The patent foramen ovale (PFO), an orphan congenital heart defect without specialty parent or guardian, has been at the center of an escalating clinical/academic battle for two decades. Pitting cardiologist against neurologist, clinician against academic, regulation against medical practice, and neurological patients against insurers, the response to broad acceptance and application of percutaneous closure has challenged the academic/industry/regulatory system of evidence-based medicine in the United States.


Cardiologists and neurologists view PFO through very different clinical lenses. This marked difference is illustrated by a survey of CLOSURE 1 neurology and cardiology investigators.

PFO Closure: Neurology vs Cardiology

CVA RISK= consider PFO a risk for stroke. CLOSE = recommend closure. ASX CLOSE= recommend asymptomatic PFO closure. SCUBA= recommend closure in scuba divers. MIGRAINE= recommend PFO closure for migraine WARFARIN= recommend warfarin as medical management.(Percentages) Stroke 2005; 65:172-3.


1. “The Tower of Babel”. A common language of anatomy, function, and clinical importance does not exist for PFO. Clear clinical differentiation from ASD II is absent. Post mortem “probe patency” (25%) is confused with true “shunt patency” (5-7%). Multiple diagnostic modalities are used (TEE, TTE, single and power m-mode TCD). Nonsensical, un-validated shunt severity schemes persist. Separating innocuous from high risk anatomy/physiology is sporadic.

2. “The Trouble with Bubbles”. Much of the confusion and conflict derives from the diagnosis of PFO based upon the appearance of a single bubble (or trace bubbles) in the left atrium by ultrasound diagnosis. Four studies have figured prominently in arguments against PFO as a cause of stroke. The majority of patients have trivial shunt and many have no evidence of any right-to-left shunt as large defects were diagnosed in the absence of bubbles if “membrane separation” was present. Membrane separation, a gap seen between secundum and primum is common and normal.

“Negative” PFO/Stroke Association Studies Bubble # Small PFO Percent Small PFO Large PFO Criteria W or W/O Shunt
PICSS (CIRCULATION 2002; 105:2625) 1-9 Bubbles 59% 2mm separation
SPARC (JACC 2006; 47:1465) 1-9 Bubbles ? 1mm separation
SPARC (MAY CLIN PROC 2006; 49:620) 1-9 Bubbles ? 1mm separation
NOMAS (JACC 2007; 49:797) ≥ Bubbles ? ?

It should come as no surprise that paradoxical embolism cannot occur if there is no shunt and is very unlikely if shunt consists of a tiny number of bubbles passing right-to-left for a few seconds intermittently during the day. There is tremendous variation in shunt severity in the published literature and particularly in randomized trials—which include high percentages of low risk, low shunt patients. The figure below demonstrates these different patient populations for trials where shunt amount is reported compared with the Utah population (UPSG) where closure was restricted to truly severe shunt. The Spencer scale is logarithmic.

shunt severity populations

(From: Sorensen SG, Developing a Successful Integrated PFO Program, in Patent Foramen Ovale, Amin Z, ed)

3. The “HDE RULE”. The FDA released the Amplatzer PFO and NMT CardioSEAL occluders for PFO closure in 2001/2002 under a Humanitarian Device Exemption (HDE).


  • Device safety and function proven; but not efficacy
  • Limited to 4000 devices/year; assumed low patient need
  • Released as a marketable, commercial device similar to PMA
  • Required IRB approval and monitoring
  • IRB could modify clinical requirements
  • Required two strokes to implant:
    • Identifying first stroke then treated with warfarin
    • Second stroke, failing therapeutic INR met closure criteria
  • Stroke not defined. TIA not included

Specifically stating no safety concerns, the FDA withdrew the PFO HDE in 2006 based upon use reaching the arbitrary 4000. Despite HDE discontinuance, the HDE Rule has defined U.S. clinical research despite ethical and clinical problems.

