A systematic review of interventions to provide genetics education for primary care

Background At least 10 % of patients seen in primary care are said to have a condition in which genetics has an influence. However, patients at risk of genetic disease may not be recognised, while those who seek advice may not be referred or managed appropriately. Primary care practitioners lack knowledge of genetics and genetic testing relevant for daily practice and feel inadequate to deliver genetic services. The aim of this systematic review was to evaluate genetics educational interventions in the context of primary care. Methods Following the process for systematic reviews developed by the Centre for Reviews and Dissemination, we conducted a search of five relevant electronic databases. Primary research papers were eligible for inclusion if they included data on outcomes of interventions regarding genetics education for primary care practitioners. The results from each paper were coded and grouped under themes. Results Eleven studies were included in the review. The five major themes identified inductively (post hoc) were: prior experience, changes in confidence, changes in knowledge, changes in practice, satisfaction and feedback. In five of the studies, knowledge of practitioners was improved following the educational programmes, but this tended to be in specific topic areas, while practitioner confidence improved in six studies. However, there was little apparent change to practice. Conclusions There are insufficient studies of relevant quality to inform educational interventions in genetics for primary care practitioners. Educational initiatives should be assessed using changes in practice, as well as in confidence and knowledge, to determine if they are effective in causing significant changes in practice in genetic risk assessment and appropriate management of patients. Electronic supplementary material The online version of this article (doi:10.1186/s12875-016-0483-2) contains supplementary material, which is available to authorized users.


Background
It is estimated that approximately 10 % of United States (US) citizens [1] and 6 % of Europeans [2] are affected by a rare disease, equating to approximately 30 million individuals in the US [1] and the same number in Europe [2]. Of all rare diseases, 80 % have a genetic component [2] and it is therefore not surprising that at least 10 % of patients seen in the primary care setting are said to have a condition in which genetics has an influence [3]. However, there is evidence that patients at risk of genetic disease may not be recognised, while those who seek advice about their risks of rare genetic diseases may not be referred or managed appropriately by health professionals [4]. This may relate to lack of awareness of the family that a condition may have a genetic component, or may be due to lack of knowledge in primary care professionals that this may be so. A Dutch study in 2003 [5] showed that 70 % of patients believed that general practitioners (GPs) were not competent to handle queries about rare genetic conditions adequately, while in the US similar concerns about the competence of primary care practitioners to manage the patients with genetic concerns appropriately have been raised [6]. Undetected genetic risk can have serious consequences for the entire family, for example through preventing access to screening or preventive drugs or surgery (e.g. for cancer), resulting in increased morbidity, mortality, family burden and healthcare costs. Raising awareness of the risks and potential management of such cases is important for the patient and the wider family.
In the age of genomics, primary care practitioners will be increasingly involved in preventive care and organisation of relevant surveillance and develop a more flexible way to deal with patients' requests for information and genetic tests. Primary care can be provided by a range of health professionals and is defined by the World Health Organisation as health care that is directly accessible by patients as the first point of contact, as well as being comprehensive and ongoing [7]. It involves prevention, and pre-symptomatic detection of disease, as well as early diagnosis [7], all of which are relevant to patients at risk of genetic disease. To illustrate this, Acheson et al. [8] reported that a sample of 100 primary care physicians had faced questions from patients related to familial cancer, neurological conditions, prenatal testing issues and a wide range of single gene disorders, confirming the need for knowledge and skill to deal with such queries. However, it has been reported that primary care physicians lack knowledge of genetics relevant for daily practice, lack oversight of genetic testing and feel inadequate to deliver genetic services [6,9,10].
As a result of a Delphi study involving experts in primary care education and genetic specialists in the United Kingdom (UK), Burke et al. [11] identified that primary care practitioners had three main responsibilities in relation to genetics: being able to identify patients at risk of a condition, contributing to medical management of such patients and appropriate communication of genetic information to patients. The authors of the paper developed a curriculum for genetics education in primary care organised according to these three activities [11]. However, genetics education is only very slowly starting to become a common part of medical curricula. This was made evident in a study by Julien-Reynier et al. [12], who studied the educational needs of 1170 GPs in five European countries and found that priorities for education differed according to country, gender and age of the physician. An average of 23.5 % (range 4.7-42 %) had received no genetics education in their undergraduate medical training, while another 29.7 % reported that the education they had received was not useful to them in their practice. A total of 984 (85.1 %) had not undertaken any continuing medical education related to genetics. Overall, the highest priority for education was 'Genetics of Common Diseases' , followed by ' Approaching Genetic Risk Assessment in Clinical Practice' , 'Psychosocial and Counseling Issues' , 'Basic Genetics and Congenital Malformations' , 'Ethical, Legal and Public Health Issues' and lastly 'Techniques and Innovation in Genetics'.
However, provision of education may not always be sufficient to ensure improvements in patient care. In a paper on utilisation of research in practice, Grol and Grimshaw [13] point out that while many health professionals undertake continuing medical education, it is difficult to link this with changes in practice. This is borne out by Weinreich et al. [14], who instituted an educational programme focussed on improving detection of patients at risk of haemoglobinopathy, but found that while GPs regarded the education positively, this was not converted into changes in screening based on the patient's ethnicity.
It is therefore clear that continuing medical education that is grounded in daily practice of GPs is urgently needed [9]. In order to provide effective genetics education, knowledge of the interventions that have been used in the past, and the outcomes of those interventions with their potential impact on health, is required. The aim of this study was to conduct a systematic review of educational interventions focussed on genetics and used in the context of primary care. The specific research question was: What are the outcomes of educational interventions in genetics for primary care professionals? As this was a mixed methods review, we used an inductive approach to search for possible outcomes. However, we anticipated that in line with the educational assessment framework proposed by Kirkpatrick [15] that is described later in this paper, outcomes would include satisfaction, changes in knowledge and skills and changes in behaviour.

