Article Text

Therapy with direct oral anticoagulants for secondary prevention of thromboembolic events in the antiphospholipid syndrome: a systematic review and meta-analysis of randomised trials
  1. Josefine B H Adelhelm1,
  2. Robin Christensen2,3,
  3. Gustavo G M Balbi4 and
  4. Anne Voss1,3
  1. 1Department of Rheumatology, Odense University Hospital, Odense, Denmark
  2. 2Section for Biostatistics and Evidence-Based Research, the Parker Institute, Copenhagen University Hospital, Copenhagen, Denmark
  3. 3Department of Clinical Research, University of Southern Denmark Faculty of Health Sciences, Odense, Denmark
  4. 4Department of Clinical Medicine, Federal University of Juiz de Fora, Juiz de Fora, Brazil
  1. Correspondence to Dr Anne Voss; anne.voss{at}rsyd.dk

Abstract

Objective Antiphospholipid syndrome (APS) is a systemic autoimmune disorder characterised by venous thrombosis (VT) or arterial thrombosis (AT) and/or pregnancy morbidity and the presence of antiphospholipid antibodies. Direct oral anticoagulants (DOACs) hold several advantages to vitamin K antagonists (VKAs) for prevention of thrombosis and we wish to evaluate DOACs compared with VKAs in secondary prevention of thromboembolic events in patients with APS.

Methods We conducted searches of the published literature using relevant data sources (MEDLINE, Embase and Cochrane CENTRAL), and of trial registers for unpublished data and ongoing trials. We included randomised trials examining individuals >18 years with APS classified according to the criteria valid when the trial was carried out. Randomised controlled trials had to examine any DOAC agent compared with any comparable drug. We tabulated all occurrences of events from all eligible randomised trials. Due to few events, ORs and 95% CIs were calculated using the Peto method.

Results 5 randomised trials comprising 624 patients met the predefined eligibility criteria. The primary outcome measure was new thrombotic events, a composite endpoint of any VT or AT, during the VKA-controlled phase of treatment. According to the I2 inconsistency index, there was evidence of statistical heterogeneity across the studies (I2=60%). Across trials, 29 and 10 thrombotic events were observed in 305 and 319 patients with APS treated with DOAC and VKA, respectively, corresponding to a combined Peto OR of 3.01 (95% CI 1.56 to 5.78, p=0.001). There was a significantly increased risk of AT while treated with DOACs compared with VKA (OR 5.5 (2.5, 12.1) p<0.0001), but no difference in the risk of VT (p=0.87). We found no significant difference in risk of bleeding.

Conclusions DOACs were associated with a significant increase in the risk of a new thrombotic event, especially AT, favouring standard prophylaxis with warfarin.

PROSPERO registration number CRD42019126720.

  • Antibodies, Anticardiolipin
  • Antiphospholipid Syndrome
  • Antibodies, Antiphospholipid
  • Cardiovascular Diseases
  • Therapeutics

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information.

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

  • Vitamin K antagonists (VKAs) are the only anticoagulants recommended for secondary prevention of thrombosis in antiphospholipid syndrome (APS), and we wish to evaluate direct oral anticoagulants (DOACs) compared with VKAs for this.

WHAT THIS STUDY ADDS

  • DOACs were associated with a significant increase in the risk of a new thrombotic event, especially arterial thrombosis, favouring standard prophylaxis with warfarin.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • The results of our meta-analysis might suggest that a subset of patients with APS with only venous thrombosis history might benefit from DOAC treatment. Since scientific studies so far report the results heterogeneously, we propose in future to use a core outcome set in this area of research.

