Table 1 Summary of included intervention studies targeting stress or burnout among medical doctors

Arora et al. [40]

RCT with participants allocated to intervention (n = 10) or active control (n = 10)

N = 18 (20 began) novice surgeons/UK

• Stress (STAI, heart rate, and salivary cortisol)

• Mental imagery

5 × 30 min mental practice sessions, each undertaken before performing a surgical procedure on a VR simulator

• Sig lower stress in intervention group than control on average and maximum heart rate and cortisol during intervention measurement points. Sig lower anxiety on STAI in intervention group. However, differences were not significant between the groups by the post measurement time point.

• Sig higher mental imagery for intervention than control group

Not reported

Not reported

• ESs and CIs of ESs unable to be estimated

• Unclear whether mental training would translate to stress reductions outside of the VR task

• No intention to treat analysis conducted

Bar-Sela et al. [46]

Single group pre-post intervention design

N = 17 oncology residents/Israel

• Burnout (MBI)

• Expectations of contribution to improvement in abilities as doctors

9 × 1.5 h Balint group sessions, held monthly

• Means were divided by experience level (junior <3 years and senior >3 years). Across the three MBI scales, scores increased from beginning to end of year with the exception of depersonalisation among junior participants, which dropped. No statistical testing of changes

Not reported

• Sig increase from pre to post on residents’ ratings of contribution of intervention to abilities, although still only mid (3.1) on 5-point scale

• No control group

• Insufficient sample size

• No statistical testing of effects

• ESs and CIs of ESs unable to be estimated

Bragard et al. [45]

RCT with participants allocated to intervention (n = 49) or waitlist control (n = 47)

N = 96 (of 2160 invited) medical residents working in oncology/Belgium

• Communication self-efficacy

• Communication stress

• Self efficacy to manage stress

(all original questionnaires)

• Burnout (MBI)

40-h communication and stress management skills training programme, delivered in small groups (n = 7) over 8 weeks

• Relative to control group, intervention group reported sig less stress to communicate (d = .60, 95% CI = .21 to 1.04), sig greater self-efficacy to communicate (d = .88, 95% CI = .55 to 1.41) and self-efficacy to manage stress (d = .81, 95% CI = .45 to 1.3)

• No sig differences on burnout subscales of EE (d = .12, 95% CI = −.29 to .52), DP (d = .02, 95% CI = −.27 to .54), or PA (d = .14, 95% CI = −.38 to .43)

Not reported

Not reported

• Original items used for stress to communicate, self-efficacy to communicate, and self- efficacy to manage stress showed poor test retest reliability (rs = .49, .51, and .58, respectively). Changes observed in scores should be interpreted with caution.

• Average adherence to training programme was 50%

• Low response rate

• CIs of ESs estimated, not reported

Ghetti et al. [47]

Single group, pre-post intervention

N = 17 (of 36 invited) obstetrics and gynaecology residents/USA

• Burnout (MBI)

• Interest and ability in psychological medicine

• Physician empathy

Balint groups, conducted once per month, each of 1-h duration, over 1-year intervention period

• No sig changes in burnout on subscales or total MBI

• No sig changes in total interest and ability in psychological medicine, however at item level sig changes on 3 items (7, 8, and 9; relating to the doctor’s capacity to integrate psychological care into patient treatment)

• No sig changes on physician empathy scale

Not reported

Not reported

• No comparison group

• ESs and CIs of ESs unable to be estimated

• Less than half of residents participating in the Balint groups completed measures

• No data on average attendance in group

• Potentially insufficient power

Gunasingam et al. [26]

RCT with participants allocated to intervention (n = 13) or waitlist control (n = 18)

N = 31 (of 52 invited) first year doctors based in a single metropolitan hospital/Australia

• Burnout (MBI)

• Acceptability and satisfaction with intervention (original questionnaire and qualitative focus groups)

Four debriefing sessions held over 8 weeks, each of 1-h duration, led by experienced senior health professional

• No sig changes in burnout on subscales or total MBI

Not reported

Well received—60% would recommend to future doctors and 90% found sessions to be a source of emotional and social support

• Weak statistical power

• No data on attendance at sessions reported

• ESs and CIs of ESs unable to be estimated

• Passive control

Isaksson et al. [25]

Single group design, measures taken pre- and 1 year following intervention. Quasi comparison group to general sample of Norwegian doctors

N = 227 doctors (185 completed follow-up) self-referred for counselling between 2003 and 2005, compared with data from survey of 390 Norwegian doctors conducted in 2003/Norway

• Burnout (MBI)

• Mental distress

• Job stress

• Emotion focused coping and active coping

• Neuroticism

• Sick leave

• Satisfaction with intervention

Doctors chose one of two interventions: a single day (6–7 h) individual counselling session or 5-day group-based counselling programme aimed at motivating reflection on the doctors’ situation and personal needs.

