Article Text
PDF
Abstract
Background Non-Hispanic Black (NHB) patients experience increased prevalence of stroke risk factors and stroke incidence compared with non-Hispanic White (NHW) patients. However, little is known about >90-day post-stroke functional outcomes following mechanical thrombectomy.
Objective To describe patient characteristics, evaluate stroke risk factors, and analyze the adjusted impact of race on long-term functional outcomes to better identify and limit sources of disparity in post-stroke care.
Methods We retrospectively reviewed 326 patients with ischemic stroke who underwent thrombectomy at two centers between 2019 and 2022. Race was self-reported as NHB, NHW, or non-Hispanic Other. Stroke risk factors, insurance status, procedural parameters, and post-stroke functional outcomes were collected. Good outcomes were defined as modified Rankin Scale score ≤2 and/or discharge disposition to home/self-care. To assess the impact of race on outcomes at 3-, 6-, and 12-months’ follow-up, we performed univariate and multivariate logistic regression.
Results Patients self-identified as NHB (42%), NHW (53%), or Other (5%). 177 (54.3%) patients were female; the median (IQR) age was 67.5 (59–77) years. The median (IQR) National Institutes of Health Stroke Scale score was 15 (10–20). On univariate analysis, NHB patients were more likely to have poor short- and long-term functional outcomes, which persisted on multivariate analysis as significant at 3 and 6 months but not at 12 months (3 months: OR=2.115, P=0.04; 6 months: OR=2.423, P=0.048; 12 months: OR=2.187, P=0.15). NHB patients were also more likely to be discharged to rehabilitation or hospice/death than NHW patients after adjusting for confounders (OR=1.940, P=0.04).
Conclusions NHB patients undergoing thrombectomy for ischemic stroke experience worse 3- and 6-month functional outcomes than NHW patients after adjusting for confounders. Interestingly, this disparity was not detected at 12 months. Future research should focus on identifying social determinants in the short-term post-stroke recovery period to improve parity in stroke care.
- thrombectomy
- intervention
- stroke
Data availability statement
Data are available upon reasonable request.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Non-Hispanic Black (NHB) patients are at a higher risk of stroke and experience more strokes than non-Hispanic White (NHW) patients. However, little is known about long-term (>90 days) post-stroke functional outcomes following mechanical thrombectomy.
WHAT THIS STUDY ADDS
This multicenter study finds that NHB patients with ischemic stroke have worse functional outcomes at 3 and 6 months post-thrombectomy than NHW patients, after adjusting for pre-stroke risk factors, health insurance coverage, occlusion site, and procedural parameters. Statistical differences in 12-month functional outcomes were not detected, which might, in part, be due to loss to follow-up.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study reveals previously undocumented racial disparities in functional outcomes of patients undergoing thrombectomy during a specific follow-up window (3 and 6 months postoperatively). These findings can guide focused healthcare efforts to identify these patients, improve parity in in-hospital care, and address social determinants acting in the short-term post-stroke recovery period.
Introduction
Stroke is the third leading cause of combined death and disability worldwide1 and the fifth leading cause of death in the USA.2 It is well documented that racial minorities, particularly non-Hispanic Black (NHB) people in the USA face a disproportionate burden of stroke.3–6 Specifically, NHB people are 50% more likely to experience stroke and 70% more likely to die from a stroke than non-Hispanic White (NHW) people.7 Several factors are thought to contribute to these disparities, including increased prevalence of stroke risk factors (eg, hypertension and diabetes) and unequal access to healthcare, which may arise from social and structural barriers that are distributed racially and/or socioeconomically.3
Mechanical thrombectomy is increasingly used as an effective approach to treat ischemic stroke and achieve good functional outcomes.8–12 Currently, very few studies have examined the efficacy of mechanical thrombectomy in attaining parity in functional outcomes among racially diverse populations while controlling for disparities in pre-stroke risk factors and access to care. Those studies that do exist report conflicting findings, and most stop short of examining outcomes at >90 days postprocedure.13–16
As endovascular therapies are continually used, it is important to understand whether approaches like mechanical thrombectomy can reduce long-term lifestyle burdens on patients, families, and healthcare systems equitably. Here we query functional outcomes in a racially diverse cohort of patients undergoing mechanical thrombectomy. Our two centers in Baltimore, Maryland, a highly diverse setting in which stroke is a leading cause of death, provide unique opportunities to elucidate findings on racial disparities in stroke outcomes. We describe patient characteristics, evaluate stroke risk factors, and analyze the adjusted impact of race on long-term functional outcomes to better identify and limit sources of disparity in post-stroke care.
