedback group would be using biofeedback skills more frequently, suggesting that their biofeedback skills remained effective in the months after training. The authors further hypothesized that more proficient use of biofeedback skills in the faded feedback group would result in lower pain intensity at the conclusion of training and at 3 mos. They also hypothesized that greater proficiency in controlling the pain response
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
in the faded feedback group would result in greater improvement in self-report physical functioning at the conclusion of biofeedback training and at 3 mos. Because of the common incidence of excessive fear with chronic pain, the authors also assessed fear of pain as a psychologic construct ostensibly mediated by biofeedback training. Therefore, they also hypothesized that the faded feedback schedule would reduce activity-related fear of pain to a greater degree than the full feedback schedule at the conclusion of training and at 3 mos. The purposes of this pilot study were to establish effect sizes for sample size estimation for a larger randomized clinical trial that will include more study arms and assessment of additional psychologic constructs and to determine whether the protocol would benefit from revision. In addition, the authors’ 3-mo follow-up did not include reassessment of skill in controlling HRV in the clinic but instead requested that patients complete several surveys. The return rate for the surveys was used as a proxy for attrition estimates for the forthcoming trial.
METHODS Participants A total of 20 adults (aged 18 yrs or older) participated. All participants had chronic pain and were participating in an outpatient cognitive behavioral therapy program for chronic pain management. The patients were eligible for the study if they had nonmalignant pain that had persisted for longer than 3 mos. The participants’ pain diagnoses were heterogeneous in nature, encompassing conditions such as musculoskeletal pain, fibromyalgia, headache, neuropathy, or reflex sympathetic dystrophy. Individuals were ineligible for the study if they were currently receiving or scheduled to receive occupational or physical therapy as a cointervention for pain during biofeedback training or within the 3-mo follow-up period. Patients with severe psychopathology, such as major depressive disorder or anxiety disorder with psychotic features, were excluded from the study to reduce the confounding influence these conditions might have on selfreport responses to inventories used as outcome measures. The setting was an outpatient medical rehabilitation center accredited by the Commission on Accreditation of Rehabilitation Facilities. All protocols were approved by the institutional review board (protocol number 1228). Informed consent was obtained from all participants before entry into the study. After consent, the participants www.ajpmr.com
were randomly assigned in a 1:1 ratio to either the full feedback group (n = 10) or the faded feedback group (n = 10). Randomization was managed by the study biostatistician not located at the center where the study occurred. Opaque, sealed envelopes containing the name of the group to which a participant was to be assigned were drawn from a container. After consent and collection of baseline inventories, the biostatistician drew and opened an envelope and contacted the investigator with the group assignment per participant.
Design This study is a prospective pilot study with a factorial design. Repeated measurements of participant status on each outcome measure before and at the conclusion of biofeedback training and in a followup assessment at 3 mos were made. Repeated measurements of participant status on controlling HRV during biofeedback training were also carried out.
Biofeedback Interventions A certified biofeedback specialist blinded to the purpose of the study conducted the biofeedback training sessions. The participants engaged in HRV biofeedback training three times a week with 1Y2 days between sessions, for a total of nine sessions in 3 wks. Two 10-min biofeedback trials were performed per session. In the full feedback group, the participants received continuous visual biofeedback of HRV, projected on a 120-cm 90-cm video screen on each trial. In the faded feedback group, the amount of time visual biofeedback was provided was faded across the nine sessions such that, in the initial training session, biofeedback was provided 100% of the time. However, for ensuing sessions, biofeedback frequency was systematically reduced by removing the visually projected image of HRV resulting in the following relative frequency feedback schedule for each corresponding day: 90%, 80%, 70%, 60%, 40%, 20%, 10%, and 0%. Two weeks before the 3-mo follow-up assessment, the participants were mailed paper copies of surveys and inventories. A cover letter informed them that clinic staff would call within the next 2 wks. During this call, the staff person administered the instruments by phone if the participant had not returned the instruments. To minimize loss to follow-up, up to five attempts were made to contact the participants by phone.
Biofeedback Session Protocols HRV was obtained through power spectral analysis of heart rate fluctuations to quantitatively Biofeedback Scheduling in Chronic Pain
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
871
evaluate beat-to-beat cardiovascular control. Three main spectral components were calculated from the 10-min biofeedback trial: low-frequency, mediumfrequency, and high-frequency components. The participant’s goal in HRV biofeedback training was to modulate the variability in beat-to-beat heart rate so that a smooth sinusoidal time series pattern that contained as much low-frequency spectral content (e.g., 0.1 Hz or less) as possible was established. Each biofeedback session began with the participant seated in a comfortable recliner chair that was 2 m from the video screen. A photoplethysmographic sensor placed on the participant’s middle finger recorded heart rate. A commercially available software program (Freeze-Framer 2.0, HeartMath Institute) recorded and displayed a time series graph of heart rate, as well as a three-bar graph in which each bar represented a bin for high-, medium-, or lowfrequency HRV spectral content. The goal for the participant was to Bpopulate[ the low-frequency bar so that the percentage of length of the low-frequency bar was greater than that of the other bars. Each session had two phases: (1) a 5-min baseline recording of HRV in which the participants viewed a series of calming nature scenes but received no HRV biofeedback and (2) two 10-min trials in which the participants received visual HRV biofeedback according to their group assignment. The full feedback group received visual biofeedback continuously throughout each 10-min trial. The faded feedback group received visual biofeedback for a given portion of each 10-min trial, after which the calming nature scenes reappeared while the participant was instructed to continue controlling HRV until the end of the 10-min trial. At the conclusion of each 10-min trial, the participants in both groups received summary results of the percentage of HRV in each bar.
Outcome Measures and Data Analysis Nominal variables were analyzed with W2 analyses. Continuous measures were analyzed with repeated-measures analyses of variance (ANOVAs). Because one purpose of this pilot study was to develop effect sizes for sample size estimation in a larger trial, the authors did not establish a criterion for statistical significance. Instead, they were interested in examining trends in addition to P values from inferential testing to understand directional influences of the interventions over time. All analyses were conducted with SPSS v. 22 (SPSS, Inc, Chicago, IL).
Heart Rate Variability The percentage of low-frequency HRV spectral content within a trial was analyzed to evaluate
872
Weeks et al.
proficiency in controlling HRV. Percentages from the two 10-min trials within a day were averaged to arrive at a single low-frequency HRV point estimate per day. These scores were analyzed with a 2 9 (group session) ANOVA with repeated measures on session. HRV has been shown to be reliably measured in various patient populations.10
Use of Biofeedback Skills After the Program Postprogram use of biofeedback skills was evaluated at the 3-mo follow-up assessment with a brief survey developed for the study. Survey items are displayed in Table 1. The frequency of responses to each choice was analyzed with group response choice W2 analyses per survey item.
