Value of HAVOC and Brown ESUS-AF scores for atrial fibrillation on implantable cardiac monitors after embolic stroke of undetermined source

Abstract

ObjectivesUp to 20 % of ischemic strokes are associated with overt atrial fibrillation (AF). Furthermore, silent AF was detected by an implantable cardiac monitor (ICM) in 1 in 3 cryptogenic strokes in the CRYSTAL AF study. An ESC position paper has suggested a HAVOC score ≥ 4 or a Brown ESUS-AF score ≥ 2 as criteria for ICM implantation after cryptogenic stroke, but neither of these criteria has been developed or validated in ICM populations. We assessed the performance of HAVOC and Brown ESUS-AF scores in a cohort of ICM patients implanted after embolic stroke of undetermined source (ESUS). MethodsAll patients implanted with an ICM for ESUS between February 2016 and February 2022 at two French University Hospitals were retrospectively included. Demographic data, cardiovascular risk factors, and clinical and biological data were collected after a review of electronic medical records. HAVOC and Brown ESUS-AF scores were calculated for all patients. FindingsAmong the 384 patients included, 106 (27 %) developed AF during a mean follow-up of 33 months. The scores performances for predicting AF during follow-up were: HAVOC= AUC: 68.5 %, C-Index: 0.662, and Brown ESUS-AF=AUC: 72.9 %, C-index 0.712. Compared with the CHA2DS2-VASc score, only the Brown ESUS-AF score showed significant improvement in NRI/IDI. Furthermore, classifying patients according to the suggested HAVOC and Brown ESUS-AF thresholds, only 24 % and 31 % of the cohort, respectively, would have received an ICM, and 58 (55 %) and 47 (44 %) of the AF patients, respectively, would not have been implanted with an ICM. ConclusionHAVOC and Brown ESUS-AF scores showed close and moderate performance in predicting AF on ICM after cryptogenic stroke, with a significant lack of sensitivity. Specific risk scores should be developed and validated in large ICM cohorts.