Screening for abdominal aortic aneurysms Malcolm Law Department of Environmental and Preventive Medicine, Wolf son Institute of Preventive Medicine, St Bartholomew's and the Royal London Hospital, London, UK

Ruptured aneurysm of the abdominal aorta is a common preventable cause of death, accounting for 2% of all deaths in men over 60 years of age. Population screening could prevent such deaths. Aortic diameter (which can be measured accurately on ultrasound) is a strong predictor of the risk of rupture, which is about 17% per year with aortic diameter £ 6 cm, but below 0.5% per year with aortic diameter < 5 cm, with uncertainty regarding risk in the range 5.0-5.9 cm. Adopting an aortic diameter cut-off of 6.0 cm, the detection rate is estimated to be 86% (that is, 86% of all men who would rupture an aortic aneurysm could be identified and offered surgery) and the false positive rate only 0.6% (that is, 0.6% of men who would not rupture an aortic aneurysm would be so identified). In men with aortic diameter 2 6 cm, the risk of rupture of 17% per year greatly outweighs the peri-operative mortality of about 5%. A national screening programme for men over 60 years of age could prevent 2000 deaths per year and should commence. Uncertainty remains regarding the frequency with which men with smaller aneurysms should be re-examined and the value of intervention among those with an aortic diameter of 5.0-5.9 cm, but the screening programme itself would generate data to help resolve these issues.

Ruptured aneurysm of the abdominal aorta is a common preventable cause of death (Table 1), with 4940 deaths in men and 2062 in women in England and Wales in 1996 - 2% of all deaths in men aged 60 years and over1. The disorder is a candidate for a population screening programme and, in this paper, the effectiveness of screening is assessed. Correspondence to: Dr Malcolm Law, Department of Environmental and Preventive Medicine, Wolfson Institute of Preventive Medicine, St Bartholomew's and the Royal London Hospital, College of Medicine and Dentistry, Charterhouse Square, London EC1M 6BQ, UK

Natural history For the abdominal aorta to rupture, an aneurysm must first form and grow to a critical size - the process has been compared to a blow out in an inner tube2. It is the section of the abdominal aorta below the renal arteries that is prone to aneurysmal enlargement, but the aneurysm may extend above the renal arteries or down into the common iliac arteries. The average rate of expansion of abdominal aortic aneurysms is relatively slow, but (like a balloon) the rate of expansion increases with

British Medical Bulletin 1998;54 (No. 4): 903-913

O The British Council 1998

Screening

Table 1 The numbers and rates of deaths attributed to ruptured abdominal* aortic aneurysm by age and sex, England and Wales 1996'

No. of deaths (% total) 80 (69) -

-

-

-

-

-

-

49-87 (69) 62-92 (75) 59-80 -

1/26 0/20

1/14 0/3 0/9

0/85 0/52 1/148 0/30 1/231 0/128 0/140

1/18 0/12 0/13 0/45 0/10 1/96 0/12 0/96 0/31 0/23

2/10 19/37 3/6 3/7 1/6 9/46*

0.15% (0.10-0.46%)

0.24% (0.10-0.95%)

1.4% (0.22-10.2%)

on

65-74 46-92 (70) 54-84(73) 65-80 65 (65) 65-75

Summary estimate of incidence per yeart (95% confidence interval)

12/56* 0/2

2.8 2.8 2.4 3.1 1.8 3.6 2.7 2.8 2.7 2.8 1.0 1.1

17% (12-23%)

*These were a 5.0 cm; they are excluded from the summary estimate. tCalculated from weighting the incidence in each study by its duration.

