Dear all,
I have been asked to perform a sample size calculation for a study that differs somewhat from my normal RCT fare. The investigators wish to compare the performance of two different screening tests for TB, in 3 different patient groups. One group is treatment-naïve, the other two groups have been receiving different drug treatments. The investigators have data on the sensitivity and specificity of each screening test in treatment-naïve patients (relative to population risk for latent TB as there is no gold standard). Sensitivity for test A is 90% and for test B it is 97%. Specificities for the two tests are 90% and 100% respectively. The investigators suspect that the drug treatments under investigation may affect the ability of test B to discriminate accurately, therefore they want to see which screening test performs best in each treatment group. Screening test B is more expensive than test A. I'm currently having difficulty grasping exactly how I should calculate sample size here. I think I need to ask the investigators how much better test B would need to be than test A for them to consider recommending its use given the extra expense, then power according to this difference in each group of patients (if this difference is not achieved then we conclude that test B is not sufficiently outperforming test A), but I need to take prevalence into account somehow (prevalence is very low, around 3.5%). I would be grateful for any advice you could give me.
Thanks,
Liz Hensor
Dr Elizabeth M A Hensor PhD
Data Analyst
Academic Unit of Musculoskeletal and Rehabilitation Medicine
36 Clarendon Road
Leeds
West Yorkshire
LS2 9NZ
Tel: +44 (0) 113 3434944
Fax: +44 (0) 113 2430366
[log in to unmask]
|