| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2007;83:1593-1602
© 2007 The Society of Thoracic Surgeons
a Department of Surgery, Columbia University College of Physicians and Surgeons, New York, New York
b Section of Cardiothoracic Surgery, Yale University School of Medicine, New Haven, Connecticut
c Department of Surgery, Landmark Medical Center/Beth Israel Deaconess Medical Center, Woonsocket, Rhode Island
d Stanford University Medical Center, Stanford, California
Accepted for publication December 11, 2006.
* Address correspondence to Dr Coady, Cardiac Surgery, Landmark Medical Center, 206 Cass Ave, Woonsocket, RI 02895 (Email: macoady{at}gmail.com).
Presented at the Forty-second Annual Meeting of The Society of Thoracic Surgeons, Chicago, IL, Jan 30–Feb 1, 2006.
| The Thoracic Surgery Directors Association (TSDA) Resident Research Award was established in 1990 to encourage resident research in cardiothoracic surgery. Abstracts submitted to The Society of Thoracic Surgeons (STS) Program Committee representing research performed by residents were forwarded to the TSDA to be considered for this award. The abstracts were selected by the TSDA Executive Committee consisting of Jeffrey Gold, MD, President; John Brown, MD, President-Elect; John Calhoon, MD, Secretary/Treasurer; Douglas Mathisen, MD, Immediate Past President; George Hicks, MD, Councillor-at-Large; Bartley Griffith, MD, Councillor-at-Large; and Leslie Kohman, MD, Councillor-at-Large. The TSDA Resident Research Award was given to Bryan A. Whitson, MD, a resident at the University of Minnesota Department of Surgery, Section of Thoracic Surgery, and Ryan R. Davies, MD, a resident of New York-Presbyterian Hospital (Columbia University Medical Center). They each received a monetary award of $1000.00 and an engraved desktop award. The TSDA makes this award annually. The resident author of the selected study is recognized at the STS meeting.
|
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Methods: Data on 195 patients with type A dissections followed up at a single institution between 1985 and 2005 were collected prospectively. Of these, 93 patients (47.2%) presented 48 hours or later after the initial onset of pain (group A), and the remaining 102 patients underwent immediate operative repair (group B). Median follow-up was 41.8 months (range, 0 to 386 months).
Results: Patients in group A were older (68.8 versus 59.3 years, p = 0.0005) and had a higher incidence of coronary artery disease (42.5% versus 14.6%, p < 0.0001), pulmonary disease (26.6% versus 8.4%, p = 0.0023), and congestive heart failure (14.1% versus 1.0%, p = 0.0004). Long-term survival was similar, although group B showed a trend toward improved 30-day mortality (16.5% versus 8.7%, p = 0.1035). Of the 92 patients in group A, 53 (57.6%) eventually underwent operative repair a median of 8.2 days after symptom onset. There was a trend toward improved long-term survival among patients undergoing repair (p = 0.1031).
Conclusions: Initial medical management with interval operative repair of selected patients referred greater than 2 days following an acute type A dissection is a viable option. Delayed repair after optimization of the clinical condition and detailed evaluation of concomitant diseases results in excellent long-term results.
Historic complication rates associated with type A aortic dissection have exceeded 1% to 2% per hour after the initial onset of symptoms [1, 2]. Survival at 48 hours has been estimated at 50%, and at 1 week declines even further to 30% in historical series [3]. With advances in critical care and perioprerative management, surgical outcomes continue to improve, but because of these mortality rates, emergency operative repair has become the nearly universal standard therapy for patients presenting with type A dissection. However, the mortality in these patients still ranges as high as 25%, and even the best series report operative mortality of approximately 10% [4, 5].
The disease course of dissection treated with modern medical therapy is not clear. Medical management, either as a bridge to surgery or as definitive treatment, typically consists of blood pressure control with morphine, ß-blockers, and vasodilators such as sodium nitroprusside or calcium-channel blockers [6]. Modern diagnostic, monitoring, and therapeutic technologies have advanced far beyond those available at the time of the early studies of type A dissection, but because the standard of care involves immediate operative intervention, the survival achievable with modern medical treatment has not been further addressed.
