Article Text
PDF
Abstract
Objective To synthesise the available evidence on the effects of interventions designed to improve the delivery of healthcare that reduces the greenhouse gas (GHG) emissions of healthcare.
Design Systematic review and structured synthesis.
Search sources Cochrane Central Register of Controlled Trials, PubMed, Web of Science and Embase from inception to 3 May 2023.
Selection criteria Randomised, quasi-randomised and non-randomised controlled trials, interrupted time series and controlled or uncontrolled before–after studies that assessed interventions primarily designed to improve the delivery of healthcare that reduces the GHG emissions of healthcare initiated by clinicians or healthcare services within any setting.
Main outcome measures Primary outcome was GHG emissions. Secondary outcomes were financial costs, effectiveness, harms, patient-relevant outcomes, engagement and acceptability.
Data collection and analysis Paired authors independently selected studies for inclusion, extracted data, and assessed risk of bias using a modified checklist for observational studies and the certainty of the evidence using Grades of Recommendation, Assessment, Development and Evaluation. Data could not be pooled because of clinical and methodological heterogeneity, so we synthesised results in a structured summary of intervention effects with vote counting based on direction of effect.
Results 21 observational studies were included. Interventions targeted delivery of anaesthesia (12 of 21), waste/recycling (5 of 21), unnecessary test requests (3 of 21) and energy (1 of 21). The primary intervention type was clinician education. Most (20 of 21) studies were judged at unclear or high risk of bias for at least one criterion. Most studies reported effect estimates favouring the intervention (GHG emissions 17 of 18, costs 13 of 15, effectiveness 18 of 20, harms 1 of 1 and staff acceptability 1 of 1 studies), but the evidence is very uncertain for all outcomes (downgraded predominantly for observational study design and risk of bias). No studies reported patient-relevant outcomes other than death or engagement with the intervention.
Conclusions Interventions designed to improve the delivery of healthcare that reduces GHG emissions may reduce GHG emissions and costs, reduce anaesthesia use, waste and unnecessary testing, be acceptable to staff and have little to no effect on energy use or unintended harms, but the evidence is very uncertain. Rigorous studies that measure GHG emissions using gold-standard life cycle assessment are needed as well as studies in more diverse areas of healthcare. It is also important that future interventions to reduce GHG emissions evaluate the effect on beneficial and harmful patient outcomes.
PROSPERO registration number CRD42022309428.
- health services research
- delivery of health care
- environment and public health
Data availability statement
Data are available upon reasonable request. Most of the extracted data and analyses are available in supplemental files. Any further data are available from the corresponding author on 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
WHAT IS ALREADY KNOWN ON THIS TOPIC
The delivery of healthcare is intrinsically associated with greenhouse gas (GHG) emissions, which are known to exacerbate global warming, thus contributing to climate change.
Clinician and health service interventions to improve the delivery of healthcare that reduces GHG emissions have been reported, yet to date, the effects of these interventions have not been systematically synthesised.
WHAT THIS STUDY ADDS
This is the first comprehensive systematic review to synthesise the existing body of evidence on the effectiveness of clinician and health service interventions to reduce the GHG emissions of clinical care.
Interventions of any type designed to improve the delivery of healthcare that reduces GHG emissions may be effective and safe in reducing GHG emissions, but the evidence is very uncertain.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Rigorous trials that measure GHG emissions of such interventions using gold-standard life cycle assessment are needed.
Introduction
The delivery of modern healthcare is inadvertently exacerbating illness and injury to populations through its own pollution. The global healthcare sector—health services and its medical supply chain—is responsible for approximately 5% of global net greenhouse gas (GHG) emissions (>2 gigatons of carbon dioxide equivalent (CO2e)) and is therefore a major contributor to climate change.1 At the same time, healthcare capacity and human health are affected by climate change, with an increasing number of people seeking medical care because of illness caused by extreme weather conditions, air pollution and degraded environmental conditions.2 In response, health systems globally are taking adaptation actions to reduce vulnerability to the effects of climate change. However, building climate-resilient and sustainable health systems encompasses both climate adaptation (eg, emergency preparedness) and emission-mitigation efforts (ie, reducing GHG emissions in the health system at their source), and the health sector has a vital role to play. Delivery of clinical care, including the production, transport and use of medical devices, consumables and pharmaceuticals, is estimated to account for ~70% of the healthcare sector’s total emissions.3 This must be reduced to uphold our responsibility to ‘first, do no harm’.4
The governments of 60 countries have committed to decarbonise their healthcare systems, with 14 planning to reach net zero by 2050.5 Because the majority of healthcare-related emissions are indirect, stemming from carbon-intensive inputs bought and consumed by the healthcare system,6–9 actions focused solely on decarbonising the direct activity of the system, such as investment in renewable energy and building efficiency, are only part of the solution. In addition to reducing waste and energy use, addressing the GHG emissions of healthcare also requires changes to clinical pathways, tests and treatments. As clinicians are the providers of clinical care, they are well positioned to influence or effect these changes.
