(Stroke. 1999;30:1621-1627.)
© 1999 American Heart Association, Inc.
Original Contributions |
From the Department of Neurosurgery, Koshigaya Hospital, Dokkyo University School of Medicine, Saitama, and Department of Neurosurgery, Dokkyo University School of Medicine, Tochigi, Japan.
Correspondence to Dr Yuhei Yoshimoto, Department of Neurosurgery, Koshigaya Hospital, Dokkyo University School of Medicine, 2-1-50 Minami-Koshigaya, Koshigaya, Saitama 343-8555, Japan.
| Abstract |
|---|
|
|
|---|
MethodsWe used a Markov model to evaluate the cost-effectiveness of screening for asymptomatic, unruptured intracranial aneurysms. The model involved a set of variables describing discrete health states. Each state was assigned a quality of life score and an associated medical cost. A comparison of the expected outcomes was then made between 2 hypothetical cohorts, one receiving screening and the other no screening. A sensitivity analysis was performed by altering the input values within clinically reasonable ranges to reflect uncertainty in the baseline analysis and then assessing the effects on outcomes.
ResultsCombining the incremental cost and effectiveness data revealed a cost per quality-adjusted life-year of $7760 for an annual rate of subarachnoid hemorrhage due to unruptured aneurysms (rupture rate) of 0.02; this cost was $39 450 for a rupture rate of 0.01. There was no benefit (negative quality-adjusted life-year benefit) for a rupture rate of 0.005, the rupture rate found in a recently published international cooperative study. The risks of surgery for unruptured aneurysms and the discounting ratio used to assess the impact of timing of costs and benefits on future outcomes also had significant effects on the results. Other variables had little impact on cost-effectiveness.
ConclusionsThe cost-effectiveness of screening for an unruptured aneurysm is highly sensitive to the annual rate of subarachnoid hemorrhage due to unruptured aneurysms. The low annual rupture rate seen in the recent large international cooperative study implies that screening asymptomatic populations to identify and treat unruptured aneurysms would not be cost cost-effective.
Key Words: aneurysm cost-benefit analysis quality of life subarachnoid hemorrhage
| Introduction |
|---|
|
|
|---|
Magnetic resonance angiography12 and 3-dimensional CT angiography have been increasingly used to detect unruptured aneurysms among the general population of Japan ("No Dokku," or brain checkup).13 As a result, unruptured aneurysms are being identified at an ever-increasing rate.14 Screening followed by surgical repair of unruptured aneurysms has been introduced to reduce the number of patients developing SAH, and many neurosurgeons have reported reasonably good results in several series of patients.9 10 13 15 16 17 18 19 20 These authors all advocate surgical treatment, but do their results indicate that all asymptomatic, unruptured aneurysms should be repaired? If so, should wide-scale screening be implemented for asymptomatic populations? Although a large-scale, hospital-based, multicenter study was recently published,21 no population-based analysis has been performed.
Society and governments alike are increasingly aware of the economic implications of certain treatment decisions, and treatments that are not cost-effective will increasingly be rejected by public funding and private insurance reimbursement systems. Does the cost of screening for asymptomatic, unruptured intracranial aneurysms outweigh the potential cost posed by risk of a future SAH? To address these questions, we used a mathematical model to predict whether a screening program designed to identify individuals with asymptomatic aneurysms would be a cost-effective strategy for SAH prevention.
| Methods |
|---|
|
|
|---|
|
We made several simplifying assumptions. Our baseline case considered screening in a cohort of 50-year-old asymptomatic individuals. In our model, surgery was the only treatment available and was assumed to remove permanently the risk of SAH due to aneurysmal rupture. Although endovascular techniques are being developed for the treatment of intracranial aneurysms, their role has yet to be established in long-term follow-up studies; therefore, they were not considered in this study.
Screening Group
Screened individuals could be shown to have no unruptured
intracranial aneurysms (well without aneurysm) or to
have an aneurysm at the beginning of the first cycle. All of
the latter group were assumed to undergo aneurysm surgery. They
might then die perioperatively (dead), survive with
neurological deficits and a repaired aneurysm (disabled without
aneurysm), or survive with normal neurological function and a
repaired aneurysm (well without aneurysm). In
succeeding years, survivors were assumed to live at the postoperative
level of functioning until they died of causes other than
aneurysmal SAH. We also assumed that the aneurysm would
never rupture after surgery and that each patient would have the same
death rate as the general population.
