Objectives Clinician well-being has been recognised as an important pillar of healthcare. However, research mainly addresses mitigating the negative aspects of stress or burnout, rather than enabling positive aspects. With the added strain of a pandemic, identifying how best to maintain and support the well-being, satisfaction and flourishing of general practitioners (GPs) is now more important than ever.
Design Systematic review.
Data sources We searched MEDLINE, PsycINFO, Embase, CINAHL and Scopus from 2000 to 2020.
Study selection Intervention studies with more than 50% GPs in the sample evaluating self-reported well-being, satisfaction and related positive outcomes were included. The Cochrane Risk of Bias 2 tool was applied.
Results We retrieved 14 792 records, 94 studies underwent full-text review. We included 19 studies in total. Six randomised controlled trials, three non-randomised, controlled trials, eight non-controlled studies of individual or organisational interventions with a total of 1141 participants. There were two quasi-experimental articles evaluating health system policy change. Quantitative and qualitative positive outcomes were extracted and analysed. Individual mindfulness interventions were the most common (k=9) with medium to large within-group (0.37–1.05) and between-group (0.5–1.5) effect sizes for mindfulness outcomes, and small-to-medium effect sizes for other positive outcomes including resilience, compassion and empathy. Studies assessing other intervention foci or other positive outcomes (including well-being, satisfaction) were of limited size and quality.
Conclusions There is remarkably little evidence on how to improve GPs well-being beyond using mindfulness interventions, particularly for interventions addressing organisational or system factors. This was further undermined by inconsistent reporting, and overall high risk of bias. We need to conduct research in this space with the same rigour with which we approach clinical intervention studies in patients.
PROSPERO registration number CRD42020164699.
Funding source Dr Diana Naehrig is funded through the Raymond Seidler PhD scholarship.
- primary care
- general medicine (see internal medicine)
- mental health
- organisational development
Data availability statement
All data relevant to the study are included in the article or as supplementary information. All included studies are published. We will consider sharing data upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Strengths and limitations of this study
While burnout has been a general focus of research, interventions to improve the well-being of general practitioners (primary care physicians) appear sparse.
The strength of this study is the extensive and systematic approach taken to evaluate interventions aimed at improving well-being, satisfaction, flourishing and other positive outcomes in general practitioners.
The systematic review was registered on PROSPERO a priori before commencing the data selection and extraction process.
English only articles were included.
The limitation of this systematic review is the dependency on the number of retrieved and included publications, and their quality of methodology and reporting.
Mental ill-health, burnout and stress among healthcare practitioners are a huge concern internationally with impacts on individual doctors and their families, patient care and the sustainability of the healthcare system.1 The well-being of clinicians has been recognised as crucial, and has been added to the more commonly shared health system goals of: improved care for individuals, better population health and reduced healthcare costs.2–5 Despite this stated aim, few studies have evaluated interventions to improve well-being, satisfaction and flourishing in general practitioners (GPs),6–12 typically taking a more traditional, problem-focussed approach, such as investigating causes and reduction of burnout and stress.
Medical doctors who provide primary care to patients are the backbone of healthcare provision. In Australia, the UK and Europe, typically the term ‘general practitioner’ (GP) is used, whereas North American articles generally refer to ‘primary care physicians’ or ‘family doctors’. Terms are used interchangeably.
Satisfaction and well-being in GPs are known to be associated with several factors. In the USA, primary care physicians are less satisfied with their job than specialists.13 14 A 10-year decline in job satisfaction for British GPs has been reported,15 and a Norwegian longitudinal study reported dwindling GP satisfaction over 7 years.16 In primary care, job satisfaction correlates with practice resources, an ordered atmosphere, a practice culture that enables communication and ease of providing quality care,13 17 18 and is inversely related to isolation and low sense of community.19 20
For well-being, Murray et al21 conducted a cross-sectional study exclusively exploring positive mental health and psychological resources (well-being, resilience, self-efficacy and hope) of GPs. GPs positive mental health was comparable to the general population and older and female GPs fared best, suggesting interventions for younger and male GPs may be most useful.
Systematic reviews and interventions aimed at improving satisfaction and well-being in GPs appear sparse. A European collaboration conducted a systematic review and qualitative study looking at positive determinants of satisfaction in GPs. They identified general work-related factors (ie, workload, responsibility, recognition and income) and GP specific factors (ie, competence, intellectual stimulus and work–life balance).22 23 However, there does not appear to be a systematic review looking at interventions to improve satisfaction exclusively in GPs.