Interventions implemented by clinicians or healthcare services to reduce the GHG emissions of clinical care have been reported,10 11 yet this area of research is in its infancy. It is important that interventions to reduce GHG emissions do not compromise quality of healthcare or patient outcomes. The aim of this study was to identify and synthesise the available evidence on the benefits and harms of interventions that have been initiated by clinicians or healthcare services specifically aimed at improving the delivery of healthcare that reduces the GHG emissions of healthcare.
Methods
The study protocol was registered on PROSPERO (CRD42022309428). We used standard Cochrane methodology and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidance.12
Criteria for considering studies for this review
Table 1 describes our study eligibility criteria. No language or date restrictions were applied.
Eligibility criteria for study selection
Search methods for identification of studies
An ‘objective approach’ was used to design the search strategy.13 This method uses text mining to develop search strategies with high sensitivity and precision and is especially helpful for complex search strategies.14 We searched PubMed via the National Library of Medicine, Embase via Elsevier, Cochrane Central Register of Controlled Trials (CENTRAL) via Wiley and Web of Science via Clarivate from inception to 3 May 2023 (online supplemental file 1). The WHO trials portal was searched via Cochrane CENTRAL. All included studies and relevant review articles were checked for additional references. We also contacted experts in the field and study authors of abstracts to identify any subsequent publications.
Supplemental material
Data collection and analysis
Study selection
Initial screening of titles and abstracts and full texts of potentially eligible articles were independently screened by pairs of authors (KP, RH, MG, TP and/or RB). Discrepancies were resolved by discussion until a consensus was reached.
Data extraction
Pairs of authors (KP, RH, MG, TP and/or RB) who independently extracted data from the included studies used a standardised form. Studies that only reported results in abstracts (such as conference proceedings) were classified as awaiting classification and their results were not included in our results synthesis. Interventions were classified according to the Cochrane Effective Practice and Organisation of Care (EPOC) taxonomy15 and described according to the Template for Intervention Description and Replication (TIDieR).16
Risk of bias assessment
Risk of bias assessment was conducted independently by pairs of authors (KP, RH, MG, TP, DAO’C and/or RB) with discrepancies resolved by a consensus. For trials, we planned to use Cochrane’s Risk of Bias 1.0 tool.17 For observational studies, we used a modified checklist adapted and used by the Cochrane Childhood Cancer Group.18–20 The domains include selection, attrition, detection and confounding bias (internal validity) and reporting bias for generalisability, adequate follow-up, appropriate outcomes and appropriate analysis (external validity).
Measurement of intervention effect and data synthesis
Our primary comparison was any intervention implemented to improve the delivery of healthcare that reduces GHG emissions compared with no intervention/usual practice. We prepared a structured summary of intervention effects table for each outcome reported at the end of the intervention delivery. Where studies were judged to be sufficiently similar in terms of study designs, interventions, settings and outcomes, we planned to pool data in a meta-analysis using a random-effects model.
However as meta-analysis was not possible, we performed vote counting based on the direction of effect (beneficial, harmful or null effect).21 Where possible, we estimated a percentage of change to give an estimate of effect size. In situations where multiple effect measures were reported with inconsistent results (eg, some beneficial measures and some harmful), we summarised the net effect according to the study authors’ overall interpretation of the intervention effect.
Summary of findings and assessment of the certainty of the evidence
We created a summary of findings table for the primary comparison for the following outcomes: GHG emissions, financial costs, effectiveness, patient-relevant outcomes, harms, acceptability and engagement.
Two review authors (RH, KP) independently assessed the certainty of the evidence (high, moderate, low and very low) using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation) approach.22 Disagreements were resolved via discussion until a consensus was reached.
Patient and public involvement
Patients were not directly involved in setting the research question or the outcome measures or in the design or implementation of the study. No patients were asked to advise on interpretation or writing up of results.
