Screening for Abdominal Aortic Aneurysm in Men Ages 60 to 80 Years: A Cost-Effectiveness Analysis

Parameters Relevant to Screening

Most studies of AAA have been done in the high-risk population of men ages 60 to 80 years. Estimates of prevalence in this population range from 2% to 7.8% [10, 11]. Groups with additional risk factors including smoking, hypertension, or atherosclerotic vascular disease have a higher prevalence [2]. The prevalence of AAA is much less in younger adults. The male-to-female ratio for death from AAA is 11:1 between ages 60 to 64 years and narrows to 3:1 between ages 85 to 90 years [12]. Few aneurysms less than 4 cm in diameter will rupture [13, 14]. The rate of rupture of untreated aneurysms larger than 4 cm is directly related to the size of the aneurysm. The rate of AAA rupture as a continuous function of size, however, is not known precisely [15]. Therefore it is difficult to compare the benefits from screening using different aneurysm sizes at which surgery would be offered. Overall, 3% to 6% of aneurysms greater than 4 cm in diameter will rupture annually [14, 16]. Without surgical intervention, rupture of an aortic aneurysm is invariably fatal. Three percent to 5% of initially small (< 4 cm) aneurysms will progress to a potentially operable size each year [9, 17]. No direct data exist on the annual incidence of new aneurysms in a previously screened population; however, an estimate of 0.1% has been made [9].

Screening Tests

Two tests, ultrasonography and palpation of the abdomen by physical examination, have been advocated as screening tests for AAA. Other tests that can detect aneurysms, including plain radiographs of the abdomen, computed tomographic scanning, and magnetic resonance imaging, are either not sensitive enough or are too expensive to warrant serious consideration.

The accuracy of physical examination in detecting AAA is not completely known. Larger aneurysms are easier to detect than small ones, and it is easier to detect aneurysms in thin people. Estimates of the sensitivity of physical examination in detecting AAA, using surgical measurement or abdominal ultrasound as the gold standard, range from 22% to 96% [18, 19]. Higher sensitivities are obtained in studies of surgical candidates for resection, a group with a high prevalence of large aneurysms. Collin and associates [20] reported a sensitivity rate of 34.8%, a specificity rate of 97%, and a positive predictive value of 36% in a community-based population with a 5.4% prevalence of AAA. Lederle and colleagues [21] observed a sensitivity rate of 50% and a positive predictive value of 35% in a high-risk group with a prevalence of AAA of 9% screened in an internal medicine clinic. Four of five aneurysms larger than 5 cm in diameter in this series were detected by palpation. In contrast, Allen and coworkers [18] reported a 22% sensitivity rate and a 94% specificity rate with a positive predictive value of 17% in a sample of a population with a 5% prevalence of aneurysm. No large-scale, community-based studies of screening for AAA by physical examination have been reported.

Ultrasonography is an extremely sensitive and specific test for AAA of all sizes, at least in cases in which the diagnosis and size of the aneurysm can be confirmed at surgery. Reported sensitivity rates range from 82% to 99%, with sensitivity approaching 100% in some series of patients with a pulsatile mass [3, 6]. Based on Medicare reimbursement rates, abdominal ultrasonography costs about $150 in the United States. It should be possible to do screening ultrasound scans for much lessperhaps as little as $50. Data from other countries [9] have reported ultrasound costs as low as $35 in 1989.

Effectiveness of Treatment

Surgical resection and repair with an artificial graft is an effective treatment for AAA. If the patient survives the immediate postoperative period, long-term survival is comparable to that in persons who never had an AAA. The elective surgical mortality rate is about 5%. The mortality rate during emergency surgery for rupture is much higher, ranging from 21% to 76% [22, 23].

