Summary of Results

Four large, population-based screening RCTs of men age 65 years or older examined the effectiveness of one-time AAA screening and found that AAA prevalence varies from 4 to 7.6 percent, with the majority of screen-detected AAAs being small in size (aortic diameter of ≤4 to 4.5 cm).^{12, 15, 113, 143, 147, 155, 170} One more contemporary screening trial in the same population solely contributed to outcomes of prevalence and number of operations and reports a 3.3 percent prevalence, reflecting a temporal decline in the disease.¹⁴⁶ The invitation for screening among men age 65 years or older was associated with a 35 percent reduction in AAA-related mortality, a 38 percent reduction in AAA rupture rate, and a 43 percent reduction in the number of emergency surgeries. There was no statistically significant difference, however, in all-cause mortality at 12- to 15-year followup.

Study Characteristics

Two fair- and two good-quality population-based screening RCTs from the United Kingdom, Denmark, and Australia assessed the efficacy of AAA screening in population-based settings: the Multicentre Aneurysm Screening Study (MASS) (n=67,770);^{12, 134, 135, 170, 171} the Chichester, United Kingdom, screening trial (n=15,382);^{13, 36, 113, 173} the Viborg County, Denmark, screening trial (n=12,639);^{14, 143–145, 147} and the Western Australia screening trial (n=38,480) (Appendix E Tables 1 and 2).^{7, 15, 154, 155, 168} These four RCTs were included in the previous review, with new long-term data reported in the Western Australia trial¹⁵ included in this update. One additional population-based screening trial in Denmark (VIVA) is discussed in detail under KQ3 but is mentioned here due to its contribution to AAA prevalence in a screened population and number of operations.¹⁴⁶ All trials identified potential participants age 64 or 65 years or older from population registries or regional health directories. The MASS trial identified participants from four centers in the United Kingdom; the Chichester trial included nine general practices in Chichester; the Viborg trial included the population from Viborg County; and the Western Australia trial included participants from a capital city and satellite towns. Reported mean (or median) ages ranged from 67.7 to 72.6 years, and the oldest study participants were age 83 years. One study, the Chichester trial,¹⁴ included women,²⁵ while the other three recruited only men. Other than age and sex, no studies reported outcomes in the screened and control groups by any other demographic information.

Two trials provide some risk factor information.^{15, 147} The Viborg trial, which described AAA-related comorbidity risk factor information from hospital discharge data, indicated that 26.5 percent of all participants had at least one cardiovascular risk factor or chronic obstructive pulmonary disease (COPD).^{145, 147} The Western Australia trial reported cardiovascular comorbidity and risk factor information for the screened group and analyzed the association between the risk factor and AAA diagnosis, but risk factor data were not collected for the control group, nor were they linked to mortality outcomes.¹⁵ Three studies had no trial exclusions; only the MASS trial excluded patients who (1) were identified by their primary care clinicians as too high risk to be screened, (2) were terminally ill, or (3) had other serious health problems or prior AAA repair.

All trials randomized participants to one of two groups: the invited group received a letter invitation for one-time ultrasound screening, while the control group received usual care. All trials considered “normal” aortic diameter to be smaller than 3 cm and defined AAA as an aortic diameter of 3.0 cm or larger. Three of the RCTs (MASS, Viborg, and Chichester) further prescribed specific postscreening surveillance protocols for AAAs with an aortic diameter of 3.0 cm or larger with repeat ultrasounds,^{113, 147, 170} while one trial (Western Australia) sent initial ultrasound results to primary care clinicians for management.¹⁵ In the MASS trial, patients with aortic diameters of 3.0 to 4.4 cm were rescanned yearly, those with aortic diameters of 4.5 to 5.4 cm were rescanned at 3-month intervals, and those with aortic diameters of 5.5 cm or larger were urgently referred to a vascular surgeon.¹⁷⁰ In the Viborg trial, individuals with an ectatic aortic size of 2.5 to 2.9 cm were offered a repeat scan at 5 years, those with an aortic size of 3.0 to 4.9 cm were offered annual scans, and those with an aortic size of 5.0 cm or larger were referred to vascular surgery.¹⁴⁷ In the Chichester trial, patients with AAAs with an aortic diameter of 3.0 to 4.4 cm were rescanned annually, those with AAAs with an aortic diameter of 4.5 to 5.9 cm were rescanned every 3 months, and those with AAAs with an aortic diameter of 6 cm or greater were referred to vascular surgeon, as were those with increase of diameter of 1 cm or more per year (Appendix E Table 1).¹¹³

The primary outcome reported in trials was AAA-specific mortality (defined as all AAA-related deaths plus all deaths within 30 days of AAA surgical repair); all four trials also reported AAA rupture rate and all-cause mortality as benefit outcomes. Mortality data and causes of death were ascertained from death certificates in all studies, and three of the RCTs additionally involved an independent blinded review of autopsy reports and/or hospital records for all AAA-related deaths. Mean or median followup in these four population-based screening trials ranged from 12.8 to 15 years, with short-term results published at 3- to 5-year intervals; this report focuses solely on the longest-term followup. Local and national health departments, research councils, and heart foundations funded these studies.