A systematic review of interventions to improve the psychological well-being of GPs identified only four articles; two cognitive-behavioural, one mindfulness-based intervention and one self-help information approach.24 They applied a dual model of languishing/flourishing and the presence of mental illness/absence of mental illness matrix.25–28
Overall, little seems known about which interventions are efficacious in promoting GPs well-being and satisfaction. In contrast to more extensive research on burnout, distress and mental ill-health with a view to treat, avoid or mitigate negative outcomes, we explicitly aimed to apply a positive lens and focus on interventions that enhance GPs’ satisfaction and well-being, or promote environments and individual behaviours that may better enable well-being. We purposely included interventions on any level; directed at the individual (ie, training, workshops), the organisation (ie, work-flow improvements in the practice) and on a systemic level (ie, policy change). It is useful to bear in mind that GPs are typically high functioning individuals and are not a priori a clinical population, which is why we believe a positive framework is most preproperate.
Currently, reports indicate that COVID-19 places enormous additional strain on health professionals which impacts their physical, mental and social well-being.29 30 GPs as the first port of call may be particularly affected, while playing a crucial role in supporting population health.31–33 Efforts must be made to understand how GPs remain well, and if necessary, put measures in place to assist this.
We systematically reviewed studies exploring the effect of any type of intervention on the well-being, satisfaction and well-being of GPs. We broaden and expand on the existing literature, deliberately including any type of intervention, and a range of positive outcomes, and explore if there have been more recent intervention studies conducted in this field.
Data sources and search strategy
We followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines,34 and consulted a specialist librarian (JKH). MEDLINE, PsycINFO, Embase, CINAHL and Scopus were searched on 13 January 2020.
PICO (Population, Intervention, Comparison, Outcome) search terms included GPs and synonyms; interventions and implementations directed at the individual, the level of the organisation or practice and the healthcare system; outcomes included well-being, satisfaction, flourishing, synonyms and antonyms (search strategy, online supplemental file 1). Titles, abstracts, text, key terms and subject headings were searched for English publications. Eligible articles and related systematic reviews were hand-searched for further relevant references, and authors were asked to supply full-text articles where relevant conference abstracts only were available.
Study selection and data extraction
Due to the large number of records, sample screening of 1% of titles (n=107) was performed and discussed by two reviewers (DN and AS) together, and excellent agreement was reached. DN screened all titles. DN and AS independently and separately examined all abstracts and full-text records for inclusion using Covidence.35 Initial inter-rater reliability showed a proportionate agreement of 0.94 with Cohen’s Kappa of 0.68 for abstracts. For full-text screening, there was a proportionate agreement of 0.84 with Cohen’s Kappa 0.56, whereby both reviewers agreed to include 15/94 full-text articles and exclude 64/94 articles. Initial disagreements on 15 articles were resolved through joint discussion, or third reviewer adjudication (NG).
Studies with more than 50% GPs (family doctors, primary care physicians) working in a practice setting or medical centre, reporting on well-being, satisfaction, flourishing, mindfulness, resilience, empathy, engagement, balance, empowerment, compassion, work-related morale and control measures were included. We excluded studies exclusively reporting on burnout, distress, or mental ill-health.
Data including author, year, type of study, participants, intervention, preoutcome and postoutcome measures, and quantitative and qualitative results were extracted (table 1).
Data synthesis and analysis
We calculated within-group and between-group absolute change and effect sizes (Hedges’ g) (see tables 2 and 3, online supplemental files 1 and 2).36 We compared mean outcome scores and SDs at baseline with postintervention scores. Where several postintervention measures were reported, the primary outcome point nominated by the authors was selected. We utilised SD*, which takes different sample sizes into account (formulae in online supplemental file 2). For within-group, we calculated the pooled SD* based on preintervention and postintervention SDs, for between-group analysis, the effect size was calculated based on the pooled SD* of control and intervention groups at baseline37 (online supplemental file 2).
Positive effect sizes indicated an effect for the intervention. Effect sizes of 0.2, 0.5 and 0.8 were considered small, moderate and large, respectively.38
Risk of bias
Two reviewers (AS and DN) independently applied the Cochrane RoB239 to randomised controlled trials (RCTs). Total RoB2 scores showed 100% agreement. Any discrepancies of subscores were discussed, and consensus was achieved. The other studies were assessed by DN.
