Background/importance Anesthesiologists frequently use truncal catheters for postoperative pain control but with limited characterization of dosing and toxicity.
Objective We reviewed the published literature to characterize local anesthetic dosing and toxicity of paravertebral and transversus abdominis plane catheters in adults.
Evidence review We searched the literature for bupivacaine or ropivacaine infusions in the paravertebral or transversus abdominis space in humans dosed for 24 hours. We evaluated bolus dosing, infusion dosing and cumulative 24-hour dosing in adults. We also identified cases of local anesthetic systemic toxicity and toxic blood levels.
Findings Following screening, we extracted data from 121 and 108 papers for ropivacaine and bupivacaine respectively with a total of 6802 patients. For ropivacaine and bupivacaine, respectively, bolus dose was 1.4 mg/kg (95% CI 0.4 to 3.0, n=2978) and 1.0 mg/kg (95% CI 0.18 to 2.1, n=2724); infusion dose was 0.26 mg/kg/hour (95% CI 0.06 to 0.63, n=3579) and 0.2 mg/kg/hour (95% CI 0.06 to 0.5, n=3199); 24-hour dose was 7.75 mg/kg (95% CI 2.1 to 15.7, n=3579) and 6.0 mg/kg (95% CI 2.1 to 13.6, n=3223). Twenty-four hour doses exceeded the package insert recommended upper limit in 28% (range: 17%–40% based on maximum and minimum patient weights) of ropivacaine infusions and 51% (range: 45%–71%) of bupivacaine infusions. Toxicity occurred in 30 patients and was associated with high 24-hour dose, bilateral catheters, cardiac surgery, cytochrome P-450 inhibitors and hypoalbuminemia.
Conclusion Practitioners frequently administer ropivacaine and bupivacaine above the package insert limits, at doses associated with toxicity. Patient safety would benefit from more specific recommendations to limit excessive dose and risk of toxicity.
- Drug-Related Side Effects and Adverse Reactions
- Nerve Block
- Pain, Postoperative
- REGIONAL ANESTHESIA
- Postoperative Complications
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
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.
- Drug-Related Side Effects and Adverse Reactions
- Nerve Block
- Pain, Postoperative
- REGIONAL ANESTHESIA
- Postoperative Complications
Eason and Wyatt reported the first truncal catheter in the paravertebral (PVB) space.1 A decade later, Sabanathan et al popularized the technique2 and Rafi identified the transversus abdominis plane (TAP) as an alternative location for truncal catheters, albeit in a fascial plane instead of the neuraxial space.3 Since then, practitioners have used truncal infusions of local anesthetic for postoperative pain control. However, there are limited data regarding delivery technique, dosing, and complications associated with these catheters. Multiple prior authors have asserted that bilateral catheters and ‘relatively large doses of local anesthetics’4 5 are safe and not associated with an increased risk of adverse events. Previous meta-analysis6 and systematic reviews7 on the topic only included randomized controlled trials (RCTs) not designed to assess for local anesthetic systemic toxicity (LAST) or other rare complications. With complication rates below 0.5% for LAST,8 thousands of patients are needed to appropriately identify these rates. Given this limitation, we reviewed the literature to characterize dosing of local anesthetic in the PVB or TAP space; we then evaluated for dosing above the upper limit suggested by the package inserts,9 10 and assessed for associations with LAST.
For full methods, see online supplemental material. A literature search was performed on PubMed and EuropePMC to identify publications between 1987 and 2021 with catheters or infusions in the TAP or PVB space intended for greater than 24 hours of use with either bupivacaine or ropivacaine. All study designs were considered. Exclusion criteria were alternative local anesthetics for the infusion (eg, lidocaine, mepivacaine, chloroprocaine), alternative infusion sites (eg, erector spinae), unavailable dosing, review articles, conference abstracts and animal studies. Non-English language papers were translated using Google Translate. If a paper was unavailable, data were extracted from the abstract. Abstract screen, full text screen and data extraction were confirmed by two authors (MF, LJ, and BB). Disagreements on inclusion or extraction were mediated by the third author. The data were split into pediatric catheters (presented in the accompanying manuscript: Fettiplace et al,11) and adult data in the current manuscript.
