Article Text

Original research
Transradial cerebral angiography becomes more efficient than transfemoral angiography: lessons from 500 consecutive angiograms
  1. D Andrew Wilkinson,
  2. Neil Majmundar,
  3. Joshua S Catapano,
  4. Vance L Fredrickson,
  5. Daniel D Cavalcanti,
  6. Jacob F Baranoski,
  7. Caleb Rutledge,
  8. Andrew F Ducruet,
  9. Felipe C Albuquerque
  1. Department of Neurosurgery, Barrow Neurological Institute, Phoenix, Arizona, USA
  1. Correspondence to Dr Felipe C Albuquerque, Department of Neurosurgery, Barrow Neurological Institute, Phoenix, AZ 85013, USA; neuropub{at}barrowneuro.org

Abstract

Background Transradial arterial access (TRA) for cerebral diagnostic angiography is associated with fewer access site complications than transfemoral access (TFA). However, concerns about increased procedure time and radiation exposure with TRA may slow its adoption. Our objective was to measure TRA rates of success and fluoroscopy time per vessel after ‘radial-first’ adoption and to compare these rates to those obtained with TFA.

Methods We examined 500 consecutive cerebral angiograms on an intent-to-treat basis during the first full year of radial-first adoption, recording patient and procedural characteristics and outcomes.

Results Over a 9-month period at a single center, 457 of 500 angiograms (91.4%) were performed with intent-to-treat via TRA, and 431 cases (86.2%) were ultimately performed via TRA. One patient (0.2%) experienced a temporary neurologic deficit in the TRA group, and none (0%) did in the TFA group (p=0.80). The mean±SD fluoroscopy time per vessel decreased significantly from the first half of the study to the second half for TRA (5.0±3.8 vs 3.4±3.5 min/vessel; p<0.001), while TFA time remained unchanged (3.7±1.8 vs 3.5±1.4 min/vessel; p=0.69). The median fluoroscopy time per vessel for TRA became faster than that for TFA after 150 angiograms.

Conclusion Of 500 consecutive angiograms performed during the first full year of radial-first implementation, 86.2% were performed successfully using TRA. TRA efficiency exceeded that of TFA after 150 angiograms. Concerns about the length of procedure or radiation exposure should not be barriers to TRA adoption.

  • angiography
  • aneurysm
  • artery
  • catheter
  • technique

Data availability statement

No data are available.

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.

Introduction

The benefits of transradial arterial access (TRA) over transfemoral access (TFA) for interventional cardiology, including reduced access-site complications, length of stay, adverse clinical events, and even mortality, have been well-documented.1–3 Although originally described almost 20 years ago,4–7 TRA for cerebral angiography was initially not widely practiced and has only recently been adopted by a number of centers.8–11 TRA for cerebral angiography has been associated with reduced complication rates12 and improved patient satisfaction13 compared with TFA.

Although the rate of TRA adoption among neuroendovascular practitioners has not been well studied, previous surveys of cardiologists identified the perception of increased radiation and procedural time of TRA compared with TFA as the greatest barrier to TRA adoption.14 Within the neurointervention literature, some studies have described the steep initial learning curve for TRA, both institutionally and individually, with 30 to 50 TRA cerebral angiograms described as necessary for proficiency.15–17 However, continued development beyond the initial learning curve, as well as efficiency directly compared with TFA, has not been described. The goal of this study was to evaluate the continued learning curve of TRA for angiography to assess rates of success and efficiency, comparing these rates directly to those for TFA.

