Objective To evaluate current evidence of the effect of specialised nutritional interventions on nutritional status, survival, quality of life and measures of functionality in patients with incurable cancer.
Methods Systematic literature review according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines using PubMed/MEDLINE, EMBASE, Scopus, LILACS and Cochrane Library databases. Clinical studies that evaluated different specialised nutritional interventions, such as nutritional counselling, oral nutritional supplementation (ONS), enteral nutrition (EN) and parenteral nutrition (PN), were eligible. Only studies classified as being of high methodological quality (ie, low or moderate risk of bias) were included.
Results A total of 22 studies reporting on 2448 patients were deemed eligible. Five types of specialised nutrition were observed: mixed (multimodal nature, ie, dietary counseling, ONS, physical activity and/or drugs) (n=12), ONS (n=5), PN (n=3), EN (n=1) and multidisciplinary team counselling (n=1). Benefits of any kind from the interventions were reported in 14 (63.6%) studies, mainly resulting from mixed intervention. Nutritional status improved in 12 (60.0%) of 20 studies and quality of life improved in eight (50.0%) of 16 studies. Few studies have evaluated the influence of nutritional interventions on survival and measure of functionality, and have not shown improvement in these outcomes.
Conclusion Despite the limited evidence, specialised nutritional interventions can yield positive effects for patients with incurable cancer, mainly in their nutritional status and quality of life.
- clinical decisions
- complementary therapy
- quality of life
- supportive care
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What was already known?
Previous studies have demonstrated inconclusive evidence of effects of nutritional interventions in patients with incurable cancer.
What are the new findings?
Although limited, current evidence suggests that specialised nutritional interventions have a beneficial impact on nutritional status and quality of life in patients with incurable cancer.
There is insufficient evidence to demonstrate the impact of nutritional interventions on the survival and functionality measures of patients with incurable cancer.
What is their significance?
Clinical: Nutritional interventions, particularly those of a multimodal nature, may have a positive effect on outcomes in patients with incurable cancer.
Research: This systematic review highlights the necessity to provide well-designed randomised controlled trials, focusing on the relevance of specialised nutrition in the context of multimodal interventions in patients with incurable cancer.
Cancer is a complex multifactorial disease and in low-income and middle-income countries its diagnosis usually comes late, when it is already at an advanced stage, with distant metastases and/or life expectancy of just a few months.1 2 As the disease progresses, nutritional status (NS) deteriorates due to increased tumour-host interactions in conjunction with other factors, such as tumour site, anticancer therapies and nutritional impact symptoms like anorexia, nausea and vomiting.3
Patients may already have malnutrition when they are diagnosed. The prevalence of malnutrition in patients with incurable cancer is high, reaching more than 80% according to the assessment method, tumour site and treatment characteristics.4 5 A poorer NS is related to unfavourable outcomes such as lower performance status (PS), longer length of stay and worse quality of life (QoL), and has recognised prognostic value.4–9
In this context, the main nutritional guidelines for cancer patients recommend assessing their NS and food intake to enable adequate nutritional care.10 11 According to these guidelines, patients with inadequate oral food intake should receive nutritional counselling. If this does not improve the situation, oral nutritional supplements (ONS), enteral nutrition (EN) and even parenteral nutrition (PN) (supplemental or total) should be considered on a case-by-case basis.2 12 13
In patients with early-stage cancer undergoing disease-modifying therapy, the recommendations in the literature are more consistent as to the potential benefits of specialised nutritional interventions on clinical outcomes, symptom control, tolerance to anticancer treatment and QoL.14 15 However, in patients with metastatic cancer or receiving exclusively palliative care, there is as yet no consensus on the benefits of such interventions.16
A systematic review by Blackwood et al 2 about the effect of oral dietary interventions in patients with incurable cancer showed moderate to quality evidence in favour of such interventions, although the clinical effect sizes were considered small. In its guidelines on cancer cachexia, the American Society of Clinical Oncology7 the strength of recommendation is moderately in favour of nutritional counselling to address cancer cachexia, and the European Society of Medical Oncology guidelines report that the benefits of nutritional and metabolic interventions in patients with cachexia and advanced disease are uncertain.17 Thus, although nutritional interventions are generally part of the management strategy for patients with incurable cancer, there is still a shortage of evidence based on comparisons of different specialised nutritional interventions and their potential benefits for such patients.
