Background
Method
Sourcing Information
Selection criteria
Reason for exclusion | Number | Studies |
---|---|---|
Population not GPs | 6 | Bolton 2001; Hankir 2014; O’Reilly 2007; Ospina-Kammer(from Krasner); Rahe 2002; Rowe 1999 |
No intervention | 11 | Bluestein 2011 - Commentary
Firth-Cozens 2001 - Proposals for interventions
Gardner 2005 - Survey
Gutkin 2003 - Commentary
Hansen 2013 - Qualitative investigation of strategies
Hickner 2014 - Commentary
Latha 2004 - Overview of clinical environment
MacLean 2009 - Commentary
Sim 1996 - Commentary
Sim 1997 - Systematic review
Taub 2006 - Ethical guidelines
|
Uncontrolled before and after study | 7 | Dunn 2007 (from Gardiner [34]); Fortney 2013; Gardiner 06; Krasner 2009; Manocha 2009; Margalit 2005; Winefield 1998. |
Cohort study | 4 | Place 2013; Ro 2007; Ro 2010; Ro 2012 |
Qualitative evaluation | 1 | Schneider 2014 |
TOTAL | 29 |
Data extraction
Quality assessment
Data synthesis
Results
Author/Year | Country | Population | Intervention | Comparator | Study design | Outcome measures/Timepoints | Numbers analysed | Results |
---|---|---|---|---|---|---|---|---|
Gardiner et al. [34] | Australia | 86 GPs elected to attend a cognitive behavioural management course for which they gained Continued Professional Development (CPD) points IG 50-59 years 32.9 %
CG 50-59 years 42.9 %
Setting – metropolitan area Adelaide | 15 h over 5 weeks. Cognitive behavioural management. Aims 1.To improve psychological well-being through stress reduction 2. Have a beneficial effect on coping styles 3. Improve morale through problem-focussed coping | 24 GPs attending similar length CPD courses. Reported as being slightly older. | Controlled before and after study | Work-related distress (WRD -7 items Max score 49 = high distress)) Work-related morale (WRM – 7 items Max score 49 = high morale. Poor = 29) Quality of working life (QoWL – Rate statements 1-7. Max score 42. ‘Poor’ = 22) GHQ-12 (12 considered above threshold for high distress) Coping with work events (CwWE) Included to assess role of coping styles in stress outcomes Outcomes collected at T1 = Pre-intervention on first night of course T2 = Post-intervention on last night of course T3 = 12 weeks after first session |
Intervention group (IG)
T1 = 86 T2 = 77 T3 = 62
Control group(CG) T1 = 24 T2 = 19
Exclusions 0
Withdrawals 0
Lost to follow-up
IG T2 = 9 T3 = 24
CG T2 = 5 | WRD- Higher = more distress
IG v CG pre v post-intervention
ANOVA F = 2.99 p = 0.09 T1-T3 in IG F = 9.8 p = 0.000 WRM Rating morale as ‘poor’
IG v CG pre v post-intervention
ANOVA F =2.1 p = 0.15 T1-T3 in IG F = 12.6 p = 0.000 QoWL Rating quality as ‘poor’
IG v CG pre v post-intervention
ANOVA F =2.0 p = 0.16 T1-T3 in IG……F = 14.0 p = 0.000 GHQ
IG v CG pre v post-intervention
F = 11.9…p = 0.001 T1 = T3 in IG F = 28.2 p = 0.000 CwWE
IG v CG pre v post intervention
No significant difference |
Gardiner et al. [34] | Australia | 312 Rural GPs in reference group used to determine actual retention rate; 69 Rural GPs in intervention group who volunteered to attend a work/life balance retreat advertised by Rural Doctors’ Workforce Agency (RDWA) 205 Rural GP respondents to RDWA survey in control group Majority were male, 2/3 aged 30-50 years. | 9 h Work/life balance retreat. Format Group and individual CB coaching.
Pre-retreat – Drs’ issues, subjective stress ratings, validated stress questionnaire.