Pain Intensity The participants rated pain intensity on a 10-cm visual analog scale (VAS) in which 0 cm corresponded to Bno pain[ and 10 cm corresponded to Bworst pain imaginable.[ The participants marked the 10-cm line representing current level of pain, and the worst pain had been in the last 24 hrs. The distance from the zero anchor to the mark in millimeters represented the pain intensity score. VASs for pain have been shown to be reliable and valid in chronic pain populations.11,12 The VAS was administered at baseline, at the conclusion of training, and at the 3-mo follow-up assessment. VAS ratings were analyzed with a 2 3 (group measurement occasion) ANOVA with repeated measures on measurement occasion.
Functional Status Perceived function was assessed with the English language version (all participants’ first language) of the Pain Disability Questionnaire (PDQ). The PDQ is a 15-item self-report instrument of functional status focusing on perceived disability in performing activities of daily living and psychosocial status in people with chronic disabling musculoskeletal disorders.13 Respondents mark their response on a 15-cm VAS, with the score derived by measuring the location of the mark to the nearest 1.5 cm. The number of 1.5-cm increments represents the item score. Summing the scores yields a total functional disability score ranging from 0 (optimal function) to 150 (total disability). The English language version of the PDQ has demonstrated strong reliability, responsiveness, and construct validity.13,14 The PDQ was administered at baseline, at the conclusion of training, and at the 3-mo follow-up assessment. PDQ scores were analyzed with a 2 3 (group measurement occasion) ANOVA with repeated measures on measurement occasion.
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
TABLE 1 Survey items administered at the 3-mo follow-up assessment to understand postprogram use of biofeedback skills Faded Feedback Group, Full Feedback Group, n (%) (n = 6) n (%) (n = 5)
Survey Item How often have you used your heart rate control skills to control your pain since the end of the biofeedback program? Never Once a month Once a week A few times a week Daily How often do the heart rate control skills help control your pain? Never Less than half the time Approximately half the time More often than half the time Every time I use the skills Have you used the skills more frequently or less frequently now than right after discharge from the biofeedback program? More frequently now Less frequently now Approximately the same then and now I have not used the skills to control my pain because they do not work I have not used the skills because my pain has been under control
Fear of Pain
0 (0) 0 (0) 0 (0) 2 (33) 4 (67)
2 (40) 1 (20) 0 (0) 0 (0) 2 (40)
0 (0) 0 (0) 3 (50) 2 (33) 0 (0)
2 (40) 2 (40) 1 (20) 1 (20) 0 (0)
1 (17) 0 (0) 5 (83) 0 (0)
0 (0) 0 (0) 3 (60) 2 (40)
0 (0)
0 (0)
Kinesiophobia is defined as excessive, irrational, and debilitating fear of physical activity resulting from a sense of vulnerability to painful injury or reinjury.15 Kinesiophobia was assessed with the English language version of the 11-item Tampa Scale of Kinesiophobia (TSK-11).16 The TSK-11 uses a four-point Likert response scale ranging from Bstrongly disagree[ (rated as 1) to Bstrongly agree[ (rated as 4). Summing item scores yields a total fear of pain score ranging from 11 (disagreement that fear of pain limits movement) to 44 (endorsement that fear of pain limits movement). The English language version of the TSK-11 has demonstrated strong reliability and construct validity.16,17 The TSK-11 was administered at baseline, at the conclusion of training, and at the 3-mo follow-up assessment. TSK-11 scores were analyzed with a 2 3 (group measurement occasion) ANOVA with repeated measures on measurement occasion.
unable to be contacted after several sessions, and one was removed from the study after a diagnosis of heart arrhythmia. Three full feedback group participants discontinued the study before the posttest: one was unable to be contacted after several sessions, one had a recurrence of breast cancer, and one could not arrange transportation after his third session. One faded feedback group participant and two full feedback group participants completed biofeedback training but failed to return follow-up surveys after five attempts to contact by phone and were therefore classified as lost to follow-up. t Tests comparing baseline inventory and pain scores of attrited participants with those who completed the study did not reveal major differences among the groups, attenuating risk for selection bias (all P 9 0.05). Further analyses were conducted on the six faded feedback group participants and five full feedback group participants who completed the study.
RESULTS
Participant Characteristics
Attrition Analyses
No important differences among the groups were detected for demographic or baseline pain characteristics (Table 2; all P 9 0.431).
Participant enrollment and flow is shown in Figure 1. Attrition rates included participants who discontinued biofeedback training and those who completed training but did not return follow-up surveys. Three faded feedback group participants discontinued the study before the posttest: two were www.ajpmr.com
Heart Rate Variability Neither the group main effect (P = 0.656) nor the group session interaction (P = 0.995) revealed Biofeedback Scheduling in Chronic Pain
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
873
FIGURE 1 Study flow diagram. important differences among the groups in controlling HRV across the nine biofeedback sessions (Fig. 2).
Use of Biofeedback Skills After the Program Table 1 displays responses by group to followup survey questions. A larger proportion in the faded feedback group continued to use skills acquired during training (P = 0.132 for the difference between the groups). In addition, a larger proportion in the faded feedback group used the skills half the time or more (P = 0.092 for the difference between the groups). Finally, a greater proportion of those in the full feedback group indicated that they had not used the skills to control pain because they did not work (P = 0.179 for the difference between the groups).
Pain Intensity Figure 3 portrays mean VAS ratings for current level of pain. The group main effect did not indicate important differences among the groups (P = 0.506); however, the group measurement occasion interaction revealed that current pain was decreased relative to baseline at the end of biofeedback training and at follow-up in the faded feedback group (P = 0.291). In contrast, current pain was unchanged in the full feedback group at the end of biofeedback training and increased over baseline at follow-up. Figure 4 portrays mean VAS ratings for worst pain in the last 24 hrs. The group main effect did not indicate important differences among the groups (P = 0.519); however, the group measurement occasion interaction revealed that worst pain was decreased relative to baseline at the end of biofeedback training and at follow-up in the faded
TABLE 2 Demographic and baseline pain characteristics of the participants in each group Group
Sex, n (%) Male Female Employment status, n (%) Full-time regular duty Part-time regular duty Part-time modified duty Have not worked/retired Age, mean (SD), yrs Baseline current pain, mean (SD), cm Baseline worst pain in 24 hrs, mean (SD), cm
874
Weeks et al.
Faded Feedback (n = 6)
Full Feedback (n = 5)
3 (50) 3 (50)
3 (60) 2 (40)
1 (17) 2 (33) 0 (0) 3 (50) 56.7 (10.8) 4.17 (1.94) 8.17 (1.72)
0 (0) 2 (40) 1 (20) 2 (40) 60.2 (8.1) 3.85 (1.92) 7.80 (2.30)
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
FIGURE 2 Percentage of time that low-frequency HRV was achieved in each group per biofeedback session. feedback group (P = 0.148). In contrast, worst pain was unchanged in the full feedback group at the end of the biofeedback training and increased over baseline at follow-up.
