Académique Documents
Professionnel Documents
Culture Documents
Background—Given the high mortality rates in patients with type A aortic dissection, predictive tools to identify patients
at increased risk of death are needed to assist clinicians for optimal treatment.
Methods and Results—Accordingly, we evaluated 547 patients with this diagnosis enrolled in the International Registry
of Acute Aortic Dissection (IRAD) between January 1996 and December 1999. Univariate testing followed by
multivariate logistic regression analysis was performed to identify independent predictors of death. In-hospital mortality
rate was 32.5% in type A dissection patients. In-hospital complications (neurological deficits, altered mental status,
Downloaded from http://circ.ahajournals.org/ by guest on December 9, 2016
myocardial or mesenteric ischemia, kidney failure, hypotension, cardiac tamponade, and limb ischemia) were increased
in patients who died compared with survivors (P⬍0.05 for all). Logistic regression identified the following presenting
variables as predictors of death: age ⱖ70 years (OR, 1.70; 95% CI, 1.05 to 2.77; P⫽0.03), abrupt onset of chest pain
(OR 2.60; 95% CI, 1.22 to 5.54; P⫽0.01), hypotension/shock/tamponade (OR, 2.97; 95% CI, 1.83 to 4.81; P⬍0.0001),
kidney failure (OR, 4.77; 95% CI, 1.80 to 12.6; P⫽0.002), pulse deficit (OR, 2.03; 95% CI, 1.25 to 3.29, P⫽0.004),
and abnormal ECG (OR, 1.77; 95% CI, 1.06 to 2.95; P⫽0.03) (area under receiver operating curve, 0.74;
Hosmer-Lemeshow statistic, P⫽0.75).
Conclusions—The in-hospital mortality rate in acute type A aortic dissection is high and can be predicted with the use of
a clinical model incorporated in a simple risk prediction tool. This tool can be used to educate patients with dissection
about their predicted risk and in clinical research for risk adjustment while comparing outcomes of different therapies.
(Circulation. 2002;105:200-206.)
Key Words: aorta 䡲 mortality 䡲 surgery 䡲 risk factors
Received August 8, 2001; revision received November 2, 2001; accepted November 5, 2001.
From the University of Michigan (R.H.M., P.G.H., J.V.C., D.E., W.F.A., K.A.E.), Ann Arbor; University of Tokyo (T.S.), Japan; Instituto Policlinico
San Donato (E.B.), Milan, Italy; Hadassah University Hospital (D.G.), Jerusalem, Israel; Hospital Universitario “12 de Octubre” (A.L., L.C.M.), Madrid,
Spain; and University of Rostock (C.A.N.), Germany.
Correspondence to Kim A. Eagle, MD, University of Michigan, 3910 TC, Ann Arbor, MI 48109. E-mail keagle@umich.edu
© 2002 American Heart Association, Inc.
Circulation is available at http://www.circulationaha.org
200
Mehta et al Predicting Death in Type A Aortic Dissection 201
TABLE 1. Demographics and History for Patients With Type A Aortic Dissection
Variable Overall Survived Died P
n (%) 547 (100) 369 (67.5) 178 (32.5)
Demographics
Age, mean⫾SD, y 61.9 (14.1) 60.7 (13.6) 64.4 (14.9) 0.004
Age ⱖ70 y 192 (35.2) 110 (30.0) 82 (46.1) 0.0002
Male 357 (65.5) 255 (69.3) 102 (57.3) 0.006
Transferred to IRAD sites 368 (67.4) 254 (69.0) 114 (64.0) 0.24
Etiology and patient history
Marfan syndrome 31 (5.8) 21 (5.8) 10 (5.9) 0.96
Hypertension 367 (69.4) 254 (70.6) 113 (66.9) 0.39
Atherosclerosis 145 (27.2) 91 (25.1) 54 (31.8) 0.11
Biscuspid aortic valve 15 (5.8) 12 (6.7) 3 (3.9) 0.38
Iatrogenic dissection 33 (6.5) 21 (6.0) 12 (7.4) 0.56
Prior aortic dissection 18 (3.4) 16 (4.4) 2 (1.2) 0.05
Prior aortic aneurysm 66 (12.4) 45 (12.4) 21 (12.5) 0.93
Downloaded from http://circ.ahajournals.org/ by guest on December 9, 2016
nosis was suspected on the basis of the history and physical Predictive Modeling
examination and was confirmed by imaging study findings, visual- Iterative logistic regression modeling for in-hospital death with the
ization at surgery, and/or postmortem examination. likelihood ratio test used for model selection was performed. Initial
modeling used elements marginally suggestive of an unadjusted
Patient Selection association to in-hospital death (defined as P⬍0.20). Variables were
All patients with acute type A aortic dissection enrolled in IRAD reviewed for clinical significance before testing. Diagnostic routines
from January 1, 1996, to December 31, 1999, were included for the (Hosmer-Lemeshow test for lack of fit, change in deviance, and
purpose of this analysis. Acute type A dissection was defined as any likelihood ratio test) were used on final model selection. SAS 8.1
dissection that involved the ascending aorta with presentation within was used for all analyses.