Our goal was to assess and quantify dosing of catheters (in mg/kg) including bolus dosing, infusion dosing and 24-hour cumulative dosing. See the ‘Data synthesis’ section in online supplemental methods for full details. We also extracted and evaluated complications, specifically cases of LAST and toxic blood levels of bupivacaine. Cases were counted as LAST if the authors of the original paper noted ‘toxicity’, commented on ‘side effects’ associated with likely ‘high serum levels’, or directly stated LAST.
Extraction and demographics
Following review, we extracted 225 papers. preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow chart and primary data are available supplemental file (online supplemental figure S1: online supplemental tables S1–S12.) This included 108 papers using bupivacaine2 12–118 with 58 RCTs, 16 case reports, 10 case series, 15 observational studies, 6 retrospective cohort studies, and 3 technical reports (online supplemental tables S1, S3–S7); along with 121 papers using ropivacaine33 41 45 47 105 119–234 with 57 RCTs, 21 case reports, 18 observational studies, 12 retrospective cohort studies, 11 case series and 2 technical reports (online supplemental tables S2, S8–S12) with 6802 total patients. Some papers included multiple infusion regimens (eg, multiple delivery rates or separate intermittent bolus and continuous infusion methods) and we extracted a total of 125 bupivacaine infusions and 149 ropivacaine infusions. Summary details about paper demographics and delivery details are included in online supplemental tables S13, S14. RCTs represented the majority of patients (53%) with a worldwide distribution. Europe (29%) and USA (20%) were the top two regions. PVB catheters primarily used continuous (91%) and unilateral infusions (90%) for thoracic surgery (62%) while TAP catheters were primarily used by general surgery (64%) with bilateral infusions (79%) and occasional intermittent bolus methods (41%). Both block locations predominantly used fixed-doses (defined as a standardized volume of local anesthetic instead of weight-based dose or volume; used in 84% of bolus delivery, 77% of infusion delivery). On-demand and breakthrough delivery were available for 12.5% and 22% of patients, respectively.
Initial Bolus dosing
Fixed-volume bolus was the predominant method of delivery (online supplemental table S14). Bolus concentration was similar for the PVB (mode=0.25%, median=0.27%, table 1, online supplemental tables S4, S15) and the TAP (mode and median=0.25%, table 1). Fixed-bolus volumes were slightly higher in the TAP group (mode=40 mL, median=30 mL) than PVB group (mode=28 mL, median=20 mL) in the setting of more frequent bilateral catheters. Taking weight-based dosing into account led to similar overall bolus doses (figure 1A,B). There was more variability in the PVB bolus dosing with higher maximum dosing and lower minimum dosing. PVB catheters were more frequently dosed above 2.0 mg/kg contributing to the higher maximum dose compared with the TAP catheters. The most frequent bolus was 0.75–1.25 mg/kg with combinations of 0.25% at either 0.3–0.4 mL/kg74 78 84 90 or 20–40 mL.107 109 112 Alternatives included 0.125 at 40cc92, 0.5 at 15cc47 or 75 mg with a 0.27% concentration for the infusion.97
Concentration was similar between TAP and PVB (median and mode=0.5%, table 1, online supplemental tables S9, S15), but with higher volumes in the TAP (median=30 mL, mode=40 mL) than PVB (median and mode=20 mL) leading to a lower bolus dose in PVB catheters compared with TAP (table 1, figure 2A,B). The most common bolus regimen for catheters was 0.75–1.25 mg/kg with either 0.2–0.375% concentration at 15–20 mL or 0.5–0.6% at 10–15 mL in unilateral catheters.122 135 160 168 199 201 205 210 222 225 Some of these papers used bilateral catheters with twice as much volume and double the dose (upper limit 2.5 mg/kg). The second most common bolus regimen for catheters was 1.75–2.25 mg/kg and with 0.5% bolus and 30cc unilateral; alternatively 0.75% 20cc unilateral.120 134 141 156 179 200 207 213 227
Infusion dosing (combined continuous and intermittent bolus)
