An 87-year-old woman with quiescent rheumatoid arthritis, not on immunosuppressive therapy, presented with unilateral arm weakness, confusion and visual hallucinations. There were no infective symptoms or history of malignancy. Cerebrospinal fluid (CSF) analysis demonstrated lymphocytosis and raised protein, without flow cytometric or cytological abnormalities. Viral, bacterial, mycobacterial and fungal testing of CSF and serum were negative. MRI brain indicated unilateral leptomeningeal enhancement. There was no evidence of occult malignancy on CT imaging of the chest, abdomen and pelvis. Rheumatoid factor and anticyclic citrullinated peptide were strongly positive. The patient declined meningeal biopsy but responded to treatment with corticosteroid therapy.

In 25% of patients presenting with embolic stroke, a cause is not determined. Atrial fibrillation (AF) is a commonly identified mechanism of stroke in this population, particularly in older patients. Conventional investigations are used to detect AF, but can we predict AF in this population and generally? We performed a systematic review to identify potential predictors of AF on 12-lead electrocardiogram (ECG).MethodWe conducted a search of EMBASE and Medline databases for prospective and retrospective cohorts, meta-analyses or case-control studies of ECG abnormalities in sinus rhythm predicting subsequent atrial fibrillation. We assessed quality of studies based on the Newcastle-Ottawa scale and data were extracted according to PRISMA guidelines.ResultsWe identified 42 studies based on our criteria. ECG patterns that predicted the risk of developing AF included interatrial block, P-wave terminal force lead V1, P-wave dispersion, abnormal P-wave-axis, abnormal P-wave amplitude, prolonged PR interval, left ventricular hypertrophy, QT prolongation, ST-T segment abnormalities and atrial premature beats. Furthermore, we identified that factors such as increased age, high CHADS-VASC, chronic renal disease further increase the positive-predictive value of some of these parameters. Several of these have been successfully incorporated into clinical scoring systems to predict AF.ConclusionThere are several ECG abnormalities that can predict AF both independently, and with improved predictive value when combined with clinical risk factors, and if incorporated into clinical risk scores. Improved and validated predictive models could streamline selection of patients for cardiac monitoring and initiation of oral anticoagulants.

No abstract