Article Text

A randomized, double-blind, placebo-controlled study to evaluate the efficacy and safety of a bupivacaine hydrochloride implant in patients undergoing abdominoplasty
  1. Alina C Beaton1,
  2. Daneshvari Solanki2,
  3. Hernan Salazar3,
  4. Steve Folkerth4,
  5. Neil Singla1,
  6. Harold S Minkowitz2,
  7. David Leiman2,
  8. Ben Vaughn5,
  9. Nina Skuban6 and
  10. Gwendolyn Niebler6
  1. 1 Lotus Clinical Research, LLC, Pasadena, California, USA
  2. 2 HD Research/First Surgical Hospital, Bellaire, Texas, USA
  3. 3 Endeavor Clinical Trials, San Antonio, Texas, USA
  4. 4 Midwest Clinical Research Center, Ohio, Dayton, USA
  5. 5 Rho, North Carolina, Durham, USA
  6. 6 Innocoll Pharmaceuticals Limited, Innocoll Biotherapeutics, Princeton, New Jersey, USA
  1. Correspondence to Dr Nina Skuban, Innocoll Pharmaceuticals Limited, Innocoll Biotherapeutics, Princeton, NJ 19073, USA; nskuban{at}innocoll.com

Abstract

Introduction Surgical site infiltration with bupivacaine hydrochloride (HCl) is a standard element of postoperative analgesia for soft tissue surgeries, but results in short-lived analgesia. A novel bupivacaine implant, XARACOLL (bupivacaine HCl), is Food and Drug Administration approved for treatment of acute postsurgical pain following adult inguinal herniorrhaphy. This study examined the efficacy and safety of the bupivacaine implant (300 mg) compared with placebo for postsurgical pain after abdominoplasty.

Methods In this double-blind, placebo-controlled study, patients undergoing abdominoplasty were randomized to three 100 mg bupivacaine implants or three placebo collagen implants, in a 1:1 ratio, implanted intraoperatively. No other analgesics were administered into the surgical site. Patients were allowed opioids and acetaminophen for postoperative pain. Patients were followed for up to 30 days after treatment. Primary outcome: the analgesic effect of the bupivacaine implants through 24 hours postsurgery, measured by the sum of time-weighted pain intensity (SPI24). Prespecified key secondary outcomes included SPI48 and SPI72, percentage of opioid-free patients through 24, 48, and 72 hours, and adverse events, which were tested sequentially to control for multiplicity (ie, if the first variable failed to reach significance, no subsequent variables were declared statistically significant).

Results The bupivacaine implant patients (n=181) reported statistically significant lower SPI24 (mean (SD) SPI24=102 (43), 95% CI 95 to 109) compared with placebo patients (n=184; SPI24=117 (45), 95% CI 111 to 123, p=0.002). SPI48 was 190 (88, 95% CI 177 to 204) for INL-001 and 206 (96, 95% CI 192 to 219) for placebo, and not significantly different between groups. The subsequent secondary variables were therefore declared not statistically significant. SPI72 was 265 (131, 95% CI 244 to 285) for INL-001 and 281 (146, 95% CI 261 to 301) for placebo. The opioid-free percentage of patients at 24, 48, and 72 hours was 19%, 17%, and 17% for INL-001 and 6.5% for placebo patients (at all timepoints). The only adverse event occurring in ≥5% of patients and for which proportion INL-001 >placebo was back pain (7.7% vs 7.6%).

Conclusion The study design was limited by not containing an active comparator. Compared with placebo, INL-001 provides postoperative analgesia that is temporally aligned with the period of maximal postsurgical pain in abdominoplasty and offers a favorable safety profile.

Trial registration number NCT04785625.

  • pain, postoperative
  • analgesics, opioid
  • drug-related side effects and adverse reactions
  • pain management
  • postoperative complications

Data availability statement

No data are available.

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, an indication of whether changes were made, 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

  • The bupivacaine hydrochloride implant (INL-001) is Food and Drug Administration approved for use in adults for placement into the surgical site to produce postsurgical local analgesia for up to 24 hours following open inguinal hernia repair.