Therefore, the aim of this systematic review is to evaluate current evidence on the effects of specialised nutritional interventions in patients with incurable cancer. The effects of interest are: survival, NS (body weight, body composition or Patient-Generated Subjective Global Assessment (PG-SGA)), measures of functionality (handgrip strength (HGS) or PS) and QoL.
Literature search and study selection
This systematic review was based on the following question: ‘In patients with incurable cancer, are specialised nutritional interventions associated with improved outcomes (survival, NS, functionality or QoL)?’ (online supplemental table 1). The research protocol was registered in PROSPERO (number: CRD42021288308) and the review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology.18 19 We also used the Assessment of Multiple Systematic Reviews20 measurement tool to ensure the methodological quality of our review.
The search was carried out in November 2021 and updated in May 2022 in the PubMed/MEDLINE, EMBASE, Scopus, LILACS and Cochrane Library databases. No restrictions were imposed on the language, publication date or sample size. The snowballing method of manually tracking publications of interest from the reference lists of selected articles and other review articles was used as a secondary search strategy. The searches of all the databases were based on MeSH controlled vocabulary as described in online supplemental table 2.
All the studies retrieved in the searches were recorded in a web tool (Rayyan, Qatar Computing Research Institute, available at http://rayyan.qcri.org), which helped expedite the initial screening of the abstracts and titles.21 All the titles and abstracts were read and duplicates were removed. Reading and selection was carried out by two researchers, independently. Whenever the abstract did not provide enough information to establish whether an article should be excluded, it was read in full. All the potentially eligible studies were then retrieved in full and appraised independently by two authors. Disagreements were discussed with a third author until consensus was reached.
The inclusion criteria for the studies were: availability of abstract online; original research; clinical trials (randomised controlled and non‐controlled); performed on humans; participants >18 years of age; evaluating the effect of any specialised nutritional intervention. There were no restrictions on the frequency, duration or intensity of interventions. Studies including inpatients or outpatients in in-home palliative care from any country were included.
Incurable cancer was deﬁned as metastatic cancer (histological, cytological or radiologic evidence) or locally recurrent advanced cancer being treated with palliative intent, meaning that some antineoplastic treatment could be administered to prolong life and/or alleviate symptoms. To be eligible, the studies also had to have defined outcome measures such as survival or mortality, and/or NS (body weight, body composition or PG-SGA), functionality (HGS or PS) and/or QoL.
Exclusion criteria were: studies involving patients undergoing active anticancer treatment with curative intent (or studies where this was not clear), nutritional interventions with exclusively only one type of nutrient (eg, proteins, amino or fatty acids) or vitamins/minerals and studies of low methodological quality (eg, with a high risk of bias). All the excluded articles were reviewed by two authors to double-check that they did not meet the eligibility criteria.
Methodological quality of studies
Two independent reviewers assessed the studies’ methodological quality, then a third reviewer completed the assessment by elucidating possible conflicts. The Cochrane risk-of-bias tool for randomised trials (RoB2)22 and risk-of-bias tool for non‐randomised studies (ROBINS‐I)23 were used as recommended.24 For each study and outcome domain, a judgement was assigned in terms of risk of bias and concerns. Risk of bias visualisation (robvis, Cochrane, UK) was used for the graphical summary of the results.25
In addition, the studies’ funding sources were considered, as this information is important because industry-funded studies sometimes favour sponsored products.26 27
Two researchers extracted the data independently using a form designed especially to record the following information: author, year of publication, study design, country of publication, sample size, characteristics of participants (eg, age and cancer type), aims, follow-up, length of recruitment period, characteristics of the nutritional intervention (eg, nutritional counselling, ONS, EN, PN or mixed interventions), route of administration and composition of the formula/product used, outcomes studied (measurement tools, measured and reported time points, baseline and follow‐up data) and main findings. Measures of effect included: change from baseline to follow‐up within the intervention group (IG), and differences between the IG and control group (CG) at baseline and at follow‐up.