14 days of symptom onset.13,22
Development of a Simple Bedside Risk
Data Collection Prediction Tool
Data were assembled with the use of a standardized data form on The variables found to be significantly associated with in-hospital
patient demographics, history, clinical presentations, physical find- death in the “best” regression model were assigned a score equal to
ings, imaging studies results, details of medical and surgical treat- their coefficients in the fitted model (natural log of their odds ratios
ment, and patient outcomes. Completed data entry forms were rounded to the nearest decimal) (Appendix). The sum of this
forwarded by the participating IRAD sites to the coordinating center numerical score in an individual patient could then be used to predict
at the University of Michigan. Data forms were reviewed for his or her in-hospital death. A risk prediction tool that plotted the
analytical internal validity, and external validation was performed various score levels against the corresponding predicted death was
through a random (5%) field selection and error audit. Data were developed to aid clinicians in predicting in-hospital death for patients
scanned electronically into an Access database. with type A aortic dissection.
TABLE 2. Clinical Presentations, Signs, and Diagnostic Imaging Results of All Patients With
Type A Aortic Dissection
Variable Overall Survived Died P
Clinical presentations and signs
Abrupt onset of pain 435 (84.5) 289 (82.3) 146 (89.0) 0.05
Migrating pain 69 (13.6) 38 (11.1) 31 (18.8) 0.02
All neurological deficits 93 (17.0) 53 (14.4) 40 (22.5) 0.02
Coma/altered consciousness 78 (15.0) 33 (9.5) 45 (26.5) ⬍0.0001
Syncope 96 (18.5) 63 (17.9) 33 (19.8) 0.60
Congestive heart failure 31 (6.1) 26 (7.5) 5 (3.0) 0.05
Mean systolic blood pressure (SD), mm Hg 127 (39.8) 133 (35.7) 115 (45.3) ⬍0.0001
Mean diastolic blood pressure (SD), mm Hg 72 (24.9) 75 (22.3) 66 (29.2) ⬍0.0001
Hypotension/shock/tamponade 154 (29.0) 72 (20.1) 82 (47.1) ⬍0.0001
Any pulse deficit 153 (30.1) 84 (24.7) 69 (41.1) 0.0002
Diagnostic imaging
Chest radiography 475 (86.8) 323 (87.5) 152 (85.4) 0.49
Downloaded from http://circ.ahajournals.org/ by guest on December 9, 2016
26.9% in surgically treated patients versus 56.2% in those Predictive Model for In-Hospital Death
treated medically without surgery. The cause of death was not Independent predictors of in-hospital death are as shown in
specified in 33.3% of these patients, whereas rupture ac- Table 4. The area under the receiver operating curve for the
counted for a third of death (33.3%), followed by neurolog- model was 0.74, suggesting good model discrimination. The
ical deficit (13.9%), visceral ischemia/kidney failure (11.5%) deviance probability value was 0.12, and the Hosmer-
and cardiac tamponade (7.9%) in the remaining. Lemeshow statistic was not significant, indicating little de-
parture from a perfect fit (2⫽4.3; df⫽7; P⫽0.75). Figure 1
Univariate Predictors of In-Hospital Death for All plots expected deaths versus observed deaths over the risk
Patients With Type A Aortic Dissection categories.
