Lederle FA, Johnson GR, Wilson SE, Ballard DJ, Jordan, Jr WD, Blebea J, Littooy FN, Freischlag JA, Bandyk D, Rapp JH, Salam AA, for the Veterans Affairs Cooperative Study #417 Investigators . Rupture Rate of Large Abdominal Aortic Aneurysms in Patients Refusing or Unfit for Elective Repair. JAMA. 2002;287(22):2968-2972. doi:10.1001/jama.287.22.2968
Author Affiliations: Department of Medicine, Veterans Affairs Medical Center, Minneapolis, Minn (Dr Lederle); Cooperative Studies Program Coordinating Center, Veterans Affairs Medical Center, West Haven, Conn (Mr Johnson); Department of Surgery, University of California, Irvine, Orange, Calif (Dr Wilson); Baylor Health Care System, Dallas, Tex (Dr Ballard); Department of Surgery, Veterans Affairs Medical Center, Birmingham, Ala (Dr Jordan); Department of Surgery, Pennsylvania State University, Hershey (Dr Blebea); Department of Surgery, Veterans Affairs Medical Center, Hines, Ill (Dr Littooy); Department of Surgery, Veterans Affairs Medical Center, Los Angeles, Calif (Dr Freischlag); Department of Surgery, Veterans Affairs Medical Center, Tampa, Fla (Dr Bandyk); Department of Surgery, Veterans Affairs Medical Center, San Francisco, Calif (Dr Rapp); Department of Surgery, Veterans Affairs Medical Center, Atlanta, Ga (Dr Salam).
Context Among patients with abdominal aortic aneurysm (AAA) who have high operative
risk, repair is usually deferred until the AAA reaches a diameter at which
rupture risk is thought to outweigh operative risk, but few data exist on
rupture risk of large AAA.
Objective To determine the incidence of rupture in patients with large AAA.
Design and Setting Prospective cohort study in 47 Veterans Affairs medical centers.
Patients Veterans (n = 198) with AAA of at least 5.5 cm for whom elective AAA
repair was not planned because of medical contraindication or patient refusal.
Patients were enrolled between April 1995 and April 2000 and followed up through
July 2000 (mean, 1.52 years).
Main Outcome Measure Incidence of AAA rupture by strata of initial and attained diameter.
Results Outcome ascertainment was complete for all patients. There were 112
deaths (57%) and the autopsy rate was 46%. Forty-five patients had probable
AAA rupture. The 1-year incidence of probable rupture by initial AAA diameter
was 9.4% for AAA of 5.5 to 5.9 cm, 10.2% for AAA of 6.0 to 6.9 cm (19.1% for
the subgroup of 6.5-6.9 cm), and 32.5% for AAA of 7.0 cm or more. Much of
the increased risk of rupture associated with initial AAA diameters of 6.5-7.9
cm was related to the likelihood that the AAA diameter would reach 8.0 cm
during follow-up, after which 25.7% ruptured within 6 months.
Conclusion The rupture rate is substantial in high-operative-risk patients with
AAA of at least 5.5 cm in diameter and increases with larger diameter.
Rupture of abdominal aortic aneurysm (AAA) can be prevented by elective
surgical repair, but because most AAA never rupture,1
elective repair is reserved for patients at high risk of rupture. The most
commonly used predictor of rupture is the maximum diameter of the AAA. Two
randomized trials found no reduction in mortality from repairing AAA smaller
than 5.5 cm in patients at low operative risk.2,3
No randomized trials are available in patients with larger AAA, and decision
making in these patients is often complicated by advanced age and serious
comorbidities. Surgery is usually deferred in high-operative-risk patients
until the AAA attains a diameter at which the risk of rupture is thought to
outweigh the operative risk. However, few data are available on the rupture
risk of large AAA, resulting in substantial disagreement among experts.4 We conducted a prospective observational Veterans
Affairs Cooperative Study to determine the incidence of rupture in patients
with large AAA for whom elective repair was not planned because of medical
contraindications or patient refusal.
