Association Between Anatomic Features of Atrial Septal

Abnormalities Obtained by Omni-Plane Transesophageal

Echocardiography and Stroke Recurrence in Cryptogenic Stroke
Patients with Patent Foramen Ovale
Jong-Young Lee, MDa, Jae-Kwan Song, MDa,*, Jong-Min Song, MDa, Duk-Hyun Kang, MDa,
Sung-Cheol Yun, PhDb, Dong-Wha Kang, MDc, Sun Uck Kwon, MDc, and Jong Sung Kim, MDc
The association between the anatomic characteristics obtained by omni-plane transesoph-
ageal echocardiography (TEE) and stroke recurrence in patients with cryptogenic stroke
and patent foramen ovale (PFO) remains unclear. In the present longitudinal follow-up
study, we sought to investigate whether PFO findings assessed by TEE can predict stroke
recurrence. Of the 1,014 consecutive patients with acute ischemic stroke referred for TEE,
184 (mean ⴞ SD age, 51 ⴞ 14 years) were classified as having cryptogenic stroke with PFO,
and follow-up data were available for 181 patients. During follow-up (median 3.5 years), 14
patients (7.7%) experienced stroke recurrence. Multivariate analysis showed that atrial
septal aneurysm or hypermobility of the atrial septum (hazard ratio 6.04, 95% confidence
interval 1.84 to 19.86, p ⴝ 0.003) and PFO size (hazard ratio 3.00, 95% confidence interval
1.96 to 4.60, p <0.0001) were independent predictors of stroke recurrence. The optimal
cutoff value of PFO to predict stroke recurrence within 3 years was 3.0 mm (95% confi-
dence interval 2.1 to 3.7 mm, area under the curve 0.889, p <0.001) with a sensitivity and
specificity of 90.0% and 79.4%, respectively. Using this cutoff, the 3-year stroke recurrence-
free survival rates differed significantly (98.9 ⴞ 1.1% vs 71.5 ⴞ 16.2%, p <0.001). In
conclusion, our data suggest that risk stratification might be possible using the findings
from TEE. The prophylactic benefit of PFO closure from these findings needs additional
investigation. © 2010 Elsevier Inc. All rights reserved. (Am J Cardiol 2010;106:129 –134)

The appropriate treatment strategy for secondary stroke
prevention in patients with cryptogenic stroke and patent
foramen ovale (PFO) remains challenging. The clinical and
anatomic variables reported to be risk factors associated
with stroke recurrence include older age,1 large PFO,2–5
large right-to-left shunting,6 and combined atrial septal an-
eurysm (ASA).7,8 However, these factors were not con-
firmed by other studies.6,8 –10 Most clinical observations
have been from the results of cross-sectional and case-
control studies, which have flaws, including poorly defined
control groups,2,3,5 case selection according to age,5,8 and
heterogeneous imaging techniques, including transthoracic9
and old mono- and biplane transesophageal echocardiogra-
phy (TEE).2,3,8,10 To overcome these limitations, the present
longitudinal follow-up study of patients with cryptogenic
stroke and PFO explored the relation between the anatomic
features of atrial septal abnormalities, evaluated by the stan-
dard omni-plane TEE, and stroke recurrence, regardless of
patient age. The aim of the present study was to determine
whether risk stratification to predict stroke recurrence was
Divisions of aCardiology and bBiostatistics, and cDepartment of Neu-
rology, Asan Medical Center, University of Ulsan College of Medicine,
Seoul, South Korea. Manuscript received December 15, 2009; manuscript
received and accepted February 4, 2010.
*Corresponding author: Tel: (82) 2-3010-3150; fax: (82) 2-486-5918.
E-mail address: jksong@amc.seoul.kr (J.-K. Song).
possible using the findings from TEE, which could help to
establish guidelines for secondary prevention.
Methods
All patients with acute ischemic stroke referred to the
Stroke Center, Asan Medical Center from January 2000 to
April 2007 were screened for enrollment. A standardized
protocol to determine the definite causes of ischemic stroke
has been used at our institution since 1995. For the diagno-
sis of cryptogenic stroke, patients who were considered to
have definite causes of stroke were excluded11: (1) large-
artery atherosclerosis (defined as stenosis ⱖ50% or occlu-
sion of the corresponding artery); (2) lacunar stroke (defined
as a small, deep infarct ⱕ20 mm in diameter without arterial
or cardiac sources of embolism; (3) cardioembolic causes,
such as atrial fibrillation, recent (within 4 months) myocar-
dial infarction, dilated cardiomyopathy, rheumatic mitral
stenosis, mitral or aortic vegetations or prostheses, left atrial
or left ventricular thrombus or tumor, akinetic left ventric-
ular segment, spontaneous echocardiographic contrast of
the left atrium, and complex atheroma of the aortic arch; and
(4) other definite causes of stroke, such as nonatheroscle-
rosis arteriopathies (e.g., dissection), coagulopathies, and
hematologic or systemic disorders (e.g., the antiphospho-
lipid-antibody syndrome).
A strictly predefined protocol, performed by experienced
echocardiographic specialists (JKS, DHK, and JMS), was
used to assess for PFO and ASA at rest and during the