The MASS trial^{12, 134, 135, 170, 171} stands out as the highest quality of the four trials; it had the greatest number of participants, the highest adherence to screening, and clear reporting of randomization, allocation, blinding of outcome assessors, and confirmation of equal followup in the invited and control groups. All trials appeared to use intention-to-treat analysis; adherence to screening varied from the lowest adherence in the Western Australia trial (62.5% of those invited attended screening) to the highest adherence in the MASS trial (80.2% adherence). Three studies reported low loss-to-followup rates in participants with AAA (MASS trial, 70% retention rate at 13-year followup in men with an AAA detected at the initial scan;¹⁷⁰ Viborg trial, 75.1% retention rate in invited group and 58.0% in control group at 52-month followup;¹⁴⁷ and Western Australia trial, 87.1% retention in invited group and 84.9% in the control group at 3.6-year followup).¹⁵

Detailed Results (Male Only; Female Results in KQ1a)

AAA Prevalence in Screened Population

AAA prevalence (aortic diameter of ≥3.0 cm) on the initial screen for male attenders in the four population-based screening trials varied from as low as 3.9 percent in the Viborg trial¹⁴⁷ to as high as 7.6 percent in the Chichester trial¹¹³ (Table 1). Notably, the Chichester and Western Australia trials reported the highest AAA prevalence rates and they recruited older participants (Chichester median age, 72 years; Western Australia mean age, 72.7 years compared to mean ages of 67.7 and 69.2 years in the Viborg and MASS trials, respectively). The VIVA trial¹⁴⁶ reported a prevalence of AAAs (619/25,078 [3.3%]) similar to the older Viborg trial¹⁴⁷; both trials were conducted in the same geographic area. Four of the five trials (MASS, Chichester, Western Australia, and VIVA)^{12, 13, 15, 113, 143, 146, 155, 170} reported the prevalence of AAA by size at initial screening, with MASS, VIVA, and Western Australia trials reporting that the majority of AAAs (87% to 93%) detected were small (measuring <5.5 cm). The overall prevalence of large AAAs (≥5 cm or ≥5.5 cm) in the screened population was consistent across studies and was reported as 0.3 to 0.6 percent (Appendix E Table 3).

Table 1

AAA Prevalence, Rupture, and Surgery Data for One-Time Screening Trials (KQ1).

Effect of Population Screening on All-Cause and AAA-Related Mortality

A meta-analysis of all-cause mortality of the four screening trials^{15, 113, 147, 170} (N=124,929) using relative risk (RR) estimates with the DerSimonian and Laird method showed a result that when rounded, yielded a nonstatistically significant pooled result (RR, 0.99 [95% CI, 0.98 to 1.00]; I²=0%) (Figure 2). None of the individual screening trials reported a statistically significant reduction in all-cause mortality (Table 2) with screening except MASS (HR, 0.97 [95% CI, 0.95 to 0.99]).¹⁷⁰ Individually calculated RR point estimates ranged from 0.98 to 1.00, with none reaching statistical significance.

Figure 2

Pooled Analysis of All-Cause Mortality (Male-Only) in One-Time Screening Trials. Abbreviations: CG = control group; CI = confidence interval; IG = intervention group; N = population size; n = sample size; RR = relative risk.

Table 2

All-Cause and AAA-Related Mortality Data for One-Time Screening Trials (KQ1).

Pooled analysis of AAA-related mortality of the four trials^{15, 113, 147, 170} (N=124,929) showed a statistically significant 35 percent reduction associated with invitation to screening (Peto OR, 0.65 [95% CI, 0.57 to 0.74]; I²=80%) (Figure 3) with high heterogeneity. We estimate a number needed to screen of 305 men (95% CI, 248 to 411) to prevent one AAA death. The results lost statistical significance, however, with the restricted maximum likelihood method (data not shown). Individual trial results demonstrate that the MASS and Viborg trials found a statistically significant AAA-related mortality benefit in the invited group compared to the control group at the longest followup time points, while the Chichester study reported an HR that was less than 1 but not statistically significant (Table 2).

Figure 3

Pooled Analysis of AAA-Related Mortality and Ruptures (Male-Only) for Rupture in One-Time Screening Trials. Abbreviations: CG = control group; CI = confidence interval; IG = intervention group; N = population size; n = sample size; OR = odds ratio.

Emergency Operations

Pooled results at the longest followup of the five trials^{15, 113, 146, 147, 170} (N=175,085) showed a reduction in emergency operations in the invited group (Peto OR, 0.57 [95% CI, 0.48 to 0.68]; I²=27%) (Figure 4). We estimate that screening 1,000 men for AAA would reduce the number of emergency procedures by 2 (95% CI, 2 to 2). Individual trial results show that results for number of emergency surgeries were as follows: MASS, Viborg, and Western Australia trials reported significantly fewer emergency surgeries in the invited group at the longest-term followup (MASS: Peto OR, 0.50 [95% CI, 0.39 to 0.64]; Viborg: Peto OR, 0.47 [95% CI, 0.29 to 0.77]; Western Australia: Peto OR, 0.60 [95% CI, 0.37 to 0.95]), while Chichester and VIVA showed similar nonsignificant trends (Chichester: Peto OR, 0.77 [95% CI, 0.41 to 1.48]; VIVA: OR, 0.82 (95% CI, 0.53 to 1.27) (Table 1, Figure 4).

Figure 4

Pooled Analysis of Operations (Male-Only) in One-Time Screening Trials. Abbreviations: CG = control group; CI = confidence interval; IG = intervention group; N = population size; n = sample size; OR = odds ratio.

Summary of Results

Available subgroup analyses are scant, and their analytical credibility is mixed. The Viborg¹⁴⁷ and Western Australia¹⁵ trials, both of which were population-based screening trials, reported subanalyses, with substantial limitations suggesting that there is no differential screening effect based on age. While our review scope included adults age 50 years or older, none of the trials recruited patients younger than age 64 years. Only the Chichester^{36, 113} trial, also a population-based screening trial, examined AAA screening in women, showing a low prevalence of AAA in women, with most screen-detected AAAs with an aortic diameter measuring 3.0 to 3.9 cm.^{13, 36} There was no difference in AAA rupture rate among women at 10-year followup or in AAA-related or all-cause mortality at 5 years between the invited and control groups, but the trial was underpowered to detect such differences. While the Western Australia trial reported that smoking is associated with a higher risk of all-cause mortality (OR, 1.59 [95% CI, 1.47 to 1.72]) and AAA-related mortality (OR, 2.95 [95% CI, 1.04 to 8.43]) in the screened group,¹⁵ it did not compare outcomes in the unscreened control for comparison and therefore does not address modification of intervention effectiveness by smoking status.

