Article Text

Original research
Comparison of Pipeline embolization device versus Tubridge embolization device in unruptured intracranial aneurysms: a multicenter, propensity score matched study
  1. Chi Huang1,
  2. Gengwu Ma1,
  3. Xin Tong2,3,
  4. Xin Feng1,
  5. Zhuohua Wen1,
  6. Mengshi Huang1,
  7. Anqi Xu1,
  8. Hao Yuan1,
  9. HongYu Shi1,
  10. Jiancheng Lin1,
  11. Can Li1,
  12. Runze Ge1,
  13. Jiwan Huang1,
  14. Chao Peng4,
  15. Yajun Zhu5,
  16. Tao Wang6,
  17. Changren Huang6,
  18. Zongduo Guo5,
  19. Shuyin Liang1,
  20. Shixing Su1,
  21. Xin Zhang1,
  22. Xifeng Li1,
  23. Aihua Liu2,3,
  24. Chuan-Zhi Duan1
    1. 1Neurosurgery Center, Department of Cerebrovascular Surgery, Engineering Technology Research Center of Education Ministry of China on Diagnosis and Treatment of Cerebrovascular Disease, Zhujiang Hospital of Southern Medical University, Guangzhou, Guangdong, China
    2. 2Neurointervention Center, Beijing Neurosurgical Institute, Beijing, Beijing, China
    3. 3Neurointervention Center, Beijing Tiantan Hospital, Beijing, Beijing, China
    4. 4Department of Neurosurgery, Guangdong Provincial People's Hospital, Guangzhou, Guangdong, China
    5. 5Department of Neurosurgery, The First Affiliated Hospital of Chongqing Medical University, Yuzhong, Chongqing, China
    6. 6Department of Neurosurgery, The Affiliated Hospital of Southwest Medical University, Southwest Medical University, Luzhou, Sichuan, China
    1. Correspondence to Dr Chuan-Zhi Duan, Neurosurgery Center, Department of Cerebrovascular Surgery, Engineering Technology Research Center of Education Ministry of China on Diagnosis and Treatment of Cerebrovascular Disease, Zhujiang Hospital of Southern Medical University, Guangzhou, Guangdong, 510280, China; doctor_duanZJ{at}; Dr Aihua Liu, Beijing Neurosurgical Institute, Beijing Tiantan Hospital, Capital Medical University, Beijing, Beijing, 100070, China; liuaihuadoctor{at}


    Background Flow diverter devices (FDs) are increasingly used for treating unruptured intracranial aneurysms (UIAs), but limited studies compared different FDs.

    Objective To conduct a propensity score matched analysis comparing the Pipeline embolization device (PED) and Tubridge embolization device (TED) for UIAs.

    Methods Patients with UIAs treated with either PED or TED between July 2016 and July 2022 were included. Propensity score matching was performed to adjust for age, sex, comorbidities, smoking, drinking, aneurysm size, morphology, neck, location, parent artery diameter, adjunctive coiling, and angiographic follow-up duration. Perioperative complications and clinical and angiographic outcomes were compared after matching.

    Results 735 patients treated by PED and 290 patients treated by TED were enrolled. Compared with the PED group, patients in the TED group had a greater number of women and patients with ischemia, a smaller proportion of vertebrobasilar and non-saccular aneurysms, a smaller size and neck, and fewer adjunctive coils and overlapping stents, but a larger parent artery diameter and lumen disparities. After adjusting for these differences, 275 pairs were matched. No differences were found in perioperative complications (4.4% vs 2.5%, P=0.350), in-stent stenosis (16.0% vs 15.6%, P>0.999), or favorable prognosis (98.9% vs 98.5%, P>0.999). However, PED showed a trend towards better complete occlusion over a median 8-month angiographic follow-up (81.8% vs 75.3%, P=0.077).

    Conclusion Compared with PED, TED provides a comparable rate of perioperative and short-term outcomes. Nevertheless, a better occlusion status in the PED group needs to be further verified over a longer follow-up period.

    • Aneurysm
    • Flow Diverter
    • Vascular Malformation

    Data availability statement

    Data are available upon reasonable request. The data supporting this study’s findings are available from the corresponding author upon reasonable request.