Table 3 Summary data from 12 studies (tabulated previously115) that identified all cases of ruptured aortic aneurysm in defined communities by combining data from hospital admissions and coroners' autopsies Total number

2019(100%)

Reached hospital

1253 (62%)

Operated on

720 (36%)

Survived surgery

318(16%)

Aortic aneurysms are associated with smoking (the risk in smokers is about 4 times that in non-smokers), blood pressure (interquartile relative risk about 2-fold) and lipids14"19. The pathology of the disease is incompletely understood - aneurysms are generally associated with areas of the aorta affected by atheromatous disease, but destruction of elastin within the aortic wall is also important. However, the widely advocated means of preventing cardiovascular disease in general - not smoking and reducing lipids and blood pressure - would also be expected to reduce mortality from abdominal aortic aneurysm. Deaths from ruptured abdominal aortic aneurysms must be undercertified, since sudden or unattended deaths are not invariably subject to autopsy. The well-recognised increase over time in the death rate from this cause in England and Wales (Fig. 1) may be due to a decline over time in under-certification. It may also be due to a true increase in incidence20, though the decline over the same period in the British Medical Bulletin 1998;54 (No. 4)

905

Screening

900 800 c

700 -

O

E 600 u. 500 Q.

« 400 o DC 300 Fig. 1 Mortality ascribed to ruptured abdominal aortic aneurysms in men aged 65-74 years, England and Wales 1951-1996.

200 1001950

I 1 960

I

1970

1980

1990

I 2000

Year

prevalence of smoking and the average blood pressure, two major causes of the disorder, weigh against this interpretation.

The screening procedure As with many diseases, age and sex are important as an initial screening enquiry. The disorder is predominantly one of men: age specific death rates in men correspond to those in women 20 or more years younger (Table 1). Of the 4940 deaths in men, 97% occurred in men aged 60 years and over. Screening might, therefore, be limited to men, and to men aged 60 years and over. Age and sex apart, maximum aortic diameter is the principal screening variable. This is the greater of the antero-posterior and transverse diameters, measured by ultrasound scanning of the abdominal aorta. This test is safe, free of discomfort, repeatable to 0.5 cm, and requires equipment which is portable21"24. It does not usually identify aneurysms as bulges, but a maximal external infrarenal diameter in the anteroposterior or transverse planes of i> 3.5 cm in men indicates aneurysmal enlargement with certainty and 3.0-3.4 cm with high probability; uncertainty occurs in the range 2.5-2.9 cm25. Patients with aortic diameters ^ 3.0 cm are, therefore, followed as further expansion can be expected. Screening programmes have operated successfully in various localities in England. Table 4 summarises published data on these programmes. The rate of acceptance of the invitation is high - generally above 75%, and the aorta can be visualised on ultrasound scanning (allowing a measurement of the aortic diameter) in 97-100% of men. 906

British Medical Bulletin 1998.54 (No. 4)

Screening for abdominal aortic aneurysms

Table 4 Population screening programmes in five localities in England Locality

Gloucester13

Age (years)

No invited

Attended*

Aorta visualised on scanning

65

5,337

4,232 (79%)

4,229(100%)

Oxford 25

65-74

843

426(51%)

426 (100%)

Birmingham 1 *-"

60-75

13,000

10,061 (76%)

Northumberland"

65-79

800

Chichester'2-"

65-80

14,057

628 (79%) 8,944 (64%)t

9,771 (97%)

Aortic diameter > 3.0cm men women 167(3.9%) 20 (4.7%) 706 (7.2%)t

612 (97%)

40(6.5%)

8,806 (98%)

294(7.5%)

62(1.3%)

'Letters sent to wrong address, illness and known aortic aneurysm account for some instances of failure to attend. tlncreasing over time; recent proportions are (men/women) 81%/73% at age 65 years, declining with age to 66%/58% at age 76-80 years29.