a. Weird Science:
As stated by the FDA repeatedly, there is no proven best medical therapy for PFO stroke prevention. RESPECT included five different anti-platelet/anti-thrombotic regimens. Warfarin for non-atrial fibrillation stroke prevention is unsafe and ineffective. DSMB (safety boards) stopped SPIRIT and WASID warfarin stroke trials. WARSS, a randomized aspirin vs warfarin trial including cryptogenic stroke, showed a dismal two year combined stroke/death/bleed end-point of 18%. The HDE Rule therefore requires giving an unsafe, ineffective treatment and then requiring failure of this failed medical therapy to permit PFO closure; all after two strokes—each with a risk of disability of 50%.

b. “The Genie Is Out of the Bottle”
Percutaneous closure was rapidly accepted impacting recruitment for PFO RCT. An irreversible tipping point had been reached due to: 1) limitations and pitfalls of life-long medical therapy, 2) advantages of percutaneous closure 3) the permanence and disability of stroke 4) biologic plausibility of PFO closure 4) reluctance to participate in RCT in general and specifically where disabling stroke was a possible outcome.

c. Regulation, Medical Ethics, and Patient Care
While the FDA approves, labels, and regulates the medical device industry, by law it does not and cannot control or regulate medical practice. Congress has specifically addressed this issue in Section 214 of the 1997 FDAMA (Modernization Act) by explicitly prohibiting any FDA intrusion into medical practice (the “Medical Practice Provision”). Within this context, The HDE Rule for PFO closure is problematic: the FDA directs “medical care” which is monitored by an IRB (a scientific body) which manages a commercially marketed device which is used by a physician who desires stroke prevention for the patient. This clear conflict with the FDAMA is the reason that IRBs could alter FDA requirements and off-label use could not be prevented.

The blind acceptance of the HDE Rule as “gospel truth” in trial design and practice clinical guidelines remains a major source of patient and provider distrust.

Battles in the PFO War

Randomized Clinical Trials

The rise of RCT as the gold standard of evidence-based medicine has greatly impacted medical research and improved health care quality. Originally designed to reduce bias in research, provide physicians reliable treatment modalities, and permit the development of safe and efficacious treatments, RCT have limitations: cost, delay, narrow selection, industry support, conflict of interest, and inability to extrapolate to daily practice.

(Bothwell LE, et al, Assessing the Gold Standard—Lessons from the History of RCTs N Engl J Med 2016; 374:2175-2181)

PFO/Stroke Trials

The CardioSEAL/STARflex Device- CLOSURE 1

The Clamshell/CardioSEAL/STARflex device has been shown to be inferior the Amplatzer PFO and HELEX occluders in two device comparison RCT (Taaffe M, AJC 2008, Hornung M Euro Heart J 2013). The recent AAN practice advisory recognizes the device differences. Use of this device was abandoned over a decade ago in Utah PFO programs and in many European centers. In 2007, slow enrollment and sponsor financial losses prompted CLOSURE 1 investigators to convince the FDA to accept a study participant number reduction from 1600 to 800 based upon a meta-analysis published 4 years earlier (Khairy P, Ann Int Med 2003). The study was statistically underpowered and failed.


  • Lack of therapeutic gold standard; no proof of any medical Rx benefit
  • Enrollment phase selection bias
  • Exclusion of patients with high risk of PDE: DVT, ASA
  • 50% of patients with trivial shunt
  • Exclusion of high risk patients by enrolling physicians
  • Study under-powered to detect clinically significant differences
  • Short duration of follow-up
  • Inferior device: residual shunt, thrombus, atrial fibrillation
  • Operatory/Center inexperience; lack of proficiency
  • Recurrences non-cryptogenic; Doubtful stroke selection effectiveness

Amplatzer PFO Occluder RCT: RESPECT and the PC TRIAL

The Amplatzer™PFO Occluder received CE mark approval in Europe in 1998 and is available in over 80 countries world-wide. It was available in the U.S. from 2001-2006.