Methods
Conducting a systematic review enables the evidence on a particular topic to be gathered, analysed and synthesised. Adherence to a rigorous set of guidelines is essential to ensure rigour and objectivity. We followed the process for systematic reviews developed by the Centre for Reviews and Dissemination [16], which involved identification of relevant search terms, selection of studies based on explicit inclusion and exclusion criteria and quality assessment of papers (Table 1). The software package EndnoteX7 [17] was used to store, organise and retrieve all citations during the review process.

Search strategy
We conducted a search of five relevant electronic databases: Medline, CINAHL, SocIndex, PsychInfo and ERIC, using relevant search terms (see Table 2).
The search terms were derived from numerous exploratory searches of the literature (as per CRD guidance) [16] and suggestions from experts in this field. We searched for the terms in titles and abstracts.
The search focussed on papers published between January 2005 and July 2015. The start date was chosen due to changes in the importance of primary care in genetic and genomic healthcare at that time, when the emergence of genomics in health shifted the focus of care in genetics from purely specialist services to primary and secondary care. For example, a seminal paper was published in 2003 urging the involvement of a wider health professional community in healthcare using genomics [19], while the same year there was recognition in the UK NHS White Paper [20] that primary care had a role in managing patients with genetic conditions, resulting in subsequent funding of posts for GPs with a special interest in genetics. It was therefore reasonable to expect that papers published in response to these initiatives might appear as early as 2005. We limited the search to studies that included human participants and were published in English. We also handsearched the reference lists of relevant papers.
Papers were eligible for inclusion if they: i) were reports of primary research using qualitative, quantitative or mixed methods designs, ii) included data on outcomes (i.e. change in knowledge or change in practice) of interventions focussed on genetics education for primary care practitioners. Papers were excluded if they: i) focussed on health professionals other than those working in primary care, ii) focussed on patient education, iii) described an intervention but did not report outcomes, iv) reported educational needs without use of an intervention v) provided information on educational interventions but the results for primary care providers could not be separated from results for other professionals. For this purpose, we used the WHO definition of primary care and included practitioners from any healthcare profession who offered ongoing and comprehensive care at the first point of contact, directly accessible to patients [7].
As the purpose of a pilot study [21] is to assess the feasibility and logistics of a further (larger) research study, rather than to test an hypothesis and generate reliable and valid findings, we did not include pilot studies. Papers identified for possible inclusion were assessed by at least two researchers, who independently classified the papers and then discussed any differences of opinion until a consensus was reached [16]. A table of excluded papers and reasons for exclusion is available included as Additional file 1.
The eligible papers were assessed for methodological rigour by two researchers (HS and either DT or MP). Each researcher independently scored the paper using the system described by Kmet et al. [22]; scores were then compared and discussed until a consensus was reached. Kmet et al. [22] provide a checklist of 14 items for a quantitative study and 10 items for qualitative studies. A score of 2 (criterion fully met), 1 (criterion partially met) or 0 (criterion not met) is allocated to each item and the scores are totalled.

