Abstract Archives of the RSNA, 2004
Bas Groot Koerkamp MD, Abstract Co-Author: Nothing to Disclose
Jeroen J. Nikken MD, Presenter: Nothing to Disclose
Theo Stijnen PhD, Abstract Co-Author: Nothing to Disclose
Abida Z. Ginai MD, Abstract Co-Author: Nothing to Disclose
Myriam Hunink PhD, Abstract Co-Author: Nothing to Disclose
If an RCT has been used to identify the optimal strategy, we are generally left with the decision whether further research should be performed to decrease uncertainty about the optimal strategy. More research is justified only if the expected benefit of information exceeds the estimated research costs. Of all analyses assessing uncertainty, only value of information analysis can quantify the cost of uncertainty and therefore the expected net health benefit of performing further research. The purpose of this study was to illustrate the usefulness of value of information analysis to direct further research concerning the use of MRI for acute knee trauma.
In an RCT for patients with acute knee injury, the effect of MRI versus MRI only in the absence of fracture versus no MRI on overall costs was compared from a societal perspective. We developed a probabilistic cost-benefit model including time and costs associated with lost productivity. We calculated the total expected value of perfect information (EVPI), population EVPI, and the partial EVPI for several parameters, using a two-level Monte Carlo simulation.
The total EVPI of all parameters was 83 euro, reflecting an expected reduction in costs, if all parameters are known exactly, of 83 euro on a total cost of 1621 euro per patient for the (non-significantly optimal) selective MRI strategy. With a population to benefit in the Netherlands of about 20,000 patients per year, population EVPI is 7.6 million euro. This relatively high population EVPI justified the exploration of partial EVPI. Only four parameters (out of 109) together were responsible for 83% of the benefit to gain with more research. These parameters represent the number of days off work due to the knee injury for each of the three strategies and the number of hospital days for the intervention strategy.
This application illustrates the usefulness and practicality of Value of Information analysis to guide further research. Instead of performing another trial to gather better estimates of all 108 parameters, our results suggest that the next study should focus on obtaining better estimates for the number of days off work for each strategy.
Groot Koerkamp, B,
Nikken, J,
Stijnen, T,
Ginai, A,
Hunink, M,
Value of Information Analysis Directing Further Research: Assessing the Use of MRI for Acute Knee Trauma. Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL.
http://archive.rsna.org/2004/4404255.html