Article Text

Clinical frailty and performance scale translation in palliative care: scoping review
  1. Felicity Dewhurst1,2,
  2. Daniel Stow1,
  3. Paul Paes3,4,
  4. Katherine Frew3 and
  5. Barbara Hanratty1
  1. 1Population Health Sciences, Newcastle University, Newcastle upon Tyne, UK
  2. 2Palliative Medicine, St Oswald's Hospice, Newcastle upon Tyne, UK
  3. 3Palliative Medicine, Northumbria Healthcare NHS Foundation Trust, North Shields, UK
  4. 4School of Medical Education, Newcastle University, Newcastle upon Tyne, UK
  1. Correspondence to Dr Felicity Dewhurst, Population Health Sciences, Newcastle University, Newcastle upon Tyne, UK; drfelicitywerrett{at}


Background Frailty is associated with advancing age and increases the risk of adverse outcomes and death. Routine assessment of frailty is becoming more common in a number of healthcare settings, but not in palliative care, where performance scales (eg, the Australia-modified Karnofsky Performance Status Scale (AKPS)) are more commonly employed. A shared understanding of performance and frailty measures could aid interspecialty collaboration in both end-of-life care research and clinical practice.

Aims To identify and synthesise evidence comparing measures of performance routinely collected in palliative care with the Clinical Frailty Scale (CFS), and create a conversion chart to support interspecialty communication.

Methods A scoping literature review with comprehensive searches of PubMed, Web of Science, Ovid SP, the Cochrane Library and reference lists. Eligible articles compared the CFS with the AKPS, Palliative Performance Scale (PPS), Karnofsky Performance Scale or Eastern Cooperative Oncology Group Performance Status or compared these performance scales, in patients aged >18 in any setting.

Results Searches retrieved 3124 articles. Two articles directly compared CFS to the PPS. Thirteen studies translated between different performance scores, facilitating subsequent conversion to CFS, specifically: AKPS/PPS 10/20=very severe frailty, AKPS/PPS 30=severe frailty, AKPS/PPS 40/50=moderate frailty, AKPS/PPS60=mild frailty.

Conclusion We present a tool for converting between the CFS and performance measures commonly used in palliative care. A small number of studies provided evidence for the direct translation between CFS and the PPS. Therefore, more primary evidence is needed from a wider range of population settings, and performance measures to support this conversion.

  • Clinical decisions
  • Communication
  • Prognosis
  • Supportive care
  • Clinical assessment

Statistics from

Request Permissions

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.

Key messages

What was already known?

  • A growing number of people are dying with frailty and multiple long-term conditions, but most palliative care services are configured to support the needs of people dying with cancer.

  • To ensure equitable access to palliative care, clinicians must be able to interact with specialties outside palliative care including general and geriatric medicine where the importance of frailty is increasingly recognised.

What are the new findings?

  • We have combined evidence from two studies that mapped the Palliative Performance Scale (PPS) to the Clinical Frailty Scale (CFS) with evidence from studies mapping between different palliative performance measures to create a simple conversion tool.

  • Equivalent scores are: Australian-modified Karnofsky Performance Status Scale (AKPS)/PPS 10 or 20=very severe frailty, AKPS/PPS 30=severe frailty, AKPS/PPS 40 or 50=moderate frailty, AKPS/PPS 60=mild frailty.

What is their significance?

a. Clinical

  • An agreed translation between performance and frailty scales could aid interspecialty communication and improve access to patient-centred palliative care.

b. Research

  • We identified a paucity of evidence: to support our proposed conversion. More primary studies are needed in a range of populations and settings and using robust methods to compare the CFS to performance scales.


The expertise and experience of both geriatric and palliative medicine physicians is required to provide optimal care for patients with frailty.1–3 The need for palliative medicine trainees to have frailty included in their training curriculum has been recently recognised.4 Consideration of the perceptions of key stakeholders, (particularly older people) is also fundamental to ensure that the recognition of frailty results in optimisation of care rather than the introduction of prejudice. Descriptors of frailty should not be solely medically based and understanding of meaning should transcend healthcare professionals. It is imperative that an individual’s holistic requirements are considered to enable a patient’s ‘level of frailty’ to be fully described and subsequently ensure their function and sense of well-being are appropriately optimised.5 6 A common language to describe these holistic needs may aid discussions between services, patients and families, and facilitate decision making to ensure timely provision of palliative care.