We have previously reported acceptable outcomes in a small series of late presenting patients with acute type A dissections managed temporarily, or permanently, without operative intervention [7]. At that time, we suggested that a subset of patients may have passed a critical phase in which the lethal complications of type A dissection typically occur, and this subset, may be managed medically or operated on semielectively with reasonable long-term survival. We therefore undertook the current study to further define this subset of patients and examine predictors of poor outcomes within this population.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Data were collected both prospectively and retrospectively from patient interviews and hospital chart review, and all data were entered into a computerized database (MS Access 2003, Microsoft Corp, Redmond, WA). Long-term follow-up was performed through office visits, telephone interviews, and office chart review. Data recovered from hospital records and computer files were cross-referenced with hospital discharge data monitored by the Connecticut Hospital Association and the Connecticut State Mortality Records as well as the Social Security Death Index (available at: http://ssdi.rootsweb.com). The patient database is maintained as part of the ongoing studies at the Yale Center for Thoracic Aortic Disease, a major referral center for southern New England. A total of 972 patient-years of follow-up after the type A dissection were available on these patients (median, 41.8 months; range, 0 to 386.8 months).
Patients were initially divided into two groups: group A consisted of 93 patients who received definitive operative management more than 48 hours after initial symptom onset or received solely medical management for their dissection, and group B consisted of 102 patients in whom immediate operative repair was performed. Group A was further subdivided into 53 patients who ultimately underwent operative repair and 40 who did not. Baseline demographic information for these patients is presented in Table 1. The primary outcome was mortality. Secondary end points among the patients undergoing operative repair included the incidence of postoperative complications, the time to extubation, the day of discharge from the intensive care unit, and length of stay.
|
2 test was used for comparisons of dichotomous risk factors with negative outcomes. The Mantel-Haenszel 2 test was used for comparisons taking into consideration disease severity (cardiac disease, pulmonary disease, renal disease, carotid disease, and tobacco use) [8] or stratifying continuous variables (aneurysm size or size index, age). The t test was used for analysis of continuous variables (p < 0.05). Where the distribution of the data was not normally distributed, in particular the measurement of times from symptom onset to presentation or operation, the nonparametric Wilcoxon rank sum test was used. The complete list of variables included in the analysis is given in the Appendix. After univariate analysis, variables were entered into the models using three selection criteria: forward, forward stepwise, and backward. All variables with a value of p < 0.40 in univariate analysis were included in the regression analyses, and the threshold for entry into the model was p < 0.20. Models were calculated using the LOGISTIC procedure and compared using the –2 Log L score and the Wald score for significance; in all cases the most predictive model is presented.
Product-limit estimates (Kaplan-Meier) for survival were calculated using the LIFETEST procedure with the log-rank test for difference between strata. The Cox regression model (using the PHREG procedure) was used to identify the most predictive variables. All variables with a value of p < 0.40 in univariate analysis were included in the regression analyses. Threshold for entry into the model for Cox regression was p < 0.20.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Patients With Delayed Presentation
Except for the increased prevalence of coronary artery disease (CAD) among patients undergoing eventual repair, the presence of comorbidities within group A did not vary between operative and medical management (Table 1). This remained true when disease severity was accounted for (data not shown).
Initial Presentation and Treatment
Chest pain represented the most common presenting symptom in all groups, and overall, presenting symptoms were similar between group A and group B (Table 2). Patients in group B did have a higher incidence of rupture (25.2% versus 1.1%, p < 0.0001) but a lower incidence of arch involvement (2.9% versus 14.1%) identified on diagnostic imaging. Median time to operation in group A was 199 hours versus 8.5 hours in group B (p < 0.0001).
|
A large proportion of patients undergoing eventual surgical repair initially presented to an outside hospital (60.4% versus 15.5% of those managed medically, p < 0.001), but the median time between presentation at the outside hospital and arrival at our institution was the same for the surgical (41 hours) and medical (39.5 hours, p = 0.9023) groups. Initial treatment within group A focused on standard medications for antiimpulse therapy and blood pressure control, including the use of ß-blockade (42.4%), sodium nitroprusside (24.0%), and calcium-channel blockade (12.0%).