Methods

We analyzed 500 consecutive adult diagnostic cerebral angiograms performed over 9 months on an intent-to-treat basis, beginning in July 2019 and concluding in March 2020, during the first full academic year of ‘radial-first’ adoption at a single center. Patient demographics, pathology, procedural details, and access sites were recorded. Patients with planned potential procedures who ultimately received diagnostic angiography only (eg, potential stroke that had recanalized or carotid stent that did not meet criteria for treatment) were not included. Access site decisions were made at the discretion of the operator but were radial first unless known patient or procedural characteristics were judged to warrant femoral access. The reasons for TFA, whether performed with intent-to-treat TFA or after TRA to TFA crossover, were recorded. All angiograms were performed by two senior endovascular fellows, with a combined previous experience of more than 400 endovascular transfemoral cases before the study but no significant transradial experience. The study was approved by the St Joseph’s Hospital and Medical Center Institutional Review Board for Human Research (Phoenix, AZ) and complied with the Health Insurance Portability and Accountability Act. Intended and ultimate access sites, sheath size, use of antiplatelet agents, use of anticoagulant, digital subtraction angiography (DSA) fluoroscopy time, vessels injected, and use of closure device were recorded. Cases were divided a priori into 10 cohorts of 50 consecutive cases each for analysis of the institutional learning curve and into 10 cohorts of 20 consecutive cases for examining the individual operator curve for each of the two neuroendovascular fellows who were the primary operators.

Technique

TRA angiograms were performed preferentially with a 5 French (5F) slender Glidesheath (Terumo Medical Corp, Somerset, NJ), placed under ultrasound guidance in either the proximal or distal radial artery, followed by injection of a cocktail of heparin (5000 U), verapamil (2.5 mg), lidocaine (20 mg), nitroglycerin (200 µg), and 10 mL of aspirated blood. The sheath was not connected to the flush catheter. Radial angiography was performed to verify favorable radial and brachial anatomy. A 5F Simmons-2 Glide catheter (Terumo Medical Corp, Somerset, NJ) was then advanced to the innominate artery and reformatted by first attempting to reformat in the descending aorta, then attempting to reformat by primarily selecting one of the carotids, and lastly reformatting off the aortic valve if the other methods failed. Selective catheterization of the vessels of interest was then performed with the use of a 0.035-inch guidewire, using the double-flush technique. Selective catheterization of internal carotid, external carotid, or vertebral arteries was preferred. Images were occasionally obtained from a more proximal position (ie, common carotid or subclavian injections) if more selective catheterization proved difficult on the initial attempt and a less selective angiogram would suffice clinically. The groin was prepared and draped in all cases to facilitate a rapid conversion to TFA in cases of TRA failure.

Statistical analysis was performed using SPSS Statistics (IBM Corp, Armonk, NY) and Excel (Microsoft, Redmond, WA). Cases were analyzed and compared on an intent-to-treat basis. TRA conversions to TFA were included in the intent-to-treat TRA group. Means with standard deviations and medians with interquartile ranges were reported for continuous variables. Independent t-tests, Mann-Whitney U tests, and χ2 analyses were performed to compare the cohorts. All results with a probability value <0.05 were considered significant.

Results

Among 500 consecutive diagnostic cerebral angiograms, 457 (91.4%) were performed with intent-to-treat via TRA, and 43 (8.6%) were performed with intent-to-treat via TFA. No differences were noted in patient sex (302/457 [66.1%] women vs 155/457 [33.9%] men for TRA and 23/43 [53.5%] women vs 20/43 [46.5%] men for TFA; p=0.13) or mean±SD age (55±16 years for TRA vs 57±15 years for TFA; p=0.26) between the TRA and TFA groups, although the TRA group had a higher number of patients receiving dual antiplatelet therapy than the TFA group (102/457 [22.4%] for TRA vs 4/43 [9.3%] for TFA; p=0.05).

Among all 500 patients, the reasons for angiography were unruptured aneurysm (n=207, 41.4%), arteriovenous malformation or arteriovenous fistula (n=121, 24.2%), subarachnoid hemorrhage (n=57, 11.4%), atherosclerosis (n=44, 8.8%), intracerebral hemorrhage (n=43, 8.6%), moyamoya disease (n=14, 2.8%), tumor (n=12, 2.4%), and unknown (n=2, 0.4%), with no statistically significant differences in pathology between the TRA and TFA groups (p=0.07). Minor vascular access site complications were lower in the TRA group than in the TFA group (2/457 [0.4%] for TRA vs 2/43 [4.7%] for TFA; p=0.04).

No major vascular access site complications occurred in either group. One patient (0.2%) in the TRA group with subarachnoid hemorrhage experienced a temporary neurologic deficit postoperatively that resolved within 1 week, and no temporary neurologic deficits occurred in the TFA group (p=0.80). No patient experienced a permanent neurologic deficit in either group (table 1).