0002-9149/10/$ – see front matter © 2010 Elsevier Inc. All rights reserved. www.ajconline.org
doi:10.1016/j.amjcard.2010.02.025
130 The American Journal of Cardiology (www.ajconline.org)

Figure 1. Representative transesophageal echocardiographic images for quantification of PFO size and shunt. At baseline (A,B), color Doppler mapping
visualized left-to-right shunt through the PFO, and the PFO size was 1.2 mm. With M-mode tracing during the Valsalva maneuver, the excursion of the atrial
septum could be easily measured (C). Using hand-agitated saline, contrast echocardiography with the Valsalva maneuver was performed for PFO size
measurement and shunt quantification. At rest and during the Valsalva maneuver (D), the interatrial septum maintained the convexity toward the right atrium,
which moved toward the left atrium (arrow) immediately after release of the maneuver, making a larger PFO opening (arrowhead) and right-to-left shunt
(E). PFO size was measured at the frame showing the maximum opening (F); 3.7 mm in this case. Ao ⫽ aorta; LA ⫽ left atrium; RA ⫽ right atrium.

Valsalva maneuver using 5-MHz omni-plane transesopha-
geal echocardiographic transducers and hand-agitated sa-
line. After careful observation of the interatrial septum using
2-dimensional echocardiography, color Doppler mapping and
M-mode tracing of the septum with the Valsalva maneuver
(Figure 1A–C), ⱖ3 intravenous contrast injections (5 ml of
hand-agitated saline) were administered through the right
antecubital vein at rest and with the Valsalva maneuver,
followed by deep inspiration (Figure 1D–F). PFO was de-
fined as passage of contrast (ⱖ3 microbubbles) from the
right to left atrium through the gap, within 3 cardiac cycles
after complete opacification of the right atrium. All diag-
nostic discrepancies were resolved by a second reviewer
unaware of the patient details.
Digitally, stored transesophageal echocardiographic im-
ages were reviewed and analyzed by an investigator (JYL)
who was unaware of the clinical outcomes. Using calipers,
the PFO size was measured as the maximum separation of
the septum primum from the septum secundum (Figure 1F).
The maximum number of the microbubbles in the left
atrium within 3 cardiac cycles after complete opacification
of the right atrium was also calculated. The PFO size and
shunt quantification were measured ⱖ3 times, and the mean
value of each was calculated. We classified grade 1 as
 Miscellaneous/PFO and Stroke Recurrence 131

Table 1
Baseline demographic and transesophageal echocardiographic findings of patients with cryptogenic stroke and patent foramen ovale (PFO) according to
treatment strategy
Variable Antiplatelet Group Anticoagulation Group Surgical Closure Group p Value
(n ⫽ 99) (n ⫽ 60) (n ⫽ 22)