No-Screening Group
Subjects who did not undergo screening would live without
knowing whether they had an intracranial aneurysm. They might
die of causes other than aneurysmal SAH or develop SAH. We
assumed that all unruptured aneurysms would rupture at a
constant rate over time and that all patients who developed
aneurysmal SAH and were admitted to hospital would undergo
surgical repair of the aneurysm. They might then die, survive
with a neurological deficit and a repaired aneurysm (disabled
without aneurysm), or survive without a neurological deficit
and a repaired aneurysm (well without aneurysm). It was
also assumed that patients with a repaired aneurysm would not
develop SAH in the future and would have the same death rate as the
general population.
Annual Death Rate of the General Population
The hazard function p(x), namely, the annual death
rate of the general population at age x, is known to
approximate well to an exponential function.24 25 We
plotted yearly death rates for each age group included in the model
using Japanese life statistics data from 1990.26 By
regression analysis, the following approximation of
p(x) was obtained2 :
![]() |
Baseline Analyses
Probabilities
Incidence of Unruptured Aneurysms
In previous autopsy27 28 29 30 31 32 and angiographic
series,11 13 29 33 34 35 there has been a wide range in the
incidence of unruptured aneurysms in the general population
(0.5% to 7.9%). For our baseline analyses, we chose an
incidence of unruptured aneurysm of 3.0% in 50-year-old
subjects.
Annual Rate of SAH Due to Unruptured Aneurysms
Many attempts have been made to clarify the natural history of
unruptured aneurysms.3 5 7 21 29 36 37 38 39 40 41 42 Juvela et
al37 followed up 142 patients with 181 aneurysms
for an average of 13.9 years (total of 1994 patient-years) and found a
yearly rupture rate of 1.4%. This rupture rate was stable over the
first, second, and third decades of follow up. Yasui et
al41 reported results for 234 patients with unruptured
intracranial aneurysms, with a mean follow-up period of 75
months, showing an average annual rupture rate of 2.3%.
However, lower annual rupture rates of unruptured intracranial aneurysms were found in a recently reported international cooperative study21 : The overall rupture rate for the 1449 patients in this retrospective cohort (12 023 patient-years of follow-up) was 0.5% per year. Because of the range of the above, we chose 3 different annual rupture rates, 0.02, 0.01, and 0.005, for our baseline analyses.
Mortality and Morbidity Due to Aneurysmal SAH
The authors of the International Cooperative Study on Timing of
Aneurysm Surgery reported that in their 3521 hospitalized
patients, the mortality rate was 26% and the morbidity rate was 16%;
58% of the patients recovered fully.43 44 However, more
recent articles have reported far higher mortality rates for SAH (52%
to 67%).21 37 40 41 45 Thus, we assumed that the
mortality rate of SAH would be 50% and that half of the 50% who
survived would have some disability and that the remainder would fully
recover.
Risks of Surgery for Unruptured Aneurysms
A review and meta-analysis of 28 separate articles
reporting the results of elective surgery for unruptured
asymptomatic aneurysms in 733 patients found a
morbidity rate of 4.1% (95% confidence interval, 2.8% to 5.8%) and
a mortality rate of 1% (95% confidence interval, 0.4% to
2.0%).46 These data may have been influenced by differing
sources of patient referral or selection and/or by publication
bias47 (ie, the tendency for excellent surgical results to
be published more often than average or poor results). A recent
prospective international study reported a much higher rate (combined
mortality and morbidity, 13.1% to 15.7%).21 Morbidity in
that study included cognitive impairment as well as physical
disability. We assumed the surgery-related mortality and morbidity
rates to be 10% in the baseline analysis. We also assumed
that, among those who had surgical complications, 20% died and the
rest (80%) survived and had some disability.
