|Home | About | Journals | Submit | Contact Us | Français|
Noninvasive imaging of atherosclerosis is being increasingly used in clinical practice, with some experts recommending to screen all healthy adults for atherosclerosis and some jurisdictions mandating insurance coverage for atherosclerosis screening. Data on the impact of such screening have not been systematically synthesized.
We aimed to assess whether atherosclerosis screening improves cardiovascular risk factors (CVRF) and clinical outcomes.
This study is a systematic review.
We searched MEDLINE and the Cochrane Clinical Trial Register without language restrictions.
We included studies examining the impact of atherosclerosis screening with noninvasive imaging (e.g., carotid ultrasound, coronary calcification) on CVRF, cardiovascular events, or mortality in adults without cardiovascular disease.
We identified four randomized controlled trials (RCT, n=709) and eight non-randomized studies comparing participants with evidence of atherosclerosis on screening to those without (n=2,994). In RCTs, atherosclerosis screening did not improve CVRF, but smoking cessation rates increased (18% vs. 6%, p=0.03) in one RCT. Non-randomized studies found improvements in several intermediate outcomes, such as increased motivation to change lifestyle and increased perception of cardiovascular risk. However, such data were conflicting and limited by the lack of a randomized control group. No studies examined the impact of screening on cardiovascular events or mortality. Heterogeneity in screening methods and studied outcomes did not permit pooling of results.
Available evidence about atherosclerosis screening is limited, with mixed results on CVRF control, increased smoking cessation in one RCT, and no data on cardiovascular events. Such screening should be validated by large clinical trials before widespread use.
Several techniques of noninvasive vascular imaging have been proposed in recent years as tools for detecting asymptomatic atherosclerosis without overt cardiovascular disease (CVD).1–3 Some measures of atherosclerosis have been shown to improve the prediction of coronary heart disease (CHD) events.4–7 However, good prediction does not necessarily lead to effective prevention.8,9 Among the important criteria to assess their potential clinical value, novel markers of cardiovascular risk should be assessed by their effect on patient management and outcomes.10
Screening for atherosclerosis and knowing the test result might help enhance patient motivation to change unhealthy behaviors,11,12 such as smoking, or improve lifestyle and adherence to medications. In 2003, a review for the U.S. Preventive Services Task Force (USPSTF) identified only two studies that examined the impact of coronary artery calcification [CAC, by computed tomography (CT)] screening on risk-reducing behaviors.13 These two studies were not randomized,14,15 and behavior outcomes were self-reported. Since this publication, new studies16–18 or studies with other screening techniques11 have been published.
A substantial amount of controversy remains about the role of screening for atherosclerosis in the clinical setting.19 Some groups have suggested that imaging should be considered only in certain subgroups of patients, while others advocate a more widespread application (Table 1). Some experts have even suggested screening most healthy adults for atherosclerosis.3 Such screening is being increasingly used in clinical practice to screen asymptomatic adults.12,13,20–22 Moreover, insurance coverage for atherosclerosis screening with carotid ultrasound is under consideration;23 the Texas Legislature passed a new law, effective Jan 1, 2010, mandating broad insurance coverage for atherosclerosis screening, which could lead to a considerable increase in screening.19 Therefore, a systematic review was performed to assess whether screening for atherosclerosis improves cardiovascular risk factors (CVRF) and clinical outcomes.