Patients eventually undergoing surgery had a nonstatistically significant increase in the amount of time from symptom onset to arrival at our institution (60 hours versus 15 hours, p = 0.2976). The median time to operative repair from symptom onset was 199 hours and from admission was 114 hours.
Early Outcomes in All Patients
There was a nonsignificant decrease in early mortality (within 30 days of symptom onset) of 8.6% among patients in group A versus 16.8% for group B (odds ratio [OR], 0.470; 95% confidence interval [CI], 0.1928 to 1.1488). Overall, rupture was the only variable significantly predictive of increased 30-day mortality after diagnosis (OR, 4.750; 95% CI, 1.834 to 12.302). This was confirmed in multivariate analysis (OR, 4.953; 95% CI, 1.812 to 13.541).
Early Outcomes in Patients Presenting After A Delay
Patients undergoing surgical repair had a nonsignificant decrease in 30-day mortality compared with those who received medical therapy alone (5.7% versus 12.5%, p = 0.2442). Other predictors of increased early mortality included tobacco use (p = 0.0014) and pulmonary disease (p < 0.0001; Table 3).
|
Operative procedure
Among all patients who underwent surgery, replacement of the ascending aorta was the most common (92.3%), with a variable need for replacement of either the entire aortic root or some portion of the aortic arch. A concomitant coronary bypass procedure was required in 14 patients (9.0%), and 59 (37.8%) also had aortic valve or aortic root replacement. The need for these procedures did not vary significantly between groups.
Operative outcomes
Patients in groups A and B had similar postoperative outcomes, including days in the intensive care unit (7.7 versus 9.3 days), days until extubation (4.6 to 5.4 days), postoperative length of stay (13.9 versus 18.3 days), and overall hospital length of stay (17.9 versus 18.3 days).
None of the reasons (Table 4) for delayed management specifically predicted 30-day postoperative mortality (data not shown). Comorbid diseases, including CAD, congestive heart failure, and renal insufficiency, also failed to predict poor postoperative outcome. Significant predictors of postoperative (30-day) mortality among all patients included aortic rupture at operation (OR, 2.74; 95% CI, 1.05 to 7.14), the presence of paraplegia before repair (OR, 14.11; 95% CI, 1.22 to 163.14), and prior aortic surgery (OR, 5.4583; 95% CI, 1.12 to 26.39).
|
There was a trend toward improved survival to hospital discharge associated with patients in group A (30-day mortality, 9.4% versus 15.5%, p = 0.2904). When this group of patients was considered alone, there were nonstatistically significant decreases in early survival associated with redo operations (30-day mortality 20.0% versus 7.0%, p = 0.2044), aortic rupture (22.2% versus 6.8%, p = 0.1497), the need for delay because of the patient’s overall condition (18.2% versus 7.1%, p = 0.2648), and patients who initially refused operative intervention (25.0% versus 8.2%, p = 0.2680).
Overall postoperative (30-day) survival was 87.5%. Survival at 1, 5, and 10 years among all patients undergoing operation was 78.3%, 55.6%, and 46.1%, respectively. There was a trend toward improved survival at 1-year in patients undergoing delayed repair, although in the long-run, as shown in Figure 1, the survival curves are nearly identical (p = 0.3669).
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Our data confirm that many patients presenting with type A aortic dissection requiring immediate repair present in significant hemodynamic compromise with aortic rupture. Patients incurring rupture had a 30-day mortality rate four times as high as those without rupture; even in patients reaching operative repair, it remained a significant predictor of postoperative mortality. These results are consistent with prior studies in which signs and symptoms of hemodynamic compromise—including rupture, renal failure, intestinal malperfusion, and large volume blood transfusion—rather than comorbid disease have been the most powerful predictors of mortality [8–10].
