Outcome of the Swedish Nationwide Abdominal Aortic Aneurysm Screening ProgramClinical Perspective
Background: A general abdominal aortic aneurysm (AAA) screening program, targeting 65-year-old men, has gradually been introduced in Sweden since 2006 and reached nationwide coverage in 2015. The aim of this study was to determine the outcome of this program.
Methods: Data on the number of invited and examined men, screening-detected AAAs, AAAs operated on, and surgical outcome were retrieved from all 21 Swedish counties for the years 2006 through 2014. AAA-specific mortality data were retrieved from the Swedish Cause of Death Registry. A linear regression analysis was used to estimate the effect on AAA-specific mortality among all men ≥65 years of age for the observed time period. The long-term effects were projected by using a validated Markov model.
Results: Of 302 957 men aged 65 years invited, 84% attended. The prevalence of screening-detected AAA was 1.5%. After a mean of 4.5 years, 29% of patients with AAA had been operated on, with a 30-day mortality rate of 0.9% (1.3% after open repair and 0.3% after endovascular repair, P<0.001). The introduction of screening was associated with a significant reduction in AAA-specific mortality (mean, 4.0% per year of screening, P=0.020). The number needed to screen and the number needed to operate on to prevent 1 premature death were 667 and 1.5, respectively. With a total population of 9.5 million, the Swedish national AAA-screening program was predicted to annually prevent 90 premature deaths from AAA and to gain 577 quality-adjusted life-years. The incremental cost-efficiency ratio was estimated to be €7770 per quality-adjusted life-years.
Conclusions: Screening 65-year-old men for AAA is an effective preventive health measure and is highly cost-effective in a contemporary setting. These findings confirm the results from earlier randomized controlled trials and model studies in a large population-based setting of the importance for future healthcare decision making.
Editorial, see p 1149
Ruptured abdominal aortic aneurysm (AAA) is a common cause of death among elderly men. Early detection by screening with ultrasound and prophylactic surgery in appropriately selected individuals has been established as an effective method to reduce the high mortality and verified in randomized controlled trials.1–5
In Sweden, with a total population of 9.5 million (2011–2012), 400 to 500 deaths in men ≥65 years of age are annually attributed to ruptured AAA.6 Since 2006, a general AAA-screening program, targeting 65-year-old men, has gradually been introduced in Sweden, reaching nationwide coverage in 20157 (Figure 1). The fully implemented program annually invites 60 000 65-year-old men. In the United Kingdom, a nationally coordinated screening program providing a single scan of all men aged 65 years was launched in 2009 and reached nationwide coverage in 2013.8 In the United States, screening for AAA has been delivered through Medicare since 20079; a single scan is offered to men aged 65 to 75 years who have ever smoked. However, because of regulatory restrictions, the use of the program has so far been limited.10 A similar program is offered through the US Department of Veteran Affairs, and various privately funded AAA-screening programs exist.9,11 So far, no nationwide AAA-screening report on the effect on AAA mortality in the population has been published from any country.
The aim of this national collaborative study was to report the outcome of the Swedish AAA-screening program. The Swedish Aneurysm Screening Study group (SASS) was created in 2006 to share experiences and to coordinate scientific projects. This is the first report from the SASS collaboration.