Methods
Study design and patients
We retrospectively reviewed an established population of 326 patients who underwent mechanical thrombectomy for ischemic stroke at two centers in our institution between January 1, 2019 and July 5, 2022, inclusive. Both centers are tertiary care facilities that offer mechanical thrombectomy and IV thrombolysis 24 hours per day. Chart review consisted of extracting data from progress notes, medication lists, and discharge summaries via electronic patient records under institutional review board approval. Patients who underwent cerebral arteriographic imaging but did not receive mechanical thrombectomy were excluded from our study. Patients with Hispanic ethnicity were also excluded owing to the limited sample size of Hispanic-identifying patients at our centers during the study period.
Demographic and clinical characteristics that might influence post-stroke outcomes and were known and/or obtainable at the time of neurosurgical consultation were collected. Racial information was based on self-identification, and patients were grouped into one of the following categories: NHB, NHW, or non-Hispanic Other (Asian, Pacific Islander, Native American, or another race). Racial identification was documented last (after outcome measures and other clinical characteristics) to minimize bias. Sex and age at time of mechanical thrombectomy were recorded, as was receipt of IV tissue plasminogen activator (tPA) and National Institutes of Health Stroke Scale (NIHSS) score on admission. Pertinent ischemic stroke risk factors for which data were available and that had an ability to prognosticate stroke incidence and outcome17 18 were documented; these included hypertension, diabetes mellitus, smoking, prior stroke, atrial fibrillation, cardiac stent, cardiovascular disease (inclusion criteria: coronary artery disease, prior myocardial infarction, congestive heart failure), chronic kidney disease, respiratory disease (inclusion criteria: asthma, chronic obstructive pulmonary disease), and intracranial atherosclerotic disease (ICAD).
Health insurance coverage (uninsured, Medicaid, Medicare, private insurance, or Veterans Affairs) was recorded. Occlusion site was recorded as follows: anterior cerebral artery, middle cerebral artery (segments M1, M2, and/or M3), internal carotid artery terminus or intracranial internal carotid artery, and posterior circulation strokes (clots of the basilar, posterior cerebral artery, cerebellar arteries, and/or vertebral arteries).19 No tandem occlusions were recorded. Procedural factors including temporal parameters (time in minutes from symptom discovery to hospital presentation, presentation to groin puncture, and groin puncture to recanalization), monitored anesthesia care, and use of a stent retriever were recorded, as were modified Thrombolysis in Cerebral Infarction (TICI) score and length of hospital stay (LOHS, days from admission to discharge). A TICI score ≥2b was considered successful reperfusion.16 20
Three hundred and twenty-six patients were included on initial assessment. Owing to loss to follow-up, 262 (80.4% of initial cohort), 221 (67.8% of initial), and 197 (60.4% of initial) patients were included in analysis of 3-, 6-, and 12- month outcomes, respectively, with all three racial groups included. After excluding patients identifying as racial group ‘Other’ due to the group’s small sample size, the initial cohort of NHB and NHW patients consisted of 310 cases, and attrition was as follows: 253 (81.6% of initial cohort), 214 (69.0% of initial), and 190 (61.3% of initial) patients at 3-, 6-, and 12- month analyses, respectively.
Outcome measures
Outcome measures were collected from electronic records at 3, 6, and 12 months following thrombectomy. Primary outcomes included modified Rankin Scale (mRS) score, mortality, and discharge disposition. A low mRS score indicating good functional outcomes (ie, functional independence) was defined as ≤2, in accordance with prior studies evaluating mRS as a measure of outcomes.13–16 Discharge disposition equal to 0, indicating routine discharge to home/self-care, was defined as a good functional outcome,14 21 while discharge dispositions of 1 and 2 were defined as poor outcomes (1=discharge to rehabilitation, 2=discharge to hospice/death). To accurately characterize patients who developed a poor post-stroke outcome in our cohort, we aimed to identify acutely available factors associated with mRS score and discharge disposition values indicating poor outcomes. We screened these factors via univariate and multivariate logistic regression.