Functional Status Neither the group main effect (P = 0.464) nor the group measurement occasion interaction (P = 0.384) revealed consistent differences among the groups in ratings of perceived disability as assessed by the PDQ. Mean (SE) PDQ scores were 36.8 (4.7) for the faded feedback group and 42.1 (5.2) for the full feedback group.
Fear of Pain Figure 5 portrays mean TSK-11 scores. The group main effect did not indicate important differences
among the groups (P = 0.474); however, the group measurement occasion interaction revealed that fear of physical activity due to pain decreased at the end of the biofeedback training in the faded feedback group (P = 0.102). The reduction in fear continued into the follow-up period. In contrast, fear of physical activity was relatively unchanged in the full feedback group.
DISCUSSION This pilot study provided evidence that feedback scheduling during skill training that reduced dependence on biofeedback to gain control of HRV had sustained benefits on several measures of reactivity to pain. As hypothesized, the group receiving a high frequency of external feedback about ability to control HRV and the group receiving a low
FIGURE 3 Mean VAS pain ratings per group for current pain across the three measurement occasions. www.ajpmr.com
Biofeedback Scheduling in Chronic Pain Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
875
FIGURE 4 Mean VAS pain rating scores per group for worst pain in the last 24 hrs across the three measurement occasions.
frequency of external feedback performed the biofeedback task similarly at the end of nine sessions of training. Thus, reduced feedback frequency was not detrimental to training performance. Again, as hypothesized, at the 3-mo follow-up assessment, the group experiencing a low frequency of feedback in training reported more frequent use of the biofeedback skills, suggesting that these skills remained effective in the months after training. Support was also revealed for the hypothesis that more proficient use of biofeedback skills in the faded feedback group would result in lower pain intensity at the conclusion of training and at the 3-mo follow-up.
Finally, as hypothesized, the greater proficiency in modulating physical symptoms afforded by the faded feedback schedule reduced activity-related fear of pain to a greater degree than that by the full feedback schedule at the conclusion of biofeedback training and at the 3-mo follow-up evaluation. The principle of reducing relative frequency of feedback in acquisition to enhance long-term skill retention has not been established for medically oriented skills. Evidence that the guidance hypothesis can be extended to medical skills, such as control over HRV, is encouraging. HRV is a measure of beat-to-beat neurocardiac function reflecting
FIGURE 5 Mean TSK-11 scores per group across the three measurement occasions.
876
Weeks et al.
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
both sympathetic and parasympathetic autonomic nervous system activity.18,19 Chronic pain tends to arouse sympathetic activity and suppress parasympathetic activity. Control over HRV affects a shift in autonomic function away from a stress-related sympathetic response toward an increased parasympathetic response, which is involved in the inhibition of pain signals and sympathetic outflow. The faded feedback condition may allow participants to engage in self-management of the parasympathetic response to pain during biofeedback training so that, at the conclusion of training, participants have greater self-sufficiency in managing pain than that afforded by the full feedback schedule. Alternately, the full feedback schedule may discourage learning to regulate HRV intrinsically because the need for self-regulation is nullified when external biofeedback is constantly present, resulting in poorer capability to control HRV at the conclusion of training and less frequent use of biofeedback skills in the long term. It is also encouraging that reducing feedback frequency may influence psychologic constructs reactive to pain, such as fear of pain. Fear of pain due to movement and activity-avoidance behaviors in anticipation of injury (or reinjury) are significant and common psychologic complications of chronic pain. The fear-avoidance model of chronic pain suggests that activity-related physiologic symptoms, such as variability in heart rate, may result in exaggerated pain sensations.20 These misinterpreted physiologic symptoms reinforce the fear that physical activity will be painful. As fear is elevated, avoidance of activity increases. Consequently, avoidance of movement leads to a cycle of physical deconditioning and further disability. Pain-related fear has been shown to be as potent a predictor of physical disability as pain severity21Y23 and can have a significant negative impact on treatment outcome.15 For this reason, it is encouraging that enhanced learning through reduced feedback frequency was associated with reduced fear-avoidance beliefs. The purposes of this pilot study were, among other things, to understand attrition rates and to obtain effect sizes for sample size estimates for a larger randomized clinical trial that includes a follow-up retention test of skill in controlling HRV. In the chronic pain population, attrition rates from psychologic-based management programs are known to be high.24 Similarly, attrition rates in this pilot study were high in both groups (40% in the faded feedback group and 60% in the full feedback group). Although the authors intentionally implemented a multiple callback feature as part of the design of this pilot study, it is clear that other www.ajpmr.com
methods to avoid attrition must be used in the forthcoming larger trial. For example, use of a runin period in which prospective participants attend several appointments before inclusion can determine participants with a high probability of being adherent to the study schedule. In addition to these protocol features, sample size estimates in the larger trial will take into consideration the potential for attrition as high as 60%. On the basis of effect sizes from this study and the large attrition rates, the estimates in this study suggest that sample size in a larger trial will need to be two-anda-half times greater to achieve statistical significance at the 0.05 level of confidence. Among other adjustments, the single outcome for which there was no trend in favor of the faded feedback schedule was self-rated assessment of disability perceptions with the PDQ. The authors posit two possible explanations for why the groups did not differ on PDQ scores: It is possible that disability orientation is not modifiable in response to biofeedback training. Alternatively, the PDQ may not be sensitive enough to measure differential effects of biofeedback schedules on perceptions of pain interference with functional status. In the larger trial, the authors will include additional measures of function and disability to enhance measurement of these constructs. Despite the study’s limitations, its data provide evidence