Results
Description of studies
Search results
After exclusion of duplicates, the search yielded 1758 studies and 151 full-text reports were assessed for eligibility (figure 1). We excluded 102 reports (online supplemental file 2) and included 21 studies (20 uncontrolled before–after studies and 1 interrupted time series).23–43 25 conference abstracts were included in studies awaiting assessment (online supplemental file 3).
Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram displaying the number of studies identified and included from databases, registers and other sources.
Setting and target populations
Table 2 and online supplemental file 4 detail the characteristics of the included studies. Published between 2011 and 2023, seven were conducted in the USA,25 26 33–35 37 42 six in the UK,24 27 31 32 38 39 three in Australia,29 36 40 and one in each of France,41 Germany,43 Portugal,30 Singapore23 and Spain.28 All studies compared a period (or number of cases) of no intervention with a period or subsequent periods/cases of intervention delivery. All included studies were conducted in hospital settings with approximately half (n=11) targeting staff within a single anaesthesia department.23–26 28 36 37 40–43
Characteristics of included studies
Interventions
Table 3 summarises the interventions in the included studies according to the EPOC taxonomy and online supplemental file 5 describes the interventions of each study according to the TIDieR checklist. Descriptions of the materials used, tailoring, modifications and intervention fidelity were sparse. Interventions targeted a change in delivery of anaesthesia (n=12),23–26 28 31 36 37 40–43 reducing waste and/or increasing recycling (n=5),27 30 33 35 38 reducing unnecessary test requests (n=3)29 32 34 and reducing energy use (n=1).39 Where reported, study intervention periods ranged in duration from 1 week30 to 18 months,41 and the follow-up data collection period ranged from 2 months34 to 32 months.32
Interventions in included studies (summary) by EPOC taxonomy and subcategory
Three studies evaluated a single intervention type (all relying solely on environmental restructuring targeting anaesthesia and waste),27 28 38 while the remainder evaluated multicomponent strategies. These ranged from two to six intervention components within the one study (table 3). Implementation strategies such as education (n=18),23–26 29–37 39–43 were most common, followed by delivery arrangements (n=16) (eg, environmental restructuring)23–28 30 31 33–36 38 40 42 43 and governance arrangements (n=2) (eg, policies to increase accountability for quality of practice).25 29
Outcome measures
The outcomes reported in each study are summarised in online supplemental file 6. 18 studies reported GHG emissions as an outcome,23 26–30 32–40 42 43 15 reported financial costs,23–25 27 29 30 32 34–37 39–42 20 reported effectiveness measures,24–27 29–36 38–40 42 43 and only single studies reported harm29 and staff acceptability.31 No studies reported engagement with the intervention or patient-relevant outcomes other than death.
GHG emissions were reported in different units such as CO2e per period of time,26 30 32 33 35 36 41 42 per patient or admission,29 34 37 38 43 wasted,39 saved23 27 or in terms of global warming potential.38 40 In one study, the unit of GHG emissions reported was unclear.28 Various methods of measuring the GHG emissions of clinical care were also used with only four studies using life cycle assessment (LCA),27 29 34 38 the international standard for assessing the environmental impact of a product or service across its life cycle.44 45
Financial costs were estimated or measured and reported differently across the studies. For example, total expenditure on volatiles before and after intervention,24 total cost savings on biochemistry tests,32 and total cost of non-hazardous waste disposal by recycling and landfill.33
Various ways of measuring effectiveness were used including volatile bottle usage,24 26 37 40 41 43 fresh gas flow rates,25 26 42 minutes of volatile agent use,42 total waste weight generated, incinerated, diverted and saved,38 electrical energy (kWh) wasted per year39 and reduction in unnecessary tests.29 32 34 Harm was measured in terms of patient mortality,29 while acceptability was measured using an anonymous staff survey.31
Risk of bias of included studies
Study risk of bias assessment is reported in online supplemental file 7. One study was judged to be at low risk of bias across all items.38 Seven studies were deemed at low risk of bias for confounding,25 26 29 30 33 38 41 three for generalisability33 38 41 and four for outcome reporting.29 30 38 40
Effects of interventions
Primary comparison: any intervention type compared with no intervention.
Primary outcome: GHG emissions.