The literature search identified 4714 records of interest; 549 remained after reading the titles and abstracts, 23 were read in full and 10 were considered eligible. An examination of the retained articles’ reference lists and other review articles recovered 14 more studies. After evaluation, two of these were excluded due to low methodological quality,28 29 leaving a total of 22 articles,30–51 (figure 1) most of which were judged to have a low risk of bias (figure 2).
Collectively, 22 studies reported on 2448 patients aged from 44.047 to 74.251 years, mostly male (n=12)30 36 37 39 40 42–45 48–50 and predominantly with gastrointestinal tract cancer (n=20).30–32 34–46 48–51 The publication years ranged from 2003 to 2021, with the majority (n=12)33–35 37–39 46–51 dating from the last decade, and half carried out in Europe (n=11)30 31 36 38 41 42 44 45 48–50 (table 1). Only three of the studies were non-controlled clinical trials.35 36 43 Of the randomised trials, five were controlled with calorie-controlled diet,38 45 48 49 51 five with nutritional counselling,33 39 42 44 47 two with ONS without eicosapentaenoic acid (EPA),30 31 one with regular diet and counselling,50 one with ONS and abstention from alcohol,34 one with medication,41 one with placebo (ONS plus medication),40 one with dietary adjustments,32 one with liquid enteral product46 and one with intravenous fluid.37 Ten of the articles received industry funding.30–33 36 38 43 45 46 48
Specialised nutritional interventions
Five types of interventions were observed (mixed (n=12),40–51 ONS (n=5),30–34 PN (n=3),36–38 EN (n=1)34 and multidisciplinary team counselling (n=1)),39 with lengths of intervention ranging from 334 43 to 3536 weeks. The mixed interventions were established with nutritional counselling and/or ONS with or without omega-3 and/or EN and/or PN and/or physical activity counselling and/or drugs.40–51 In the ONS interventions, three studies used omega-330–32 and one used wine34 (tables 2 and 3).
Benefits of any kind from the interventions were reported in 14 (63.6%) studies,32 33 35 36 39 41–44 47–51 mainly of mixed intervention, in patients with incurable cancer. NS improved in 12 (60.0%)30 31 33 37 39–42 45 46 48 of 20 studies and QoL improved in eight (50.0%)33 35 42–44 47 49 51 of 16 studies (figure 3).
Eleven of the studies had survival as an outcome.30 33 34 37 38 40 41 45 46 48 50 The median survival of the patients ranged from 7 days46 to 21.7 months50 in the IGs vs 8 days37 to 16 months50 in the CGs. Only two41 50 studies showed any benefit in terms of survival, having performed mixed interventions (tables 2 and 3).
Previous NS was described in 12 of the studies, finding the patients to be malnourished and/or cachectic before the nutritional intervention.30 31 34–36 38 40 42 43 47 48 51 NS was the most frequently outcome observed among the studies (n=20),30–36 38–45 47–51 being evaluated in terms of body weight (n=19),30–35 38–45 47–51 body composition (n=13)30–33 35 36 41–43 47–50 and PG-SGA (n=5).43 44 47 48 51 Nine studies showed a statistically significant increase32 39 42–44 48 50 in or maintenance33 35 of body weight in the IGs. Four of the studies that investigated PG-SGA43 44 47 51 reported no significant difference in the PG-SGA score between the IGs and CGs (tables 2 and 3).
Nine studies assessed body composition by bioelectrical impedance analysis30–33 36 42 43 47 49 and the majority (n=5)33 36 42 43 47 showed some benefit from specialised nutrition interventions. Breitkreut et al 42 found that body cell mass was preserved and fat-free mass was higher in the IG, with significant intergroup differences. Kapoor et al 47 reported increased body fat in the IG and a significant difference compared with the CG. Pelzer et al 36 observed that the median extracellular mass/body cell mass index decreased and median phase angle increased by 10% in the IG. Differently, Uster et al 49 did not observe differences in phase angle. Evaluating lean body mass (LBM), Sánchez-Lara et al 33 observed an increase in the IG compared with the CG, while Fearon et al,30 Moses et al 31 and Bauer et al 32 found no statistically significant differences between the groups. Read et al 43 found no change in the IG.
