Objectives: To assess the clinical and cost-effectiveness of computed tomography (CT) screening for asymptomatic coronary artery disease; also to establish whether coronary artery calcification (CAC) predicts coronary events and adds anything to risk factor scores, and whether measuring CAC changes treatment.

Data sources: Main electronic databases were searched up to 2005, with a MEDLINE update in February 2006.

Methods: A systematic review of screening studies and economic evaluations was carried out. Studies were included in the review if screening for coronary heart disease was the principal theme of the study, and if data were provided that allowed comparison of CT screening with current practice, which was taken to be risk factor scoring. Mismatches between CAC scores and risk factor scoring were of particular interest. A review of the case for screening against the criteria used by the National Screening Committee (NSC) for assessing screening programmes was also undertaken.

Results: No randomised control trials (RCTs) were found that assessed the value of CT screening in reducing cardiac events. Seven studies were identified that assessed the association between CAC scores on CT and cardiac outcomes in asymptomatic people and included 30,599 people. Six used electron-beam CT. The relative risk of a cardiac event was 4.4 if CAC was present, compared to there being no CAC. As CAC score increased, so did the risk of cardiac events. The correlation between CAC and cardiac risk was consistent across studies. There was evidence that CAC scores varied among people with the same Framingham risk factor scores, and that within the same Framingham bands, people with higher CAC scores had significantly higher cardiac event rates. This applied mainly when the CAC scores exceeded 300. There was little difference in event rates among the groups with no CAC, and scores of 1-100 and 101-300. In one study, CAC score was a better predictor of cardiac events than the Framingham risk scores. No studies were found that showed whether the addition of CAC scores to standard risk factor assessment would improve outcomes. There were reports from two observational studies that lowering of low-density lipoprotein cholesterol to about 3 mmol/l; or below with statin treatment modestly reduced CAC scores, but this was not confirmed in two RCTs. In three studies examining whether knowledge of CAC scores would affect compliance with lifestyle measures, perception of risk was affected, but it did not improve smoking cessation rates, although it did increase anxiety. There were a few economic studies of CT screening for heart disease, which provided useful data on costs of scans, other investigations and treatment, but relied on a number of assumptions, and were unable to provide definitive answers. One modelling study estimated that adding CT screening to risk factor scoring, and only giving statins to those with CAC score over 100, would save money, based on a cost per CT screen of US$400 and statin costs of US$1000 per annum per patient. However, the arrival of generic statins has reduced the price dramatically, and these savings no longer apply.

Conclusions: CT examination of the coronary arteries can detect calcification indicative of arterial disease in asymptomatic people, many of whom would be at low risk when assessed by traditional risk factors. The higher the CAC score, the higher the risk. Treatment with statins can reduce that risk. However, CT screening would miss many of the most dangerous patches of arterial disease, because they are not yet calcified, and so there would be false-negative results: normal CT followed by a heart attack. There would also be false-positive results in that many calcified arteries will have normal blood flow and will not be affected by clinically apparent thrombosis: abnormal CT not followed by a heart attack. For CT screening to be cost-effective, it has to add value over risk factor scoring, by producing sufficient additional information to change treatment and hence cardiac outcomes, at an affordable cost per quality-adjusted life-year. There was insufficient evidence to support this. Most of the NSC criteria were either not met or only partially met. It would be useful to have more data on the distributions of risk scores and CAC scores in asymptomatic people, and the level of concordance between risk factor and CAC scores, the risk of cardiac events per annum according to CAC score and risk factor scores, information on the acceptability of CT screening, after information about the radiation dose, and an RCT of adding CT screening to current risk factor-based practice.