that systematically reducing the frequency of biofeedback results given to the learner with chronic pain is associated with several positive outcomes over an extended period. This information could be useful in facilitating a more costeffective biofeedback regimen for people with chronic pain. The extent to which these findings apply generally to patients with chronic pain will require larger studies to determine whether the findings of this study can be replicated and, as such, establish their generalizability. REFERENCES 1. Giggins OM, Persson UM, Caulfield B: Biofeedback in rehabilitation. J Neuroeng Rehabil 2013;10:60 2. Morley S, Eccleston C, Williams A: Systematic review and meta-analysis of randomized controlled trials of cognitive behaviour therapy and behaviour therapy for chronic pain in adults, excluding headache. Pain 1999;80:1Y13 3. Henschke N, Ostelo RWJG, van Tulder MW, et al: Behavioural treatment for chronic low-back pain. Cochrane Database Syst Rev 2010 4. Winstein CJ, Schmidt RA: Reduced frequency of knowledge of results enhances motor skill learning. J Exp Psychol Learn Mem Cogn 1990;16:677Y91
Biofeedback Scheduling in Chronic Pain Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
877
5. Weeks DL, Kordus RA: Relative frequency of knowledge of performance and motor skill learning. Res Q Exerc Sport 1998;69:224Y30
15. Kori SH, Miller RP, Todd DD: Kinesiophobia: A new view of chronic pain behavior. Pain Manag 1990;3:35Y43
6. Anderson DI, Magill RA, Sekiya H, et al: Support for an explanation of the guidance effect in motor skill learning. J Mot Behav 2005;37:231Y8
16. Woby SR, Roach NK, Urmston M, et al: Psychometric properties of the TSK-11: A shortened version of the Tampa Scale for Kinesiophobia. Pain 2005;117:137Y44
7. Salmoni AW, Schmidt RA, Walter CB: Knowledge of results and motor learning: A review and critical reappraisal. Psychol Bull 1984;95:355Y86
17. Tkachuk GA, Harris CA: Psychometric properties of the Tampa Scale for Kinesiophobia-11 (TSK-11). J Pain 2012;13:970Y7
8. Wulf G, Shea C, Lewthwaite R: Motor skill learning and performance: A review of influential factors. Med Educ 2010;44:75Y84
18. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology: Heart rate variability: Standards of measurement, physiological interpretation, and clinical use. Circulation 1996;93:1043Y65
9. Ronsse R, Puttemans V, Coxon JP, et al: Motor learning with augmented feedback: Modalitydependent behavioral and neural consequences. Cereb Cortex 2011;21:1283Y94 10. Perini R, Veicsteinas A: Heart rate variability and autonomic activity at rest and during exercise in various physiological conditions. Eur J Appl Physiol 2003;90:317Y25 11. Jensen MP, Turner JA, Romano JM, et al: Comparative reliability and validity of chronic pain intensity measures. Pain 1999;83:157Y62 12. Price DD, Bush FM, Long S, et al: A comparison of pain measurement characteristics of mechanical visual analogue and simple numerical rating scales. Pain 1994;56:217Y26
878
19. McCraty R, Atkinson M, Tiller WA, et al: The effects of emotions on short-term heart rate variability using power spectrum analysis. Am J Cardiol 1995;76:1089Y93 20. Norton PJ, Asmundson GJG: Amending the fearavoidance model of chronic pain: What is the role of physiological arousal? Behav Ther 2003;34:17Y30 21. Crombez G, Vlaeyen JW, Heuts PH, et al: Fear of pain is more disabling than pain itself: Evidence on the role of pain-related fear in chronic back pain disability. Pain 1999;80:329Y40 22. McCracken L, Faber SD, Janeck AS: Pain-related anxiety predicts non-specific physical complaints in persons with chronic pain. Behav Res Ther 1998;36: 621Y30
13. Anagnostis C, Gatchel R, Mayer TG: The pain disability questionnaire: A new psychometrically sound measure for chronic musculoskeletal disorders. Spine 2004;29:2290Y302
23. Vlaeyen JW, Linton SJ: Fear-avoidance and its consequences in chronic musculoskeletal pain: A state of the art. Pain 2000;85:317Y32
14. Gatchel RJ, Mayer TG, Theodore BR: The Pain Disability Questionnaire: Relationship to one-year functional and psychosocial rehabilitation outcomes. J Occup Rehabil 2006;16:75Y94
24. Williams AC, Eccleston C, Morley S: Psychological therapies for the management of chronic pain (excluding headache) in adults. Cochrane Database Syst Rev 2012;11
Weeks et al.
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Pain Intervention
Affiliations: From the Rehabilitation Research Department (DLW), Rehabilitation Psychology Clinic (AAW, AGT), and Physical Medicine and Rehabilitation Department (GTC), St. Luke’s Rehabilitation Institute, Spokane, Washington.
Correspondence: All correspondence and requests for reprints should be addressed to: Douglas L. Weeks, PhD, Rehabilitation Research Department, St. Luke’s Rehabilitation Institute, 711 S Cowley St, Spokane, WA 99202.
No funding, grants, or equipment from any source supported this work. The authors received no financial benefits from the study. No part of this work has been published or presented. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
DOI: 10.1097/PHM.0000000000000285
Pilot Randomized Trial Comparing Intersession Scheduling of Biofeedback Results to Individuals with Chronic Pain Influence on Psychologic Function and Pain Intensity ABSTRACT
Disclosures:
0894-9115/15/9410-0869 American Journal of Physical Medicine & Rehabilitation Copyright * 2015 Wolters Kluwer Health, Inc. All rights reserved.
ORIGINAL RESEARCH ARTICLE
Weeks DL, Whitney AA, Tindall AG, Carter GT: Pilot randomized trial comparing intersession scheduling of biofeedback results to individuals with chronic pain: influence on psychologic function and pain intensity. Am J Phys Med Rehabil 2015;94:869Y878.
Objective: The objective of this study was to compare the effectiveness of two biofeedback schedules on long-term improvement in physical and psychologic reactivity to chronic nonmalignant pain.
Design: This study is a prospective, randomized pilot trial. Methods: Twenty adults with chronic pain engaged in heart rate variability (HRV) biofeedback training for nine sessions with HRV presented visually. Two groups, formed by random assignment, were compared: The faded feedback group received concurrent visual HRV biofeedback in session 1, with the amount of biofeedback systematically reduced for ensuing sessions so that, by session 9, the participants were controlling HRV without external feedback. The full feedback group received visual HRV biofeedback continuously across all sessions. Outcome measures assessed at baseline, immediately after the program, and 3 mos after the program included pain intensity, fear-avoidance beliefs, and self-report physical functioning. Use of biofeedback skills was also assessed 3 mos after the program. Nominal variables were analyzed with W2. Continuous measures were analyzed with repeated-measures analyses of variance.
Results: The faded feedback schedule resulted in greater use of biofeedback skills at 3 mos and improved pain intensity and fear-avoidance beliefs after the program and at 3 mos. Physical functioning did not differ between groups.