Patient and public involvement
No patients are involved.
The database searches rendered 14 792 records in total. After removing duplicates, studies conducted before the year 2000, and adding 12 studies through hand search—which included contacting authors for full text papers of relevant conference abstracts—10 759 studies were screened. We eliminated 9682 records by title, and 983 by abstract, leaving 94 studies for full-text assessment.
Study characteristics and design
We included 19 studies in the systematic review2 10 40–56 (tables 1 and 2, and PRISMA-Flowchart figure 1). Six RCTs, three non-RCTs (controlled before and after trial, CBA), eight non-controlled interventions (non-controlled before and after trial, NCBA) and two reports from a longitudinal cohort during which a health policy change was introduced, which we considered as ‘naturalistic’ interventions.41 56 Five studies included a qualitative component. Only one RCT50 and two CBAs44 47 utilised active controls. Five RCTs and one CBA40 had a waitlist control arm. Publications from the USA (8/19, 42%), Europe (8/19, 42%), Australia (2/19) and Israel (1/19) were included (table 1).
The total population enrolled was 1141 for the 17 intervention studies (average participants per study 67.1, range 6–290). The two studies reporting on the same panel survey population41 56 included approximately 2000 participants each year. Mean age overall ranged from 40 to 54.9 years, and sex from 8%–76.9% male participants (table 1). Attrition for intervention groups varied from 0% to 20%, for controls from 0% to 24%. One outlier had a total attrition rate of 80%.51 Eight studies reported follow-up measures, timepoints ranged from 3 to 48 months postbaseline (mean 14.6 months).
We found considerable variation in intervention type, length and dose-intensity. Three groups were distinguished based on the focus of the intervention: individual/personal (13/19, 68% of studies, n=930), organisational (4/19, 21%, n=211) and naturalistic interventions on a systemic level (policy change in the UK) (2/19, 11%).
Individual mindfulness-based interventions were most common (9/13, 69%), followed by educational training or experiential workshops (3/13, 23%) with one coaching intervention.46 Two organisational interventions trialled the addition of clerical support or scribes, and two explored an organisational improvement programme. Two studies from the UK examined the effects of the introduction of a pay for performance scheme41 56 (table 2).
Outcomes and their measures
The definitions of outcomes and measurement tools varied considerably. Only one study clearly stated one a priori primary outcome,51 with most using a battery of self-reported outcome measures (online supplemental file 4). These included a range of 12 validated tools (BAER, BRS, CD-RISC-10, FFMQ, GRAS, JSPE, JSS-WCW, MAAS, PANAS, SCS, UWES and WHO-5)57–75 as well as 13 measures where no validation information was obtainable. A range of job satisfaction measures were applied in eight studies, mindfulness in six, resilience in four, compassion and empathy tools were each used thrice, the positive and negative affect scale was used twice. The WHO-5 well-being index was used once (online supplemental file 4). Not one study evaluated flourishing. Negative outcome measures were often concurrently reported. Sixteen studies employed the Maslach Burnout Inventory or other stress-related measures. As the a priori aim of the study was to explore effects of interventions on well-being, satisfaction and other positive outcomes, we did not extract and report results for negative outcome measures, nor examine possible inter-relationships between positive and negative outcome measures.
The between-group change for controlled studies and within-group change for intervention arms are presented in tables 2 and 3, online supplemental file 3).
(a) Individual focussed interventions
(1) Mindfulness (k=9)
Six mindfulness interventions (3 RCTs, 1 CBA and 2 NCBA) evaluated mindfulness outcomes (FFMQ, MAAS and BAER) and reported moderate to high between-group effect sizes (k=4) ranging from 0.5 to 0.88 for mindfulness with an outlier at 1.5 (42). Within-group ES (k=6) showed moderate effect sizes (range 0.47–0.78) with one study outlier at 0.37 and one at 1.05 (tables 2 and 3).