No post- intervention outcomes reported, only 12-month follow-up

(At 1 year)

• Sig reductions in EE (d = .55, 95% CI = .33 to .77), DP (d = .28, 95% CI = .06 to .49), and PA (d = .08, 95% CI = −.13 to .29).

• Sig reductions in mental distress, job stress (d = .65, 95% CI = .44 to .87), emotion focused coping and neuroticism

• No sig change in active coping

• Sig fewer doctors on full time sick leave

Descriptive statistics not reported, however among male doctors satisfaction with the intervention independently predicted reduction in EE

• ESs and CIs of ESs not reported, but estimated from presented data

• No comparison condition

• Poorly defined treatment

• Effects of the two different counselling interventions not investigated separately

• Some participants participated in psychotherapy after the programme, effect not accounted for in analyses

• Reduction in work-hours confounds treatment effect and not controlled for

Isaksson et al. [50]

Single group design, measures taken 3 years following the intervention described in

N = 227 doctors (184 completed follow-up) self-referred for counselling between 2003 and 2005/Norway

• Burnout (MBI)

• Job stress

• Emotion focused coping and active coping

• Neuroticism

• Sick leave

Doctors chose one of two interventions: a single day (6–7 h) individual counselling session or 5-day group-based counselling programme aimed at motivating reflection on the doctors’ situation and personal needs.

No post-intervention outcomes reported

(At 3 years)

• No sig change in EE, job stress, emotion focused coping, and active coping from 1 to 3 years

• Sig reduction in neuroticism from 1 to 3 years

• From baseline (Isaksson et al. [25]) to 3-year follow-up, sig reductions in EE (d = .70, 95% CI = .48 to .93), and job stress (d = .77, 95% CI = .55 to .98)

Not reported

• ESs and CIs of ESs not reported, but estimated from presented data

• No comparison condition

• Poorly defined treatment

• Effects of the two different counselling interventions not investigated separately

• Reduction in work-hours confounds treatment effect and not controlled for

Kotb et al. [32]

Single group design, measures taken pre- and post- intervention

N = 33 family practice physicians/Egypt

• Burnout (MBI)

• Satisfaction with programme

Educational intervention consisting of 7 sessions, each of 60-min duration, focussing on CBT skills such as cognitive restructuring and relaxation

(At 6 months)

• No sig changes in MBI subscales of EE (d < .01, 95% CI = −.75 to .74), DP (d = −.59, 95% CI = −1.37 to .18), or PA (d = −.08, 95% CI = −.82 to .67)

Not reported

81% of participants reported they were satisfied (score ≥60%) with the intervention

• ESs and CIs of ESs not reported, but estimated from presented data

• No comparison condition

Krasner et al. [19]

Single group design, measures taken pre- (×2), post intensive period, post maintenance period, and 3 months following intervention

N = 70 (of 871 invited, 51 provided data at last assessment) primary care doctors in a continuing medical education course/USA

• Burnout (MBI)

• Mindfulness

• Physician Empathy

• Beliefs about psychosocial aspects of patient care

• Personality

• Mood states

Mindfulness, awareness, and communication training intervention, with 8-week intensive period (27 h total) and 10-month maintenance period (2.5-h session each month)

(Post maintenance)

• Sig improvements on EE and DP, but not PA subscales of MBI

• Sig improvements in total mood disturbance and physician beliefs

• Sig increases in physician empathy, mindfulness, and Big 5 personality Mini-Markers

(At 3 months)

• Sig reductions (from baseline) on EE (d = .62), DP (d = .45) and increase on PA (d = .44) subscales of MBI remained sig at follow-up

• Changes observed on total scales for secondary outcomes from pre- to-post remained sig at follow-up

Not reported

• No comparison condition

• Average attendance low (33.6 of possible 52 h) and not explored in relation to treatment phase or outcomes

• ESs and CIs of ESs not able to be calculated for post time point

Lemaire et al. [37]

RCT with participants allocated to intervention (n = 21) or waitlist control (n = 19). Control group received intervention during follow-up period.