Data extraction and synthesis
Data relating to the methods and main outcomes of each study were extracted and presented in a table. Relevant results or findings from each eligible study were coded and entered into a master table. The results/findings were grouped under themes that were identified inductively (post hoc) (see supplemental material for more detail on the codes and themes). Coding was undertaken by two researchers and reviewed by a third researcher. As the papers were heterogeneous in terms of methods, interventions and outcome measures, neither metaanalysis nor meta-synthesis was possible and we present the findings in a narrative form [16].
Because there were multiple outcomes for numerous studies, we reported every outcome and categorised these according to Kirkpatrick's framework [15] for evaluating educational outcomes. This model can be used to distinguish the impact of education at four levels: valuation (level 1; satisfaction), learning (level 2; knowledge and knowledge retention), behaviour (level 3: applying To determine the outcomes of educational interventions in genetics for primary care professionals.

Participants
Sample of any size of primary care professionals who were recipients of an educational intervention focussed on genetics. Participants could be based in any country and be from any profession involved in delivering primary care Intervention Any educational intervention directed at qualified health professionals in primary care and focussed on genetic healthcare, including but not restricted to online or face to face education lasting one hour or more.

Comparators
None; control group of comparable professionals; control group of comparable practices.

Outcomes
Any outcomes, including but not restricted to: satisfaction with the educational intervention, changes in knowledge, changes in confidence changes in skills, changes in clinical behaviour, use of acquired genetics competencies, impact on organizational change or impact on patient health.

Study design
Any study design, including RCTs, quasi-experimental studies, cohort studies and qualitative studies. knowledge and consultation skills regarding timely recognition of patients at risk) and effects on patient health and organization (level 4: changes in actual practice performance [i.e. referral] and results). The impact on society, or patient safety in genetic medical care, is part of level 4. The first learning step, according to Moore [23], would be to try and inform primary care health workers and aim for a better understanding of genetics. The first level of Kirkpatrick's framework for evaluating the educational outcome then assesses satisfaction with the genetic modules. The higher the level, the more complex the potential learning outcome of the genetic module and the greater the possibility of having an impact on (genetic) health. We used Moore's model of CPD curriculum design [23], identifying individual learning steps with their educational objectives and used the Kirkpatrick framework as a model to evaluate the genetic modules [15]. These analyses are presented in Table 5.

Results
There were twelve papers [24][25][26][27][28][29][30][31][32][33][34][35] identified for potential detailed review. However, two papers reported the same study [34,35], so eleven papers were reviewed. The characteristics of the eleven individual studies [24][25][26][27][28][29][30][31][32][33][34] are presented in Table 3. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart [18] in Fig. 1 shows the selection process and the reasons for exclusion of papers. A list of excluded papers (with reasons for exclusion) is provided in the supplemental material. Regarding quality of the papers reviewed, the consensus quality score totals ranged (as a percentage of the maximum possible score) from 70 to 100 % and as we set the minimum acceptable score at 70 %, all eleven papers were deemed eligible for inclusion in the detailed review. We have included the scores and comments in Table 3 to enable judgements on the quality of the evidence to be made.

Overview of the studies
In the eleven papers included, three studies were undertaken in the United States (US) [27,31,33], two in Canada [25,26], three in the United Kingdom (UK) [24,28,34], two in the Netherlands [29,30] and one in Australia [32]. The style of delivery and topics covered are reported in Table 4. Six of the interventions were specifically aimed at GPs [26,27,29,30,33,34]: the remaining four were aimed more generally at primary care professionals. In all studies, the intervention was assessed by testing changes in knowledge and/or practice. Houwink et al. [29] also tested the practitioner's knowledge, attitudes and skills using a standardised patient. Laberge et al. [31] assessed the impact of their intervention on both clinical practice and genetic teaching of primary care specialists in university settings.
The findings of the synthesis of evidence are presented under five main themes: prior experience, changes in confidence, changes in knowledge, changes in practice, satisfaction and feedback. A detailed table showing categories and themes derived from the paper is available in Additional file 2.

