A systematic review of the impact of mindfulness on the wellbeing of healthcare professionals

OBJECTIVE
Among efforts to improve the well-being of healthcare professionals are initiatives based around mindfulness meditation. To understand the value of such initiatives, we conducted a systematic review of empirical studies pertaining to mindfulness in healthcare professionals.


METHOD
Databases were reviewed from the start of records to January 2016. Eligibility criteria included empirical analyses of mindfulness and well-being outcomes acquired in relation to practice. 81 papers met the eligibility criteria, comprising a total of 3,805 participants. Studies were principally examined for outcomes such as burnout, distress, anxiety, depression, and stress.


RESULTS
Mindfulness was generally associated with positive outcomes in relation to most measures (although results were more equivocal with respect to some outcomes, most notably burnout).


CONCLUSION
Overall, mindfulness does appear to improve the well-being of healthcare professionals. However, the quality of the studies was inconsistent, so further research is needed, particularly high-quality randomized controlled trials.


INTRODUCTION
Healthcare professionals (HCPs) can face particular challenges that can be detrimental to their physical and mental health. A wealth of research has accumulated indicating that HCPs are liable to experience a range of mental health issues, including anxiety (Gao et al., 2012), burnout (Khamisa, Oldenburg, Peltzer, & Ilic, 2015), depression (Givens & Tjia, 2002), and stress (Bidwal, Ip, Shah, & Serino, 2015). Moreover, these problems may be particularly acute among HCPs relative to people in other professions (Brooks, Gerada, & Chalder, 2011). A recent survey of over 3,700 public sector workers in the United Kingdom found that staff working for the National Health Service were the most stressed, we are able to stand back and witness it" (p. 377). This process, also known as "decentring," is defined as "the ability to observe one's thoughts and feelings as temporary, objective events in the mind, as opposed to reflections of the self that are necessarily true" (Fresco et al., 2007, p. 234). This ability is theorized as having a positive effect upon wellbeing. In MBIs, the aim is not to change participants' thoughts/feelings per se, as cognitive therapy might seek to, but to help people "become more aware of, and relate differently to" this content (Shapiro, Astin, Bishop, & Cordova, 2005, p. 165). For example, in Mindfulness-Based Cognitive Therapy (MBCT), designed to prevent depressive relapse, people are taught to decenter from their cognitions, thus helping prevent a "downward spiral" of negative thoughts and worsening negative affect, which could otherwise trigger relapse (Segal, Williams, & Teasdale, 2002).
Thus, MBCT, and MBIs generally, involve "retraining awareness" so that people have greater choice in how they relate and respond to their subjective experience, rather than habitually responding in maladaptive ways (Chambers, Gullone, & Allen, 2009, p. 659). For instance, the development of decentring can help people tolerate distressing qualia, which is important given that inability to tolerate such qualia is a transdiagnostic factor underlying diverse psychopathologies (Aldao, Nolen-Hoeksema, & Schweizer, 2010).
Mindfulness interventions were initially limited to clinical settings, such as Kabat-Zinn's (1982) MBSR program and subsequent adaptations like MBCT (Segal et al., 2002). However, since the late 1990s, there has been increasing use of mindfulness in occupational contexts, for not only staff who might be suffering with stress and mental health issues but also workers "in general" (e.g., as a protective measure against future issues). To assess the state of this literature with regard to HCPs, we conducted a systematic review of relevant research.
Although a number of reviews have already been conducted in this area, these have tended to have fairly narrow remits in terms of population and/or outcome. These include reviews focused only on certain healthcare professions, such as general practitioners (Murray, Murray, & Donnelly, 2016), social workers (Trowbridge & Lawson, 2016), and nurses (Botha, Gwin, & Purpora, 2015), all of which featured small numbers of studies. Or, such reviews have concentrated on HCPs more generally but were concerned only with specific outcomes, such as stress, as in the case of Burton, Burgess, Dean, Koutsopoulou, and Hugh-Jones (2017), who included only nine studies, or empathy and emotional competencies, as in the case of Lamothe, Rondeau, Malboeuf-Hurtubise, Duval, and Sultan (2016), who focused just on MBSR and identified 14 such studies. By contrast, the current paper aims for greater inclusivity, reporting the results of a far broader systematic review, encompassing: (a) workers across all HCP contexts, (b) a wide range of wellbeing outcomes, and (c) the impact of mindfulness generally (not limited to any one intervention).