17 of the 18 studies (94%) that measured GHG emissions reported effect estimates favouring the intervention (table 4).23 26–30 32–38 40–43 This included the single study judged to be at overall low risk of bias.38
Summary of intervention effects for GHG emissions
Nine of 12 studies targeting anaesthesia reported a reduction in GHG emissions ranging from 25% after three Plan–Do–Study–Act cycles (incorporating environmental restructuring, education, audit and feedback, reminders and strategies to optimise organisational culture) to increase the use of low-flow anaesthesia and decrease sevoflurane use26 to 100% following implementation of an anaesthetic gas capture technology to absorb and recycle expelled halogenated anaesthetic gases by the patient.31 The remaining three studies did not measure the effect on GHG emissions.
Three of five studies evaluating interventions targeting waste disposal reported a reduction in GHG emissions ranging from 32% following a multicomponent intervention including clinician education, audit and feedback, recycling implementation and relocation of landfill and medical waste bins30 to 85% after converting from single-use to reusable sharps containers in acute care hospitals.38 Two studies reported reduced GHG emissions,33 35 but percentage reductions could not be estimated as before–after values were not reported. All three studies29 32 34 evaluating interventions targeting unnecessary testing reported reductions in GHG emissions ranging from 9.5% following implementation of a telehealth preoperative evaluation process and a clinical practice guideline34 to 37% after implementation of a policy to reduce non-urgent pathology testing to 2 days per week combined with provider education.29
The single study evaluating an intervention to reduce energy use found no improvement in GHG emissions wasted following an audit and feedback and educational intervention aimed at reducing the number of workplace computers left on overnight and on weekends.39
Overall, interventions designed to improve the delivery of healthcare that reduces GHG emissions may reduce GHG emissions, but the evidence is very uncertain (table 5). According to GRADE, evidence from observational studies starts at low certainty and we further downgraded for bias (confounding and generalisability, and inadequate follow-up and outcome reporting bias) and indirectness (outcome dissimilarity). Changes in anaesthetic use, waste and costs were used as a surrogate for GHG emissions.
Secondary outcomes
Financial costs
13 of 15 (86%) studies measuring financial costs, none judged at low risk of bias on all criteria, reported effect estimates favouring the intervention (online supplemental file 8).23 24 27 29 30 32 33 35–37 40–42 Reduction in costs ranged from 14%33 to 63%,27 and in three, we were unable to estimate a percentage change.23 35 37 One additional study reported a non-significant cost reduction (p=0.81)25 and one reported a small increase in costs.39 Overall, interventions designed to improve the delivery of healthcare that reduces GHG emissions may reduce financial costs, but the evidence is very uncertain (downgraded due to observational study design, bias and outcome dissimilarity) (table 5).
Effectiveness
18 of 20 (89%) studies that reported effectiveness, including the study judged at overall low risk of bias,38 reported effect estimates favouring the intervention (online supplemental file 9).24–27 29–33 38 40 42 43 Of these, 11 of 11 (100%) studies targeting anaesthesia use were beneficial, 5 of 5 (100%) studies targeting waste were beneficial and 2 of 3 (67%) studies targeting unnecessary tests were beneficial. One study targeting energy use reported no meaningful change in effectiveness,39 and one targeting unnecessary tests reported mixed beneficial and harmful results.34 Overall, interventions designed to improve the delivery of healthcare that reduces GHG emissions may reduce anaesthesia use, waste and unnecessary tests but lead to little or no change in energy use, but the evidence is very uncertain (downgraded due to observational study design and bias) (table 5).
Harms
The single study that reported harms, judged at unclear risk of generalisability bias, reported that in-hospital mortality did not differ before or after implementation of a policy to reduce non-urgent pathology testing (OR 1.09 (95% CI 0.75 to 1.59)).29 Based on this study, interventions designed to improve the delivery of healthcare that reduces GHG emissions may have little to no effect on patient harms, but the evidence is very uncertain (downgraded due to observational study design and two levels due to imprecision) (table 5).
Acceptability
The single study that measured acceptability (judged at high risk of confounding bias and unclear risk of generalisability, follow-up and outcome reporting bias) reported that most staff using nitrous oxide cracking equipment stated it was easy or very easy to set up (16 of 22, 73%), explain how to use to patients (19 of 22, 86%) and change the masks and filters between patients (19 of 22, 86%).31 However, 36% of staff (8 of 22) reported concerns about the size of the equipment and problems with the technology. Based on this, interventions designed to improve the delivery of healthcare that reduces GHG emissions may be acceptable to staff, but the evidence is very uncertain (downgraded due to observational study design, bias and two levels due to imprecision) (table 5).
