Diagnosing coronary artery disease and cost of care

Can cost of testing and evaluation help select appropriate choices in health care? While it would obviously seem to be the case, showing that it is so is quite complicated. Over the last several decades, we have come to define tests by their sensitivity and specificity and their application in diagnosis by predictive value. Sensitivity and specificity define a test, while predictive value is dependent on disease prevalence. Then simple Bayesian statistics can use sensitivity, specificity, and pre-test prevalence to calculate predictive value. Apps for this purpose for hand-held devices are readily available. However, studies of the sensitivity and specificity of testing often suffer from biases (e.g. verification bias) and pre-test prevalence is generally unknown. Sensitivity and specificity also relate to each other, such that depending on the cutoff point for a positive test, as sensitivity rises specificity falls, a relationship known as a receiver operating characteristics (ROC). In addition, sensitivity is dependent on severity of disease and specificity on co-morbidity. Determining test characteristics also depends on there being an agreed to “gold standard”, which is often not agreed upon or is itself hard to measure. These problems make evaluation and use of testing difficult and often yield uncertain results. However, more fundamentally, these measures do not really tell you what you want to know, which is whether the decision on choice of testing can influence outcome. Outcome can be divided into clinical outcomes, whether events or health status, and resource utilization. Resource utilization can be scaled by applying cost weights to different resources and then totally these specific costs to simply calculate total cost. However, both clinical outcomes and cost pose difficulty in measurement. Another fundamental problem when assessing outcome of testing is that the influence of testing on subsequent tests and therapy may vary considerably. Outcomes, both clinical and economic, while difficult to measure initially, are even more so long-term, in the best of circumstances. When the influence of testing on subsequent care is uncertain, the assessment of outcome becomes that much more challenging. Nonetheless, billions of dollars are spent annually in the United States alone on testing to evaluate the presence and severity of obstructive coronary disease. In principle, we would like to know that we are getting good value for the money spent. However, this is generally unknown for reasons noted above. Thus, studies in the real world must address smaller issues, acknowledging that the larger questions cannot be easily answered.