Statistical analyses
Continuous variables were presented as mean±SD, and ordinal variables were presented as median and IQR. Samples with missing data (due to loss to follow-up, unobtainable information, or other reasons) were excluded from analysis. Bivariate and trivariate statistical differences were calculated using Student’s t-test or one-way analysis of variance where appropriate for continuous variables, and Mann Whitney U or Kruskal-Wallis test for ordinal variables, where appropriate. Categorical variables were presented as total number and percentage, and comparisons were made using χ2 analyses. Power analysis was conducted to confirm the appropriateness of statistical comparisons. Where appropriate, post hoc tests for one-way analysis of variance and χ2 comparisons were conducted using Tukey’s and Bonferroni adjustments, respectively. Each variable (pre-stroke risk factors, health insurance, occlusion site, procedural parameters, NIHSS score) was considered a univariate predictor via logistic regression to assess its relationship with functional outcomes. Predictors with a P value <0.25 association with functional outcomes at any of the three follow-up time points were included in a multivariate logistic regression to identify the impact of race on functional outcomes at each of the three follow-up times. Significance was determined at the P<0.05 level, and all P values are two-sided. Statistical analysis was conducted via RStudio, version 4.2.2. This study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) observational cohort guideline.
Results
Baseline differences among patients with ischemic stroke undergoing mechanical thrombectomy
Of the 326 cases in this study, 137 (42.0%) were NHB, 173 (53.1%) were NHW, and 16 (4.9%) were non-Hispanic of another race. Owing to loss to follow-up, sample size at 3, 6, and 12 months consisted of 262, 221, and 197 patients, respectively. Analysis of the initial cohort revealed a significant difference between the mean ages of NHB, NHW, and Other patients (P<0.001), and three-way χ2 analysis confirmed this difference for age >55 years (P=0.002). However, no significant pairwise differences in age were found after post hoc adjustment. χ2 analysis revealed that respiratory disease was more prevalent among NHW than among NHB patients (P=0.03). While rates of uninsurance were statistically similar between all three groups, NHB patients were more likely than NHW patients to have Medicaid coverage (P<0.001), and NHW patients were more likely than NHB patients to have Medicare coverage (P=0.001). NHW patients were also more likely to present with M1 segment occlusions than NHB patients (P=0.03); there was no difference in the distribution of occlusion sites otherwise.19 While there was a three-way difference in mean length of hospital stay (P=0.01), this difference was not significant on adjusted pairwise comparison. No statistical differences in sex, hypertension (HTN), diabetes mellitus (DM), smoking, prior stroke, atrial fibrillation, ICAD, presence of cardiac stent, cardiovascular disease, chronic kidney disease, or IV tPA administration were observed between racial groups. Procedural parameters including time from symptom discovery to hospital presentation, presentation to groin puncture, and puncture to recanalization, as well as monitored anesthesia care and stent retriever use were not statistically different between groups. Unsuccessful reperfusion (TICI <2b) was not different between groups, and neither was median baseline NIHSS score (table 1).
Characteristics of patients undergoing mechanical thrombectomy by race (n=326 cases)
Frequency of poor functional outcomes (mRS score >2) at 12 months’ follow-up was significantly different between the three groups on χ2 analysis before adjustment for confounding variables (online supplemental table S1); however, this comparison was not significant on post hoc pairwise analysis. The absolute difference in frequency of mRS score >2 between NHB and NHW patients was 11.9% at 3 months, 12.2% at 6 months, and 11.6% at 12 months (online supplemental table S1), with NHB patients showing higher frequencies of mRS >2 at all three time points. Median mRS score, frequency of mRS score >2, and mortality rates were not statistically different between groups at other time points assessed before adjustment for confounding variables. The distribution of mRS scores by race at the time points assessed is depicted by online supplemental figure S1A–C. Frequency of discharge disposition greater than 0, indicating discharge to rehabilitation or hospice/death, was not significantly different between groups (online supplemental table S2). No differences in median discharge disposition were observed between racial groups. Given the relatively small sample size of patients identifying as Other race, and that this group demonstrated no significant differences in outcomes or pre-stroke risk factors on post hoc pairwise comparisons, patients of ‘Other’ race were excluded for the remainder of the analysis.