Discussion
Statement of principal findings
Based on the available evidence, which included 21 observational, hospital-based studies from eight high-income countries, interventions of any type designed to improve the delivery of healthcare that reduces GHG emissions may be effective and safe in reducing healthcare GHG emissions, but the evidence is very uncertain. No studies reported patient-relevant outcomes other than death or engagement with the intervention. More than half of the interventions targeted delivery of anaesthesia, while others targeted waste, unnecessary tests and energy use. Healthcare provider education was the most common type of intervention, used in all but three studies, followed by environmental restructuring. Only one in five studies (4 of 21) measured GHG emissions using the gold-standard life cycle analysis method. All but one study was susceptible to bias: in particular, confounding, generalisability and outcome reporting biases. As there were no comparative effective studies, we cannot draw conclusions about which intervention/s may have greater effects.
Strengths and limitations of this study
A strength of our review is that it includes the spectrum of interventions aiming to reduce GHG emissions generated by clinical care that have been studied and reported across healthcare settings worldwide. We also used accepted Cochrane and GRADE methods22 to synthesise the available evidence and appraise its certainty. However, the available evidence on which our review is based has several limitations. First, no trials were identified, and we included 21 clinically and methodologically diverse studies, of which all but one were judged to be at risk of various biases. The applicability of our findings is also limited to interventions implemented in the hospitals of high-income countries and mostly targeting delivery of anaesthesia.
It was also not possible to prespecify a minimal important difference in GHG emissions as this is currently unknown. This impacts interpretation of our primary outcome. Related to this, few studies used LCA, the internationally standardised method for quantifying GHG emissions, which may also have contributed to measurement error. Finally, 25 conference abstracts were included as studies awaiting assessment as we were unable to confirm their eligibility. As the methodology and characteristics of these studies were similar to those included in this review, it is unlikely that their inclusion would have appreciably changed our conclusions. Similarly, the identification of additional observational research is unlikely to alter our conclusions.
Comparison with previous research
Studies investigating the effectiveness of interventions aimed at improving the delivery of healthcare that reduces GHG emissions are few in number, limiting direct comparisons with other research. In keeping with our findings, a number of small-scale observational quality improvement projects have been undertaken in the UK and have reported reductions in GHG emissions.46–48 However, these studies were case reports and therefore did not meet our eligibility criteria.
A systematic review, performed by members of our team, that limited study inclusion to only those that evaluated behaviour change interventions designed to reduce the GHG emissions generated by clinical care included six of the same studies we identified.49 The current review includes all intervention types (eg, delivery, financial or policy arrangements) and found that while most included at least one component that targeted the behaviour of healthcare clinicians (eg, education), many also used environmental restructuring at the hospital level (eg, changes to anaesthetic machines, conversion from disposable to reusable equipment) to improve the delivery of greener clinical care, while others used governance arrangements. Another systematic review summarised interventions aimed at improving operating theatre environmental sustainability.50 Of the 34 included studies, only 1 was included in our review.37 The other studies were comparative footprint studies that did not implement an intervention, or evaluated interventions primarily designed to reduce waste rather than GHG emissions. Neither of these reviews synthesised the evidence systematically or used the GRADE approach to appraise the certainty of the evidence.
A systematic review published after our final search date that assessed the effectiveness of telemedicine programmes reported GHG emission savings ranging from 0.9 to 900 kg CO2 per teleconsultation in all 48 included studies.51 Of note, none of the included studies in this review met our eligibility criteria as either they did not implement an intervention that was designed to improve the delivery of telemedicine or there was no control or non-intervention group.
Other reviews have examined hospital environmental sustainability more generally,52 sought to quantify the environmental impact of health services and clinical pathways,53–56 or identified opportunities to improve the sustainability of health and social care57 but did not identify any studies that actively addressed these opportunities.
