Article Text

Download PDFPDF

Original research
#Fake news: a systematic review of mechanical thrombectomy results among neurointerventional stroke surgeons on Twitter
  1. Adam Andrew Dmytriw1,
  2. Thomas Joseph Sorenson2,3,
  3. Jonathan M Morris4,
  4. Patrick J Nicholson1,
  5. Christopher Alan Hilditch1,
  6. Christopher S Graffeo3,
  7. Waleed Brinjikji3,4
  1. 1 Department of Medical Imaging, University of Toronto, Toronto, Ontario, Canada
  2. 2 School of Medicine, University of Minnesota, Minneapolis, Minnesota, USA
  3. 3 Department of Neurologic Surgery, Mayo Clinic, Rochester, Minnesota, USA
  4. 4 Department of Radiology, Mayo Clinic, Rochester, Minnesota, USA
  1. Correspondence to Dr Waleed Brinjikji, Departmentof Radiology, MayoClinic, Rochester MN 55902, USA; brinjikji.waleed{at}mayo.edu

Abstract

Objective Twitter is a popular social media platform among physicians. Neurointerventionalists frequently document their lifesaving mechanical thrombectomy cases on Twitter with very favorable results. We fear that there may be some social media publication bias to tweeted mechanical thrombectomy cases with neurointerventionalists being more likely to tweet cases with favorable outcomes. We used these publicly documented cases to analyze post-intervention Twitter-reported outcomes and compared these outcomes with the data provided in the gold standard literature.

Methods Two reviewers performed a search of Twitter for tweeted cases of acute ischemic strokes treated with mechanical thrombectomy. Data were abstracted from each tweet regarding baseline characteristics and outcomes. Twitter-reported outcomes were compared with the Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke (HERMES) trial individual patient meta-analysis.

Results When comparing the tweeted results to HERMES, tweeted cases had a higher post-intervention rate of modified Thrombolysis In Cerebral Infarction (mTICI) scale score of 2c/3 (94% vs 71%, respectively; p<0.0001) and rate of National Institutes of Health Stroke Scale (NIHSS) score ≤2 (81% vs 21%, respectively; p<0.0001). There were no reported complications; thus, tweeted cases also had significantly lower rates of complications, including symptomatic intracerebral hemorrhage (0% vs 4.4%, respectively; p<0.0001), type 2 parenchymal hemorrhage (0% vs 5.1%, respectively; p<0.0001), and mortality (0% vs 15.3%, respectively; p<0.0001).

Conclusions There is a significant difference between social media and reality even within the ‘MedTwitter’ sphere, which is likely due to a strong publication bias in Twitter-reported cases. Content on ‘MedTwitter’, as with most social media, should be accepted cautiously.

  • stroke
  • thrombectomy
  • technology
  • technique
View Full Text

Statistics from Altmetric.com

Introduction

Twitter is a widely-used social media platform for interacting with an ever-expanding network of electronic friends, or ‘followers’. As with most social media platforms (eg, Facebook, Instagram, LinkedIn), Twitter is regularly used for publicly sharing personal successes with the intention of garnering immediate external gratification from followers. A medical subculture within Twitter, termed ‘MedTwitter’, has recently exploded and, unsurprisingly, adopted similar practices. Neurointerventionalists, especially those who perform mechanical thrombectomies, are not immune to these habits.

Due to the relatively high frequency of endovascular-treated acute ischemic stroke (AIS) and the highly competitive nature of neurointerventionalists, social media documentation of ‘lifesaving’ mechanical thrombectomy performances and post-recanalization bravado (eg, ‘helping patients cheat death’) is abundantly used as ammunition in the perpetual battle for increased social capital among peers. In fact, mechanical thrombectomy cases are among the most easily tweetable cases in medicine since a comprehensive medical history (ie, age/gender, baseline National Institutes of Health Stroke Scale (NIHSS) score and occlusion location), as well as treatment results (ie, final modified Thrombolysis In Cerebral Infarction (mTICI) score, final NIHSS score, number of passes), can easily be objectively conveyed within the 260-character limit. The inclusion of pre- and post-treatment angiograms and a dramatic picture of the clot are also often included to further enhance the online reputation of the neurointerventionalist or stroke center.

While many of these tweets are educational and instructive to fellow practitioners, we suspect that there may be some social media publication bias among tweeted cases. To better study the publication bias among tweeted thrombectomy cases, we performed a systematic review of neurointerventional MedTwitter to analyze post-intervention Twitter-reported outcomes and compare these outcomes with the data provided in the established gold standard literature.