WHAT THIS STUDY ADDS

  • In abdominoplasty, bupivacaine implants provide postoperative analgesia through 24 hours.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • Future research needs to determine whether the bupivacaine implants provide additional analgesia compared with standard bupivacaine infiltration.

Introduction

Approximately 80% of patients who undergo surgery will experience acute postoperative pain which, if inadequately managed, can be associated with delayed recovery time, and an increase in healthcare costs.1

While opioids are used in postoperative pain management, recent efforts to improve acute postsurgical pain management have focused on limiting opioid use to reduce the risk for opioid-related adverse events (AEs).2 It has also been reported that postoperative opioid treatment can lead to persistent opioid use among opioid-naïve patients, with rates of 5.9%–6.5%.3 A multimodal approach to analgesia, involving the administration of drugs with a variety of mechanisms of action, is recommended to decrease the need for opioids, and the infiltration of local anesthetics, such as bupivacaine, at the surgical site has become a common component of multimodal analgesic strategies.4 However, local anesthetic infiltration into the surgical site induces short-lived analgesia (4–8 hours).5 Infiltration also carries other risks such as preparation dosing errors and accidental intravascular injection that can result in local anesthetic systemic toxicity, associated with significant morbidity and mortality.6–9

XARACOLL (Innocoll Pharmaceuticals, Ireland), also known as INL-001, is a US Food and Drug Administration (FDA)-approved bioresorbable proprietary collagen-matrix drug-delivery implant containing bupivacaine hydrochloride (HCl) for treatment of postsurgical pain for up to 24 hours following open inguinal hernia repair. The collagen-matrix is composed of type I purified bovine collagen and acts as an extended drug delivery system, releasing bupivacaine immediately and over time in the surgical wound. Each implant contains 100 mg of bupivacaine HCl and the approved dose consists of three implants placed into the surgical site (total dose 300 mg). Two phase III studies showed improvement in pain intensity through 24 hours and less opioid use through 72 hours postsurgery, compared with placebo.10 11 Whereas these studies established INL-001’s ability to safely reduce pain intensity and opioid use following open inguinal hernia repair surgery, there was limited data evaluating its analgesic properties in other soft tissue surgeries.

Abdominoplasty is a commonly performed surgical procedure with significant postoperative pain.12 13 It is a recognized model for characterizing a drug’s effect on acute postoperative soft tissue pain and has been validated as an FDA-accepted registration model.14 This phase III study evaluated the postoperative analgesic efficacy and safety of INL-001 in patients undergoing abdominoplasty.

Methods

This was a multicenter, randomized, double-blind, placebo-controlled efficacy and safety study of INL-001 (300 mg) for postsurgical pain in patients undergoing abdominoplasty. The study was registered on ClinicalTrials.gov (March 8, 2021; NCT04785625) and conducted from April 2021 (first patient enrolled: April 29, 2021) to October 2021 (last patient complete: October 27, 2021) at four centers in the USA (https://clinicaltrials.gov/ct2/show/NCT04785625). The primary objective of the study was to evaluate the analgesic effect of treatment with INL-001 compared with placebo collagen matrix implants. The secondary objective was to assess the safety and tolerability of INL-001.

Study population

Men and women (aged 18–65 years), with a body mass index (BMI) of 18–35 kg/m2 undergoing elective abdominoplasty surgery with rectus sheath plication, were eligible for participation. Patients were excluded from the study if they had known hypersensitivity to amide-type local anesthetics, fentanyl, morphine, oxycodone, acetaminophen, non-steroidal anti-inflammatory drugs, or bovine products, were scheduled for other significant concurrent surgical procedures, had undergone major surgery within 3 months before the scheduled abdominoplasty, or planned to undergo another surgical procedure within 30 days after the study surgery.

During the screening period (up to 45 days before surgery), eligibility was confirmed, and informed consent provided. Patients were monitored during an inpatient period of approximately 4 days and an outpatient follow-up period of up to 30 days (phone-call on day 7 and a clinic visit on days 15 and 30).