Implications for clinicians and policymakers
While clinical care alternatives that reduce GHG emissions have been identified (eg, low-flow anaesthesia,58 telehealth,59 dry-powder instead of metred-dose inhalers60), the uptake of most of these initiatives is reliant on changes in healthcare practice. This review identified a range of interventions that have been implemented with most relying on behavioural change of individual clinicians. However, successful implementation often requires whole-of-system change.61 A multifaceted approach is therefore likely needed for rapid decarbonisation of healthcare, including policy improvements and system-wide interventions, in addition to targeting the behaviour of individual clinicians and consumers. Only two studies in this review evaluated a policy change/governance arrangement at the health service level, but scalability and sustainability of the intervention are unknown.25 29
More than half of the interventions evaluated in our review targeted delivery of anaesthesia, a recognised hotspot of GHG emissions in the healthcare sector and where mitigation efforts are anticipated to have high impact.62 However, 2 of the 11 studies used technological solutions for capturing or destroying inhaled anaesthetic waste despite a recommendation from the American Society of Anesthesiologists against this practice as a high mitigation priority because reuse of the captured gas is not currently approved.58
In addition, care which provides little or no value, or may even harm patients (ie, low-value care), is a less recognised but significant source of negative environmental impact that requires urgent attention. While many healthcare organisations have implemented interventions aimed at reducing low-value care, this review identified only three studies that measured the impact of such interventions on GHG emissions. Identifying and eliminating sources of low-value care present an opportunity to accelerate reductions in avoidable GHG emissions, healthcare expenditure and iatrogenic harms in a direct and immediate way.
Implications for research
Our review has several implications for future research in this field. The body of evidence for interventions designed to reduce the GHG emissions of healthcare would be strengthened by using rigorous study designs, such as randomised controlled trials, and for research to be extended beyond hospital settings (particularly anaesthesia) and high-income countries. Despite the exponential growth in the number of healthcare LCAs conducted over the past two decades, existing LCAs cover only a small proportion of available healthcare products and processes across a few geographical settings.63 Future research should focus on building this evidence base and improving the standardisation among healthcare LCA studies, to provide the foundation for effective interventions and targeted mitigation strategies that prioritise the biggest reductions.
Future studies should also improve the reporting of intervention descriptions according to the TIDieR guidelines,16 including the theory or framework underpinning the interventions and intervention fidelity to facilitate reproducibility and comparability. Measurement of implementation outcomes such as acceptability, adoption, fidelity and sustainability will be integral to ensure optimal uptake of interventions in clinical practice. Finally, research and international consensus to determine a minimal important difference in GHG emissions is needed to guide interpretation of effects that are meaningful. Future studies should also include measurement of patient health outcomes and monitoring for potential adverse or unintended consequences of interventions aiming to reduce GHG emissions.
Although telemedicine and the replacement of desflurane with sevoflurane are known to reduce GHG emissions,51 64 further research evaluating effective methods of increasing these practices is warranted. Further, while this review focused specifically on what can be done by clinicians and health services to reduce GHG emissions of clinical care, it should be acknowledged that a comprehensive climate change mitigation strategy should also address energy efficiency, building design, production of renewable energy and transportation.
Conclusion
Interventions of any type designed to improve the delivery of healthcare that reduces GHG emissions may be effective and safe in reducing GHG emissions, but the evidence is very uncertain. Rigorous studies that measure environmental impacts using gold-standard LCA in addition to patient outcomes are needed to determine their true effects.
Summary of findings (GRADE)
Data availability statement
Data are available upon reasonable request. Most of the extracted data and analyses are available in supplemental files. Any further data are available from the corresponding author on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
Ethics approval was not required for this systematic review.
Acknowledgments
We would like to thank Justin Clark for conducting our search strategy and Anna Scott for assistance in using MethodsWizard, a semiautomated tool to aid in writing our protocol.
References
Supplementary materials
Supplementary Data
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.
Footnotes
KP and RH are joint first authors.
X @PicklesKristen, @DrRomiHaas, @Thanya_Indu, @AlexBarrattDr, @RachelleBuchbin
Contributors KP and RH contributed equally to this paper. KP and RH jointly coordinated the project administration. KP, RH, MG, TP and RB conceived the project. All authors contributed to the methodology. KP, RH, MG, TP, DAO'C and RB extracted data. All authors contributed to the formal analysis and validation. RB supervised the work. KP, RH and RB wrote the original draft. All authors discussed the results, contributed to the editing of the manuscript, had full access to all the data in the study and had final responsibility for the decision to submit for publication. KP, RH and RB are the guarantors.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests All authors have completed the ICMJE uniform disclosure form and declare: no financial support for the submitted work; financial relationships with the Australian National Health and Medical Research Council (NHMRC), Australian Department of Health, HCF Foundation, Cabrini Foundation and Arthritis Australia in the previous 3 years supporting other work; no other relationships or activities that could appear to have influenced the submitted work.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
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.