Materials and methods

This observational review adhered to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) recommendations as best as possible given the nature of this review.1

Literature search, search selection, and eligibility criteria

One board-certified neuroradiologist and one radiology resident conducted a comprehensive search of Twitter for tweets tweeted in English from July 2006 to July 2018. Searched hashtags were: #stroke, #leavenoELVObehind, #radialfirst, #survivestroke, #ELVO, #TICI, #TICI3orBust, #thrombolysis, and #GTFVO. In addition, all remaining tweets from neurointerventional surgeons who used the above-listed hashtags were screened for additional self-reported stroke cases that did not use one of the above hashtags. Tweets were included in our analysis if they (1) reported a tweeter-performed mechanical thrombectomy for acute ischemic stroke secondary to a large vessel occlusion and (2) they reported final mTICI score, pre- and post-neurological status, and/or technical details from the procedure. Tweets were excluded if the mechanical thrombectomy was not performed by the tweeter or if the tweeter reported a mechanical thrombectomy but did not provide any of these data publicly on Twitter.

Data abstraction

Once the final list of tweets meeting the inclusion criteria was compiled, the following data were abstracted from each tweet and/or tweeter to the best of our ability: name of tweeter, number of mechanical thrombectomy cases tweeted during the reviewed period, and case details (location of occlusion; how access was obtained (femoral vs radial); number of passes; pre- and post-intervention NIHSS score; post-intervention mTICI score; and objective statements of post-intervention neurological status). Any angiographic images included in the tweet were also reviewed by a board-certified neuroradiologist (WB) to determine if final mTICI scores were reasonably reported and proper Twitter decorum was followed.

Outcomes

The primary outcomes of this review were rate of ‘substantial endovascular reperfusion’ (defined as post-mechanical thrombectomy mTICI score 2c or 3),2 incidence of post-treatment NIHSS score ≤2, rate of ‘major early neurological recovery’ (defined as a reduction in NIHSS score of at least eight points or reaching 0–1 within 24 hours),3 rate of ‘first pass effect’ (defined as one-pass mTICI 3 revascularization without rescue therapy),4 and complication rate (ie, vessel perforation, dissection, vasospasm, etc). Tweeted cases were only included in these analyses if objective data were provided; any subjective comments about post-treatment status (eg, ‘significant improvement’, ‘complete recovery’, ‘fine next day’) were not used in place of a NIHSS score. To better understand the publication bias of tweeted cases, outcomes from our series of tweeted cases were compared with outcomes reported in the Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke (HERMES) trial individual patient meta-analysis of the mechanical thrombectomy stroke trials published in 2015.2–4

Statistical analysis

Descriptive statistics were reported as mean and SD for continuous variables and proportion and percentage for categorical variables. Comparisons of means were performed with a two-tailed t-test. Comparisons of proportions were performed with an ‘N−1’ χ2 test.5 6 Before running any statistical tests, the alpha (α) level was set at 0.05. All statistical analyses were performed using commercially available software (JMP 13, SAS).

Results

Demographics

Thirty-six neurointerventional surgeons tweeted 115 cases of mechanical thrombectomy for the treatment of AIS secondary to large vessel occlusion. The majority (100/115; 87%) of cases were tweeted by academic neurointerventionalists rather than private practitioners. The median number of followers for the neurointerventionalist tweeters was 782 (range 54–2459) people. Cases were frequently ‘retweeted’ and ‘favorited’ by both physicians and non-physicians alike. Most (16/36; 44%) neurointerventional surgeons tweeted only one case. Location of the occlusion was provided in 113 (113/115; 98%) tweeted cases; the most common location of occlusion was the middle cerebral artery (71/113; 63%). The most common access method was through the femoral artery (105/115; 91%). Full demographic data are summarized in table 1.

Table 1

Summary of demographic information

Outcomes

One hundred and four (104/115; 90%) tweeted cases reported post-intervention mTICI scores. mTICI 3 was the most common score (95/104; 91%) with three patients (3/104; 2.9%) scoring mTICI 2c, five (5/104; 4.8%) scoring mTICI 2b, and one (1/104; 0.96%) scoring mTICI 0 (after seven passes at a basilar artery occlusion).

Twenty-six (26/115; 23%) tweeted cases included both a pre- and post-mechanical thrombectomy NIHSS score. The mean±SD pre-treatment NIHSS score was 17.8±5.07 and the mean±SD post-intervention NIHSS score was 2.6±3.2. This difference in pre- and post-intervention NIHSS score was highly statistically significantly (p<0.0001). Seventeen (17/26; 81%) tweeted cases reported a post-intervention NIHSS score ≤2; 25 (25/26; 96%) tweeted cases reported ‘major early neurological recovery’.3

Only seventeen (17/115; 15%) tweeted cases reported the number of passes required for recanalization of the vessel, but most (13/17; 76%) experienced ‘first pass effect’. Full data of outcomes are summarized in table 2.

Table 2

Summary of outcomes

Of the 115 tweeted cases, there were no cases of vasospasm, vessel perforation, dissection, or emboli to previously non-affected territories. There were also no reported cases of symptomatic intracerebral hemorrhage, type 2 parenchymal hematoma, or mortality.