Table 1

Patient and procedural characteristics of 500 consecutive diagnostic angiograms

TRA versus TFA performance

Among cases with intent-to-treat TRA, 26 of 457 (5.7%) crossed over to TFA. There were no crossovers from TFA to TRA. Ultimately, 431 of 500 total cases (86.2%) were performed via TRA, including 222 (51.5% of TRA) via distal radial access, 207 (48.0% of TRA) via proximal radial access, and 2 (0.5% of TRA) via ulnar access (ulnar was included in the TRA group for analysis). The remaining 69 cases (13.8% of total) were performed via TFA (figure 1). The most common reasons for TFA access among the 43 cases with intent-to-treat TFA were the need for left vertebral artery catheterization, judged likely to be difficult via TRA (n=19, 44.2%), anatomy otherwise unspecified (n=9, 20.9%), unknown (n=6, 13.9%), and previously known radial artery occlusion (n=4, 9.3%). The most common reasons for TFA access among the 26 patients who were intent-to-treat TRA but crossed over were failed radial artery access (n=6, 23.0%), inability to catheterize an intracranial vessel of interest (n=4, 15.4%), radial loop (n=4, 15.4%), and previously unknown radial artery occlusion (n=4, 15.4%) (online supplemental table S1).

Figure 1

Flowchart depicting access sites for 500 consecutive cerebral angiograms. TFA, transfemoral access; TRA, transradial access. Reproduced with permission from Barrow Neurological Institute, Phoenix, Arizona.

When examining temporal trends in the total number of cases performed via TRA vs TFA, with the cases divided into 10 consecutive cohorts of 50 cases each, we found no significant temporal trends noted in the number of cases performed with intent-to-treat TFA, or that crossed over from TRA to TFA (online supplemental figure S1). When the total number of cases was divided into subgroups of the first and second 250 cases, there was no significant difference in the number of cases performed with intent-to-treat TFA in the first versus the second half of the cases (n=20, 8.0% for the first half vs n=23, 9.2% for the second half; p=0.63), or the number of cases that crossed over from TRA to TFA (n=15, 6.0% for the first half vs n=11, 4.4% for the second half; p=0.42). Likewise, the total number of TFA cases was virtually unchanged from the first 250 cases to the second (n=35, 14.0% for the first half vs n=34, 13.6% for the second half; p=0.90).

DSA fluoroscopy times

Fluoroscopy time per vessel was calculated for each cohort of 50 consecutive patients, with median times for each cohort shown in figure 2. Median fluoroscopy time per vessel decreased from each cohort to the next, with two exceptions (cohorts 4–5 (cases 151–200 vs 201–250) and cohorts 9–10 (cases 401–450 vs 451–500)). The median time per vessel was significantly greater in the first half of the series than the second (3.8 min/vessel, IQR 2.8–5.7 vs 2.6 min/vessel, IQR 2.0–3.7; p<0.001). The mean time per vessel was significantly higher in the first half of the series than the second (4.9±3.7 min/vessel for cohorts 1–5 vs 3.4±3.4 min/vessel for cohorts 6–10; p<0.001). The median time per vessel for the first cohort was over twice that of the last (5.2 min/vessel vs 2.3 min/vessel).

Figure 2

Median TRA + TFA (IQR) fluoroscopy time (min) per vessel. TFA, transfemoral access; TRA, transradial access. Reproduced with permission from Barrow Neurological Institute, Phoenix, Arizona.