Clinical characteristics that showed significant association
included advanced age, female sex, abrupt onset and migra-
tory chest pain, hypotension/shock/tamponade on presenta- Variables Score and the Bedside Risk
tion, and evidence of neurological or pulse deficits at presen- Prediction Tool
tation (Tables 1 through 3). Furthermore, the presence of a The risk variables and their allocated scores are listed in
widened mediastinum on chest radiography, electrocardio- Table 4. Kidney failure was given the highest score (1.6),
graphic evidence of new Q waves and/or ST-segment devia- whereas sex was given the lowest score (0.3), with other
tions, or a lack of a normal ECG were associated with higher variable scores being in between these values [hypoten-
in-hospital mortality rates. As expected, complications such sion/shock/tamponade (1.1), abrupt onset of pain (1.0), age
as neurological deficit, myocardial ischemia, hypotension, ⱖ70 (0.5), and abnormal ECG on presentation (0.6)].
kidney failure, and limb ischemia were also more frequent in There was a good agreement between observed and pre-
patients who died than in those who survived. On the other dicted death over score categories (Figure 2.). The simple
hand, a medical history of hypertension, diabetes, aneurysm, bedside risk prediction tool is shown in the Appendix and
or heart surgery was similar in the two groups of patients. in Table 4.
Mehta et al Predicting Death in Type A Aortic Dissection 203
prevented by earlier and more aggressive surgical treatment. dissection, irrespective of treatment strategies (surgical or
If complications have already developed, attempts should be medical), were analyzed in contrast to most prior investiga-
made to try and minimize their deleterious effect by estab- tions, which evaluated only the surgical cohort (Table 5).
lishing effective visceral (renal, cerebral, mesenteric) blood Thus, the findings of our study can be generalized to patients
flow and relieving organ ischemia in a timely fashion either with acute type A dissection having a broad spectrum of
through surgery or by percutaneous techniques such as clinical presentation across diverse medical institutions. Fi-
fenestration and/or endoluminal stents followed by definitive nally, this study provides an important risk prediction tool
surgical repair. This latter approach is less invasive and can that allows accurate estimation of a patient’s in-hospital risk
quickly restore vital organ perfusion so that the patient with of death at the bedside through the use of readily available
malperfusion can be stabilized and then go on to have a clinical information at presentation. This tool could also be
successful repair. Preliminary results of this approach at our valuable in outcome research to estimate risk-adjusted out-
institution as well as other sites have been encouraging.25–28 comes while comparing different subgroups of patients with
type A dissection.
Risk Prediction Bedside Tool for Predicting Death
The present study incorporated variables at presentation that Limitations of Our Study
were independently associated with death into a simple risk The results of our study should not be generalized to patients
prediction tool that is relatively accurate in predicting risk of with chronic stable type A dissection or to patients who are
death in patients with acute type A aortic dissection (Table 4). selected to undergo surgery. We measured only one easily
The model risk prediction estimates should help clinicians and reliably measurable outcome domain, in-hospital death,
while making recommendations about different treatment which, although important to patients, is not sufficient for the
options for individual patients with type A dissection. Like full evaluation of patients with type A aortic dissection. Our
risk prediction tools for other cardiovascular diseases, this focus on death as the only outcome should not diminish the
importance of other outcome variables such as nonfatal
adverse events, patient functional status, patient satisfaction,
and resource use. Future studies are needed to address the
best approach for evaluating the predictors of these other
domains.
Conclusions
Our study supports prior observations that acute type A
dissection is associated with high morbidity and mortality
rates despite recent advances in treatment. This study further
indicates that multiple factors affect in-hospital death and
help to identify these predictors. In addition, it provides a
useful, simple, bedside risk prediction tool that could be used
by physicians not only for determining the prognosis of
patients and counseling them and their families regarding
their risks but also as a research tool to better understand the
impact of newer diagnostic and therapeutic technological
Figure 2. Observed vs predicted death for acute type A aortic
dissection based on risk score. Note that as the score advances in the treatment of patients with acute type A aortic
increases, the predicted and observed mortality rates increased. dissection.