Introduction

Background

Antiphospholipid syndrome (APS) is a systemic autoimmune disorder characterised by venous thrombosis (VT) or arterial thrombosis (AT) and/or pregnancy morbidity in the presence of antiphospholipid antibodies (aPL) on two or more occasions at least 12 weeks apart. The aPL include lupus anticoagulant, anticardiolipin antibody or anti-β2 glycoprotein I IgG or IgM antibodies.1 APS occurs as a primary condition, or secondary in the presence of, for example, SLE.2 Evidence suggests that the aPL profile is prognostic, and triple positivity increases the risk of thromboembolic events.3

The European Alliance of Associations for Rheumatology recommends for secondary prophylaxis of thrombosis APS ‘treatment with VKAs (vitamin K antagonists) with INR (internationalised normalised ratio) 2–3 or INR 3–4’ considering the individual’s risk of bleeding and recurrent thrombosis. Treatment with VKA with INR 2–3 plus low-dose aspirin may also be considered.4

The pharmacodynamics of direct oral anticoagulants (DOACs) are inhibition of either factor IIa (thrombin; for example, dabigatran etexilate) or factor Xa (eg, rivaroxaban, edoxaban or apixaban). DOACs are easy to use with simple dosing, anticoagulation monitoring is not indicated and drug plasma levels should not be followed. However, dosage should be adjusted in patients with impaired renal or liver function.5

Rationale

Lifelong treatment with VKA implies frequent monitoring of INR and may be experienced as a burden by the patient, as indicated by scientific studies demonstrating a decrease in quality of life.6 Furthermore, the dose–response relationship between coumarins and INR is affected by many factors including dietary habits, genetic interactions, drug interactions, etc, which may increase the risk of bleeding including life-threatening episodes.7 8

DOACs are recommended for secondary prophylaxis in patients with deep vein thrombosis and pulmonary embolism not related to APS,5 and it is relevant to explore the potential of DOACs in secondary prevention of thromboembolic events in APS. If DOACs could replace VKA, partially or completely, we hypothesise that it could potentially reduce the risk of bleeding episodes and change the patient’s perception of own illness. In 2016–2017, two authors9 10 showed positive case reports on 23 and 24 patients with APS, respectively, treated with DOAC for secondary prophylaxis.

In 2021, a meta-analysis of randomised controlled trials (RCTs) by Dufrost et al11 found a significantly higher risk of recurrent AT, but not for VT, when comparing DOACs with VKA for secondary prophylaxis. The same year, Aibar and Schulman published a meta-analysis on RCTs and cohorts comparing any antithrombotic regimen in APS,12 and found that VKA was more effective than DOAC (relative risk (RR): 0.25; 95% CI: 0.07 to 0.93) to prevent recurrent AT. Both concluded that DOACs should not be used for patients with APS with a history of AT. Khairani et al confirmed this including four RCTs and emphasised that patients with thrombotic APS on DOACs compared with VKA have increased risk of AT.13 Recently, Shah et al also found increased risk of stroke among patients with APS treated with DOACs and calculated RRs that may indicate DOACs to be associated with higher risks of thrombotic events.14 In the present meta-analysis, we included data from all five present RCTs on patients with APS published 2016–2022 as seen below.

Objectives

Our objectives were to examine whether DOACs reduce the incidence of secondary APS-related AT and VT, by reviewing randomised trials that assess the efficacy and safety of these drugs for secondary prophylaxis in patients with APS.

Methods

Protocol and registration

The review protocol was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols guidelines15 and registered with the International Prospective Register of Systematic Reviews on 12 April 2019 (CRD42019126720); the original protocol is available as online supplemental appendix 1. The reporting of the systematic review and meta-analysis follows the recommendations from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.16

Supplemental material

Information sources and search strategy

Literature search strategies were developed in collaboration with a research librarian (LØ). We searched Cochrane Central Register of Controlled Trials, MEDLINE and Embase in May 2022 (see online supplemental appendix 2 for search strategy). The electronic database search was supplemented by searching ongoing trial registers: https://www.clinicaltrials.gov/; https://www.clinicaltrialsregister.eu/; https://www.who.int/ictrp/en and https://clinicaltrials.bayer.com/. We also scanned the reference lists of included studies and relevant reviews identified through the search. No language limits were imposed on the search.

