Author Affiliations: Cecil G. Sheps Center for Health Services Research and Department of Medicine, University of North Carolina, Chapel Hill (Drs Harris and Sheridan); and Department of Health Behavior and Health Education, University of North Carolina School of Public Health, Chapel Hill (Dr Kinsinger).
The 2011 Institute of Medicine report titled “Clinical Practice Guidelines We Can Trust”1 recommends that guidelines be updated at a prespecified time and that new evidence be reviewed regularly to determine the need for earlier updating. New evidence, however, is not the only reason for updating guidelines. Another is to reconceptualize benefits and harms with the advantage of more experience and new understanding. This would seem to be the essence of critical thinking, of avoiding the cognitive “confirmation” bias whereby we simply continue to think the way we have, without serious questioning. An article in this issue of the Archives2 considers screening for abdominal aortic aneurysm (AAA). Both new evidence about benefits and new concepts of harms suggest that it may now be time to rethink AAA screening.
The most recent US Preventive Services Task Force (USPSTF) recommendation for screening for AAA was in 2005.3 At that time, a systematic review of the evidence found 4 randomized controlled trials (RCTs) of screening for AAA using the open aneurysm repair (OAR) technique for AAAs of at least 5.5 cm in diameter. Pooled results showed that screening reduced AAA-associated mortality by a relative 43% over about 5 years; absolute risk reduction among higher-risk men (ages 65 to 75 who had ever smoked) was from about 4.6 to about 2.6 AAA-associated deaths for every 1000 men over 5 years. There was no difference in all-cause mortality between screened and unscreened groups. Based on these findings, the USPSTF recommended a 1-time ultrasonographic screening in men aged 65 to 75 who had ever smoked (grade B recommendation). Owing to lower AAA-associated mortality, the USPSTF did not recommend screening routinely for men aged 65 to 75 who never smoked (grade C recommendation), and recommended against screening women (grade D recommendation).
Since the 2005 USPSTF review,3 evidence has changed. Although we are unaware of new population-based RCTs of screening for AAA, there is new evidence that the burden of AAA is rapidly decreasing. According to studies from the United States and other Western countries, mortality from ruptured AAA has decreased by as much as 50% in the past 10 to 15 years.4,5 This reduction closely parallels the reduction in smoking prevalence and myocardial infarction incidence, and it started before AAA screening began. This argues that the true incidence of AAA is decreasing, perhaps with little contribution from screening. Thus, the potential benefit from AAA screening is decreasing.
In recent years, clinical practice patterns have also changed in a way that raises a concern about increasing the harm of screening. Today, an estimated 70% to 80% of AAA repairs are performed using the endovascular aneurysm repair (EVAR) technique rather than OAR. Two RCTs comparing OAR with EVAR found a reduction in 30-day AAA mortality with EVAR (0.5% vs 3.0% in one study6; 1.8% vs 4.3% in the other).7 In both trials, however, there was no difference in AAA-associated mortality after 2 to 5 years of follow-up. The concern, however, is that the simpler (although not less expensive) EVAR technique may encourage intervention for aneurysms smaller than 5.5 cm, aneurysms that do not need repair. One study used a medical imaging database to examine 10 228 patients undergoing EVAR, finding that 59% had a maximum AAA diameter below the 5.5-cm treatment threshold suggested by the RCTs.8 Screening finds many people with these small “aneurysms” that pose little threat to the patient's life. In the largest screening RCT, for example, 71% of AAAs detected by screening were 3.0 to 4.4 cm in diameter, below the 5.5-cm threshold for surgery.9 If more people with these smaller AAAs are treated unnecessarily by EVAR, then the current treatment strategy may harm more people overall than was documented in the screening RCTs.
The harms of either OAR or EVAR are not limited to increased mortality. In 1 study, 25% of EVAR patients had an endoleak, and 4% required a secondary therapeutic procedure over a mean follow-up of 1.8 years.6 Reintervention was needed in 30% of EVAR patients (compared with 28% of OAR patients). Other nonfatal but serious medical complications push overall complication rates higher still.3
The harms of treatment are not the only harms associated with screening. In recent years, our conceptualization of other harms has evolved.10 Patients are playing an increasing role in health care decisions and often value outcomes differently than physicians or investigators. The psychological effects of being “labeled” with having an AAA that will not be repaired immediately (if at all) have not been fully explored, but at least 2 RCTs have shown negative psychological effects from diagnosis of small AAAs. This is a form of overdiagnosis that needs further study. The literature we do have raises concerns that these effects should not be taken lightly.
In sum, the USPSTF has a difficult job ahead in updating the recommendation on screening for AAAs. New evidence suggests that the population benefit of screening is lower now than it was in 2005 and may have been even lower then than the USPSTF knew. The harms of treatment may not have declined at all (and may have even increased), partly because of the apparently less stringent criteria being used by EVAR interventionists, with more procedures being performed for probably little benefit. Perhaps most importantly, our understanding of the harms of screening has evolved, teaching us to look more closely, and with the patient's perspective, at the potential harms of labeling, overdiagnosis, and nonfatal complications of intervention that are important features of AAA screening. Given these considerations, our weighing of benefits and harms for AAA screening may have changed.
In the current issue of the Archives, Shreibati et al2 find that, among eligible Medicare enrollees, reimbursement for AAA screening was associated with a “modest” 15% increase in the use of abdominal ultrasonography. They note several reasons that may help explain why the increase was not greater. Many will see the findings of this study as yet another example of unjustifiably slow diffusion of effective medical care into actual clinical practice. We suggest a different interpretation. Sometimes rapid diffusion is not desirable. Sometimes it takes time to fully understand the effects of a new screening policy within the context in which it is introduced. If AAA mortality is declining without much contribution from AAA screening; if EVAR is a technology that is hard to control; if AAA screening leads to substantial harms by labeling and overdiagnosis, then perhaps it is a good thing that the USPSTF will get a chance to reevaluate its positive recommendation before more and more people develop the habit of screening. The outcome this time may be different.
Correspondence: Dr Harris, Cecil G. Sheps Center for Health Services Research, 725 Martin Luther King Blvd, CB# 7590, University of North Carolina, Chapel Hill, NC 27599-7590 (email@example.com).
Published Online: September 17, 2012. doi:10.1001/archinternmed.2012.4280
Financial Disclosure: None reported.
Additional Information: Dr Harris was a member of the US Preventive Services Task Force in 2005, when it made its recommendation on screening for abdominal aortic aneurysm. He voted for the recommendation.
Harris R, Sheridan S, Kinsinger L. Time to Rethink Screening for Abdominal Aortic Aneurysm?Comment on “Impact of the Screening Abdominal Aortic Aneurysms Very Efficiently (SAAAVE) Act on Abdominal Ultrasonography Use Among Medicare Beneficiaries”. Arch Intern Med. 2012;172(19):1462-1463. doi:10.1001/archinternmed.2012.4280