METHOD
The literature search was conducted using the MEDLINE and Scopus electronic databases. The search was conducted as part of a broader ongoing systematic review into mindfulness in all occupations. The criteria for the broader review were as follows: mindfulness AND work OR occupation OR profession OR staff, in all fields in MEDLINE, and limited to article title, abstract, and keywords in Scopus. The dates selected were from the start of the database records to January 10, 2016.
For this current review into HCPs, in terms of participants, interventions, comparisons, outcomes, and study design, the key inclusion criteria were as follows: participants-currently employed in a healthcare context; outcomes-any pertaining to mindfulness, well-being, and job performance; and study design-any empirical study featuring data collection. Exclusion criteria were theoretical articles or commentaries without statistical or qualitative analyses.
Although we were principally interested in studies of MBIs in healthcare workplaces, as a secondary concern we were also interested in nonintervention studies, such as regression analyses of the association between trait mindfulness and well-being outcomes. Studies were required to be published, or in press, in English in a peer-reviewed academic journal. The review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Moher, Liberati, Tetzlaff, & Altman, 2009). The review protocol for the broader systematic review was registered with the International Prospective Register of Systematic Reviews database on January 5, 2016 (registration number: CRD42016032899). Papers were divided into intervention studies and nonintervention studies. For intervention studies, the following variables were extracted from each paper: type of design (e.g., randomized controlled trial [RCT] vs. convenience sample); occupation of participants; number of experimental participants; number of control participants (if applicable); type of MBI; length of MBI; nature of control; principal well-being and performance outcomes; and the significance level and effect size of principal outcomes. For nonintervention studies, the following variables were extracted: type of analysis; occupation of participants; number of experimental participants; principal well-being and performance outcomes; and the significance level of principal outcomes.
The primary summary measures were mindfulness and well-being outcomes. These were principally psychometric scales pertaining to mindfulness, mental health, and physical health. Secondary summary measures of interest were outcomes that pertain to well-being, such as compassion and empathy. Tertiary summary measures of interest were outcomes relating to job performance.
The Quality Assessment Tool for Quantitative Studies (QATQS; National Collaborating Centre for Methods and Tools, 2008) was used to assess the quality of the studies. QATQS assesses methodological rigor in six areas: (a) selection bias, (b) design, (c) confounders, (d) blinding, (e) data collection method, and (f) withdrawals and dropouts. Each area is assessed on a quality score of 1 to 3: 1 = strong; 2 = moderate; 3 = weak. Scores for each area were collated, and a global score assigned to each study. If there are no weak ratings, then the study overall is scored 1; one weak rating leads to a 2; and two or more weak ratings generates a 3. QATQS scoring was conducted by the third author, and checked independently by the first author. Any discrepancies were resolved by discussion with agreement reached in all cases.

