Biousse, Valérie

ImportanceAtrial fibrillation (AF) is the most common, chronic, cardiac arrythmia in older US adults. It is not known whether AF is independently associated with increased risk of retinal stroke (central retinal artery occlusion), a subtype of ischemic stroke that causes severely disabling visual loss in most cases and is a harbinger of further vascular events.ObjectiveTo determine whether there is an association between AF and retinal stroke.Design, Setting, and ParticipantsThis retrospective cohort study was performed between July 2023 and May 2024 using computerized inpatient, outpatient, emergency department, and skilled nursing facility claims files for a 5% sample of US fee-for-service Medicare beneficiaries aged 66 years and older between 2000 and 2020. Follow-up ended at death, conclusion of fee-for-service Medicare coverage, end of the study period, or loss to follow-up of 85% of the study cohort.ExposureAF, based on validated International Classification of Diseases, Ninth Revision, Clinical Modification and International Statistical Classification of Diseases, Tenth Revision, Clinical Modification diagnosis codes.Main Outcomes and MeasuresThe primary end point was incident retinal stroke in the primary diagnostic position of a single claim in any venue of care. Secondary end points included retinal stroke in any position of a single claim, 1 positive control end point (cerebral ischemic stroke), and 4 negative control end points (central retinal vein occlusion, urinary tract infection, humeral fracture, and cataract). Unadjusted and adjusted hazard ratios (HRs) and rate differences were computed across matched and overlap-weighted cohorts with and without AF (defined as 1 inpatient claim or 2 outpatient claims within a 365-day period).ResultsIn total, 1 090 144 patients (591 400 female [54.3%]; mean [SD] age, 76.92 [7.09] years) were included in the study; 545 072 patients had AF and 545 072 were matched controls. The median (IQR) follow-up was 45 (18 to 90) months. In total, 1333 patients with AF (rate, 0.55 per 1000 person-years) and 1082 AF-free matched controls (rate, 0.50 per 1000 person-years) experienced retinal stroke. The cause-specific, adjusted HR of retinal stroke after overlap weighting was 1.14 (95% CI, 1.02 to 1.28; adjusted rate difference, 0.05 [95% CI, −0.01 to 0.11]). AF was associated with cerebral ischemic stroke (adjusted HR, 1.73 [95% CI, 1.69 to 1.76]; adjusted rate difference, 10.11 [95% CI, 9.72 to 10.49]). Of 4 prespecified negative control end points, AF was not associated with central retinal vein occlusion but was associated with urinary tract infection, cataract, and humeral fracture.Conclusions and RelevanceIn this cohort study of Medicare beneficiaries aged 66 years and older, AF was independently associated with retinal stroke. The magnitude of the association was small, and a contribution from residual, unmeasured confounding could not be excluded.

ObjectiveTo compare the diagnostic performance of an artificial intelligence deep learning system with that of expert neuro‐ophthalmologists in classifying optic disc appearance.MethodsThe deep learning system was previously trained and validated on 14,341 ocular fundus photographs from 19 international centers. The performance of the system was evaluated on 800 new fundus photographs (400 normal optic discs, 201 papilledema [disc edema from elevated intracranial pressure], 199 other optic disc abnormalities) and compared with that of 2 expert neuro‐ophthalmologists who independently reviewed the same randomly presented images without clinical information. Area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity were calculated.ResultsThe system correctly classified 678 of 800 (84.7%) photographs, compared with 675 of 800 (84.4%) for Expert 1 and 641 of 800 (80.1%) for Expert 2. The system yielded areas under the receiver operating characteristic curve of 0.97 (95% confidence interval [CI] = 0.96–0.98), 0.96 (95% CI = 0.94–0.97), and 0.89 (95% CI = 0.87–0.92) for the detection of normal discs, papilledema, and other disc abnormalities, respectively. The accuracy, sensitivity, and specificity of the system's classification of optic discs were similar to or better than the 2 experts. Intergrader agreement at the eye level was 0.71 (95% CI = 0.67–0.76) between Expert 1 and Expert 2, 0.72 (95% CI = 0.68–0.76) between the system and Expert 1, and 0.65 (95% CI = 0.61–0.70) between the system and Expert 2.InterpretationThe performance of this deep learning system at classifying optic disc abnormalities was at least as good as 2 expert neuro‐ophthalmologists. Future prospective studies are needed to validate this system as a diagnostic aid in relevant clinical settings. ANN NEUROL 2020;88:785–795