The public healthcare system in Sweden is run by 21 independent counties, each deciding autonomously whether screening should be introduced or not, and how to design the program. With the exchange of ideas and experiences within the medical profession, however, it has been possible to harmonize the design and methodology throughout the country. All men aged 65 years, consecutively identified through the National Population-based Registry, updated every third month, were invited to an ultrasound examination of the aorta, regardless of any known AAAs. According to a detailed analysis from middle Sweden, the vast majority of men with a known AAA did not attend the screening examination, where the prevalence of known AAAs among 65-year-old men was 0.5% and the prevalence of screening-detected, previously unknown AAAs was 1.7%.12 If they did attend, however, the AAA was included in the prevalence estimate. The examination fee varied between the different counties, from €0 to €20. Travel expenses were not reimbursed. The baseline examination includes a single ultrasound scan where the maximum infrarenal anteroposterior diameter is measured according to the leading edge–to–leading edge method13 with the ultrasound transducer longitudinally to the aorta. An AAA is defined as an aortic diameter ≥30 mm.14
A common study protocol was applied for the 21 counties, including data on the annual number of invited and examined men aged 65 years, annual number of screening-detected AAAs and their size, annual number of screening-detected AAAs operated on, the operative technique used, and surgical outcome. The prespecified analysis plan included the following outcome parameters: number of men invited, attendance rate, prevalence, AAA-size distribution, repair rate, repair technique, and repair outcome. It also included a plan to analyze the effect on AAA-specific mortality in the population and to estimate the cost-effectiveness of the program. From the Swedish Cause of Death Registry6 the annual overall AAA-specific mortality rates for all men aged ≥65 years were retrieved for each county for the years 2000 to 2014. The International Classification of Diseases codes used were I71.3 (Abdominal aortic aneurysm, ruptured) and I71.4 (Abdominal aortic aneurysm, without rupture) if classified as causing or contributing cause of death. The corresponding annual population statistics were retrieved from Statistics Sweden.15
The study complies with the Declaration of Helsinki. No individual data were collected, and, according to Swedish law, ethical approval and informed consent are not required for aggregated registry-based data on group levels.
Data from the 21 counties was aggregated, analyzed, and presented on a national level, with the exception of the analysis of change over time in mortality rates, which were calculated separately for each county.
The association between screening and reduction in AAA-specific mortality rate was analyzed by 3 different approaches;
A linear regression analysis was used to estimate the relationship between change in AAA mortality and duration of screening. The first regional AAA-screening program was initiated in 2006. Consequently, the years 2000 to 2005 were used as baseline for AAA mortality rate and were compared with the mortality rate for the years 2011 to 2014, when the potential effect of screening was expected to become evident in this particular population. Counties with equal length of screening were aggregated into common data points. The resulting β-coefficient indicates the annual change in AAA mortality after screening has been initiated. AAA-related mortality rate was calculated as deaths/100 000 men ≥65 years.
Using the same statistics, time trends in AAA mortality rate were compared between 2 groups; those counties that, at the end of the study, had screened ≥6 years, and those counties that, at the end of the study, had screened <4 years. From the MASS trial (Multicentre Aneurysm Screening Study), we know that the effect of screening on AAA mortality in the population gradually increases with longer follow-up, and is judged to become detectable after a minimum of 6 years of screening, but undetectable after <4 years of screening.2 The relative changes in AAA mortality over time for the 2 groups were calculated with 95% confidence intervals, as described by Altman and Bland.16 For comparison, the corresponding relative changes in all-cause mortality, mortality from ischemic heart disease, and mortality from cancer were calculated.
A stepped-wedge cluster randomized trial design was used to estimate the effect of screening 65-year-old men on the incidence of AAA-specific mortality among elderly men (≥65 years). A stepped-wedge design involves the sequential roll-out of an intervention to clusters of participants over time. By the end of the study, all participants will have received the intervention in a random order. The design is suitable for evaluating interventions during routine implementation,17 in particular, community-level public health interventions that have been proven effective in individual-level trials.18 The first empirical example of this design being used is in the Gambia Hepatitis Study,19 and it has since been used to evaluate a number of various interventions.20 The method was also used in one of the first population-based AAA-screening studies in Huntingdon, England.21 In this study, all men ≥65 years in Sweden were consecutively allocated to either the screening group or the control group on the basis of intention to screen in the respective county. Thus, the population in the screening group acts as its own control for the period before being invited to screening. Over the study period, the entire study population was scrutinized for AAA-specific death. AAA deaths were allocated to the control group if the death occurred before screening had started and to the screening group if the death occurred after screening had started. The observation period was measured in person-years, and the AAA mortality rate was calculated per 100 000 person-years of men ≥65 years of age.