Comparison of outcomes with current literature

When comparing the tweeted results with the HERMES results, the tweeted cases had a higher rate of ‘substantial endovascular reperfusion’ than the published rate (94% vs 71.0%, respectively; p<0.0001)2 and a higher rate of post-intervention NIHSS score ≤2 than the published rate (81% vs 21%, respectively; p<0.0001).3 When comparing tweeted results with the study by Zaidat et al, the tweeted cases had a higher rate of ‘first pass effect’ than the published rate (76% vs 25.1%, respectively; p<0.0001).4

Tweeted cases had significantly lower rates of symptomatic intracerebral hemorrhage (0% vs 4.4%, respectively; p<0.0001), type 2 parenchymal hemorrhage (0% vs 5.1%, respectively; p<0.0001), and mortality (0% vs 15.3%, respectively; p<0.0001).3 Comparisons of outcomes are summarized in table 3.

Table 3

Comparison of outcomes from tweeted cases with outcomes reported in the published literature

Discussion

Our review of Twitter found that post-intervention rates of ‘substantial endovascular reperfusion,’ NIHSS score ≤2, and rates of ‘first pass effect’ as well as complication rates for tweeted cases are extremely favorable. Differences between outcomes reported on Twitter and outcomes reported in the literature were strongly statistically significant and heavily favored Twitter-reported cases. We also found that a few individuals (8.3%) dominated the social media stroke scene, reporting over 40% of the tweeted cases. Additionally, despite the push on Twitter for ‘#radialfirst,’ the femoral artery was still overwhelmingly the most popular route of access for tweeted cases (91%), with only one report of direct puncture of the carotid artery.

The substantially improved outcomes reported on Twitter are hypothesized to be the result of two possible factors (which are not mutually exclusive): (1) higher mechanical thrombectomy abilities (and, thus, better results) among tweeting neurointerventionalists and/or (2) social media publication bias. We strongly suspect the latter. Although social media has been shown to be a valuable tool in medical communication, our findings raise the specter that MedTwitter is not entirely reflective of reality. This notion brings into question why we, as physicians, are portraying ourselves on social media with outcomes that are hyper-idealistic? MedTwitter is surely growing into a place of reference for medical education, and the implications of sharing cases publicly on social media are thus becoming progressively greater. As the influence of this community grows, it is very important that realistic expectations are provided to patients, families, and caregivers (who increasingly turn to this community for reference) regarding the results that can be routinely achieved. As in the traditional face-to-face practice of medicine, authenticity remains paramount. While discrepancies in outcomes between self-adjudicated registry data and blinded data, in addition to local proceduralist and core laboratory interpretation, are well known,7–9 the volume of mechanical thrombectomy cases being publicly reported on social media platforms renders this a pressing issue for the neurointerventional community in particular.

It is essential to recognize that, while Twitter first and foremost represents a social media and advertising platform, the public may not consistently recognize this while reviewing posts of healthcare professionals. This is not unique, as many other media sources also perpetuate the impression that mechanical thrombectomy is miraculous,10 11 but remains problematic, nonetheless. Although positive public opinion is appreciated, there is undoubtedly the caveat of inflated unrealistic expectations, and these might be a responsibility to medical practitioners to temper the unbridled evangelism of novel ‘miraculous’ therapies. Further perpetuating unrealistic expectations on a public-facing platform may be counterproductive for the neurointerventional community as a whole. Additionally, the reporting of anonymized cases without expressed consent of the patient is an ethical quandary with continuous debate.12

However, despite the above concerns, we believe that Twitter remains a highly valuable tool for widely sharing information and convening with a like-minded diverse group of individuals. Anecdotally, the senior author, who is an active tweeter of thrombectomy cases, has experienced subjective improvement in his practice by learning of and implementing new techniques based on recommendations from his colleagues on Twitter. Connections with neurointerventional colleagues across the globe are enhanced through these social media interactions, which can foster real collaborative scientific efforts. However, there is still a significant difference between social media and reality.

Limitations of the study

The main limitation with our results is that data were abstracted as best as possible from each tweet, but there was no consistent tweet format, so data were also inconsistent. Further standardization of mechanical thrombectomy tweets is recommended including standardization of outcomes, clinical presentation data, and hashtags. Lastly, the conclusions of this review are based on non-verifiable, non-reviewed, possibly biased reports of cases as they were publicly distributed on Twitter, so detailed analytic measures of the included tweets is largely impossible. Nevertheless, this represents the first and largest systematic review on the disparity between neurointerventional MedTwitter and real life, which will only become progressively more relevant with time.

Conclusions

Differences between outcomes reported on Twitter and outcomes reported in the literature were strongly statistically significant and heavily favored Twitter-reported cases. These findings suggest that there is a strong publication bias in Twitter-reported cases and that content on MedTwitter, as with most social media, should be accepted cautiously.

References

View Abstract

Footnotes

  • Contributors AAD made substantial contributions to the conception and design of the work, data acquisition, and analysis. TJS and WB made substantial contributions to the conception and design of the work, data acquisition, analysis and interpretation, and drafting of the work. JMM, PJN, CAH, and CSG made substantial contributions to the conception and design of the work and revising it critically for important intellectual content. All authors gave final approval of the version to be published and agree to be accountable for all aspects of the work.

  • 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.

  • Patient consent Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data were gathered from publically available tweets and are available upon request from the corresponding author.

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.