Abstract 84: An Index to Identify Cryptogenic Stroke Patients whose Stroke is likely Attributable to Patent Foramen Ovale
Background: A patent foramen ovale (PFO) discovered in the setting of a cryptogenic stroke (CS) may be incidental or pathogenic. Based on Bayes theorem, the proportion of CS that is PFO-attributable among patients found to have a PFO has been shown to be related to PFO prevalence in CS versus control patients. However, the prevalence of PFO in CS patients is itself dependent on the presence or absence of other risk factors. We exploited this relationship to create an index to stratify CS patients with PFO by their likelihood that the CS is PFO-attributable.
Methods This project is part of the Risk of Paradoxical Embolism (RoPE) Study, an international, multicenter collaboration that has combined data for patients with CS and cryptogenic TIA who have known PFO status from 12 component studies (n=3665). For this study, we included those subjects within the 7 databases enrolling subjects both with and without PFO (n=3023). We used generalized linear mixed models to identify variables associated with the presence of a PFO, accounting for clustering within study. Based on this model, we created a simple index. Bayes theorem was used to estimate the PFO-attributable fraction in each stratum assuming a PFO prevalence in the general population of ∼25%.
Results: Variables negatively associated with the presence of a PFO included: age (odds ratio [OR] = 0.97 per 1 year increase, p <0.0001); diabetes (OR= 0.65, p < 0.001); hypertension (OR =0.68, p < 0.0001); smoking (OR = 0.70, p<0.60); prior stroke or TIA (OR = 0.78, p=0.04). Cortical stroke on neuroimaging (OR = 1.46, p < 0.001) was also associated with PFO. Based on this, a simple index was created in which the absence of each stroke risk factor was assigned a point, with age dichotomized at 50 years. PFO prevalence in each stratum is shown in the table for patients < age 60, i.e. the subset of patients likely to be considered for PFO closure trials.
Conclusion: Even among CS patients in the younger age range considered eligible for closure trials, there is considerable heterogeneity in the distribution of risk factors for stroke and other characteristics that identify subgroups of CS patients with variation in PFO prevalence. This reflects substantial and clinically important variation in the probability that a discovered PFO is likely to be pathogenic rather than incidental. This score may be useful in selecting patients for closure trials, or for stratification within trials, particularly if combined with a recurrence risk model.
- © 2012 by American Heart Association, Inc.