Supplemental material
Univariate predictors associated with poor functional outcomes in patients with ischemic stroke
Several predictors showed significant univariate relationships with poor mRS outcomes on logistic regression analysis at 3-month (online supplemental table S3), 6-month (online supplemental table S4), and 12-month (online supplemental table S5) follow-up. Many of these predictors also showed significant univariate relationships with poor discharge disposition (online supplemental table S6). NHB patients were more likely to develop poor functional outcomes (mRS score >2) at 3 months (P=0.04) and 6 months (P=0.04) than NHW patients. Patients with poor mRS outcomes were also more likely to be older (3 months: P<0.001; 6 months: P<0.001; 12 months: P<0.001), have HTN (3 months: P=0.02; 6 months: P=0.02; 12 months: P=0.006), and/or have DM (3 months: P=0.005; 6 months: P=0.02; 12 months: P=0.002). Prior stroke was likewise associated with mRS score >2 on univariate analysis (3 months: P=0.01; 6 months: P=0.008; 12 months: P=0.02), as were chronic kidney disease (3 months: P=0.02; 6 months: P=0.03; 12 months: P=0.02), Medicare coverage (3 months: P=0.004; 6 months: P=0.01; 12 months: P=0.009), TICI score <2b (3 months: P=0.04), LOHS (3 months: P=0.01; 6 months: P=0.009; 12 months: P<0.001), and high NIHSS score (3 months: P=0.003; 6 months: P<0.001; 12 months: P=0.003). Private insurance coverage (3 months: P=0.02; 6 months: P=0.03) and IV tPA administration (3 months: P=0.007; 6 months: P=0.04; 12 months: P=0.04) were negatively associated with poor mRS outcomes (online supplemental tables S3–S5). NHB patients were more likely to be discharged to rehabilitation or hospice compared with NHW patients on univariate analysis (P=0.04). Age (P=0.006), DM (P=0.03), TICI score <2 (P=0.004), LOHS (P<0.001), and NIHSS (P<0.001) were also associated with poor discharge disposition (online supplemental table S6). No univariate association between poor functional outcomes (mRS score >2 or discharge disposition >0) and sex, smoking, atrial fibrillation, ICAD, presence of cardiac stent, cardiovascular disease, respiratory disease, occlusion site, or procedural parameters was observed.
Multivariate logistic regression verifies race as a predictor of poor functional outcomes in patients with ischemic stroke after adjusting for pre-stroke risk factors
Univariate predictors with P<0.25 were included in a multivariate regression analysis to determine the relationship between race and poor functional outcomes following mechanical thrombectomy. After adjustment, NHB patients were more likely than NHW patients to develop moderate–severe disability at 3 months (OR=2.115, P=0.04) (table 2) and 6 months (OR=2.423, P=0.048) (table 3), but not at 12 months (OR=2.187, P=0.15) (table 4) follow-up. Interestingly, LOHS was not significant on multivariate analysis at 3- and 6- but was significant at 12-months’ follow-up (OR=1.136, P=0.002), suggesting that longer hospital stays are associated with worse long-term, but not necessarily worse short-term, outcomes. Multivariate analysis also revealed that NHB patients were more likely to be discharged to rehabilitation, hospice, or discharged as dead compared with NHW patients (OR=1.940, P=0.04) (table 5).
Analysis of characteristics of patients who were lost to follow-up (online supplemental table S7) revealed a significant decrease in attrition at 3 months compared with 6 months (P<0.001), and at 3 months compared with 12 months (P<0.001). No significant differences were detected in the distribution of NHB compared with NHW cases lost to follow-up within or between time points. Additionally, no significant differences in frequency of poor discharge disposition were detected in cases lost to follow-up over time; however, most patients (>50%) lost to follow-up were discharged to rehabilitation or hospice care and had Medicare insurance.