Four studies assessed body composition by gold-standard methods.35 41 48 50 Using CT, Solheim et al 48 observed no differences in muscle mass and van der Werf et al 50 found no differences in skeletal muscle area and muscle density between the IG and CG. Using dual energy X-ray absorptiometry, Gresham et al 35 observed an increase in LBM and appendicular lean mass, but found no statistically significant difference in body fat after intervention. Lundholm et al 41 did not find any statistically significant differences in LBM between the groups, but did observe increased body fat in the IG (tables 2 and 3).
Quality of life
In 15 studies that assessed QoL,30 32–35 38 42–47 49–51 Core Quality of Life questionnaire (QLQ-C30) of the European Organisation for Research and Treatment of Cancer was the most used tool (n=9).30 32 33 35 44 45 47 49 51 Most of the studies (n=8)33 35 42–44 47 49 51 demonstrated benefits of nutritional intervention in QoL. Significant improvements were found in global health,33 34 appetite loss,33 35 47 fatigue,33 44 47 role function,35 social function,44 nausea and vomiting,49 depression,35 pain,35 leisure activities,42 energy levels43 and psychological condition42 in the IGs. Molassiotis et al 51 also found improvement in Functional Assessment of Anorexia/Cachexia Treatment (FAACT) scores in the IG throughout the study. As for the CGs, these showed worsening in the domains nausea and vomiting,33 49 neuropathy,33 global health47 and social function47 (tables 2 and 3).
This systematic review examined the impact of different nutritional strategies on clinical outcomes. The results suggest that specialised nutritional interventions, especially of a multimodal nature (ie, dietary counselling, ONS, physical activity counselling and/or drugs), can have some positive effect on the NS and QoL of patients with incurable cancer, but there is insufficient evidence to demonstrate the impact of nutritional interventions on the survival and functionality measures In this population, where poor NS often results in decreased QoL, PS and survival, maintaining NS can be important, even in the context of incurable disease.4–9
Despite the scarce and heterogeneous evidence and the multiplicity of parameters and interventions evaluated, making it difficult to compare the studies, this systematic review provides a useful overview of the role of specialised nutritional care for patients with incurable cancer. It should be noted that the evidence already reported in the literature on patients at the earlier stages of the disease is also unclear when it comes to the impact specialised nutritional intervention can have on the target outcomes especially in terms of survival.14 15
There is scant evidence in the literature on the effect of specialised nutritional interventions on the survival of patients with incurable cancer. Among the studies selected for this review, only two (18.2%) clinical trials41 50 observed statistical differences in survival after mixed intervention. Although NS is known to be associated with worse prognosis and increased risk of mortality, previous systematic reviews16 52 53 have failed to identify any positive effect of nutritional intervention on survival, regardless of the stage of the cancer. Two previous meta-analyses evaluated the effects of ONS and nutritional counselling on malnourished patients with cancer receiving active treatment or palliative care and found no improvement in survival.16 52 In another recent systematic review by Hamaker et al,53 the six studies that assessed nutritional support alone or in combination with dietary counselling in malnourished patients found no benefit in terms of survival, and only one of four found any benefit in terms of survival among patients receiving nutritional counselling vis-a-vis those receiving regular care.
Cotogni et al,54 who compared malnourished cancer patients in palliative care receiving home PN or artificial hydration, found that median overall survival was significantly longer in the patients receiving home PN (4.3 vs 1.5 months, p<0.001). Amano et al,55 in a multicentre prospective cohort study showed significant differences in the survival of the patients from the EN (n=730), PN (n=190), when compared with CG (n=533) (log-rank p<0.001). In the multivariate-adjusted model, a significantly lower risk of mortality was observed in the EN (hazard ratio (HR) 0.43, 95% confidence interval (CI) 0.37 to 0.49) and PN (HR 0.52, 95% CI 0.44 to 0.62) groups than in the CG. In addition, Ruggeri et al,56 in a multicentre retrospective study with patients in exclusive palliative care, found that those who started EN or PN with precachexia had longer survival than those who started nutritional support with cachexia or refractory cachexia stages.
