Article Text

Download PDFPDF
P-019 Clinical Conviction: Selection Bias In Recent Endovascular Stroke Trials
  1. S Sheth1,
  2. J Saver1,
  3. R Jahan2,
  4. S Starkman1,
  5. G Duckwiler2,
  6. S Tateshima2,
  7. N Gonzalez2,
  8. D Liebeskind1,
  9. Ucla Stroke Investigators1
  1. 1Department of Neurology, University of California, Los Angeles, CA, USA
  2. 2Division of Interventional Neuroradiology, University of California, Los Angeles, CA, USA


Introduction Differing biases may affect enrollment patterns and result in trial participants that are not representative of clinical practice nor ideal candidates for therapy. We sought to evaluate the stringency of inclusion/exclusion criteria for five recent endovascular stroke treatment (EST) trials and then measure any bias in the decision making process for EST trial enrollment at our institution.

Materials and methods From a prospectively maintained registry, all patients treated in clinical practice with EST from April 2004 to June 2013 were identified. The decision to perform EST at our institution is based upon the clinical history, examination, and favorable imaging pattern of proximal vessel occlusion without evidence of significant ischemic injury. Trial eligibility criteria from EST trials active at our institution were used to determine the proportion and characteristics of patients eligible for each trial.

Results Among 299 patients treated with EST, average age was 68, 57% were female, NIHSS was 17, and 45% received IV tPA. 125/269 (47%) would have been eligible for MR RESCUE, 33/208 (16%) for IMS III, 10/20 (50%) for SWIFT, 32/41 (78%) for MultiMERCI, 6/11 (55%) for TREVO2. Patients enrolled in trials numbered 57, 21% of all patients in the study time period and 17–50% of eligible patients. Using the trial with the most enrollments as an example (22 patients, MR RESCUE), reasons for inability to enrol included concurrent enrollment in another trial (43%), referring physician refusal for randomization (35%), inability to complete imaging (8%) and inability to obtain consent (7%). Compared to patients enrolled in the trial, those eligible but treated outside of a trial presented with similar distributions of age (68 vs. 69, p = 0.73) and gender (female 52% vs. 55%, p = 1.0). Eligible patients treated outside of trials presented with lower NIHSS (17 vs. 20, p < 0.05), arrived to the ED sooner after symptom onset (185 vs. 285 min, p < 0.05) and began EST sooner after symptom onset (315 vs. 417 min, p < 0.05). TICI2b/3 recanalization rates were similar (53% vs. 63%, p = 0.72). Outcomes for eligible patients treated outside of trials were improved: more were discharged home (21% vs. 0%, p < 0.05) and achieved discharge modified Rankin scale 0–3 (39% vs. 5%, p < 0.05) compared to patients enrolled in the study. Specifically, the subset that was not enrolled due to refusal of randomization also presented sooner after symptom onset (198 vs. 285 min, p < 0.05), had smaller infarct volumes at presentation (26 vs. 49 mL, p < 0.05), and had improved outcomes at discharge compared to those enrolled (mRS 0–3 36% vs. 5%, p < 0.05).

Conclusions Endovascular trials excluded the majority of patients deemed suitable for ERT in routine practice. Competing interests from other trials as well as reluctance to randomise against EST limited enrollment. The population of patients enrolled in trials tended to present later, have more severe strokes, and suffer worsened outcomes. Conviction in the benefit of EST led to the treatment of patients most likely to benefit from EST outside of trials. As a result, recent EST trials may have not reflected the treatment population and outcomes of routine practice.

Disclosures S. Sheth: None. J. Saver: None. R. Jahan: None. S. Starkman: None. G. Duckwiler: None. S. Tateshima: None. N. Gonzalez: None. D. Liebeskind: None.. . UCLA Stroke Investigators: None.

Statistics from

Request Permissions

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