Integration of meta-analysis and economic decision modeling for evaluating diagnostic tests
The creation of structures in the UK (i.e. National Institute for Health and Clinical Excellence) and worldwide to facilitate evidence-based health policy decision-making has highlighted the role that systematic reviews including, where appropriate, meta-analysis, and economic evaluations have to play in the decision-making process. These methodologies provide answers to fundamental questions such as: Does the technology work, for whom, at what cost, and how does it compare with alternatives? In the area of diagnostic test performance such evidence-based evaluations are crucial to the decision making process as early diagnosis can lead to diseases being treated more successfully than if treatment were delayed.
Although policy relevant questions relating to diagnostic tests could often be evaluated through randomised controlled trial designs, such an approach would be prohibitively expensive and time-consuming in many situations. For this reason, decisions relating to the use of diagnostic tests are often evaluated through decision modelling methods. Where multiple sources of evidence exist which relate to the decision model parameters it is important that this evidence is synthesised appropriately. Evidence synthesis of diagnostic test accuracy data is more complicated than for intervention studies due to additional issues relating to variable test threshold levels, dependence between outcomes (i.e. sensitivity and specificity) and use of multiple tests in combination. A number of alternative evidence synthesis models exist ranging from naive synthesis of independent sensitivity and specificity through to optimisation of a hierarchical summary receiver operating characteristic (HSROC) curve.
To date, there has been little research into how the results obtained from evidence synthesis of diagnostic tests interfaces with corresponding economic decision modelling to address policy questions such as, “At which threshold value is the test most beneficial/cost-effective?” The objective of this presentation is to outline a single coherent framework for synthesising the diagnostic test accuracy data, inputting it directly into the decision model together with its corresponding uncertainty, and evaluating the decision model. The proposed method is applied to the example of deep vein thrombosis and the results are presented in terms of receiver operating characteristic (ROC) curves and cost-effectiveness acceptability curves.