PVB catheters frequently utilized continuous infusions (94.3%, online supplemental table S14) whereas TAP catheters predominantly used intermittent bolus delivery (62%). Both TAP and PVB catheters primarily used fixed-volume doses (74% of infusions). Median hourly dose in PVB catheters (0.23 mL/kg/hour, table 2, online supplemental tables S5, S6, S15) was higher than TAP catheters (0.125 mL/kg/hour) with frequent infusions above 0.3 mg/kg/hour in PVB catheters compared TAP catheters (figure 1C,D). The most common infusion dose was 0.175–0.225 mg/kg/hour, including individual infusions of 0.25% at 5–10 mL/hour13 70 80 83 93 97 103 or 0.125% up to 10 mL/hour or 0.25 mL/kg/hour.70 104 111 117 While no TAP papers used doses above 0.4 mg/kg/hour, 11 PVB papers used doses of 0.5 mg/kg/hour or higher primarily in the 1990s.14 15 18 20 24 25 27 30 31 36 40
Authors predominantly reported continuous infusion methods in 88.8% of PVB infusions and 81.9% of TAP infusions . Median concentrations in continuous infusions were similar (0.2% in both PVB and TAP, table 2, online supplemental tables S10, S11, S15, figure 2C,D) but higher in PVB than TAP due to higher upper limits (eg, 0.6%). Meanwhile, continuous infusion volumes were higher in TAP than PVB due to a higher upper limit of volume (25 vs 22 mL). For intermittent boluses, intervals were similar but concentration and volume were higher in the TAP groups, resulting in lower doses in PVB (median: 0.23 mg/kg/hour, figure 2C) compared with TAP (median: 0.32 mg/kg/hour, figure 2D). The most common infusion dosing for catheters was 0.15–0.25 mg/kg/hour, with variable concentrations of 0.075%–0.45% and volumes of 6–25 mL/hour.122 143 179 187 199 201 205 210 232
Total dosing (bolus plus infusion plus breakthrough)
Consistent with bolus and infusion dosing, median 24-hour dose was higher in PVB catheters (6.1 mg/kg, table 3, online supplemental table 15, figure 1E) than TAP catheters (4 mg/kg, figure 1F). Maximum 24-hour dose was also higher for PVB catheters (8.9 mg/kg) than TAP catheters (4 mg/kg). The most common 24-hour dose fell between 3.5 and 5.5 mg/kg over 24 hours.26 28 29 42 47 61 67 76–78 84 89 90 93 95 97 102 105 110 Of the 108 papers, 66 (61%) reported maximum 24-hour doses over 400 mg with 13 (~20% of the 66) occurring with bilateral catheters. Assuming a 70 kg patient, 51.1% of subjects (n=1646, range: 45.1–71%) exceeded the upper limit of 400 mg of bupivacaine in 24 hours recommended by the manufacturer10 and experts.235
Consistent with bolus and infusion dosing, median 24-hour dose in PVB catheters (7.3 mg/kg, table 3, online supplemental table S15, figure 2E) was lower than then in TAP catheters (8.0 mg/kg, figure 2F). Maximum 24-hour dose in PVB catheters (9.3 mg/kg) was also lower than in TAP catheters (10.3 mg/kg). The most common median 24-hour dosing regimen for catheters fell between 4.5 and 7.5.45 105 145 158 165 188 190 194–196 198 200 209 222 226 231 234 Infusions with 24-hour maximum doses of greater than 17 mg/kg accounted for 217/3579 (6%) of subjects, driven by high dose infusions (eg, >16 cc/hour of 0.2%)187 or high concentration infusions (eg, 0.4%–0.5% infusions).153 160 161 163 167 168 172 184 187 203 Of the 121 papers, 41 (34%) reported maximum doses above 770 mg with 23 (~56% of all high dose) using bilateral catheters. Assuming a 70 kg patient, 28.4% of subjects (n=1018, range: 17–40%) exceeded the upper limit of 770 mg of ropivacaine in 24 hours.
Duration of infusion
The median duration of infusion in bupivacaine catheters (online supplemental table S7) was 60 hours (range: 24–504) with a duration of 48 hours in TAPs (range: 24–96) and 72 hours in PVBs (range: 24–504). The longest duration was 504 hours used in patients with postherpetic neuralgia.44
The median duration of infusion in ropivacaine catheters (online supplemental table S12) was 48 hours (range: 24–1056) with a duration of 48 hours in TAPs (range: 24–240 hours) and 48 hours in PVBs (range: 24–1056 hours). The PVB catheters used for over 1000 hours were from a case series by Esch et al131 highlighting ambulatory catheters in a terminally ill patient.