Large population studies have demonstrated that many poor health outcomes are more strongly associated with frailty than with chronological age. However, the tool used to measure frailty in these studies has varied.7 8 The Clinical Frailty Scale (CFS) was developed for the Canadian Study of Health and Ageing.9 It uses clinical judgement to evaluate specific domains including multiple long-term conditions, function and cognition to summarise the overall level and impact of frailty. The CFS was originally developed as a 7-point scale and was subsequently updated to a 9-point scale in 2007; 1 is very fit, with 8 being very severely frail and 9 being terminally ill (less than 6 months to live but not otherwise demonstrating signs of frailty).9 The use of category 9 may be contentious in palliative care given all patients are ‘terminally ill’, omission of this point on the scale has been suggested in this setting.5–7 Assessments can be reliably conducted by healthcare professionals from a range of professions and disciplines.9 10 The CFS is viewed as the gold standard measurement of frailty following a Comprehensive Geriatric Assessment by the National Institute for Health and Care Excellence, the British Geriatric Society and the National Health Service (NHS) specialised clinical frailty network. The CFS has been supported as the most appropriate scale for triage and individualisation of patient care, particularly as patients near the end of their lives.11 Frailty, measured with the CFS, has been used as a clinical decision-making tool for optimising patient care during the pandemic.12 Details of the features of the CFS are summarised in box 1.

Box 1

The Clinical Frailty Scale (CFS)’s key characteristics


  • The CFS was designed to summarise the results of a Comprehensive Geriatric Assessment and consequently facilitate interdisciplinary and multidisciplinary discussion.10 41 42

  • It evaluates specific domains including comorbidity, function and cognition to generate a frailty score ranging from 1 (very fit) to 8 (very severely frail).43

  • Category 9 denotes patients who are ‘terminally ill’, defined as having less than 6 months to live but not otherwise demonstrating signs of frailty. The use of this category may be contentious in palliative care given all patients are ‘terminally ill’.9 44

  • The CFS is used commonly in both research and clinical care.


  • The CFS is a clinical judgement-based frailty tool.9.


  • There has been a significant recent increase in the number of publications using the CFS, this likely reflects the increasing appreciation that frailty is an important construct worthy of consideration.10 45

  • Research has demonstrated that the CFS can be reliably performed by researchers and a variety of professionals.9

  • Researchers and clinicians value the ease and efficiency of the CFS while appreciating the way it combines clinical judgement and objective measurement.


  • When categorising patients as frail and not frail, a CFS score of 5 (mildly frail) is the most widely used cut-off.9

  • The CFS can be used for risk stratification, it is a significant predictor of outcomes, specifically, mortality, multiple long-term conditions, disability, functional decline, mobility, length of hospitalisation, readmission, institutionalisation, cognitive function and falls.9 45

  • Studies have demonstrated a significant association between CFS scores and other measures of frailty.9


  • Research has demonstrated that in almost 90% of cases increasing frailty measured using the CFS is significantly associated with an increasing risk of mortality.9


  • The CFS was originally developed as a 7-point scale but there is an updated 9-point version which is now favoured, this complicates the meta-analysis of data.9

  • The CFS has been used internationally, however, publications detailing its use predominantly originate from high income countries, specifically Canada and the UK.9

  • The CFS has been applied to a range of inpatient and outpatient populations. However, it has mostly been used in hospital settings (63%), particularly within geriatric, renal and general internal medicine, cardiology, intensive care and surgery. Information from other medical specialties and from the community is sparse.9

  • The CFS has not commonly been used to predict patient-oriented measures such as quality of life and this has been suggested as an important focus of future studies.9

  • It has been proposed that the CFS can do more than just predict outcomes. It may also be used to evaluate interventions’ ability to mitigate frailty and prevent and treat disease. Further research into the potential uses for the CFS is warranted.

  • One novel area may be whether routine measurement could improve end-of-life care for older adults. Rapid decline may trigger referral to specialist services in patients who previously have had limited access to palliative care.9 44

Frailty assessments are not routinely performed in palliative care. Instead, performance status is recorded using a variety of measures, the Karnofsky Performance Scale (KPS), the Eastern Cooperative Oncology Group Performance Status (ECOG)), Australia-modified Karnofsky Performance Status Scale (AKPS) and the Palliative Performance Scale (PPS), the latter two of which are now most commonly used in end-of-life care settings.