Costs
Each health state in the Markov model is associated with one of
the following cost considerations: (1) examination costs of
screening, (2) hospital costs of surgery for unruptured
aneurysms, (3) hospital costs of surgery for ruptured
aneurysms and acute care for SAH, and (4) rehabilitation and
nursing home costs in subsequent years of survival. We selected
magnetic resonance angiography as the screening method because of its
noninvasiveness and reliability,12 and we set the
screening cost at $200 (we are aware of much higher costs in other
countries). Thirty percent of SAH patients were assumed to die before
admission (prehospital death, zero cost). Accurate data on treatment
costs were not available for our baseline analysis; we
estimated the cost of a surgical admission for SAH, including acute
care, to be $20 000 and that for unruptured aneurysm to be
$15 000; these amounts are based on the scale of reimbursement used by
the Japanese health insurance system. We estimated the average cost of
long-term care for disabled patients after surgery or SAH to be
$10 000 per year. Our evaluation included only direct medical costs;
indirect costs, such as loss of earnings through inability to work,
were omitted.
Quality-Adjusted LifeYears
We expressed the outcome of each treatment strategy in terms of
the expected number of quality-adjusted life-years (QALYs) gained. To
establish the expected number of QALYs, we identified a "QALY
weight" score, ranging between 0.00 and 1.00, for each health state
in the model. Lifetime QALYs then equaled the sum of the number of
years spent in each health state, multiplied by the QALY weight
associated with that state.
Death
By convention, death is assigned a value of zero.
Postoperative or Post-SAH Recovery Period
On the basis of clinical experience, we assigned all patients
who underwent surgery for an unruptured aneurysm a
postoperative recovery period of 1 month. Three months were assigned
for those who underwent surgery after SAH due to a ruptured
aneurysm. For the postoperative recovery period, we assumed a
quality of life score equal to 60% of the full value.
Postoperative or Post-SAH Deficits
Several methods of measuring the quality of life after stroke
have been devised.48 49 Previous work in decision
analysis has also estimated the value of living with a
neurological deficit as 0.50 to 0.75. Thus, we assumed a value of 0.70
for survival with a postoperative or post-SAH neurological
deficit.50
Well Without Aneurysm
Normal function without an aneurysm or with a repaired
aneurysm was assigned a value of 1.00.
Well Without Screening
Normal function without knowing whether an unruptured
aneurysm was present was also assigned a value of 1.00.
Discounting
Benefits and costs are more significant if they occur earlier.
To account for this phenomenon, current values for both benefits and
costs were calculated by discounting the original values at the rate of
3% per year.
Cost-Effectiveness Ratios
We calculated the expected benefits and costs associated with
screening. A treatment strategy that yielded the greatest
quality-adjusted survival for the least expensive intervention would
obviously become the treatment of choice. However, for a strategy that
yielded greater quality-adjusted survival but was more expensive,
additional analysis would be necessary to determine whether the
longer survival justified the extra costs. This was calculated
using the incremental cost-effectiveness ratio:
Cost-effectiveness=(Cost A-Cost
B)/(Effectiveness A-Effectiveness B),
where A and B are 2 different treatment
options.
Sensitivity Analysis
The sensitivity analysis was performed by altering the
input values for individual variables within a clinically
reasonable range to assess the effects of uncertainties in the
assumptions made in the baseline analyses. The ranges of
variables tested included (1) incidence of unruptured
aneurysm (0.01 to 0.10), (2) mortality due to SAH (0.40 to
0.60), (3) overall mortality and morbidity due to surgery for an
unruptured aneurysm (0.05 to 0.15), (4) cost of surgery for an
unruptured aneurysm ($10 000 to $20 000), (5) cost of surgery
for a ruptured aneurysm plus acute care costs ($10 000 to
$30 000), (6) cost of long-term care for a disability ($5000 to
$15 000 per year), (7) examination costs of screening (magnetic
resonance angiography) ($100 to $300), (8) QALY value for postoperative
or post-SAH recovery period (0.50 to 0.70), (9) QALY value for living
with neurological deficits (0.60 to 0.80), (10) QALY value for living
without the knowledge of whether an unruptured aneurysm is
present (without having screening) (0.999 or 0.998), (11)
subject age (40 to 60 years), and (12) discount rate (0.05).