Eligible patients were those evaluated at 47 Veterans Affairs medical
centers who were diagnosed as having AAA of at least 5.5 cm in diameter by
ultrasonography or computed tomography (CT) within 3 months prior to enrollment
and for whom elective repair was not expected in the next 6 months because
of medical contraindications to surgery or patient refusal. Patients with
the following were excluded: symptoms or radiological evidence of rupture,
previous aortic surgery, dissection of the thoracic aorta, known condition
associated with secondary AAA (eg, Marfan disease), or death expected in the
next 30 days. Informed consent and signature by the patient and next of kin
of a nonbinding form indicating willingness to have autopsy were required.
Patients were identified when they were referred to the vascular surgery service,
a process facilitated by use of notices and reminders sent to other physicians.
The protocol required the vascular surgery team at each center to offer entry
to all eligible patients.
Follow-up began at enrollment by telephone call to the study's central
office confirming eligibility and consent, so the study was entirely prospective.
Subsequent measurements of AAA were obtained by ultrasonography at 6-month
intervals throughout the study. The maximum outside AAA diameter was used,
as determined by the radiologist's reading at the participating medical center,
consistent with usual clinical practice. Follow-up ended at the time of elective
AAA repair, following successful repair of rupture, at death, or at the end
of the study.
The main outcome measure was incidence of rupture by initial and attained
AAA diameter. Attained AAA diameter was determined by allocating patients
to a diameter stratum based on the day that the AAA attained that diameter,
either initially or later during follow-up, at which time they were censored
out of any previous smaller AAA stratum (so events were only counted once).
Patients were not moved from larger to smaller diameter strata. Measurements
obtained within 7 days before a possible rupture (which could not be used
for decisions regarding elective repair and could reflect acute prerupture
AAA expansion) were not used to assign new strata or to compute AAA enlargement
rate as a possible predictor of rupture.
We planned to enroll 120 patients in each of 5 strata of AAA diameter:
5.5 to 5.9 cm, 6.0 to 6.4 cm, 6.5 to 6.9 cm, 7.0 to 7.9 cm, and 8.0 cm or
larger. Because fewer patients were actually enrolled, resulting in low numbers
in some strata, we collapsed the data into 3 strata, 5.5 to 5.9 cm, 6.0 to
6.9 cm, and 7.0 cm or larger, selected because they had been used in previous
reports5,6 and resulting in at
least 50 patients per stratum.
Medical records, including hospital records, nursing home notes, and
imaging and autopsy reports, were requested for deaths or AAA repair procedures.
Eyewitness accounts were obtained by telephone for deaths that occurred outside
of health care facilities. Death certificates were not used to assign cause
Rupture rates were generated by product-limit estimates (SAS PROC LIFETEST,
SAS Institute Inc, Cary, NC) for definite, probable, and possible ruptures.
Definite ruptures were those confirmed at surgery or autopsy or by CT. Probable
ruptures included, in addition to definite ruptures, cases in which patients
died with symptoms consistent with rupture (severe abdominal, flank, or back
pain7) but without objective confirmation of
rupture, and also patients who had urgent repair of AAA that had not ruptured
but had developed symptoms consistent with rupture, a presentation that often
heralds imminent rupture.7,8 Possible
ruptures included probable ruptures and cases in which patients had sudden
unexplained or unwitnessed deaths without autopsy. Many patients in the latter
group would be expected to have died of cardiac or pulmonary disease, but
AAA is also a leading cause of sudden death.9
We expected that the true rupture rate should be between the definite rupture
rate and the possible rupture rate, and we consider the probable rupture rate
to be the best estimate of the true rate.
We used Cox regression models (SAS PROC PHREG) to assess baseline variables
as predictors of rupture and logistic regression models (SAS PROC LOGIST)
that included last measured AAA diameter to assess AAA enlargement rate as
a predictor of rupture. Enlargement rate was calculated using the first and
last imaging test for each patient after excluding measurements within 7 days
of possible rupture and patients with less than 120 days between the first
and last measurements (because of the greater distortion due to measurement
variation). The number of ruptures observed was not large enough to generate
valid predictive models using all collected variables simultaneously,10 so models using AAA diameter plus 1 other variable
Among 266 eligible patients, 68 refused enrollment. The remaining 198
patients were enrolled between April 1995 and April 2000 and followed up through
July 2000. All patients but 1 were men, nearly all had a history of smoking
(Table 1), reflecting both the
veteran population and the population at risk for AAA,11
and most were elderly and had high rates of comorbidities, especially coronary
artery disease and chronic obstructive pulmonary disease.