The KPS is the original performance scale produced over 70 years ago. Modifications to produce the other scales have aimed to more accurately reflect the health status of patients with life limiting illnesses.13 The KPS was originally modified to the Thorne Karnofsky Performance Status (TKPS) for use with home hospice patients, The TKPS avoids reference to location of care and had new descriptors for assessing the frequency of professional visits and the proportion of time spent in bed. The PPS and the AKPS were subsequent attempts to adapt performance assessments to the nuances of a palliative population. The PPS was first described in 1996 as a modification of the KPS is routinely used in America and Canada.14 Modification of the KPS and TKPS to AKPS was motivated in part to ensure its applicability in any care setting (KPS and TKPS are more relevant when considering hospital admissions and palliative care at home respectively). As a result, AKPS focuses on functional ability rather than place of care. AKPS has been adopted as part of The Palliative Care Outcomes Collaboration, Australia, and the Outcome Assessment and Complexity Collaborative, UK.15

The ECOG is a 40-year-old adaption of the KPS produced to reflect oncology patient’s suitability for treatments and clinical trials. A recent overview of the ECOG, highlighted some of its limitations (including inability to account for multiple long-term conditions, frailty or cognition) and recommended a move to routine use of the CFS to enhance patient assessment and care.16 However, performance assessments are an established language in specialist palliative care (SPC) and oncology, leading others to suggest that rather than being disregarded altogether, the evidence for the clinical value of the scales should be reviewed and expanded.17 Prior research on these performance measures link level of function to survival, however, the variation in which measure is used means the impact of this research is lessened. An accurate translation between the scales would enable published data to be used collectively.17 18 Details of the features of the Performance Scales are summarised in box 2.

Box 2

Performance Scales’ Key Characteristics


  • The Karnofsky Performance Scale (KPS), Palliative Performance Scale (PPS) and Australia-modified Karnofsky Performance Status Scale (AKPS) are linear scales from 100 (normally active, without evidence of disease) to 0 (dead) summarising ability to perform daily activities, and level of assistance required.

  • The PPS is a modification of the KPS and incorporates five observer-rated parameters (ambulation, activity and evidence of disease, self-care, intake and level of consciousness).46 47

  • The AKPS is also a modification of the KPS. It is said to be a less complex alternative, easier to use with each clinical encounter, as simple descriptors divide patients into categories by assessing performance across three domains (work, activity and self-care). The AKPS was modified to ensure its applicability in any care setting, particularly those specific to SPC, therefore, AKPS focuses on functional ability rather than place of care.18

  • The Eastern Cooperative Oncology Group Performance Status (ECOG) is a six-point scale ranging from 0 (fully active) to 5 (dead), assessing level of function, ambulation and capability to self-care.48.


  • The calculations of all the performance scales are subjective, they are easy to perform, taking only minutes to complete with no specific equipment being required.17


  • Inter-rater reliability is similar for all of the performance scales with evidence demonstrating that training of healthcare professionals is more important than clinical experience or rater profession.23 49

  • Patients can make valuable and reliable self-assessments using all scales as evidenced from the moderate to good degree of correlation of patient-assigned scores with those of healthcare professionals.24 28 50

  • The PPS has been shown to have good overall inter-rater reliability across all palliative care settings including outpatients.46 47

  • The KPS may provide greater inter-rater consistency across different oncology professionals compared with ECOG and PPS.35


  • The evidence of validity of the KPS is longstanding.51 It and the ECOG correlate well with functional and fatigue scores.30

  • The validity of the PPS is backed up by significant evidence. It can reliably be used for disease monitoring, care planning, resource allocation, communication with patients, families and other healthcare professionals and with research.14 47 52

  • AKPS is seen as an efficient and pragmatic way of summarising the performance of a patient and predicting appropriate interventions regardless of care location.18 53

  • The Outcome Assessment and Complexity Collaborative suite of measures includes the palliative care phase (stable, unstable, deteriorating, dying and deceased). The phase is predictive of the need for SPC services. When phase deteriorates AKPS has been shown to reduce by a median of 10. Together they are said to provide a more meaningful picture of the patient’s status, supporting the use of AKPS in SPC.12

  • AKPS is more reliable in patients with non-malignant disease and 53multiple long-term conditions than the other performance measures. It has been shown to be independently associated (p<0.05) with age, daily step count, forced expiratory volume in one second (FEV1), and forced vital capacity (FVC) in chronic obstructive pulmonary disease (COPD).


