Article Text

Original research
Healthcare utilisation and quality of life according to atrial fibrillation burden, episode frequency and duration
  1. Maria Hee Jung Park Frausing1,2,
  2. Martijn Van De Lande3,
  3. Dominik Linz4,5,
  4. Harry J G M Crijns4,
  5. Robert G Tieleman6,
  6. Martin E W Hemels7,8,
  7. Mirko De Melis9,
  8. Ulrich Schotten10,
  9. Mads Brix Kronborg1,2,
  10. Jens C Nielsen1,2,
  11. Isabelle Van Gelder3,
  12. Michiel Rienstra3
  1. 1 Department of Clinical Medicine, Aarhus University, Aarhus, Denmark
  2. 2 Department of Cardiology, Aarhus University Hospital, Aarhus N, Denmark
  3. 3 Department of Cardiology, University Medical Center Groningen, Groningen, The Netherlands
  4. 4 Cardiovascular Research Institute Maastricht (CARIM), Maastricht Universitair Medisch Centrum+, Maastricht, The Netherlands
  5. 5 Department of Biomedical Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
  6. 6 Department of Cardiology, Martini Hospital Groningen, Groningen, The Netherlands
  7. 7 Department of Cardiology, Rijnstate Ziekenhuis Arnhem, Arnhem, The Netherlands
  8. 8 Department of Cardiology, Radboud University Medical Center, Nijmegen, The Netherlands
  9. 9 Cardiovascular Diagnostics, Medtronic Bakken Research Center, Maastricht, The Netherlands
  10. 10 Department of Physiology, Maastricht University Medical Centre, Maastricht, The Netherlands
  1. Correspondence to Dr Maria Hee Jung Park Frausing, Department of Cardiology, Aarhus University Hospital, Aarhus N, Denmark; mariseje{at}rm.dk

Abstract

Background We aimed to evaluate the association between atrial fibrillation (AF) burden, duration and number of episodes with healthcare utilisation and quality of life in patients with early paroxysmal AF without a history of AF.

Methods In this observational cohort study, we included 417 patients with paroxysmal AF from the Reappraisal of Atrial Fibrillation: interaction between hyperCoagulability, Electrical remodelling and Vascular destabilisation in the progression of AF (RACE V) Study. Patients were monitored with an insertable cardiac monitor for 1 year. Outcomes collected were healthcare utilisation, and quality of life assessed using the Atrial Fibrillation Severity Scale and EuroQol EQ-5D-5L questionnaires.

Results During 1 year of follow-up, 63 973 AF episodes were detected in 353 (85%) patients. The median AF burden was 0.7% (IQR 0.1–4.0%). AF ablation was performed more frequently in patients with intermediate-to-high AF burdens (>0.2%) (16.2% vs 5.9%, p=0.01) and longer AF episode duration (>1 hour) (15.8% vs 2.0%, p=0.01), whereas cardioversions were more frequent in patients with longer episode duration (>1 hour) (9.5% vs 0%, p=0.04) and intermediate (0.2–1.9%) (but not high) AF burdens (13.6% vs 4.2%, p=0.01). Patients with many episodes (>147) reported higher symptom severity (p=0.001). No differences in symptom severity nor in EQ-5D-5L scores according to AF burden or duration were observed.

Conclusion In patients with early paroxysmal AF, higher AF burden and longer episode duration were associated with increased rates of healthcare utilisation but not with symptoms and quality of life. Patients with a higher number of episodes experienced more severe symptoms.

Trial registration number NCT02726698.

  • atrial fibrillation

Data availability statement

Data are available upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

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.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Paroxysmal atrial fibrillation (AF) is a highly heterogeneous cardiac arrhythmia with variable clinical presentation.

WHAT THIS STUDY ADDS

  • In patients with paroxysmal AF, burden, episode duration and number of episodes were highly variable. AF ablation and cardioversion occurred more frequently in patients with higher AF burdens and longer AF episode duration. No differences in symptoms or quality of life according to AF burden or episode duration were observed, but patients with frequent episodes reported more symptoms.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • Results from this study indicate that AF burden, duration and number of episodes could become important metrics in AF clinical and prognostic assessments.

