Echocardiographic strain markers and subsequent atrial fibrillation detection after cryptogenic stroke: an updated systematic review and meta-analysis

Background Undetected atrial fibrillation (AF) is an important potential mechanism after cryptogenic stroke. Speckle-tracking echocardiography-derived left atrial (LA) phasic strain and left ventricular global longitudinal strain (LV GLS) have been reported to differ between patients with and without AF subsequently detected after stroke, but the magnitude and consistency of these group-level differences remain uncertain. Methods PubMed, Embase, and Web of Science were searched from inception to May 3, 2026. Eligible reports enrolled adults with cryptogenic ischemic stroke or embolic stroke of undetermined source (ESUS); mixed undetermined-source cerebrovascular cohorts were retained only when the source report explicitly framed the population as cryptogenic stroke/ESUS or undetermined-source cerebrovascular disease. No numerical percentage threshold for cohort composition had been prespecified in the registered protocol. Reports excluded known AF at baseline, measured LV GLS or LA phasic strain by speckle-tracking echocardiography, and reported AF detected during follow-up. Random-effects meta-analyses used raw mean differences (MDs) and complementary Hedges g standardized mean differences (SMDs), with restricted maximum-likelihood estimation, Hartung-Knapp confidence intervals (CIs), and prediction intervals. Prespecified outcome-level deduplication was retained; revision-stage analyses included a monitoring-duration subgroup analysis, a cohort-only reservoir-strain sensitivity analysis, and a conduit-strain sensitivity analysis addressing unresolved values in Bufano et al. Results Nineteen reports representing 17 cohort clusters were included. Outcome-specific analyses comprised 7 datasets for LV GLS ( n = 1,403), 17 for LA reservoir strain ( n = 2,755), 10 for the primary LA conduit-strain analysis excluding Bufano et al. ( n = 1,958), and 14 for LA contractile strain ( n = 2,455). Patients with subsequently detected AF had lower LA reservoir strain (MD, -7.44% points; 95% CI, -9.52 to -5.36; SMD, -0.87; 95% CI, -1.11 to -0.62), LA conduit strain in the primary analysis excluding Bufano et al. (MD, -3.29; 95% CI, -5.17 to -1.41; SMD, -0.45; 95% CI, -0.65 to -0.24), and LA contractile strain (MD, -4.31; 95% CI, -5.96 to -2.65; SMD, -0.80; 95% CI, -1.10 to -0.50). LV GLS did not differ significantly (MD, -1.15; 95% CI, -2.89 to 0.60; SMD, -0.37; 95% CI, -0.99 to 0.25). Prediction intervals crossed the null for all outcomes. In the exploratory monitoring-duration subgroup analysis of 15 LA reservoir-strain datasets, there was no statistically significant between-category difference ( P = 0.9275). Conclusions Baseline LA phasic strain was lower, on average, in patients with AF subsequently detected after cryptogenic stroke or ESUS, with the largest standardized group separation observed for LA reservoir strain. However, substantial heterogeneity, null-crossing prediction intervals, and predominantly aggregate unadjusted data preclude conclusions about independent predictive performance or individualized discrimination. These findings support prospective evaluation of LA strain as one component of multivariable strategies for selecting patients for prolonged rhythm monitoring, rather than as a standalone diagnostic or treatment-decision tool.