  • Evidence shows that the KPS, PPS and the ECOG delineate prognosis well in advanced cancer.25

  • The ECOG may be seen as a simpler way of discriminating patients with widely different prognosis and therefore is said to be appropriate for use when assessing suitability for oncological treatment33

  • The PPS appears to be a valid, reliable and useful tool with the ability to predict outcome and prognosis, irrespective of other clinical or demographic variation in patients admitted to palliative care units (PCU) in both in Canada and Australia.14 47 52 54–57Downing et al performed an international large data meta-analysis on 1808 patients from four original datasets to reveal a strong association between PPS and survival. The Kaplan-Meier survival curves show each PPS level as distinct, with a strong ordering effect in which higher PPS levels are associated with increased length of survival.54 The PPS sensitively identifies functional decline which is directly related to outcome (both death and discharge).14 54 58 59

  • Performance measured using AKPS is directly related to survival (the Spearman correlation coefficient of baseline AKPS and overall survival is 0.26, p<0.001) in both intermediate AKPS 50%–70% and lower AKPS 10%–40% ranges. Specifically, it is more predictive of survival and has better face validity at lower values than the KPS. AKPS has longitudinal test–retest reliability and trends consistent with the established palliative care trajectories of illness.18


  • The KPS has limited sensitivity and reliability at the lower end of the scale in patients with life limiting conditions particularly in those who choose to avoid hospital admissions and who use informal/family carers.60 It is also not reliable for the assessment of patients with neurological deficit where ECOG is said to perform better.32 The KPS has been validated in a younger patient cohort and less applicable to an older heterogeneous patient group.18

  • The ECOG It has not been tested in non-malignant palliative care populations.48 It is poorly discriminatory in those with moderate to high levels of disability.48 60 Poor reliability and agreement among observers is often higher when the overall patient population has a good performance status compared with when the group is more heterogeneous with respect to functional performance.60

  • In one Japanese study, the PPS was shown not to accurately predict survival in palliative care patients with a mean survival of 49 days.56

  • The research on the AKPS (although promising particularly in the context of non-malignant disease and multimorbidity) is sparse and therefore more work is required to prove reliability and validity in all patient conditions and settings.53


To identify and synthesise evidence comparing measures of performance routinely collected in palliative care with the CFS and create a conversion chart to support interspecialty communication.


This was a scoping review designed to map key concepts on frailty and performance scales. Scoping reviews are an increasingly common approach for seeking and mapping evidence. They are frequently used for the clarification of definitions and when the subject has theoretical divisions or is relevant across specialty boundaries. We used the scoping review methodology for these aforementioned reasons and because literature had not yet been comprehensively reviewed and was likely to be of a heterogeneous nature. It was felt to be an appropriate method to summarise and disseminate research findings and in this case to identify research gaps, and to make recommendations for the future research while aiding evidence-based clinical practice. The methods used in this scoping review are those recommended by Joanna Briggs Institute and as such this review has been conducted and reported to an appropriate standard to ensure robustness of the results.19

Systematic search strategy

This study was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.20 We systematically identified literature that compared the performance measures with the CFS and each other. Articles were identified by a comprehensive search of PubMed, Web of Science, Ovid SP (including Medline, Embase, Health Management Information Consortium and PsycINFO), the Cochrane Library from earliest index date to end December 2021, using tailored strategies (online supplemental table 1). Searches were supplemented with screening of references, bibliographies and citations and targeted searches of key author’s work. All searches limited to English Language.

Search terms

Synonyms were explored, and relevance ensured using the Boolean operators OR and AND, respectively. Search terms and their limitations were as follows; (Karnofsky OR AKPS OR PPS OR KPS OR Performance (Title, Abstract or Key word)) AND (“Clinical Frailty Scale” OR CFS (NOT Chronic Fatigue Syndrome) (Title, Abstract or Key word)); Karnofsky OR AKPS (Title); (Cancer OR Malignancy OR Oncology OR Failure (Title)) AND (Performance (Title)) AND (Compare OR Comparison OR Contrast (Title)). Frailty OR Palliative (Cochrane only). The search terms and results are demonstrated in detail in online supplemental table 1. The results annotated with a * were reviewed for inclusion. Priority was given to ensuring all relevant articles were included. As such the number of titles and abstracts reviewed were numerous and therefore only reviewed by one of the authors.

