Background Several studies have suggested that after ischemic stroke, continuous electrocardiographic (ECG) monitoring (CEM) increases the atrial fibrillation (AF) detection rate. However, optimal ...CEM terms of use are not clear. The aim of our study was to evaluate the usefulness of CEM in detecting AF and define optimal terms of the use of CEM. Methods We prospectively enrolled consecutive patients with acute ischemic stroke who were admitted to the stroke unit without AF on baseline ECG. We compared 2 strategies of AF detection: the first using CEM and the second with routine clinical practice (24-hour Holter ECG and additional ECGs). Adjusted odds ratios for the association between AF diagnosis and the use of CEM stratified by monitoring duration were calculated using multivariate logistic regression analysis. Results Of the 1166 patients included, 220 (18.87%) had AF on baseline ECG and were excluded. Of the 946 remaining patients, 592 underwent CEM. The prevalence of AF using CEM was 12.50% compared 2.26% using the routine strategy. After adjustment (demographic data, vascular risk factors, and National Institutes of Health Stroke Scale scores), using CEM increased 5.29 fold the odds of finding AF (95% confidence interval CI 2.43-11.55) compared to the routine strategy. The adjusted odds ratio (9.82; 95% CI 3.01-32.07) was maximum for the first day of monitoring and decreased later. Beyond 5 days, CEM usefulness was not significantly higher than the routine strategy. Conclusions We suggest that in order to enhance the detection rate of AF, CEM could be generalized in the stroke unit. It must be started early in patients with acute stroke and prolonged over a minimum of 4 days.
Detection of new atrial fibrillation (AF) after ischemic stroke is challenging. The aim of the TARGET-AF study was to identify relevant markers for ruling out delayed AF in stroke patients. Early and ...prolonged Holter electrocardiography (ECG) monitoring during hospitalization was performed systematically in consecutive acute stroke patients naive to AF (no history of AF or no AF on baseline ECG). All clinical and paraclinical data for routine etiologic assessment were collected. The diagnostic value of all parameters significantly associated with AF was assessed by comparison of area under the receiver operating characteristic curve (AUC). Of the 300 stroke patients enrolled (mean age, 62.5 ± 15.5 years; sex ratio: 1.7; mean National Institutes of Health Stroke Scale score, 7.1 ± 7.9, median duration of Holter ECG monitoring, 6.8 days), 52 (17.3%) had newly diagnosed AF. Parameters significantly associated with AF were classified by increasing AUC: anterior circulation localization (AUC, 0.604; 95% confidence interval CI, 0.546-0.660), P-wave initial force (AUC, 0.608; 95% CI, 0.545-0.669), left atrial dilatation (AUC, 0.657; 95% CI, 0.600-0.711), National Institutes of Health Stroke Scale score (AUC, 0.667; 95% CI, 0.611-0.720), sex (AUC, 0.683; 95% CI, 0.627-0.736), age (AUC, 0.755; 95% CI, 0.707-0.797), CHA2 DS2 -VASc score (AUC, 0.796; 95% CI, 0.746-0.841), STAF (score for the targeting of AF) score (AUC, 0.842; 95% CI, 0.796-0.882), and plasma brain natriuretic peptide (BNP) level (AUC, 0.868; 95% CI, 0.825-0.904). The use of all parameters combined (AUC, 0.910; 95% CI, 0.872-0.940) was not significantly more efficient in diagnosing AF than BNP alone ( P = .248). At the Youden plot, the diagnostic properties for BNP >131 pg/mL were sensitivity, 98.1% (95% CI, 89.7-99.7); specificity, 71.4% (95% CI, 65.3-76.9); and negative predictive value, 99.4% (95% CI, 96.9-99.9). Our data indicate that a BNP level ≤131 pg/mL might rule out delayed AF in stroke survivors and could be included in algorithms for AF detection.