Demographic data
Age (years) 53 ⫾ 13 53 ⫾ 13 41 ⫾ 12*† 0.001
Men 69 (70%) 48 (80%) 15 (63%) 0.34
Hypertension 54 (55%) 28 (47%) 2 (9%)*† 0.001
Diabetes mellitus 17 (17%) 10 (17%) 0 (0%)* 0.03
Cholesterol ⬎240 mg/dl 25 (25%) 14 (24%) 6 (27%) 0.85
Smoker 35 (35%) 23 (38%) 8 (36%) 0.87
Family history of stroke 5 (5%) 2 (3%) 10 (5%) 0.85
Echocardiographic data
Left to right shunt 27 (27%) 24 (40%) 16 (69%)*† ⬍0.001
Right-to-left shunt at rest 8 (8%) 4 (7%) 1 (5%) 0.13
Shunt grade 3 14 (14%) 8 (13%) 11 (50%)*† ⬍0.001
Atrial septal aneurysm or hypermobility 11 (11%) 7 (12%) 1 (5%) 0.65
Patent foramen ovale size (mm) 2.0 ⫾ 1.2 1.9 ⫾ 1.1 2.4 ⫾ 1.2 0.10

Data are presented as mean ⫾ SD or numbers (%).

* p ⬍0.05 for antiplatelet versus closure; † p ⬍0.05 for anticoagulation versus closure.

Figure 2. Recurrent stroke-free survival curves according to treatment option in patients with cryptogenic stroke and patent foramen ovale.

minimal (ⱕ5 bubbles), grade 2 as moderate (6 to 20 bub-
bles), and grade 3 as severe (⬎20 bubbles).12 ASA or
hypermobility was defined as ⱖ10 mm of phasic septal
excursion either into the atrium or a sum total excursion of
⬎15 mm during the cardiorespiratory cycle, with a base of
ⱖ15 mm.13 The maximum length of the eustachian valve
was also measured in the digital images.
Except for 3 patients who refused additional procedures
or medication, the treatment strategy for secondary preven-
tion in patients with cryptogenic stroke and PFO was de-
termined at the discretion of the attending stroke neurologist
and was determined from the stroke pattern, patient age,
co-morbidities, and cardiovascular risk factors. Patients as-
signed to medical treatment were treated with either a vita-
min K antagonist or antiplatelet therapy. Warfarin was ad-
justed to a target international normalized ratio of 2.0 to 3.0.
In some patients, PFO closure was performed percutane-
ously with the patient under general anesthesia and with
transesophageal echocardiographic guidance.
The clinical data from patients with cryptogenic stroke and
PFO were regularly recorded by neurologists during the regu-
lar medical visits. Patients who failed to make a regular visit
were interviewed by telephone to determine whether they had
experienced recurrent ischemic stroke, the primary end point
of the present study. Recurrence of ischemic stroke was con-
firmed by a new lesion on computed tomography or magnetic
resonance imaging. The clinical follow-up data obtained thor-
ough the end of August 2008 were used in the final analysis.
132 The American Journal of Cardiology (www.ajconline.org)