In patients presenting to the hospital soon after symptom onset, immediate repair remains the only appropriate management. In those patients presenting more than 48 hours after symptom onset, however, surgery may be temporarily delayed to a semielective or urgent setting. Our early outcomes in a prior small series of patients were encouraging [7]. Here, we report results in a large subset of these highly selected, late-presenting patients in whom definitive surgical repair was delayed either temporarily or permanently.
Our original study showed that although acute type A aortic dissection is a lethal disease and generally requires immediate surgery, those patients who have survived several days after the onset of symptoms have survived the initial deadly period of the disease [7]. This suggested that this group of late presenters may undergo surgery semielectively after medical optimization or may receive aggressive medical therapy alone if they are not considered operative candidates.
These conclusions were based on the findings of similar morbidity and mortality between those initially treated with surgery and those with delayed treatment. Further study of this subset of the patient population was undertaken for three important reasons: (1) to define the reasons why late presenters are delayed in their surgical treatment and to examine their treatment course before surgery, (2) to compare the morbidity and mortality results of those who receive treatment within 48 hours and those who do not, and (3) to better define the natural history of type A aortic dissections in the current era.
Reasons for Delay
Many patients in our cohort were delayed in their treatment because of delays in their presentation to health professionals or because of delays intrinsic to the medical system: transfer between hospitals, further imaging to confirm diagnosis, interval invasive studies, and consultation from other specialists (Table 4). Overall, patients who eventually underwent surgery more than 48 hours after the initial onset of symptoms were more likely to have been admitted to an outside hospital (60.8% in group A versus 12.8% in group B). These two findings would indicate that delays in patient presentation and delays in diagnosis or transfer to an appropriate care facility both contribute to delays in treatment. Amongst patients for whom the reason for delay in surgery was known, a delay in diagnosis was more prevalent in those eventually undergoing surgical repair (51.9% in group A versus 25.0% in group B), indicating that delays in diagnosis brought about delays in surgical correction and that more expedient diagnosis would likely lead to more expedient surgery.
Although patients who survived this initial delay had excellent outcomes, the number of nonsurvivors cannot be estimated from our data set, but may be high given the 8 patients (about 4%) who were excluded from this analysis because they died within 48 hours of symptom onset. The need for rapid diagnosis and operative repair in the early stages of dissection cannot be overemphasized.
Patients who have already made it through the early period appear to have a decreased likelihood of rupture, which allows for additional testing and medical optimization. In the absence of hemodynamic compromis, these patients may be stabilized on optimal blood pressure medication and afterload reduction and further diagnostic testing pursued while their medical condition is optimized. A decision can then be made to pursue surgical intervention in an urgent rather than an emergency setting.
Consults and other imaging or diagnostic studies can contribute to clinical decision-making and may diagnose related or concomitant medical conditions that necessitate changing the operative procedure (ie, a cardiac catheterization revealing unknown CAD that would change a procedure to include a concomitant coronary artery bypass grafting). Although controversy exists about the prevalence of CAD in the dissection population, it may be as high as 30% [4], suggesting that some patients may benefit from concomitant revascularization.
Outcomes in Patients With Delayed Repair
Traditionally, medically managed patients have fared much worse than have surgically treated patients [4, 9]. Because emergency surgery is the current standard treatment, little is known about the natural history of those patients who present after 48 hours into the course of their dissection. These data suggest that a select number of late-presenting patients, having survived the initial danger period, are less likely to rupture than previously appreciated. In our study, no increased 30-day mortality occurred amongst the 39 patients who received solely medical management, and long-term survival was only minimally worse than those who underwent operations after a delay.
These patients were almost universally excluded from surgical treatment because of the severity of their comorbid diseases. Patients in group A were older than those in group B (median age of 68.8 versus 59.3, respectively) and had a higher incidence of chronic obstructive pulmonary disease (26.6% versus 8.4%), a higher incidence of congestive heart failure (14.1% versus 1.0%), and a higher incidence of CAD (43.5% versus 14.6%), indicating the poor clinical status of patients deemed unsuitable for immediate operation. Of interest was that they also had a lower incidence of valve involvement; this may have resulted in less severe symptomatology and a higher likelihood of refusal of operative repair.