Post retreat
– letter to self at 4 weeks, e-mail follow-up and support for 5-6 weeks, Interview to assess goals at 10 weeks, validated stress questionnaire. Over 3 years 8 retreats were held. | Baseline data from RDWA survey reported in Gardiner 2005. 205/440 respondents to survey were used as the control group for the intervention. The entire population of rural GPs (312) was used to calculate actual retention rate at 42 months after the intervention | Controlled before and after study. | Rural Doctor Distress (RDD) (Customised 10 item scale graded 1-7 where 1 = not at all) Doctors’ Intention to leave rural practice (ITL) (Rated by IG on scale 1-7 where 1 = not at all) Actual retention rate of rural GPs (ARR) (Calculated by comparison of IG with de-identified data from RDWA database) Outcomes collected at T1 = Pre-intervention T2 = 3 months after intervention T3 = 42 months after intervention RDD reported T1 T2 ITL reported T1 T2 ARR reported T3 |
Intervention group (IG) n = 40
Control group n = 205 Of 69 GPs who volunteered to attend 48 completed post-intervention questionnaires T2 but only 40 were eligible due to inconsistencies in personal codes. Missing – Number analysed in intervention group at T1 not specified. CG Intention to leave at T2 = 10 IG Intention to leave at T2 = 3 | Rural Doctor Distress (RDD)
IG v CG at baseline Not significantly different p > 0.05
IG T1-T2 Significant t-test for all 10 items.
p = 0.05 Doctors’ intention to leave
IG T1-T2 Chi2
p = 0.023 % intending to leave practice
CG v IG T1 v IG T2 47.5 % v 81.1 % v 40 % Actual retention rate
IG v CG
93.9 % v 79.5 %
Chi
2
p = 0.027 |
Holt & Del Mar [36] | Australia | 819 GPs respondents to questionnaire 233 GPs eligible for inclusion in study as had GHQ-12 ≥ 3 Of the 819 questionnaire respondents 552 were male. No gender data on the 233 eligible for the intervention. Questionnaire sent to 1356 GPs from 8 divisions in 2 Australian states. | Aim Need for broader organizational approach to occupational stress Format Mailed intervention consisted of letter with feedback on GHQ score, interpretation of score and self-help sheet which addressed issues identified in baseline data. | Questionnaire respondents with GHQ-12 score ≥3 were divided into 2 groups.
Control group (CG) n = 113
Intervention group (IG) n = 120 | Controlled trial | GHQ-12 item Used to detect psychological distress and changes within the same population. Scores classified as 0-2 = none to mild psychological distress 3-7 = mild to moderate 8-12 = moderate to severe. Outcomes collected and reported at T1 = pre-intervention/baseline T2 = 3 months post-intervention |
Intervention group
n = 78
Control group
n = 83 Analysed both as intention to treat and then excluding the 26 GPs who attended a concurrent educational programme. Exclusions = 14 in the Intervention group who enrolled in a concurrent educational program that used similar material. Lost to follow-up
IG T2 = 42
CGT2 = 30 | GHQ-12 scores Analysed by intention to treat
IG pre and post-intervention
Change = 3.39
CG pre and post-intervention
Change = 2.25
IG v CG
Difference of means 1.14 (0.07,2.27) p = 0.05 Analysed after excluding data from 26 GPs Difference of means 1.44 (0.18, 2.7) p = 0.03 Results for GPs attending the educational program showed that 62 % scored ≤2 on GHQ-12 |
Martin Asuero et al. [37] | Spain | 68 Primary care professionals elected to attend a mindfulness education programme. Mean age 47 92 % women 60 % doctors 33.3 % nurses 6.7 % social workers and clinical psychologists. Intervention group1 n = 21 Intervention group2 n = 22 Authors report no significant baseline differences in intervention groups. Setting – Primary health care centres in Catalonia. | 28 h over 8 weeks. Mindfulness-based group psychoeducational activities. Aims To assess the effectiveness of a training programme designed to reduce burnout and mood disturbance, to increase empathy and to develop mindfulness. Format Intervention was delivered by the same trained instructor to both intervention groups Weekly sessions included educational presentations; formal mindfulness meditation; narrative and appreciative inquiry exercises and group discussion. There was an 8 h guided silent mindfulness session. Materials/Homework Participants paid $68 for packs containing a CD recording of exercises with an explanatory booklet. Home practice was expected. | 25 Primary care professionals who were offered the intervention after study completion. | Controlled clinical trial | Maslach Burnout Inventory (MBI) 22items 3 subscales. Higher scores on emotional exhaustion and depersonalization; lower scores on personal accomplishment indicate a higher degree of burnout. Possible scores 0-140 Profile of mood states (POMS) Short version- 15 items- 5 subscales: tension-anxiety, depression-dejection, anger-hostility, vigour-activity, fatigue-inertia. Higher scores indicate a worse mood state (except vigour) Scores range from 0-60 morale. Jefferson Scale of Physician Empathy (JSPE) 20 items 3 subscales. Higher scores on compassionate care, perspective taking and ‘standing in the patient’s shoes’ indicate higher degree of empathy Five facets mindfulness questionnaire (FFMQ). Observing, describing, acting with awareness, non-judging, non-reactivity rated on 5 point Likert where 1 = never/very rarely -5 = very often/always .39 items Evaluation questionnaire Translated into Spanish from University of Massachusetts Center for Mindfulness All measured at baseline T1and 8/52 post-intervention T2 |
Intervention group (IG)
T1 = 43 T2 = 43
Control group(CG) T1 = 25 T2 = 25
Exclusions 0
Withdrawals 0
Lost to follow-up
IG T2 = 0
CG T2 = 0 | MBI
SRM IG = 0.43 CG = -0.11
Mean between group difference
-7 (-13.4 to -0.6 (95 % CI)) p < 0.05
SES 0.74
POMS SRM IG = 0.62 CG = -0.1
Mean between group difference -0.71 (-11 to -3) p < 0.01 SES 1.15 JSPE SRM IG = 0.31 CG = -0.24
Mean between group difference
5.2 (0.2 to 10.3)
p < 0.05
SES 0.71
FFMQ
SRM IG = 0.65 CG = 0.1
Mean between group difference
SES 0.9
SRM = standardized response mean. Calculated as mean change in score divided by the standard deviation of the change. SES = standardized effect size. Calculated as mean difference between groups divided by standard deviation of the control group. Values > 0.8 = large changes |
Bias | Gardiner et al. [34] | Holt and Del Mar | Gardiner et al. [34] | Asuero et al. [37] |
---|---|---|---|---|
Random sequence generation (selection bias) | High – Allocation by preference of participant | Unclear –Insufficient information provided about sequence generation | High – Allocation by judgement of participant | Unclear – Insufficient information provided about sequence generation |
Allocation Concealment (selection bias) | High – Explicitly unconcealed procedure | Unclear – Insufficient information provided | High – Explicitly unconcealed procedure | Unclear-Insufficient information provided |
Blinding of participants and personnel (performance bias) | High – Blinding of participants and personnel was not possible | High – Blinding of participants and personnel was not possible | High – Blinding of participants and personnel was not possible | High – Blinding of participants and personnel was not possible |
Blinding of outcome assessment (detection bias) | High – Self-reported outcomes | High – Self-reported outcomes | High – Self-report for Rural Doctor Distress and Intention to leave. | High Self-reported outcomes |
Incomplete outcome data (attrition bias) | Unclear – Insufficient reporting of attrition to justify ‘low’ risk | Low – Clear participant flow reported | Unclear – Insufficient reporting of attrition to justify ‘low’ risk | Unclear–Baseline table indicates there were dropouts in the intervention group. No details provided |
Selective reporting (reporting bias) | Low – The published report includes all expected outcomes | Low – All outcomes reported | Low – The published report includes all expected outcomes | Low-All outcomes reported. |
Other bias | Unclear – Insufficient information to assess whether other important risk of bias exits | High – Concurrent educational program effecting 26 participants. 14 in intervention group did not receive the intervention as a consequence. Control group contamination possible. | Unclear – Insufficient information to assess whether another important risk of bias exits |
Components | Gardiner et al [34] | Holt and Del Mar [36] | Gardiner et al [34] | Asuero et al. [37] |
---|---|---|---|---|
Selection Bias 1. Are the individuals selected to participate likely to be representative of the target populations? | Self-referred/elected therefore using dictionary definition this scores 3 = NOT LIKELY | Participants were those respondents to a questionnaire found to score above a threshold. Questionnaire sent to all GPs in 8 Divisions of General Practice in Australia. 2 = Somewhat likely | Self-referred therefore using dictionary definition this scores 3 = NOT LIKELY | Self-referred/elected to attend. Subsequent stratified randomization reported. 2 = Somewhat likely |