Last and toxic serum levels
Twelve patients25 30 52 65 109 (0.37% of all patients) experienced bupivacaine LAST (table 4) including 11 cases out of 3145 total patients (0.4%) when excluding case series/reports. This also included five cases of interpleural delivery; excluding those cases the rate was 0.2%. Neurological symptoms included dysgeusia,109 confusion,25 30 convulsions,52 65 and seizures precipitating a fatal aspiration.65 Cardiac symptoms included hypotension,109 arrhythmia, myocardial infarction, heart failure and death25 30 but with other potential causes or contributing factors (eg, pneumonectomy). All LAST patients received doses greater than 10 mg/kg over 24 hours and infusions greater than 0.35 mg/kg/hour (75th dosing percentile). Other associations included rib fractures52 109 low body weight,65 cytochrome P-450 (CYP) 3A4 inhibiting medication, and breakthrough boluses. An additional 28 patients had bupivacaine levels greater than 2.1 mg/L known to correlate with LAST (table 5),236 but without explicit symptoms. Of the 29 patients with toxic blood levels, only a single patient received less than 5.7 mg/kg over 24 hours (eg, 400 mg in a 70 kg patient).29
Eighteen patients (0.48% of all catheters) experienced ropivacaine LAST (table 4). Excluding case reports/series there were 16 cases out of 3520 catheters (0.45%). Eleven cases occurred following cardiac surgery. Excluding both cardiac and case reports, the rate dropped to 0.15% (5 out of 3259). Twelve cases received greater than the 11.5 mg/kg dose (consistent with a 770 mg 24-hour upper limit recommended in a 70 kg by the manufacturer and approximate 75th dosing percentile).9 Below this limit, toxicity occurred in 0.24% of catheters (6 out of 2334) and above this, toxicity occurred in 1.0% of catheters (12/1145). Among the 11 toxicity subjects with measured toxic plasma concentrations, 10 had blood levels above the median toxic venous blood level of 2.2 μg/mL reported by Knudsen et al.236 An additional 57 patients with measured plasma concentrations had levels above 2.2 μg/mL but without toxic symptoms (table 5). A prolonged Qtc was observed in one case161 but not attributed to LAST because the unbound plasma level was 0.12 mg/L, less than the median value of 0.15 mg/L found to be toxic in Knudsen et al; discounting that the range of toxic unbound venous plasma levels was as low a 0.01 mg/L in Knudsen.236 The same was true for Ollier et al162 who reported an unbound venous level of 0.09 and reported it as safe because it was below the median toxic value in Knudsen, but not below the minimal toxic value.
We evaluated local anesthetic dosing and toxicity of ropivacaine and bupivacaine infusions in TAP and PVB catheters in adults in the published literature. We assessed these two blocks because of their historical character and need for large volumes/doses. While the sites have different absorption kinetics, the terminal elimination half-lives are similar for neuraxial237 238 and TAP239 240 sites based on prior pharmacokinetic studies. Consistent with accumulation/elimination (instead of absorption) as a risk factor, practitioners reported 30 cases of toxicity out of the 6802 infusions (0.4% of all infusions) with only 5 cases occurring before the 24-hour mark. Of the 30 cases, 22 were in the top dosing quartile and 26 received doses in excess of the 24-hour limit when adjusted for weight. Comparatively, median 24-hour doses were 420 and 540 mg for bupivacaine and ropivacaine, respectively, with over 51% and 28% of patients receiving doses above the package-insert 24-hour limit of 400 mg and 770 mg for bupivacaine10 and ropivacaine,9 respectively. Intermittent bolus methods did not cause any toxic events but breakthrough bolus methods did,65 167 identifying the need for additional study.