In men judged at sufficiently high risk of rupture of the aorta, the remedy is an elective surgical insertion of a prosthetic aorta. The perioperative mortality of this operation is about 5% on average5*21"28. It varies according to centre, being between 1-3% in the most experienced centres12-14-26'27'29'30. The most representative study is one of all surgical cases in The Netherlands in 1990, in which the overall peri-operative mortality was 6.7%, increasing markedly with age (2%, 4%, 10% and 17% in persons in their 50s, 60s, 70s and 80s)28. Peri-operative mortality will be lower in a screening programme than in clinical practice because sick individuals are less likely to attend for screening. During a 5 year screening study29, mortality from all causes was almost twice as high in those who did not accept screening as in those who did (19.0% versus 10.4%), indicating that sick people tend to decline screening. Conceptually, the aim of screening is not to identify all aneurysms but only those that will rupture. The prevalence in men of aortic diameter £ 3.0 cm diameter is about 5% at age 65 years and 9% at age 75-80 years12"14'27"29, but screening would be inefficient if it directed all these men to medical attention when only a minority will rupture (2% of men die from this cause). Growth rates are slow (see preceding section) and most men die from other causes before their aneurysm is sufficiently large to be at significant risk of rupture. If the remedy was simple and safe it might be offered to all men with detectable aneurysms, even the smallest. In view of the 5% peri-operative mortality intervention must be limited to men in whom the risk of imminent rupture is judged sufficiently high to warrant this risk. The objective of screening is to select a 'cut-off value of aortic diameter that identifies a large proportion of persons who will rupture the aorta (high detection rate) while intervening in a small proportion who will not (low false positive rate). British Medical Bulletin 1998,54 (No. 4)

907

Screening

Estimating detection rates and false-positive rates Data such as those in Table 2 are insufficient to assess the value of a screening programme. It is necessary to estimate the detection rate and the false-positive rate associated with a specific cut-off value of aortic diameter. This requires knowledge of the distribution of aortic diameter in subjects with (affected) and without (unaffected) a ruptured aortic aneurysm. In affected subjects, estimates of the distribution of maximum aortic diameter in cases of ruptured aortic aneurysm are available from two sources - patients presenting for emergency surgery for acute rupture (measuring aortic diameter with callipers), and studies reporting the distribution from measurements taken at autopsy in patients who died of acute rupture. Data are available from three surgical studies and two autopsy studies; their results (similar in the two types of study) have been summarised2 (the average aortic diameter at rupture is about 8.5 cm). From this distribution, it is simple to calculate the cumulative distribution of the percentage of affected subjects with aortic diameter greater than or equal to specified values. This is shown in Table 5 as the detection rate. The distribution of maximum aortic diameter in unaffected subjects can be determined from ultrasound surveys in men of similar age distribution to that of the affected subjects in the studies used to determine the detection rate2. From this, the percentage of unaffected subjects with aortic diameter greater than or equal to specified values is shown in Table 5 as the false-positive rate. The distributions of aortic diameter may underestimate (by 2-3 mm) the true measurement in both affected and the unaffected individuals - in the affected because the measurements were made at zero blood pressure when aneurysms collapse slightly8, and in the unaffected because measurements on ultrasound are slightly lower than on computed tomography or at operation23-24. The estimates of detection rate and false positive rate are based on the difference between the two distributions, so the similar measurement error in both will not affect the estimate of screening performance. Table 5 shows that, given a policy of offering surgery to all men with aortic aneurysms of diameter of 6 cm or greater, about 86% of all men who would rupture an aortic aneurysm could be identified and offered surgery (detection rate 86%). Only 0.6% of the men who would not have ruptured an aortic aneurysm would be offered surgery (falsepositive rate 0.6%). The ratio of the two, 143, is the likelihood ratio that is, the 'concentrating power' of the screening test or the risk in screen positive men relative to that in the general population. Multiplying this likelihood ratio by the mortality rate provides an estimate of the risk of rupture over the next year in screen positive men. This is 18% - remarkably close to the estimate of 17% from prospective 908

British Medical Bulletin 1998;54 (No. 4)

Screening for abdominal aortic aneurysms

Table 5 Estimated screening parameters, and risk of rupture calculated from them 2 Diameter of 'screen-positive' aorta (cm) £3.0 £ 4.0 a 5.0 £6.0 £7.0

Detection rate*

False positive ratet

Likelihood ratio

(i)

(ii)

(i)/(ii)

100% 100% 97% 86% 68%

7.4% 2.7% 1.1% 0.6% 0.2%

14 37 88 143 340

Risk of rupture in 1 year*

1.7% 4.6% 11% 18% 42%

'Proportion of all subjects with a ruptured abdominal aortic aneurysm who are identified by the cut-off. tProportion of all subjects without a ruptured abdominal aortic aneurysms who are identified by the cut-off. ^From the likelihood ratio multiplied by the annual mortality in men aged 65-79 years, increased to take account that 16% of ruptures are not fatal (Table 3).

studies shown in Table 2. There is uncertainty, however, about the relative value of intervention when aortic diameter is in the range 5.0-5.9 cm. The most rational screening policy, therefore, would use a conservative intervention policy initially - say aortic diameter > 6.0 cm (this may later be revised). The risk of rupture is about 17% in the next year or 50% in the next three years - an order of magnitude greater than the perioperative mortality after elective surgery of about 5%5>21>28. The prevalence of large aneurysms increases with age, but no major influence of age on the risk of rupture for a given aortic diameter is recognised. A randomised trial has confirmed fewer deaths from, or emergency operations for, ruptured aortic aneurysm in 3205 men offered screening than in 3228 men not offered screening (9 versus 20; P = 0.06)29. In women, the number of events was too small to allow a conclusion (3 versus 2). The trial confirms net benefit from early detection in men, but trials necessarily underestimate the size of the benefit. Events in those who declined the screening test must be included in the intention-to-treat analysis. Also, deaths from acute rupture may have been undernumerated in the unscreened group if sudden or unattended deaths were wrongly ascribed to heart disease or other causes; this will not have occurred in the screened group as the presence of an aneurysm was known.

Screening policy A national screening programme should be established. It could save 2000 lives per year (from the number of deaths in England and Wales in British Medical Bulletin 1998,54 (No. 4)

909

Screening

men aged 60-79 years; Table 1) if the detection rate was 86% (Table 5) and 75% of all men accepted screening (Table 4)). Screening might commence at age 60 years and be repeated at age 70 years; alternatively, the number of deaths prevented by scanning each man only once may not be much fewer27. Invitations to attend for an ultrasound scan would be sent to men using a population age-sex register. Those with aortic diameter £ 3 cm would be scanned again at appropriate intervals. There is uncertainty over where to set the cut-off value of aortic diameter for surgery, but this is not a reason to defer the initiation of screening. Screening can be introduced using a high cut-off (6.0 cm) which could be lowered later in the light of data generated by the screening programme and other studies. The coordinators of the screening programme would be notified of all deaths and emergency hospital admissions due to ruptured aortic aneurysm so that risk in men with smaller aneurysms could be determined. The appropriate intervals at which aortic diameter should be remeasured in men with small aneurysms is also not known: there are sufficient data to estimate average rate of expansion but insufficient to estimate the fastest rate (say, the 95th centile). This too is not a reason to defer screening; frequent (say 6-12 monthly) ultrasound examinations would be performed until sufficient data were obtained, and then the interval between examinations would be lengthened. Detecting men with smaller aneurysms may also be worthwhile in itself. Patients should be encouraged to stop smoking as this slows the rate of expansion31. Treatment with beta-blockers also slows expansion32-33, but it is uncertain whether the benefit justifies the cost and side effects of treatment. There is concern that screening for abdominal aortic aneurysm may not be worthwhile because of the associated risk of atherosclerotic disease: even after successful elective surgery, susceptibility to heart disease, stroke and other circulatory diseases must shorten life expectancy. This excess risk has been quantified: life expectancy is reduced by only 2 years2. The average number of years of life lost by men who die from ruptured abdominal aortic aneurysms between the ages of 60 and 79 years is 9.1 years; the population average life expectancy in the same age range in England and Wales is 11 years2. Screening to prevent deaths in people over 60 years of age may be perceived as not worthwhile. Routine invitations for breast and cervical screening are discontinued at age 65 years in Britain; it would be inconsistent to terminate two screening programmes at an age when another is commenced. Resolving the issue by value judgements on the quality of life at different ages is invidious; the pragmatic solution is to allow individuals themselves to decide whether the quality of their life justifies a screening test which may extend it. The high uptake of invitations for abdominal aneurysm screening (about 75%, Table 4) confirms that most older people judge that it is. 910