Randomization and blinding

Randomization was stratified by study center and BMI (<30 kg/m2 and ≥30 kg/m2). On the day of surgery, eligible patients were randomly assigned (1:1) to treatment with INL-001 (three 100 mg implants containing a total dose of 300 mg of bupivacaine HCl) or placebo collagen implants, which were identical in appearance and provided in blinded, numbered kits to the investigative site. A centralized blocked randomization was performed using an electronic randomization system. Site staff, including surgeons, the study sponsor personnel including the statistician, and patients were blinded to treatment. No unblinding to treatment occurred during the study. Unblinding was conducted by an independent data management group after all patients completed the study and the database was locked.

Intervention and procedures

Abdominoplasty was performed according to standard surgical techniques under general anesthesia, with no other local anesthetic used at the surgical site (online supplemental material 1 contains the anesthesia protocol). The abdominoplasty procedure was performed with an anterior approach, and with an incision not extending above the umbilicus. In general, the incision was from one anterior superior iliac spine to the other, but the exact incision length could vary depending on the patient’s anatomy. Drains could be placed at the discretion of the investigator, but ancillary procedures (eg, liposuction) at the same time were prohibited.

Supplemental material

Prior to incision closure, the INL-001 or placebo implants were placed in the surgical site. Following tissue removal and repair of the abdominal musculature, two implants were placed on the rectus diastasis at the site of rectus sheath plication and one implant was placed below the abdominal incision between Scarpa’s fascia and the subcutaneous fat. The implants were, to the greatest extent possible, placed so they spanned the fascia that was exposed prior to surgical closure. Implants could be cut in two halves to accommodate placement. Patients in both treatment groups were permitted rescue medication (opioids—morphine or oxycodone, and/or acetaminophen) to manage breakthrough pain (see online supplemental material 2).

Outcome measures

Efficacy

Pain intensity was measured using a patient-completed 11-point Numerical Pain Rating Scale (NPRS; 0 (no pain) to 10 (worst possible pain)). NPRS scores were recorded at prespecified multiple timepoints through 72 hours post-treatment. These timepoints were: 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, 18, 20, 24, 28, 32, 36, 48, and 72 hours after treatment (placement of implants). Pain intensity was also recorded prior to rescue pain medication use.

The sum of pain intensity (SPI) was calculated from time 0 (placement of implants) through 24 hours post-treatment (SPI24), through 48 hours (SPI48), and through 72 hours (SPI72). SPI was calculated using area under the curve (AUC) of pain intensity (NPRS) and the actual assessment times in hours (ie, the pain-time curve). The primary efficacy outcome measure for the study was SPI24. SPI48 and SPI72 were considered key secondary efficacy outcome measures. Pain scores obtained prior to rescue medication use were also included in the calculation of SPI. Pain scores subsequent to the administration of pain medication rescue use were censored through the duration of efficacy of the rescue (2 hours for morphine, 4 hours for acetaminophen, oxycodone), with the exception that pain scores in this period that were higher than the rescue pain score were retained. This approach assumes that the pain profile without rescue would have transitioned linearly from the pre-rescue value to the first value after the duration of effect for the rescue; the censoring removes the dip in pain scores attributable to rescue.

Opioid use was collected throughout the study. The proportion of patients who were opioid free for 0–24, 0–48, and 0–72 hours after surgery were prespecified key secondary efficacy outcomes.

Safety

Patients were monitored for AEs through day 30 after surgery. AEs were a prespecified secondary outcome measure of this study. Following hospital discharge, patients reported AEs by telephone and provided incidence, duration, and any associated treatment. Additionally, AE information was recorded during the follow-up phone call (day 7) and visits (days 15 and 30). AEs were assessed for severity and relatedness to treatment.

In addition, a signs and symptoms checklist (online supplemental material 3) was used at prespecified timepoints (hours 0.5, 1, 2, 3, 4, 5, 7, 9, 12, 15, 18, 24, 48, and 72, and days 7 and 15) to systematically assess for any AEs suggestive of systemic bupivacaine toxicity. Blood samples could be collected to assess for bupivacaine plasma concentrations at the discretion of the investigator (online supplemental material 4).