Our study identified heterogeneous dosing within the literature for both ropivacaine and bupivacaine in both TAP and PVB blocks. This is consistent with pediatric data where weight-based dosing is more well studied than in adults.241–243 Bolus doses for bupivacaine and ropivacaine frequently comport with dose limits of 175 mg and 200–250 mg for each respective drug.9 10 At least one case of toxicity arose from a bolus of 200 mg of ropivacaine in a patient with an actual body weight of 45 kg150 illustrating the problem with fixed-volume doses.244
Infusion limits for bupivacaine in adults are lacking but pediatric textbooks advise an upper limit of 0.3–0.375 mg/kg/hour245 and the package insert recommends an upper limits of 400 mg in 24 hours.10 The package insert dosing would limit infusions to 0.23 mg/kg/hour in a 70 kg patient without a bolus.10 This dose mirrors the median hourly dose in PVB catheters but with the addition of a bolus dose, 51% of patients exceeded the 24-hour maximum. Using the pediatric limit of 0.3 mg/kg/hour, 26% received excess doses (eg, 7.2 mg/kg/day, 504 mg in a 70 kg patient) but with no cases of toxicity below this dose. In contrast to bupivacaine, the ropivacaine package insert recommendeds an upper limit of 28 mg/hour (0.4 mg/kg/hour in a 70 kg patient) and 770 mg 24-hour limit.9 Many ropivacaine infusions used doses greater than 0.5 mg/kg/hour leading to 12 mg/kg or greater in 24 hours (or 840 mg in a 70 kg patient).
The impact of these large doses is uncertain but our toxicity data would indicate a risk associated with the higher doses. Kotzé et al attempted to assess the effect of dose on outcome and adverse events in a prior meta-regression.6 The study reported that doses of 1000 mg a day of bupivacaine were more efficacious and not more risky. However, that analysis only included 577 patients in the evaluation of adverse outcomes. Complicating this issue, many trials with high doses do not even monitor (or report monitoring) for potential LAST.125 127 160 203 As such, there is a significant limitation to systematic reviews or meta-analysis of RCT data of rare events.
Local anesthetic systemic toxicity
Our search yielded 30 cases of LAST out of 6802 total infusions with a rate of toxicity of 0.37%–0.48% in bupivacaine and ropivacaine catheters respectively based on naïve pooling. This is consistent with the upper limits of estimated rates of LAST.8 It also comports with rates from a prior systematic review of PVB catheters that observed 3 cases out of 383 published infusions (0.8%).246 The risk of death was 0.03%, representing 10% of all LAST cases, consistent with prior publications.247 248 Of those cases with associated timings, only 1 (3%) occurred following the initial bolus and 5 (17%) occurred before 24 hours, with the rest resulting from ongoing infusions. Of the 30 cases, 26 used doses above the recommend 24 hour limits (when adjusting for a 70 kg patient) with a median dose of 901 mg and 810 mg of bupivacaine and ropivacaine respectively. All cases of bupivacaine toxicity used doses above 5.7 mg/kg (~400 mg in a 70 kg patient) and 66% of the cases of ropivacaine toxicity used above 11 mg/kg (~770 mg in a 70 kg patient).
A number of other risk factors were associated with toxicity including hypoalbuminemia,65 157 177 CYP-inhibiting medications,65 177 breakthrough bolus doses,65 167 bilateral catheters109 122 150 154 173 and cardiac surgery.122 173 Neurological symptoms were the most common symptom, with confusion, somnolence, combativeness and agitation occurring in 22 of the 30 cases (73%) whereas paresthesias or tremors occurred in 6 out of 30 cases (20%). Hypotension or generalized seizures only occurred in 4 cases (13%). Cardiac manifestations were seen for both bupivacaine and ropivacaine with sequelae including two deaths.
Prior reviews of LAST in the literature characterized data from 1979 to 2020,247–250 and identified 243 cases of toxicity with 20 cases of catheter-associated toxicity in adults. Those reviews included 10 cases of bupivacaine toxicity251–258 and 6 cases of ropivacaine toxicity259–263 not already included in this review. Of those cases with documented doses, 66% used large doses (minimum 9.8 mg/kg per day or 0.4 mg/kg/hour for bupivacaine and 12.3 mg/kg per day or 0.5 mg/kg/hour for ropivacaine).58 252 253 255–257 259 Other risk factors included multiple catheters,260 low body weight (50 kg or less),252 257 liver dysfunction/resection,257 prolonged duration (5+ days),252 262 breakthrough doses,257 258 260 262 CYP-inhibition,251 albumin-binding displacers (eg, amitriptyline),244 264 hypoalbuminemia,262 an intrathecal catheter,258 and an intravenous catheter.263 The 46 cases (30 in this review and 16 additional from prior reviews) included 24 cases of ropivacaine toxicity and 22 cases of bupivacaine toxicity, with convergent risk factors (box 1). The addition of 26 cases in this review illustrates the limitation of targeted literature reviews247–250 which do not capture toxicity unmentioned in the title or abstract. The risk factors also comport with risk factors in the pediatric literature (accompanying manuscript: Fettiplace et al11).