Study Characteristics and Results

Summary of Results

No trial-level evidence examined the effectiveness of one-time screening plus rescreening compared to one-time screening alone. Eight heterogeneous, prospective cohort studies recruited screen-negative participants and administered various rescreening protocols (rescreening every 1 to 5 years with 1 to 6 repeated scans) and reported the proportion of initially normal or ectatic aneurysms that reach an aortic diameter of 5.0 or 5.5 cm at the repeat scan. These studies report that AAA-related mortality over 5 to 12 years is rare (< 3%) among those with normal aortas (<3 cm) on the initial scan. Upon rescreening, few aortas grew to larger than 5 cm (0% to 2.2%) at 5 years,^{121, 123, 138} and 0 to 15 percent had progressed at 10 years.¹²⁰ Four studies reported no AAA ruptures or AAA-related deaths^{121, 138, 167, 169} at 4- to 5-year followup; one study reported 2.4 percent ruptures at median 7.9-year followup.¹²⁰ Overall, this heterogenous body of literature was too limited to make conclusions about the effectiveness of rescreening.

Study Characteristics

Five fair-quality prospective studies,^{121, 123, 165, 167, 169} two good-quality cohort studies,^{119, 120, 125, 138, 151, 156} and one fair-quality case-control study¹⁴⁸ examined the yield of rescreening participants who initially screened negative for AAA (Appendix E Table 4). Six of these studies were available in the previous review¹⁰⁵ and two of the studies^{167, 169} are new; one screening program has updated data.¹⁵⁶ Two of these studies analyzed subsamples of the Aneurysm Detection and Management (ADAM) Veterans Affairs trial of open repair vs. surveillance strategies for small AAAs.^{121, 138} Additionally, three studies were based in the United Kingdom: a subsample of the Gloucestershire screening study,¹⁵⁶ a hospital screening program,¹²³ and a Chichester screening program separate from the Chichester trial.¹⁶⁵ Two of the cohort studies were conducted in Sweden,^{167, 169} and one of these was conducted exclusively in women.^{167, 169} The case-control study was a subsample of the Viborg screening trial; however, data were only considered from the cohort of participants with ectatic aortas who were offered rescreening. The size of the rescreened cohorts ranged from 33¹⁶⁷ to 2,692¹⁶⁹ participants; samples of those with normal or ectatic aortas were derived from larger screening programs and had mean followup ranging from 4 to 10 years. The Gloucestershire population-based screening program published their observations over the 25-year history of the program with a proportion being followed for at least 10 years^{120, 156} (Appendix E Table 4).

The definition of “normal” or “ectatic” aortas differed; inclusion criteria for selection of the rescreening cohort based on aortic diameter were defined as follows: 2.5 or 2.6 to 2.9 cm,^{121, 123, 148, 156, 167, 169} smaller than 2.5 cm,¹⁶⁹ or 3 cm or smaller.^{138, 165, 169} Ultrasound measurement techniques varied, with some measurements obtained using inner-to-inner wall measurements,^{119, 120, 125, 151, 156} leading edge,^{167, 169} or unspecified measurements.^{121, 123, 138, 148, 165} Repeat screening occurred at various intervals after the initial normal scan as follows: annually,^{121, 123,156} once at 3 to 5 years,^{138, 148, 167, 169} and every 2 years.¹⁶⁵ Participants had a total of one,^{138, 148, 167, 169} five,^{121, 165} or six¹⁵⁶ scans after initial screening over the study duration; one study¹²³ did not report total number of scans.

Benefit outcomes (Tables 3 and 4) reported in these trials included all-cause mortality,^{121, 156, 167, 169} AAA-related mortality,^{121, 138, 156, 167, 169} AAA rupture,^{121, 138, 156, 165, 167, 169} and large AAA incidence.^{121, 123, 138, 148, 156, 165, 167} Six of these studies reported the use of procedures (Table 3).^{121, 138, 148, 156, 167, 169}

Table 3

AAA Prevalence, Rupture, and Surgery Data for Rescreening Studies (KQ2).

Table 4

All-Cause and AAA-Related Mortality Data for Rescreening Studies (KQ2).

Most studies included men age 65 years or older, with only one study including men age 50 years and older (Appendix E Table 5).¹³⁸ The mean age of participants ranged from 65 to 70 years. One Swedish study¹⁶⁷ solely recruited 70-year-old women, and one study¹³⁸ reported the inclusion of few female participants (2.4% women). The remaining studies solely recruited men. Additionally, the two studies that followed subgroups from the ADAM trial^{121, 138} and one Swedish study¹⁶⁷ included risk factor information.

Overall, this group of observational studies is limited because of the following: a small number of participants with normal aortas was included; all but a single study exclusively recruited men; cohort study designs did not have matched controls, and the primary focus of most studies was growth rate because the followup time for most studies was 5 years. This time frame would be too short to expect the development of AAA-related health outcomes.

Detailed Results

Summary of Results

There were no trials available to examine the differential effectiveness of rescreening by subpopulation. Nearly all rescreening cohort studies were performed in Caucasian men and most did not perform subgroup analyses. Scant available data had low credibility. Only one good-quality rescreening study¹³⁸ reported AAA-related health outcomes at 4 years and captured risk factor information; however, there were no AAA-related deaths or ruptures reported in this study. A single rescreening study among women (N=25)¹⁶⁷ reported AAA-related health outcomes, but it was too small to make any comparisons with the other all-male studies. Two studies^{138, 169} reported multivariable regression analyses suggesting that current smoking is an independent risk factor for the development of AAA at rescreening, and another study’s univariate analysis shows a similar trend for smoking status among women.¹⁶⁷

Study Details

Only four of the rescreening studies provided data on screening in subgroups.^{138, 165, 167, 169} For all but one study,¹³⁸ it was unclear whether the subanalyses were prespecified. Additionally, there was a lack of a proper control group in these studies and the overall study sizes were small, making the credibility of the subgroup analysis low. The conclusions about the effects of rescreening in subgroups were limited by lack of adequate reporting of health outcomes by subpopulation, as well as by the heterogeneity of study rescreening protocols and short followup times.