The institutional review board of Asan Medical Center ap- Table 2

proved the present retrospective study. Baseline demographic and transesophageal echocardiographic
The primary end point of the study was stroke recur- characteristics in patients with cryptogenic stroke and patent foramen
ovale (PFO) stratified by recurrent stroke during medical treatment
rence. The data are presented as the mean or median ⫾ SD
or median and range, as appropriate. Between group com- Variable Recurrence p Value
parisons of the baseline and transesophageal echocardio- No Yes
graphic variables were done using Student’s t test, the Wilcoxon (n ⫽ 145) (n ⫽ 14)
rank sum test, analysis of variance, or the Kruskal-Wallis
test for continuous variables and the chi-square test or Demographic data
Fisher’s exact test for categorical variables, as appropriate. Age (years) 52 ⫾ 13 57 ⫾ 15 0.24
Men 108 (75%) 9 (64%) 0.52
Unadjusted cumulative event rates were estimated using the
Hypertension 72 (50%) 10 (71%) 0.12
Kaplan-Meier method and were compared using the log- Diabetes mellitus 27 (19%) 0 (0%) 0.13
rank test. Independent predictors of recurrence were ana- Smoking 55 (38%) 3 (21%) 0.22
lyzed using the Cox proportional hazard model. For recur- Family history of stroke 7 (5%) 1 (7%) 0.52
rence, owing to the infrequent occurrence of events, the Echocardiographic data
predictive value of univariate findings was subsequently Left-to-right shunt 48 (33%) 3 (21%) 0.55
tested with a bootstrap resampling procedure in which mul- Right-to-left shunt 11 (8%) 1 (7%) 0.34
tivariate Cox regression analysis with a backward elimination Shunt grade 3 15 (11%) 6 (43%) 0.010
process was repeated for 1,000 bootstrap resampling.14 The Atrial septal aneurysm or 13 (9%) 5 (36%) 0.011
relative frequency of selection of the bootstrap resampling hypermobility
Patent foramen ovale size (mm) 1.8 ⫾ 1.0 3.9 ⫾ 1.0 ⬍0.001
⬎50% was used as the criterion for inclusion of predictors
Length of eustachian valve (mm) 13.0 ⫾ 5.7 13.6 ⫾ 7.2 0.908
in the final multivariate model. The proportional hazards
assumption was confirmed by testing of partial (Schoenfeld)
residuals, and no relevant violations were found. To assess
the cutoff value of PFO size for predicting recurrence within no significant change was found in PFO size according to age
3 years, the receiver operating characteristic curve was (p ⫽ 0.586). However, women had larger PFOs than men (2.3 ⫾
used. We excluded censoring data within 3 years, and re- 1.4 vs 1.9 ⫾ 1.1 mm, p ⫽ 0.041).
currence after 3 years was censored. The optimal cutoff Of the 184 patients, 22 underwent percutaneous closure
value was defined as the value with the maximum sum of of the PFO using an Amplatzer device (18 mm in 10 and 25
sensitivity and specificity. The reported 95% confidence mm in 11 patients) or a CardioSEAL (17 mm in 1 patient;
intervals for the cutoff value, area under the curve, sensi- NMT Medical, Boston, Massachusetts). TEE immediately
tivity, and specificity were obtained using the bootstrap with after the procedure confirmed successful closure in all 22
percentile method (220 replicate). patients, and no serious complications occurred. Follow-up
The intra- and interobserver variability of PFO size mea- echocardiography (TEE and transthoracic echocardiography
surement were evaluated in 50 randomly selected patients. in 5 and 14 patients, respectively) was performed in 19
The correlation coefficient of the 2 measurements and the patients within 2 months after the procedure. Only 1 patient
coefficient of variation, defined as the SD of the measured (4.5%) showed a minimal grade 1 residual right to left shunt
difference as a percentage of the average of the 2 series, with contrast echocardiography using hand-agitated saline.
were calculated. The intraobserver correlation coefficient No patient had a thrombus on the device. During clinical
