Introduction
What is new?
- •
Two recent examples of randomized clinical trials provide an accessible illustration of value of information (VOI) methods for assessing the current evidence regarding the decision to adopt a new health care intervention.
- •
Where the evidence in insufficient, the use of VOI methods for determining optimal sample size for future studies is illustrated, with reference to downloadable MS-Excel files for that purpose.
- •
Methods for examining the robustness of VOI solutions are illustrated.
- •
Researchers should consider using VOI methods for planning and analyzing clinical studies to optimize decision making and research planning.
In comparison to a cast, the clinical effectiveness of a removable brace has been demonstrated for children with acute, symptomatic low-risk ankle fractures [1]. Similarly, a prefabricated wrist splint was demonstrated to be at least as effective as casting for acceptably angulated distal radius greenstick or transverse fractures [2]. In both studies, the measure of effectiveness was recovery of physical function as measured by the Activities Scale for Kids—performance version (ASKp) [3]. The use of the standard hypothesis testing approach for analyzing and designing randomized clinical trials (RCTs) has been criticized for the arbitrariness of the levels used for the type I and II errors probabilities [4]. The same levels, 5% for type I errors and 20% for type II errors, are used for most trials, although the cost of such errors would vary considerably between trials. Furthermore, standard procedures do not determine if a clinical trial has been or will be cost-effective. Value of information (VOI) methods, based on Bayesian decision theory, have been proposed as an alternative. In the present article, VOI methods are applied to the data from the fracture studies [1], [2] with the purpose of illustrating how the methods can be used to determine if the evidence is sufficient for decision making. If the evidence is insufficient, VOI methods can determine the optimal sample size of the RCT needed to provide additional evidence. An additional objective of this article is to introduce into the clinical literature a decision-analytic approach for assessing cost-effectiveness data from RCTs. The decision-analytic approach provides an alternative to the standard hypothesis testing approach. Although VOI methods, as alternative to standard hypothesis testing, have been described elsewhere and are generally appreciated, for many researchers, they remain abstract and theoretical. The purpose of this article is to provide a more accessible introduction using two recently completed cost-effectiveness analyses as examples. These two examples are ideal because although they are similar in design and outcome, their VOI analyses differ with respect to the adequacy of evidence and optimal sample size for future research. We also illustrate a means for exploring the sensitivity with respect to important variables, such as time horizon and the threshold value for the for health outcomes, something not covered particularly well in the current literature.
Details of the both studies are given in the Methods section, along with an introduction to VOI methods. In the Results section, the methods are illustrated using the evidence from the studies. Further information is provided in the Discussion section.