Wound healing was assessed by use of a signs and symptoms checklist (online supplemental material 3) at 24, 48, and 72 hours, and days 7, 15 and 30, to systematically assess for any AEs indicative of adverse wound healing. An additional assessment of wound healing was done through the completion of the Southampton Wound Grading System15 72 hours postsurgery and at days 15 and 30 (online supplemental material 5).

Additional safety variables included clinical laboratory test results, vital signs, electrocardiography findings, and concomitant medications.

Sample size

Sample size calculation was based on previous clinical studies for INL-001 in inguinal hernia repair.10 The effect size of these inguinal hernia repair studies were 0.53 for SPI24 (based on a least squares means (LSmeans) difference of 24.5 with SD 46.5) and 0.25 for SPI48 (based on an LSmeans difference of 23.4 with SD 92.4). Using a two-sample t-test with a 95% CI, a sample size of 360 was expected to yield a power of >99% for SPI24, the prespecified primary outcome measure, and this drove the sample size choice. The power to show separation at 48 hours was recognized as low (66%) but it was thought that the ultimate power of the study to detect a difference at that timepoint might be higher given that abdominoplasty is generally a more painful procedure than inguinal hernia.

Analysis populations

The intent-to-treat (ITT) population consisted of patients randomly assigned to study treatment who may have received any dose of study drug (INL-001 or placebo). The ITT population was used for disposition patient-count purposes. The modified ITT (mITT) population consisted of all patients randomly assigned to study treatment who received any dose of study drug (INL-001 or placebo) and was used for efficacy analysis. The safety population consisted of all patients who received any dose of study drug (INL-001 or placebo) and was used for safety analyses.

Statistical analysis

Patient demographics, baseline characteristics, efficacy, and safety variables are summarized using descriptive statistics.

Efficacy variables were compared between groups (INL-001 vs placebo). For continuous variables, an analysis of covariance was used with treatment as the main effect (two levels) and a covariate for BMI (continuous). For outcomes of proportions, the difference was tested with the Pearson’s χ2. In case of low counts (any expected cell count ≤5), a Fisher’s exact test was used. For the safety variables, no formal statistical tests are performed.

Multiplicity

Prespecified key secondary efficacy variables were tested sequentially at the 0.05 level to control the overall type I error rate. Each key secondary efficacy variable was tested in order:

  1. SPI48.

  2. Proportion of patients who were opioid free post-treatment through 24 hours.

  3. Proportion of patients who were opioid free post-treatment through 48 hours.

  4. SPI72.

  5. Proportion of patients that were opioid free post-treatment through 72 hours.

If a variable failed to reach statistical significance, no subsequent variables would be declared statistically significant.

Results

Of the 802 patients screened, 366 patients were randomly assigned to the study groups (mITT population; INL-001=181, placebo=184). All treated patients completed the full inpatient period. In total, 174 INL-001 patients (96.1%) and 163 placebo patients (88.1%) completed the entire study (figure 1). Demographics and baseline characteristics were similar between groups (table 1).

Table 1

Demographic and other baseline characteristics

Figure 1

Flow diagram of study population. ITT, intent-to-treat; mITT, modified ITT.

Efficacy results

Primary outcome: SPI24

INL-001 patients reported statistically significantly lower mean SPI24 compared with placebo patients (mean (SD; 95% CI) SPI24=102 (42; 96 to 109) vs 117 (45; 111 to 123); p=0.002; table 2).

Table 2

Primary and key secondary efficacy variables (mITT population): pain intensity and opioid use

Key secondary outcomes: SPI48, SPI72 and percentage opioid free

SPI48 (mean (SD; 95% CI)) was 190 (88; 177 to 204) for INL-001 and 206 (96; 192 to 219) for placebo and was not significantly different between groups (p>0.05, see table 2). Therefore, the subsequent variables were declared not statistically significantly different between groups. SPI72 was 265 (131; 244 to 285) for INL-001 and 281 (146; 261 to 301) for placebo.

At the periods 0–24, 0–48, and 0–72 hours, 18.8% (n=34), 17.1% (n=31), and 17.1% (n=31) of INL-001 patients remained opioid free. For all periods, 6.5% (n=12) of placebo patients remained opioid free (table 2). A statistically significant difference between groups for prespecified secondary outcome measures was not achieved.