The PC TRIAL randomized 204 device and 210 medically treated patients showed a non-significant 36% reduction for death/non-fatal CVA/TIA/PDE and a non-significant 86% reduction in non-fatal stroke (0.5% vs 2.4%).

RESPECT is reviewed in detail on the “News” section on this website including 5 year, 10 year, and 13 year follow-up results. The Amplatzer™PFO Occluder was PMA approved by the FDA October 2016.


Two new randomized clinical trials of percutaneous patent foramen ovale (PFO) device closure have shown superiority of device closure over medical management for the prevention of recurrent stroke. The results of these trials were reported and the European Stroke Organization Conference in Prague.


The REDUCE trial used the Gore HELEX and Gore Cardioform Septal Occluder at 63 sites in 7 countries in the US and Europe. 664 patients were randomized 2:1 with device closure versus antiplatelet therapy in patients between 18 and 59 years of age. Both co-primary end-points by intention-to-treat analysis were met.

  1. 77% relative reduction in recurrent ischemic stroke. (p=0.001)
  2. 50% reduction in MRI silent brain infarcts** (p=0.024)
  3. Low complications rates. Atrial fibrillation occurred on 2.3%, but 80% were peri-procedural, had rapid resolution, and were not considered significant by research teams.

**MRI “silent” brain infarcts are associated with significant increase risk of dementia, functional and cognitive impairment, death, and recurrent clinical ischemic stroke. Silent brain infarcts (MRI T2/FLAIR) are now recognized within the new definition of stroke (AHA/ASA Expert Consensus Document; An Updated Definition of Stroke for the 21st Century, Stroke 2013; 44:2064-2089


The CLOSE trial, performed at 32 sites in France and 2 in Germany, randomized 663 patients using three different PFO closure devices.

  1. 97% relative reduction of recurrent stroke by PFO closure over medical management
  2. Low complication rates and atrial fibrillation similar to REDUCE.

These two trials and thirteen year follow-up data from RESPECT should change practice patterns and Guidelines. “REDUCE, in particular, provides solid support for practice change, because it included MRI-based assessments of silent infarcts as part of one of its primary end-points. That represents strong, objective evidence that closure is reducing events, and that’s not really subject to bias.” (Scott Kasner MD, REDUCE Neurology PI)


The failed, delayed, controversial, and completed randomized trials of PFO closure for the prevention/reduction of migraine are reviewed in detail on the Migraine and Shunt page of this website.

Clinical Practice Guidelines

Clinical Practice Guidelines as defined by the Institute of Medicine, HHS, AHRQ, and the National Guidelines Clearinghouse (NGC):

“Clinical practice guidelines are statements that include recommendations intended to optimize patient care that are informed by a systematic review of evidence and the assessment of the benefits and harms of alternative options (IOM definition)

These guidelines are not fixed protocols that must be followed, but are intended for health care professionals and providers to consider. While they identify and describe generally recommended courses of intervention, they are not presented as a substitute for the advice of a physician or other knowledgeable health care professional or provider. Individual patients may require different treatments from those specified in a given guideline. Guidelines are not entirely inclusive or exclusive of all methods of reasonable care that can obtain/produce the same results. While guidelines can be written that take into account the variations in clinical settings, resources, or common patient characteristics, they cannot address the unique needs of each patient nor the combination of resources available to a particular community or health care professional or provider. Deviations from clinical practice guidelines may be justified by individual circumstances. Thus, guidelines must be applied based on individual patient needs using professional judgement.” (U.S. Department of Health and Human Resources/ AHRQ/NGC/Disclaimer/definition and selection of evidence-based guidelines).