Introduction

Atrial fibrillation (AF) is typically categorised into paroxysmal self-terminating AF, persistent non-self-terminating AF or permanent AF.1 It has long been known that outcomes with more sustained forms of AF are worse than with paroxysmal AF.2 Only recently, it became evident that paroxysmal self-terminating AF is a highly diverse cardiac arrhythmia that varies considerably in burden, duration and number of episodes.3 4 In the clinical setting, it is relevant to determine which parameter—burden, duration or number of episodes—is more closely associated with outcomes and healthcare utilisation. In patients with paroxysmal AF monitored for 14 days and post-ablation, AF burden has been proposed as a more sensitive marker of AF-associated patient outcomes as opposed to time to first episode,5 6 although other studies suggest that episode duration is a more important parameter.7 However, the clinical relevance of AF assessment based on burden, duration or number of episodes in patients with paroxysmal AF who have not undergone ablation and thus are in an earlier phase of their arrhythmia has not previously been investigated. The aim of this study was therefore to evaluate the association of AF burden, duration and number of episodes with clinical outcomes in patients with paroxysmal AF.

Methods

Patient population

Patients from the Reappraisal of Atrial Fibrillation: interaction between hyperCoagulability, Electrical remodelling and Vascular destabilisation in the progression of AF Study (RACE V, ClinicalTrials.gov no. NCT02726698) with 1 year of validated implantable loop recorder (ILR)-recorded AF episodes were eligible for inclusion. The RACE V Study is a Dutch, prospective, single-arm, multicentre, observational study aimed to investigate predictors of AF progression in patients with self-terminating paroxysmal AF. A detailed description of the RACE V Study was previously published.3 8–10 Patients were eligible for inclusion in case they were ≥18 years of age, had documented, self-terminating paroxysmal AF of a less than 10-year duration with no other indication for oral anticoagulants, and were willing to undergo implantation of an ILR or had a cardiac implantable electronic device (with an atrial lead) compatible with the CareLink remote monitoring system. Exclusion criteria were triggered AF, congenital heart disease, prior or planned pulmonary vein isolation, current or expected initiation of amiodarone treatment, pregnancy or life expectancy less than 2.5 years. All patients completed 1 year of continuous rhythm monitoring.

Patients were not involved in the design, conduct, reporting or dissemination of this study.

Clinical assessments and healthcare utilisation

All patients underwent extensive baseline assessments including a clinical examination and echocardiography. Follow-up visits were scheduled at 12 months. Information concerning serious adverse effects resulting in death, disability or incapacity, requiring prolonged or unexpected hospitalisation or life-threatening events, was recorded prospectively throughout follow-up. Healthcare utilisation included planned cardioversions and invasive rhythm control (ablation). Healthcare utilisation was evaluated based on frequency of AF-associated hospital admissions over 1 year of follow-up. This was defined as any hospitalisation resulting in chemical and/or electrical cardioversion (ECV) and planned/unplanned AF ablation or AF-related surgery. Choice in treatment strategy for AF was at the discretion of the physician and decided in accordance with the prevailing European Society of Cardiology AF guidelines.1 11 Information about implantation of an ILR was only provided per request from the treating physicians.

Assessment of quality of life

AF symptoms and health-related quality of life were assessed using the Atrial Fibrillation Severity Scale (AFSS)12 and EuroQol EQ-5D-5L questionnaires administered at baseline and at the 1-year follow-up visit.