Supplemental material

Study selection and data extraction

Literature search results were uploaded to Covidence. The first review author (JBHA) screened the titles and abstracts yielded by the search against the eligibility criteria. We included RCTs examining individuals >18 years with APS classified according to the criteria valid when the trial was carried out. RCTs had to examine any DOAC agent compared with any comparable drug (ie, both active and placebo comparators). We obtained full reports for all titles that appeared to meet the eligibility criteria. Review authors (JBHA/AV) independently screened full-text reports and decided whether these met the inclusion criteria; we resolved disagreement through discussion (RC). The first review author (JBHA) extracted data from the included trials, using a customised Microsoft Excel spreadsheet database. All analyses were based on data reported on the intention-to-treat (ITT) principle whenever possible. The major efficacy outcome was incident thromboembolic events; other major outcomes were (1) bleeding and (2) death. Major bleeding was defined by the International Society on Thrombosis and Haemostasis as clinically overt bleeding associated with any of the following: (1) fatal outcome; (2) involvement of a critical anatomical site; and (3) fall in haemoglobin concentration of at least 20 g/L or the need for transfusion of ≥2 units of packed red blood cells or whole blood.17

Risk of bias in individual studies

We used the Cochrane Collaboration tool18 to facilitate the assessment of possible risk of bias and evaluated five bias domains: selection bias (random sequence generation and allocation concealment); performance bias (blinding of participants and personnel); detection bias (blinding of outcome assessment); attrition bias (incomplete outcome data) and reporting bias (selective reporting). Each bias domain was graded low risk, high risk or unclear risk. Then, each RCT was assigned an overall risk of bias in terms of low risk (low for all key domains), high risk (high for ≥1 key domain) and unclear risk (unclear for ≥1 key domain).

Statistical analysis

Anticipating that the major outcomes would correspond to rare outcome events, we followed recommendations of Bradburn et al19 and used Peto ORs and 95% CIs as the primary analysis approach to compare the DOAC and comparator groups.

Because all trials had similar duration of follow-up for all treatment groups, the use of ORs represents a valid approach to assessing the risk associated with the use of DOAC. Trials in which patients had no events in either group were excluded from analyses. P values are two sided. We tested for heterogeneity with the Cochran’s Q-test and used the method proposed by Higgins et al to measure inconsistency, where I2 is interpreted as the percentage of total variation across several studies due to heterogeneity.20 Results in forest plots present Peto OR estimates and 95% confidence for each major outcome, to give a visual suggestion of the amount of study heterogeneity and of the overall combined results of the included studies. While the primary meta-analyses were based on ORs and 95% CIs calculated with the use of the Peto method, we also performed meta-analyses using absolute risk differences as the effect measure, applying both a fixed and random-effects approach. Subgroup analyses and sensitivity analyses using alternative meta-analysis approaches are presented in online supplemental appendices 3–6. Data were analysed with the use of Review Manager V.5.3 (The Cochrane Collaboration).

Supplemental material

Supplemental material

Supplemental material

Supplemental material

Outcome Reporting Bias In Trials

Outcome reporting bias (ORB) occurs when variables are selected for publication based on their results.21 To explore the risk of ORB, an outcome matrix was produced to help identify missing study outcome data. In the outcome matrix, the outcomes of interest in the review and how they were reported in the trial are listed in the columns and the different studies listed in the rows. The Outcome Reporting Bias In Trials (ORBIT) Matrix enabled us to evaluate the risk of ORB in the qualitative evidence synthesis. Further, the following were also done: (1) checking the reasons, when available, for excluding studies to ensure that no studies were excluded because they did not report the outcomes of interest in the review; and (2) assessing the eligible studies as to whether the review outcomes of interest were reported and what other core outcomes were reported in the included trials. If important outcomes were not reported, authors were contacted for information.

Patient and public involvement

The present research questions were conceived from patients’ inquiries in the outpatient clinic. Patient partners were involved when the review protocol was prepared. The results will be presented in national patient partner groups.

Results

Results of the search

As illustrated in figure 1, 750 studies were identified from the databases after de-duplication. After title and abstract screening by JBHA, 738 articles were excluded, leaving 12 articles for full-text scrutiny by JBHA and AV. Searching trial registries, we found one ongoing trial (ClinicalTrials.gov no. NCT03684564, RISAPS). Six publications met the inclusion criteria22–27 for qualitative analysis, but just five studies were included in the review, since Arachchillage et al23 analysed data from the RAPS trial.24 Goldhaber et al26 performed post-hoc subgroup analysis based on data from RE-COVER I+II28 29 and RE-MEDY.30

Figure 1

PRISMA flow chart of studies included. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; RCT, randomised controlled trial.