Twelve of the studies included in our systematic review demonstrated positive benefits on NS of specialised nutritional intervention. There is evidence that the ‘anabolic potential’ of patients with incurable cancer can be enhanced under certain conditions.57 Therefore, it is important that all such patients be evaluated for their current NS and prognosis in order to provide timely decision-making about individualised nutritional interventions focused on maintenance or improved of NS.14 15
From the studies reviewed here, it is not clear under what patient conditions the respective interventions were made, which may have influenced the effects shown in the results. For example, patients with substantial weight loss may also have refractory cachexia, and therefore, limited anabolic potential. These findings suggest the importance of assessing NS at the time of initiation of nutritional intervention, irrespective of the stage of the disease and also highlight the importance of standardising measures to assess NS outcomes in clinical studies.
Body weight was the most common parameter used to assess the effect of nutritional interventions, which were found to yield positive benefits in 9 of these 19 studies. The diverging results of the studies that evaluated body weight can be explained by differences in nutritional intervention, sample size, study duration and methodological quality.30–35 38–45 47–50 In the meta-analysis performed by van der Schueren et al,14 positive effects on body weight were observed in patients receiving ONS during chemo(radio)therapy compared with the CG. In Lee et al’s58 study, nutritional counselling with or without ONS appeared to be more effective than ONS alone in improving nutritional outcomes (body weight, PG-SGA and BMI).
Although PG-SGA is a complete, non-invasive nutritional assessment instrument, addressing aspects such as weight change, food intake, symptoms of NS impact, and PS,59 it was not widely used in the studies and most of those that did use it found no improvement in the patients’ scores. We believe that these results may be due to the limited number of participants in these studies,43 44 adherence to the intervention,44 45 baseline NS of the participants40 and length of intervention.43 48 51
Some studies demonstrated a greater increase in muscle mass in the IG.33 35 44 In the studies carried out by Breitkreutz et al 42 and Sánchez-Lara et al,33 muscle mass increase can be explained by the use of fat-enriched ONS or ONS including EPA. In ONS, formulas enriched with omega-3 fatty acids may provide some positive results in terms of weight gain, muscle mass, nutritional intake and QoL.2 EPA can help modulate body composition due to its anti-inflammatory effect, which results in reduced proteolysis and increased muscle synthesis. Pappalardo et al 60 found EPA supplementation to yield positive effects on LBM compared with standard supplementation. Muscle gain has been linked to stable disease and may represent a response to successful cancer therapy. This highlights the importance of recognising therapeutic windows for intervention before refractory cachexia is sets in.61 Studies that did not show statistically significant differences in NS employed different nutritional interventions, such as the simultaneous use of medication and ONS, the use of ONS enriched with EPA and dietary compliance prior to the use of ONS.31 32 41
Functionality measures are often used to reflect the impact of cancer on patients and their ability to perform everyday activities,62 and are related to NS.63 64 Few studies included in our review evaluated this outcome and all of them reported no beneficial effects of nutritional interventions on functionality measures, even those that showed improvement in NS.35 48 50 van der Werf et al 50 observed increased survival in the IG without any improvement in HGS.
These results can be explained by the fact that in advanced cancer patients with significant loss of body weight or muscle mass, function begins to decline at an increased rate to the point that even the most effective interventions are unable to have a favourable impact on muscle function, since the recovery of body weight and muscle mass long precedes functional recovery, making it unlikely for a synchronous increase in muscle mass and function to be observed.65 Another potential explanation for these findings is that the mechanisms of muscle function are related not only to improved NS or increased muscle mass, but also to the ‘quality’ of the muscle.63 64
These facts have been used to explain why, for example, recent clinical trials that have succeeded in demonstrating increases in muscle mass have been unable to show associated increases in patient physical function.66 Thus, although maintaining or improving functionality for as long as possible is one of the main goals of care for patients with incurable cancer,7 17 evidence suggests that the complexity of this effect makes it hard to obtain.
Quality of life
Patients with incurable cancer often present nutritional impact symptoms, decline in performance status, deterioration of NS, and consequently, impaired QoL due to disease progression and previous anticancer therapies.6 60 Maintaining or improving QoL in such patients is fundamental as it is one of the primary goals of care in such cases.1 58
The findings of this systematic review showed differing results regarding the effects of specialised nutritional interventions on QoL. Eight of the 16 studies that investigated QoL reporting positive effects, and six of these evaluated mixed interventions. The mixed-intervention studies that showed better QoL results were mainly those that combined nutritional counselling with ONS. In fact, as QoL is a subjective multidimensional construct,1 the adoption of a multipronged approach could be more successful because of the different dimensions involved.