Studies frequently utilised resilience, compassion and empathy measures with overall low-to moderate effect sizes. One RCT and two NCBAs measured resilience (BRS, RS-14 and CD-RISC), whereby between-group ES (k=1) was moderate at 0.61, while within-group (k=3) effect sizes were low to moderate (range 0.17–0.51). Compassion measures (SCBC and SCS) were reported in three studies (1 RCT and 2 NCBAs). Between-group ES (k=1) was 0.73, while within-group ES (k=3) varied considerably (−0.04 to 0.77). Three studies reported on empathy (JSPE) (1 RCT, 1 CBA and 1 NCBA) with very low 0.02, respectively moderate between-group 0.44 ES (k=2), and within-group ES ranging from 0.2 to 0.44 (k=3) (tables 2 and 3).
Two mindfulness studies (NCBAs) measured positive affect (PANAS), only one reported a within-group ES (0.52). One NCBA reported a within-group ES (0.52) for well-being (WHO-5), another NCBA reported an ES of 0.46 for self-reflection.
These effect sizes are generally supported by the results reported in the individual studies (table 1). Several interventions included repeated measures at later time-points, that is, during maintenance phase,48 showing an ongoing impact of mindfulness practice. Qualitative results suggested increased well-being and compassion towards self and others,40 respectively, improved awareness, acceptance, peacefulness and openness55 after the intervention.
(ii) Training, workshops and coaching (k=4)
For training, workshops and coaching interventions, we were only able to obtain data to calculate the ES of one RCT46 and one CBA.47 Low between-group effect sizes for work-related morale (0.3), quality of work-life (0.27) and low ES for both measures within-group (0.43 and 0.45, respectively) were found for Gardiner’s CBA.47 Very low effect sizes for job satisfaction and resilience both for between-group (0.06, 0.13) and within-group (0.13, 0.24) change were observed in Dyrbye’s RCT.46
These results are reflected in the individual study results (table 1). Barcons44 did not detect any significant between-group differences for overall job satisfaction, while Margalit50 demonstrated significant improvement in self-esteem between-groups.
(b) Organisational interventions (k=4)
One RCT and three NCBAs trialled organisational interventions. Means and SD were not provided; therefore, we were not able to calculate effect sizes. Linzer et al49 (RCT, n=166) demonstrated that workflow interventions, communication and overall quality improvements benefited satisfaction in the intervention group. While Dunn et al2 (NCBA, n=32) showed that quality improvement projects in the workplace showed significant improvement in quality work competence ratings but fluctuating satisfaction levels. Two smaller uncontrolled trials (n=13 in total) investigated the addition of clerical staff to the practice. Pozdnyakova et al52 showed that the addition of clerical staff led to an improvement in satisfaction with the clinic workflow from 2/6 to all 6 GPs in a single practice but did not report on any other measures of well-being. Similarly, Contratto et al45 reported improved quality of life and professional balance for seven general medicine physicians in a mixed-methods approach.
(c) Systemic interventions (k=2)
The introduction of a new contract with pay for performance scheme showed a significant improvement in job satisfaction56 with an effect size of 0.44 between 2004 and 2005. Allen et al41 used the same data and included a 2008 survey to look at satisfaction as a function of the exposure of GPs to the pay per performance scheme. While job satisfaction declined again in 2008, the exposure to the scheme did not affect satisfaction.
Risk of bias
The types of intervention and study settings precluded blinding for randomised controlled studies (no allocation concealment for waitlist control groups), and the outcome measures were participant reported throughout, and as such all studies were rated as high risk of bias by the Cochrane RoB2.
Strengths, limitations, and interpretation of evidence
We identified 19 studies, half of which were published in the last 4 years, demonstrating an increased interest in the improvement of well-being and satisfaction of GPs. In comparison, a systematic review from 2016 looking at interventions to reduce burnout in physicians included fifteen RCTs and 37 cohort studies with 20 studies conducted before 2010,76 suggesting that burnout has been a research focus for longer. Or this may possibly indicate that the focus is more generally shifting from a disease and ‘dis-abled’ to an ‘en-abled’ approach when trying to design interventions for healthcare professionals.
The considerable heterogeneity in the definition and measurement of constructs, study design, participant numbers, intervention types, intervention dose (ranging from 6 to 53 hours), follow-up periods, quality and reporting precluded a meta-analysis. It is challenging to draw robust conclusions regarding the (comparative) effectiveness of the different types of interventions reviewed.