and coefficient of variation was 0.960 (p ⬍0.001) and 9.9%, follow-up of 2.9 ⫾ 4.0 years (median 2.5), no recurrence
respectively, and the interobserver correlation coefficient developed of any embolic stroke.
and the coefficient of variation was 0.922 (p ⬍0.001) and Of the 99 patients who received antiplatelet medication
18.8%, respectively. (antiplatelet group), acetylsalicylic acid (100 mg/day) was
All p values are 2-sided, and p ⬍0.05 was considered the main regimen for 78 patients, either alone (n ⫽ 14) or
statistically significant. Statistical Analysis System soft- combined with other antiplatelet drugs, such as clopidogrel
ware, version 9.1 (SAS Institute, Cary, North Carolina) and (75 mg/day, n ⫽ 49) and cilostazol (100 mg twice daily,
R programming language with DiagnosisMed library were n ⫽ 11). The remaining patients received monotherapy with
used for statistical analysis. clopidogrel (n ⫽ 10) or cilostazol (n ⫽ 6) or a combination
of these 2 medications. Of the 60 patients who received
Results warfarin (anticoagulation group), 27 continued taking this
drug, and 33 changed to antiplatelet drugs at 6 to 12 months
During the study period, 4,543 patients were admitted be- after the initial ischemic stroke. No significant major bleed-
cause of acute (⬍7 days after onset) ischemic stroke. Of these, ing episodes causing interruption of therapy occurred but
1,014 patients (22.3%) were referred for echocardiography to minor bleeding episodes (e.g., epistaxis, bruising, ecchymo-
determine the potential cardiac source of embolism. PFOs sis, and petechiae) were observed in 5 patients (8.3%) in the
were identified in 229 patients (22.6%); 45 patients had coex- anticoagulation group. Table 1 lists the baseline and echo-
isting high-risk cardioembolic sources of stroke. Thus, the cardiographic variables according to treatment group in the
remaining 184 were classified as having cryptogenic stroke patients with cryptogenic stroke and PFO.
and PFO and were the subjects of the present study. Their Three patients refused either PFO closure or medical
mean ⫾ SD age was 51 ⫾ 14 years, and 135 (73.4%) were treatment, and the clinical follow-up data of the remaining
men. The mean ⫾ SD PFO diameter was 2.0 ⫾ 1.2 mm, and 181 patients were used for the evaluation of stroke recur-
 Miscellaneous/PFO and Stroke Recurrence
Figure 3. Receiver operating characteristic curve testing diagnostic value of PFO size in predicting recurrent stroke within 3 years (Left) and recurrent
stroke-free survival curves according to PFO size in patients with cryptogenic stroke and PFO (Right).
rence. During follow-up (median 3.5 years), recurrent isch-
emic stroke developed in 14 patients (7.7%) and was con-
firmed as repeat cryptogenic stroke after clinical evaluation.
Although the unadjusted 5-year stroke recurrence-free sur-
vival rates were different according to the treatment options
(Figure 2), on multivariate analysis, it was not an indepen-
dent predictor. Table 2 lists the summary of the baseline and
transesophageal echocardiographic variables in patients who
did and did not develop recurrent stroke during medical
treatment.
On multivariate Cox proportional hazard model analysis,
ASA or hypermobility of the atrial septum (hazard ratio
6.04, 95% confidence interval 1.84 to 19.86, p ⫽ 0.003) and
PFO size (hazard ratio 3.00, 95% confidence interval 1.96 to
4.60, p ⬍0.0001) were identified as independent predictors
of stroke recurrence. Using receiver operating characteristic
curve analysis, the best cutoff value of PFO size was 3.0
mm (95% confidence interval 2.1 to 3.7) and the area under
the curve was 0.889 (95% confidence interval 0.801 to
0.955, p ⬍0.001), with a sensitivity and specificity of 90.0%
(95% confidence interval 83.3% to 100%) and 79.4% (95%
confidence interval 62.6% to 90.6%), respectively. Using
this cutoff value, the 3-year stroke recurrence-free survival