Conclusions: Systematically reducing the frequency of external visual feedback during HRV biofeedback training was associated with reduced reactivity to chronic pain. Results of this pilot study should be confirmed with a larger randomized study. Key Words:
www.ajpmr.com
Biofeedback, Chronic Pain, Motor Learning, Psychologic Feedback
Biofeedback Scheduling in Chronic Pain Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
869
B
iofeedback therapy is a common component of multidisciplinary intervention for chronic pain.1 Biofeedback techniques provide patients with realtime visual or auditory information about a biologic signal that is theoretically reactive to pain but otherwise unknown to the patient. Typical biologic signals fed back to patients are muscle contraction activity of a target muscle, heart rate, heart rate variability (HRV), galvanic skin response, temperature, or respiratory rate. The patient’s goal in biofeedback therapy is to bring the biologic signal under intentional control in hopes of disrupting reactivity to pain. Comprehensive systematic reviews have concluded that there is low-quality evidence that biofeedback therapy is more effective for reducing pain or psychologic symptoms at the conclusion of a series of training sessions (e.g., in the short term) than a wait list control condition.2,3 However, these reviews have failed to find long-term benefit for pain relief or psychologic symptoms. In fact, most studies failed to reassess study groups in follow-up tests that are separated from original training, leading to an absence of evidence for long-term benefit. Thus, further well-designed trials are needed to establish whether progress in biofeedback training is sustained in the long term. The acquisition of pain-modifying strategies through biofeedback training results from learning to volitionally control a physiologic signal, that is, psychomotor learning. Laboratory research on variables that influence psychomotor learning has examined the schedule for providing external feedback within a training session that is optimal for long-term retention of skill. In most of these feedbackscheduling studies, a paradoxical relationship has been established as optimal for long-term retention: scheduling methods that reduce the frequency at which external feedback is provided to a learner during training (acquisition phase) result in greater levels of skill in retention tests separated from acquisition.4Y6 In these retention tests, presentation of external feedback is extinguished to assess the learner’s performance of the criterion skill based on intrinsic capability. This same research has shown that feedback schedules providing frequent feedback during acquisition encourage near-perfect performance while the feedback is present but result in poorer performance in retention tests. A theoretical explanation for this paradoxical relationship between reduced provision of feedback and enhanced learning is the guidance hypothesis.7Y9 The hypothesis suggests that provision of external
870
Weeks et al.
feedback at high frequencies relative to skill attempts during acquisition aids in guiding the learner to the correct response. However, external feedback becomes so guiding that the learner gains a dependence on the feedback to support performance, which is revealed as poor performance in retention testing. Thus, a high frequency of feedback (e.g., feedback given on 100% of skill attempts) provides learners (and clinicians) with the false sense that they have learned the skill well enough to perform in the absence of external feedback. In contrast, a feedback schedule that reduces the frequency with which the learner is provided with feedback reduces the opportunity for dependence on feedback. Instead, the learner must engage in more self-regulation of performance during acquisition attempts when external feedback is not present. Practice with self-regulation of performance enables the learner to perform at a higher level of skill in retention testing when external feedback is absent. One scheduling method to reduce feedback frequency is to fade the presentation of feedback across training so that the relative frequency is systematically reduced as experience accrues.4 On the basis of the guidance hypothesis, continuous visual feedback offered in biofeedback training should be reduced as early as possible in acquisition to reduce the likelihood for dependence on external feedback to control the biologic process in the long term. To assess this premise, the authors conducted a randomized pilot study comparing two biofeedback schedules for encouraging retention of pain modification ability and improvement in physical and psychologic functioning in people with chronic pain. Participants engaged in HRV biofeedback training for 3 wks and then were contacted after 3 mos to collect follow-up information. Two groups were compared: The faded feedback group received a high relative frequency of HRV biofeedback early in acquisition, with the amount of biofeedback systematically reduced across sessions. In contrast, the full feedback group received HRV biofeedback continuously in all sessions. The authors hypothesized that the groups would not differ in ability to control HRV at the conclusion of biofeedback training. They also hypothesized that, at 3 mos, the faded feedback group would be using biofeedback skills more frequently, suggesting that their biofeedback skills remained effective in the months after training. The authors further hypothesized that more proficient use of biofeedback skills in the faded feedback group would result in lower pain intensity at the conclusion of training and at 3 mos. They also hypothesized that greater proficiency in controlling the pain response
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
in the faded feedback group would result in greater improvement in self-report physical functioning at the conclusion of biofeedback training and at 3 mos. Because of the common incidence of excessive fear with chronic pain, the authors also assessed fear of pain as a psychologic construct ostensibly mediated by biofeedback training. Therefore, they also hypothesized that the faded feedback schedule would reduce activity-related fear of pain to a greater degree than the full feedback schedule at the conclusion of training and at 3 mos. The purposes of this pilot study were to establish effect sizes for sample size estimation for a larger randomized clinical trial that will include more study arms and assessment of additional psychologic constructs and to determine whether the protocol would benefit from revision. In addition, the authors’ 3-mo follow-up did not include reassessment of skill in controlling HRV in the clinic but instead requested that patients complete several surveys. The return rate for the surveys was used as a proxy for attrition estimates for the forthcoming trial.
METHODS Participants A total of 20 adults (aged 18 yrs or older) participated. All participants had chronic pain and were participating in an outpatient cognitive behavioral therapy program for chronic pain management. The patients were eligible for the study if they had nonmalignant pain that had persisted for longer than 3 mos. The participants’ pain diagnoses were heterogeneous in nature, encompassing conditions such as musculoskeletal pain, fibromyalgia, headache, neuropathy, or reflex sympathetic dystrophy. Individuals were ineligible for the study if they were currently receiving or scheduled to receive occupational or physical therapy as a cointervention for pain during biofeedback training or within the 3-mo follow-up period. Patients with severe psychopathology, such as major depressive disorder or anxiety disorder with psychotic features, were excluded from the study to reduce the confounding influence these conditions might have on selfreport responses to inventories used as outcome measures. The setting was an outpatient medical rehabilitation center accredited by the Commission on Accreditation of Rehabilitation Facilities. All protocols were approved by the institutional review board (protocol number 1228). Informed consent was obtained from all participants before entry into the study. After consent, the participants www.ajpmr.com
were randomly assigned in a 1:1 ratio to either the full feedback group (n = 10) or the faded feedback group (n = 10). Randomization was managed by the study biostatistician not located at the center where the study occurred. Opaque, sealed envelopes containing the name of the group to which a participant was to be assigned were drawn from a container. After consent and collection of baseline inventories, the biostatistician drew and opened an envelope and contacted the investigator with the group assignment per participant.
Design This study is a prospective pilot study with a factorial design. Repeated measurements of participant status on each outcome measure before and at the conclusion of biofeedback training and in a followup assessment at 3 mos were made. Repeated measurements of participant status on controlling HRV during biofeedback training were also carried out.
Biofeedback Interventions A certified biofeedback specialist blinded to the purpose of the study conducted the biofeedback training sessions. The participants engaged in HRV biofeedback training three times a week with 1Y2 days between sessions, for a total of nine sessions in 3 wks. Two 10-min biofeedback trials were performed per session. In the full feedback group, the participants received continuous visual biofeedback of HRV, projected on a 120-cm 90-cm video screen on each trial. In the faded feedback group, the amount of time visual biofeedback was provided was faded across the nine sessions such that, in the initial training session, biofeedback was provided 100% of the time. However, for ensuing sessions, biofeedback frequency was systematically reduced by removing the visually projected image of HRV resulting in the following relative frequency feedback schedule for each corresponding day: 90%, 80%, 70%, 60%, 40%, 20%, 10%, and 0%. Two weeks before the 3-mo follow-up assessment, the participants were mailed paper copies of surveys and inventories. A cover letter informed them that clinic staff would call within the next 2 wks. During this call, the staff person administered the instruments by phone if the participant had not returned the instruments. To minimize loss to follow-up, up to five attempts were made to contact the participants by phone.