The long-term effects were projected by using a previously described and validated Markov model,22 designed to analyze the health-economy of 1-time screening of 65-year-old men. In summary, the model used a lifetime perspective and assumed a screening program up and running in steady state. Actual observed screening-detected prevalence from the national AAA-screening program was used. The incidental detection rate was extrapolated from the MASS trial, corresponding to 25% at 4 years.2 Time-dependent probability of AAA progression to rupture and surgery was based on the MASS study corrected with recent Swedish data.2,23 Contemporary age-specific operative outcome data were retrieved from the Swedvasc registry.23 General population age-dependent all-cause mortalities were based on contemporary population statistics from Statistics Sweden,15 with a relative 5-year survival rate of 95% for individuals surviving elective AAA repair and 90% after surgery for ruptured AAA.24 Sex- and age-dependent quality-adjusted life-years (QALYs) were retrieved from a Swedish EQ-5D health-related quality-of-life population estimate.25 Screening and surveillance costs were based on actual total costs in the national screening program. The cost of surgery and follow-up was retrieved from a recent cost analysis study from Sweden.26 Outcomes in the model were life-years and QALYs gained, incremental cost-efficiency ratio in Euros/QALY (the cost of extending 1 individual’s life with 1 QALY by invitation to screening), relative risk reduction and absolute risk reduction from AAA death, number needed to screen, and number needed to treat to prevent 1 AAA-related death, change in elective and emergency AAA repairs workload, and annual total cost of the screening program. Cost and effects were discounted at 3% annually.
Statistical analysis was performed with IBM SPSS Statistics for Windows, Version 22.0 (IBM Corp), and the Markov model was developed in the TreeAge Pro 2012 Healthcare software package (TreeAge Software).
A total of 312 784 men aged 65 years were invited between 2006 and 2014. One county (Halland), representing 3% of the target population, had incomplete records on numbers screened and numbers repaired, and was excluded from the subsequent analysis. Of 302 957 remaining invited men, 253 896 (84%; 95% confidence interval [CI], 84–84) attended screening between 2006 and 2014 (Figure 1). The uptake was slightly higher in the counties with a fee than when the examination was free: 84% (95% CI, 84%–85%) versus 81% (95% CI, 81%–81%), P<0.001. The total number of men with screening-detected AAAs was 3891, a prevalence of 1.5% (95% CI, 1.5%–1.6%) with an annual detection rate at steady state of ≈650. The proportion of AAAs ≥4 cm was 18%, the proportion of AAAs ≥5 cm was 12%, and the proportion of AAAs ≥5.5 cm was 7%.
One additional county (Jönköping), representing 4% of the population, had incomplete data on AAA repair rate specifically for screening-detected AAAs. Consequently, this population was excluded from analysis regarding AAA repairs, but was included in other analyses. Thus, AAA repair analysis was based on a total of 3787 patients with screening-detected AAAs, of whom 683 (18%) were operated on during the study period: 394 (58%) with open repair and 289 (42%) with endovascular aneurysm repair (EVAR). The cumulative repair rate of screening-detected AAAs per year after screening is displayed in Figure 2 and was 29% after a mean follow-up of 4.5 years. The overall operative 30-day mortality for screening-detected AAAs was 0.9% (95% CI, 0.2–1.6), 1.3% (95% CI, 0.2–2.4) after open repair, and 0.3% (95% CI, 0.0–1.0) after EVAR.
For the total population of men ≥65 years AAA-specific mortality declined from 74/100 000 in the year 2000 to 45/100 000 in the year 2014 (–39%). Comparing AAA-specific mortality between counties having screened ≥6 years (mean, 7.1 years) and counties having screened <4 years (mean, 1.5 years) showed a more prominent decline in mortality during the past 4 years in counties having screened ≥6 years (Figure 3). The relative change in AAA mortality for counties screened ≥6 years versus counties screened <4 years was 0.73 (95% CI, 0.62–0.88; P<0.001). The corresponding relative change in all-cause mortality, mortality from ischemic heart disease, and mortality from cancer were 0.98 (95% CI, 0.96–1.00; P=0.10), 0.93 (95% CI, 0.89–0.97; P=0.002), and 1.03 (95% CI, 0.99–1.07; P=0.22), respectively (Figure 3).
A linear regression analysis of the change in mortality from baseline as a function of the duration of screening is displayed in Figure 4. The resulting β-coefficient of –4.0 (95% CI, –7.2 to –0.8; P=0.020) indicates a mean 4.0% reduction in AAA mortality for each year of screening.