RESULTS
For the broader systematic review (i.e., mindfulness across all occupations), after removing duplicate citations, 721 potentially relevant papers were identified. In the current systematic review, focusing specifically on HCPs, from reviewing the abstract, 543 papers were excluded. From the full-text reviews of 178 papers, 97 further papers were excluded. Thus, a total of 81 papers were included in the systematic analysis: 66 intervention studies and 15 nonintervention studies. Two of these papers pertained to the same trial (Cohen-Katz, Wiley, Capuano, Baker, Kimmel, et al., 2005;Cohen-Katz, Wiley, Capuano, Baker, Deitrick, et al., 2005), and so the 81 papers included in the analysis represented results from 80 independent participant samples.
The studies comprised a total of 3,805 participants, discounting participants not included in analyses because of attrition. There were 2,645 participants in the intervention studies, as shown in Table 1, including 1,869 undertaking MBIs. There were 663 separate control participants, excluding Singh, Singh, Sabaawi, Myers, and Wahler (2006), in which participants acted their own controls, plus Grepmair, Mitterlehner, Loew, and Nickel (2007), in which participants were not HCPs per se but patients being treated by them. There were 1,160 participants in nonintervention studies, as detailed in Table 2. The studies covered a range of occupations, including physicians (n = 9), nurses (n = 16), disability professionals (n = 4), therapists, psychologists and counsellors (n = 24), mixed (nonspecific) mental health professionals (n = 8), and mixed (nonspecific) healthcare professionals (n = 20).
An overview of the findings is shown in Table 3 below. This shows whether outcomes were (a) increased in relation to an MBI, (b) did not change in relation to an MBI (or in exceptional cases, changed in a "negative" direction), or (c) were found in nonintervention studies to be associated with mindfulness. A more detailed breakdown of the results is included below in the discussion, featuring tables detailing all the studies that assessed a given outcome.

DISCUSSION
MBIs generally had a positive effect upon all outcome measures. However, there were some areas in which findings were more equivocal, including burnout, health, resilience, and generic "well-being." This discussion will run through the main outcomes in turn, beginning with mindfulness and awareness itself. Note. All statistically significant results are reported. Effect sizes were calculated when means and standard deviations were available; otherwise, just statistically significant differences are offered.  Note. a Studies showing worsening outcomes in relation to mindfulness. In instances where the total number of studies does not appear to be an accurate product of the other three columns (e.g., in the case of anxiety), this is because some studies used multiple measures with respect to a given outcome, and observed both a significant impact and no significant change.

Mindfulness and awareness
MBIs certainly appear effective at engendering mindfulness, with a small-to-medium effect size (d = .36), as assessed by 33 intervention studies, shown in Table 4  Two Factor Mindfulness Scale (Krasner et al., 2009) Note. Authors in bold denote RCT studies.
variation shows the value of drilling down into the fine-grained details of studies. Furthermore, it highlights the notion that-so far as multidimensional scales are concerned-mindfulness is not a monolithic construct, rather it comprises nuances, upon which there may be differential rates of change and development.

Anxiety
Turning now to the various well-being outcomes, first, on balance, mindfulness appears to have a beneficial impact upon anxiety, as shown in Table 5 below, with a medium effect size (d = −.51). While nine studies reported an improvement in relation to an MBI, six observed no change, although one further study (Rimes & Wingrove, 2011) reported worsening levels of anxiety.
In addition, of the nonintervention studies, Westphal et al. (2015) reported an inverse correlation between anxiety and mindfulness. Given the high prevalence and burden of anxiety among healthcare professionals (e.g., a survey of Chinese nurses found the prevalence of clinically significant anxiety symptoms to be as high as 43.4%), the improvements in anxiety linked to MBIs are noteworthy, modest though they are. As with mindfulness, a range of scales were Note. Authors in bold denote RCT studies; ! in third column = poorer outcome in relation to mindfulness.
deployed. The most prominent were Spielberger, Gorsuch, and Lushene's (1970) State-Trait Anxiety Inventory, and Lovibond and Lovibond's (1995) Depression Anxiety Stress Scale (DASS). The multidimensional DASS is particularly useful because it also covers depression and stress; therefore, it enables more ground to be covered with the one scale, thus reducing the empirical demands placed on participants.

