Author's response to Spectraldomain optical coherence tomography in subjects over 60 years of age, and its implications for designing clinical trials

摘要

I would like to thank Comyn et al for their interest in our published article.1 I agree that different methodologies, different assumptions, or even analyses on different patient collectives might result in a different conclusion or a different sample size needed for randomised clinical trials.(i and ii) Power: the sample size calculation used with power of 80% was based on studies, such as the Age-Related Eye Disease Study trial.2 Using 90% power, a=0.05 and 10% loss to follow-up, I calculated once more the sample size needed for hypothetical studies (table 1).(iii) I integrated the findings from Comyn et al3 into our sample size calculations, (table 1), that is difference to detect in central retinal thickness (CRT)=30 +8 mum. Using the standard deviations in the described data from Comyn et al of 100 mum, a sample size of 83 or 264 is needed for a single- or double-arm studies. A hypothetical study using a device with less precision would, theoretically, further increase the standard deviation and, therefore, subsequently increase the sample size needed. In case of diabetic macular oedema and age-related macular degeneration, whereas there is a high degree of asymmetry between both eyes of the same individual, I agree that a single-arm study design might not be the most appropriate method. In other clinical trials, such as those listed as examples, they however do.