| Results |
|---|
|
|
|---|
Combining the incremental cost and effectiveness data in the baseline
analyses revealed a cost per QALY of $7760 and $39 450 in the
first and second analyses, respectively, which corresponded to
the additional cost for the gain in QALYs between the screening and
no-screening strategies. In the third analysis, a cost per QALY
was undefined because of a negative QALY benefit. Figure 2
shows the effects of annual rate of SAH
from unruptured aneurysms on the cost per QALY generated by
screening, indicating the marked impact of this variable.
|
Sensitivity Analysis
The sensitivity analyses for other selected model
variables are shown in the Table
. Overall mortality and morbidity
of surgery for unruptured aneurysms had significant effects on
the results in all analyses. The relationship between the risks
of surgery for unruptured aneurysms and the cost-effectiveness
ratio is demonstrated in Figure 3A
. As
the overall mortality and morbidity rose from 0.05 to 0.15, the cost
per QALY rose from $3360 to $13 810 in the first analysis,
from $17 290 to $138 200 in the second analysis, and from
$73 160 to undefined in the third analysis, indicating a
significant impact of this parameter in the second and
third analyses. Cost-effectiveness ratios are undefined when
screening is less effective and more expensive than no screening.
|
|
The effects of the incidence of unruptured aneurysms on the
cost-effectiveness relationship are shown in Figure 3B
. In the
first and second analyses, the costs per QALY were relatively
stable for this variable, varying from $5989 to $13 380 and
$33 410 to $57 640, respectively. In the third analysis,
screening was less effective and more expensive than no screening,
regardless of the incidence of unruptured aneurysms.
Age and the discounting ratio also have some influence on the cost per QALY: Older subjects and a high discount ratio (0.05) are less favored for screening. It is to be noted that input QALY scores of 0.999 and 0.998 for being well without screening significantly lowered the cost per QALY even in the third analysis (rupture rate=0.005). Other variables, including mortality of SAH and all aspects of the cost and QALY scores, had relatively little impact on the cost per QALY in all analyses.
| Discussion |
|---|
|
|
|---|
In the baseline calculations of our study, screening for unruptured aneurysms produced an incremental cost-effectiveness ratio of $7760 and $39 450 per QALY in the first and second analyses (annual rupture rate of 0.02 and 0.01, respectively), both of which are comparable to those of other widely practiced prophylactic interventions. There was no benefit from screening in our analysis for an annual rupture rate of 0.005.
In sensitivity analyses, we examined other factors that might
make screening more or less cost-effective. The risk of surgery for
unruptured aneurysms was a significant factor: At the 0.005
annual rupture rate, only extremely low surgical mortality and
morbidity would justify screening (Figure 3A
). Age had
significant effects on the results, as did discount rate. At higher
discount rates, screening techniques with early costs and risks
(screening and aneurysmal surgery) and delayed benefits
(increased quality-adjusted survival and decreased long-term care cost)
are less favored. Conversely, screening would be preferable at a lower
discount rate.
In contrast, the cost-effectiveness ratio was relatively robust when most other parameters, such as mortality due to SAH, costs of screening and surgery, and QALY scores in each health state, were considered.
The cost-effectiveness of screening was sensitive to the incidence of
unruptured aneurysms to some extent (see our second
analysis). The effects, however, seem to be limited, because
application of higher values resulted in a higher cost as well as a
greater QALY gain. This is relevant, because some populations are known
to have a greater-than-normal tendency to develop intracranial
aneurysms and SAH, eg, patients with polycystic kidney
disease58 and familial aneurysm.59 60
Our results do not support elective screening in such populations. We
recognized that some of these high-risk individuals have an excessive
fear of future SAH, even if they have no clinical symptoms. If such
patients have even a slight devaluation of their quality of life (cf
Table
, QALY score of well without screening: 0.999 or 0.998),
the cost-effectiveness analysis of screening is markedly
affected, regardless of whether the potentially higher incidence of an
unruptured aneurysm is considered. Screening of such patients
might be recommended, because this may ameliorate their psychological
distress about the possible future occurrence of SAH. We advocate that
the decision of whether to undergo screening should be left to
individuals and do not propose the use of public resources for this
purpose. Rupture of an aneurysm is a long-term risk over many
years, but screening and surgery generate an immediate risk. Some
patients may prefer to avoid surgery, even at the cost of a later
excess in risk; others may not.4
It must be recognized that the assumptions necessary for the construction of a mathematical model place limitations on the reliability of the conclusions. Thus, we assumed that all patients with unruptured aneurysms would undergo surgery and at the same level of risk. It may be reasonable to repair only aneurysms larger than 10 mm in diameter, because the recently published international study21 suggested that surgical intervention is not likely to improve the natural history of patients with aneurysms smaller than 10 mm. Satisfactory data on the risk of surgery by size of aneurysm would be of value but are not available.