rate differed significantly (98.9 ⫾ 1.1% vs 71.5 ⫾ 16.2%,
p ⬍0.001, Figure 3).
Discussion
Appropriate definition of high-risk patients requires lon-
gitudinal follow-up data from patients with cryptogenic
stroke and PFO, who have been diagnosed and measured
using a standardized protocol. To date, data are available
from 4 longitudinal follow-up studies using both bi- and
omni-plane TEE. However, these studies showed contradic-
tory conclusions on the effect of PFO size or anatomic
abnormalities of the interatrial septum. In a single-center
study, the PFO size was not a risk factor, but patients with
right-to-left shunt at rest or hypermobility of the atrial
septum had a significantly greater recurrence rate than those
without (12.5% vs 4.3%) during a median follow-up of 31
months.6 In patients aged ⱕ55 years, PFO alone was not
associated with stroke recurrence; however, patients with
both PFO and ASA were found to have a substantial risk.8
133
However, in another multicenter longitudinal study, patients
with both PFO and ASA were reported to have no increased
risk of adverse events with medical therapy.10 These inves-
tigators later confirmed that in patients aged ⱖ65 years,
PFO significantly increased the risk of adverse events but
failed to demonstrate a PFO size-dependent increase in
adverse events.1
In our study, PFO size was the most powerful contin-
uous variable associated with stroke recurrence. Using
receiver operating characteristic analysis, we calculated a
cutoff value for the PFO size. The most important dif-
ference between our study and previous longitudinal fol-
low-up studies was that we used a standardized protocol
and omni-plane TEE in all patients. Although omni-plane
TEE has not been shown superior to bi- or mono-plane
TEE in the quantification of PFO size and shunt, the only
case-control study demonstrating a positive relation be-
tween PFO size and embolic cerebrovascular events had
performed omni-plane TEE in all patients.5 PFO detec-
tion is highly dependent on operator technique, the in-
jection site of the contrast,15 and the imaging acquisition
method.16 Thus, failure to document a close association
between PFO size and stroke recurrence in previous stud-
ies might have resulted, at least in part, from the use of
outdated techniques (mono- or biplane TEE) for adequate
quantification of PFO size and shunt.
Several limitations of our study need to be explained.
The stroke recurrence rate was not high, which might have
affected our statistical analysis. However, our subjects were
enrolled from a large registry database that included ⬎1,000
patients with ischemic stroke, the largest database in a
single-center study. The prevalence of PFO in patients with
ischemic stroke referred for TEE and the stroke recurrence
rate (8.8% during a median follow-up of 3.6 years with
medication) were comparable to the results of previous
meta-analyses.17 Thus, we believe our results were not bi-
ased, and our subjects might be representative of patients
with cryptogenic stroke. In our study, patients receiving
anticoagulation had a greater recurrent stroke-free survival
rate than those receiving antiplatelet treatment alone. How-
ever, we believe that randomization of the medical therapy
would be necessary to clarify the risk status of antiplatelet
134 The American Journal of Cardiology (www.ajconline.org)
therapy. This was confirmed in our multivariate analyses,
which showed a lack of association between the treatment
options and stroke recurrence.
Despite these limitations, we believe that risk stratifica-
tion to predict stroke recurrence was possible using the
findings from omni-plane TEE in patients with cryptogenic
stroke and PFO. PFO size was found to be the most impor-
tant risk factor associated with stroke recurrence and the
best cutoff value of PFO size was 3.0 mm. Large, random-
ized, controlled trials are necessary to determine the optimal