In a stepped-wedge cluster randomized trial design the AAA-related mortality rate was 25% lower in the screening group than in the nonscreening group, relative risk 0.77 (95% CI, 0.72–0.81) (Table, Figure 1).
The predicted long-term effects of this national AAA-screening program in a population of 9.5 million were as follows. The estimated relative risk reduction of AAA-specific death was 40% and the absolute risk reduction was 15 prevented deaths from AAA per 10 000 men invited to screening. The number needed to screen was 667, and the number needed to treat (ie, number of screening-detected AAAs needed to operate on to prevent 1 premature death) was 1.5. The estimated annual number of prevented premature deaths from AAA was 90, and there were 577 gained QALYs. Of the screening-detected AAAs, model predictions indicated that 40% will eventually be repaired, with an annual surgical case load of screening-detected AAAs in men ≥65 years of 360 elective AAA repairs (+109% in comparison with no screening) and 36 ruptured AAA repairs (–59% in comparison with no screening). The incremental cost-efficiency ratio was estimated to be €7770 per QALY. The national annual total cost of the Swedish screening program was estimated to be €12 620 000 in comparison with €8 140 000 (difference + €4 480 000) for the traditional system of the opportunistic detection of AAAs.
This is the first evaluation of a nationwide AAA-screening program on the effect on AAA mortality in the population. The Swedish experience presented here indicates that a population-based AAA-screening program can be implemented in a simple way, at low cost, and that it is well accepted by the population. These contemporary real-world data also suggest that 1-time screening of 65-year-old men for AAA is an effective preventive health measure, and highly cost-effective, which is in agreement with what have been suggested from earlier randomized controlled trials1–5 and recent reviews1,27,28 and model studies.22,29
The improvement in the AAA-specific mortality rate since the implementation of the screening program may also be attributable in part to other factors, including a falling prevalence of the disease, increased use of endovascular repair, better perioperative outcome, and increased life expectancy. However, despite a falling prevalence, contemporary AAA screening in men remains cost-effective, probably by counterbalancing the lower prevalence with improved surgical outcome and increased longevity. In a recent model study, we showed that screening remained cost-effective with prevalence as low as 0.5%.22 This may have implications for populations with a low or decreasing prevalence rate. A recent review of the first 5 years of screening in England reported a similar prevalence but a slightly lower attendance rate than the Swedish experience.8
The use of EVAR was lower in this setting (42%) in comparison with current Swedish general practice (59%).15 This is likely a result of the lower age in the screening group, and the use of EVAR is anticipated to increase to the general level over time. In part, for the same reason, the surgical outcome was excellent (30-day mortality, 0.9%), and similar to the early experience from England.8 Although some reports suggest that screening-detected AAAs have a lower operative mortality than opportunistically detected AAAs,30 we cannot rule out that this also eventually will approach the general perioperative mortality for AAA repair seen in Sweden, at present ≈1.5%.23 A higher EVAR use results in fewer patients considered unfit for repair and lower perioperative mortality. However, EVAR is associated with a higher cost; in a contemporary Swedish setting, the perioperative cost is fairly equal between EVAR and open repair, whereas EVAR has a slightly higher cost during follow-up26. Altogether, the effect of variation in EVAR use on cost-effectiveness for AAA screening is very small.22
The gradual introduction of screening in Sweden from 2006 to 2015 (Figure 1) made it possible to demonstrate a mortality reduction from screening, because populations not yet screened could serve as controls for screened populations when analyzing AAA-specific mortality. A possible general small continuous decline in AAA rupture incidence during the implementation of the screening program, unrelated to screening, is an important potential confounder when evaluating the effect of screening on AAA-specific mortality, especially in a stepped-wedge design. Different statistical methods were used to assess the potential effect on AAA-specific mortality; however, all showed a significant reduction related to screening. In addition, the lack of a similar decrease in all-cause mortality and mortality from ischemic heart disease and cancer (Figure 3) further supports the assumption that screening is a main driver behind the observed major decrease in AAA-related mortality. With today’s low autopsy rate, the cause of death can be difficult to assign accurately. The latest major audit of the Swedish Cause of Death Registry showed that overall the underlying cause was correct in 77% of cases.31 Thus, death caused by the rupture of an unknown AAA could be misclassified as attributable to ischemic heart disease, and death caused by ischemic heart disease in a person with a small AAA could be misclassified as attributable to rupture. Such misclassification, however, would likely cause a small bias toward the cost-ineffectiveness of screening. Another limitation is that we used a model design for long-term predictions. The model predicted a 27% repair rate at 4.5 years of follow-up, consistent with the observation of 29% at 4.5 years in this study (Figure 2), which is also similar to the repair rate of 26% at 4 years in the MASS study.2 Thus, the model was found to be reliable and robust when validated against real-world data, strengthening the validity of the long-term predictions.