The AFSS questionnaire is a disease-specific tool designed to evaluate symptom severity in patients with AF. It encompasses seven items that evaluate symptoms experienced at rest (palpitations, dyspnoea, fatigue, dizziness and chest pain) and during exercise (dyspnoea, fatigue). Each symptom is scored on a range from 0 to 5, and the maximum attainable score is 35 with higher scores indicating higher symptom severity.12 13

The EQ-5D-5L is a standardised and validated questionnaire used for assessment of health-related quality of life in diverse patient populations. It is comprised by five dimensions of health (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) using five response levels (no problems, slight problems, moderate problems, severe problems or extreme problems). A second component constitutes a Visual Analogue Scale (VAS) (the EuroQol VAS (EQ-VAS)) in which patients rate their health on a range from 0 to 100 with 100 representing perfect health.14

AF patterns and burden

Patients received a LINQ Reveal ILR (Medtronic, Minnesota, USA) connected to a CareLink remote monitoring system (Medtronic, Minnesota, USA). The ILR was set to automated detection of AF ≥2 min (regardless of rate), tachycardias ≥182 beats per minute (bpm), pauses ≥5 s and bradycardias ≤50 bpm. To prevent data loss, patients were instructed to perform daily automated and weekly manual transmissions. All episodes of AF were manually adjudicated and corrected by five independent physicians. For this analysis, only episodes of AF occurring within the first year after ILR implantation were assessed. We defined AF burden as percentage of time spent in AF over 1 year of follow-up. Patients were grouped into tertiles based on total AF burden: no AF, lower tertile (<0.2%), middle (intermediate) tertile (0.2–1.9%) and upper (high) tertile (>1.9%). For descriptive purposes, we defined short episodes as episodes lasting less than 10 min, intermediate episodes as episodes lasting 10 min–6 hours and long episodes as episodes lasting longer than 6 hours. Episodes lasting >24 hours were also assessed. Patients were additionally grouped according to duration of their longest episode using these cut-offs and into quartiles according to the total number of AF episodes experienced over follow-up: <9 episodes, 9–32 episodes, 33–147 episodes and >147 episodes. Due to the limited sample size, tertiles and quartiles were chosen to ensure sufficient observations in each group.

Statistical analyses

Categorical data are presented as frequencies with percentages, and continuous data as means with SDs or medians with IQRs, as appropriate. Between-group differences were analysed using the Mann-Whitney two-sample test. The association of AF burden, number of AF episodes and duration of longest AF episode with healthcare utilisation and quality of life was further assessed using logistic and Poisson regression models with and without adjustment for potential confounders (age, sex, coronary artery disease, use of class I or III antiarrhythmic drugs or beta-blocker usage at baseline). One patient with missing information in the baseline assessments was excluded from the multivariable regression analyses. None of our patients had missing ILR data or missing information about healthcare utilisation, but a total of 24 (AFSS) and 27 (EQ-5D-5L) patients had missing information in the 1-year follow-up quality-of-life questionnaires. Missing information was in all cases considered missing at random and was addressed using multiple chained imputation (n=20 imputed sets). We considered two-sided p values below 0.05 to be statistically significant. Statistical analyses and graphs were completed using Stata V.18.0 for Windows (StataCorp, Minnesota, USA).

Results

We included 417 patients with paroxysmal AF from the RACE V Study. A total of 63 973 episodes of AF were detected in 353 (85%) patients during 1 year of continuous rhythm monitoring; 64 (15%) patients did not experience AF recurrence. Baseline characteristics according to AF burden are shown in table 1 and according to episode duration and number of episodes in online supplemental tables 1 and 2. The median age was 65 years (IQR 58–71 years), 179 (43%) were women, 310 (74%) had a CHA2DS2-VASc score of 2–4, and 96 (27%) patients were in pharmacological rhythm control therapy with class I or III antiarrhythmics at baseline (online supplemental figure 1). Baseline characteristics according to episode duration and number of AF episodes were comparable. The median AF burden was 0.7% (IQR 0.1–4.0%) and the median time to first AF episode in patients with AF recurrence was 13 days (IQR 3–54 days). We observed large variability in AF patterns over follow-up; this was most pronounced in patients with AF burdens in the intermediate-to-high tertiles (figure 1). An overview of duration and number of AF episodes according to AF burden is listed in table 2 and visualised in figure 2.

Supplemental material

Figure 1

Graphical abstract. AF, atrial fibrillation; RACE V, Reappraisal of Atrial Fibrillation: interaction between hyperCoagulability, Electrical remodelling and Vascular destabilisation in the progression of AF.