Two studies with patients with accessory feeding routes (PN or EN) had different results. Gresham et al 35 found improved QoL in the EN group, but Bouleuc et al 38 saw better deterioration-free survival for functionality in the CG. Therefore, the question as to whether EN benefits QoL remains unanswered, probably because there is no consensus on the best route of administration or when to start and stop the intervention. In such cases, individual patient assessment and multiprofessional discussion are recommended to ensure conduct that benefits the patient’s life and does not cause harm. Wiegert et al 67 compared the use of nutritional interventions in patients with incurable cancer in palliative care and in non-palliative care and observed that the prevalence of EN and PN was similar in both settings, but that their use decreased significantly with the approach of death in palliative care.
As for the four studies that made exclusive use of ONS, only Sánchez-Lara et al 33 found ONS containing EPA to benefit QoL, with decreased fatigue and loss of appetite in e patients with GI cancer. However, there are reports in the literature of satisfactory improvement of QoL after such interventions. The systematic review by Lee et al 58 found improvements in QoL scores after the use of ONS that were related to improved nutritional intake or status. Blackwood et al 2 also report improved QoL parameters after nutrition intervention, especially ONS.
These conflicting results can be explained by several factors that may not be exclusively associated with the nutritional intervention received by the patient, but which influence general QoL, such as physiological distress due to cancer treatment and tumour-related symptoms.
Limitations and strengths
Limitations of this review include the lack of homogeneity across the studies reviewed, such as type of nutritional intervention, follow-up, adherence to intervention and outcome assessments, precluding a meta-analysis. The results of each study should therefore be interpreted with care. As strengths, it is worth noting that the methodological assessment instruments used were able to validate and qualify the inclusion of studies of high methodological quality (ie, with either low or moderate risk of bias). There are still few studies dedicated to evaluating nutritional support and its results in patients with incurable cancer, and some fundamental questions remain, such as the ideal time for these interventions to be implemented (eg, cachexia stage, disease stage and length of anticancer therapy) and the most appropriate type of intervention and outcome measures.
Finally, although controlled trials are needed, the challenges of clinical trial recruitment in patients with progressive incurable disease or in ‘palliative care’ are well documented. Largely, these focus on ethical and logistical issues. Ethical issues related to the burden and intrusiveness of study measures on the participants, concerns regarding randomisation and ethics committees. Logistical issues include a lack of research infrastructure and funding, and collaborative centres.68 Furthermore, palliative care patients have an expected trajectory of deterioration and death that may complicate the ethical and enrolment issues in this population and increase the risk of low-quality clinical trials.
Despite the limited evidence it brought to light, this systematic review suggests that specialised nutritional interventions can yield positive effects on NS and QoL in patients with incurable cancer. Accordingly, paradigm shifts are necessary to address the complexity of providing support for such patients, adjusting the goals of care. The lack of consistent clinical evidence is still a challenge and further studies are needed to elucidate which patients with incurable cancer could benefit from nutritional intervention and how to best deliver such interventions. The experiences of patients and caregivers are critical and probably the most important proxy of effect. These are often not paramount outcomes in clinical trials, but are central to what is done clinically. Thus, patients’ and caregivers’ views are needed as part of t the main outcome and clinically relevant practice.
Patient consent for publication
Contributors LCdO, LC-L and EVMW contributed to study design. LCdO, LC-L EVMW, GdCC, NFdS, RdS-S and TdS designed the protocol. GdCC, TdS and EVMW conducted the database searches. RS-S and TdS performed the screening and quality assessment. GdCC, NFdS, TSMF and DAdS performed the data extraction and analyzed the data. GdCC, NFdS, RS-S, TdS, TSMF, DAdS and CdMS performed the analysis of the results. LCdO, LC-L and EVMW reviewed the paper and had overall responsibility for the final content. All the authors contributed to the synthesis and have approved the paper. LCdO is the author acting as guarantor and is the corresponding author.
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.