Prior experience
Authors of only one study reported prior experience that could influence provision of genetic healthcare. Srinivasan et al. [33] found that only 69 (25 %) of the 279 residents training in primary care in their study had experience of genetic conditions amongst family and friends. Of those 69, 35 (12.5 % of total sample) had personal experience of involvement in the care of an affected person and over three quarters had cared for a patient with a genetic condition. While 55 % had ordered a genetic test, none had offered more than a few minutes of counselling to accompany those tests, although it is not stated what proportion of these tests may have been primarily to establish a diagnosis rather than being of a predictive nature. Of 111 GPs attending a workshop on prenatal tests [32], 40 % reported recording the full family history of the patient 'always or often' (p594).

Changes in confidence in offering genetic healthcare
Practitioner confidence was shown to improve in six of the ten relevant studies, although the methods of assessing confidence differed across the studies. General practitioner confidence measured by Wilson et al. [34] did not significantly alter with respect to taking family history or ability to respond to patient queries about breast cancer risk. Similarly, only 23 % of the 22 practitioners interviewed by Laberge et al. [31] felt more confident in offering genetic healthcare.
However, Bethea et al. [24] reported participants in the intervention practices were more confident than controls (Odds Ratio (OR) 2.50, p = 0.03), especially related to family history collection (OR 2.39, p = 0.02), risk assessment (OR 3.65, p = 0.01) and reassuring those patients whom they assessed as low risk (OR 3.94, p = 0.01). While confidence scores increased across a range of 13 different topics when assessed by Carroll et al. [25], significant differences were found for only six activities or topics: assessing genetic risk (p = 0.033), making appropriate referrals (p = 0.033), discussing prenatal testing (p = 0.034), discussing benefits and limitations of genetic testing (p = 0.033), discussing content of a genetic counselling session (p = 0.022), and genetics of adult-onset disorders (p = 0.005) [25]. Participants in that study also gained confidence in being able to act as a resource for other colleagues with respect to genetics issues. Selfreported confidence in managing prenatal screening issues improved in the study by Metcalfe et al. [32], with 15.9 % Self-assessed confidence in skills related to managing adult-onset conditions increased from pre-course mean score of 2.3 of a possible score of 5 (SD = 0.7) to post-course mean scores of 3.0 (SD = 0.9), (p = 0.005). Self-assessed confidence in skills related to prenatal genetics did not increase: the mean score pre-course was 3.4 (SD = 1.1) and post score was 3.6 (SD = 1.1     of GPs feeling quite or very confident pre-workshop, compared with 81.9 % post-workshop and 45.1 % six to eight months later. Those in the intervention group studied later by Carroll et al. [26] reported more confidence across a range of competencies (adjusted mean score of 37.9/55 in control group compared to 47.0/55 in intervention group) and scored higher than the control group on the topic of making referral decisions (7.8/10 compared with 6.4/10). Confidence was assessed in only the lead clinicians in Emery et al.'s study [28]: their self-reported confidence in managing patients with a family history of cancer improved after training (p < 0.001) and remained higher than baseline levels for the following twelve months.

Changes in knowledge
Authors of five studies reported improvements in knowledge of attendees. Carroll et al. [25] tested a small sample of 21 participants' awareness of genetic services pre and post course and found no significant differences. Knowledge of breast and ovarian cancer was not shown to increase in participants, however a question on colorectal cancer was answered correctly by significantly more primary care providers (24 % pre-course increasing to 50 % post-course, p = 0.031). Although intervention participants had higher knowledge scores than controls in another study led by Carroll [26] (adjusted mean score of 47.0/55 in the intervention group compared to 37.9/55 in control group), the difference was not significant, nor was overall knowledge of genetics significantly changed in the study by Srinivasan et al. [33].
However, Houwink et al. [30] measured genetics knowledge using a test comprising 20 multiple choice items at three time points and found that knowledge had improved in the intervention group post-course and    [26] Not stated One face to face interactive workshop session of 60 min. Portfolio of tools that could be used in primary care practice. Information sheet on current topics sent to participants every two weeks during the trial.
To improve referral decisions, confidence and knowledge relevant to primary care genetics.
• Practical medical genetics information • Risks, benefits and limitations of genetic testing including psychosocial risks, confidentiality and insurance issues. Over two years, a 60 min face to face lecture each quarter (8 lectures in total) and a monthly 45 min didactic seminar.
To enable family physicians to provide appropriate services to patients with family or personal history of birth defects or mental retardation.  six months post-course. Professionals in participating practices correctly assigned risk in relation to familial breast cancer more frequently than those from nonparticipating practices (p = 0.04) in the study by Bethea et al. [24], while Clyman et al. [27] showed significant improvement in knowledge of family physicians postcourse (p < 1×10 −10) and that they were less likely to over-refer or under-refer in the context of a hypothetical scenario. Metcalfe et al. [32] tested knowledge of prenatal screening, appropriate prenatal tests and risks of recessive inheritance and Down syndrome and the scores improved significantly from a mean of 51.2 preworkshop to 62.88 post-workshop (p < 0.001) and 58.92 at follow up (p < 0.001).

