OBJECTIVES: Cost-effectiveness analysis (CEA) of a medical test can require extensive modeling if test results influence treatment decisions and disease progression. We applied the assessment hierarchy of Schaafsma et al. (2009) to a CEA of a high definition CT-scanner (one of the first assessments in the NICE/ UK Diagnostics Guidance Programme). METHODS: Schaafsma presents four steps to evaluate the value of diagnostic tests: 1:accuracy assessment, 2:evaluation of added value, 3:clinical outcome assessment and 4:cost-effectiveness analysis. RESULTS: In the assessment of diagnostic tests in coronary artery disease (CAD) modeling was unavoidable, since a major problem in CEAs is the unavailability of randomized controlled trials (RCT) that capture diagnosis, prognosis and treatment. Moreover, most RCTs in the field of diagnostic test yield only information about sensitivity, specificity, and short-term complication rates. When the evaluation is limited to the added value of the high definition CT-scanner, one model, estimating the proportions correctly diagnosed and complications associated with the CT-scanner, is sufficient. However, since incorrect test results can result in major health loss through incorrect or delayed treatment this method is inappropriate. Most tests aim to improve prognosis thus, step 3 and 4 were applied in our assessment. Therefore, five existing models were combined (diagnosis, CAD management, stroke complication, radiation, non-CAD mortality) to create a meta-model that estimates the cost-effectiveness. CONCLUSIONS: CEAs of a medical test can be performed in various ways described by Schaafsma. If the aim is to conduct a comprehensive analysis that includes various economic and health impacts, a synthesis of existing models to create a meta-model is one way to achieve this. These models need to be grafted together carefully to avoid invalid or irrelevant results; literature and expert opinion can assist in that endeavour. One critical pitfall is the use of models created for dissimilar patient populations.