British Medical Bulletin 1998;54 (No. 4)

Screening for abdominal aortic aneurysms

Financial costs Some studies have indicated that the costs associated with screening for abdominal aortic aneurysm are prohibitively high. These high costs have been based largely on inappropriate calculations - screening younger men (in whom the death rate is too low to warrant screening), using a lower cut-off of aortic diameter (and thereby a high false-positive rate), screening that included physical examination by a doctor (which is expensive and adds nothing to the ultrasound measurement), or using American data with relatively high health care costs. Based on British costs of screening and of elective surgery in 1994, and taking into account the savings on hospital resuscitation and emergency surgery for acute aneurysms prevented, the cost per life saved was an estimated £6787, or £746 per year of life saved2.

Endovascular prostheses Use of an endovascular prosthesis (inserted through the femoral artery) has potential advantages over open abdominal surgery22'34"36, including lower peri-operative mortality and morbidity and lower cost. The procedure might permit a less conservative policy on surgery, though at present the proportion of patients in whom conversion to an open abdominal procedure is necessary is too high for this to be considered. The ultimate success of the procedure remains to be determined.

Conclusions A national screening programme for abdominal aortic aneurysms should commence for men over 60 years of age. The benefits outweigh the costs and 2000 deaths in men in England and Wales could be prevented each year. Life expectancy after surgery in these patients is equivalent to that of an average man 2 years younger. References 1 2 3

Office for National Statistics. Mortality Statistics, cause: England and Wales 1996. London: The Stationery Office, 1998 Law MR, Morris J, Wald NJ. Screening for abdominal aortic aneurysms. / Med Screen 1994; 1: 110-5 Klippel AP, Butcher HR. The unoperated abdominal aortic aneurysm. Am J Surg 1966; 111: 629-31

British Medical Bulletin 1998;54 (No. 4)