N = 40 hospital based physicians/Canada

• Stress (purposely constructed questionnaire using items from previously validated stress measures)

• Adherence

• Heart rate

• Blood pressure

• Salivary cortisol

Biofeedback intervention delivered over 28 days, with 1 workshop (30 min), twice weekly meetings for intervention group, and practice 3 times per day for 5 min each.

• Sig reduction in stress (d = .44, 95% CI = −.19 to 1.07) for intervention group but not control group

• No sig changes in blood pressure (systolic d < .01, 95% CI = −.62 to .62, diastolic d = .21, 95% CI = −.41 to .83), heart rate (d = .18, 95% CI = −.44 to .80), or salivary cortisol (d = .07, 95% CI = −.69 to .55)

• 30% participants met criteria for “good” adherence

(At 4 weeks)

• Sig reduction in stress for control group (after exposed to unsupported version of intervention)

• Changes maintained for intervention group

• No sig changes on physiological measures

Not reported

• Measure of stress not validated

• Groups not equivalent on stress or heart rate at baseline

• ESs and CIs of ESs not reported, but estimated from presented data

Maher et al. [42]

Blinded, matched, quasi-experimental design, with participants allocated to intervention (n = 11) or passive control (n = 15) conditions

N = 26 surgical residents/USA

• Stress (measure by STAI, heart rate, and subjective stress scale)

• Surgical performance (blinded observer and self-assessed on OSATS)

• Acceptability of intervention

Stress management workshops (3 × 3 h duration, held weekly) focusing on identification of triggers and development of stress management techniques

• No significant effects on any measures, including between group differences on the STAI (d = .39) or heart rate variability (d = .56)

Not reported

• 91% of residents rated the intervention as valuable, 100% rated the intervention as good, very good, or excellent

• Non-random assignment of participants

• Small sample/low statistical power

• Use of ITT reported but final completion sample size not reported

• Unclear if groups equivalent at baseline on DVs

• ESs and CIs of ESs not reported, but estimated from presented data

Margalit et al. [36]

RCT with participants randomly allocated to didactic (n = 22) or interactive intervention (n = 22)

N = 44 (of 102 invited) general practitioners/Israel

• Knowledge

• Management Intentions

• Attitudes (including burnout assessed by a combination of items drawn from previously published measures of burnout)

N/A

Didactic or interactive (role play, Balint groups, individual teaching) instruction in bio-psychosocial approach to patient care. Workshops once per week for 4–6 h over 12-week period.

(At 6 months)

• Sig increase in burnout (d = .46, 95% CI = .04 to .88) and burnout guilt feelings (d = .48, 95% CI = .06 to .90) for both groups

• Sig increases in knowledge, intentions, patient centred attitudes, and self-esteem

• Sig greater improvements on self-esteem and intentions for interactive learning group

Not reported

Not reported

• Adherence/participation not measured

• Satisfaction not measured

• Low opt in

• Un-validated questionnaire

• Poor internal consistency for work strain and burnout associated with guilt feelings subscales

• CIs of ESs not reported, but estimated from presented data

McCue and Sachs [44]

Quasi-experimental trial, non-random allocation of doctors available to the intervention (n = 43) and doctors unavailable allocated to the control (n = 21)

N = 64 resident doctors in medicine, paediatrics, and medicine-paediatrics/USA

• Burnout (MBI)

• Stress (ESSI)

• Life experiences (only completed by intervention participants)

Stress management workshop of 4-h duration emphasising personal management, relationship, outlook, and stamina skills

(At 8 weeks)

• Sig reduction in stress

• No sig changes on MBI subscales

Not reported

Evaluations indicated satisfaction with intervention, although no formal quantitative or qualitative analysis reported.

• Non-random allocation limits findings

• ESs and CIs of ESs unable to be estimated

• Pre-intervention equivalency of groups not established

Milstein et al. [43]

RCT with participants allocated to intervention (n = 7) or waitlist control (n = 8)

N = 15 (of 33 invited) paediatric house officers/USA

• Burnout (MBI)

• Utility, effectiveness, and deterrents of BATHE technique (qualitative interviews)

Instruction (45 min) in use of BATHE psychotherapeutic tool, focusing on awareness and self-empathy. Encouraged practise of tool 3 times per week for next 3 months.