Selection Bias 2. What percentage of the selected individuals agreed to participate? | 1 = 80-100 %. By electing to attend participants were agreeing to participate. | Baseline questionnaire response rate 819/1356 = 60 % 60 % = 2 | 69 Volunteered to attend but cannot tell how many actually participated 5 = Can’t tell | 1 = 80-100 % All eligible volunteers agreed to participate. |
SELECTION BIAS RATING | WEAK | MODERATE | WEAK | MODERATE |
Study design | Controlled before and after study | Controlled clinical trial | Controlled before and after study | Controlled clinical trial |
Was the study described as randomized? | No | Yes | No | Yes |
Was the method of randomization described? | No | No | No | No |
Was the randomization process appropriate? | Not applicable | No | Not applicable | No |
STUDY DESIGN RATING | MODERATE | MODERATE | MODERATE | MODERATE |
Were there important differences between groups prior to the intervention? | 1 = Yes Control group more likely to be in solo practice, older and had more years in practice | 3 = Can’t tell Authors report mean comparison of baseline GHQ scores showed no significant difference prior to the intervention (p = 0.09). No other information provided on pre-intervention confounders | 3 = Can’t tell. Control group for psychological well-being outcome were respondents to a survey. Control group for actual retention were entire population of rural GPs . | 3 = Can’t tell Authors report that intervention group was larger due to high interest in the intervention. |
What percentage of relevant confounders were controlled? | Can’t tell = 4 Controlling for confounders not explicit. | Can’t tell = 4 | Can’t tell = 4 | Can’t tell = 4 |
CONFOUNDERS RATING | WEAK | WEAK | WEAK | WEAK |
Were the outcome assessors aware of the intervention status of participants? | Yes = 1 | Yes = 1 | Yes = 1 | Yes = 1 |
Were the participants aware of the research question? | Yes = 1 | Yes = 1 | Yes = 1 | Yes = 1 |
BLINDING RATING | WEAK | WEAK | WEAK | WEAK |
Were data collection tools shown to be valid? | Yes = 1 | Yes = 1 | Yes = 1 | Yes = 1 |
Were data collections tools shown to be reliable? | Yes = 1 | Yes = 1 | Yes = 1 | Yes = 1 |
DATA COLLECTION RATING | STRONG | STRONG | STRONG | STRONG |
Were withdrawals and drop-outs reported in terms of numbers/reasons? | Yes = 1 | Yes = 1 | No = 2 69 volunteers, 48 questionnaires completed post-intervention. No information on those 21 given. | No = 2 Drop-outs from intervention group mentioned in baseline table. No details provided however results in scales approximate in remainder of tables. |
Percentage of participants completing the study | 84 % = 1 89 % IG 79 % CG | 161/233 = 69 % = 2 | 57 % = 3 63 % IG 51 % CG | 100 % = 1 |
WITHDRAWALS AND DROP OUTS RATING | STRONG | MODERATE | WEAK | STRONG |
What percentage of participants received the allocated intervention? | Follow-up data for 77. Cannot tell if all 86 received the intervention. | 106/120 = 88 % Score = 1 | 48/68 = 60 % Score = 2 | 100 % Score = 1 |
Was the consistency of the intervention measured? | Not explicitly Cannot tell = 3 | Not explicitly Cannot tell = 3 | Not explicitly Cannot tell = 3 | Described as ‘essentially the same’ and delivered by the same qualified instructor. No explicit report of measurement of consistency. Cannot tell = 3 |
Is it likely that subjects received an unintended intervention that may influence results? | No = 5 | Yes = 4 Concurrent educational programme which 26 of the study participants attended. Analyses were made with and without them. | No = 5 | No = 5 |
Unit of allocation | Individual | Individual | Individual | Individual |
Unit of analysis | Individual | Individual | Individual | Individual |
Are the statistical methods appropriate for the study design? | Yes = 1 | Yes = 1 | Yes = 1 | Yes = 1 |
Is the analysis performed by intervention allocation status (ITT) rather than actual intervention received? | No = 2 | Yes = 1 | No = 2 | No = 1 |
Component | Gardiner et al [34] | Holt and Del Mar [36] | Gardiner et al [34] | Asuero et al. [37] |
---|---|---|---|---|
Selection Bias | Weak | Moderate | Weak | Moderate |
Study Design | Moderate | Moderate | Moderate | Moderate |
Confounders | Weak | Weak | Weak | Weak |
Blinding | Weak | Weak | Weak | Weak |
Data Collection Methods | Strong | Strong | Strong | Strong |
Withdrawals and Dropouts | Strong | Moderate | Weak | Strong |
GLOBAL RATING | WEAK | WEAK | WEAK | WEAK |