Figure 2

Episode characteristics (duration and number) and AFSS symptom scores (1-year follow-up) according to AF burden. Each vertical bar represents one patient. Patients are listed in the same order in all graphs (n=353). Higher AFSS scores and lower EQ-5D-5L index scores represent more symptoms and worse health situation, respectively. A total of 27 and 42 patients did not complete the AFSS and EQ-5D-5L questionnaires at the 1-year follow-up visit (shown as grey lines). AF, atrial fibrillation; AFSS, Atrial Fibrillation Severity Scale; EQ-VAS, EuroQol Visual Analogue Scale.

Table 1

Baseline clinical and echocardiographic characteristics of patients without AF recurrence and across tertiles of AF burden

Table 2

Episode characteristics of patients according to AF burden

Higher AF burdens were observed in patients with longer episode duration (table 2 and online supplemental table 3); patients with episode duration >24 hours had a median AF burden of 7.6% (IQR 1.7–19.2%, n=95), whereas patients with longest episodes lasting <1 hour had a median AF burden of 0.01% (IQR 0.002–0.04%, n=51). Likewise, a higher number of AF episodes were detected in the high AF burden group: median 164 (IQR 58–593) episodes among patients in the upper tertile vs 8 (IQR 3–17) episodes for patients in the lower tertile of AF burden (table 2).

Healthcare utilisation

Patients with longer AF episodes (>1 hour) and patients with intermediate AF burdens (0.2–1.9%) underwent significantly more cardioversions (0% vs 9.5%, p=0.04, and 4.2% vs 13.6%, p=0.01, respectively) and ablation procedures (2.0% vs 15.8%, p=0.01, and 5.9% vs 16.2%, p=0.01, respectively) than patients with lower burdens or shorter episodes (table 2 and figures 3 and 4). These associations were maintained in the adjusted regression analyses (online supplemental table 4). Of interest, there was no difference in the number of cardioversions performed in patients with lower versus higher AF burdens (4.2% vs 2.6%, p=0.48). We found no association between number of AF episodes and any measure of healthcare utilisation in the unadjusted analyses (online supplemental figure 2 and online supplemental table 5); however, cardioversions appeared less frequent in patients with multiple episodes (>147) in the adjusted analyses.

Figure 3

Box plots and bar graphs showing the association of atrial fibrillation burden with episode frequency, duration of longest episode and healthcare utilisation (n=353).

Figure 4

Box plots and bar graphs showing the association of duration of longest episode with atrial fibrillation (AF) burden, episode frequency and healthcare utilisation (n=353).

Quality of life and AF-related symptoms

Health-related quality of life was assessed based on AFSS and EQ-5D-5L questionnaire responses. A total of 389 of 417 patients completed both questionnaires at 1 year of follow-up. The median AFSS score was 4 (IQR 1–9) in the population (maximum attainable score is 35). With or without adjustment, patients with AF recurrence generally reported more symptoms than patients without AF recurrence (table 3 and online supplemental table 6); median AFSS score was 2 (IQR 0–5) for patients without recurrent AF vs 5 (IQR 3–9) for patients with AF burdens in the upper tertile. Results according to AF duration and number of episodes were comparable. Across AF patterns, AFSS scores varied remarkably, and particularly, we found no significant differences in symptom severity with AF burden or duration in patients with AF recurrence during follow-up (figures 2 and 5 and online supplemental table 6). A high number of AF episodes was, however, associated with more symptoms in this population; the median AFSS score was 3 (IQR 0–8) for patients with few (<9) episodes vs a median score of 7 (IQR 3–12) for patients with a high number of episodes (>147) (p=0.001) (figure 6 and online supplemental tables 5 and 6).

Figure 5

Box plots showing the association between AF burden (in tertiles), number of episodes (in quartiles) and duration of the longest episode with AFSS symptom scores at the 1-year follow-up. Stars (*) indicate reference group. AF, atrial fibrillation; AFSS, Atrial Fibrillation Severity Scale (n=417).