Pain scores over time are provided in figure 2 and descriptive values for additional SPI timepoints are provided in online supplemental material 6.

Figure 2

Mean (±95% CI) of Numerical Pain Rating Scores (NPRS) by treatment and timepoint.

Safety results

The safety population consisted of 181 INL-001 and 184 placebo patients. The incidence of AEs was similar between treatment groups (INL-001: 84%; placebo: 87%). No patients in either treatment group had an AE leading to discontinuation, death, or removal of the implants. Table 3 lists the AEs that occurred in ≥2% of patients and had higher occurrence in the INL-001 vs placebo group. Most patients reported AEs as mild or moderate (1% of INL-001 and 2% of placebo patients had an AE considered severe) and most AEs were considered unrelated to treatment (6% of INL-001 and 8% of placebo patients had an AE considered definitely or probably related to treatment). AEs considered definitely related to treatment were mild and resolved during the study.

Table 3

Adverse events occurring in ≥2% of patients and for which INL-001 >placebo

A review of AEs derived from the systematic assessment of any signs or symptoms of bupivacaine toxicity during the study did not indicate that any patient experienced systemic bupivacaine toxicity. Eight patients (four in each group) had bupivacaine plasma concentrations assessed (all ≤135 ng/mL) (see online supplemental material 4).

The most common (≥ 2% overall) AEs derived from the systematic assessment of wound healing during the study were: wound dehiscence (INL-001: n=21 (11.6%); placebo: n=24 (13.0%)), incision site swelling (INL-001: n=5 (2.8%); placebo: n=5 (2.7%)), incision site erythema (INL-001: n=5 (2.8%); placebo: n=4 (2.2%)), postprocedural discharge (INL-001: n=5 (2.8%); placebo: n=2 (1.1%)), incision site edema (INL-001: n=8 (4.4%); placebo: n=9 (4.9%)).

Most wound-healing AEs were mild or moderate. There was one serious AE reported in the study. The patient (placebo) experienced a wound infection at a drain site that required hospitalization. The patient recovered. The majority of reported wound dehiscence events occurred at one center and was likely influenced by surgical postoperative management and medical experience of the assessor. Results from the Southampton Wound Grading System are provided in online supplemental material 5.

There were no clinically meaningful effects on laboratory test results, vital sign parameters, physical examinations, electrocardiography parameters, and concomitant medication (other than described for rescue pain medication).

Discussion

Treatment with INL-001 was associated with significantly less pain during the first 24 hours following abdominoplasty surgery compared with placebo. This is similar to the results of open inguinal hernia studies conducted with INL-00110 and is clinically relevant since the first 24 hours after surgery are generally the most painful for soft tissue surgeries like abdominoplasty.16–19

The primary analgesic outcome of this study was pain intensity, specifically SPI24 resulting from repeat pain intensity measurements. While there is no established clinical threshold for SPI endpoints, repeat measurement of pain intensity with analysis over a period of time was chosen as the primary efficacy outcome based on the FDA recommendation to use a well-defined and reliable patient-reported outcome measure of the subject’s pain intensity which allows for frequent pain intensity measurements at preselected timepoints during the trial to measure the effect of a non-opioid analgesic effect over time.20 In addition, SPI is a common measure in pain trials.14 20–23

This study was underpowered for the first key secondary endpoint (SPI48) in the statistical hierarchy and not powered for the remaining key secondary endpoints of this study. There was in increase in pain intensity variability between patients as time from surgery increased, which is not surprising given the subjective nature of pain measurement and differences in recovery between patients postoperatively. This variability impacted the ability to demonstrate statistical significance. Given the numeric differences in favor of INL-001 on the key secondary endpoints, further study of these outcomes with INL-001 could be warranted.