Cases were separated and analyzed by intent-to-treat via TFA and TRA groups, as shown in online supplemental figure S2. Median fluoroscopy times for TFA cases stayed relatively stable throughout the study time, with some variations apparent in median times because of low case numbers (online supplemental figure S2A). TRA cases showed the largest reduction in fluoroscopy time from cohorts 1 to 2, although they continued to drop substantially throughout the rest of the study period (online supplemental figure S2B). Mean time per vessel for TFA cases was similar in the first half of the series to that in the second half (3.7±1.8 min/vessel for cohorts 1–5 vs 3.5±1.4 min/vessel for cohorts 6–10; p=0.69). Times for TRA cases decreased significantly from the first half of the series to the second (5.0±3.8 min/vessel for cohorts 1–5 vs 3.4±3.5 min/vessel for cohorts 6–10; p<0.001). When comparing techniques for the whole series, we found that mean TFA times were slightly lower than TRA (3.6±1.6 min/vessel for TFA vs 4.3±3.6 min/vessel for TRA; p=0.03), whereas median TFA times were equivalent to TRA (3.3 [IQR 2.6–5.3] min/vessel for TFA vs 3.1 [IQR 2.3–4.9] min/vessel for TRA; p=0.16). However, if cases of TRA conversion to TFA were excluded, TFA and TRA times were equivalent (3.6±1.6 min/vessel for TFA vs 4.1±3.2 min/vessel for TRA; p=0.12). Furthermore, although mean time per vessel was higher for TRA for the first half of the series (5.0±3.8 min/vessel for TRA vs 3.7±1.8 min/vessel for TFA; p=0.01), times were equivalent in the second half of the series (3.4±3.5 min/vessel for TRA vs 3.5±1.4 min/vessel for TFA; p=0.86). When plotted over time against the median overall TFA time, median TRA fluoroscopy time became faster than median TFA time from angiograms 151–200 and were faster than TFA for all but one of the remaining six cohorts (figure 3).

Figure 3

Transradial access (TRA) versus median transfemoral access (TFA) fluoroscopy time (min) per vessel. Reproduced with permission from Barrow Neurological Institute, Phoenix, Arizona.

Fluoroscopy time for the individual operators was also examined for the two fellows who were the primary operators for the cases. Operator A was the primary operator for 228 angiograms, while operator B was the primary operator for 272. Individual plots were created for the fluoroscopy time for the first 200 angiograms of each operator cohort, divided into 10 cohorts of 20 angiograms each. Both operators showed significant decreases in TRA fluoroscopy times, as shown in online supplemental figure S3, with consistently lower median fluoroscopic times than TFA after 60–100 individual angiograms for each operator. When compared with each other, both operators’ fluoroscopy times decreased in a similar manner, with operator A having lower times for cohorts 1, 2, 6, and 10 and operator B having lower times for cohorts 3–5 and 7–9. Overall median times for the first cohort were almost twice that of the last cohort for both fellows (5.1 vs 2.5 min/vessel for operator A; 5.8 vs 3.3 min/vessel for operator B). The mean fluoroscopy time per vessel for the first half of the series was significantly higher than for the second half of the series for both operators (5.6±4.5 vs 4.0±4.2, p=0.02 for operator A; 5.0±3.4 vs 3.3±1.8, p<0.001 for operator B). To examine the continued efficiency beyond the initial learning curve, we also examined fluoroscopy times after excluding the first 40 angiograms for each operator, finding that times continued to decrease significantly with continued experience from angiograms 41–120 vs 121–200 for each operator (5.4±5.1 vs 3.8±4.4 min/vessel, p=0.04 for operator A; 4.0±2.6 vs 3.3±1.7 min/vessel, p=0.03 for operator B).

Selective catheterization

To examine the selectivity of catheterization, a review of specific vessels catheterized was performed for 50 cases, including 25 from the first cohort and 25 from the last cohort. The exact vessel catheterized was recorded and classified as either selective (internal carotid artery, external carotid artery, vertebral) or nonselective (common carotid artery or subclavian). For angiograms imaging the right carotid artery, selective catheterization was performed in 27 of 37 (72.9%) angiograms. For the left carotid, selective catheterization was performed in 30 of 35 (85.7%) angiograms. For the right vertebral artery, selective catheterization was performed in 27 of 27 angiograms, and for the left vertebral artery, selective catheterization was performed in 20 of 22 (92.8%) angiograms. When comparing the first and last cohorts, 51 of 61 vessels (83.6%) were selectively catheterized in the first cohort sample, compared with 53 of 60 vessels (88.3%) in the last cohort sample (p=0.45).