Qualitative evidence synthesis

Description of included studies

Study and patient characteristics of the five eligible randomised trials22 24–27 are listed in table 1. The included studies comprised 305 patients with APS treated with rivaroxaban 20 mg/day or dabigatran etexilate 150 mg two times per day and 320 comparators treated with VKA (warfarin, target INR 2.0–3.0). The patients included in Goldhaber et al26 and the RAPS trial24 had a history of VT only, whereas patients in the other studies had histories of both venous and arterial events. The pooled mean age for the DOAC group was 47.2 years (SD 13.8) and 47.7 years (SD 15.5) for the VKA group. The pooled body mass index (BMI) for the DOAC group was 28 kg/m2 (SD 6) and 28.8 kg/m2 (SD 6) for the VKA group. Overall, the groups were comparable in terms of age, BMI and sex. However, the percentage of patients with APS with SLE varies noticeably, from 9% in the Woller et al22 DOAC group to 33% in the Ordi-Ros et al27 DOAC group; only the RAPS was stratified for SLE. Woller et al, Ordi-Ros et al and TRAPS25 have collected data on cardiovascular risk factors; 36% of the DOAC group and 40% in the VKA group were smokers; 32% of the DOAC group and 29% of the VKA group had known hyperlipidaemia; 32% of the DOAC group and 35% of the VKA group had hypertension. Less than 10% in each group had diabetes.

Table 1

Characteristics of included randomised trials

Description of ongoing studies

Still underway, the RISAPS trial (ClinicalTrials.gov ID: NCT03684564) will compare higher-intensity rivaroxaban 15 mg two times per day versus higher-intensity warfarin (INR 3.0–4.0) for 24 months, in patients with APS, with or without SLE, after experiencing a stroke, a transient ischaemic attack or other ischaemic brain damage caused by blood clots in the brain arteries or smaller blood vessels. Planned completion is end of 2024, and the trial manager has been contacted for an update in June 2022.

Thromboembolic events

As seen in figure 2A, the pooled number of thromboembolic events was 29 in the DOAC group and 10 in the warfarin group. The summary OR for thromboembolic events was statistically significant (3.01 (95% CI 1.56 to 5.78)) with a moderate-to-large degree of inconsistency (I2=60%), also visualised by individual OR values ranging from 0.84 (95% CI 0.18 to 3.82) to 10.33 (95% CI 1.9 to 56.3).

Figure 2

Forest plots of outcomes (A–D) for comparison of DOACs versus warfarin (Peto OR). DOAC, direct oral anticoagulant.

Bleeding and death

Bleeding events are shown as overall events in figure 2B and the subgroup of major bleeding events is shown in figure 2C. For overall bleeding events, there were a total of 63 events in the DOAC group(s) and 70 in the warfarin group, corresponding to a summary OR for bleeding events of 0.92 (95% CI 0.62 to 1.37) which was not significant. All trials further subgrouped for major bleeding events, as seen in figure 2C. With 11 major bleeding events in the DOAC group and 12 in the warfarin group and an OR 0.94 (95% CI 0.41 to 2.17, p=0.88), there was also no significant difference between treatment groups for major bleeding events. Figure 2D shows that there were no deaths related to treatment, but only three studies report this outcome. One cardiovascular death was observed in the TRAPS trial25 in a patient from the DOAC group with known heart failure; the death occurred 433 days after suspension of DOAC while the patient was back on warfarin. Unfortunately, due to the outcome reporting method of Goldhaber et al,26 it is unclear whether Goldhaber et al observed any deaths related to VT in the study period or only non-fatal VTs.

Subgroup analyses

Risk of thrombosis in subgroups of the trial population was analysed based on type of thrombosis, thrombosis history prior to trial inclusion and triple aPL positivity (forest plots of ORs in online supplemental appendix 6). There was a significantly increased risk of AT while treated with DOACs compared with VKA (OR 5.5, 95% CI 2.5 to 12.1, p<0.0001), whereas there was no difference in the risk of VT (p=0.87).