(At 3 months)

• No sig differences between groups or over time on MBI

Not reported

Not reported

• ESs and CIs of ESs unable to be estimated

• Insufficient statistical power

• Passive control condition

• Minimum statistical detail not reported in results (no group Ms or SDs)

Ospina-Kammerer and Figley [35]

Quasi-experimental trial, non-random allocation of doctors available to the intervention (n = 14) and doctors unavailable allocated to the control (n = 10)

N = 24 family practice residents/USA

• Burnout (EE subscale of MBI)

N/A

Relaxation/meditation training using Respiratory One Method (ROM). Participation in 4 workshops, held weekly, of 1-h duration.

• Sig lower EE in treatment group than control group at post treatment

Not reported

Not reported

• ESs and CIs of ESs unable to be estimated

• Non-random allocation limits findings

• Average attendance at workshops not reported

• Small sample size

Pflugeisen et al. [39]

Single group pre-post intervention design (with 8-week follow-up)

N = 19 (of 23 originally enrolled) community hospital physicians/USA

• Burnout (MBI)

• Stress (PSS)

• Mindfulness skills

• Participant use of intervention aspects

Mindfulness intervention delivered via 3 live sessions (90 min each), 8 online training videos (5–7 min each), and weekly teleconference coaching calls (1 h each) delivered over 8 weeks.

• Sig increase in PA (d = .57, 95% CI = −.07 to 1.20) and reduction in stress (d = .88, 95% CI = .23 to 1.54). No sig changes for EE (d = .47, 95% CI = −.16 to 1.10) or DP (d = .33, 95% CI = −.30 to .95)

• Sig increases in all mindfulness skills

(At 8 weeks)

• Sig decrease in EE (d = .69, 95% CI = .05 to 1.33), stress (d = 1.14, 95% CI = .47 to 1.81), and increase in PA (d = .62, 95% CI = −.02 to 1.25) compared to baseline

• Sig increases in all mindfulness skills except observing

Not reported

• Low participant adherence to reporting usage of intervention skills to researchers

• Participation/attendance not reported

• No control group

• CIs of ESs not reported, but estimated from presented data

Popenoe [33]

Quasi- experimental trial, with participants allocated to burnout intervention (n = 3) or Balint group (n = 3) by pre-existing work groups

N = 6 family practice residents/USA

• Burnout (MBI)

• Anxiety (STAI)

• Coping skills

• Satisfaction with intervention

Four sessions (once a week, of 1-h duration) focusing on understanding burnout and developing coping skills to manage burnout (burnout group) or focusing on interactions with patients (Balint group)

• Burnout increased for all Balint group members and decreased for burnout group members, but no tests of significance conducted

• No consistent differences between groups on anxiety or coping

Not reported

Stronger participant support for the value of the burnout group than for the Balint group

• No group level analyses of significance or individual level analyses of clinical change

• ESs and CIs of ESs unable to be estimated

• Small sample size

• No baseline checks of equivalence

• Non-random allocation limits findings

Shinefield [38]

Quasi-experimental trial, with non-random allocation of participants to intervention (n = 25) or waitlist control (n = 25)

N = 50 hospital based physicians/USA

• Burnout (MBI)

• Personality

• Coping strategies

• Intervention satisfaction

Six-week stress reduction training programme based on cognitive behavioural principles, with sessions held weekly for 2-h duration

• The treatment group reported sig improvement on burnout (MBI subscales of EE d = 1.06, 95% CI = .64 to 1.49, DP d = .59, 95% CI = .18 to 1.00, and PA d = .30, 95% CI = −.11 to .70) relative to the control group

• No sig interactions between personality type and intervention effects

Not reported

Overall satisfaction was high. Components most preferred by participants were relaxation and assertion, self-care, and exercise.

• Non-random allocation limits findings

• ESs and CIs of ESs not reported, but estimated from presented data

• Baseline differences between groups on number of patients seen per week, burnout, and job satisfaction not controlled.