Burnout
Regarding burnout, the results were more equivocal, as shown in Table 6 below. Of the 22 intervention studies examining this, only 11 registered a significant improvement, while equally 11 reported no significant change. Nevertheless, the overall effect size in this outcome was small to medium (d = −.33).
In addition, three nonintervention studies observed an inverse correlation between burnout and mindfulness. One possible explanation for the relatively equivocal results with respect to the MBIs may lie in the relatively small sample sizes of many studies. Some intervention studies that did not find a significant improvement in burnout certainly observed trends in the predicted direction (e.g., Mealer et al., 2014;Poulin, Mackenzie, Soloway, & Karayolas, 2008;Raab, Sogge, Parker, & Flament, 2015;Shapiro et al., 2005), although De Vibe et al. (2013 found trends in the other direction. Larger sample sizes may allow any effect of MBIs on burnout to be clearer. Another possible explanation is the multifaceted nature of the construct. The dominant measure used was the Maslach Burnout Inventory (Maslach, Jackson, & Leiter, 1986), which has three dimensions: emotional exhaustion, cynicism/depersonalization, and professional efficacy/accomplishment. When considering the components separately, several studies found that MBIs tended to have a stronger positive effect, albeit still nonsignificant, on emotional exhaustion compared to the other two components (e.g., Barbosa et al., 2013;Moody et al., 2013;Poulin et al., 2008).

Depression
The results were generally favourable with respect to depression, as shown in Table 7  an inverse correlation between depression and mindfulness. The relatively favorable results for this outcome are welcome, given the relatively high incidence of depression in HCPs. For instance, a study by Caplan (1994)  There are many hypothesized reasons for greater liability to depression among HCPs, including personality traits like perfectionism, burdens of clinical responsibility, and reluctance to seek treatment (Bright & Krahn, 2011). Whatever the reasons, it is encouraging that, on balance, MBIs appear to help in this regard-although it bears repeating that over one third of intervention studies reported no significant change-reflecting the more established efficacy of MBIs such as MBCT with respect to depression (Segal et al., 2002).

Stress and strain
More consistent results were found for stress, by far the outcome receiving the most attention, as shown in Table 8 below. Of the 37 intervention studies examining this, 25 registered a significant improvement in relation to an MBI, while 12 reported no significant change, although, in addition, Brooker et al. (2013) observed worsening levels. The global effect size for this outcome was small to medium (d = −.42).
Three nonintervention studies also observed an inverse correlation between stress and mindfulness. These generally positive results are again welcome: As with the other outcomes, stress is generally higher among HCPs than in the general population. Firth-Cozens (2003) reported that the proportion of HCPs being above threshold levels of stress is around 28% in surveys, compared with about 18% in the general working population. As with depression, a similar range of factors has been implicated in elevated stress levels among HCPs, from long working hours to the burden of clinical responsibility (Sochos, Bowers, & Kinman, 2012).
Unfortunately, as highlighted above, these burdens have only increased over recent years, due to factors such as curbs on healthcare spending, meaning that overwork has become even more acute. As noted above, in a survey of NHS staff, 61% reported feeling stress all or most of the time and 59% reported that the stress is worse this year than last year (Dudman et al., 2015). Thus, while it is encouraging that MBIs may help alleviate or prevent stress, it is of course vital that these underlying structural causes are also addressed.

