Approximately one third of the population has a patent foramen ovale (PFO); about the same frequency as having brown hair. Can a normal finding be important in the cause of stroke, and should it change management? In particular should we close a normal hole with an artificial device?

The incidence of PFO in patients with cerebral infarction is about 20–30%, which is similar to that in control subjects.1 ,2 This has been taken to imply that PFO is coincidental.2 However, the incidence of PFO is usually higher, about 50–60%,3 in cryptogenic cerebral infarcts (those associated with normal carotid ultrasonography and haematological analysis in patients with no atrial fibrillation or other clinical evidence of heart disease). Furthermore, PFOs in patients with cryptogenic stroke are larger than in patients with another potential cause for stroke4 and in control subjects.5 Finally, the recurrence rate of cryptogenic stroke is higher if a PFO is detected (14%v 7% per 100 patient years).6

These facts suggest a role for a PFO, but to establish a high likelihood of causation requires finding a source for emboli. Unless venous investigation is conducted immediately, it is difficult to differentiate primary from secondary deep …