    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:

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


    • The Pipeline embolization device (PED) and Tubridge embolization device (TED) are two differently designed flow diverter devices (FDs) widely used for the treatment of unruptured intracranial aneurysms (UIAs). The PED is composed of cobalt–chromium–nickel alloy, while the TED is made of nitinol.


    • In this propensity score matched study, the perioperative and short-term outcomes of PED and TED treatments for UIAs were comparable.

    • Compared with the TED, the PED was more frequently used for vertebrobasilar and non-saccular aneurysms, or aneurysms with a larger size and wider neck but a smaller parent vessel diameter and lumen disparities.


    • This study provides valuable insight into the short-term outcomes of two different designs of FDs and serves as a reference for neurointerventionalists when selecting an appropriate FD for intracranial aneurysm treatment.


    With the advancement of endovascular techniques, flow diverter devices (FDs), represented by the Pipeline embolization device (PED; Medtronic, Minneapolis, USA), have emerged as a new option for treating unruptured intracranial aneurysms (UIAs). Although all FDs share the same mechanism of vessel reconstruction, there may be concerns about the potential differences intrinsic to the material and structure of various products. However, few comparative studies evaluating different FDs have been published.1 2 The PED is composed of a braided cobalt–chromium–nickel mesh and functions by impeding the flow within the aneurysm, thereby progressively leading to its occlusion.3 4

    Additionally, the Tubridge embolization device (TED; MicroPort NeuroTech, Shanghai, China) has also gained widespread application. The TED is a self-expandable braided stent with flared ends designed by a nickel–titanium alloy and has been reported to exhibit favorable efficacy in treating UIAs with the advantages of super-elasticity and shape-holding memory.5 6 The composition of the PED consists of a cobalt–chromium alloy, which enhances its visibility by increasing stiffness.7 Conversely, the nickel–titanium design of TED allows for more flexibility, but provides a weaker radial force, making it susceptible to shortening due to better vascular adaptation.8 9 Although the two most commonly used commercial products serve the same purpose, their performance properties and therapeutic indications are not exactly same due to disparities in manufacturing processes.6 10 11 Therefore, further evaluation is imperative to scrutinize potential nuances in efficacy between these different devices.

    Although previous studies have conducted a comparative analysis on the treatment of UIAs with the two devices,8 12 valid comparability was limited due to variations in neurointerventionalists' experience, their small sample size, and heterogeneity in demographic and aneurysmal characteristics. Hence, we performed a multicenter analysis to assess the efficacy of these two FDs for UIAs. Propensity score matching (PSM) was employed to mitigate the confounding bias commonly encountered in observational studies.


    Study design

    We conducted a retrospective analysis of a prospectively maintained database at five advanced stroke centers, all of which performed more than 100 FD treatments annually (online supplemental table S1). Consecutive patients treated with a first-ever FD between July 2016 and July 2022 were retrospectively reviewed. Demographics and aneurysm characteristics were collected prospectively at admission.

    Supplemental material

    The inclusion criteria included the following: (1) age 18 to 80 years; (2) UIAs treated with PED (Classic/Flex) or TED. The exclusion criteria included the following: (1) ruptured aneurysms; (2) aneurysms that had previously received endovascular or microneurosurgical treatment; (3) tandem treatment (multiple aneurysms covered by a single FD); (4) poor image quality; (5) presence of other cerebrovascular disease such as arteriovenous malformations, arteriovenous fistulas, or Moyamoya disease. Additionally, since the primary goal of this study was short-term angiographic outcomes, patients without angiographic follow-up were excluded. Ultimately, we enrolled a total of 1025 patients (figure 1). This study received approval from the institutional review boards of all participating centers, and informed consent was waived owing to the anonymous design of the study.

    Figure 1

    Flow diagram of patient selection. PED, Pipeline embolization device; TED, Tubridge embolization device.

    Endovascular procedures

    Before the procedure, patients received dual antiplatelet therapy consisting of daily aspirin 100 mg and clopidogrel 75 mg for at least 3 days. All patients underwent thromboelastography or platelet function testing at least 1 day before the scheduled operation. Ticagrelor was used for alternative antiplatelet management for individuals non-responsive to clopidogrel. General anesthesia and full heparinization were performed during the operation for all patients. Dual antiplatelet therapy was continued for 6–12 months after the procedure, followed by single antiplatelet therapy with daily aspirin 100 mg for an additional 6–9 months.

    The transfemoral arterial approach using a triaxial system was commonly selected as the standard method to access the target aneurysm: a 6/8 Fr long sheath (Neuron MAX, Penumbra, California, USA) was employed; 5/6 Fr intermediate catheter: Navien (Medtronic) or Passageway (Precision Medical, Nanjing, China) was inserted into the desired artery using a guidewire; based on the path diagram, a 0.014″ microwire was employed to navigate the microcatheter: Marksman (Medtronic) or Fastrack (MicroPort NeuroTech) through the aneurysm neck and into the farthest-reaching normal vessel for FD delivery and deployment.

    The decision on stent selection was based on several factors, including the experience and preferences of the participating centers. In certain scenarios, such as aneurysms with a larger vessel diameter, surgeons preferred selecting a TED owing to its broader vascular adaptability. The following conditions were considered for adjunctive coiling: (1) there was a risk of stent shortening or migration; (2) a high risk of recurrence or rupture due to rapid blood flow velocities at the neck. For large aneurysms with a wide neck, overlapping stent techniques were used to enhance local metal coverage.

    Data collection and follow-up outcomes

    Demographic characteristics included age, sex, comorbidities, smoking and drinking habits, and preoperative modified Rankin Scale (mRS) score. Two authors assessed morphological features, including location, morphology, aneurysm size, neck width, the parent vessel diameter (the mean diameter of the vessel at the proximal and distal ends of the aneurysms), and lumen disparities (the absolute value of the difference between the proximal and distal parent diameter). Three neurointerventionalists with more than 10 years of experience supervised these measurements. Procedural details were collected from the electronic medical record system, including the type of FDs, coiling assistance, and overlapping stent technique. Perioperative complications and device-related challenges were also evaluated. Perioperative complications were defined as intraoperative rupture of the aneurysms, in-stent thrombosis, postoperative subarachnoid hemorrhage and intraparenchymal hemorrhage, transient ischemic attack (a transient neurological deficit without corroborative imaging), mild ischemia (a change in the National Institutes of Health Stroke Scale (NIHSS) score of ≤4 lasting for <7 days), and severe ischemia (a change in the NIHSS score of >4 lasting for >7 days).13 Device-related challenges including stent re-apposition, poor wall apposition (needed balloon assistance), stent shortening or migration, and intraoperative perforation.

    The initial follow-up for digital subtraction angiography was scheduled to occur 3–6 months after the procedure. Subsequently, angiography was performed every 6–12 months, using various imaging techniques, including digital subtraction angiography, CT angiography, and MR angiography. Aneurysm occlusion status was classified into three groups according to the O’Kelly-Marotta classification: complete occlusion (D, entirely non-filling), near-complete occlusion (C, only entry remnant), and incomplete occlusion (A and B, subtotal and total filling).14 Aneurysm occlusion was defined as complete or near-complete occlusion. In-stent stenosis (ISS) referred to a narrowing of the FD diameter exceeding 25% without pre-existing significant artery stenosis; symptomatic ISS denoted cases with associated symptoms. Clinical follow-up was conducted by face-to-face outpatient or telephone follow-up. Based on the mRS, clinical outcomes were categorized as favorable (0–2), morbidity (3–5), and mortality (6).15

    Statistical analyses

    All statistical analysis was performed using R studio software (version, 4.2.2). Normality of continuous variables was assessed with Kolmogorov-Smirnov test. Continuous variables are expressed as the mean and SD for normally distributed variables, and as median and IQR for non-normally distributed variables. These variables are tested using the t-test or Wilcoxon signed-rank test. Categorical variables are reported as frequency and proportions, and analyzed using the Χ2 test or Fisher’s exact test. Additionally, PSM analysis employing the nearest-neighbor method was conducted to adjust for potential confounders between patients treated with the PED or TED. A 1:1 matching without replacement was performed based on a caliper defined as 0.1 SD. Statistical significance for all tests was considered at P<0.05.


    Demographic and aneurysm characteristics

    A total of 1025 consecutive patients with 1025 UIAs were enrolled in the study, with a median age of 55 years (IQR 46–62). Among them, 735 (71.7%) patients received treatment with PED, while 290 (28.3%) patients underwent TED treatment. The baseline characteristics are detailed in table 1. Compared with TED, the PED was used more frequently for vertebrobasilar (16.2% vs 3.1%, P<0.001), non-saccular aneurysms (17.6% vs 5.9%, P<0.001), larger size (7.59 mm vs 4.40 mm, P<0.001) and neck width (5.45 mm vs 4.18 mm, P<0.001), but less frequently for a larger parent vessel diameter (3.71 mm vs 4.00 mm, P<0.001) and lumen disparities (0.60 mm vs 0.98 mm, P<0.001). Other baseline characteristics, except for gender, ischemic stroke history, and smoking and drinking habit, were similar.

    Table 1

    Comparison for the type of flow diverter device used before propensity score matching

    Procedure details

    All 1025 procedures were successfully performed, and 97.0% of FDs were successfully implanted in the initial attempt. Balloon assistance was used in only 10 cases with poor apposition (1.0%). A higher proportion of patients in the PED group received adjunctive coiling (32.9% vs 14.1%, P<0.001). Additionally, use of the overlapping stent technique was more prevalent in the PED group than in the TED group (6.4% vs 2.1%, P=0.008). The rate of immediate complete occlusion was lower in the TED group than in the PED group (6.6% vs 12.0%, P=0.015).

    Device-related challenges and perioperative complications

    The rate of device-related challenges was 3.9% for PED and 4.8% for TED. Stent re-apposition, poor wall apposition, stent shortening, stent migration, and intraoperative perforation were comparable for PED and TED. A 3.2% (33/1025) overall rate of perioperative complications was observed. Twenty-five patients developed perioperative complications in the PED and eight in the TED group. The eight cases of in-stent thrombosis received intra-arterial tirofiban injection, followed by postoperative intravenous tirofiban infusion for 24 hours and dual antiplatelet bridging therapy. Of these eight cases, postoperative ischemia was observed in only two patients, while one patient experienced a transient ischemic attack (online supplemental table S2). Indeed, both groups had similar rates of device-related challenges and perioperative complications (table 1).

    Short-term follow-up outcomes

    The PED group had a shorter angiographic follow-up duration than the TED group (median 7 vs 9 months, P<0.001). Aneurysm occlusion (complete/near-complete) rates did not differ between the two devices, but there was a trend towards more complete aneurysm occlusion in the PED group than the TED group (80.5% vs 75.5%, P=0.089). No significant differences were observed in ISS (17.1% vs 16.6%, P=0.893) and symptomatic ISS (1.9% vs 1.7%, P>0.999). Additionally, no aneurysm recurrence was observed in the present study.

    As for clinical outcomes, the follow-up durations were the same between the two groups (median 18 vs median 18 months, P=0.503). There was no statistical difference in the rate of favorable prognosis (98.9% vs 98.3%, P=0.610) and surgery-related morbidity (0.4% vs 0.7%, P=0.625). No mortality was observed during this period.

    Propensity score matching analysis

    After adjusting for age, sex, hypertension, diabetes, hyperlipidemia, coronary disease, ischemic stroke history, smoking and drinking habits, location, morphology (saccular or non-saccular), aneurysm size, neck width, parent vessel diameter, adjunctive coiling, and angiographic follow-up duration, matching was successful for 275 PED and TED treatment pairs in the PSM cohort (table 2). After matching for baseline characteristics, only the parent vessel diameter (3.74 mm vs 4.02 mm, P<0.001) and lumen disparities (0.59 vs 0.97, P<0.001) remained significantly different between the two groups. There were no significant differences in perioperative and short-term outcomes, but elevated occurrence of transient ischemic attack in the PED group was a matter of concern (2.2% vs 0, P=0.030). The rate of aneurysm occlusion (86.5% vs 86.9%, P>0.999), ISS (16.0% vs 15.6%, P>0.999), and favorable prognosis (98.9% vs 98.5%, P>0.999) were comparative in the matched cohort. However, there was a slightly higher trend towards complete aneurysm occlusion within the PED group compared with the TED group (81.8% vs 75.3%, P=0.077).

    Table 2

    Comparison for the type of flow diverter device used after propensity score matching


    With the recent advancements in interventional devices, an increasing number of centers have adopted newly designed FDs for several neurovascular treatments. However, the rapid transition to newer devices might introduce unknown risks for patients. To our knowledge, this study represents the first propensity-matched analysis comparing treatment outcomes between PED and TED for patients with UIAs. By adjusting for confounding factors, this study provides compelling evidence supporting comparable short-term treatment outcomes between PED and TED. Despite sharing a similar mechanism of flow diversion, it is important to explore both the similarities and differences between the two devices due to variations in material composition and design conception.

    Our findings demonstrate that the rate of incomplete occlusion, ISS, and favorable clinical prognosis were similar between the two devices. These results align with previously published studies.4 8 16 17 Additionally, we also found a tendency for better occlusion (only for complete) results with the PED, which has been observed in prior studies.12

    Comparison with previous studies of PED for UIAs

    In the past decade, numerous studies have assessed the flow diversion treatment of UIAs with PED.3 18–20 The rate of complete occlusion with PED treatment has been proved to be 80–85% in several large prospective trials.3 17 21 Similarly, a large retrospective study of 1171 patients showed that rates of occlusion and ISS were 81.4% and 8.4%, respectively.4 However, perioperative complications were noted in approximately 12.8% of cases. Our findings were consistent with previously published studies except for a lower complication rate (3.4%).

    Comparison with previous studies of TED for UIAs

    The TED is currently not available in the USA but is widely used elsewhere. TED differs from the PED, in that it is a braided self-expanding stent composed of flared ends made from nickel–titanium alloy material. Owing to its good therapeutic effect, it has been demonstrated to be a reliable FD with clinical evidence.6 16 22 Recently, several studies with limited sample sizes have reported favorable outcomes when employing TED for UIAs. Zhang et al23 reported that complete occlusion was achieved in 71.4% of aneurysms without perioperative complications. Another case series involving 77 intracranial aneurysms treated with TED revealed that perioperative adverse events occurred in only six patients (10.5%). With three cases lost to follow-up, all remaining aneurysms displayed occlusion within 1 year and only one case of intra-stent stenosis was observed (1.4%).5 In our study encompassing the therapy of 290 UIAs using TED devices, the rates of operative complications and complete aneurysm occlusion were 3.2% and 79.1%, respectively, which is consistent with published studies.6 16 24

    Comparison with previous studies comparing PED with TED for UIAs

    PED provides stronger radial force to ensure good wall apposition in a tortuous vessel,9 25 whereas TED offers a wider range of size options, resulting in better adaption for larger-diameter target arteries.6 Our results showed that PED may be more suitable for cases characterized by uniform or smaller vessel diameters, whereas TED may be recommended for arteries with larger diameter or large distal and proximal lumen disparities. As shown in table 2, both parent artery diameter (3.74 mm vs 4.02 mm, P<0.001) and lumen disparities (0.59 mm vs 0.97 mm, P<0.001) remain significantly larger in the TED group after matching.

    Prior to our study, rare studies compared PED with TED in intracranial aneurysms. A previous study with a relatively small sample size found no significant difference in treatment outcomes between the two groups.8 However, in a study focusing on posterior circulation aneurysms, it showed that patients treated with TED had a higher rate of periprocedural complications (15.9% vs 3.5%, P=0.001)12; this trend was not observed in our study. Our results showed comparable rates of perioperative complications, aneurysm occlusion, ISS, and favorable clinical prognosis between the two FDs for treating UIAs. Possibly, the same design conception and procedural techniques might allow operators to accumulate experience from vast PED treatments, thereby enhancing the success rate of TED treatment.