Article selection

Titles and abstracts were screened by a researcher experienced in SPC and geriatric medicine, including carrying out frailty and performance assessments (FD). Full-text screening of potentially eligible articles was used to determine eligibility. Article selection is shown in figure 1.

Figure 1

Preferred reporting items for reviews flow diagram. AKPS, Australia-modified Karnofsky Performance Status Scale; CFS, Clinical Frailty Scale; ECOG, Eastern Cooperative Oncology Group Performance Status; KPS, Karnofsky Performance Scale; PPS, Palliative Performance Scale.

Inclusion criteria

We considered an article to be eligible for this review if it compared the CFS with the AKPS, PPS, KPS or ECOG. We also included articles that compared one or more of these performance measures to each other in a palliative care setting where patients were described as having any life-limiting illness (malignant or non-malignant). Original peer reviewed articles published in English were included.

Exclusion criteria

We excluded reviews of articles, articles published only as abstracts, articles in which the eligible patients were under 18 and articles not published in English. Studies were not excluded based on specific diseases.

Data extraction

FD extracted the data from each of the identified articles. The following data was included; year of publication, country of origin, study setting, study population, performance and frailty measures compared, sample size and summary of findings. For each of the performance scales (AKPS, PPS, KPS, ECOG) a description of the tool, nature of assessment, author described feasibility, limitations and correlation with outcomes were also extracted.

Data synthesis

First, studies comparing CFS to any of the performance measures were reviewed. We summarised information on how scores related to prognosis, short-term and long-term outcomes, suitability for treatment and scale descriptors. This process was repeated for included studies that compared performance scales only. Together, this provided evidence for translation between a variety of performance scales and the CFS. We combined this synthesis to produce a translation tool between the performance scales and the CFS.

Study characteristics

Searches identified 3124 articles, of which 71 were accessed for full text screening (figure 1).

Mapping CFS to PPS

Two articles made direct comparisons between PPS and the CFS (table 1). Both studies were from North America (USA21 and Canada22) and considered patients in long term care21 and from an inpatient palliative care unit, chronic care, palliative referrals and outpatient geriatric clinics.22 Both articles21 22 compared the CFS to the PPS.

Table 1

Studies comparing frailty and performance scales

One study22 set out to directly map CFS scores to the PPS on the basis of clinical judgement. In this study, participants were scored on the PPS by a palliative care physician and palliative care nurse, and on the CFS by a geriatric physician and geriatric nurse (four scores per participant). The cross-scale pairings with the highest level of agreement (assessed via Cohen’s kappa on the mean of the physician and nurse scores for either scale) were used to generate the comparison tool. The values of PPS mapped to CFS were as follows 90=3, 70–80=4, 60=5, 40–50=6, 10–30=7. The authors reported the marginal hit rate (the proportion of people at each level of one scale that are correctly assigned to the same level by the other scale) for each level of the conversion tool, with ranges between 43% and 80% being correctly allocated.

The second study took CFS and PPS scores from 171 long term care residents and examined the association between these and two outcomes (hospital readmission and mortality). The basis of the comparison between CFS and PPS in this study was the similar strength (and direction) of the association between the scales and outcomes in the same patient group. Mean scores from this study across the different comparisons suggest the >6 CFS ≤40 PPS are equivalent. Patients with a 1-year survival of less than 10% had a CFS of >6 and a PPS <40. Patients who died within 1 year had a mean CFS of 7 and PPS 33.6.

Translating between PPS

Thirteen articles assessed the inter-operability of multiple PPS. Five were from Canada,23–27 three from Australia,18 28 29 two from Brazil,30 31 one from France,32 one from Italy33 and one from Spain.34 These 13 articles provided evidence to allow mapping between different performance scales. The details are described in detail in table 2.

Table 2

Studies comparing different palliative performance scales

We have extrapolated finding from all study’s findings of these studies have then been extrapolated to allow a conversion chart to be produced (table 3).

Table 3

Comparable scoring of the CFS, PPS, KPS, AKPS and ECOG


This scoping review systematically identified two papers that examined directly the relationship between CFS and the PPS. Further systematic searches found 13 studies directly comparing different measures of performance used in palliative care. By combining this information, we present a tool to support conversion between the CFS and a range of performance scales routinely used in palliative care, thus avoiding inaccuracies or discrepancies in conversion that may result from the use of clinical judgement alone.