Risk factors for local anesthetic systemic toxicity identified in catheters
High weight-based dose
Prolonged duration (>72 hours).
Albumin displacing drugs (amitriptyline).
Cardiac surgery/ischemic heart disease.
Cytochrome P450 inhibiting medications.
Conclusion and remaining questions
In conclusion, we investigated the use of bupivacaine and ropivacaine in PVB and TAP catheters across the literature and found frequent use of doses in excess of the package insert upper limit recommendations. We also identified cumulative dose in adult patients as a consistent risk factor for toxicity, contrary to prior meta-analysis6 and expert opinion.4 5 However, high doses were generally well tolerated and many cases of toxicity occurred in the setting of concurrent risk factors (box 1). Given the elevated risk of toxicity in the upper quartile of dosing, weight-based dosing (using ideal or lean body weight) restricted to the 75th percentile with an absolute upper limit in the 90th–95th percentile would likely limit toxic events. This reflects practice in the pediatric literature.265–268 Patients with comorbidities would benefit by limiting rates to the median dose commensurate with neonatal dosing recommendations245 265:
Bupivacaine bolus 0.75–1.5 mg/kg (0.25% at 0.3–0.6 mL/kg), max 2.0 mg/kg.
Bupivacaine infusion of 0.1–0.35 mg/kg/hour (0.1% at 0.1–0.35 mL/kg/hour).
Absolute max of 0.4–0.5 mg/kg/hour.
Max of 0.2–0.25 mg/kg/hour in the setting of liver dysfunction or CYP inhibitors, cardiac surgery or heart failure, cachexia, protein deficiency/malnutrition or albumin displacing drugs.
Ropivacaine bolus 1.0–2.0 mg/kg (0.5% at 0.25–0.5 mL/kg); max 3.0 mg/kg.
Ropivacaine infusion 0.15–0.45 mg/kg/hour (0.2% at 0.75–0.225 mL/kg/hour).
Absolute max of 0.5–0.6 mg/kg/hour.
Max of 0.25–0.3 mg/kg/hour in the setting of liver dysfunction or CYP inhibitors, cardiac surgery or heart failure, cachexia, protein deficiency/malnutrition or albumin displacing drugs.
Duration: 48–72 hours ; max 120 hours.
Practitioners should maintain vigilance toward dose, along with other risk factors, and consider methods to reduce total dosing (eg, lower infusion rate, lower infusion concentration, unilateral catheters and intermittent bolus methods). Patients with risk factors or on prolonged infusions (greater than 72 hours) may also benefit from monitoring for high drug concentrations.
There are a number of remaining questions that this review identifies. First, the field would benefit from a follow-up meta-analysis that incorporates case reports, observational trials along with RCTs to accurately describe the rate of toxicity in truncal catheters in the published literature. Second, there are questions about the safety of local anesthetic catheters in cardiac surgery, the need for dose adjustment in susceptible patients (eg, with CYP-inhibiting medication or hypoalbuminemia), the use of breakthrough doses, the safety of intermittent bolus methods and recommendations for monitoring adverse events in clinical trials of local anesthetics. Finally, rational guidelines based on these data would help clinicians and improve patient safety.
Patient consent for publication
BB and LJ are joint first authors.
Twitter @BrittaniBungart, @mfettiplace
BB and LJ contributed equally.
Contributors MF conceived the project and designed the extraction. BB, LJ and MF all contributed to literature screening, data extraction, data preparation, manuscript drafting and manuscript editing.
Funding MF is supported by an NIH T32 training grant 5T32GM007592-42.
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.