911

Screening

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

29

912

Foster JH, Bolasny BL, Gobbel WG, Scott HW. Comparative study of elective resection and expectant treatment of abdominal aortic aneurysm. Surg Gynecol Obstet 1969; 129: 1-9 Bernstein EF, Chan EL. Abdominal aortic aneurysm in high-risk patients. Ann Surg 1984; 200: 255-63 Kremer H, Weigold B, Dobnnski W, Schreiber MA, Zollner N. Sonographische Verlaufsbeobachtungen von Bauchaortenaneurysmen. Klin Wochenscbr 1984; 62: 1120-5 Delin A, OhJsen H, Swedenborg J. Growth rate of abdominal aortic aneurysms as measured by computed tomography. Br J Surg 1985; 72: 530-2 Nevirt MP, Ballard DJ, Hallett JW. Prognosis of abdominal aortic aneurysms. N Engl J Med 1989; 321: 1009-14 Collin J, Heather B, Walton J. Growth rates of subclinical abdominal aortic aneurysms implications for review and rescreening programmes. Eur J Vase Surg 1991; 5: 141-4 Guirguis EM, Barber GG. The natural history of abdominal aortic aneurysms. Am J Surg 1991; 162: 481-3 Bengtsson H, Bergqvist D, Ekberg O, Janzon L. A population based screening of abdominal aortic aneurysms. Eur ] Vase Surg 1991; 5: 53-7 Scott RAP, Wilson NM, Ashton HA, Kay DN. Is surgery necessary for abdominal aortic aneurysm less than 6 cm in diameter? Lancet 1993; 342: 1395-6 Lucarotti M, Shaw E, Poskitt K, Heather B. The Gloucestershire Aneurysm Screening Programme: the first 2 years' experience. Eur J Vase Surg 1993; 7: 397-401 Smith FCT, Grimshaw GM, Paterson IS, Shearman CP, Hamer JD. Ultrasonographic screening for abdominal aortic aneurysm in an urban community. Br ] Surg 1993; 80: 1406—9 Bengtsson H, Bergqvist D. Ruptured abdominal aortic aneurysm: a population-based study. / Vase Surg 1993; 18: 74-80 Reed D, Reed C, Stemmermann G, Hayashi T. Are aortic aneurysms caused by atherosclerosis? Circulation 1992; 85: 205-11 Strachan DP. Predictors of death from aortic aneurysm among middle-aged men: the Whitehall study. Br] Surg 1991; 78: 401-4 Grimshaw GM, Thompson JM, Hamer JD. Prevalence of abdominal aortic aneurysm associated with hypertension in an urban population. / Med Screen 1994; 1: 226-8 Doll R, Peto R, Wheatley K, Gray R, Sutherland I. Mortality in relation to smoking: 40 years' observations on male British doctors. BMJ 1994; 309: 901-11 Fowkes FGR, Macintyre CCA, Ruckley CV. Increasing incidence of aortic aneurysms in England and Wales. BMJ 1989; 298: 33-5 Ernst CB. Abdominal aortic aneurysm. N Engl ] Med 1993; 328: 1167-72 Harris PL. Reducing the mortality from abdominal aortic aneurysms: need for a national screening programme. BM] 1992; 305: 697-9 Lederle FA, Wilson SE, Johnson GR et al. Variability in measurement of abdominal aortic aneurysms. / Vase Surg 1995; 21: 945-52 McGregor JC, Pollock JG, Anton HC. The value of ulrrasonography in the diagnosis of abdominal aortic aneurysm. Scot Med J 1975; 20: 133—7 Collin J, Araujo L, Walton J, Iindsell D. Oxford screening programme for abdominal aortic aneurysms in men aged 65 to 74 years. Lancet 1988; 2: 613-5 Holdsworth JD. Screening for abdominal aortic aneurysm in Northumberland. Br } Surg 1994; 81: 710-2 Khoo DE, Ashton H, Scott RAP. Is screening once at age 65 an effective method for detection of abdominal aortic aneurysms? / Med Screen 1994; 1: 223-5 Akkersdijk GJM, van der Graaf Y, van Bockel JH, de Vries AC, Eikelboom BC. Mortality rates associated with operative treatment of infrarenal abdominal aortic aneurysm in The Netherlands. Br ] Surg 1994; 81: 706-9 Scott RAP, Wilson NM, Ashton HA, Kay DN. Influence of screening on the incidence of ruptured abdominal aortic aneurysm: 5-year results of a randomized controlled study. Br J Surg 1995; 82: 1066-70

British Medial Bulletin 1998;54 (No. 4)

Screening for abdominal aortic aneurysms

30 Mutirangura P, Stonebridge PA, Clason AE et al. Ten-year review of non-ruptured aortic aneurysms. Br J Surg 1989; 76: 1 2 5 1 ^ 31 MacSweeney STR, Ellis M, Worrell PC, Greenhalgh RM, Powell JT. Smoking and growth rate of small abdominal aortic aneurysms. Lancet 1994; 344: 651-2 32 Gadowsld GR, Pilcher DB, Rica MA. Abdominal aortic aneurysm expansion rate: effect of size and beta-adrenergic blockade. / Vase Surg 1994; 19: 727-31 33 Leach SD, Toole AL, Stern H, DeNatale RW, Tilson D. Effect of fJ-adrenergic blockade on the growth rate of abdominal aortic aneurysms. Arch Surg 1988; 123: 606-9 34 Marin ML, Veith FJ. Transfemoral repair of abdominal aortic aneurysm. N Engl J Med 1994; 331:26 35 Paaski WP, Laustsen J. Early results of 132 aortic-iliac anerial reconstructions with the new stretch ePTFE vascular prosthesis. Int Angiol 1994; 13: 296-9 36 Chiesa R, Melissano G, Castellano R et al. A new ePTFE stretch graft for aorto-iliac reconstructions. Surgical evaluation and one year follow-up with magnetic resonance imaging. / Cardiovasc Surg 1995; 36: 134-41

British Medical Bulletin 1998;54 (No. 4)

913