Mehta et al Predicting Death in Type A Aortic Dissection 205
14. Fann JI, Smith JA, Miller DC, et al. Surgical management of aortic dissection 22. Daily PO, Trueblood HW, Stinson EB, et al. Management of acute
during a 30-year period. Circulation. 1995;92(suppl II):II-113-II-121. dissections. Ann Thorac Surg. 1970;10:237–247.
15. Mazzucotelli JP, Deleuze PH, Baufreton C, et al. Preservation of the 23. Eagle KA, Guyton RA, Davidoff R, et al. ACC/AHA guidelines for
aortic valve in acute aortic dissection: long-term echocardiographic coronary artery bypass graft surgery: executive summary and recommen-
assessment and clinical outcome. Ann Thorac Surg. 1993;55:1513–1517. dations: a report of the American College of Cardiology/American Heart
16. Pansini S, Gagliardotto PV, Pompei E, et al. Early and late risk factors in Association Task Force on Practice Guidelines (Committee to Revise the
surgical treatment of acute type A aortic dissection. Ann Thorac Surg. 1991 Guidelines for Coronary Artery Bypass Graft Surgery). Circulation.
1998;66:779 –784. 1999;100:1464 –1480.
17. Chirillo F, Marchiori L, Andriolo L, et al. Outcome of 290 patients with 24. Lytle BW, Mahfood SS, Cosgrove DM, et al. Replacement of the
aortic dissection, a 12 year multicenter experience. Eur Heart J. 1990; ascending aorta: early and late results. J Thorac Cardiovasc Surg. 1990;
11:311–319.
99:651– 658.
18. Scholl FG, Coady MA, Davies R, et al. Interval or permanent non-
25. Deeb GM, Williams DM, Bolling SF, et al. Surgical delay for acute type
operative management of acute type A aortic dissection. Arch Surg.
A dissection with malperfusion. Ann Thorac Surg. 1997;64:1669 –1677.
1999;134:402– 406.
19. Haverich A, Miller DC, Scott WC, et al. Acute and chronic aortic 26. Chavan A, Hausmann D, Dresler C, et al. Intravascular ultra-sound
dissections: determinants of long-term outcome for operative survivors. guided percutaneous fenestration of the intimal flap in the dissected aorta.
Circulation. 1985;72(suppl II):II-22-II-34. Circulation. 1997;96:2124 –2127.
20. Niederhauser U, Kunzli A, Seifert B, et al. Conservative treatment of the 27. Walker PJ, Dake MD, Mitchell RS, et al. The use of endovascular stent
aortic root in acute type A dissection. Eur J Cardiothorac Surg. 1999; techniques for the treatment of complications of aortic dissection. J Vasc
15:571–578. Surg. 1993;18:1042–1051.
21. Safi HJ, Miller CC, Reardon MJ, et al. Operation for acute and chronic 28. Slonim SM, Nyman U, Semba CP, et al. Aortic dissection: percutaneous
aortic dissection: recent outcome with regard to neurologic deficit and management of ischemic complications with endovascular stents and
early death. Ann Thorac Surg. 1998;66:402– 411. balloon fenestration. J Vasc Surg. 1996;23:241–253.
Downloaded from http://circ.ahajournals.org/ by guest on December 9, 2016
Predicting Death in Patients With Acute Type A Aortic Dissection
Rajendra H. Mehta, Toru Suzuki, Peter G. Hagan, Eduardo Bossone, Dan Gilon, Alfredo Llovet,
Luis C. Maroto, Jeanna V. Cooper, Dean E. Smith, William F. Armstrong, Christoph A.
Nienaber and Kim A. Eagle
on Behalf of the International Registry of Acute Aortic Dissection (IRAD) Investigators
Downloaded from http://circ.ahajournals.org/ by guest on December 9, 2016
Circulation. 2002;105:200-206
doi: 10.1161/hc0202.102246
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2002 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circ.ahajournals.org/content/105/2/200
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located,
click Request Permissions in the middle column of the Web page under Services. Further information about
this process is available in the Permissions and Rights Question and Answer document.