Figure 6

Box plots showing the association between AF burden (in tertiles), number of episodes (in quartiles) and duration of the longest episode with EQ-5D-5L anxiety scores at the 1-year follow-up. Colour intensity indicates severity of symptoms with lightest colours indicating the least perceived problems. Stars (*) indicate reference group. AF, atrial fibrillation.

Table 3

Healthcare utilisation and quality of life at the 1-year follow-up visit according to AF burden

Despite overall high self-evaluated health (median 80, IQR 70–85), we also observed high variability in EQ-5D-5L index scores regardless of AF burden, duration or number of episodes. Results according to each domain of health are shown in table 3 and online supplemental tables 3 and 5. Anxiety prevalence according to AF burden, duration and episode number is shown in figure 6. Patient anxiety is an important cause for acute hospitalisation in patients with AF and was therefore included as a separate parameter in this analysis. No statistically significant between-group differences were observed.

Discussion

The purpose of this study was to investigate the association of AF burden, duration and number of episodes with healthcare utilisation and quality of life in patients with early self-terminating paroxysmal AF without a history of ablation. Based on data obtained during 1 year of continuous rhythm monitoring in 417 patients, we demonstrated that (1) a high degree of variability in mode of AF recurrences in terms of total AF burden, episode duration and number of episodes exists; (2) healthcare utilisation assessed by AF-associated hospitalisation was increased in patients with higher AF burdens and longer episode duration, but for cardioversions, specifically duration of longest episode was more influential than AF burden; and (3) no differences in AF symptom scores were observed when analysed based on AF burden and AF duration, but patients with many AF episodes, however, reported more severe symptoms. Finally, we observed remarkable variation in AF symptoms and EQ-VAS.

Recently, the conventional definition of AF recurrence in rhythm control trials as time to first sustained AF episode lasting longer than 30 s15 16 has been challenged due to lack of clinical argumentation to support this threshold. This includes non-differentiation between patients with low versus high AF burden, or between patients with symptomatic versus non-symptomatic episodes.6 17 As opposed to studies on subclinical AF,18–20 the RACE V Study applied ILR-based monitoring in patients with clinical AF. Our findings suggest a positive correlation between increasing AF burden, duration and episode number with increased healthcare utilisation, thus indicating that a significant gradient in AF severity exists within this patient group. Evidence in favour of a more refined approach to AF was also demonstrated in a recent subanalysis of the Cryoballoon vs Contact-Force Atrial Fibrillation (CIRCA-DOSE) trial,6 which associated longer-duration AF episodes (>1 hour) and AF burdens >0.1% (defined either as proportion of time in AF) with increased rates of healthcare utilisation post-ablation.

Results from this study indicate that depending on the outcome of interest, AF burden, duration and number of episodes could be appropriate parameters to assess. AF symptom severity was more closely related to higher AF episode frequency than to AF burden, which contrasts findings from CIRCA-DOSE6 where both episode duration >24 hours and AF burden >5% were associated with an attenuated improvement in AF Effect on QualiTy-of-life Questionnaire scores after catheter ablation. The variation in symptoms with number of AF episodes in the CIRCA-DOSE population could relate to the more frequent onset and termination of AF episodes. Similarly, duration of longest AF episode correlated better with the need for cardioversions than burden of AF. In line with results from CIRCA-DOSE, we found no association between generic quality-of-life scores regardless of AF classification method, but AF recurrence versus non-recurrence did influence perceived symptoms as assessed by the AFSS—and as noted, so did number of episodes. Of interest was the enormous variation in symptoms and EQ-VAS score between patients, which concurs with previous data showing poor symptom–rhythm correlation in AF.21 22

Traditional AF classification into paroxysmal, persistent or permanent AF may be insufficient as our findings support a more refined perception of paroxysmal AF; how it presents could provide aetiological as well as prognostic clues. Because continuous rhythm monitoring cannot be established for all patients, the optimal time frame for measurement of AF burden, duration and number of episodes, and the association with AF assessed using opportunistic or intermittent monitoring are of interest and need further clarification, as does the potential of non-invasive continuous rhythm monitoring using novel digital devices in this patient group.23 An improved understanding of the relationship between specific clinical characteristics and different patterns of AF might impact how and in whom early initiation of invasive or non-invasive rhythm control therapy is more appropriate. Importantly, a high number of episodes was associated with more reported symptoms, and thus, treatment to reduce AF targeted towards these individuals might also provide incremental value from the patient perspective.