Discussion

Although it has been almost 20 years since TRA for neuroendovascular procedures was first described, most cases are still performed via TFA. The cardiology experience suggests that even after introducing TRA and several studies suggesting better outcomes, as well as guidelines advocating its use preferentially,18 TRA adoption was and continues to be inconsistent, with significant variation in usage rates across operators and institutions.19 Some cardiology studies suggest that TRA continues to be associated with a modest degree of crossover to TFA and slightly longer procedural times than TFA.20 TRA has also been associated with a small but significant increase in radiation exposure in cardiology.21 The perception of increased radiation exposure and procedural length were cited in one study as the greatest barriers to TRA adoption among cardiologists surveyed.14 However, increasing operator experience with TRA in cardiology brought lower crossover rates, lower contrast doses, and lower fluoroscopy times.22 For neurovascular cases, increased fluoroscopy time has been associated with a higher rate of angiography-associated emboli causing MRI changes.23 Thus, reducing fluoroscopy time and demonstrating equivalency or improvement over TFA are paramount if TRA is to be more widely adopted for neuroendovascular cases.

We found that, over time, the rates of success with TRA performance remained relatively stable, with stable rates of intent-to-treat via TFA and stable rates of TRA to TFA crossover, such that almost 90% of cases were successfully performed via TRA. Efficiency at DSA performance, however, continued to improve over the entire 500 angiograms. This increased efficiency (as measured by fluoroscopy times) was due to improvement in TRA, but not TFA, performance. Increased efficiency was noted for TRA performance both at the institutional level and at the individual operator level. TFA was faster than TRA in the first part of the series, but TRA became more efficient institutionally after 150 angiograms, and it became more efficient per primary operator after about 75 angiograms. As expected, vascular access site complication rates were higher for TFA than for TRA.

Several factors may account for continued TRA efficiency gains while TFA efficiency remained stable. TRA access and sheath placement may be more technically challenging at first, especially when attempting distal radial access, a site that may give a greater chance of repeat access success24 and was used in over half of our TRA procedures. During the first full year of TRA training, techniques which were not previously known institutionally were acquired. For example, selection of the left internal carotid before the left external carotid and of the right external carotid before the right internal carotid may speed transitions between vessels. TRA also encompasses unique technical challenges that are not as prominent in TFA, such as catheter formation and reformation techniques. Learning optimal sequences to improve efficiency and reduce the number of needed guidewire manipulations, acquiring techniques for left vertebral artery catheterization, and maintaining a reverse catheter engaged in the great vessels can all be challenging and are somewhat unique to TRA. Furthermore, the larger number of TRA procedures in our series, compared with TFA procedures, may have allowed for more efficiency gains. Although we did not record procedural times, which could have found delays early on in nonfluoroscopic portions of the procedure, such as set up and access, we found that once the operators and, perhaps more importantly, the nurses and technicians became adept at the setup of the patient and room for TRA cases, nonfluoroscopy procedural times quickly became similar between TRA and TFA cases. Because we routinely prepare the groin and use ultrasound for every TRA case, even crossover from TRA to TFA (which accounted for only 5% of cases in this study) can be accomplished quickly.

In our experience, reasons for failure of radial access may be quite different from reasons for failure of catheterization. Femoral access difficulty is often due to femoral atherosclerotic disease and correlates well with difficulty in catheterization, likely because vascular disease of the femoral artery and aortic arch share common risk factors (hypertension, hyperlipidemia, and age) and often result in common conditions such as tortuosity and calcification, which likewise make catheterization more difficult. In contrast, access from a radial approach may in fact be most difficult in young, anxious patients, with vessels especially prone to spasm, but whose aortic arch and great vessels may be relatively straight and not particularly hard to navigate.

Overall, our results show that we were able to achieve high rates of selective catheterization using TRA (over 80%), and we would expect those numbers to improve with experience. Although exact reasons for nonselective (eg, common carotid or subclavian) catheterization could not be discerned retrospectively from the records, in general our approach for the carotid circulation was to attempt selective catheterization but not spend excess time attempting selective catheterization if it proved difficult on the initial attempt and if a common carotid angiogram would suffice clinically. For the posterior circulation, if only one vertebral artery was needed, we generally favored the right, and we were usually successful (success rate 92.8%) in attempts when left vertebral catheterization was judged both necessary clinically and feasible on the basis of preoperative imaging. We note that 21 of 500 total angiograms (4.1%) were performed for TFA due to the preoperative impression that left vertebral catheterization was necessary and would be easier via a femoral approach.

