Several methodologic difficulties exist in assessing biofeedback. For example, most interventions that include biofeedback are multimodal and include relaxation and behavioral instruction, which may have effects separate from those that may occur due to biofeedback. While studies may report a beneficial effect of multimodality treatment, without appropriate control conditions, it is impossible to isolate the specific contribution of biofeedback to the overall treatment effect. For example, relaxation, attention, or suggestion may account for the successful results that have been attributed to biofeedback. These are nonspecific therapeutic factors, some of which can be considered placebo effects. Moreover, it is important that studies demonstrate that biofeedback improves disease-related health outcomes, as opposed to potentially affecting only physiologic, intermediate outcomes, and that they address the durability of effects beyond the initial, short-term biofeedback training period.
A 1995 Blue Cross Blue Shield Association Technology Evaluation Center (TEC) Assessment reviewed the literature on the use of biofeedback in the treatment of nine different conditions: anxiety disorders, headaches, hypertension, movement disorders, incontinence, pain, asthma, Raynaud’s disease, and insomnia. (1) The Assessment concluded that, due to methodologic limitations of the literature, there was insufficient evidence to conclude that biofeedback provides benefit in treating any of the nine conditions. While a substantial number of studies reported improvement in the biofeedback group relative to the no-treatment group, there were generally no differences when the isolated effect of biofeedback was compared with relaxation or behavioral therapy alone. In addition, although there was evidence that feedback on physiologic processes provides patients with an enhanced ability to control these processes, there was, nevertheless, no consistent evidence of any relationship between a patient’s ability to exert control over the targeted physiologic process and any health benefits of the intervention.
Following is a summary of updated literature on topics covered in the 1995 TEC Assessment, as well as in new literature on biofeedback for any other miscellaneous conditions not considered in other medical policies. Updated literature searches have focused on identifying randomized controlled trials (RCTs) and meta-analyses.
A systematic review of studies on biofeedback for hypertension was published by Greenhalgh and colleagues in 2010. (2) The investigators searched for RCTs that included adults with essential hypertension (defined as at least 140/90 mm Hg) and that compared biofeedback interventions, alone or in combination with other therapies, to medication, sham biofeedback, no treatment, or another behavioral intervention. A total of 36 trials (n=1,660) met inclusion criteria. Trials generally had small sample sizes; only 4 included more than 100 patients. All were single-center, and most were conducted in the United States. Trials used a variety of biofeedback techniques including thermal biofeedback, galvanized skin response, pulse wave velocity, and heart rate variability; some trials used more than one modality. Twenty studies evaluated biofeedback alone, fifteen evaluated biofeedback combined with another intervention, and one had multiple arms and evaluated both types of interventions; only four trials included a sham biofeedback comparison group. The authors stated that they did not pool study findings due to differences in interventions and outcomes and the generally poor quality of the studies.
The investigators reported that trials comparing biofeedback alone versus no treatment or another behavioral intervention did not provide convincing evidence of the superiority of biofeedback. Only one of five trials that compared a biofeedback combination intervention (most commonly combined with relaxation) to a different behavioral treatment found the biofeedback intervention to be superior. Approximately half of the trials comparing a biofeedback combination to no treatment found a significant benefit to the biofeedback combination, but the specific effects of biofeedback cannot be determined from this analysis. Only one trial was identified that compared a biofeedback combination intervention to sham biofeedback, and this study did not find a significant difference in the efficacy of the two interventions. Four studies on biofeedback alone and another four on a combined biofeedback intervention reported data beyond six months; most of these found no significant differences in efficacy between the biofeedback and control groups. Greenhalgh and colleagues concluded, “…we found no convincing evidence that consistently demonstrates the effectiveness of the use of any particular biofeedback treatment in the control of essential hypertension when compared with pharmacotherapy, placebo, no intervention or other behavioral therapies.”
In a previous meta-analysis, published in 2003, Nakao and colleagues found that biofeedback was effective in lowering systolic and diastolic blood pressure but only when the biofeedback was combined with relaxation techniques. (3) The authors further noted that study is needed to determine whether biofeedback has any blood pressure lowering effect without relaxation techniques.
Motor function after stroke, injury, or lower-limb surgery
Several systematic reviews have been published; none of these conducted quantitative pooling of results due to heterogeneity among study populations, interventions, and outcome measures. A 2010 systematic review by Silkman and McKeon evaluated the effectiveness of electromyography (EMG) biofeedback for improving muscle function during knee rehabilitation after injury. (4) Four RCTs that compare knee rehabilitation exercise programs with and without biofeedback were identified. Sample sizes in individual studies ranged from 26 to 60 patients. Two of the four studies found a statistically significantly greater benefit in the programs that included biofeedback, and the other two did not find a significant difference between groups. The positive studies assessed intermediate outcomes e.g., contraction values of the quadriceps muscles. None of the studies were designed to assess functional outcomes.
