Cost-effectiveness analysis for imaging techniques with a focus on cardiovascular magnetic resonance

Abstract

With the need for healthcare cost-containment, increased scrutiny will be placed on new medical therapeutic or diagnostic technologies. Several challenges exist for a new diagnostic test to demonstrate cost-effectiveness. New diagnostic tests differ from therapeutic procedures due to the fact that diagnostic tests do not generally directly affect long-term patient outcomes. Instead, the results of diagnostic tests can influence management decisions for patients and by this route, diagnostic tests indirectly affect long-term outcomes. The benefits from a specific diagnostic technology depend therefore not only on its performance characteristics, but also on other factors such as prevalence of disease, and effectiveness of existing treatments for the disease of interest. We review the concepts and theories of cost-effectiveness analyses (CEA) as they apply to diagnostic tests in general. The limitations of CEA across different study designs and geographic regions are discussed, and we also examine the strengths and weakness of the existing publications where CMR was the focus of CEA compared to other diagnostic options.

Figure 1

Non-invasive imaging services provided by cardiologists per 1000 Medicare beneficiaries from 1999–2008. CMR accounts for a very small percentage of the total expenditures for cardiac imaging amongst Medicare beneficiaries. Data adapted from Andrus et al. [5].

Figure 2

Quality adjusted life years and utility. By convention perfect health is assigned a value of 1 and death is assigned a value of 0. Various health states such as myocardial infarction, percutaneous coronary intervention, coronary artery bypass surgery are assigned a utility weight which reflects the estimated effect on total health. Utility weights in this example were obtained from http://www.cearegistry.org.

Figure 3

A hypothetic decision analysis when 3 competing management options exist for a serious condition with a high disease prevalence of 40%. Management options include a) treat all without testing, b) imaging test first and treatment guided by imaging findings, and c) treat none. The only available treatment is associated with a complication (5%) and only a fraction (70%) of all affected patients will respond to the treatment. The imaging test available has positive and negative predictive values of both 90%. Note that in this decision tree, the decision node is indicated by a square and chance nodes are indicated by circles. Probabilities of outcomes that branch from a chance node always add up to 1. In this example, when one only considers probability of survival, folding back and averaging out the decision tree will yield an average survival rate of 84.3%, which compares favorably to 83.9% of imaging first and 68% treatment none. However, when QALY was considered (after adjustment for the poor quality of life in patients who suffered from treatment complications), the treat all option only yielded an average of 82.9% which was inferior to the imaging first option with an average of 83.6%. Note that in this example, complication from the imaging test itself was not modelled.

Figure 4

MRI units per one million populations across different European nations. Source: OECD [48].