Biofeedback Session Protocols HRV was obtained through power spectral analysis of heart rate fluctuations to quantitatively Biofeedback Scheduling in Chronic Pain
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
871
evaluate beat-to-beat cardiovascular control. Three main spectral components were calculated from the 10-min biofeedback trial: low-frequency, mediumfrequency, and high-frequency components. The participant’s goal in HRV biofeedback training was to modulate the variability in beat-to-beat heart rate so that a smooth sinusoidal time series pattern that contained as much low-frequency spectral content (e.g., 0.1 Hz or less) as possible was established. Each biofeedback session began with the participant seated in a comfortable recliner chair that was 2 m from the video screen. A photoplethysmographic sensor placed on the participant’s middle finger recorded heart rate. A commercially available software program (Freeze-Framer 2.0, HeartMath Institute) recorded and displayed a time series graph of heart rate, as well as a three-bar graph in which each bar represented a bin for high-, medium-, or lowfrequency HRV spectral content. The goal for the participant was to Bpopulate[ the low-frequency bar so that the percentage of length of the low-frequency bar was greater than that of the other bars. Each session had two phases: (1) a 5-min baseline recording of HRV in which the participants viewed a series of calming nature scenes but received no HRV biofeedback and (2) two 10-min trials in which the participants received visual HRV biofeedback according to their group assignment. The full feedback group received visual biofeedback continuously throughout each 10-min trial. The faded feedback group received visual biofeedback for a given portion of each 10-min trial, after which the calming nature scenes reappeared while the participant was instructed to continue controlling HRV until the end of the 10-min trial. At the conclusion of each 10-min trial, the participants in both groups received summary results of the percentage of HRV in each bar.
Outcome Measures and Data Analysis Nominal variables were analyzed with W2 analyses. Continuous measures were analyzed with repeated-measures analyses of variance (ANOVAs). Because one purpose of this pilot study was to develop effect sizes for sample size estimation in a larger trial, the authors did not establish a criterion for statistical significance. Instead, they were interested in examining trends in addition to P values from inferential testing to understand directional influences of the interventions over time. All analyses were conducted with SPSS v. 22 (SPSS, Inc, Chicago, IL).
Heart Rate Variability The percentage of low-frequency HRV spectral content within a trial was analyzed to evaluate
872
Weeks et al.
proficiency in controlling HRV. Percentages from the two 10-min trials within a day were averaged to arrive at a single low-frequency HRV point estimate per day. These scores were analyzed with a 2 9 (group session) ANOVA with repeated measures on session. HRV has been shown to be reliably measured in various patient populations.10
Use of Biofeedback Skills After the Program Postprogram use of biofeedback skills was evaluated at the 3-mo follow-up assessment with a brief survey developed for the study. Survey items are displayed in Table 1. The frequency of responses to each choice was analyzed with group response choice W2 analyses per survey item.
Pain Intensity The participants rated pain intensity on a 10-cm visual analog scale (VAS) in which 0 cm corresponded to Bno pain[ and 10 cm corresponded to Bworst pain imaginable.[ The participants marked the 10-cm line representing current level of pain, and the worst pain had been in the last 24 hrs. The distance from the zero anchor to the mark in millimeters represented the pain intensity score. VASs for pain have been shown to be reliable and valid in chronic pain populations.11,12 The VAS was administered at baseline, at the conclusion of training, and at the 3-mo follow-up assessment. VAS ratings were analyzed with a 2 3 (group measurement occasion) ANOVA with repeated measures on measurement occasion.
Functional Status Perceived function was assessed with the English language version (all participants’ first language) of the Pain Disability Questionnaire (PDQ). The PDQ is a 15-item self-report instrument of functional status focusing on perceived disability in performing activities of daily living and psychosocial status in people with chronic disabling musculoskeletal disorders.13 Respondents mark their response on a 15-cm VAS, with the score derived by measuring the location of the mark to the nearest 1.5 cm. The number of 1.5-cm increments represents the item score. Summing the scores yields a total functional disability score ranging from 0 (optimal function) to 150 (total disability). The English language version of the PDQ has demonstrated strong reliability, responsiveness, and construct validity.13,14 The PDQ was administered at baseline, at the conclusion of training, and at the 3-mo follow-up assessment. PDQ scores were analyzed with a 2 3 (group measurement occasion) ANOVA with repeated measures on measurement occasion.
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
TABLE 1 Survey items administered at the 3-mo follow-up assessment to understand postprogram use of biofeedback skills Faded Feedback Group, Full Feedback Group, n (%) (n = 6) n (%) (n = 5)
Survey Item How often have you used your heart rate control skills to control your pain since the end of the biofeedback program? Never Once a month Once a week A few times a week Daily How often do the heart rate control skills help control your pain? Never Less than half the time Approximately half the time More often than half the time Every time I use the skills Have you used the skills more frequently or less frequently now than right after discharge from the biofeedback program? More frequently now Less frequently now Approximately the same then and now I have not used the skills to control my pain because they do not work I have not used the skills because my pain has been under control
Fear of Pain
0 (0) 0 (0) 0 (0) 2 (33) 4 (67)
2 (40) 1 (20) 0 (0) 0 (0) 2 (40)
0 (0) 0 (0) 3 (50) 2 (33) 0 (0)
2 (40) 2 (40) 1 (20) 1 (20) 0 (0)
1 (17) 0 (0) 5 (83) 0 (0)
0 (0) 0 (0) 3 (60) 2 (40)
0 (0)
0 (0)
Kinesiophobia is defined as excessive, irrational, and debilitating fear of physical activity resulting from a sense of vulnerability to painful injury or reinjury.15 Kinesiophobia was assessed with the English language version of the 11-item Tampa Scale of Kinesiophobia (TSK-11).16 The TSK-11 uses a four-point Likert response scale ranging from Bstrongly disagree[ (rated as 1) to Bstrongly agree[ (rated as 4). Summing item scores yields a total fear of pain score ranging from 11 (disagreement that fear of pain limits movement) to 44 (endorsement that fear of pain limits movement). The English language version of the TSK-11 has demonstrated strong reliability and construct validity.16,17 The TSK-11 was administered at baseline, at the conclusion of training, and at the 3-mo follow-up assessment. TSK-11 scores were analyzed with a 2 3 (group measurement occasion) ANOVA with repeated measures on measurement occasion.
unable to be contacted after several sessions, and one was removed from the study after a diagnosis of heart arrhythmia. Three full feedback group participants discontinued the study before the posttest: one was unable to be contacted after several sessions, one had a recurrence of breast cancer, and one could not arrange transportation after his third session. One faded feedback group participant and two full feedback group participants completed biofeedback training but failed to return follow-up surveys after five attempts to contact by phone and were therefore classified as lost to follow-up. t Tests comparing baseline inventory and pain scores of attrited participants with those who completed the study did not reveal major differences among the groups, attenuating risk for selection bias (all P 9 0.05). Further analyses were conducted on the six faded feedback group participants and five full feedback group participants who completed the study.