A Cochrane review that assessed electromyographic (EMG) biofeedback for the recovery of motor function after stroke was published in 2007. (5) It included thirteen randomized or quasi-randomized studies with a total of 269 patients. All of the trials compared EMG biofeedback plus standard physiotherapy to standard physiotherapy; in addition to standard physiotherapy, several studies also included a sham biofeedback group. The studies tended to be small and poorly designed. The authors did not find support for EMG biofeedback to improve motor power, functional recovery, or gait quality when compared to physiotherapy alone.
A systematic review by Zijlstra and colleagues, published in 2010, searched for studies evaluating biofeedback-based training to improve mobility and balance in adults older than 60 years of age. (6) Although the review was not limited to studies on motor function after stroke, more than half of the studies included older adults post-stroke. For inclusion in this review, studies needed to include a control group of patients who did not receive biofeedback and to assess at least one objective outcome measure. A total of 97 potentially relevant articles were identified, and 21 (22%) studies, including 17 RCTs, met the selection criteria. Twelve of the 21 (57%) studies included individuals post-stroke, 3 included older adults who had lower-limb surgery, and 6 included frail older adults without a specific medical condition. Individual studies were small; sample sizes ranged from 5 to 30 patients. The added benefit of using biofeedback could be evaluated in 13 of 21 (62%) studies. Nine of the 13 studies found a significantly greater benefit with interventions that used biofeedback compared to control interventions. However, the outcomes assessed were generally not clinical outcomes but were laboratory-based measures related to executing a task, e.g., moving from sitting to standing in a laboratory setting and platform-based measures of postural sway. The applicability of improvements in these types of measures to clinical outcomes such as the ability to perform activities of daily living or the rate of falls, is unknown. Only one study cited in this review reported an improvement in fall rates, and this trial could not isolate the effect of biofeedback from other components of treatment. In addition, only three studies reported long-term outcomes, and none of these reported a significant effect of biofeedback. Conclusions about the efficacy of biofeedback for improving mobility and balance in older adults cannot be drawn from these data due to the lack of evidence on clinical outcomes. Other methodologic limitations include limited data on the durability of effects and the inability to isolate the effect of biofeedback in many studies.
A 2010 RCT, not included in the Zijlstra et al. review, evaluated biofeedback to improve motor function in patients who were at least 6 months post-stroke. (7) The study, conducted in Italy by Jonsdottir and colleagues, randomized 20 patients to 20 sessions of EMG biofeedback (n=10) or standard rehabilitation (n=10). Patients in both groups received sessions lasting 45 minutes, 3 times a week. The biofeedback consisted of an acoustic signal; patients in the intervention group wore a biofeedback belt device. All patients completed the 20 sessions, and 9 in each group (a total of 90%) were available for the follow-up 6 weeks after completion of the intervention. The analyses found statistically significant effects of the biofeedback intervention on the outcome variables of ankle power peak, velocity, and stride length but not knee flexion peak from baseline evaluation to the final follow-up. For example, in the treatment group, stride length (percent height per second) increased from 44.1 pre-treatment to 51.1 at final follow-up, and stride length in the control group increased from 33.4 pre-treatment to 35.2 at final follow-up. Although positive, data from this study alone cannot change the conclusion of an insufficient body of evidence on biofeedback to improve motor function after stroke. Moreover, the study did not evaluate outcomes related to activities of daily living, and the biofeedback protocol used in the study has not been replicated in other studies.
A 2009 systematic review on complementary and alternative medicine in the treatment of Raynaud’s disease included an examination of the literature on biofeedback. (8) The authors identified five trials, and these reported a variety of outcomes. A pooled analysis of findings from four trials (total n=110) on the change in frequency of attacks favored the sham control group over the biofeedback group (weighted mean difference: -1.21; 95% confidence interval [CI]: -1.68 to -0.73; p<0.00001). Several trials had more than 2 arms; in the preceding analysis, only the arms comparing active and sham biofeedback were included.
The trial that was given the highest quality rating by the authors of the systematic review and had the largest sample size was the Raynaud’s Treatment Study, published in 2000. (9) This was a randomized comparison of sustained-release nifedipine and thermal biofeedback in 313 patients with primary Raynaud’s disease. In addition to these two treatment groups, there were two control treatments: pill placebo and EMG biofeedback. EMG biofeedback was chosen as a control because it did not address the physiologic mechanism of Raynaud’s disease. The mean attack rate at one year, the primary study outcome, was 0.16 in the thermal biofeedback group, 0.23 in the EMG biofeedback group, 0.07 in the nifedipine group, and 0.21 in the placebo group. Nifedipine significantly reduced Raynaud’s attacks compared with placebo (p<0.002), but thermal feedback did not differ significantly from EMG biofeedback (0 37). There was not a significant difference in attack rates in the nifedipine and thermal biofeedback groups for the primary outcome (p=0.08). However, several secondary outcomes including all attacks and verified attacks at two months significantly favored nifedipine over thermal biofeedback.
One small randomized study (n=57) examined changes in sleep bruxism following treatment with a cognitive behavioral therapy program consisting of problem-solving, progressive muscle relaxation, nocturnal biofeedback, and training of recreation and enjoyment. (10) Similar improvements were observed for the occlusal splint group as for the multicomponent cognitive behavioral program. The effects of biofeedback were not isolated in this study and thus conclusions cannot be drawn about its effectiveness compared to occlusal splinting.