Multivariate predictors of poor functional outcome (mRS score >2) in patients with ischemic stroke undergoing mechanical thrombectomy at 3 months’ follow-up (n=253 cases)
Multivariate predictors of poor functional outcome (mRS score >2) in patients with ischemic stroke undergoing mechanical thrombectomy at 6 months’ follow-up (n=214 cases)
Multivariate predictors of poor functional outcome (mRS score >2) in patients with ischemic stroke undergoing mechanical thrombectomy at 12 months’ follow-up (n=190 cases)
Multivariate predictors of poor discharge disposition in patients with ischemic stroke undergoing mechanical thrombectomy
Discussion
We find that NHB patients undergoing mechanical thrombectomy for acute ischemic stroke experience poorer 3-month and 6-month functional outcomes as measured by mRS and discharge disposition compared with NHW patients after adjusting for pre-stroke risk factors, health insurance coverage, occlusion site, and procedural parameters. Statistical difference in 12- month functional outcomes was not detected.
It is well known that NHB people in the USA face an increasingly higher risk of stroke than NHW people.18 22–24 This difference is in part due to racial disparities in the prevalence of stroke risk factors, where NHB individuals are at significantly higher risk of hypertension, diabetes, and other known comorbidities.25 Yet, little is known about the long-term post-stroke functional outcomes of racially diverse patient populations. Of the few recent studies examining the relationship between race and mortality or mRS,13–16 most stop short of assessing outcomes >90 days. Here we evaluate racial disparities in mechanical thrombectomy outcomes at 3 months, 6 months, and 12 months in a cohort of patients treated at two centers in Baltimore. Understanding the long-term relationships between race, ischemic stroke risk factors, and functional outcomes of current stroke therapies like mechanical thrombectomy presents critical opportunities to identify and attenuate disparities that may exist in in-hospital and post-interventional care. The racially diverse setting of the two centers represented in this study is particularly suited to studying this relationship.
Given the rise in implementation of mechanical thrombectomy as an effective approach to treating ischemic stroke12 and the presence of racial disparities in stroke incidence,5 6 18 22–24 26 examining the equitability of thrombectomy outcomes is imperative to improving the quality of stroke care. After stroke intervention, patients are discharged to home, rehabilitation, or hospice/death. At subsequent follow-up appointments, patients are evaluated for neurologic function and ability to carry out activities of daily living with or without assistance, among other examinations. These evaluations, in addition to discharge disposition, comprised the basis for our study of functional outcomes.
We demonstrate a significant association between NHB race and poor 3-month and 6-month mechanical thrombectomy outcomes after adjusting for pre-stroke risk factors, in-hospital factors (administration of tPA, TICI reperfusion score), initial NIHSS score, occlusion site, health insurance coverage, and procedural parameters. This disparity was not statistically detected at 12 months postprocedure; however, NHB patients were observed to have a poor outcome (mRS score>2) frequency of 80.3% compared with 68.7% in NHW patients at 12 months (online supplemental table S1).
There were no significant predictors of positive functional outcomes on multivariate analysis unique to 12-month follow-up, suggesting that the variables accounted for in this study, including access to healthcare insurance and time to hospital presentation (suggestive of access to care), do not explain the lack of outcomes disparity at 12 months compared with 3 and 6 months. It is possible that the considerable loss to follow-up (38.7%) at 12 months, which reduced the sample size at this time point, decreased statistical power and the ability of the study to detect significant differences in nominal outcomes between NHW and NHB patients at 12 months. It is also possible that post-stroke rehabilitation and recovery programs diminish these disparities noticeably at 12 but not 6 months following thrombectomy. Short-term post-stroke care may present unique challenges that yield such disparities. There may be additional factors that explain the observed disparities—for example, the availability of social support networks in post-stroke recovery can affect functional outcomes,27 as can the duration of pre-stroke comorbidities like HTN, which occurs earlier in NHB patients.28
Notably, the absolute difference in frequency of poor outcome (mRS score >2) between NHB and NHW patients remained consistent over time (online supplemental table S1), with NHB patients showing higher frequencies of mRS score >2 at all three time points. This suggests that racial disparities in post-thrombectomy functional outcomes may be persistent despite the present study’s inability to detect a difference between racial groups in the adjusted odds of developing poor outcomes at 12 months on multivariate analysis. The findings here can help to direct healthcare efforts toward monitoring the equitability of early in-hospital and, at minimum, short-term (from discharge to 6-month follow-up) postoperative care to meet the needs of the individual.