Studies had to meet four inclusion criteria: (1) the study population had to be adults without preexisting CVD [myocardial infarction, stroke, transient ischemic attack, or peripheral arterial disease (PAD)]; (2) imaging of atherosclerosis had to be done by noninvasive techniques,20 such as ultrasonography (carotid intima–media thickness or carotid plaques), conventional abdominal radiography (aortic calcifications), CT (CAC), ankle/brachial index (ABI), or flow-mediated brachial artery endothelial vasodilatation to assess endothelial dysfunction; (3) studies had to report follow-up clinical events, such as all-cause or cardiovascular (CV) mortality, acute coronary syndrome (ACS), stroke, documented PAD, or intermediate clinical outcomes, such as CVRF control (smoking cessation, blood lipid control, diet change, weight reduction, or physical activity improvement) or changes in health behavior (e.g., increased motivation to change lifestyle, increased perception of cardiovascular risk, adherence to medication); and (4) the intervention group had to be compared with a control group of adults who did not receive the screening intervention (randomized controlled trials, RCTs) or a control group who did not have evidence of atherosclerosis upon screening, as reported in available non-randomized studies; these non-randomized studies did not report comparisons with a control group of adults who were not screened for markers of atherosclerosis. The search strategy was not limited to RCTs because of the likely low numbers of such studies13 and to assess whether observational studies showed patterns that might be consistent with potential impact of atherosclerosis screening. Studies that assessed the impact of surgical procedures (e.g., screening for aortic aneurysm followed by surgery) were excluded. The numerous studies that assessed risk prediction of atherosclerosis markers5 or that assessed cardiac stress imaging tests were not included as such studies have been previously reviewed.24–27 We did not include studies that only examined changes in prescription rates of CV preventive medications (e.g., aspirin, statins) after atherosclerosis screening since prescription rates are mainly related to physicians’ decisions; the aim of the present systematic review was to assess whether the results of screening given to patients lead to increased motivation12 (e.g., lifestyle changes, adherence rates to CV preventive medication), with subsequent improvement in CVRF control and clinical outcomes. If follow-up outcomes other than prescription rates were reported, such studies were not excluded.
The primary outcomes were clinical events, such as all-cause and CV mortality, ACS, stroke, documented PAD, or a change in CVRF control at follow-up (i.e., differences in smoking cessation rates, lipid, and blood pressure changes). Secondary outcomes were modification in adherence rates to CV preventive medication, risk perception of developing heart disease, quality of life, and changes in health behavior (e.g., increased motivation to change lifestyle, physical activity, diet).
MEDLINE (1966 through April 2009) and the Cochrane Controlled Clinical Trial register (1996 through April 2009) were searched using a recommended approach for systematic reviews of RCTs with a predefined search strategy.28 All languages were considered eligible. For MEDLINE, three comprehensive search themes were combined using the Boolean operator “and.” The first theme representing the patient population of interest was created using the following terms which appeared as exploded MeSH headings—CV diseases/prevention and control—or as text words—coronary or aortic or aorta. The second theme for the screening method of interest used the following terms—magnetic resonance angiography or radiography, abdominal or radiography, thoracic or tomography, x-ray, computed or ultrasonography, or ankle/blood supply or brachial artery/blood supply—or the terms appearing as text words—computed tomography or ultrasonography or ankle–brachial index, or flow-mediated vasodilatation. The third theme, which defined the outcome of interest, was created using the exploded MeSH headings: counseling or preventive health services or motivation or behavior, including smoking cessation, as well as coronary disease and mortality for RCTs. For the Cochrane Controlled Clinical Trial register, a similar search strategy used text words, for which the comprehensive search themes “patient population,” “screening method,” and “studied outcome” were combined using the Boolean term “and.”
In a two-step selection process, two investigators (RA and VdB) independently reviewed the titles and abstracts of all citations to identify studies meeting the inclusion criteria. When in doubt about eligibility on the first screen, full-text articles were obtained. Articles selected by either both or only one author were selected for full-text review. In addition, one investigator (RA) screened the reference list of identified studies on the first screen and major reviews on the topic for other potentially relevant studies.12,13,29–31 The same two investigators reviewed the relevant reports in full text for eligibility and independently extracted data from all studies fulfilling eligibility criteria. Data extraction included characteristics of the screening intervention, type of study, baseline clinical characteristics of the participants, and relevant outcomes. Authors of the studies were contacted for additional information when needed.32
Study design was considered as the primary study quality measure;33 studies with a control group that did not receive the screening intervention (RCTs) were considered stronger than those with a control group that had the screening intervention but did not present evidence of atherosclerosis (non-randomized studies). For the RCTs, the Jadad quality score was reported, adapting it to the present situation;34 blinding of study participants to intervention was not included as this is difficult to achieve in studies of lifestyle interventions. Items used to assess study quality were: methods of randomization (two points) and reporting of losses to follow-up (one point), thus leading to a maximum score of three points.