Detailed Results

Summary of Results

All four population-based screening RCTs (2 fair-quality, 2 good-quality) provide information on operative mortality and the number of surgeries for AAA, showing an increase in elective surgeries among the intervention group compared to the control group, but no difference in operative mortality.^{12, 15, 113, 147, 168, 170} One new fair-quality population-based screening trial, VIVA,^{22, 146} which looked at the impact of screening for multiple cardiovascular conditions, provides the number of elective and emergency operations among those screened. Overall, there were approximately 40 percent more surgeries in the invited group than in the control group (k=5; N=175,085; Peto OR, 1.44 [95% CI, 1.34 to 1.55]), which is largely driven by elective operations (k=5; N=175,085; Peto OR, 1.75 [95% CI, 1.61 to 1.90]). There was no statistically significant difference in 30-day mortality rates among the invited vs. control groups for either elective surgeries or emergency surgeries at 12- to 15-year followup.

Three small quality of life observational studies report mixed results.^{141, 150, 174} These studies, in addition to the two population-based screening RCTs that reported quality of life scores,^{12, 15, 168, 170} generally showed no substantial differences in quality of life, anxiety, or depression scores between those who screened positive and those who were unscreened or screened negative for AAAs based on one-time screening.

Study Characteristics

Four population-based AAA screening trials previously described for KQ1 also provide data on harms (Tables 1 and 2).^{15, 113, 147, 170} These four RCTs were included in the previous review, with the addition of updated long-term followup data available from the Western Australia trial.¹⁵ One additional new multicomponent population-based Danish screening trial, VIVA, was included in this current review of harms to estimate elective surgeries.¹⁴⁶ VIVA enrolled 50,156 men ages 65 to 74 years from 2008 to 2011. Participants were randomized to screening vs. no screening for hypertension, peripheral artery disease, and AAA. After screening, VIVA participants who had confirmed AAA or peripheral artery disease were counseled on the need to initiate preventive interventions including walking, smoking cessation, a low-fat diet, and cholesterol testing, with aspirin and statin therapy prescribed to those meeting a total cholesterol threshold value (Appendix E Tables 1 and 2).¹⁴⁶ An interim analysis at a median of 4.4 years of followup reported number of operations, all-cause mortality, and AAA-related outcomes, including causes of death based on death certificates. The effects of AAA screening alone could not be independently assessed with respect to all-cause or AAA-mortality because there were multicomponent screening interventions administered; however, the number of surgeries were included in this review as these would almost exclusively be expected as a result of AAA screening.

One study (MASS) reported quality of life differences over time between screened and unscreened populations.^{12, 170} This MASS subsample^{12, 170} plus four additional studies measured various quality of life questionnaires for screened populations with and without AAA diagnoses, comparing prescreening baseline scores to repeated scores at 1 to 15 months after screening (Appendix E Tables 1 and 6).^{15, 141, 150, 168, 174} All of these quality of life studies were available in the previous review.¹⁰⁵ Two studies were subsamples from the screened arms in the MASS (N=1,956; 599 AAAs, 631 normal aortas, 726 not invited for screening)^{12, 170} and Western Australia (N=365; 120 AAAs, 245 normal aortas) trials;^{15, 168} two additional observational comparisons were analyzed from population-based screening programs in Gloucestershire (N=161; 61 AAAs, 100 normal aortas)¹⁵⁰ and Sweden (N=69; 24 AAAs, 45 normal aortas);¹⁷⁴ and one small observational study analyzed screened participants in a rural Australian screening study (N=183 completed postscreening questionnaire; 35 AAAs, 89 normal aortas).¹⁴¹ The studies reported quality of life outcomes using the 36-Item Short-Form Health Survey (SF-36)^{12, 15, 141, 168, 170, 174} and EuroQOL EQ-5D.^{12, 15, 168, 170} Mood (anxiety or depression) was measured with the Hospital Anxiety and Depression scale (HADS),^{15, 141, 168} the General Health Questionnaire (GHQ),¹⁵⁰ and an 80-point linear analogue anxiety scale.¹⁵⁰ Questionnaires were administered prior to screening, after screening, or in selected subgroups of participants with screen-detected AAAs undergoing surgery or surveillance. Timing of questionnaires was up to 12 months after a specific event.

Detailed Results

Operative Mortality

30-Day Postoperative Mortality From Elective Surgery

Two of the four population-based screening trials report 30-day operative mortality from elective surgery, showing no difference among participants invited to screening and those in the control group (Table 2).^{15, 170} The MASS¹⁷⁰ and Western Australia¹⁵ trials reported no statistically significant difference in 30-day mortality from elective surgery between the invited and control groups at 12.8- to 13.1-year followup (n=1,827; MASS: RR, 0.76 [95% CI, 0.40 to 1.45]; Western Australia: RR, 0.82 [95% CI, 0.43 to 1.57]) (Figure 5).

Figure 5

Forest Plot of 30-Day Mortality (Male Only) Due to Elective and Emergency Surgery in One-Time Screening Trials. Abbreviations: CG = control group; CI = confidence interval; IG = intervention group; N = population size; n = sample size; RR = relative risk. (more...)

30-Day Postoperative Mortality From Emergency Surgery

The MASS¹⁷⁰ and Western Australia¹⁵ screening trials report 30-day operative mortality from emergency surgery and show similar results (Table 2). They showed no statistically significant difference in 30-day mortality from emergency surgery between the invited and control groups at 12.8- to 13.1-year followup (n=316; MASS: RR, 0.98 [95% CI, 0.68 to 1.43]; Western Australia: RR, 1.43 [95% CI, 0.90 to 2.25] (Figure 5).

