RESUMO
In the assessment of the safety of a new drug, a large battery of toxicological studies is performed. In toxicological studies it is common practice to analyse the data from the sexes separately. We argue that this is not best practice and that much is to be gained from combining data from both sexes when evaluating studies. The main benefits and arguments in favour of combining the data are: (i) Improved efficiency, allowing fewer test animals and/or better precision. (ii) Other phases of drug development, clinical trials in particular, combine sexes. (iii) Most substances act similarly on both sexes and most drugs are prescribed without sex differentiation. (iv) If sex differences are of interest, combined data facilitates the quantification of the difference. (v) Reduced number of false positive and false negative findings. Although there might be some grounds to analysing the sexes separately, we argue that these are comparatively minor. We do not promote simply pooling the data and neglecting the existence of two sexes when comparing treatment groups. Rather, the analysis of combined data should allow for sex differences, for instance by using stratified procedures or by including a sex factor in a statistical model. We illustrate from real study data where the approach we propose has substantial benefits over traditional methods. A theoretical illustration is provided to quantify mathematically the potential gains and the corresponding sample size reduction, i.e. a reduction of animal use of ca. 50%, that would be possible without impairing the precision of studies.