A flowchart summarized the number of trials identified, excluded and included. Due to major variations in both screening methods and studied outcomes, the results were not combined in a meta-analysis and only individual study results were reported.
A total of 2,634 unique citations were identified, including 1,783 from MEDLINE and 851 additional from the Cochrane Controlled Clinical Trial register (Fig. 1). After a two-step screening process, 12 articles fulfilled the inclusion criteria.11,14–18,32,35–39 Using the abstract and title on the first screen, 13 disagreements concerning eligibility for a full-text review occurred between the two reviewers (kappa=0.81). For the second screen, which was based on the full-text review of 41 studies, there was one disagreement (kappa=0.94). This disagreement was resolved by consensus.
Four RCTs included a total of 709 participants.11,16,32,35 Heterogeneity in screening methods and the studied outcomes did not allow pooling of results or description of overall results. Semiquantitative assessments of the study results are described in Table 2. No RCTs examined the impact of screening on CV events or mortality. Overall, screening for atherosclerosis did not improve CVRF, except for increased smoking cessation rates (18% vs. 6%, p=0.03) in one RCT;11 other RCTs included only few smokers (Table 3). Two RCTs found no improvement on quality of life.
In 153 smokers randomly allocated to carotid ultrasound, providing pictures of their own atherosclerotic plaques improved the rates of smoking cessation from 18% in the screened group vs. 6% in the group without screening (p=0.03, Table 3) in addition to brief advice for smoking cessation. The absence of biochemical validation of smoking cessation was the major limitation of this trial. It was undertaken in the Seychelles islands, in a population with low nicotine dependence (mean: ten cigarettes/day, about half the amount as in Western countries). No other RCTs that evaluated the effect of carotid ultrasound atherosclerotic plaque screening on risk-reducing behavior were found.
This RCT included 450 asymptomatic active-duty US Army personnel aged 39–45 years.16 CAC screening by CT did not lead to improved CVRF control or change in the Framingham Risk Score (Table 3). This study has the highest Jadad quality score (3/3). Screening had little effect on smoking rates (5/13 in intervention vs. 4/17 in the control group), but few smokers were randomized. Among its limitations, the study participants were young adults with a low 10-year CHD risk of 6% and a low prevalence of CAC (15%).12
A subgroup analysis showed that the presence of coronary calcifications (“risk marker”) might be a motivating factor. Among those with CAC (n=59), those who learned that they had calcifications had a smaller, albeit non-statistically significant, increase in the 10-year Framingham Risk Score compared with those who did not receive such information (0.21% vs. 1.52% increase in risk score, p=0.13).16 Another potential limitation of this trial was the 2×2 factorial design with two interventions (information on calcification vs. no information intensive vs. usual care) as this may have had an effect on the same outcomes; this might have biased the results toward the null hypothesis.40
This study of 56 postmenopausal women randomized to either CAC screening plus a counseling session or the control group (conventional counseling) found that the intervention CT group had less improvement in LDL-cholesterol compared with the control group (Table 3). The high proportion of women with low CAC score (mean score=1.37) might partly explain the results as these women screened for CAC might have been reassured that they were not at high risk of CHD.
In a pilot study, 50 white adults self-referred from advertisements were randomized to total body scan with CAC screening or no screening and followed for 2 years. Among 3 of the 25 screened subjects who had a CAC score ≥100, cardiac catheterization showed three-vessel occlusive disease in one patient who subsequently underwent coronary bypass grafting; the two others had normal cardiac stress testing and no clinical events. No other participant had clinical events during follow-up, but the small sample size limits the interpretation of the results. Self-reported quality of life measured by the SF-36 did not differ between screened and control subjects over time, except for one of the eight subscales (fewer role limitation due to physical health problem in screened subjects).
