EXPECTED VALUE OF INFORMATION ANALYSIS TO EVALUATE THE POTENTIAL GAINS OF AN RCT COMPARING WOUND DRESSINGS

Author(s)

Welton NJ¹, Blazeby JM¹, Reeves B¹, Rogers C¹, Dumville J²
¹University of Bristol, Bristol, UK, ²University of Manchester, Manchester, UK

Presentation Documents

PRM237--strong-u-welton-nj-u-sup-1-sup-strong-blazeby-jm-sup-1-sup-reeves-b-sup-1-sup-rogers-c-sup-1-sup-dumville-j-sup-2-sup-br-sup-1-sup-university-of-bristol-bristol-uk-sup-2-sup-university-of-manchester-manchester-uk

OBJECTIVES:

A variety of wound dressings are available for general surgery, categorised as Simple, Advanced, Glue and no dressing (Exposed). These differ in costs and may differ in surgical site infection (SSI) risk. This study aims to explore the use of value of information methods to assess if a new RCT is a cost-effective use of resources, to compare the value of different study designs, and compare results with standard sample size calculations.

METHODS:

We develop a simple decision tree model to compare the costs and quality of life resulting from different wound dressing types, which may differ in SSI risk. We use results from a network meta-analysis to inform the relative efficacy of the different dressing types. Expected Value of Sample Information (EVSI) is computed to evaluate the value of reducing uncertainty by running an RCT for various included interventions and sample size. Population EVSI is presented for 1.208m wounds over a 5-year time horizon. We compare the results with those from standard sample size calculations.

RESULTS:

There was considerable uncertainty as to which dressing type is the most cost-effective. Population EVSI indicates that an RCT comparing dressings is likely to be of value and that designs that compare Simple Dressings vs Glue have much higher value than designs that do not make this comparison. Balanced designs comparing Simple Dressings vs Glue vs Exposed have population EVSI of £1515m, £2057m, and £2150m for total sample sizes of 500, 2500, 5000 resp. Standard sample size calculations suggest a much larger sample (around 25000).

CONCLUSIONS:

We discuss possible reasons for obtaining a smaller sample size with the value of information approach compared to standard sample size calculations and implications for trial design.

Conference/Value in Health Info

2017-11, ISPOR Europe 2017, Glasgow, Scotland

Value in Health, Vol. 20, No. 9 (October 2017)

Code

PRM237

Topic

Study Approaches

Disease

Multiple Diseases

Presentation