When interpreting the reduction in mortality from AAA screening, it is important to bear in mind that the screening was introduced in a population with an already ongoing caseload of preventive AAA surgery based on opportunistic detection. In this study, the doubling of the elective surgery rate secondary to introducing screening was predicted to prevent an additional 90 premature deaths annually. Consequently, the total number of premature deaths prevented annually in Sweden by preventive AAA surgery could be estimated to be 180. The introduction of screening coincides with a falling prevalence of the disease,12 that is, with an overall falling elective and ruptured AAA repair rate. Thus, the increased workload of screening-detected AAAs requiring elective repair is counterbalanced by a decreased elective repair rate in the general population, and the reduction in ruptured AAA repair rate was accelerated by screening.
A fundamental ethical principle is to minimize any harm associated with a screening program. Although AAA screening saves lives by significantly reducing the number of AAA ruptures (–44% in the current report), the subsequent increased number of elective repairs (+109% in the current report) puts patients at risk of perioperative mortality. The results show that this problem is very limited, however, with only 1 to 2 extra men (0.1%–0.3% of all men with screening-detected AAAs) dying of elective repair per year. The reason for this is the exceptionally high efficiency of prophylactic AAA repair. The high mortality associated with rupture combined with a very low elective perioperative risk results in a number needed to treat of only 1.5, that is, 3 individuals with screening detected AAAs have to be electively repaired to prevent 2 premature deaths from AAA. Furthermore, men undergoing prophylactic AAA repair have a large AAAs with high rupture risk, that is, a poor prognosis without surgery. Some authors have claimed that screening for AAA results in significant overtreatment.32 The Swedish data suggest, however, that a substantial proportion of those with a screening-detected AAA significantly benefit from screening. With a 40% repair rate and a number needed to treat of 1.5, at least one-fourth of all patients with a screening-detected AAAs will experience a longer life as a result of attending the screening program.
Until now, the Swedish AAA-screening program has been independently organized by the 21 individual counties. The National Board of Health and Welfare (Socialstyrelsen), a government agency under the Swedish Ministry of Health and Social Affairs, recently decided that all Swedish 65-year-old men should be offered an aortic examination. This recommendation was based on an extensive review of the current evidence by the Swedish Agency for Health Technology Assessment and Assessment of Social Services. Our report further supports this decision.
In conclusion, the screening of 65-year-old men for AAA is an effective preventive health measure and is highly cost-effective in a contemporary setting. These findings confirm results from earlier randomized controlled trials and model studies in a large population-based setting of importance for future healthcare decision making.