Also, we assumed that SAH did not develop after screening showed no aneurysms or after an aneurysm was obliterated by surgery. This assumption may be false. Because aneurysms are rarely seen in younger subjects,32 it is obvious that they are not congenital but mostly develop during middle age. Therefore, after screening or successful surgery, some de novo aneurysms could grow, enlarge, and eventually rupture. However, to date, no data on the rate of de novo formation of aneurysms have been published. We postulate that the effect of de novo aneurysm formation on our results would be relatively small, because they would develop at the same rate in both the screening and no-screening populations.
Finally, the cost aspect of our study was restricted to direct medical costs.
Conclusion
Our study indicated that the cost-effectiveness of screening for
unruptured aneurysms was highly sensitive to assumptions about
the annual rate of SAH from an unruptured aneurysm. Screening
asymptomatic populations to avoid future SAH might be
cost-effective in a population with a high rupture rate (0.01 to 0.02
per year). In contrast, such screening is neither cost-effective nor
beneficial if the rupture rate is 0.005 per year (as recently
reported21 ). Under such circumstances, only extremely low
rates of surgical mortality and morbidity would make screening
justifiable.
Received February 23, 1999; revision received May 24, 1999; accepted May 24, 1999.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
T Baba, K Houkin, and S Kuroda Novel epidemiological features of moyamoya disease J. Neurol. Neurosurg. Psychiatry, August 1, 2008; 79(8): 900 - 904. [Abstract] [Full Text] [PDF] |
||||
![]() |
K. Nael, J. P. Villablanca, L. Mossaz, W. Pope, A. Juncosa, G. Laub, and J. P. Finn 3-T Contrast-Enhanced MR Angiography in Evaluation of Suspected Intracranial Aneurysm: Comparison with MDCT Angiography Am. J. Roentgenol., February 1, 2008; 190(2): 389 - 395. [Abstract] [Full Text] [PDF] |
||||
![]() |
D J A Connolly, D Birchall, and A Gholkar Current theory in imaging of intracranial vascular disease Imaging, October 1, 2002; 14(5): 396 - 408. [Abstract] [Full Text] [PDF] |
||||
![]() |
A J P Goddard, D Annesley-Williams, and A Gholkar Endovascular management of unruptured intracranial aneurysms: does outcome justify treatment? J. Neurol. Neurosurg. Psychiatry, April 1, 2002; 72(4): 485 - 490. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. B. Bederson, I. A. Awad, D. O. Wiebers, D. Piepgras, E. C. Haley Jr, T. Brott, G. Hademenos, D. Chyatte, R. Rosenwasser, and C. Caroselli Recommendations for the Management of Patients With Unruptured Intracranial Aneurysms : A Statement for Healthcare Professionals From the Stroke Council of the American Heart Association Stroke, November 1, 2000; 31(11): 2742 - 2750. [Full Text] [PDF] |
||||
![]() |
J. B. Bederson, I. A. Awad, D. O. Wiebers, D. Piepgras, E. C. Haley Jr, T. Brott, G. Hademenos, D. Chyatte, R. Rosenwasser, and C. Caroselli Recommendations for the Management of Patients With Unruptured Intracranial Aneurysms : A Statement for Healthcare Professionals From the Stroke Council of the American Heart Association Circulation, October 31, 2000; 102(18): 2300 - 2308. [Full Text] [PDF] |
||||
![]() |
Y. Yoshimoto, K. Yamashita, S. Kashiwagi, and S. Kato Treatment for Ruptured Aneurysms and Screening for Unruptured Aneurysms • Response Stroke, December 1, 1999; 30 (12): 2759 - 2768. [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Stroke Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 1999 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |