| To Register SMDM Homepage | |
|
|
|
||||
Purpose: Various diagnostic imaging tests are available to diagnose coronary artery disease. The choice among these tests is difficult due to uncertainty about parameters, such as test characteristics and risk reductions of interventions. Our objective was to apply value of information (VOI) analysis to identify key parameter uncertainties – regarding the choice between imaging tests for coronary artery disease - for which future clinical research is justified.
Methods: We developed a probabilistic Markov model comparing cost-effectiveness from the health-care system perspective for four diagnostic tests for the diagnosis of coronary artery disease: exercise echocardiography, exercise single-photon-emission computed tomography (SPECT), computed tomographic angiography (CTA), and digital subtraction angiography (DSA). The expected value of perfect information (EVPI) was estimated per patient and subsequently for the entire USA, assuming an annual patient population of 250,000 for a duration of 5 years. Finally, the value of obtaining more information for particular (sets of) parameters was estimated. We considered 12 cohorts of patients defined by age, sex, and type and severity of chest pain.
Results: At a willingness-to-pay threshold of $50,000/QALY, CTA yielded the highest net health benefit, which was only 1 quality-adjusted life-day (QALD) greater than DSA - the next best strategy - and 40 QALD greater than not testing. The total EVPI was 2.5 QALD per patient. This implies that resolving all decision uncertainty is expected to improve net benefit with on average 2.5 QALD per patient. The population EVPI was $425 million. Uncertainty about the quality-of-life weights for varying severity of chest pain was responsible for 30% of the expected benefit. This is explained by the considerable uncertainty about these quality-of-life weights and their covariance structure, as well as their impact on false negative test results. Other important parameters were the prior probability of disease and the sensitivity and specificity of CTA. Resolving uncertainty about the probabilities of disease-related mortality and morbidity, other test characteristics, and health state transitions had a negligible expected benefit.
Conclusion: The total population EVPI is sufficiently large to justify gathering more evidence regarding the choice between these tests. Our results suggest that an observational study to obtain better estimates of quality-of-life weights, the prior probability of coronary artery disease, and test characteristics of CTA is the most useful next clinical study.
See more of Oral Concurrent Session B - Clinical Effectiveness and Quality of Life
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)