There is considerable debate about whether patients with aneurysms measuring between 4 and 5 cm in diameter should be observed or offered surgery [2, 13, 22]. Several studies are in progress to resolve this issue [2]; however, results will not be available for several years. A recent decision analysis [15] concluded that in the majority of scenarios. early surgery is preferred to watchful waiting for patients with AAAs less than 5 cm in diameter. Reported prevalence data use 4 cm to divide small and large aneurysms. Prevalence data using 5 cm as the dividing line between large and small aneurysms are not available. For this reason we have used 4 cm as the cutoff point above which surgery might be offered. The analysis recognizes that not all patients with an aneurysm larger than 4 cm will be offered or elect to have surgery. Changing the proportion of patients undergoing surgery allows consideration of only operating on larger aneurysms in the analysis (Table 1).

Methods

A systematic review of the literature was done as part of work carried out for the Canadian Task Force on the Periodic Health Examination and the United States Preventive Services Task Force. The resulting background papers were reviewed for accuracy and completeness by outside experts. Data relevant to screening for AAA were obtained from this review (Table 1). A range of possible values is shown for most variables as well as a value considered most probable by the authors. Some of these data can be accepted with a high degree of confidence whereas others are speculative.

The Simulation Model

We used a computer spreadsheet program (Quattro Pro 4.0; Borland International, Inc.; Scotts Valley, California) to simulate possible AAA screening programs for a cohort of 10 000 men ages 60 to 79 years at the initiation of screening and followed for 20 years. The initial distribution of ages in the cohort was presumed to be that of the U.S. male population in this age range: 60 to 64 years, 35%; 65 to 69 years, 29%; 70 to 74 years, 22%; and 75 to 79 years, 14%.

The cohort is presumed to be unscreened until the first year of the simulation. The model simulates several processes affecting this cohort simultaneously over time: onset and growth of AAAs; clinical surfacing of AAAs in the absence of screening; rupture and death from AAAs; aging and death from causes other than AAA rupture or surgical complication; case finding by screening; and repair by elective or emergency surgery.

Members of the cohort are considered to be in one of seven states at any given time: 1) alive without AAA; 2) having an AAA less than 4 cm in diameter that has not been discovered; 3) having an AAA greater than 4 cm in diameter that has not been discovered; 4) having an AAA less than 4 cm that has been discovered; 5) having an AAA greater than 4 cm, previously discovered when less than 4 cm and not re-examined since; 6) having an AAA greater than 4 cm, known to the medical system; 7) dead.

Because we used deterministic calculations in a simplified model, these processes were represented as sequential events occurring each year in the cohort. The status of men in the cohort is represented as transitions among the seven states as the result of the events occurring.

The sequence in each year is presumed as follows:

Day 1. If this is a year in which screening is done, all men without a previously discovered AAA will be screened on this day; elective surgery is considered for persons in whom an aneurysm 4 cm or larger is found. Screening moves men from states 2 and 3 to states 4 and 6. Elective surgery is presumed to return the patient to state 1 (no AAA), if successful, or to state 7 (dead) if not. Not all men with aneurysms greater than 4 cm diameter will be offered or choose to undergo surgical repair; the proportion not having surgery is a parameter in the model.

Day 2. All previously discovered AAAs (except those found on screening yesterday) are examined by abdominal palpation and with ultrasound. Any men who are found to now have an AAA 4 cm or larger are considered for elective surgery. Those that undergo repair may be returned to state 1 or die of operative complications.

Day 3. On this day all interval cases that would surface during the year by means other than screening do, in fact, surface. (We call this process interval case finding to differentiate it from cases found by screening.) Elective surgery is considered for those with AAAs 4 cm or larger, as above.

Day 4. Any AAAs that will rupture during the year do so today. Some proportion of these arrive at a hospital in time for emergency surgery; the rest die. Those undergoing emergency surgery are cured or they die.

Day 364. All deaths from other causes happen on this penultimate day of the year. The fraction dying is computed using age-specific mortality rates for American men.

Day 365. All new AAAs occur on this day. Some fraction of existing AAAs that were less than 4 cm grow to 4 cm or larger (these men advance from state 2 to 3 or state 4 to 5).

This sequence is highly stylized for convenience of programming as these events occur at random times during the year for individuals. The order was selected to favor screening. At each step any costs that are generated are computed and discounted to present value using a discount rate set as a parameter of the program (5% for results reported here). At the end of each simulated year, survivors and those dying of competing causes at year's end each accumulate 1 life-year. These are discounted to present value at the same discount rate. The program can weight life-years by a quality factor to account for living with a known AAA; however, differential weighting was not used for the results reported here.

Results

The program is designed to compute incremental cost-effectiveness of a screening program compared to a baseline program [24]. Table 2 contains computational results using the most probable values in Table 1. The first row of Table 2 shows the life-years that would be accumulated by the cohort in the absence of medical care for AAA. The present value (discounting at a rate of 5%) of the life-years experienced in this case is $85 341 during the 20 years we followed the cohort. (All figures reported from this point on will be for the 20-year period and discounted to present value at this discounted rate; mention of this will be omitted for brevity.)

The purpose of medical care for AAAemergency surgery for ruptured AAAs or elective surgery for AAAs found incidentally or by screeningis to gain life-years compared to no treatment. Because any improvement will be gained at some dollar cost, we will focus on the cost per life-year of gains beyond this baseline. For example, if only emergency surgery for ruptured AAA were available (that is, there is no elective surgery), the cohort would experience 85 457 life-years, with the increment over baseline being 116 life-years. The incremental cost to the cohort is $1 806 505 (for emergency surgery), thus the incremental cost per incremental life-year is $15 573 (Table 2).

Without a screening program, cases surface by rupturing, exhibiting symptoms, or by incidental detection; this cohort accumulates 85 647 life-years. The elective and emergency surgical care and follow-up office visits and ultrasound examinations would cost a total of $6 797 577.

If the cohort were screened once by ultrasound at the start of the first year, then not screened again, the result would be an increment of 57 life-years at an additional cost of $2 368 348, as shown in Table 2. Thus a screening program consisting of one screening ultrasound examination in the first year would gain additional life-years for the cohort at a rate of $41 550 per life-year.

To evaluate the benefit of repeated screening, we used the simulation to compare a protocol consisting of an initial ultrasound scan plus another scan 5 years later for all men in the cohort who were alive and did not have previously diagnosed and untreated AAAs to the program described above. The second ultrasound 5 years later gains 1 life-year for the entire cohort compared to the one-time screen and accrues additional costs of $906 789. The additional screen at 5 years thus gains life-years at a cost of $906 789 per life-year.

Table 3 shows similar results for a different screening protocol. Here we assume men are screened with a physical examination, costing $20, and a follow-up ultrasound, costing $150, if results of palpation are positive. The assumed sensitivity of physical examination is 35% (Table 1); the specificity of the protocol is 100% (that of ultrasound). If we assume a 5.4% prevalence of AAA, we can calculate the number of true- and false-positive screens; hence the average cost to each of the 10 000 men in the cohort is $27.09 per screen. This protocol of abdominal palpation for the initial screen followed by ultrasound, if positive, gains an additional 20 life-years at an incremental cost of $28 741 per life-year under the most probable case assumptions. The incremental cost of a life-year gained by a second screen with the same protocol after 5 years is $746 752.

Table 4 displays similar calculations using the parameter values most favorable and least favorable to screening compared to the most probable case values. If all the parameters in Table 1 most favorable to screening were to be simultaneously true, then the one-time screen with ultrasound gains life-years at a rate of $5389 per life-year, and the additional screen at 5 years gains a few more life-years at a cost of $154 932 each. The comparable costs for the protocol of physical examination confirmed by ultrasound are $3739 and $54 999 per life-year. If all the parameter values least favorable to screening were to occur simultaneously, a screening program loses slightly more life-years due to elective surgical deaths than it gains by avoiding AAA deaths.

Discussion

This analysis has several limitations. 1) Much of the data are uncertain. For example, there are no direct data describing the percentage of aneurysms discovered by incidental case finding in the absence of formal screening. If this percentage is greater than the 33% used in our analysis, the benefit of screening is less. 2) We assume ultrasound is 100% sensitive for AAA. It is certainly close to 100% sensitive, but any decrement less than 100% will decrease the benefit from screening. 3) We did not consider complications other than death from surgery for AAA. As noted in the RAND review [2], other complications do occur, but time-based probabilities from population-based cohort studies have not been reported. It would thus be very difficult to factor these complications into the analysis. 4) We assumed that life expectancy for a person with successful surgical repair was the same as that of the general population of the same age. In fact many persons with AAA have co-existing coronary artery disease that shortens their life expectancy.

Most of these limitations err toward increasing the potential benefit of screening for AAA. Thus our analysis should be viewed as being optimistic toward screening. The true benefit may be less than we calculate. The results show the greatest benefit (190 life-years) accrues from elective surgical treatment of incidentally discovered aneurysms. A single screening by ultrasound has a modest additional benefit (57 life-years) gained at significant cost per life-year. The protocol of abdominal palpation followed by ultrasound is more cost-effective than screening by ultrasound alone but gains only 20 life-years. Ironically, as shown in Table 2, the most cost-effective approach from a purely financial standpoint is to only offer emergency surgery because it would be offered only to persons who would otherwise die from ruptured AAA. No costs are incurred treating persons who would have done well anyway. Thus even though emergency surgery has a comparatively low survival rate and high unit cost, it is more cost-effective than the other preventive interventions.

A legitimate question to ask is what is the ideal age to screen for AAA? rather than screening everyone in a 20-year age range. The precise answer to this question is not known because it depends on the prevalence of AAA at each age. Our model does allow simulating specific 5-year age groups. If, for example, we screen men aged 60 to 64 years by abdominal ultrasound using the most probable values from Table 1 (including prevalence), 71 life-years are gained at a cost of $31 913 per life-year. However, the prevalence is probably lower in this younger age group. If we use a lower prevalence rate of 2% for aneurysms less than 4 cm in diameter and 1% for aneurysms greater than 4 cm, then screening by ultrasound gains 33 life-years at a cost of $57 136 per life-year. In older cohorts, competing causes of mortality increase and a higher prevalence of AAA is necessary to make screening cost-effective.

How much is too much to spend to save life-years? A recent review of the cost-effectiveness of cholesterol-lowering programs notes that there are health intervention programs with public support costing less than $40 000/life-year but programs costing more than $60 000/life-year appear controversial because of their expense [25]. Using our most probable assumptions, a one-time screening program of physical examination, followed by ultrasound if results are positive, falls into the favorable range. One-time screening with ultrasound of men aged 60 to 80 years from the general population has a less favorable cost-effectiveness ratio, bordering the high end of this range. Both ratios become more favorable if screening is targeted to a higher prevalence population such as men with known coronary or peripheral vascular disease.

Several important questions regarding the detection and treatment of AAA are currently uncertain and need further research. Better data on the prevalence of AAA in the general population and the annual incidence of new aneurysms in a previously screened population are needed. The percentage of aneurysms discovered by incidental case finding is uncertain. The sensitivity and specificity of abdominal palpation by primary care providers for the detection of AAA has been studied on only a limited basis. Research is ongoing to determine the optimal size aneurysm for consideration of elective surgery [2].

Our most probable case analysis suggests that a single screen for AAA in men aged 60 to 80 years either by ultrasound scan or abdominal palpation might be considered cost-effective although of modest benefit. Repeated screening, however, is not cost-effective. This conclusion must be interpreted with caution due to the uncertainty and variance of the data on which it is based. Screening for AAA can be made to look very attractive or harmful depending on the data used in the screening model. More definitive studies are needed before a recommendation that physicians routinely screen men ages 60 to 80 years for AAA can be justified.

This paper is article 54 published at the Madison Veterans Affairs Geriatric Research and Education Clinical Center, Madison, Wisconsin.