    Aneurysm occlusion after FD treatment

    The rate of immediate complete occlusion in the PED group was significantly higher than that in the TED group (12.0% vs 6.6%, P = 0.015) before matching. This disparity may be attributed to a greater number of patients receiving coiling assistance during PED treatment. However, both groups exhibited similar rates of aneurysm occlusion (86.5% vs 86.6%, P>0.999) at median follow-ups of 7 and 9 months, respectively. After propensity score matching, with a median follow-up of 8 months, there is a tendency for better complete occlusion in PED group (81.8% vs 75.3%, P=0.077). Indeed, near-complete aneurysm healing was significantly higher in the TED group (6.0% vs 11.0 %, P=0.008) and there was no difference in aneurysm occlusion (complete/near-complete occlusion) between the two groups (86.5% vs 86.9%, P>0.999). As is shown in previous studies, aneurysms generally exhibit rapid healing within 1 year following initial FD treatment.6 17 26 27 Thus, longer-term follow-up is necessary for further assessment.

    In-stent stenosis after FD treatment

    In this study, the stenosis rates of PED and TED were similar (17.1% vs 16.6%, P = 0.893), consistent with previously reported rates of 10–20%.12 18 24 28 The majority of patients with ISS are asymptomatic, with only a small number (19/174, 10.9%) experiencing symptoms. Moreover, our results also confirmed that FDs have a higher rate of ISS compared with conventional stents,29 30 which remains a clinical issue requiring attention and resolution. Veteran surgeons in our centers ensured that good wall apposition helped to overcome several differences in stent material and design. Thus, it is uncertain whether the nickel–titanium TED reduces the occurrence of ISS compared with the cobalt–chromium PED through its good vascular adaption. Further prospective investigation with longer duration and larger sample size is required to prove this property.


    The main advantage of our work is that it conducts a comparative analysis, enrolling the largest sample size of intracranial aneurysms treated with different FDs. Our findings offer valuable insights into the short-term outcomes between two differently designed FDs.

    However, it is important to acknowledge several noteworthy limitations in our research. First, although the multicenter design ensures broad representation, potential bias might arise due to the retrospective nature of the study and variations in perioperative management among different centers. Second, our follow-up time was not long enough, and dynamic changes in complete occlusion and ISS should be observed over a long follow-up period. Third, despite controlling for relevant confounders through propensity score matching, there remains a possibility of selection biases between the two devices. Given that PED has been on the market for longer than TED, increased operator proficiency with PED might influence stent selection. With increased surgical experience in TED application, future studies might reveal a more balanced distribution between the two devices. Thus, a prospective trial including a large number of random samples and long-term follow-up is required to verify our conclusions.


    In this propensity score matched study, the clinical and angiographic outcomes of the PED and TED were found to be comparable. At the short-term follow-up, comparative analysis of PED and TED for the treatment of UIAs did not identify any significant differences in total perioperative complications, in-stent stenosis, and clinical prognosis. Nevertheless, a better occlusion status in the PED group needs to be further verified via randomized controlled trials with a longer follow-up period.

    Data availability statement

    Data are available upon reasonable request. The data supporting this study’s findings are available from the corresponding author upon reasonable request.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and was approved by the institutional review board of Zhujiang Hospital (2023-KY-023-02), Beijing Tiantan Hospital, Capital Medical University (KY2022-102-02), Guangdong Provincial People's Hospital (KY-Q-2022-344-02), the First Affiliated Hospital of Chongqing Medical University (CY2023-067-01), and the Affiliated Hospital of Southwest Medical University (KY2023219).


    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.


    • CH and GM are joint first authors.

    • CH and GM contributed equally.

    • Contributors All authors contributed to the paper, satisfying the ICMJE guidelines for authorship. ChiH: designed, conceptualized the study, and drafted the manuscript; GM, XT, XF: analyzed and interpreted the data; ZW, MH, AX, HY, HS, JL, CL, RG, JH, CP, YZ, TW: collected the data; ChangrenH, ZG, SL, SS, XZ, and XL: critically revised the study outcomes; AL and C-ZD: study supervision, and critical revision of the manuscript. AL and C-ZD are guarantors for the paper.

    • Funding XF, AL, and C-ZD were founders of this study. Natural Science Foundation of China (82201427, 82171290) and the Foundation of National Heath Commission Capacity Building and Continuing Education Center (GWJJ2022100102).

    • Competing interests None declared.

    • 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.