Effects of mindfulness‐based interventions on fatigue and psychological wellbeing in women with cancer: A systematic review and meta‐analysis of randomised control trials

Abstract Background Cancer diagnosis and treatment can cause fatigue, stress and anxiety which can have a detrimental effect on patients, families and the wider community. Mindfulness‐based interventions appear to have positive effects on managing these cancer‐related symptoms. Objective To investigate the efficacy of mindfulness on cancer related fatigue (CRF) and psychological well‐being in female cancer patients. Methods Five databases (CINHAL, Ovid Medline, Ovid Psych Info, Scopus, and Cochrane), and two trial registers (WHO and Clinicaltrials.gov) were searched for randomised control trials from inception to April 2021 and updated in August 2022. Meta‐analysis was performed using Review Manager 5.4. The standardised mean difference (SMD) and 95% confidence intervals (CI) were used to determine the intervention effect. Subgroup analysis was performed for adaptation to types of mindfulness, length of intervention and types of comparator used. Results Twenty‐one studies with a total of 2326 participants were identified. Mindfulness significantly improved CRF (SMD −0.81, 95% CI −1.17 to −0.44), depression (SMD−0.74, 95% CI −1.08 to −0.39) and anxiety (SMD −0.92, 95% CI −1.50 to −0.33). No effect was observed for quality of life (SMD 0.32, 95% CI −0.13–0.87) and sleep (SMD −0.65, 95% CI −1.34–0.04). Subgroup analysis revealed that there was little difference in SMD for adapted type of mindfulness (p = 0.42), wait list control compared to active comparator (p = 0.05) or length of intervention (p = 0.29). Conclusion Mindfulness appears to be effective in reducing CRF and other cancer related symptoms in women. Adaptations to mindfulness delivery did not have negative impact on results which may aid delivery in the clinical settings.


| BACKGROUND
Worldwide there are an estimated 8.5 million women with cancer, while in the UK there are more than 182,000 new cases per year. 1,2 Symptoms associated with diagnosis and treatment of cancer include depression, anxiety, sleep deprivation and fatigue. Cancer-related fatigue (CRF) prevalence is higher in females and it is rated as the fourth most common unmet need among those living with and beyond gynaecology cancer. [3][4][5][6] The definition of CRF provided by the National Comprehensive Cancer Network illustrates its pervasiveness, as it is deemed to be a persistent feeling of tiredness that is physical and cognitive, it is not related to activity level and is not relieved by sleep. 7 CRF is multifaceted in nature and presentation, hence the interventions to manage this symptom need to be reflective of this. However, currently, the advice to manage CRF continues to be keeping active, eating a healthy diet and sleep hygiene. 8 However, research suggests that interventions such as exercise have mixed results for managing CRF, with some studies showing little or no effect. [9][10][11][12][13][14] Furthermore, it seems that exercise alone may not successfully target all aspects of CRF, for example, emotional or cognitive fatigue, may actually hinder participants' engagement in interventions such as exercise. [15][16][17] Diet and sleep hygiene although have evidence that may impact CRF, there remains a lack of studies that incorporate the role of diet in the direct management of CRF or sleep interventions that improve sleep substantially. 18 This lack of improvement in sleep can directly or indirectly effect CRF, this is known as a phenomena called 'cluster symptoms' where more than one symptom cluster together and effect each other positively or negatively, for CRF the symptoms identified include sleep, anxiety and depression. 19 The presence of these symptoms can lead to stress which may also lead to a further increase in the cluster symptoms. 20 The reduction of stress for cancer patients may be a factor in enhancing the management of CRF and interventions that can have the mind-body impact that may result in improvements in lifestyle include interventions like mindfulness.
Mindfulness has been described as the intention of being aware of the present moment without judgement. 21 And the research for cancer in this area has seen exponential growth over the last 10 years. However, the majority of this research has focused on reduction of depression, anxiety and stress symptoms associated with CRF. [22][23][24][25] Practising mindfulness has been shown to help emotional self-regulation, the development of positive coping mechanisms, and stress reduction, leading to improved quality of life (QoL) in women with breast cancer. 26,27 Mindfulness alone may be helpful in managing CRF or it can be the starting point that will allow individuals to access interventions that may further enhance the management of CRF.
Previous systematic reviews have evaluated the impact of mindfulness on psychological wellbeing, [28][29][30] showing a moderate effect on reducing anxiety and depression. However, some of these reviews were restricted in terms of the types of mindfulness for example, mindfulness-based stress reduction (MBSR) or mindfulnessbased cognitive therapy (MBCT), and others only included certain cancer population, for example, breast cancer. 28,29,31,32 The reviews that evaluated mindfulness in relation to CRF have also focused on specific types of mindfulness or have only included certain populations or types of cancer. 33,34 However, a recent review 34 reported the positive effects that mindfulness had on both CRF and vitality, the authors of this review assessed CRF as tiredness and exhaustion and vitality as energy and levels of function, each was evaluated separately. This review included any type of cancer, males and females, at any stage of cancer and any type of mindfulness. To date, no systematic review has evaluated research studies that included women with cancer, any type of mindfulness and its impact on CRF.
Hence, the aim of this review was to assess evidence for the impact of any type of mindfulness on CRF in women with cancer. The secondary aim was to consider the impact of mindfulness on psychological wellbeing, which was defined in the current review as depression, anxiety, and sleep, all of which are described as part of the symptom cluster of CRF.

| Search strategy
This review followed the preferred reporting of items of systematic reviews and meta-analysis (PRISMA) statement. 35 The protocol was registered in PROSPERO CRD42021240439. 36 (2) fatigue or tiredness or lethargy, and (3) mindfulness or meditation.
(Additional search strategy provided in supporting material).

| Inclusion and exclusion criteria
Studies were eligible for inclusion if they were randomised control trials (RCT) testing a mindfulness based intervention (e.g., MBSR, mindfulness-based cognitive therapy (MBCT), mindfulness-based art therapy (MBAT)), participants were female, over 18 with a diagnosis of cancer, fatigue was measured at baseline and one other timepoint post intervention, and studies were published in English. Comparison groups included treatment as usual/waitlist control or active treatments such as supportive care or education. Studies were excluded if mindfulness was not the main component (e.g., Acceptance and Commitment Therapy) and if reported as poster and or conference reports or abstracts.

| Study selection
Studies were initially screen by titles and abstracts and excluded if they did not focus on fatigue and women with cancer. Full text articles were retrieved and checked for eligibility, by a team of reviewers and any disagreements resolved by consensus. RefWorks bibliography software was used to export, manage and deduplicate search results, with additional hand removal of duplicates.

| Data extraction
Data were individually extracted (KMCC) and checked, by a team of reviewers, and any discrepancies were resolved through discussion.
Data extracted included: author, year of publication, country of origin, age, stage of cancer, type of cancer, treatment status, intervention arms, outcomes, measures, eligibility criteria, assessment timepoints, the results such as the mean and standard deviations (effect size) for CRF, anxiety, depression, sleep and QoL and adverse events. An Excel spreadsheet was designed to capture these data.

| Risk of bias assessment
The risk of bias was assessed for all 23 using the Cochrane risk-ofbias tool for randomised trials (RoB 2) 37 which includes 5 domains: randomisation, deviations from interventions, outcome measures, missing outcome data, reporting of results all of which are judged as low, some concerns and high with a summarised overall risk. This was completed by (KMCC) and 21% independently checked by other reviewers any discrepancies were discussed and consensus reached within the team.

| Quality of assessment
The Grading of Recommendations, Assessment, Development and Evaluation (GRADE) was performed to assess the quality of evidence of the included studies. This assessment goes beyond the risk of bias and includes 5 domains for assessment: risk of bias, inconsistency of the results, indirectness, imprecision, and publication bias possible ratings include high, moderate, low and very low. 38 GRADEpro was used to perform this assessment, and data for each outcome generated a 'Summary of Findings Table' with associated footnotes that explained any decisions on downgrading of the quality of evidence.

| Data analysis
Review Manager software (RevMan version 5.4) was used to conduct the meta-analysis, using a random-effects (inverse variance method) as heterogeneity in treatment effects was anticipated due to between-study variations in clinical factors (e.g., content of intervention). 39 Effects sizes were calculated as standardised mean differences (SMDs) with 95% confidence intervals (CIs) indicating the difference in means between groups divided by the pooled standard deviation (SD). Effect sizes were categorised by Cohen's classifications: SMD 0.2-0.5 small effect, SMD 0.5-0.8 medium effect, and SMD >0.8 large effect. 40 A negative SMD for CRF, depression, anxiety and sleep indicated a larger improvement in these outcomes due to the mindfulness intervention. For QoL a positive SMD indicated a larger reduction. Heterogeneity was evaluated using the I 2 statistics with values of 0%-40% representing might not be important, 30%-60% representing moderate, 50%-90% representing substantial and 75%-100% considerable heterogeneity. 41 Data was extracted for pre, post and first follow up for both intervention and control groups.
Effect sizes were calculated for pre to post-treatment and for pretreatment to last follow up.

| Addressing missing data
In those studies that did not provide all the required data, authors were contacted via email to request SDs and means. If it was possible to calculate this from other data provided such as standard error (SE), CIs, or t-values, then this was performed. Where it was not possible to access SDs and mean, these were excluded from the meta-analysis.

| Sensitivity analysis
Sensitivity analysis was carried out to ensure the robustness of the effect. For studies with a high risk of bias, all removal of these studies was performed to determine sensitivity to the results. If substantial or considerable heterogeneity was evident while performing metaanalysis, then sensitivity analysis was also performed to explore reasoning. This may be performed by removing outliers with effect sizes that are two standard deviations from the pooled effect size. 42

| Pooled effects at post-intervention
A forest plot for effect size of CRF is shown in Figure 2 17 studies were included in this analysis. As shown in Table 2  Sensitivity analysis was performed by removing outliers, which reduced the heterogeneity for CRF, anxiety and sleep from considerate to moderate or not being present, but did not appear to change the levels of heterogeneity for depression or QoL (See Table 2). The removal of outliers also reduced the pooled effect size from large to between moderate or small for all outcomes. Further sensitivity analysis was performed by removing studies with poor study quality which resulted in a reduction in heterogeneity from considerate to where it may not be considered important for CRF, anxiety and sleep but resulted in little change in heterogeneity for depression.
Although the pooled effect size was also reduced from large to moderate or small for all outcomes ( Table 2).

| Sub-group analysis
Sub group analysis was performed for type the of mindfulness, length of intervention and comparator used.
The effects of the mindfulness intervention were analysed as two groups, those studies that used MBSR as the mindfulness intervention compared with studies that used other types of mindfulness that  Table S1).
In terms of length of intervention, studies of 8 weeks duration in length were compared to those that were less than 8 weeks. One study was of 12 weeks duration as this was considered an outlier, it was omitted from the analysis. For CRF, anxiety, and depression, there appeared to be no significant difference between sub-groups with both showing a favourable response to mindfulness (SMD range −0.56 to −1.03). Mindfulness did not have a significant effect on QOL and sleep (Table S1 in supporting materials).
In the comparator analysis studies that used a WLC or UC were grouped together, and studies that used what was defined as AC were grouped together. For CRF, and anxiety this analysis indicated no significant difference between groups however, the pooled effect and overall effect showed a favourable response to the WLC group (CRF WLC; p = <0.0001 anxiety WLC; p = 0.03) with little or no effect for AC groups (CRF AC p = 0.13; anxiety AC p = 0.08) heterogeneity remained considerable for both these outcomes. Although there was a statistical significant between groups for depression the favourable response like the other outcomes, of CRF and anxiety, remained large for WLC (SMD −1.14) whereas AC showed little or no effect and demonstrated no heterogeneity within in this analysis (Table S1 in supporting materials). The subgroups for depression differed in the number of studies and number of participants suggesting that subgroup analysis may not be able to detect differences.
For QOL and sleep both groups did not show a favourable response to mindfulness.

| Pooled effects at follow up
Only (n = 12) studies provided data that could be included in an analysis Table 2     with 14 out of the 20 studies assessed as low. This was expected, as most of the studies accounted for missing data through intention-totreat analysis. Risk of bias from the randomisation process was also low (domain 1), as most studies provided adequate descriptions, however, many of the studies failed to adequately describe the allocation. The level of some concern was also high for reporting of studies, as very few trials were either pre-registered or had a predefined analysis plan (See Figure 3).

| Grading of recommendations, assessment, development and evaluation
Using GRADE the overall quality of evidence was rated as low or very low suggesting a low level of confidence in the effect estimate. The level of evidence for RCT was downgraded from high to low for sleep and anxiety and very low for fatigue, depression and QoL. This reduction from high to low or very low was due to serious concerns regarding levels of heterogeneity, risk of bias and publication bias.
Overall, no serious concern were found for inconsistency or indirectness (See Table S2 Summary of findings table in supporting materials).

| Publication bias
Publication bias was assessed using funnel plots for fatigue and, QoL, outcome measures that have fewer than 10 studies were not assessed as a minimum of 10 studies is required for a funnel plot to detect bias. 65 The funnel plots demonstrated asymmetry which would be suggestive of publication bias ( Figure S1 in supporting material).

| DISCUSSION
To our knowledge this is the first study to conduct a meta-analysis on the impact of mindfulness on fatigue for women with cancer. The results of this meta-analysis suggest that mindfulness led to a reduction in CRF post intervention and at follow up. For the secondary aim, mindfulness reduced anxiety and depression post intervention, with slight reductions at follow up, however no significant improvements were noted in sleep or QoL. The analysis showed that there was considerable heterogeneity between studies for all outcomes an attempt to account for this was performed through sensitivity analysis and removal of outliers although there was some reduction it continued to be considerable (Table S1 in  These conflicting results may have been due to differences in the types of mindfulness, types of cancer, outcomes and aims within these reviews. Reviews on mindfulness and CRF have to date focused on either a specific type of mindfulness (MBSR) or specific cancers, such as breast, additionally reviews that have included many types of mindfulness have included various types and stages of cancer. 32-34 To date no review has assessed whether any type of mindfulness has an effect on CRF for women with cancer, therefore, this review has contributed new knowledge in this area.
Psychological well-being, which included depression and anxiety, showed that mindfulness improved this symptom in this review.
Many of the included studies had the primary aim of psychological well-being. [43][44][45][46]48,[52][53][54][55]58,59,63,69  be because some studies failed to report these data, but may also be a result of differences in the length of follow-ups being, weeks or several months after the intervention.
Five studies made reference to their being no adverse events associated with their interventions, the remaining 75% of studies ignored this as part of their reporting. Even though the likelihood of adverse events for mindfulness is viewed as being low, this should be reported in all studies, as mindfulness can have a negative impact on people with anxiety disorders. 75 Most of the included studies in this review had exclusion criteria associated with mental health contraindications for participating in mindfulness. [43][44][45][46][48][49][50][51][52][53][54][55][56][57][58][59][60][61] However, this may suggest that recruitment into mindfulness studies to ameliorate CRF may not be representative of the general cancer population and this has implications for clinical practice as discussed below.

| Limitations
Among the strengths of this current review are its emphases on CRF and the inclusion of all types of mindfulness. This allowed for the Currently, nearly all the studies within this review were performed face to face, but as this review shows, adaptations to types and duration of mindfulness interventions did not have a detrimental effect on the overall effect, which may make it feasible to further adapt this type of intervention to a digital platform. Covid-19 has resulted in a trigger for the use of online services to manage health and has been received in the most part positively however, it still has its issues such as poor bandwidth and users technology skills. 76 Never the less the delivery of mindfulness interventions through these types of platforms would be worth further investigations as it would permit scalability and the ability to reach a larger population especially those in hard to reach rural area. 77 Alongside this, the need for more robust studies with active comparators such as exercise that focus on outcomes such as CRF are needed.
Characteristics of mindfulness interventions particularly the type and duration did not demonstrate within this review a difference between groups. Although there appeared to be a slightly larger effect for the non-MBSR and the less than 8 weeks in duration group all types of mindfulness demonstrated a favourable response to the outcomes of CRF, anxiety and depression. Cillessen et al., 2019, 72 also found that there was no difference in efficacy between types of mindfulness intervention and that all types were effective however, they did establish that there was a larger effect if adherence to the original protocol was maintained. These results may demonstrate that reducing the length of the intervention or adapting the type of mindfulness does not appear to have a detrimental effect on the overall effect and indeed may aid delivery in future clinical settings, as compliance may be easier with shorter interventions. This would also have financial implications for clinical settings and possibly reduce participant burden.
When assessing types of comparator, the analysis showed there were no differences between sub groups the comparison against WLC appears to show a greater favourable response in this review for all outcomes, which has also been identified in other reviews. This is not unexpected as studies that have a WLC or UC can result in a larger effect size in the intervention arms. 34 The small number of studies that used AC as the control arm in this review (n = 6, 32%) may have contributed to the findings. Other reviews also found similar findings where there were either too few studies to perform sub group analysis or the results were more favourable for the intervention groups compared to an AC group. 31,34 Furthermore, the type of AC may have an effect, as other reviews have suggested that if the comparator interventions are not developed for the management of particular symptoms, such as fatigue, then the results for these outcomes may not be reflective. 34 The use of specific interventions for control arms was demonstrated by Monti 2013 et al., in their 3 arm study with women with breast cancer and the effect of mindfulness or educational support on stress and QoL. They found that even though the MBAT maybe more beneficial for stress reduction and improved QoL than the support group or untreated group, the active control group was still beneficial and therefore still a worthy intervention in this population. Alongside this comparing or combining mindfulness with other interventions such as exercise 78 which is already supported by guidelines for managing fatigue, may provide evidence to enable health care professionals to make informed decisions with regard to interventions that could be offered to their patients.

| CONCLUSION
In conclusion, there appeared to be a large effect of mindfulness on fatigue, anxiety and depression however, there was little or no effect intervention that is well received with few adverse events and is adaptable, which will make it a transferrable and scalable intervention within the clinical setting.