Mindfulness-based stress reduction (MBSR) for improving health, quality of life and social functioning in adults: a systematic review and meta-analysis
Journal article, Peer reviewed
MetadataShow full item record
Original versionCampbell systematic reviews. 2017, 13 (11). 10.4073/csr.2017.11
This review summarizes all studies that compare the effect of a MBSR program to a control group intervention, in which the participants had been randomly allocated to be in either the MBSR group or a control group. The review summarizes the results in two categories. First, where the effect of the MBSR program was compared to an inactive group (either a wait list group or one receiving ordinary care also received by the MSBR group). Second, where MBSR was compared with an alternative active group intervention. What are the main results? MBSR has a moderately large effect on outcome measures of mental health, somatic health, and quality of life including social function at post-intervention when compared to an inactive control. If 100 people go through the MBSR program, 21 more people will have a favourable mental health outcome compared to if they had been put on a wait-list or gotten only the usual treatment. These results may be inflated by underreporting of negative trials and moderate heterogeneity (indicating differences between the trials). MBSR has a small but significant effect on improving mental health at post-intervention compared to other active treatments. MBSR has the same effect as other active interventions on somatic health, and quality of life (including social function). There was no underreporting of negative trials, and heterogeneity (differences between trials) were small for mental health, moderate for quality of life and large for somatic health. The effects were similar across all target groups and were generally maintained at follow-up (1–34 months). The effects were largely independent of gender and study sample. The effects seemed also largely independent of duration and compliance with the MBSR intervention. No studies report results regarding side-effects or costs. Effects were strongly correlated to the effects on measures of mindfulness, indicating that the effects may be related to the increase in self-reported mindfulness. Two thirds of the included studies showed a considerable risk of bias, which was higher among studies with inactive than active control groups. Studies of higher quality reported lower effects than studies with low quality. The overall quality of the evidence was moderate, indicating moderate confidence in the reported effect sizes. Further research may change the estimate of effect. Background There is an increasing focus on mind-body interventions for relieving stress, and improving health and quality of life, accompanied by a growing body of research trying to evaluate such interventions. One of the most well-known Programs is Mindfulness-Based Stress Reduction (MBSR), which was developed by Kabat-Zinn in 1979. Mindfulness is paying attention to the present moment in a non-judgmental way. The Program is based on old contemplative traditions and involves regular meditation practice. A number of reviews and meta-analyses have been carried out to evaluate the effects of meditation and mindfulness training, but few have adhered to the meta-analytic protocol set out by the Cochrane Collaboration and Campbell Collaboration, or focused on MBSR only. The first edition of this review was published in 2012 with a literature search done in 2010, comprising 31 studies. As the field is rapidly developing, an update is called for. Objectives To evaluate the effect of Mindfulness-Based Stress Reduction (MBSR) on health, quality of life and social functioning in adults. Search methods The following sources were searched, most recently in November 2015: PsycINFO (Ovid), MEDLINE (Ovid), EMBASE (Ovid), AMED (Allied and Complementary Medicine) (Ovid), CINAHL (Ebsco), Ovid Nursing Full Text Plus (Ovid), Cochrane Central Register of Controlled Trials (CENTRAL), British Nursing Index, (ProQuest), Eric (ProQuest), ProQuest Medical Library, ProQuest Nursing & Allied Health Source, ProQuest Psychology Journals, Web of Science, SveMed+, Social Services Abstracts, Sociological Abstracts and International Bibliography of Social Sciences. Selection criteria The review included randomised controlled trials (RCTs) where the intervention followed the MBSR protocol developed by Kabat-Zinn, allowing for variations in the length of the MBSR courses. All target groups were accepted, as were all types of control groups, and no language restrictions were imposed. Data collection and analysis Two reviewers read titles, retrieved studies, and extracted data from all included studies. Standardized mean differences (as Hedges' g) from all study outcomes were calculated using the software Comprehensive Meta Analysis. The meta-analyses were carried out using the Robumeta Package within the statistical program R, with a technique for handling clusters of internally correlated effect estimates. We performed separate meta-analyses for MBSR compared to either waitlists or treatment as usual (WL/TAU – named inactive), and for MBSR compared to control groups that were offered another active intervention. Results The review identified 101 RCTs including the 31 from the first review, with a total of 8,135 participants. Twenty-two trials included persons with mild or moderate psychological problems, 47 targeted people with various somatic conditions and 32 of the