Table 2.

Summary of Systematic Reviews of Mind-Body Therapies (MBTs).

First Author, YearConditionNumber and Type of Studies ReviewedTotal Number of SubjectsInterventionsResultsQuality Rating13
Anderson & Lyttkens35Tinnitus8 randomized controlled trials700Cognitive-behavioral, relaxation, hypnosis, biofeedback, education, problem solvingEffect sizes:* Annoyance = 0.86 Loudness = 0.68 Negative affect = 0.48 Sleep = 0.26 Effects diminished at follow-up4
Astin et al36Rheumatoid arthritis24 randomized controlled trials1,189Psychological-psychosocial approachesEffect sizes: Pain = 0.22 (0.06 at follow-up) Disability = 0.36 (0.20 at follow-up)5
Berghmans et al37Urinary incontinence11 randomized controlled trials240Pelvic floor muscle exercises with or without biofeedbackNo evidence that addition of biofeedback is helpful4
Brown38Diabetes82 trials (26 single group design)Not providedPatient education including relaxation, stress reduction, and cognitive-behavioral counselingMetabolic control (0.16–0.41)5
Carroll & Seers39Chronic pain9 randomized trials414RelaxationInsufficient evidence to support use of relaxation4
Devine & Pearcy40Chronic obstructive pulmonary disease65 studies (54% controlled; 34% randomized)3,642Psychoeducational care (see Table 1)Endurance = 0.77 Functional status = 0.637
Devine41Asthma31 studies (58% randomized)1,860Psychoeducational careAsthma attacks = 0.56 Respiratory volume = 0.34 Expiratory flow rate = 0.29 Functional status = 0.46 Medication use = 0.625
Devine42Presurgery191 studies (69% randomized)Not availablePresurgical psychoeducational interventionsRecovery = 0.43 Pain = 0.387
Dusseldorp43Coronary heart disease37 studies (75% randomized)9,699Health education and stress management34% reduction in cardiac mortality 29% reduction in recurrent events5
Eisenberg et al44Hypertension26 randomized trials1,264Cognitive behavioral techniques (including meditation, relaxation, biofeedback, stress management)Significant blood pressure declines compared with usual care or wait-list control, but not significant compared with placebo or sham condition5
Glanz et al45Post stroke rehabilitation8 randomized trials180BiofeedbackLower extremity range of motion = 0.5 Upper extremity = 2.37
Haddock et al46Chronic benign headache20 randomized trialsNot availableHome and clinic-based behavioral treatments46% (home based) and 53% (clinic based) showed significant improvement (ie, 50% reduction in headache activity)5
Hadhazy et al47Fibromyalgia13 randomized trials802Mind-body therapies including autogenic training, relaxation, meditation, biofeedback, cognitive therapy, hypnosisLimited evidence that MBTs are effective in fibromyalgia5
Hermann et al48Pediatric migraine17 studies (9 control group design)92Behavioral treatments (typically biofeedback or relaxation, or multi-component MBT)All behavioral treatments more effective than placebo or wait-list although thermal biofeedback alone or in combination with relaxation appeared most effective4
Holyrod & Penzien49Migraine35 studies (63% used between-group comparisons)2,445Biofeedback and/or relaxationRelaxation and biofeedback appear equally effective compared with pharmacologic approaches (43% reduction in migraine activity)3
Johnston & Vogele50Surgery34 randomized trials1,774Psychological preparationPain = 0.85 Pain medication = 0.60 Length of stay = 0.65 Recovery = 0.612
Linden & Chambers51Hypertension89 randomized trials1,651Psychologically based nonpharmacologic treatmentsIndividualized therapy appears most effective, ES = 0.65 for systolic blood pressure Multicomponent ES = 0.51 Single method ES = 0.472
Linden et al52Coronary artery disease23 randomized trials2,024Addition of psychosocial treatment to standard cardiac rehabilitationPatients not receiving psychosocial adjunct had adjusted odds ratio for 2-y mortality of 1.7 and 1.84 for recurrence—effects weakened at follow-up3
Meyer & Mark53Cancer45 randomized trials2,840Psychosocial interventions (eg, cognitive behavioral, social support, educational, and multimodal interventions)Effect sizes: Functional adjustment = 0.19 Treatment and disease-related symptoms = 0.263
Moreland et al54Stroke8 randomized trialsNot availableBiofeedbackSignificant changes in ankle muscle strength, ES = 1.27
Morin et al55Insomnia59 controlled trials2,102Psychological interventions (eg, stimulus control, relaxation)Significant changes in sleep latency, ES = 0.88 Time awake after sleep, ES = 0.653
Morley et al56Chronic pain25 randomized trials1,672Cognitive behavioral and behavioral therapy (including biofeedback, relaxation)Mean ES = 0.5 across all domains (eg, pain, function, mood)7
Mullen et al57Arthritis15 controlled trials (13 randomized)Not availablePsychoeducational interventionsEffect sizes: Pain = 0.20 Disability = 0.062
Murtagh & Greenwood58Insomnia66 controlled trials1,907Psychological treatmentsSleep latency = 0.87 Total sleep time = 0.49 Awakenings = 0.63 Sleep quality = 0.945
Scott et al59Childbirth11 randomized trials4,230Emotional supportContinuous support resulted in shorter labor (mean difference of −1.64 h), less analgesia (OR = 0.64), oxytocin (OR = 0.29), forceps (OR = 0.43) and cesarean section (OR = 0.49)4
Seers & Carroll60Acute pain7 randomized trials821RelaxationSome weak evidence (3/7 studies) suggesting positive effect of relaxation5
Superio-Cabuslay61Osteoarthritis19 controlled trials3,148Patient education (eg, arthritis self-management program)Effect sizes: Pain = 0.17 Disability = 0.03 Tender joints (RA) = 0.281
van Tulder et al62Low back pain20 randomized trials1,349Behavioral and cognitive-behavioral treatmentsEffect sizes: Pain = 0.62 Functional status = 0.357
  • * Effect sizes are typically derived from Cohen’s d statistic,34 which represents the difference of the group means divided by their pooled standard deviation. Effect sizes therefore represent the difference between groups expressed in standard deviation units (ie, an effect size of 1.00 signifies a difference of one standard deviation in the outcome measure between the treatment and control groups). As suggested by Cohen, effect sizes between 0.20 and 0.50 are considered small, those between 0.50 and 0.80 moderate, and those greater than 0.80 large. Effect sizes in the behavioral sciences tend to be in the small to moderate range. In addition, even small effect sizes can represent important clinical phenomena, particularly when they represent changes in an outcome, such as mortality, or could potentially effect large numbers in the population.

    ES = effect size, OR = odds ratio, RA = rheumatoid arthritis.