Increasingly, explanations for differences in post-stroke disability are categorized into pre-stroke, acute stroke, early recovery, and community-living periods, where factors contributing to each period are likely to drive disability differentially during these periods and possibly into the longer term.29 Since this study included patients who had undergone mechanical thrombectomy with similar NIHSS score on arrival and no differences in stroke etiology or stroke subtypes, acute stroke period factors are unlikely to account for the observed short-term outcome disparities. There are, however, factors during the recovery and community-living periods that could contribute to mortality and long-term disability. Access to post-stroke rehabilitation and follow-up appointments can affect functional recovery.14 Additionally, post-stroke comorbidities can significantly affect the progression of post-stroke outcomes, where NHB patients have been reported to fare worse as a result of these factors.28 It remains unknown, however, why racial disparities in clinical and functional outcomes 3–6 months post-discharge persist after controlling for common pre-stroke risk factors, such as CKD, which disproportionately affects NHB people.30 Future research should focus on reducing the disparities identified here by elucidating the effects of recovery and community-living period factors on functional outcomes.
Limitations
Limitations of this study include its retrospective nature and analysis of data from only two centers. Outcome measures were reported dichotomously, which could limit the ability to detect minute differences in stroke risk factors and outcomes associated with race. Several patients were lost to follow-up (specifically, 18.4%, 31.0%, and 38.7% of the initial cohort were lost to follow-up at 3, 6, and 12 months, respectively, in univariate and multivariate analyses), limiting the statistical power of the study; however, power analysis was performed and statistical significance detected at current levels, suggesting that the observed effect may persist with greater sample sizes. Discharge disposition of 0 (home) was assumed to indicate a positive outcome, including the assumption that acute/subacute rehabilitation was available to all cohort patients. While this study captures physical disability, cognitive, behavioral, and emotional outcomes are important considerations that contribute to activities of daily living in stroke survivors and should be examined in future. Given that our institution is located in a city with a higher percentage of NHB patients than the national percentage, that some stroke risk factors might have disproportionate prevalence in this region, and that the two centers sampled are close in proximity, the conclusions drawn from this study might be limited in generalizability. However, simultaneously, the proximity of the centers sampled in this study might be a strength in that it allows the inclusion of patients exposed to infrastructure within the same area, whereas studies comparing outcomes across regions with variable infrastructure, which might be difficult to account for, could be limited by confounding factors that are difficult to measure. Finally, this study investigates only a few social determinants, and future work should examine the role of socioeconomic status, obesity, psychosocial well-being, and other determinants on stroke outcomes.
Despite these limitations, our study establishes disparities in outcomes measured in two ways (mRS score and discharge disposition) at 3- and 6-months’ follow-up, which was previously uncharacterized. These findings serve as an important basis for investigating and attenuating causes of increased post-stroke disability in racial minorities.
Conclusions
Racial disparities in rates of stroke risk factors and stroke incidence are well documented. However, little is known about the relationship between race and long-term post-stroke functional outcomes, particularly following mechanical thrombectomy, a technique that is now widely adopted. Our statistical analysis finds that, in ischemic stroke treated by mechanical thrombectomy, NHB patients have significantly higher odds of developing moderate to severe disability at 3 and 6 months but not at 12 months postoperatively compared with NHW patients after adjusting for stroke risk factors, health insurance coverage, occlusion site, and procedural parameters. NHB patients are also more likely than NHW patients to be discharged to rehabilitation and hospice. This study suggests that there are factors beyond pre-stroke health status and procedural considerations that contribute to short-term but not long-term differences in health outcomes between NHB and NHW stroke patients. Additional research to determine the effects of social determinants during the stroke recovery period on short-term functional outcomes could improve parity in stroke treatment and recovery.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
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
X @vsyedavalli, @fegomd
Contributors TS and RX designed the study, monitored the data collection, and revised the paper. TS is the guarantor. TS collected the data from the online registry, analyzed the data, and drafted and revised the paper. KR and SKN collected part of the data from the online registry. AHu, CCY, RJT, JH, EM, AHi, VY, VU, PG, JMC, LFG, and RX contributed to clinical data collection. All authors critically reviewed the manuscript.
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 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.