studies recruited people from the general population. Seventy-two studies compared MBSR to a WL/TAU control group, while 37 compared MBSR to an active control intervention. Seven studies compared MBSR to both a WL/TAU condition and to an active control group. Ninety-six studies contributed to the meta-analyses (based on information from 7,647 participants). Two thirds of the included studies showed a considerable risk of bias, and risk of bias was higher among studies with inactive than active control groups. Post-intervention Hedges’ g effect sizes for MBSR versus WL/TAU for the outcome measures of mental health, somatic health, and quality of life including social function were, respectively, 0.54 (95% CI 0.44, 0.63), 0.39 (95% CI 0.24, 0.54), and 0.44 (95% CI 0.31, 0.56). Some funnel-plot asymmetry points to a small degree of underreporting of negative trials. Heterogeneity was moderate for mental health and quality of life, and high for somatic health. Assuming a favourable outcome for 50% of the control group, the main finding of an effect size of 0.54 for improving mental health corresponds to a 65% chance that a random person from the treatment group will have a higher score than a person picked at random from the control group (probability of superiority). Another way of putting it, is that in order to have one more favourable mental health outcome in the treatment group compared to the control group at end of intervention, five people need to be treated (NNT=4.9, 95% CI 4.2, 5.9). Thus, if 100 people go through the treatment, 21 more people will have a favourable outcome compared to if they had been put on a wait-list or gotten the usual treatment. For 21 studies with follow-up data, the effect size was generally maintained at follow-up (1–32 months). For the comparison of MBSR versus alternative psychosocial interventions at post-intervention there was a small, statistically significant difference in favour of MBSR improving mental health with a Hedges’ g effect of 0.18 (95% CI 0.05, 0.30), and MBSR was not more effective than other active interventions on outcome measures of somatic health, 0.13 (95% CI -0.08, 0.34) and quality of life (including social function), 0.17 (95% CI -0.02, 0.35). Heterogeneity was low for mental health, moderate for quality of life and high for somatic health, and there was no funnel-plot asymmetry. Assuming a favourable outcome for 50% of the control group, the main finding of an effect size of 0.18 for improving mental health corresponds to a 57% chance that a random person from the treatment group will have a higher score than a person picked at random from the control group and the NNT=14, 95% CI 8, 50). Since the measure of mental health includes outcomes from a larger proportion of the included studies compared to somatic health or quality of life, it is a more robust measure for the effect of the MBSR intervention. It is therefore treated as the main primary outcome for the meta-analyses. For all comparisons effect sizes were fairly similar across the range of target groups and the effects were generally maintained at follow-up (1–34 months). Effect sizes for measures of mental health were not particularly influenced by length of intervention, attendance or self-reported practice, but they were strongly correlated to the effects on measures of mindfulness, indicating that the effects of the MBSR intervention may be related to the increase in self-reported mindfulness. Sensitivity analyses with exclusion of studies with exceptional findings did not substantially change the results. A majority of studies suffered from risk of bias, and studies of higher quality reported lower effects than studies with low quality. We found no reports of side-effects or costs in any of the trials. The overall quality of the evidence was moderate, indicating moderate confidence in the reported effect sizes. However, further research could impact on our confidence in the estimate of effect and may change the estimate. Authors’ conclusions MBSR has moderate effect on mental health across a number of outcome measures, for a range of target groups and in a variety of settings, compared to a WL or TAU control group. NNT was 4.9 (95% CI 4.2, 5.9) post-intervention; on par with other well-established interventions in the health service. The effect on somatic health is smaller, but still statistically significant. MBSR also seems to improve measures of quality of life and social function when compared to inactive control groups. MBSR improved mental health compared to other active psychosocial interventions, with a NNT = 14 (95% CI 8, 50), and had a similar effect on improving somatic health, and quality of life and social function. For all comparisons, the effects were maintained at follow-up and correlated to effects on mindfulness. The quality of the evidence was moderate and should be improved in future studies. There were many studies with considerable bias, and heterogeneity was mostly moderate. In addition, there is indication of underreporting of negative studies when MBSR was compared to inactive controls. These factors might have influenced the results found. MBSR might be an attractive option to improve health, handle stress, and cope with the strains of life. Ways to further strengthen the effect should be sought. All new trials should include measures of mindfulness and explore moderators and mediators of effects. New studies should register study protocols and adhere to guidelines for reporting of randomized controlled trials.