In the real world of clinical medicine, the adherence to CPG by physicians remains low: in the US: 55% overall and 55% for preventative care, acute care, and chronic conditions (McGlyn), psychiatry-27% (Bauer), primary care treatment CPG-HT 43%, lipid management-40%. DM-25%, CHF-25% (Nichols) (Institute of Medicine, Guidelines We Can Trust, 2011 National Academy Press, Washington DC). PFO/Stroke/ AHA-ASA 2006 Guideline adherence in 10 published studies: 35% mean, lowest 10% (Khairy meta analysis). Clearly, CPG do not represent the standard-of-care. The Institute of Medicine, at the request of the U.S. Congress has developed rigorous, standards for trustworthy guidelines. However, IOM standards remain largely non-utilized after two decades. Guideline controversies have been widely reported in the press. Perhaps, the longest and most contentious area has been in stroke management:

  1. “Why we can’t trust clinical guidelines: Lenzer J, BMJ 2013; 346:f3830. Lenzer has raised issue of financial conflict, panel selection, and industry relationships to the AHA/ASA endorsement of thrombolytic therapy of stroke with altaplase. (AAN reply BMJ 2013; 347:f5324)
  1. The American College of Emergency Physicians (ACEP) and American Academy of Neurology (AAN) reached a consensus on tPA management guidelines of acute stroke in 2013 after 8 years of disagreement and conflict. ACEP argued that the data did not support AAN/ASA guidelines which placed emergency physicians in an ethical and legal/malpractice dilemma. The 2013 document was immediately rejected and a new ACEP guideline (sans AAN) was published in 2015 (Ann Emerg Med 2015; 66:322-33).


  1. AAN Practice Parameter: Recurrent Stroke with Patent Foramen Ovale and Atrial Septal Aneurysm. Neurology 2004; 62:1042-1050.
    The AAN reviewed 129 studies rejecting 125. The conclusions were:

    1. PFO is not associated with increased risk of stroke or death
    2. PFO with ASA possibly increase risk of stroke but not death
    3. Insufficient data for aspirin or warfarin
    4. Insufficient evidence for PFO closure (but Khairy P meta-analysis excluded)

These conclusions differ from those 4 of 129 studies found “acceptable”:

PICCS: “Paradoxical embolism through a PFO is undoubtedly a cause of stroke…it becomes important to identify those at high risk for recurrent event.(Circ 2002; 105:2625)
MAS. PFO and ASA Study Group: “Patients with both PFO and ASA who had a stroke constitute a subgroup at substantial risk for recurrent stroke, and prevention strategies other than aspirin should be considered”. (NEJM 2001; 345:1740)
LA SPIENZA: “The association of right-to-left shunt at rest and high membrane mobility seem to identify PFO patients with cerebrovascular ischemic events who are at high risk for recurrent brain embolism.” (Stroke 2000; 31: 2407)
LAUSANNE STUDY: “…the first brain event in patients with PFO may often be devastating, as one-half of the patients suffered a severe initial stroke.. The demonstration of factors associated with an increased risk of recurrence in our study and other studies suggests that high risk patients with PFO exist.”(J Neurol 1996; 46:1301)

  1. AHA/ASA Secondary Stroke Prevention Guidelines 2006 and 2011: Stroke 2006; 37:577-617; Stroke 2011; 42:227-272.
    Insufficient information to recommend primary closure
    PFO closure recommendation Class IIb (may be considered)
  2. AHA/ASA Secondary Stroke Prevention Guidelines 2014: Stroke 2014; 45:2160-236.
    PFO closure changed from Class IIb to Class III (of no benefit or harmful)
    IVC filters (??) for DVT/ PFO patients were added; Class IIa.A public criticism and published call to revise the 2014 AHA/ASA Stroke Guidelines followed (Thaler, TCT 2014; Saver et al, Stroke 2015; 46:e85).

    “….it is an extreme violation of the norms of evidence-based medicine to conclude that there is definitely no benefit of an intervention when the best estimate of treatment effect available from randomized controlled trials actually suggests benefit.”

    “I submit that the recommendation of a class III evidence needs to be changed. The (AHA/ASA) committee erred in interpreting negative trials. I think they should have supported at least class IIb recommendation maybe even class IIa, but certainly not class III.”