DOACs were significantly worse than VKA for secondary prophylaxis, especially among patients with a history of AT (OR 5.5 (95% CI 2.1 to 14.7) p=0.0006). Although less harmful, DOACs were also inferior to VKA in patients with a history of VT (OR 2.7 (95% CI 1.2 to 6.1) p=0.01). The risk of thrombosis for the subgroup of aPL triple-positive patients was higher with DOACs than VKA (OR 3.8 (95% CI 1.66 to 8.65) p=0.002). Unfortunately, not all trials characterised how many patients had a history of both AT and VT or described the patients’ aPL profiles.

Other outcomes

Due to three strokes in the DOAC arm, Woller et al22 doubled the daily dose of DOAC, after inclusion of 25 patients. Nevertheless, three more events occurred and subsequently all patients with a history of AT were excluded. Ordi-Ros et al27 registered catastrophic APS in one patient receiving DOAC. The RAPS trial24 measured thrombin generation as a primary efficacy endpoint and the endogenous thrombin potential was significantly higher in the DOAC group as compared with the warfarin group at day 42, but it did not reach the prespecified non-inferiority threshold of less than 20% difference in mean percentage change. RAPS trial also measured quality of life and found no difference between treatment groups in terms of health utility, but a small difference in the visual analogue score favoured the DOAC group (mean difference 6.5 (95% CI 1.4 to 11.5) p=0.013). Patient satisfaction with DOAC was significantly higher than with warfarin, assessed by Woller et al.22

Risk of bias in included studies

Figure 3A presents our risk of bias assessments for each of the eligible studies, supported by figure 3B illustrating each risk of bias item presented as percentages across all included studies.

Figure 3

(A) Risk of bias summary and (B) risk of bias graph. QoL, quality of life; SAE, serious adverse event.

All of the included trials used correct methods of randomisation and correct allocation concealment, minimising risk of selection bias on the included individuals. Only Goldhaber et al26 used double-blinding, the other trials found the need to do open-label studies. For outcomes such as thrombosis and bleeding, we estimate that they have been objectively evaluated and most trials had blinded committees assess potential outcomes as prespecified in the protocols. However, because of a lack of blinding of participants reporting on their perceived change in quality of life and satisfaction with anticoagulant treatment, we judge a high risk of performance bias, due to the subjective nature of the outcome measures. There are no available data on the reason for 6.3% attrition in each treatment group in Ordi-Ros et al,27 and Goldhaber et al26 do not comment on the reason for a large proportion of loss to follow-up in RE-MEDY (20%) or the proportion of patients with APS lost to follow-up. Hence, risk of attrition bias is high in these studies. Selective reporting bias is generally low, but when compared with the study protocol, Woller et al’s study22 lacks outcome data on metrics of ability to include patients, compliance and nuisance bleeding.

Discussion

In our comprehensive literature search, five clinical trials were identified, two had to modify the study protocol and both terminated early, due to excess of events in the intervention drug arm and due to low patient accrual, respectively.22 25 The results of the meta-analysis have some limitations: only five RCTs are available, with a total study population of just 625 patients. Additionally, each individual study has several limitations. For instance, the study by Goldhaber et al26 was a post-hoc analysis of three RCTs, which were not designed to examine patients with APS, as they did not test all patients for thrombophilia and positivity for aPL was not confirmed after a minimum of 12 weeks, as required by classification criteria.1

Besides APS history and type of anticoagulant treatment, other risk factors such as age and smoking should be considered, when evaluating a person’s cardiovascular risk. A 2013 cross-sectional study31 found that the combination of smoking and aPL antibodies was strongly associated with vascular events. Hence, 35–40% of the trial population in this meta-analysis are smokers and none of the included RCTs took this confounder into account. Previously, a ‘two-hit hypothesis’ has been suggested for APS. In addition to persistent positivity for aPL, a ‘second hit’ is required to ‘trigger’ events. Factors such as inflammation, infection, genetic predispositions, age, smoking and traditional cardiovascular risk factors (hypertension, diabetes mellitus, hyperlipidaemia, obesity, etc) are all suggested as ‘second hits’ and should be controlled for when trying to understand the real impact of DOACs in patients with APS.32 33