It is important to determine if comorbidities are more dangerous if left uncorrected for emergency surgery, or if treating comorbidities at the cost of delayed or declined surgery is appropriate to improve operative candidacy and outcomes. Our results would indicate that in those patients presenting after a delay, the cost of delayed surgery may be less than previously thought, and correcting comorbidities before surgery or avoiding surgery will result in acceptable outcomes.
Natural History
Our data demonstrate that once patients have survived the initial postdissection period, their dissections may have stabilized. Long-term survival is nearly identical for patients in group A and group B. Medical management in these patients, especially those with limited life spans because of comorbid conditions or age, results in better outcomes than previously thought. These findings should be incorporated into the preoperative planning and discussions with the patient.
Limitations
Several limitations of these data should be enumerated:
First, selection bias is unavoidable in a retrospective cohort review. The sickest patients with type A dissection often die in a prehospital setting. Because the patients analyzed here were all seen at a large referral center, those patients seen at smaller institutions and deemed unsuitable for transfer were excluded. Thus, patients included here were likely healthier and more stable than the average patient presenting with an acute type A dissection to a local emergency department.
Second, several factors limit the completeness of the data obtainable: (1) patients thought not to be operative candidates may not have received follow-up care within our institution (although survival data is complete, the incidence of major complications during follow-up may be underestimated); (2) in the absence of prospective data collection with clearly defined terms, this study relies on medical record documentation by a wide range of providers; (3) as a referral population, complete data collection for the early disease course relies on records transmitted from outside institutions.
Finally, the small sample size necessarily limits the statistical power of several findings.
Nevertheless, these data clearly demonstrate that in a highly selected population of patients surviving the initial 48 hours after acute type A dissection, interval or permanent nonoperative management provides a viable alternative.
Conclusion
Traditional management of type A dissection with emergency surgery continues to carry high perioperative morbidity and mortality, likely related to the frequency of significant hemodynamic compromise and rupture in patients presenting acutely and requiring immediate operative repair. The cohort of patients who survive the initial 48 hours after an acute dissection may have acceptable outcomes with delayed management, despite their significant comorbidities. It must be assumed, however, that many others most likely died before presentation.
Although we are not advocating a nonoperative treatment approach for these late-presenting patients, delayed management carries several specific advantages: stabilization of aortic tissue, optimization of clinical condition, obtaining consults that can contribute to clinical decision-making, and diagnosis of related medical conditions. These factors allow for more informed decisions about patient care; for example, in patients with significant risk factors for coronary artery disease, preoperative cardiac catheterization may lead to the addition of coronary bypass grafting to the procedure. Determinations can also be made whether to treat these patients nonoperatively (where the risks outweigh the benefits) or surgically (in the urgent setting, where the surgeon is fully-informed and the patient has been stabilized). This strategy may result in decreased perioperative mortality in these patients compared with those who require immediate repair.
We continue to recommend immediate operative repair in patients presenting soon after their acute type A dissections, but in those who have survived a significant delay, an urgent (as opposed to emergency) operation after completion of additional diagnostic studies and careful attention to improve the overall clinical status offers acceptable perioperative survival. In patients with severe comorbidities presenting with a contraindication to surgical repair, a nonoperative approach with optimization of the clinical condition is a viable strategy.
|
Appendix |
|---|
|
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
DR DAVIES: Clearly, we do not recommend that patients with an acute dissection be delayed and managed medically. We are looking at a very specific subset of patients who present to a tertiary medical facility. These patients have already survived their stay at home, they have survived transport to the hospital, and maybe even survived transfer to a tertiary medical facility. That select group of patients has survived what we called in the last slides "the eye of the storm"; their dissection may be more stable and allow for urgent, rather than emergent repair. The additional time gained may be used both for medical stabilization and further diagnostic work-up such as cardiac catheterizations.
DR EDWARD B. SAVAGE (St. Louis, MO): I think it is important to reinforce one of the points you made, and one of your graphs concerns me, because you are comparing mortality rates and results between two different groups, and when you publish this, if you put that into the paper, it may be interpreted that everyone should wait 48 hours. It is very important that you reinforce that these are not comparable groups and that the patients in the delayed group have been selected out.