RESULTS
Participant Characteristics
Attrition Analyses
No important differences among the groups were detected for demographic or baseline pain characteristics (Table 2; all P 9 0.431).
Participant enrollment and flow is shown in Figure 1. Attrition rates included participants who discontinued biofeedback training and those who completed training but did not return follow-up surveys. Three faded feedback group participants discontinued the study before the posttest: two were www.ajpmr.com
Heart Rate Variability Neither the group main effect (P = 0.656) nor the group session interaction (P = 0.995) revealed Biofeedback Scheduling in Chronic Pain
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
873
FIGURE 1 Study flow diagram. important differences among the groups in controlling HRV across the nine biofeedback sessions (Fig. 2).
Use of Biofeedback Skills After the Program Table 1 displays responses by group to followup survey questions. A larger proportion in the faded feedback group continued to use skills acquired during training (P = 0.132 for the difference between the groups). In addition, a larger proportion in the faded feedback group used the skills half the time or more (P = 0.092 for the difference between the groups). Finally, a greater proportion of those in the full feedback group indicated that they had not used the skills to control pain because they did not work (P = 0.179 for the difference between the groups).
Pain Intensity Figure 3 portrays mean VAS ratings for current level of pain. The group main effect did not indicate important differences among the groups (P = 0.506); however, the group measurement occasion interaction revealed that current pain was decreased relative to baseline at the end of biofeedback training and at follow-up in the faded feedback group (P = 0.291). In contrast, current pain was unchanged in the full feedback group at the end of biofeedback training and increased over baseline at follow-up. Figure 4 portrays mean VAS ratings for worst pain in the last 24 hrs. The group main effect did not indicate important differences among the groups (P = 0.519); however, the group measurement occasion interaction revealed that worst pain was decreased relative to baseline at the end of biofeedback training and at follow-up in the faded
TABLE 2 Demographic and baseline pain characteristics of the participants in each group Group
Sex, n (%) Male Female Employment status, n (%) Full-time regular duty Part-time regular duty Part-time modified duty Have not worked/retired Age, mean (SD), yrs Baseline current pain, mean (SD), cm Baseline worst pain in 24 hrs, mean (SD), cm
874
Weeks et al.
Faded Feedback (n = 6)
Full Feedback (n = 5)
3 (50) 3 (50)
3 (60) 2 (40)
1 (17) 2 (33) 0 (0) 3 (50) 56.7 (10.8) 4.17 (1.94) 8.17 (1.72)
0 (0) 2 (40) 1 (20) 2 (40) 60.2 (8.1) 3.85 (1.92) 7.80 (2.30)
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
FIGURE 2 Percentage of time that low-frequency HRV was achieved in each group per biofeedback session. feedback group (P = 0.148). In contrast, worst pain was unchanged in the full feedback group at the end of the biofeedback training and increased over baseline at follow-up.
Functional Status Neither the group main effect (P = 0.464) nor the group measurement occasion interaction (P = 0.384) revealed consistent differences among the groups in ratings of perceived disability as assessed by the PDQ. Mean (SE) PDQ scores were 36.8 (4.7) for the faded feedback group and 42.1 (5.2) for the full feedback group.
Fear of Pain Figure 5 portrays mean TSK-11 scores. The group main effect did not indicate important differences
among the groups (P = 0.474); however, the group measurement occasion interaction revealed that fear of physical activity due to pain decreased at the end of the biofeedback training in the faded feedback group (P = 0.102). The reduction in fear continued into the follow-up period. In contrast, fear of physical activity was relatively unchanged in the full feedback group.
DISCUSSION This pilot study provided evidence that feedback scheduling during skill training that reduced dependence on biofeedback to gain control of HRV had sustained benefits on several measures of reactivity to pain. As hypothesized, the group receiving a high frequency of external feedback about ability to control HRV and the group receiving a low
FIGURE 3 Mean VAS pain ratings per group for current pain across the three measurement occasions. www.ajpmr.com
Biofeedback Scheduling in Chronic Pain Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
875
FIGURE 4 Mean VAS pain rating scores per group for worst pain in the last 24 hrs across the three measurement occasions.
frequency of external feedback performed the biofeedback task similarly at the end of nine sessions of training. Thus, reduced feedback frequency was not detrimental to training performance. Again, as hypothesized, at the 3-mo follow-up assessment, the group experiencing a low frequency of feedback in training reported more frequent use of the biofeedback skills, suggesting that these skills remained effective in the months after training. Support was also revealed for the hypothesis that more proficient use of biofeedback skills in the faded feedback group would result in lower pain intensity at the conclusion of training and at the 3-mo follow-up.
Finally, as hypothesized, the greater proficiency in modulating physical symptoms afforded by the faded feedback schedule reduced activity-related fear of pain to a greater degree than that by the full feedback schedule at the conclusion of biofeedback training and at the 3-mo follow-up evaluation. The principle of reducing relative frequency of feedback in acquisition to enhance long-term skill retention has not been established for medically oriented skills. Evidence that the guidance hypothesis can be extended to medical skills, such as control over HRV, is encouraging. HRV is a measure of beat-to-beat neurocardiac function reflecting
FIGURE 5 Mean TSK-11 scores per group across the three measurement occasions.
876
Weeks et al.
