Objective The recently published arteriovenous malformation-related intracerebral haemorrhage (AVICH) score showed better outcome prediction for patients with arteriovenous malformation (AVM)-related intracerebral haemorrhage (ICH) than other AVM or ICH scores. Here we present the results of a multicentre, external validation of the AVICH score.

Methods All participating centres (n=11) provided anonymous data on 325 patients to form the Spetzler-Martin (SM) grade, the supplemented SM (sSM) grade, the ICH score and the AVICH score. Modified Rankin score (mRS) at last follow-up (mean 25.6 months) was dichotomized into favourable (mRS 0-2, n=210) and unfavourable (mRS 3-6;n=115). Univariate and AUROC analyses were performed to validate the AVICH score.

Results Except nidus structure and AVM size, all single parameters forming the SM, sSM, ICH and AVICH score and the scores itself were significantly different between both outcome groups in the univariate analysis. The AVICH score was confirmed to be the highest predictive outcome score with an AUROC of 0.765 compared with 0.705 for the ICH score and 0.682 for the sSM grade.

Conclusion The multicentre-validated AVICH score predicts clinical outcome superior to pre-existing scores. We suggest the routine use of this score for future clinical outcome prediction and in clinical research.

Trial registration number NCT02920645.