The studies comparing CFS and the PPS directly did so on the basis of either statistical comparison of clinical judgements,22 or loosely on univariate association between the scales and prognosis or risk of hospital admission.21 These comparisons do not account for patient centred outcomes, nor do they usefully indicate patient needs beyond these metrics.

Evidence demonstrating the equivalent scores of PPS, KPS and AKPS, together with the scale descriptors, suggests that these scales can be interconverted relatively easily, facilitating subsequent translation to the CFS. However, given the broad divisions used in the ECOG, we suggest that it can be derived from the CFS, PPS, KPS and AKPS, but not vice versa.27 34 The ECOG may still be seen as a simpler way of discriminating between patients with widely different prognosis and therefore appropriately assessing suitability for oncological treatment.33 34 More primary studies are needed in a range of populations and settings to support our proposed conversion. Previous studies have demonstrated that translating between scores at lower performance levels may introduce a higher degree of error. As such, future research should include those with high levels of disability.35

The National Institute for Health and Care Research and the Care Quality Commission state that improved understanding of how to provide palliative care to the growing number of people living and dying with frailty is an international priority and a key strategic area for research.36 Current literature describes patients care needs and prognosis using a variety of performance and frailty scales. Palliative care services have likely been providing palliative and end-of-life care for frail patients for decades. However, in palliative care modelling patients are defined by their diagnosis and performance level rather than a frailty score. Frail patients with cancer get access to services whereas frail patients with a non-malignant diagnosis do not.37 Describing all patients in the same way may reduce the inequity.

If we agree that everyone is entitled to excellent palliative care regardless of diagnosis, the important questions are around timing including which are the most appropriate indicators of decline and what patient’s specific needs are. Referral of patients at the right time is a key consideration.38 The literature suggests that all those with life limiting illness are all living with a degree of frailty, regardless of age or diagnosis and that increasing frailty indicates that end of life is approaching.39 Malignant and non-malignant progressive, life-limiting diseases may behave similarly and share indicators of decline.40 Similarly the limited existing research suggests that individuals with non-malignant disease and multiple long-term conditions have a similar symptom burden to those with malignant disease.37 A conversion between performance and frailty scales may enable immediate expansion of the literature. Recent evidence has demonstrated that Palliative medicine physicians are unfamiliar with the definition of frailty and the opportunities and rationale for measuring it.4 A scale that is understood across all disciplines would likely aid decision making and access to appropriate care. It could reduce the issue of individual specialty areas working in silos. This is fundamental because we must acknowledge that when considering the management of patients with frailty, primary and social care and geriatric and palliative medicine all have a part to play.

Strengths and limitations

This review was performed as a scoping review in anticipation of identifying a paucity of literature. The searches were carried out systematically with clear inclusion and exclusion criteria but quality appraisal was not part of the design. Given the small amount of literature identified we were unwilling to exclude any previously published evidence on quality metrics.

We have systematically identified a gap in the literature and given specific recommendations on how to close that gap. The small number of studies (or ‘evidence of a lack of evidence’) is a key and important finding.


Improving palliative care for ageing populations is an international priority. A shared understanding of performance and frailty could aid interspecialty collaboration in both end-of-life care research and clinical practice. A small number of studies provided evidence for translation between CFS and performance scales routinely used in palliative care. More primary evidence is needed from a range of populations and settings to support this conversion. An agreed translation between performance and frailty scales could improve access to patient-centred palliative care.

Ethics statements

Patient consent for publication


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.


  • Contributors All authors: FD, DS, BH, PP and KF met the following conditions (1) Made a substantial contribution to the concept or design of the work, or acquisition, analysis, or interpretation of data, (2) Drafted the article or revised it critically for important intellectual content, (3) Approved the version to be published and (4) Have participated sufficiently in the work to take public responsibility for appropriate portions of the content. FD, DS, BH, PP and KF were responsible for the designing and conducting the review. FD produced the initial draft of the manuscript. All authors critically reviewed the manuscript and contributed to redrafting. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work in ensuring that questions relating to the accuracy or the integrity of the work are appropriately investigated and resolved.

  • 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 None declared.

  • Provenance and peer review Not commissioned; internally 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.