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
both sympathetic and parasympathetic autonomic nervous system activity.18,19 Chronic pain tends to arouse sympathetic activity and suppress parasympathetic activity. Control over HRV affects a shift in autonomic function away from a stress-related sympathetic response toward an increased parasympathetic response, which is involved in the inhibition of pain signals and sympathetic outflow. The faded feedback condition may allow participants to engage in self-management of the parasympathetic response to pain during biofeedback training so that, at the conclusion of training, participants have greater self-sufficiency in managing pain than that afforded by the full feedback schedule. Alternately, the full feedback schedule may discourage learning to regulate HRV intrinsically because the need for self-regulation is nullified when external biofeedback is constantly present, resulting in poorer capability to control HRV at the conclusion of training and less frequent use of biofeedback skills in the long term. It is also encouraging that reducing feedback frequency may influence psychologic constructs reactive to pain, such as fear of pain. Fear of pain due to movement and activity-avoidance behaviors in anticipation of injury (or reinjury) are significant and common psychologic complications of chronic pain. The fear-avoidance model of chronic pain suggests that activity-related physiologic symptoms, such as variability in heart rate, may result in exaggerated pain sensations.20 These misinterpreted physiologic symptoms reinforce the fear that physical activity will be painful. As fear is elevated, avoidance of activity increases. Consequently, avoidance of movement leads to a cycle of physical deconditioning and further disability. Pain-related fear has been shown to be as potent a predictor of physical disability as pain severity21Y23 and can have a significant negative impact on treatment outcome.15 For this reason, it is encouraging that enhanced learning through reduced feedback frequency was associated with reduced fear-avoidance beliefs. The purposes of this pilot study were, among other things, to understand attrition rates and to obtain effect sizes for sample size estimates for a larger randomized clinical trial that includes a follow-up retention test of skill in controlling HRV. In the chronic pain population, attrition rates from psychologic-based management programs are known to be high.24 Similarly, attrition rates in this pilot study were high in both groups (40% in the faded feedback group and 60% in the full feedback group). Although the authors intentionally implemented a multiple callback feature as part of the design of this pilot study, it is clear that other www.ajpmr.com
methods to avoid attrition must be used in the forthcoming larger trial. For example, use of a runin period in which prospective participants attend several appointments before inclusion can determine participants with a high probability of being adherent to the study schedule. In addition to these protocol features, sample size estimates in the larger trial will take into consideration the potential for attrition as high as 60%. On the basis of effect sizes from this study and the large attrition rates, the estimates in this study suggest that sample size in a larger trial will need to be two-anda-half times greater to achieve statistical significance at the 0.05 level of confidence. Among other adjustments, the single outcome for which there was no trend in favor of the faded feedback schedule was self-rated assessment of disability perceptions with the PDQ. The authors posit two possible explanations for why the groups did not differ on PDQ scores: It is possible that disability orientation is not modifiable in response to biofeedback training. Alternatively, the PDQ may not be sensitive enough to measure differential effects of biofeedback schedules on perceptions of pain interference with functional status. In the larger trial, the authors will include additional measures of function and disability to enhance measurement of these constructs. Despite the study’s limitations, its data provide evidence that systematically reducing the frequency of biofeedback results given to the learner with chronic pain is associated with several positive outcomes over an extended period. This information could be useful in facilitating a more costeffective biofeedback regimen for people with chronic pain. The extent to which these findings apply generally to patients with chronic pain will require larger studies to determine whether the findings of this study can be replicated and, as such, establish their generalizability. REFERENCES 1. Giggins OM, Persson UM, Caulfield B: Biofeedback in rehabilitation. J Neuroeng Rehabil 2013;10:60 2. Morley S, Eccleston C, Williams A: Systematic review and meta-analysis of randomized controlled trials of cognitive behaviour therapy and behaviour therapy for chronic pain in adults, excluding headache. Pain 1999;80:1Y13 3. Henschke N, Ostelo RWJG, van Tulder MW, et al: Behavioural treatment for chronic low-back pain. Cochrane Database Syst Rev 2010 4. Winstein CJ, Schmidt RA: Reduced frequency of knowledge of results enhances motor skill learning. J Exp Psychol Learn Mem Cogn 1990;16:677Y91
Biofeedback Scheduling in Chronic Pain Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
877
5. Weeks DL, Kordus RA: Relative frequency of knowledge of performance and motor skill learning. Res Q Exerc Sport 1998;69:224Y30
15. Kori SH, Miller RP, Todd DD: Kinesiophobia: A new view of chronic pain behavior. Pain Manag 1990;3:35Y43
6. Anderson DI, Magill RA, Sekiya H, et al: Support for an explanation of the guidance effect in motor skill learning. J Mot Behav 2005;37:231Y8
16. Woby SR, Roach NK, Urmston M, et al: Psychometric properties of the TSK-11: A shortened version of the Tampa Scale for Kinesiophobia. Pain 2005;117:137Y44
7. Salmoni AW, Schmidt RA, Walter CB: Knowledge of results and motor learning: A review and critical reappraisal. Psychol Bull 1984;95:355Y86
17. Tkachuk GA, Harris CA: Psychometric properties of the Tampa Scale for Kinesiophobia-11 (TSK-11). J Pain 2012;13:970Y7
8. Wulf G, Shea C, Lewthwaite R: Motor skill learning and performance: A review of influential factors. Med Educ 2010;44:75Y84
18. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology: Heart rate variability: Standards of measurement, physiological interpretation, and clinical use. Circulation 1996;93:1043Y65
9. Ronsse R, Puttemans V, Coxon JP, et al: Motor learning with augmented feedback: Modalitydependent behavioral and neural consequences. Cereb Cortex 2011;21:1283Y94 10. Perini R, Veicsteinas A: Heart rate variability and autonomic activity at rest and during exercise in various physiological conditions. Eur J Appl Physiol 2003;90:317Y25 11. Jensen MP, Turner JA, Romano JM, et al: Comparative reliability and validity of chronic pain intensity measures. Pain 1999;83:157Y62 12. Price DD, Bush FM, Long S, et al: A comparison of pain measurement characteristics of mechanical visual analogue and simple numerical rating scales. Pain 1994;56:217Y26
878
19. McCraty R, Atkinson M, Tiller WA, et al: The effects of emotions on short-term heart rate variability using power spectrum analysis. Am J Cardiol 1995;76:1089Y93 20. Norton PJ, Asmundson GJG: Amending the fearavoidance model of chronic pain: What is the role of physiological arousal? Behav Ther 2003;34:17Y30 21. Crombez G, Vlaeyen JW, Heuts PH, et al: Fear of pain is more disabling than pain itself: Evidence on the role of pain-related fear in chronic back pain disability. Pain 1999;80:329Y40 22. McCracken L, Faber SD, Janeck AS: Pain-related anxiety predicts non-specific physical complaints in persons with chronic pain. Behav Res Ther 1998;36: 621Y30
13. Anagnostis C, Gatchel R, Mayer TG: The pain disability questionnaire: A new psychometrically sound measure for chronic musculoskeletal disorders. Spine 2004;29:2290Y302
23. Vlaeyen JW, Linton SJ: Fear-avoidance and its consequences in chronic musculoskeletal pain: A state of the art. Pain 2000;85:317Y32
14. Gatchel RJ, Mayer TG, Theodore BR: The Pain Disability Questionnaire: Relationship to one-year functional and psychosocial rehabilitation outcomes. J Occup Rehabil 2006;16:75Y94
24. Williams AC, Eccleston C, Morley S: Psychological therapies for the management of chronic pain (excluding headache) in adults. Cochrane Database Syst Rev 2012;11
Weeks et al.
Am. J. Phys. Med. Rehabil. & Vol. 94, No. 10, October 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