An RCT by Weise et al. (11) investigated the efficacy of a biofeedback-based cognitive-behavioral treatment for tinnitus in Germany. Tinnitus patients (n=130) were randomly assigned to an intervention or a waiting-list control group. Treatment consisted of twelve sessions of a biofeedback-based behavioral intervention over a 3-month period. The primary outcome measures were global tinnitus annoyance and a daily rating of tinnitus disturbance measured by a Tinnitus Questionnaire (TQ) and a daily diary using visual analog scale (VAS) scores. Patients in the waiting-list group participated in the treatment after the intervention group had completed the treatment. Results showed improvements regarding the following: tinnitus annoyance; diary ratings of loudness; feelings of controllability; changes in coping cognitions; changes in depressive symptoms; TQ: total score (range 0–84) pre-assessment mean 54.7, post-assessment mean 32.52; TQ: emotional distress (range 0–24) pre-assessment mean 16.00, post-assessment mean 8.15; and diary: loudness VAS (range 0–10) pre-assessment mean 5.68, post-assessment mean 4.38. Improvements were maintained over a 6-month follow-up period in which variable effect sizes were observed. The study does not investigate the possible additive effect of biofeedback with cognitive-behavioral therapy and did not include an active treatment control group. In conclusion, these data are insufficient to draw clinical conclusions regarding the role of biofeedback for the treatment of tinnitus.
In 2008, Cardoso et al. (12) published a systematic review of studies on the effects of facial exercises on symptoms of Bell's palsy. Studies including patients with unilateral idiopathic facial palsy treated with facial exercises associated with mirror and/or EMG biofeedback were included in this review. Four studies (n=132) met the eligibility criteria. The studies described mime therapy versus control (n=50), mirror biofeedback exercise versus control (n=27), "small" mirror movements versus conventional neuromuscular retraining (n=10), and EMG biofeedback plus mirror training versus mirror training alone. The treatment length varied from 1 to 12 months. The authors concluded that “…because of the small number of randomized controlled trials, it was not possible to analyze if the exercises, associated either with mirror or EMG biofeedback, were effective. In summary, the available evidence from randomized controlled trials is not yet strong enough to become integrated into clinical practice.”
Orthostatic hypotension in patients with a spinal cord injury
Gillis et al. (13) conducted a systematic review to identify and describe the body of literature pertaining to nonpharmacologic management of orthostatic hypotension during the early rehabilitation of persons with a spinal cord injury. Participants with any level or degree of completeness of spinal cord injury and any time elapsed since their injuries were included. Interventions must have measured at least systolic blood pressure and have induced orthostatic stress in a controlled manner and have attempted to control orthostatic hypotension during an orthostatic challenge. Four distinct nonpharmacologic interventions for orthostatic hypotension were identified: application of compression and pressure to the abdominal region and/or legs, upper body exercise, functional electrical stimulation applied to the legs, and biofeedback. Methodologic quality varied dramatically between studies. The authors concluded that “…The clinical usefulness of compression/pressure, upper body exercise and biofeedback for treating OH [orthostatic hypotension] has not been proven.”
Ongoing clinical trials
Improving function after knee arthroplasty with weight-bearing biofeedback (NCT01333189) (14): This randomized controlled trial is comparing a weight-bearing exercise program using biofeedback to a usual-care exercise program. The primary study outcome is weight-bearing symmetry 6 weeks post-surgery; secondary outcomes include the 6-minute walk test and stair climb test. The estimated date of study completion is December 2012.
A 1995 TEC Assessment found insufficient evidence to demonstrate the effectiveness of biofeedback for asthma, anxiety disorders, insomnia, movement disorders, Raynaud’s disease, and hypertension. There continues to be insufficient evidence that biofeedback benefits these conditions. In addition, literature reviews have found insufficient evidence from randomized controlled trials to support the efficacy of biofeedback for treating sleep bruxism, tinnitus, Bell’s palsy, motor function after stroke, injury or lower-limb surgery, and orthostatic hypotension in patients with spinal cord injury. Studies either failed to show any beneficial impact of biofeedback or had design flaws that leave the durability of effects in question or create uncertainty about the contribution of nonspecific factors such as attention or placebo effects. Thus, biofeedback is considered experimental, investigational and unproven for these miscellaneous conditions.
Practice Guidelines and Position Statements
In 2006, an American Academy of Sleep Medicine Report update was released entitled Practice Parameters for the Psychological and Behavioral Treatment of Insomnia. (15) In the section, Recommendations for Specific Therapies, item 3.9, the report states that “Biofeedback is effective and recommended therapy in the treatment of chronic insomnia. (Guideline)” The American Academy of Sleep Medicine (AASM) definition for guideline is “a patient-care strategy, which reflects a moderate degree of clinical certainty. The term guideline implies the use of Level II Evidence (randomized trials with high alpha and beta error) or a consensus of Level III Evidence (non-randomized concurrently controlled studies).”
No other relevant guideline or position statement was identified on the National Guideline Clearinghouse web site.