Summary of Results

Overall, there is very little information on screening harms reported by subpopulation and what is reported has low credibility as it is unlikely to have been prespecified or powered to detect subpopulation differences. Available scant data from a single trial (Western Australia) suggest a trend of higher postoperative mortality after elective repair associated with screening (compared to no screening) in older age groups but no differential screening effect on number of operations by age.¹⁵

Detailed Results

Summary of Results

Four trials evaluated the comparative effectiveness of early surgical repair vs. surveillance for small aneurysms (diameter of 4 to 5.4 cm): two evaluated the effectiveness of early open surgery^{140, 163} and two evaluated early EVAR interventions.^{118, 158} All four of these trials showed no difference in all-cause or AAA-related mortality between the intervention and control groups at up to 8 to 12 years for open repair^{140, 163} and up to 2.6 years for EVAR.^{118, 158}

Seven pharmacotherapy efficacy trials examining antibiotics, antihypertensive medications, or mast cell stabilizers were identified in this updated review.^{114, 132, 133, 152, 153, 164, 166} In all included trials, pharmacotherapy intervention showed no overall impact on AAA growth compared to placebo. Conclusions are limited by a small number of trials evaluating each medication and trial durations, which were too short to expect the development of AAA-related events or changes in health outcomes.

Study Characteristics

Detailed Results

Open Repair vs. Surveillance

All-Cause Mortality

Both trials found no significant differences in all-cause mortality at any followup time between participants receiving early open repair vs. surveillance (Table 5).^{140, 162, 163} At 5-year followup, the ADAM trial reported slightly more deaths in the intervention group than in the control group (25.1% vs. 21.5%; RR, 1.21 [95% CI, 0.95 to 1.54]),¹⁴⁰ while UKSAT reported slightly more deaths among those in the control group (30.6% vs. 46.7%, RR, 0.91 [95% CI, 0.72 to 1.16]),^{162, 163} but neither finding was statistically significant. At 12-year followup, UKSAT reported a similar trend; however, the difference was not statistically significant (adjRR, 0.88 [95% CI, 0.75 to 1.02]). An individual patient data meta-analysis (n=2,226) of patients randomized to both trials showed no survival benefit at approximately 5 years in both the unadjusted (HR, 0.96, [95% CI, 0.81 to 1.14]) and adjusted analyses (HR, 0.99, [95% CI, 0.83 to 1.18]).¹²⁶

Table 5

All-Cause and AAA-Related Mortality Data for Open vs. Surveillance Trials for Small AAA (KQ4).

AAA-Related Mortality

Similar to the findings for all-cause mortality, both trials found no significant differences in AAA-related mortality at any followup time period (Table 5).^{140, 162, 163} At 5-year followup, the ADAM trial reported nearly identical rates of AAA-related mortality among participants who received open repair vs. surveillance (3.0% vs. 2.6%; HR, 1.15 [95% CI, 0.58 to 2.31]).¹⁴⁰ Mortality rates in UKSAT were slightly higher at 5 years of followup and more deaths were observed in the surveillance group compared to the group receiving surgery, but the difference was not significant (5.7% [IG] vs. 6.6% [CG]; RR, 0.86 [95% CI, 0.54 to 1.36]).^{162, 163} At 12 years of followup, UKSAT reports an AAA-related mortality rate of 6.9 percent in the intervention group and 9.5 percent in the surveillance group.¹⁶³

Rupture

Ruptures were rare events in both trials; however, early open repair significantly reduced the rate of rupture compared to the group undergoing surveillance at each followup interval (Table 6).^{140, 162, 163} The ADAM trial reported a rupture rate of 0.4 percent in the early intervention group compared to 1.9 percent in the surveillance group (RR, 0.18 [95% CI, 0.04 to 0.81]).¹⁴⁰ Rates were slightly higher in UKSAT at 5-year followup but remain significantly different between treatment groups (1.2% vs. 3.2%; RR, 0.33 [95% CI, 0.13 to 0.83]).^{162, 163} At 12 years of followup, UKSAT reported that 2.3 percent of the early surgery group and 4.5 percent of the surveillance group experienced AAA rupture (RR, 0.51 [95% CI, 0.26 to 0.99]).¹⁶³

Table 6

AAA Growth Rate, Rupture, and Surgery Data for Open vs. Surveillance Trials for Small AAAs (KQ4 and KQ5).

All Operations

As expected, overall there were more surgical interventions in the early surgery groups than in the surveillance group (Table 6). The ADAM trial reported only overall procedures (IG: 92.6% vs. CG: 61.6%) and did not break them down into emergency vs. elective.¹⁴⁰ In UKSAT, the majority of surgical interventions were elective at each followup timepoint.^{162, 163} At 12 years of followup, there were notably fewer emergency surgeries in the early surgery group compared to the group undergoing surveillance (3 [0.5%] vs. 6 [1.1%]), but emergency surgeries overall were rare.¹⁶³

Early EVAR vs. Surveillance

All-Cause Mortality

Both EVAR trials found no significant differences in all-cause mortality at 1.7- to 2.6-year followup between participants receiving early EVAR and those undergoing surveillance (Table 7).^{118, 158} At 2.6 years, CAESAR reports similar rates of all-cause mortality between treatment groups (5.5% vs. 4.5%), with an HR (CG vs. IG) of 0.76 (95% CI, 0.30 to 1.93).¹¹⁸ Likewise, PIVOTAL found no difference in all-cause mortality between participants receiving early EVAR vs. surveillance at 1.7 years (4.1% vs. 4.1%; HR, 1.01 [95% CI, 0.49 to 2.07]).¹⁵⁸

Table 7

All-Cause and AAA-Related Mortality Data for EVAR vs. Surveillance Trials for Small AAA (KQ4).

AAA-Related Mortality

Similar to the findings for all-cause mortality, both trials found no significant difference in AAA-related mortality at any followup time period (Table 7).^{118, 158} Events were rare and findings in both trials were nearly identical in each treatment group (1 [0.5%] vs. 1 [0.6%] in CAESAR; 2 [0.5%] vs. 1 [0.3%] in PIVOTAL). The relative risks were nonsignificant, with wide CIs. Conclusions are limited by low event rates.