The results might partly be biased by the heterogeneity regarding thrombosis history and autoantibody profile in the trial populations, for example, patients in Goldhaber et al and RAPS24 had a history of only VT and unknown antibody profile, whereas patients in the TRAPS trial25 had previous VT or AT and were triple aPL positive. In the as-treated analysis of thrombotic events in Woller et al,22 after excluding patients with a history of AT, the rate of events in the DOAC arm was one-third of the ITT analysis. Through subgroup analyses, we found the risk of thrombosis in patients with a history of AT to be twice as high as those with a history of VT, but with reservations that some patients had a history of both AT and VT. In short, this might indicate a potential role of DOACs in selected populations with APS, and the choice of treatment may be stratified according to aPL profile and/or whether the thromboembolic history included venous or arterial events.

The follow-up time in the included studies varies from 7 months in RAPS and Goldhaber et al to 36 months in Ordi-Ros et al, the latter reporting the highest number of thromboembolic events and bleeding events. Indeed, the APS ACTION registry showed that even though higher than the general population, incident thrombotic events in APS are rare, 2.09 events per 100 patient-years based on almost 4000 patient-years of follow-up.34 Hence, more events might potentially have been observed in the other trials, if patients had been followed for a longer period of time.

Due to the limited number of available publications in this area, it is not surprising that a core outcome set has not yet been developed. Nevertheless, the major core outcomes were measured in all five included studies. Inspired by table 2, we suggest the following three tiers of outcome measurements for future trials: tier 1 core outcomes: new thrombotic event (venous/arterial/microvascular), bleeding (clinically relevant, major/minor), all-cause death and cardiovascular death. Tier 2 consideration for most trials: quality of life/patient satisfaction, anatomical location of thrombotic event, anatomical location of bleeding event, compliance, catastrophic APS. Tier 3 consideration for some trials: time in therapeutic range for VKA, significant decrease in haemoglobin level, need for blood transfusion.

Table 2

ORBIT Matrix for assessment of outcome reporting bias in included trials

All three major review outcomes regarding benefit and harm were reported in all of the included trials as shown in the ORBIT Matrix in table 2.

Differences between protocol and review: we prespecified bleeding as a review outcome of interest, but did not distinguish between major, minor, clinically relevant, etc. In the meta-analysis, we chose to arrange the analysis into overall bleeding and major bleeding, to make results more accurate and clinically relevant.

Conclusion

After a comprehensive literature search, this systematic review and meta-analysis summarises all available RCTs for the use of secondary prophylaxis with DOACs versus VKA in patients with APS, including the latest ASTRO-APS Study. We found no clinical value in choosing DOACs over VKA for secondary thrombosis prophylaxis in patients with APS. In fact, DOACs seem to be less effective, especially to those experiencing incident AT. A change from VKA to DOAC does not occur to be beneficial for the patients with APS in terms of risk of bleeding.

However, a subset of patients with APS with only VT history might benefit from DOAC treatment, but this should be addressed by well-designed trials. We suggest the trial outcomes mentioned above as the core outcome set in this area of research and encourage the Outcome Measures in Rheumatology Initiative to further define a core domain set for trials in patients with APS.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.

Acknowledgments

We acknowledge Lasse Østengaard (LØ) from the Research Department of Evidence Based Medicine, University Library of Southern Denmark, Denmark for assisting in developing a literature search strategy.

References

Supplementary materials

Footnotes

  • Contributors JBHA—guarantor, conceptualisation, methodology, formal analysis, investigation, writing and visualisation. RC—methodology, formal analysis and writing. GGMB—validation, writing, reviewing and editing. AV—conceptualisation, validation, investigation, writing and supervision.

  • Funding The Parker Institute, Bispebjerg and Frederiksberg Hospital is supported by a core grant from the Oak Foundation (OCAY-18-774-OFIL).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • 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.