treatment strategy for cryptogenic stroke with PFO and to
test the prophylactic benefit of percutaneous PFO closure in
these selected patients.
1. Homma S, Di Tullio MR, Sacco RL, Sciacca RR, Mohr JP; PICSS
Investigators. Age as a determinant of adverse events in medically
treated cryptogenic stroke patients with patent foramen ovale. Stroke
2004;35:2145–2149.
2. Homma S, Di Tullio MR, Sacco RL, Mihalatos D, Li Mandri G, Mohr
JP. Characteristics of patent foramen ovale associated with crypto-
genic stroke: a biplane transesophageal echocardiographic study.
Stroke 1994;25:582–586.
3. Hausmann D, Mugge A, Daniel WG. Identification of patent foramen
ovale permitting paradoxic embolism. J Am Coll Cardiol 1995;26:
1030 –1038.
4. Stone DA, Godard J, Corretti MC, Kittner SJ, Sample C, Price TR,
Plotnick GD. Patent foramen ovale: association between the degree of
shunt by contrast transesophageal echocardiography and the risk of
future ischemic neurologic events. Am Heart J 1996;131:158 –161.
5. Schuchlenz HW, Weihs W, Horner S, Quehenberger F. The associa-
tion between the diameter of a patent foramen ovale and the risk of
embolic cerebrovascular events. Am J Med 2000;109:456 – 462.
6. De Castro S, Cartoni D, Fiorelli M, Rasura M, Anzini A, Zanette EM,
Beccia M, Colonnese C, Fedele F, Fieschi C, Pandian NG. Morpho-
logical and functional characteristics of patent foramen ovale and their
embolic implications. Stroke 2000;31:2407–2413.
7. Meissner I, Khandheria BK, Heit JA, Petty GW, Sheps SG, Schwartz
GL, Whisnant JP, Wiebers DO, Covalt JL, Petterson TM, Christianson
TJH, Agmon Y. Patent foramen ovale: innocent or guilty? Evidence
from a prospective population-based study. J Am Coll Cardiol
2006;47:440 – 445.
8. Mas JL, Arquizan C, Lamy C, Zuber M, Cabanes L, Derumeaux G,
Coste J; Patent Foramen Ovale and Atrial Septal Aneurysm Study
Group. Recurrent cerebrovascular events associated with patent fora-
men ovale, atrial septal aneurysm, or both. N Engl J Med 2001;345:
1740 –1746.
9. Di Tullio MR, Sacco RL, Sciacca RR, Jin Z, Homma S. Patent
foramen ovale and the risk of ischemic stroke in a multiethnic popu-
lation. J Am Coll Cardiol 2007;49:797– 802.
10. Homma S, Sacco RL, Di Tullio MR, Sciacca RR, Mohr JP. Atrial
anatomy in non-cardioembolic stroke patients: effect of medical ther-
apy. J Am Coll Cardiol 2003;42:1066 –1072.
11. Adams HP Jr, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon
DL, Marsh EE III. Classification of subtype of acute ischemic stroke:
definitions for use in a multicenter clinical trial. TOAST: Trial of Org
10172 in Acute Stroke Treatment. Stroke 1993;24:35– 41.
12. Di Tullio M, Sacco RL, Venketasubramanian N, Sherman D, Mohr JP,
Homma S. Comparison of diagnostic techniques for the detection of a
patent foramen ovale in stroke patients. Stroke 1993;24:1020 –1024.
13. Mugge A, Daniel WG, Angermann C, Spes C, Khandheria BK, Kro-
nzon I, Freedberg RS, Keren A, Dennig K, Engberding R, Sutherland
GR, Vered Z, Erbel R, Visser CA, Lindert O, Hausmann D, Wenzlaff
P. Atrial septal aneurysm in adult patients: a multicenter study using
transthoracic and transesophageal echocardiography. Circulation
1995;91:2785–2792.
14. Sauerbrei W, Schumacher MA. Bootstrap resampling procedure for
model building: application to the Cox regression model. Stat Med
1992;11:2093–2109.
15. Hamann GF, Schatzer-Klotz D, Frohlig G, Strittmatter M, Jost V, Berg
G, Stopp M, Schimrigk K, Schieffer H. Femoral injection of echo
contrast medium may increase the sensitivity of testing for a patent
foramen ovale. Neurology 1998;50:1423–1428.
16. Ha JW, Shin MS, Kang S, Pyun WB, Jang KJ, Byun KH, Rim SJ, Huh
J, Lee BI, Chung N. Enhanced detection of right-to-left shunt through
patent foramen ovale by transthoracic contrast echocardiography using
harmonic imaging. Am J Cardiol 2001;87:669 – 671.
17. Homma S, Sacco RL. Patent foramen ovale and stroke. Circulation
2005;112:1063–1072.