Rupture

Ruptures were rarely reported in either trial, making comparisons challenging (Table 8). Both trials report zero ruptures among participants receiving early EVAR and few events among those undergoing surveillance (2 [1.1%] in CAESAR; 1 [0.3%] in PIVOTAL).^{118, 158}

Table 8

AAA Growth Rate, Rupture, and Surgery Data for EVAR vs. Surveillance Trials for Small AAAs (KQ4 and KQ5).

All Operations

Again, as anticipated, there were more total surgical interventions in the early surgery groups than in the surveillance groups (Table 8). In both trials, the majority of surgeries were elective. with more elective surgeries reported in the intervention groups (CAESAR: 94.0% vs 39.9%; PIVOTAL: 88.9% vs 30.1% received surgery).^{118, 158} Emergency surgeries were rare events. PIVOTAL reported only a single emergency surgery in the surveillance group, and there was mention of emergency surgeries in the early EVAR group. CAESAR did not report emergency surgeries as an outcome.

Pharmacotherapy vs. Placebo

Detailed Results

AAA Growth Rates

All seven trials reported no significant beneficial impact of pharmacotherapy on AAA growth (Table 9).^{114, 131–133, 136, 152, 153, 164, 166, 172} One trial of doxycycline compared with placebo reported that the intervention group had a statistically significantly greater growth in aneurysm size compared with those taking placebo (4.1 vs. 3.3 mm; difference, 0.8 mm [95% CI, 0.1 to 1.4 mm]).¹⁵² This trial included participants with larger AAAs who were unfit for surgery. The remaining six trials showed no statistically significant difference in growth rates between the intervention and control groups.^{114, 132, 133, 153, 164, 166} Four of these trials^{114, 132, 153, 164} showed a nonsignificant lower mean growth rate in the intervention group, with differences ranging from 0.5 mm/year¹⁶⁴ to 1.5 mm/year.¹⁵³ Conversely, one trial showed identical mean growth rates of 2.2 mm/year (median interquartile range) in the intervention and control groups,¹³³ and one trial of pemirolast showed a nonsignificant greater aneurysm growth rate in the intervention group compared to those taking placebo (2.71 vs. 2.04 mm/year).¹⁶⁶

Table 9

AAA Growth Rate, Rupture, and Surgery Data for Pharmacotherapy vs. Placebo Trials for Small AAAs (KQ4 and KQ5).

All-Cause Mortality

Few studies reported the impact of pharmacotherapy on all-cause mortality, and those that did found mixed trends without statistical significance (Table 10). Three antibiotic trials^{133, 152, 153} and one beta-blocker trial¹⁶⁴ report rates of all-cause mortality. PAT reported a higher mortality rate among participants in the treatment group compared to those taking placebo; however, the difference was nonsignificant (12% vs. 9.6%; p=0.36).¹⁶⁴ This same trend was reported in the study of doxycycline by Mosorin et al (23.5% vs. 20.0%; significance not reported).¹⁵³ In the remaining studies, there were more deaths observed in the control group; however, these events overall were too rare to make any conclusions about effects on all-cause mortality at 1.5- to 2.5-year followup.

Table 10

All-Cause and AAA-Related Mortality Data for Pharmacotherapy vs. Placebo Trials for Small AAA (KQ4).

AAA-Related Mortality

Again, few studies reported the impact of pharmacotherapy on AAA-related mortality and those that did found mixed results (Table 10). Only three trials—two antibiotic trials^{133, 152} and one beta-blocker trial¹⁶⁴—reported this outcome. Two of the three trials report the same number of events between treatment groups, while one trial reports a slightly higher rate among those taking placebo.¹⁵² Events overall were rare (0 to 2 events in each group), thereby limiting conclusions.

AAA Rupture

Five trials reported instances of AAA rupture between treatment groups and again found mixed results (Table 9).^{114, 152, 153, 164, 166} The event rates were very low in these trials; two trials reported no ruptures in both the intervention and control groups,^{114, 166} and three trials had 0 to 2 events in each group (control group rates, 0% to 1.4%), making it difficult to make conclusions about the medication’s effect on rupture at 1- to 2.5-year followup.

All Operations

Total AAA-related procedures are reported in all seven pharmacotherapy trials; however, only five specify surgeries as emergency or elective procedures.^{132, 133, 152, 153, 164} Results showed a mixed pattern and statistical testing was not performed. Rates of total AAA-related procedures varied widely by study (range, 2% to 40% in control groups), with two studies^{133, 166} reporting more surgeries in the intervention group, three studies^{152, 153, 164} reporting more surgeries in the control group, and one study¹¹⁴ reporting nearly identical rates of surgery in the intervention and control groups (Table 9). One trial did not report procedures by treatment group.¹³²

Two doxycycline trials^{152, 153} and one propranolol trial¹⁶⁴ reported elective AAA surgeries with mixed results (Table 9). One doxycycline study¹⁵² reported similar rates of elective surgeries between treatment groups (IG: 14.6% vs. CG: 15.5%), while the other doxycycline trial¹⁵³ reported more elective repairs in the control group compared to the intervention group (IG: 11.8% vs. CG: 40%; statistical information not reported). Similarly, PAT reported more elective surgeries in the control group than in the treatment group (20.3% vs. 26.5%; statistical information not reported).¹⁶⁴

The same three studies reporting elective surgeries also reported emergency surgeries, finding mixed results. (Table 9).^{152, 153, 164} These emergency surgeries were rare, with 0 to 2 events in each group (0% to 1.4% in the control groups), again limiting conclusions about these medications’ effects on emergency surgery rates at 1.5- to 2.5-year followup.