Mindfulness interventions provided the most comprehensive and robust data with moderate to large effect sizes on mindfulness outcomes, and low-to-moderate effect sizes on compassion, resilience and empathy. Our results are supported by two reviews looking at the effects of mindfulness-based interventions on well-being,77 78 in healthcare professionals more generally. Lomas et al77 conducted a systematic review and meta-analysis and assessed 41 studies with approximately 2100 participants. They found a wide range of self-reported outcomes (with both positive and negative measures of well-being). Reported within-group effect sizes for mindfulness, positive well-being (or life satisfaction) and compassion hovered around a moderate 0.5 mark, ES for empathy was 0.31; while for RCTs, the between-group ES for mindfulness, life satisfaction, and compassion were around 0.3.
Scheepers et al78 contributed a narrative review of 23 studies looking at mindfulness-based interventions for well-being in doctors of all ages and specialities. Review authors noted some caveats; considerable variation in type and dose-intensity of practice, sparse long-term outcome data and methodological limitations. They cautioned that mindfulness practice involves time and dedication, which is not always feasible for busy healthcare professionals. In summary, the conclusions they drew are tentatively positive.
In contrast to Lomas et al77 our ES for mindfulness is higher between-group than within-group, which is somewhat unusual. This may be explained by one study42 whose positive outcome appeared determined by the decline in mindfulness over time in their control group, rather than the intervention being effective.
We identified four studies evaluating coaching and experiential workshops for GPs, which showed low effect sizes for satisfaction measures and moderate ES for work-related morale and quality of work life. There does not appear to be much literature on coaching for healthcare professionals. One quasi-experimental study by Gardiner et al79 looked at ‘cognitive behavioural coaching’ in rural Australian GPs and demonstrated a significant within-group reduction in distress levels for the coachees. Resilience training for a range of different physicians who had completed training was investigated in a recent systematic review. Four RCTs and five observational studies were included. The authors flag heterogeneous study design and use of outcome measures, as well as quality issues with weak evidence for the interventions, while indicating potential for improvement of resilience.80
We found four small-scale organisational interventions that suggested improved (job) satisfaction, as did one large-scale health policy intervention of performance-related pay in the UK. For burnout, a paucity of interventions trials delivered at organisational and systemic levels has been previously reported,81 82 the authors suggest to actively design such trials. Similarly, Dyrbye et al46 concluded that while useful, an individual intervention such as coaching is no replacement for organisational improvement. Shanafelt and colleagues have collated their vast research into burnout and put forward nine organisational strategies to address burnout and physician well-being through leadership.83 Despite calls for action, these avenues have not been adequately addressed or reported to date, at least not for GPs, and warrant further exploration. Considering the time it takes to gather and report data, it is understandable that organisations might feel pressure to implement programmes based on preliminary data.
Commendably, Dutch researchers recently investigated the effects of a mindful leadership course in hospital-based medical specialists.84 85 Both a qualitative interview and a preself-evaluation and postself-evaluation suggested an overall benefit of the intervention with improved mindfulness, life satisfaction and leadership, reduced burnout and positive change in attitudes and behaviours towards self and others. Not all participants benefited equally, suggesting a need to provide a range of interventions to meet defend participants’ needs. Future investigation will need to explore what the impact on individuals’ leadership style and on their teams is.
We included English publications only, although purposely extending our search globally. We excluded studies before the year 2000, because well-being literature in medicine is a more recent development, and general practice is now likely quite different than two decades ago. Self-reported outcome measures are typically subject to bias, particularly considering studies included GPs from different settings and cultures, potentially introducing cultural bias, rendering comparisons challenging.
Suggestions for future research
Based on our findings, we provide some suggestions which may be useful for future research into well-being and satisfaction for GPs. Stronger collaboration among researchers in this space may also lead to improved results.
(a) Selection of outcomes & outcome measures
The reduction in burnout and stress is often equated with an improvement in well-being or satisfaction. We argue that the improvement of negative outcomes does not necessarily indicate a presence of satisfaction or well-being. This aligns with the dual continuum model of mental health/mental ill-health and flourishing/languishing.25–28 Good mental health is not automatically linked to flourishing, nor is mental ill-health an indicator of languishing. Other authors have made similar statements.10 24 42 77 We did not find a single study about flourishing in GPs.