    Others, noting 2008 European Stroke Organization clinical practice guidelines recommending primary stroke prevention PFO closure in high risk patients, have promoted the revision of all guidelines to include consideration of primary closure in high risk patients with careful registry follow-up in a 2016 editorial:

    Suggested guideline:

    “PFO closure for primary prevention can be considered in patients at high risk for paradoxical embolism due to tendency for venous thrombosis, vocational or recreational activities fostering right-to-left shunts, or aggravating PFO attributes (ASA. EV, CN, etc.), or those who can expect a collateral benefit (e.g. patients suffering from migraine).”

    The editorial concludes:

    “Stroke is among the most devastating diseases for patients and their families. It can cause loss of self-determination and independence. In a society where life expectancy of the general population is steadily increasing, it is the role of physicians to take every effort not only to prolong life, but also to preserve the quality of life.

    The role of cardiology is crucial to achieve this goal. Cardiology has to offer many therapies that prolong life and increase quality of life for patients. Patent foramen ovale closure is one of them. It is a simple, safe, and effective treatment that can make a difference for patients. It is a once-in-a-lifetime intervention and has therefore, been referred to as a mechanical vaccination against some forms of myocardial infarction, stroke and systemic embolism.

    The current evidence does not seem to be sufficiently reflected in the European and American guidelines and a more pro-active role for PFO closure for secondary prevention and for high-risk patients appears justified….”

    Nietlispach F, Meier B, Percutaneous closure of patent foramen ovale: An underutilized procedure. European Heart J 2016; 37:2023-8

  3. American Academy of Neurology. Practice Advisory: Recurrent Stroke with Patent Foramen Ovale. (Update of2004 Practice Parameter). Neurology 2016; 87; 1-7    

    Published in July 2016, this advisory came immediately after the presentation of long-term results of the RESPECT Trial (November 2015) and the FDA Circulatory Executive Review (May 2016) and immediately before FDA approval of the Amplatzer™PFO occlude.  


    a. Clinicians must counsel patients considering PFO closure:

    -It is impossible to determine whether PFO caused their stroke
    -The effectiveness of the procedure is uncertain
    -The procedure is associated with potentially serious complications

    b. Clinicians should not routinely offer percutaneous PFO closure to patients outside of a research setting

    c. In rare circumstances, such as multiple strokes despite adequate medical therapy , clinicians may offer the Amplatzer PFO occluder

    d. Disclaimer: “This information: Does not mandate any particular course of medical care; is not intended to substitute for independent professional judgement as the information does not account for individual variation among patients. The AAN assumes no responsibility for any injury or damage to persons arising out of or related to any use of this information.”

Responding to criticisms of not including recent RESPECT findings and the thinly veiled rush to publication, Dr. Gary Gronseth on behalf of the AAN Practice Advisory stated:

“Just because it’s approved, doesn’t mean it’s a good idea. It’s still an open question”

All scientific inquiry is, of course, “still an open question”. By its very nature-probabilistic and inherently uncertain- evidence based medicine can never achieve absolute certainty. Like Zeno’s Paradox of the Fletcher’s Arrow which cannot reach target (reductio ad absurdum), PFO closure cannot find acceptance by the neurological community through more scientific inquiry. These arguments are no longer scientific arguments, but a clever political tool used for over a century to defeat or delay regulatory or organizational action; an approach so common in regulatory debates that it has been given its own designation: Scientific Certainty Argumentation Method. This tactic is most familiar from tobacco industry defense and opposition to global warming arguments.

The results of RESPECT, REDUCE, AND CLOSE are as close to certainty as will ever be achievable. The striking confirmation by these studies of the PFO closure/stroke meta-analyses and PFO risk stratification of over a decade ago lead to a troubling conclusion: over one million preventable, disabling strokes for young patients; the casualties of the two decade PFO War.