Summary of Results

Subpopulation information was rarely reported among treatment studies for small AAAs. For trials of early intervention via open surgery compared to surveillance, the available trials reported limited data on the subpopulations of sex, age, and smoking status. Only UKSAT, with a 12-year followup, and the individual patient data meta-analysis, which pooled 5- to 8-year followup data from ADAM and UKSAT, reported all-cause mortality by sex, finding no differential effect.^{126, 162, 163} Both trials reported the impact of age on all-cause mortality, utilizing slightly different age cutoffs, but did not report differential all-cause mortality treatment effect by age at 4.9 and 12-year followup. UKSAT reported no differences in all-cause mortality by smoking status but did not report outcomes in smokers for each treatment arm, so no conclusions could be made about differential treatment effect of early surgery by smoking status. Neither trial of early EVAR compared to surveillance reports data by subpopulation.

None of the pharmacotherapy trials reported health outcomes by subgroup. One small doxcycline trial and one propranolol trial performed limited subgroup analyses, which did not support treatment effect modification by age or smoking status. These available analyses would be considered exploratory at best, particularly given that the subgroup methodologies were of low quality and overall trial results do not support an AAA growth benefit.

Detailed Results

Summary of Results

Two trials, ADAM and UKSAT,^{127, 137, 140, 163} reported harms of open repair vs. surveillance. These trials report rates of intervention and associated mortality, finding a 50 percent higher rate of procedures in the early intervention group compared to the control group but no difference in 30-day postoperative mortality. Readmission rates at 30 days postoperation were higher in the early intervention group in the ADAM trial than in the surveillance group; however, major surgical complications were lower in the early intervention group. Quality of life results were mixed in the early open repair vs. surveillance trials, but generally showed declines in both treatment groups over time, with no statistically significant difference observed between the groups up to 1 to 2 years postrandomization. The ADAM trial showed higher general health scores in the early repair group in the first 2 years; however, this difference did not persist over time. One trial reported higher rates of impotence in the early repair group.

Two trials of early EVAR vs. surveillance (PIVOTAL¹⁵⁸ and CAESAR¹¹⁸) report an approximately 100 percent higher procedure rate in the early intervention group compared to the surveillance group but no difference in 30-day postoperative mortality (Table 11). In the CAESAR trial, the rate of complications was consistently higher among those receiving early EVAR compared to those undergoing surveillance (the number of patients with any adverse events, any morbidity at 30 days postoperation related to repair, endoleaks at 1 year, and reinterventions). Rates were similar for any major morbidity over the trial duration between treatment groups. Conversely, the PIVOTAL trial largely reported similar rates of adverse events between groups at 30 days and 1-year postoperation and reinterventions.

Table 11

Harms Data in Studies of Treatment for Small AAAs (KQ5).

In general, national and international registries report clinically important harms rates that were similar to those reported in the trials.

With the exception of the two propranolol trials reporting high rates of discontinuation due to adverse events, other medications (including other antihypertensive medications [ACE inhibitors, calcium-channel blockers] and antibiotics) appear to be well tolerated based on few trial withdrawals reported from a small number of studies per drug class.

Study Characteristics

The two trials of early open surgery^{127, 137, 140, 163} and the two trials of early EVAR^{118, 122, 124, 158} discussed in detail above report data on harms (Table 11). The surgical harms considered in this review included surgical procedures, 30-day postoperative mortality, surgical complications including readmissions, and quality of life. In addition to the seven pharmacotherapy RCTs described in KQ4,^{114, 132, 133, 152, 153, 164, 166} one additional RCT of propranolol vs. placebo provided harms data only.¹⁴²

Five fair-quality registry studies assessing outcomes after EVAR reported harms data for small aneurysm repair^{116, 129, 149, 159, 160} (Table 12, Appendix E Table 9). Three^{116, 149, 159} of the five registries are new since the last review.¹⁰⁵ The five registry publications that prospectively examined EVAR complication rates with subgroup reporting for small-sized AAAs measuring smaller than 5.5 cm were the Vascunet international registry (N=12,610 small AAAs),¹¹⁶ the Australian Safety and Efficacy Register of New Interventional Procedures-Surgical (ASERNIP-S)¹²⁹ (N=478), the U.S. Vascular Study Group New England (VSGNE) (N=1,336)¹⁴⁹ and the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) (N=5,126),¹⁵⁹ and the European Collaborators on Stent/Graft Techniques for Aortic Aneurysm Repair (EUROSTAR) (N=1,962).¹⁶⁰

Table 12

Harms Data in Registry Studies (KQ5).

The two largest and most contemporary registries were the Vascunet¹¹⁶ and ACS NSQIP.¹⁵⁹ The Vascunet international registry from 11 countries reported data on small AAA (<5.5 cm) repair with EVAR and open repair from 2010 to 2013; this registry represents the largest registry included in the review and captured greater than 90 percent of repairs done in the majority of the participating countries during the time period represented. The mean age and sex of the subgroup with small AAAs were not reported, nor was the mean followup time. The ACS NSQIP is a nationally validated, U.S.-based, risk-adjusted dataset of surgical procedures that provided complication rates for open repair and EVAR procedures of small AAAs from 2011 to 2015 with outcomes reported by size quartile, including 3.5 to 5 cm and 5.01 to 5.5 cm. The mean age was 72.3 years and the population comprised 21.9 percent females. Approximately one-third were smokers. The ASERNIP-S registry from Australia reported complication rates for small AAA (≤5.5 cm) repair by EVAR from 1999 to 2001, with mandatory reporting by vascular surgeons during the time period. The mean age was 75 years, with a population comprising 15.9 percent women and 11 percent smokers; the median followup was 3.2 years. The VSGNE registry reported complication rates for small AAAs (<5.5 cm) repaired by EVAR and open surgery from 2003 to 2011 from a voluntary collaboration among vascular surgeons, cardiologists, and radiologists from 30 community and academic hospitals in New England. The mean age was 71 years, and the population comprised 26.2 percent women and 88.5 percent smokers; the mean followup was 1 year. EUROSTAR is an international registry from 17 European countries and 110 institutions reporting complications of elective EVAR for small AAA (4.0 to 5.4 cm) repair from 1997 to 2002. The mean age of the population was 69.7 years and participants were primarily men (only 7% were female); mean followup was 1.7 years (Appendix E Table 9).