Two thousand nine hundred ninety-four participants were included in eight non-randomized prospective or retrospective studies comparing participants with evidence of atherosclerosis on screening to those without (Table 4). No studies examined the impact of screening on CV events or mortality or on quality of life. These studies assessed the impact of atherosclerosis screening on patient behaviors and CVRFs. Heterogeneity in screening methods and studied outcomes did not permit pooling of results. Semiquantitative and detailed assessments of the study results are described in Tables 2 and and4.4. Four studies found increased risk perception of developing CHD. Four studies found self-reported improvement in diet. However, two RCTs did not show an effect on improvement in diet. One study found improvement in blood lipid levels, and two of three studies showed increased motivation to make lifestyle changes. One study found increased motivation for smoking cessation in smokers with carotid plaques compared to those without (8.7 vs. 7.2/10, p=0.008) and a pattern of higher quit rates (73% vs. 38%, p=0.10);36 the three other studies did not find improved smoking cessation rates. Physical activity did not differ between the two groups, except in one study. These behaviors were all self-reported. Concerning drug adherence, one study found that individuals with CAC scores in the fourth quartile were more likely (multivariate-adjusted OR=9.26, 95%CI=4.13–20.76) to continue statins compared to those with CAC scores in the first quartile.17 However, no formal measurement of adherence such as drug dispensers with electronic monitoring or pill count was performed. The study populations were often not population-based as participants were frequently referred for atherosclerosis screening by their general practitioner.
This systematic review found that available evidence about atherosclerosis screening using noninvasive imaging was limited and yielded mixed results. In the RCTs, screening for atherosclerosis did not improve CVRF, but an increased smoking cessation rate (18% vs. 6%, p=0.03) was found in a single RCT.11 Non-randomized studies showed potential positive effects of atherosclerosis screening on “intermediate” outcomes, such as increased motivation to change lifestyle36,37 and an increased perception of CV risk.14,18,37,39 However, such data were based on self-report and limited by the lack of a randomized control group. We found no studies that evaluated the impact of screening on CV events or mortality. These results are important in the context of substantial controversy about the role of screening for atherosclerosis.19
Consistent with the results of atherosclerosis screening on smoking cessation, mainly derived from the RCT by Bovet et al.,11 a recent Cochrane review evaluated the impact of visual feedback of medical images in changing health behavior41 and reported a statistically significant increase in smoking cessation behaviors (OR=2.81, 95% CI=1.23–6.41) after pooling the data from three studies.11,16,42 Because of the exclusion of studies in patients with preexisting CVD, the present review did not include the study by Shahab et al.42 and did not pool data on smoking cessation behaviors because of clinical heterogeneity (different timing for assessment of smoking cessation behaviors, very few smokers in one RCT).16 These data on smoking cessation behaviors require confirmation with a larger RCT that includes smokers with higher daily cigarette consumption than in the Seychelles islands,11 as well as a biochemical validation of smoking cessation, one being currently performed.43
Increased cardiovascular risk perception after atherosclerosis screening was found, consistent with recent systematic review findings that receiving global CHD risk information increased the accuracy in CHD risk perception.44 However, no study compared the incremental effectiveness of providing feedback on atherosclerosis imaging in addition to global CHD risk. Moreover, data from the present review showing increased CHD risk perception were all derived from non-randomized studies. Overall, other results are also consistent with this recent Cochrane review on feedback of medical imaging described above.41 This review (that did not specifically examine atherosclerosis screening) found mixed results concerning the impact of visual feedback of images on health behavior. Risk perception and clinical events were not assessed.