Vital and operative status as of July 2000 was confirmed for all patients.
Hospital records were obtained for all cases of AAA repair, and firsthand
information (including hospital and nursing home records, autopsy reports,
and eyewitness reports, as applicable in each case) was obtained for all deaths.
During a mean follow-up of 1.52 years, 112 patients (57%) died. All-cause
mortality was 29.8% at 1 year, 55.5% at 2 years, and 75.4% at 3 years. Autopsy
was performed on 52 (46%) of the 112 deaths. Rupture was confirmed by surgery
or autopsy in 35 patients (17.7%), of whom 17 had attempted aneurysm repair
(2 endovascular repairs) and 6 survived. Four patients had urgent repair of
symptomatic unruptured AAA, 3 of whom survived. Six patients (3.0%) died with
symptoms consistent with AAA rupture. Seven patients (3.5%) had sudden unexplained
or unwitnessed deaths with no autopsy. Follow-up was terminated because of
elective repair of asymptomatic AAA in 21 patients (10.6%), 2 of whom had
endovascular repair, with 3 postoperative deaths (all in patients who had
open repair). Sixty-nine patients (34.8%) died from a known cause other than
AAA (21 cardiac, 20 respiratory, 6 stroke, 14 cancer, and 8 miscellaneous
deaths). Fifty-six patients (28.3%) were alive without AAA rupture at the
end of the study.
The cumulative incidence of possible, probable, or definite rupture
for 3 strata of initial AAA diameter is shown in Table 2 and of probable rupture for 3 strata of initial AAA diameter
in Figure 1. Analysis of the data
using the originally planned 5 strata revealed that AAA of 6.5 to 6.9 cm had
an intermediate rupture risk between AAA of 5.5 to 6.4 cm and AAA of 7.0 cm
or more. The rates of probable rupture for AAA with initial diameter of 6.5
to 6.9 cm at 6, 12, and 18 months (after which <10 patients remained in
observation) were 10.3%, 19.1%, and 19.1%, respectively.
Assessment of rupture rate by attained AAA diameter required follow-up
imaging measurements. The percentage of ultrasonography follow-up visits completed
within 1 month of the due date was 72%. For all patients who had possible
rupture, imaging measurements were available for all but 4 patients within
6 months before the event and for all but 1 patient within 12 months before
the event. Cumulative incidence of rupture by 3 strata of attained AAA diameter
is shown in Table 3. Figure 2 shows the probable rupture rate
by attained AAA diameter using the originally planned 5 strata. The risk of
rupture increased significantly when the AAA reached at least 8.0 cm (Figure 2), indicating that much of the increased
risk associated with initial diameters of 6.5 to 7.9 cm was related to an
increased likelihood that the aneurysm would reach 8.0 cm during follow-up.
The rates of probable rupture for attained AAA diameter of at least 8.0 cm
at 6, 12, 18, and 24 months (after which <10 patients remained in observation)
were 25.7%, 36.4%, 39.5%, and 54.7%, respectively.
Diameter of the AAA was the strongest predictor of rupture in terms
of variance explained (relative risk [RR], 1.39 per 1 cm; 95% confidence interval
[CI], 1.11-1.73). After adjustment for AAA diameter at entry, the following
other factors were also significant predictors of probable rupture: renal
artery involvement of the AAA (RR, 2.36; 95% CI, 1.12-4.97), lower weight
(RR, 0.75 per 10 kg; 95% CI, 0.61-0.91), and, paradoxically, no history of
smoking (RR, 0.30; 95% CI, 0.11-0.84), no myocardial infarction (RR, 0.46;
95% CI, 0.24-0.88), and no coronary artery bypass graft surgery (RR, 0.30;
95% CI, 0.12-0.71). Age (RR, 1.00 per year; 95% CI, 0.96-1.05), family history
of AAA (RR, 0.79; 95% CI, 0.27-2.31), systolic blood pressure (RR, 1.00 per
mm Hg; 95% CI, 0.98-1.01), diastolic blood pressure (RR, 1.01; 95% CI, 0.98-1.04),
chronic obstructive pulmonary disease (RR, 0.76; 95% CI, 0.41-1.38), use of β-blockers
(RR, 0.53; 95% CI, 0.24-1.21), and poor medical condition (compared with refusal
alone as the reason for no planned repair, RR, 0.81; 95% CI, 0.40-1.65) were
not significant predictors.