The two EVAR-only registries, ASERNIP¹²⁹ and EUROSTAR,¹⁶⁰ reported endoleaks and reinterventions, as well as 30-day operative mortality. The three registries that included both EVAR and open repair reported 30-day operative mortality,^{116, 149, 159} reinterventions,¹⁵⁹ readmissions,¹⁵⁹ and/or other complications.¹⁵⁹

Harms Associated With Early Open Surgery vs. Surveillance

Harms Associated With Early EVAR vs. Surveillance

Harms Associated With Pharmacotherapy

Overall, the trials of antibiotic interventions show that these medications were well tolerated. Both doxycycline trials report similar withdrawals or discontinuations due to adverse events in the treatment and placebo groups at a mean of 1.5 years of followup (Table 13). The doxycycline trial by Mosorin et al¹⁵³ (N=32) reported that the active drug was well tolerated in most patients; one patient in the doxycycline group and one in the placebo group discontinued the drug due to an allergic reaction (5.9% vs. 6.7%). The other doxycycline trial (N=286 analyzed) suggested no statistically significant difference in adverse events leading to study withdrawal (7.6% vs. 2.1%; difference, 5.5 [95% CI, −6.3 to 17.1]).¹⁵² In the azithromycin trial¹³³ (n=211), 13 patients in the treatment group and 8 patients in the control group reported side effects due to gastrointestinal side effects, arthralgias, or allergic reactions at 1.5 years (12.3% vs. 7.6%; no statistical difference reported). In the Roxithromycin trial (N=84),¹³² there were no adverse events or drop outs reported at 2-year followup.

Table 13

Harms Data in Studies of Treatment for Small AAA (KQ5): Pharmacotherapy vs. Placebo.

Both propranolol trials report high rates of discontinuation due to adverse events,^{142, 164} while the Aortic Aneurysmal Regression of Dilation: Value of ACE-Inhibition on Risk (AARDVARK) trial of an ACE inhibitor and a calcium-channel blocker suggests that these other antihypertensive medications are generally well tolerated (Table 13).¹¹⁴ In the PAT trial (N=539), there was a higher rate of drug discontinuation in the propranolol group compared to placebo at 2.5 years (42.4% vs. 26.8%; difference, 15.6% [95% CI, 7.6% to 23.5%]).¹⁶⁴ Likewise, the propranolol group had a higher rate of trial withdrawals due to adverse events (37.7% vs. 21.3%; difference, 16.4% [95% CI, 8.7% to 24.0%]). In the small Danish propranolol trial (N=54), a substantial and statistically significantly greater proportion of patients dropped out compared to the placebo group at 2 years of followup (60% vs. 28%; RR 5.7 [95% CI, 1.5 to 22.2]).¹⁴² The most commonly cited reasons for these dropouts were death, serious cardiac arrhythmia, dyspepsia, headache, and dizziness in the intervention group. The AARDVARK trial (N=224) with the active ACE inhibitor and calcium-channel blocker arms had generally low withdrawals due to adverse events attributed to study medications at 2 years of followup (2.7% and 5.6% in the ACE inhibitor and calcium-channel blocker groups, respectively).¹¹⁴ Additionally, 4.1 percent (3/73) and 1.4 percent (1/72) from the ACE inhibitor and calcium-channel blocker groups, respectively, switched to another class of medication due to cough. Similarly, the only trial of a mast cell stabilizer (Anti-Inflammatory Oral Treatment of AAA [AORTA] trial) (N=168 analyzed single active drug vs. placebo arm) reported high rates of any adverse events (approximately 80%) and serious adverse events (approximately 18%) in both the intervention and placebo groups; a higher percentage of participants withdrew in the placebo group compared to the treatment arms at 1-year followup (1/84 [2.5%] vs. 8/84 [9.5%]).¹⁶⁶

Two propranolol trials^{142, 164} report quality of life, generally showing no difference between the intervention and control groups. One study reports a decline in SCREENQOL in both the intervention and control groups through 2 years without a significant difference between the groups,¹⁴² and the other study reports no significant difference in SF-36 scores between the propranolol and control groups at 1 month postrandomization.¹⁶⁴

Summary of Results

Scant data have been reported for harms among subpopulations. The only available evidence comes from surgical registries due to no trial data being available to examine harms in subpopulations. Existing evidence from three surgical registries shows higher 30-day operative mortality and secondary complications among women compared with men for both EVAR and open repair of small aneurysms.

Detailed Results

Overall, registry data show a higher rate of postoperative mortality following elective repair of small AAAs in women compared to men, regardless of the surgical technique. One registry study¹¹⁶ reports higher rates of 30-day postoperative mortality in women compared to men for both open repair and EVAR for small AAAs; this pattern remains consistent regardless of age (≥80 and <80 years). For example, in participants younger than age 80 years, the 30-day postoperative mortality is 2.5 percent in men and 3.0 percent in women for open repair, and 0.4 percent in men and 1.1 percent in women for EVAR. In those age 80 years or older, postoperative mortality for open repair was 3.4 percent in men and 9.7 percent in women, while intervention via EVAR was associated with a postoperative mortality of 0.6 percent in men and 1.3 percent in women. One registry shows a similar pattern for AAA repair (EVAR and open repair reported combined), with a 30-day postoperative mortality of 0.7 percent in men and 1.1 percent in women.¹⁴⁹ Further, one registry reports lower clinical success rates following surgery in women at 3- and 5-year followup compared to men. The difference observed was greatest at 5 years, with 93 percent (±1) of men and only 70 percent (±16%) of women having clinical success.¹²⁹

The two early open repair vs. surveillance trials, ADAM and UKSAT,^{140, 163} and the two early EVAR vs. surveillance trials, CAESAR and PIVOTAL,^{118, 158} did not report harms by subpopulation. Additionally, none of the pharmacotherapy trials reported harms data by subpopulation.