Changes in practice
Successful educational outcomes are those that lead to applications in practice. Such outcomes are difficult to assess when methods for reporting vary, and behaviour changes as a result of applying new knowledge may not reflect improvements in practice.
Houwink et al. [29] used standardised patient consultations (using trained actors) in a randomised controlled trial (RCT) to measure improvement in GP responses to familial cancer situations in practice pre-course, one month post-course and three months after oncogenetics training for GPs. Regression analyses revealed a moderate effect size for changes in performance in the intervention group between pre-test and post test score (standardised regression co-efficient = 0.34) and between one month post-test and three months post-test (standardised regression co-efficient = 0.28). Similarly, GPs who did a prenatal workshop were more likely to report discussing testing with all their pregnant patients (69.4 % pre workshop compared with 95.2 % post workshop and 85.7 % at follow up) [32].
Clyman et al. [27] showed no changes in referral pattern of course attendees. Wilson et al. [34] reported that patient referrals to genetic services did not significantly alter (p = 0.56) in the intervention group after the course and software package had been offered. However, those who had undergone the intervention were more likely to refer patients who were eventually assessed as having a high or moderate risk of inherited breast cancer, indicating a change in appropriateness of referrals. Referrals to cancer genetic services increased from those in the GRAIDS trial intervention practices [28], however when risk assessments of referred patients were made by specialist genetic staff, there were no differences found between risks to those referred by intervention practices and those referred by comparison practices.
When asked about whether they used the content of the programme in their practice following the course, 13 % of residents surveyed by Srinivasan et al. [33] replied that they used it often, 57 % occasionally and 29 % never. Houwink et al. [30] reported that 90 % of practitioners applied the knowledge at least once a month after the online course, 5 % indicated a frequency of use of at least once a week, while no participant reported daily application and another 5 % reported not encountering genetic issues in their practice. However, the authors did not assess the use of genetic knowledge prior to the course, so it is difficult to be certain how much difference the course actually made. In addition, while participants reported changes in knowledge, they did not report improvements in clinical skills. The primary care practitioners interviewed by Laberge et al. [31] were clinicians who were also involved in providing medical education. During telephone interviews with 22 respondents, 28 % reported changes in referral patterns for genetic healthcare, 23 % gave more consideration to genetics in the differential diagnosis, and 23 % said they put more emphasis on family history. Overall, 82 % indicated that the course had changed their clinical practice in some way, however this was not explored further. In another study 17 , 76 % (n = 32) attendees said their practice had changed a little, 21 % (n = 9) believed it had changed a lot. Where changes occurred, the nature of the change was unclear.
Primary care providers attending the course offered by Carroll et al. [25] were asked if their practice would change: 15 (86 %) intended to improve their family history taking and 10 (48 %) intended to increase their teaching activity in genetics.

Satisfaction and feedback
Overall, participants reported satisfaction with the educational programmes studied. The residents in the study by Srinivasan et al. [33] reported that the programme delivered via the web offered more flexibility, but less contact with teaching staff than their usual programme. They did indicate that the web-based course was effective in teaching communication strategies, medical management and ethical, legal and social (ELSI) issues. The mean rating for satisfaction by GPs who took the course provided by Houwink et al. [29] was 7.7 of a possible 10, based on two Likert scales. In the other study by those authors [30], the mean satisfaction score was 7.9/10 (Standard deviation (SD) = 1.3). While participation in the genetics in primary care educational project [31] was seen as worthwhile by all participants, the additional workload and time required to participate were seen as reasons to advise colleagues to consider carefully before committing to it. The majority of those who evaluated the course offered by Carroll et al. [25] believed the course was relevant to them (24/28, 86 %) and that learning genetics with other primary care professionals was helpful (26/28, 96 %), while the Genetikit [26] course and materials were rated as either useful or very useful by over three-quarters of attendees. Eight lectures offered by Clyman et al. [27] were regarded as of good quality by 96 % of attendees and 81 % believed the course would be useful in practice and 97.1 % GPs attending the prenatal screening workshop provided by Metcalfe et al. [32] believed it would be useful for practice.
Finally, we present in Table 5 the findings of each paper aligned to the educational models of Kirkpatrick [15] and Moore [23], showing that authors of only four studies addressed Level IV, assessing organisational and healthcare impact.