The median rate of change in AAA diameter was an increase of 0.43 cm
per year (interquartile range, 0.08-0.79 cm per year). Unadjusted for AAA
diameter, patients with probable rupture had a significantly greater mean
AAA enlargement rate than patients without rupture (0.75 cm per year vs 0.41
cm per year; P = .003). In models that included both
enlargement rate and the last measurement of AAA diameter as independent variables,
enlargement rate was not a significant predictor of probable rupture (RR per
0.1 cm per year, 1.07; 95% CI, 0.99-1.15; P = .09).
We report a prospective, multicenter, observational study of the rupture
rate of large AAA in mostly high-operative-risk male patients for whom outcome
ascertainment was complete and based on firsthand reports. Rupture risk was
substantial for all AAA greater than 5.5 cm and increased markedly when initial
diameter exceeded 6.5 or 7.0 cm or attained diameter exceeded 8.0 cm. These
diameters could serve as useful thresholds for decision making in high-operative-risk
male patients but may not apply to women12
or to patients who have low operative risk. The 9.4% 1-year probable rupture
rate we observed for initial AAA of 5.5 to 5.9 cm is 10-fold higher than the
rates observed in 2 recent randomized trials for AAA of 4.0 to 5.4 cm.2,3 While this large difference likely
results in part from the difference in AAA diameters, it may also reflect
a difference in rupture rates between our mostly high-operative-risk patients
with severe comorbidities and the healthier patients randomized into the trials.
Consistent with this possibility, ineligible patients followed up outside
the United Kingdom Small Aneurysm Trial had a higher rupture rate than randomized
patients with comparable AAA diameters.12
Because most large AAA are surgically repaired, natural history data
are difficult to obtain, even in patients at high operative risk, as evidenced
by our 47 medical centers requiring 5 years to enroll 198 patients. For this
reason, few data exist in the literature, and most predate the introduction
of accurate AAA measurement techniques. In 1966, Szilagyi et al13
reported rupture in 61 of 141 AAA of at least 7 cm among patients who were
followed up for a mean of 17 months, a crude rate of 30.5% per year. Recently,
Powell et al12 observed 0.28 ruptures per patient-year
in 100 patients with AAA greater than 5.5 cm. Several other studies have also
reported rupture rates that are generally consistent with our findings.5,6,14
As expected, AAA diameter was the strongest predictor of rupture in
our study. The unexpected protective effects of smoking, myocardial infarction,
and coronary artery bypass graft surgery may be artifacts of the high mortality
rate. Enlargement rate of AAA was a significant univariate predictor of rupture,
but the effect was not significant after adjustment for last measured AAA
diameter. An enlargement rate of at least 1.0 cm/y has been used as an independent
indication for repair of small AAA in 3 large randomized trials.2,3,15
However, we are aware of no previous studies that measured the impact of enlargement
rate on rupture risk. Our data suggest but do not confirm an independent effect,
and further studies are needed to resolve this question.
Outcome ascertainment is problematic in studies of AAA rupture because
causes of sudden death are difficult to distinguish. Our requirement of written
intent to have autopsy, cosigned by next of kin, resulted in a rate 4 times
the national rate16 but still less than 50%.
Hospital and nursing home records and eyewitness accounts were therefore important
for optimizing outcome assessment in our study.
We conclude that the rupture rate is substantial in patients with high
operative risk and AAA greater than 5.5 cm, and increases markedly with diameter.