Re: Compliance with requirement to report results on the EU Clinical Trials Register: cohort study and web resource
The article is really an eye opener as far as compliance with the requirement to report results on a trial register is concerned. It gives the wide spectrum of issues concerning results disclosure -- an issue which needs attention from registries operating in different parts of the globe. The article also tries to find an association of result disclosure with various other exploratory variables.
Though the article gives valuable information, there is some statistical concern. The computation of confidence interval of trials reporting results disclosure among 7274 trials in theEUCTR may not be very appropriate. We feel that this number 7274 is a population itself for EUCTR and not a sample for which confidence intervals are computed (1). Hence the number of trials reporting results is a parameter itself and not a ‘statistic’. Assuming this as a subset for the total trials registered in all registries, which to our understanding the authors intended, may not be reasonable as no population has been defined. Moreover the outcome of results disclosure may be quite different in various registries because of varying regulatory conditions prevailing in that region.
1. Sedgwick Philip. Understanding confidence intervals BMJ 2014; 349 :g6051
Competing interests: No competing interests