DR DAVIES: That is a very good point. These are patients who are highly selected, and we are not doing the selecting. The patients have been selected out by their disease process. So that it is really patients who have waited 48 hours to come in or through a transfer process it has taken 48 hours to get them to our center. It is a very specific subset of patients.
DR SAVAGE: Just make sure you include that, as an important component of your recommendations in the paper.
DR DAVIES: Thank you.
DR MARC MOON (St. Louis, MO): Dr Davies, I would like to congratulate you on addressing a difficult but very important topic—the timing of urgent versus emergent surgery. Your data suggests that we could wait from 2 o’clock in the morning until 7 o’clock in the morning to repair a subacute dissection. Can we wait from 2 o’clock in the morning on Friday until Monday morning now? What exactly is the time frame you are talking about in recommending it is safe to wait?
DR DAVIES: It is difficult to recommend waiting a whole weekend in this disease process. There are certainly patients in this group who are followed for a few days and stabilize medically, but based on these data, it is hard to conclude with a specific number. All of these patients still require what we like to call semi-elective surgery within 24 to 48 hours at the latest. However, it is hard to define exactly what the cutoff should be, because patients are dying at each time point. There is a mortality cost.
DR MOON: I think your results are consistent with previous studies based on the fact that these are patients who don’t have cardiogenic shock, don’t have flow complications, or any of the other multivariate factors that we have found to impact survival. Very nicely done.
DR JORGE A. WERNLEY (Albuquerque, NM): Nice paper, nice presentation. We certainly respect your group very much. This data is in parallel with some Japanese publications which have already suggested that after 48 hours the natural history of the disease changes dramatically. I have two questions, though. What do you gain by waiting? In the example that you gave, it was a marginal advantage of having a better team in the OR, but I would like to understand when you say stabilizing the patient better. What do you gain by the waiting? And then if you could better define who are the patients who you would not operate on? What are the reasons, what are the pathways that lead you to a decision of no surgery?
DR DAVIES: To answer the first question, I think the benefit of waiting is primarily really of operating with your own team and making sure people are in a stable and optimized condition for the operating room. I think there is a detriment. There have been studies that have shown that operative outcomes with a team in the middle of the night are not as good as elective operations during the day. I think that is probably the primary thing that we can suggest from this data.
As far as reasons for not operating, malignancy is certainly one, extremes of age—80 to 90-year-olds—is an another, and severe comorbid disease that presents a contraindication to surgery. Malignancy and extremes of age are probably the two that account for most of the patients who are managed nonoperatively in this group.
DR JOSEPH S. COSELLI (Houston, TX): Congratulations on some very difficult data, but your definition of "delayed" is specifically "delayed" to you, and there are at least two groups of delayed possibly, those that are at home having had chest pain for two or three days and those who have been at another institution under medical care for a couple of three days before they present to you. Is there any difference in those groups?
And to follow-up on another question, which was really already asked but I guess I don’t fully understand your answer, and that is, your suggested recommendation of 2 o’clock in the morning until 7 o’clock in the morning as a delay, does your data really answer that question? I mean, it is not as if you took a group of patients and then randomized them and provided those two separate opportunities and had the results. But does your analysis of the data really answer that in a way that we can feel comfortable with? Thank you.
DR DAVIES: In terms of where the patients were delayed and does it make a difference, we didn’t really analyze that data separately, partly because it is very difficult to obtain data on outside hospitalizations and get accurate information from a variety of outside hospitals that are referring into our center.
As to the second question, obviously, we cannot randomize these patients. Our primary goal was to look at a group of patients who have already waited 48 hours, whether it is at home or in the hospital, and examine their survival. In doing so, we found that medical management of those patients, including those never undergoing operation, had better survival than historical series. Furthermore, patients who were operated on had strong trends towards improved outcomes versus those requiring immediate repair. Again, it is not a randomized sample, these patients are pre-selected, but I think these results give some support to the idea that we don’t have to operate within an hour or two on all patients who come in with a type A dissection.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
