The flow of participants in the trial is as reported previously,5
except for two features. Firstly, of the 1334 men with abdominal aortic aneurysm detected at initial scan, 72% (n=963) had complete clinical follow-up to 10 years according to the protocol; this compares with 76% at seven years. Secondly, inability to follow up deaths because some men may have moved was 2.7% at 10 years, compared with 2.1% at seven years; these men were censored at the time they were last known to be alive.
Overall, 155 deaths related to abdominal aortic aneurysm (absolute risk 0.46%) occurred in the invited group compared with 296 (0.87%) in the control group, a relative risk reduction of 48% (hazard ratio 0.52, 95% confidence interval 0.43 to 0.63; table 1). The benefit seen in earlier years of follow-up was maintained in the later years of follow-up, with continued divergence of the cumulative curves of deaths related to abdominal aortic aneurysm in the two groups (fig 1). The mean age at death was similar in the invited and control groups (75.0 v 75.4 years). Non-fatal ruptures of abdominal aortic aneurysms were also about halved in the invited group (table 1). Twenty one men in the invited group died within 30 days of elective surgery, and another six men after more than 30 days. Additionally, despite being invited for screening, 170 men subsequently had a ruptured aneurysm. Many of these were, however, excluded from the potential benefit of screening—for example, among those who did not attend screening, those who did not keep an outpatient appointment, those who refused surgery, and those who were considered unfit for surgery (table 2).
Table 1 Deaths related to abdominal aortic aneurysm*, ruptured abdominal aortic aneurysm, and other causes of death
Fig 1Cumulative deaths related to abdominal aortic aneurysm, by time since randomisation
Table 2 Timing of incidence of ruptured abdominal aortic aneurysm and deaths in 33 883 men aged 65 or more invited to screening
Among those who could potentially benefit from screening, some aneurysms ruptured between recall scans, pending a decision about surgery and while awaiting surgery (table 2). Twenty five ruptures also occurred after the men had had normal initial scans, of which 19 were fatal. The rate of these ruptures increased noticeably after eight years of follow-up (fig 2). In years 8, 9, and 10 (when censoring impacts on the follow-up data available) six, six, and three ruptures occurred, respectively, with corresponding rupture rates per 10
000 person years of 3.0, 3.8, and 5.7. Time since initial scan, rather than age, was the main determinant of this increased risk of rupture.
Fig 2Rate of ruptured abdominal aortic aneurysms (number of ruptures in brackets) in men originally screened as normal, by time since randomisation. Three more ruptures were recorded in the limited follow-up after 10 years
Over the 10 years 552 elective operations took place in the invited group and 226 in the control group. The respective 30 day mortality rates of 4% (21/552) and 6% (13/226) were not significantly different (P=0.23). Sixty two men underwent emergency surgery in the invited group compared with 141 in the control group. The respective 30 day mortality rates of 29% (18/62) and 36% (50/141) were not significantly different (P=0.37). Nearly all the operations in MASS were open surgical repairs, with endovascular repair occurring only in the later period of follow-up. Two endovascular repairs were undertaken as emergency procedures (both patients died within 30 days) and 68 as elective procedures, representing 9% (68/778) of all elective operations. The 30 day mortality rate for elective endovascular repair was 3% (2/68).
The benefit of a screening programme is diluted by those who, despite being invited, do not attend: the unbiased estimate of the reduction in deaths related to abdominal aortic aneurysm among men who were screened was 60% (hazard ratio 0.40, 95% confidence interval 0.32 to 0.50). This estimate is relevant when providing information to individuals about the benefit of screening or when considering the benefit from a screening programme that achieves an attendance rate different to the 80% achieved in MASS.
Total mortality at 10 years was about 30% in each group (table 1). Because deaths related to abdominal aortic aneurysm comprise about 2% of all deaths and there were no clear differences in any other causes of death, only a small difference was found in all cause mortality (hazard ratio 0.97, 0.95 to 1.00). Although 124 fewer deaths from ischaemic heart disease occurred in the invited group, this difference was not statistically convincing (P=0.06), and the mean age of the men who died was 74.7 in both groups. These findings do not suggest any major general differences in health care between the groups as a result of screening.
The costs per person were greater in the invited group (through the costs of screening and more elective surgery but offset by fewer emergency operations), by an average of £100 (table 3). The extent of reduction in number of deaths related to abdominal aortic aneurysm in the invited group led to an estimated incremental cost effectiveness ratio of £7600 (95% confidence interval £5100 to £13
000) per life year gained over the 10 years of the trial.
Table 3 Discounted mean costs and effects per person, based on 10 year follow-up in Multicentre Aneurysm Screening Study (MASS)