• Passive control condition

Sood et al. [48]

RCT with participants allocated to intervention (n = 20) or waitlist control (n = 20)

N = 32 (of 40 enrolled) internal medicine physicians/USA

• Stress (PSS)

• Resilience (CD-RISC)

• Anxiety

• Fatigue

• Quality of life

Single, 90-min individual training in SMART resiliency and stress management programme, and training in a paced breathing meditation

(At 8 weeks)

• Sig increase in resilience (d = 1.16, 95% CI = .38 to 1.98), decrease in stress (d = 1.01, 95% CI = .22 to 1.79), decrease in anxiety (d = 1.32, 95% CI = .51 to 2.15), and increase in quality of life (d = .83, 95% CI = .06 to 1.61) for intervention group compared to control.

• No sig change for fatigue (d = .23, 95% CI = −.48 to 1.02)

Not reported

Not reported

• Differential attrition between control and intervention groups

• Passive control condition

• CIs of ESs not reported, but estimated from presented data

Sood et al. [27]

RCT with participants allocated to intervention (n = 13) or waitlist control (n = 13)

N = 26 radiology physicians/USA

• Stress (PSS)

• Resilience (CD-RISC)

• Anxiety

• Mindful awareness

• Quality of life

Single, 90-min individual training in SMART resiliency and stress management programme, with two follow-up phone calls and optional 30-min booster session

(At 12 weeks)

• Sig decreases in stress (d = .98, 95% CI = .12 to 1.83), anxiety (d = .86, 95% CI = .02 to 1.71), and increases in mindful attention awareness (d = 1.24, 95% CI = .36 to 2.13), and quality of life (d = .83, 95% CI = −.01 to 1.68) for intervention group compared to control

• No sig difference for resilience (d = .35, 95% CI = −.47 to 1.16)

Not reported

Not reported

• Incomplete data regarding attrition

• Apriori sample size not reached

• Passive control group

• CIs of ESs not reported, but estimated from presented data

West et al. [49]

RCT with participants allocated to intervention (n = 37) or active control (n = 37)

N = 74 internal medicine physicians/USA

• Burnout (MBI)

• Stress (PSS)

• Job satisfaction

• Quality of life

• Fatigue

• Physical and mental health

• Depression

• Physician empathy

Guided group discussions (1-h duration, held fortnightly over 9 months, 19 sessions in total) focussing on mindfulness, reflection, and shared experiences

• No sig differences on for total burnout (d = .03, 95% CI = -.44 to .50) or stress (d = .02, 95% CI = -.44 to .50)

(At 3 months)

• Sig increased engagement and reduced high item DP (on MBI) for intervention group compared to control d = .70, 95% CI = .23 to 1.19, with effects maintained at 12-month follow-up

Not reported

• Intervention attendance low (average of 11.7 of 19 sessions attended)

Wetzel et al. [41]

RCT with participants allocated to intervention (n = 8) or waitlist control (n = 8)

N = 16 surgical residents/UK

• Stress (heart rate, salivary cortisol, and observer rating)

• Anxiety (STAI)

• Applied surgical coping strategies

• Acceptability of intervention

Stress management training including relaxation, coping, and mental rehearsal strategies, duration not provided

• Sig lower coefficient of heart rate variability (d = 1.70, 95% CI = .55 to 2.84), higher observational team work in surgery (d = .71, 95% CI = −.30 to 1.72) & greater coping skills in intervention

• No sig differences on STAI (d = .18, 95% CI = .80 to 1.16) or cortisol (d = −.36, 95% CI = −1.35 to .63)

Not reported

Qualitative feedback indicated perceptions of improved skills and confidence as a result of the intervention

• ESs and CIs of ESs not reported, but estimated from presented data

• Passive control group

• Insufficient statistical power

Winefield et al. [34]

Single group design, measures taken pre- and post-intervention

N = 20 female general practitioners/Australia

• Burnout (MBI)

• Psychological distress (GHQ-12)

• Job satisfaction

• Job satisfaction

• Satisfaction with intervention

Three educational seminars, held fortnightly, each of 3-h duration, focusing on relaxation training, social support, managing self-expectations, and practice management

(At 4 weeks post intervention)

• Sig reductions in EE subscale of MBI (d = .37, 95% CI = −.24 to .98) and psychological distress (d = 1.13, 95% CI = .49 to 1.80)

• No sig differences for DP (d = .19, 95% CI = −.43 to .79) or PA (d = .25, 95% CI = −.36 to .86) subscales of MBI, or job satisfaction

Not reported

High satisfaction with programme, with group discussions rated as most helpful activity

• No comparison group

• ESs and CIs of ESs not reported, but estimated from presented data

• Some evidence of regression to the mean