Limitations

We acknowledge several limitations. Patients were included in a non-consecutive manner based on specific inclusion and exclusion criteria and resultingly, do not reflect the population of patients with paroxysmal AF at large. Moreover, our sample size was moderate, which reduced statistical power to detect smaller but potentially clinically significant differences between groups. No correction was made for multiple comparisons, and results from this study should be considered as hypothesis generating only. The overall burden of AF was low in this population, and 64 (15%) patients did not experience AF recurrence during follow-up. Also, only the first year of monitoring was included for simplicity, and how these parameters influence clinical outcomes on the longer term cannot be deduced from our data. For this study, we did not censor patients at cardioversion and therefore, we cannot exclude that termination of AF episodes by ECV could have reduced the AF burden in these patients. Hence, this could have influenced the lack of association between rate of ECV with high AF burden. Meanwhile, only 24 (5%) patients underwent ECV during the study period. Patients with AF may account for repeated emergency department admissions, placing additional (often unnecessary) burden to the healthcare system. Emergency department visits were not included as a formal outcome in RACE V and could therefore not be included in the present study. Also, it is possible that physicians were influenced by information from the ILR when deciding whether to proceed with ECV or ablation. Finally, our findings rely on the capability of the ILR to accurately detect all episodes of AF. Previous studies indicate a sensitivity >90%,24 25 and to ensure high specificity, all episodes of AF were manually adjudicated.

Conclusion

In a contemporary cohort of patients with self-terminating paroxysmal AF, we observed high variability in AF recurrence patterns. Higher AF burden and longer AF episode duration were associated with more healthcare utilisation but not with symptoms and quality of life. Depending on the outcome of interest, AF burden, duration and number of episodes could become incremental metrics in clinical and prognostic assessments in patients with AF.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by Medische Ethische Toetsings Commissie van de Universiteit Medisch Centrum Groningen (METc UMCG) (METc no: METc 2016/16) and all participating centres. The RACE V Study was conducted in concordance with the Declaration of Helsinki, and all patients gave written, informed consent.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • MHJPF and MVDL are joint first authors.

  • X @harry_crijns, @mirkogray

  • Contributors MHJPF was involved in the study design and data acquisition, conducted the statistical analyses and wrote the manuscript. IVG, MR and MVDL were involved with the study design, data acquisition and interpretation of results, and critically reviewed the manuscript. MHJPF and MVDL share the first authorship. HJGMC, JCN, MEWH, RGT, MDM, DL and US were involved in the interpretation of the data and critically reviewed the manuscript. All authors approved the final version. IVG is the guarantor of this manuscript.

  • Funding The Netherlands Cardiovascular Research Initiative: an initiative with support of the Dutch Heart Foundation (CVON 2014-9: Reappraisal of Atrial Fibrillation: interaction between hyperCoagulability, Electrical remodelling and Vascular destabilisation in the progression of AF (RACE V)). Unrestricted grant support from Medtronic Trading.

  • Competing interests MHJPF received consulting fees from Medtronic outside this work. MBK received speaker’s honoraria from Abbott outside this work. JCN was supported by a grant from the Novo Nordisk Foundation (NNF16OC0018658). MR received consultancy fees from Bayer and InCarda Therapeutics (to the institution). MDM is an employee of Medtronic Bakken Research Center. RGT reports grants and personal fees from Medtronic and grants from St Jude Medical outside this work. In addition, RGT has a patent as co-inventor of the MyDiagnostick issued. The remaining authors declare no conflicts of interest.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.