Clearly defining the constructs ‘well-being’ and ‘satisfaction’, while utilising validated well-being and satisfaction measures, will enhance clarity, consistency and comparability of study design and reporting. We suggest drawing on existing frameworks, models and definitions in the psychological literature (for different types of well-being, satisfaction or flourishing).27 86–88 To measure well-being, we suggest the Warwick Edinburgh Mental Well-being Scale (WEMWBS) and for Job satisfaction the Warr-Cook-Wall scale (WCW-JS), both of which have been validated in medical populations.89 90 Brady et al, who conducted a systematic review looking at the definition and measurement of ‘physician wellness’, similarly stated that there needs to be consensus and clarity of definition, if we want to improve the quality and comparability of research in this space.91 While this would improve the next phase of studies, the urgency in calls for actions may need to be balanced against the calls for consistency among studies.
(b) Organisational and systemic interventions
With the dearth of research in this space, and the relatively small effects for individual interventions, we believe it is worthwhile to explore system and organisational interventions (ie, mindful leadership training describe above) in the context of well-being.
Considering what is known about burnout (drivers being organisational culture, workplace conditions, lack of control and autonomy), it is not surprising that individual interventions are not as effective as desired.49 81 92–94 Hence, more combined approaches targeting both individuals and organisations have been proposed.
A 2017 British meta-analysis contrasted different types of interventions for burnout on the individual doctor and on the systemic level, whereby systemic interventions appear more effective.95 Similarly, groups in the USA state that the approach must be combined and include organisational interventions,1 96 97 mostly focusing on time management, rostering, workflow management, staffing and use of information technology solutions. Overall, there is a scarcity of organisational interventions aimed at reducing burnout,98 and conclusions from the two meta-analyses of interventions to reduce burnout should be considered provisional.
In summary, we endorse an intensified effort to explore organisational interventions to improve well-being and satisfaction, and believe a focus on leadership and improving the culture at work is a good place to start.
(c) Physical interventions
We did not find any physical interventions (ie, exercise and nutrition) geared towards improving GPs’ well-being. Sparse research on exercise or diet interventions for doctors exists. A Pakistani cross-sectional survey revealed that 76% of nearly 1200 doctors, nurses and dentists did not exercise at all, and only one participant ate according to U.S. department of agriculture (USDA) dietary guidelines.99 While a US cross-sectional survey of 303 physicians found that less than 25% knew the American Heart Association (AHA) dietary recommendations, while around half knew and followed their physical activity guidelines.100 Two systematic reviews looked at exercise and burnout in the general population, one was inconclusive,101 the other stated that physical activity effectively reduces burnout.102 Both identified methodological issues and no long-term follow-up. Seeing the paucity of data, this might provide an avenue for further investigation.
(d) Quality and risk of bias
Areas for risk of bias are inherent in this type of research. However, measures can be taken to reduce bias for example by using active controls in randomised studies as suggested by other review authors,78 by consistently publishing study protocols a priori, and controlling for participant attributes, such as prior engagement in mindfulness practice. Ideally, the same rigorous approach should be applied to intervention studies for clinicians, as to clinical interventions studies for patients.
Despite increasing interest in the improvement of well-being and satisfaction, there is remarkably little evidence beyond mindfulness interventions aimed at individual GPs. Few studies utilise validated measures of well-being and satisfaction, and favour burnout tools. Studies looking into organisational and systemic interventions remain sparse, and conclusions about their effectiveness may be premature.
Considering the COVID-19 pandemic and the added strain to primary care, programmes to support and research GP well-being should be prioritised by policymakers and governments worldwide.
Data availability statement
All data relevant to the study are included in the article or as supplementary information. All included studies are published. We will consider sharing data upon reasonable request.
Registration the protocol is registered on PROSPERO CRD42020164699.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Contributors DN is the guarantor and corresponding author and attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. Authors contributed to the study conception and design (DN, JKH, RE, IBH and NG), the acquisition (DN, AS and NG), analysis (DN, AS and NG) and interpretation (DN, AS, RE, IBH and NG) and the drafting or critical revision of important intellectual content and final approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding Dr DN was supported through the Raymond Seidler PhD scholarship. The funding source had no influence on the study design, collection, analysis or interpretation of data, the writing of the manuscript nor the decision to submit the article for publication. Award/grant number is not applicable.
Competing interests All authors have completed the Unified Competing Interest Form. Beyond the Raymond Seidler PhD scholarship for DN there was no support from any organisation for the submitted work. The funding source had no influence on the study design, collection, analysis or interpretation of data, the writing of the report nor the decision to submit the article for publication. IBH has declared financial relationships outside the submitted work in the previous 3 years and other relationships or activities.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.