Discussion
The aim of this review was to ascertain the effectiveness of educational interventions on genetics for primary care. Practitioner confidence was positively affected in six of the eight studies where that factor was investigated, however even in those studies confidence was generally not enhanced across the entire range of possible topics. The interventions were therefore limited in increasing confidence. Knowledge of genetics topics improved in five studies, but again this tended to be subject specific and long term follow up indicated that the level of knowledge gain was not sustained. In only one of three RCTs where knowledge was tested were the changes significant. Overall, significant changes in practice were not demonstrated, with several studies showing no change and one RCT showing only a moderate effect size. While authors of some studies reported changes in practice, these were quasi-experimental or qualitative studies where the evidence was based on self-reports. While learners appeared satisfied with the programmes offered, the range of different questions posed by authors with respect to satisfaction makes the results difficult to interpret.
There are limitations to this study. Although the authors of the review followed a rigorous process to retrieve, select and analyse the available studies, the smallness of the samples and diversity of the studies made it difficult to draw coherent and robust conclusions from the evidence. In particular, we were unable to conduct a mathematical synthesis of the results. It should also be borne in mind that the search was restricted to papers published in English, so there may be evidence published in other languages that has not been included. We searched for peer-reviewed publications, so there may also be some bias due to the omission of unpublished data or results that were only available through the grey literature. However, the rigour of the study was enhanced by the use of more than one researcher to select, appraise and analyse the included papers and the search strategy was constructed based upon numerous trial database and hand searches, so we are reasonably confident that the relevant papers were retrieved.
Previous studies have cited that lack of knowledge about genetics issues, such as genetic tests, acted as a barrier to physician use [25]. In some studies included in this review, knowledge was not significantly altered: where changes occurred these tended to be as a result of prolonged exposure to genetics information, rather than framework and organisational change is needed to ensure that patients benefit from genomic advances.
If GPs stay genetically uneducated and therefore incompetent related to the use of genomic information in general practice, individual genetic medical care is likely to be unhelpful and may possibly be even harmful [13]. We believe the results of this review should be used in the near future to guide the implementation and level of assessment of genetics education internationally. Although the majority of the issues investigated cover genetics-related knowledge, skills and attitudes essential for every medical care provider, further studies will have to determine whether the results are relevant to other medical specialties as well.

Conclusions
From this review it can be seen that long-term impact of genetics educational initiatives on patient management and policy has been poorly assessed and there is a dearth of high quality RCTs to provide a strong evidence base for educational practice. From the existing evidence, it appears that short-term educational initiatives alone are unlikely to cause significant changes in practice in the areas of recognition of genetic risk, assessment of risk and appropriate management of patients in patient care. This may in part be due to challenges in measuring change, but perhaps changes in knowledge should not be the main aim. Rather, effecting changes in genetic awareness and the ability to find the relevant information, when needed, may be better objectives. We therefore suggest that more research is undertaken to address whether, in addition to educational programmes aiming for high impact on organizational change and health gain, sustained change in practice behaviour and use of acquired genetics competences, provision of resources to supply 'just in time' information, decision aids and other clinical tools that are accessible during the clinical encounter contribute to changing primary care practice and genetic care improvement.

Additional files
Additional file 1: Papers read in full text but not included in the review. Abbreviations ELSI, ethical, legal and social issues; GP, general practitioner; OR, odds ratio; PICOS, patient population or problem, intervention (treatment/test), comparison (group or treatment), outcomes, and setting; PRISMA, preferred reporting items for systematic reviews and meta-analyses; RCT, randomised controlled trial; SD, standard deviation; UK, United Kingdom; US, United States.

Funding
The funding for this study was provided as part of the Gen-Equip project by the Erasmus + programme. The material in this paper reflects only the authors' views and the European Commission and Ecorys UK are not responsible for any use that may be made of the information it contains.

Availability of data and materials
All data were extracted from published sources listed in Table 3.