Linda Lyttkens, RN, Department of Surgical Sciences, Section of Vascular Surgery, Uppsala University, Uppsala; Ewa Pihl, RN, Department of Surgery, Falun County Hospital, Falun; Tomas Wetterling, MD, Department of Surgery, Kristianstad County Hospital, Kristianstad; Per Kjellin, MD, Department of Surgery Helsingborg County Hospital, Helsingborg; Ken Eliasson, MD, Department of Vascular Surgery, Örebro University Hospital, Örebro; Erik Wellander, MD, Department of Surgery, Länssjukhuset Ryhov, Jönköping; Azin Narbani, RN, Department of Surgery, Visby County Hospital, Visby; Elisabet Skagius, MD and Alexandra Hollsten, RN, Department of Surgery, Sundsvalls County Hospital, Sundsvall; Martin Welander, MD, Department of Thoracic and Vascular Surgery, Linköping University Hospital, Linköping; Toste Länne, MD, PhD, Division of Cardiovascular Medicine, Department of Medical and Health Sciences Linköping University, Linköping; Bibbi Fröst, RN, Department of Radiology, Oskarshamn County Hospital, Oskarshamn; David Korman, MD, Department of Surgery, Östersund County Hospital, Östersund; Sven-Erik Persson, MD, Department of Surgical and Perioperative Sciences, Umeå University, Umeå; Birgitta Sigvant, MD, PhD, Department of Vascular Surgery, Karlstad Central Hospital, Karlstad, and Department of Clinical Science and Education, Södersjukhuset, Karolinska Institute, Stockholm; Thomas Troëng, MD, PhD, Blekinge County Council, Karlskrona, and Department of Surgical Sciences, Section of Vascular Surgery, Uppsala University, Uppsala; Markus Palm, MD, Department of Surgery, Sunderby Hospital, Sunderbyn; Eva Ansgarius, BMA, Department of Physiology, Kullbergska County Hospital, Katrineholm; Nils-Peter Gilgen, MD, Department of Surgery, Mälarhospital, Eskilstuna; Christina Sjöström, RN, and Khatereh Djavani Gidlund, MD, PhD, Department of Surgery, Gävle County Hospital, Gävle; Peter Danielsson, MD, Region Halland, Halland; Adam Bersztel, MD, Department of Vascular Surgery, Västerås Central Hospital, Västerås; Tomas Jonasson, MD, Department of Surgery, Central Hospital, Växjö.
This work was done on behalf of the Swedish Aneurysm Screening Study group, and the authors acknowledge the collaborators.
Sources Of Funding
Sources of Funding, see page 1147
Circulation is available at http://circ.ahajournals.org.
- Received March 2, 2016.
- Accepted August 5, 2016.
- © 2016 American Heart Association, Inc.
- Ashton HA,
- Buxton MJ,
- Day NE,
- Kim LG,
- Marteau TM,
- Scott RA,
- Thompson SG,
- Walker NM
- Lindholt JS,
- Juul S,
- Fasting H,
- Henneberg EW
- Norman PE,
- Jamrozik K,
- Lawrence-Brown MM,
- Le MT,
- Spencer CA,
- Tuohy RJ,
- Parsons RW,
- Dickinson JA
- 6.↵Swedish Cause of Death Registry. http://www.socialstyrelsen.se/register/dodsorsaksregistret. Accessed December 31, 2015.
- Lederle FA
- Svensjö S,
- Björck M,
- Gürtelschmid M,
- Djavani Gidlund K,
- Hellberg A,
- Wanhainen A
- McGregor JC,
- Pollock JG,
- Anton HC
- 15.↵Statistics Sweden. www.scb.se. Accessed December 31, 2015.
- Altman DG,
- Bland JM
- Mdege ND,
- Man MS,
- Taylor Nee Brown CA,
- Torgerson DJ
- 19.↵The Gambia Hepatitis Study Group. The Gambia Hepatitis Intervention Study. Cancer Res. 1987;47:5782–5787.
- 23.↵Swedvasc. The Swedish National Registry for Vascular Surgery. www.ucr.uu.se/swedvasc. Accessed December 31, 2015.
- Mani K,
- Björck M,
- Lundkvist J,
- Wanhainen A
- Mani K,
- Björck M,
- Lundkvist J,
- Wanhainen A
- Linné A,
- Smidfelt K,
- Langenskiöld M,
- Hultgren R,
- Nordanstig J,
- Kragsterman B,
- Lindström D
- Johansson LA,
- Björkenstam C,
- Westerling R
What Is New?
This first evaluation of the Swedish nationwide abdominal aortic aneurysm–screening program indicates that a population-based abdominal aortic aneurysm-screening program can be implemented in a simple way and at low cost, and that it is well accepted by the population. These contemporary real-world data also suggest that 1-time screening of 65-year-old men for abdominal aortic aneurysm is an effective preventive health measure and highly cost-effective from a public health perspective.
What Are the Clinical Implications?
These findings confirm results from earlier randomized controlled trials in a large population-based setting of the importance for future healthcare decision making.