Toggle Main Menu Toggle Search

Open Access padlockePrints

Prediction of Outcome and Endovascular Treatment Benefit: Validation and Update of the MR PREDICTS Decision Tool

Lookup NU author(s): Dr Philip White



This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).


© 2021 Georg Thieme Verlag. All rights reserved.Background and Purpose: Benefit of early endovascular treatment (EVT) for ischemic stroke varies considerably among patients. The MR PREDICTS decision tool, derived from MR CLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands), predicts outcome and treatment benefit based on baseline characteristics. Our aim was to externally validate and update MR PREDICTS with data from international trials and daily clinical practice. Methods: We used individual patient data from 6 randomized controlled trials within the HERMES (Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke Trials) collaboration to validate the original model. Then, we updated the model and performed a second validation with data from the observational MR CLEAN Registry. Primary outcome was functional independence (defined as modified Rankin Scale score 0-2) 3 months after stroke. Treatment benefit was defined as the difference between the probability of functional independence with and without EVT. Discriminative performance was evaluated using a concordance (C) statistic. Results: We included 1242 patients from HERMES (633 assigned to EVT, 609 assigned to control) and 3156 patients from the MR CLEAN Registry (all of whom underwent EVT within 6.5 hours). The C-statistic for functional independence was 0.74 (95% CI, 0.72-0.77) in HERMES and, after model updating, 0.80 (0.78-0.82) in the Registry. Median predicted treatment benefit of routinely treated patients (Registry) was 10.3% (interquartile range, 5.8%-14.4%). Patients with low (<1%) predicted treatment benefit (n=135/3156 [4.3%]) had low rates of functional independence, irrespective of reperfusion status, suggesting potential absence of treatment benefit. The updated model was made available online for clinicians and researchers at Conclusions: Because of the substantial treatment effect and small potential harm of EVT, most patients arriving within 6 hours at an endovascular-capable center should be treated regardless of their clinical characteristics. MR PREDICTS can be used to support clinical judgement when there is uncertainty about the treatment indication, when resources are limited, or before a patient is to be transferred to an endovascular-capable center.

Publication metadata

Author(s): Venema E, Roozenbeek B, Mulder MJHL, Brown S, Majoie CBLM, Steyerberg EW, Demchuk AM, Muir KW, Davalos A, Mitchell PJ, Bracard S, Berkhemer OA, Lycklama A Nijeholt GJ, Van Oostenbrugge RJ, Roos YBWEM, Van Zwam WH, Van Der Lugt A, Hill MD, White P, Campbell BCV, Guillemin F, Saver JL, Jovin TG, Goyal M, Dippel DWJ, Lingsma HF

Publication type: Article

Publication status: Published

Journal: Stroke

Year: 2021

Volume: 52

Issue: 9

Pages: 2764-2772

Print publication date: 01/09/2021

Online publication date: 16/07/2021

Acceptance date: 29/03/2021

Date deposited: 13/09/2021

ISSN (print): 0039-2499

ISSN (electronic): 1524-4628

Publisher: Lippincott Williams and Wilkins


DOI: 10.1161/STROKEAHA.120.032935

PubMed id: 34266308


Altmetrics provided by Altmetric