Natural courses of atrial fibrillation following cryptogenic stroke: A systematic review and meta-analysis.

Background: Cryptogenic stroke (CS) presents challenges in management and increases risk of stroke recurrence. The clinical relevance of atrial fibrillation detected after stroke (AFDAS) remains equivocal.

Methods: Our study was conducted in a meta-analysis design, including all observational studies and utilizing random-effects model. The objectives were to explore the association between predictors and AFDAS incidence, and clinical differences between AF and non-AF patients following CS.

Results: This meta-analysis included 59 studies with 18,683 patients with CS. AFDAS incidence was 20.8 %. In univariable analysis, significant predictors of AFDAS were age >75 years (OR 6.94, 95 % CI: 3.12-15.43, I² = 0 %) and hypertension (OR 2.16, 95 % CI: 1.28-3.63, I² = 0.0 %). In comparison to non-AF group, individuals with AF showed significantly higher levels of left atrial volume index (MD 7.50 mL/m², 95 % CI: 6.20-8.80, I² = 0.0 %), left atrial diameter (MD 0.17 cm, 95 % CI: 0.02-0.32, I² = 48.6 %), D-dimer (MD 0.20 mg/L, 95 % CI: 0.05-0.36, I² = 0.0 %), CHA₂DS₂VASc score (MD 0.76 point, 95 % CI: 0.53-0.98, I² = 64.2 %), and P-wave duration (MD 9.49 ms, 95 % CI: 0.75-18.23, I² = 30.4 %). Significantly lower levels were observed in left atrial contractile strain (MD -5.68 %, 95 % CI: 7.45 to -3.91, I² = 58.0 %) and left atrial reservoir strain (MD -9.05 %, 95 % CI: 12.2 to -5.91, I² = 71.9 %). The mean time from CS to AF detection was 162 days and from cardiac monitoring to AF detection was 49 days.

Conclusion: Our study highlighted the potential link between certain comorbidities and demonstrated the delay in AF detection following CS. The significant parameters identified should be considered as new recommendations to the guideline for extended cardiac monitoring in CS patients. It is time to change the level of surveillance in this population.