What are potential harms of atherosclerosis imaging? One study showed an increase in anxiety levels after such screening,15 which was not found in two other studies16,43 (Tables 3 and and4).4). Another study found that atherosclerosis screening may result in subsequent invasive testing and increased healthcare utilization.32 However, other potential harms of atherosclerosis screening, such as radiation exposure and subsequent malignancy,8 were not assessed in the reviewed studies. Another potential harm might be false reassurance, with the pursuit of unhealthy lifestyle. The slightly lower smoking cessation rate in those without plaques (5%) compared with the non-screened group (6%) in one RCT11 might be related to false reassurance, although data are conflicting on the impact of false reassurance after a negative screening test.45
The major limitation of the present systematic review, inherent to the available studies, is the small number of available RCTs. The reported results mainly rely on data from non-randomized studies, these studies showing more positive results than RCTs, except for smoking cessation. There was also clinical heterogeneity in the screening methods used and the studied outcomes. The pooling results in this case seemed inadequate. For smoking cessation, no studies provided a confirmation of smoking cessation by biochemical validation, as recommended.46 These limitations were inherent to the available studies and confirm the need for more RCTs in this field.47 No studies examined the impact of screening on CV events or mortality. This is likely related to the need of a very large sample size for such trials; the NHLBI working group has estimated that a trial on the impact of such screening on clinical cardiovascular events would likely require >10,000 participants.20
What are the potential clinical and research implications of these findings? The mixed findings and limited data on atherosclerosis screening are important in the context of controversy about the role of screening for atherosclerosis,19 with controversial recommendations (Table 1). Recognizing the absence of current evidence for improved net health outcomes from atherosclerosis screening, the ACC/AHA has recently suggested that it may be reasonable to measure carotid intima–media thickness (IMT) or CAC among asymptomatic adults at intermediate CHD risk.48 The USPSTF did not recommend any of the available imaging modalities.49 It is common that new, often expensive, technologies do not undergo formal evaluation prior to being implemented within standard of care.50,51 Population-wide screening might concern a large population, with substantial public health and cost implications.23 For example, the Society of Cardiovascular Computed Tomography estimates that more than 200,000 Americans had CAC screening in 2008 at a cost of about US $50 million.22 Such screening might become a profitable business, including sometimes by using patients’ concerns as the motor for screening.52 Cost effectiveness studies should assess whether costs associated with atherosclerosis screening23 and potential long-term harms, such as radiation exposure and subsequent malignancy,8 might be outweighed by potential benefits, such as increased smoking cessation and improved adherence (albeit with limited data), more effective targeting of preventive therapy to those who really need it, or a reduction in major cardiovascular events.47 Modeling might help determine the screening procedures and the target population to design future large-scale, expensive RCTs.47 Given limited RCT data, scientific societies should make cautious recommendations on extensive use of atherosclerosis screening.
In summary, this systematic review shows that available evidence about atherosclerosis screening using noninvasive imaging is limited, with mixed results on CVRF control and increased smoking cessation in a single RCT. Absence of proof of benefit is certainly not proof of absence of benefit. However, the potential advantages of atherosclerosis screening need to be demonstrated by large-scale RCTs.8,50 Not conducting such trials would leave clinicians with no scientific basis for making decisions regarding newly proposed, and sometimes expensive, methods for identifying high CHD risk adults.20,47 Such trials should likely target intermediate-risk adults, as suggested by others,1,20,47 and/or those with a high likelihood of atherosclerosis,12 given the pattern of more benefits of atherosclerosis screening in the small group with calcification in the highest quality trial.16 Such trials should assess the impact of atherosclerosis screening on relevant clinical outcomes, including cardiovascular risk factors, smoking cessation, and, ideally, cardiovascular events,47 before its widespread implementation.
Dr. Rodondi’s and Dr. Cornuz’s researches on atherosclerosis screening, as well as a randomized controlled trial of atherosclerosis screening on smoking cessation (PI: Dr N. Rodondi), are supported by grants from the Swiss National Science Foundation (SNSF 3200B0-116097), the Swiss Tobacco Prevention Funds (Federal Office of Public Health, FPT 08.002282), and the Swiss Heart Foundation. Dr. Auer’s research on cardiovascular prevention is supported by grants from the Swiss National Science Foundation (SNSF SPUM 33CM30-124112 and PBLAP3-136774).
Conflict of Interest None disclosed.