In this study, INL-001 was well tolerated. The percentage of INL-001 patients (84%) reporting any AEs was similar to the placebo group (87%). The types of AEs in this study are expected after surgery, general anesthesia, or abdominoplasty specifically, or can be seen with opioid use.14 18 19 24 25 In addition, the incidence and types of AEs were comparable or lower than reported in placebo or bupivacaine-treated groups in other abdominoplasty studies.14 18 19 24 25

There was no evidence of systemic bupivacaine toxicity when INL-001 was used in abdominoplasty. As with other surgeries, there is a recognized risk of postoperative adverse wound healing after abdominoplasty. The incidence of wound-related AEs in this study was low and consistent with the abdominoplasty literature.14 18 19 25–27 Wound dehiscence was the most reported AE associated with the surgical wound, but the incidence was consistent with that reported in another abdominoplasty study that administered saline or bupivacaine in the surgical site24 and occurred mostly at one investigator site.

A limitation is that this study did not contain an active comparator treatment arm. Patients used rescue medication as needed, rather than on a scheduled basis as might be done in clinical practice to allow for a clearer assessment of efficacy. Additionally, most patients were female, however, this is not unexpected for abdominoplasty surgery.14

Conclusion

This study showed that, compared with placebo, INL-001 was effective in providing postoperative analgesia in abdominoplasty, a model of soft tissue surgery. Treatment with INL-001 was well tolerated in this patient population, with no evidence of systemic bupivacaine toxicity or impaired wound healing. These results may support expanding the indication of INL-001 for use in soft tissue surgeries beyond inguinal hernia repair.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

Ethics approval

The protocol and statement of informed consent were approved by the Institutional Review Boards (WCG IRB-IRB00000533) prior to each center’s initiation. All procedures were in accordance with the ethical standards of the national research committee and with the 1964 Helsinki Declaration and its later amendment or comparable ethical standards. The study was conducted in compliance with the International Council for Harmonization principles of Good Clinical Practice and other applicable local regulations and guidelines. Written informed consent was obtained from all patients included in the study.

Acknowledgments

We thank the patients who participated in this study, the clinical study teams, and the investigational centers that participated.

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

  • Correction notice This article has been corrected since it published Online First. A supplementary file has been removed.

  • Contributors All authors participated in the design and analytical approach of the study and contributed to the manuscript development. Results were summarized and interpreted in collaboration with all authors. All authors had full access to all data in this study and take complete responsibility for the integrity of the data and accuracy of the data analysis. NS is the author responsible for the overall content as the guarantor.

  • Funding Funding to support this study and the journal’s article processing charges and Open Access fee was provided by Innocoll Pharmaceuticals, Monksland, Athlone, Ireland. Esther Smits, PhD of Avania Clinical (Sydney, Australia) provided medical writing and editorial assistance, which was funded by Innocoll Pharmaceuticals.

  • Competing interests ACB declares that she has received consulting fees from Vertex Pharmaceuticals. NSi declares that he is the chief scientific officer and part owner of Lotus Clinical Research, LLC, which received funding from Innocoll in connection with clinical trial services performed for the study discussed in the manuscript. NSi also declares that he serves on the board and that he is part owner (holding stock or stock options) of the parent company of Lotus Clinical Research LLC (ERG Holding Co). HSM declares that he has received support from Innocoll Pharmaceuticals, Durect, and Pain Reform; grants or contracts from Innocoll Pharmaceuticals, Pacira Pharmaceuticals, Heron Therapeutics, Acacia Pharma, and Vertex Pharmaceuticals; consulting fees from Avenue Therapeutics; and payment or honoraria from Pacira Pharmaceuticals, AcelRx Pharmaceuticals, and Heron Therapeutics. DL declares that he received support for the present manuscript from Pacira Pharmaceuticals, Heron Therapeutics, Merck, Innocoll Pharmaceuticals, AcelRx Pharmaceuticals, Acacia Pharma (including consulting fees), Durect, The Medicines Company, and Pain Reform. BV declares that he is an employee of a contract research organization that is paid to provide statistical services to Innocoll Pharmaceuticals. NSk and GN declare that, at the time of the study, they were an employee of Innocoll Pharmaceuticals and that they have patents planned, issued, or pending with Innocoll Pharmaceuticals, and stock or stock options from Innocoll Pharmaceuticals.

  • 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.