Other well-being outcomes
This general pattern of mindfulness being associated with well-being was followed across the other outcomes. For example, 15 studies examined the relationship between mindfulness and distress or anger, and generally found it to have a positive effect, as shown in Table 9, with a total medium-to-large effect size (d = .60): 13 registered an improvement, whereas only two reported no change. Mindfulness was also associated with various more "positive" well-being outcomes, although the results overall were equivocal, as shown in Table 10, with an overall small-to-medium effect size (d = .36). Of the 21 intervention studies examining outcomes in this area, while 12 registered an improvement, 11 reported no change. (The nonadditive nature of the numbers in that last sentence reflects the fact that two studies used multiple well-being measures, and reported both significant and nonsignificant outcomes in relation to these.) In addition, two nonintervention studies observed a correlation with mindfulness, while McCracken and Yang (2008) actually observed an inverse correlation. Mindfulness also appeared conductive to health with a medium-to-large effect size (d = .62), although there were fewer studies focusing on such outcomes, as seen in Table 11 below. Of the 5 intervention studies examining this, three registered an improvement, while two reported no change; additionally, two nonintervention studies observed a correlation with mindfulness.
In addition to these primary well-being outcomes, mindfulness was also linked to various factors and qualities associated with well-being-including relationships, resilience, and emotional intelligence-which may provide an explanation for the generally positive outcomes adumbrated above. Regarding relationships, mindfulness practice seems to have a Note. Authors in bold denote RCT studies; ! in third column = poorer outcome in relation to mindfulness.
positive impact, as seen in Table 12, with a small-to-medium effect size (d = .46). Most of the 13 studies analyzing this outcome found either improvement or benign association with regard to mindfulness, while only two failed to provide significant results. Similarly, mindfulness was also linked to resilience, although the results were somewhat equivocal: As shown in Table 13, of the five intervention studies examining this, three observed an improvement and two reported no significant change. The overall effect size for this outcome was small (d = .21). Meanwhile, Kemper, Mo, & Khayat (2015) observed a correlation with mindfulness.
Mindfulness appeared to also affect emotional intelligence and regulation, as shown in Table 14. Of the 14 intervention studies examining this, 12 observed an improvement and only two reported no significant change. Nevertheless, Note. Authors in bold denote RCT studies.
this time no effect size was found (d = .18). In addition, seven nonintervention studies observed a correlation with mindfulness. The significance of this particular outcome is that, as outlined above, a key mechanism through which mindfulness is thought to exert its positive effects is reperceiving , also known as decentering (Fresco et al., 2007). This ability, which means that people are better able to detach themselves from distressing qualia that might otherwise precipitate feelings of stress etc., could be regarded as an aspect of a more general capacity of emotion regulation (Walsh & Shapiro, 2006).
The suggestion is that mindfulness might positively affect well-being in the following ways: (a) mindfulness involves introspective practices that facilitate the development of attention and awareness skills; (b) development of these skills leads to enhanced emotional regulation (including abilities such as reperceiving); and (c) emotional regulation is a metaskill that subserves multiple well-being outcomes (while, conversely, poor regulation skills are a transdiagnostic factor underlying diverse psychopathologies; Aldao et al., 2010). Future work may help to elucidate these hypothesized causal chains further (e.g., through longitudinal studies deploying regression analyses).
Finally, the effect of mindfulness was not limited to the well-being of HCPs, but was also associated with enhanced job performance. The dominant outcome in this respect was compassion and/or empathy, as shown in Table 15. Of the 28 intervention studies examining this, 16 observed an improvement and 9 reported no significant change, showing an overall small-to-medium effect size (d = 31); meanwhile, three nonintervention studies observed a correlation with mindfulness. Mindfulness was also associated with a broad range of other aspects of job performance, as shown in Table 16. Of the seven intervention studies examining outcomes in this area, six observed an improvement and only one found no change, with a large global effect size (d = .82). Six nonintervention studies also observed a correlation with mindfulness.

Summary and recommendations
Overall, MBIs had a positive impact upon most outcome measures, although some outcomes were rather equivocal, such as burnout. Moreover, a fairly large evidence base regarding the use of mindfulness in healthcare settings is gradually accumulating, with 81 papers included in the current review, comprising a total of 3,805 participants. Together, The preferred reporting items for systematic reviews and meta-analyses flow diagram work engagement, social capital, and creativity). Finally, where possible, trials should involve established MBIs, rather than bespoke adaptations, to better enable comparison across studies. However, there is also a need to move beyond MBIs developed for clinical contexts (e.g., MBSR) and to explore MBIs created specifically for the workplace.

CONCLUSION
Despite the issues with the current research base, the evidence of the value of mindfulness for HCPs is strong. Overall, mindfulness does appear to improve the wellbeing and job performance of HCPs on most metrics. Given the current pace of research into mindfulness, one might speculate that empirical support for the value of mindfulness in occupations such as healthcare will only strengthen over the years ahead.