Previous studies have examined the early TRA learning curves in neurovascular procedures. One study examined individual operators, finding a significant reduction in fluoroscopy time per vessel catheterized when comparing angiograms 1–5 to angiograms 11–15.8 Almallouhi et al found a rapid reduction in the number of TRA crossovers to TFA over the first 6 weeks of TRA implementation.25 Zussman et al examined their institution’s first 100 cases of TRA, finding higher rates of achieving their primary outcome in their second 50 angiograms compared with the rates in their first 50 angiograms (98% vs 88%), with a lower crossover rate (2% vs 8%).16 17 Another center examined their first 121 TRA cases, including diagnostic and interventional cases, finding a decrease in total procedural time after the first quarter, with subsequent quarters that continued to decrease.26 In a study comparing 206 TRA procedures to 844 performed via TFA, fluoroscopy times averaged 4 min less for TFA interventional cases, whereas TRA times decreased with time.12 These studies lead to the conclusion that 30–50 cases were needed to progress past the steepest part of the learning curve. Our study suggests increasing efficiency throughout the learning curve, meaning operators who may be reluctant to adopt TRA because of concerns about efficiency, safety, and radiation can expect to maintain, and even exceed, their TFA efficiency with continued TRA use.

Limitations

Our study examined fluoroscopy time and did not evaluate total procedural time, meaning any procedural time not occurring with live fluoroscopy was not recorded, including time spent gaining arterial access, as well as time between runs, such as while setting up runs or interpreting data. We hypothesized that increasing experience with TRA would be associated with reduced fluoroscopy times, further lowering the procedural time. Similarly, we measured fluoroscopy times and not direct radiation exposure, although fluoroscopy time is a strong predictor of radiation dose. We also examined learning curves during our institution’s first full year of TRA utilization, so that the learning curves incorporated not only operator learning but also institutional learning. Further trainees may benefit from previous institutional knowledge and may consequently shorten their individual learning curves. Although both primary operators in this study had significant TFA experience before adoption of TRA (more than 400 cases combined), they were still trainees, and their relative lack of TFA experience compared with more senior practitioners adopting TRA may have affected results. Additionally, the allocation of treatment (TRA vs TFA) on the basis of operator discretion instead of randomization certainly introduces bias and does not provide a direct comparison of the techniques for every case. However, we feel that it does provide a practical analysis of real-world pragmatic conditions, where the treatment decisions will be customized to each patient to optimize both safety and efficiency.

Conclusions

Most angiograms (86.2%) were successfully completed via TRA during the first 500 angiograms of the radial-first training program. The rate of TFA performance or crossover to TFA did not change with time, and TRA efficiency exceeded that of TFA after 150 angiograms. Concerns about procedure time or radiation exposure should not be barriers to TRA adoption.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

Acknowledgments

DAW is currently with the Department of Neurosurgery, Penn State Health Milton S Hershey Medical Center, Hershey, PA. DDC is currently with St Vincent’s Medical Center, Ayer Neuroscience Institute, Bridgeport, CT. The authors thank the staff of Neuroscience Publications at Barrow Neurological Institute for assistance with manuscript preparation.

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

  • Contributors All authors made substantial contributions to the conception or design of the work. DAW: study design, data collection, data analysis, manuscript editing and writing, and statistical analysis. NM: study design, data collection, manuscript editing and writing. JSC: data collection and manuscript editing. VLF: data collection and manuscript editing. DDC: data collection and manuscript editing. JFB: manuscript editing and data analysis. CR: data collection and manuscript editing. AFD: study design and manuscript editing. FCA: study design, manuscript editing, and guarantor.

  • 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 AFD is a consultant for Stryker (Kalamazoo, MI), Cerenovus (Johnson & Johnson, New Brunswick, NJ), Medtronic (Dublin, Ireland), Penumbra (Alameda, CA), and Koswire, Inc (Flowery Branch, GA); and serves on the editorial board of Journal of NeuroInterventional Surgery. FCA serves on the editorial board of Journal of NeuroInterventional Surgery.

  • 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.