Journal of Molecular Medicine, 2019, Vol.97(6), pp.871-877
Byline: Thorsten Buch (1), Katharina Moos (2,3), Filipa M. Ferreira (1), Holger Frohlich (4), Catherine Gebhard (5), Achim Tresch (2,3) Keywords: Sex; Animal experimentation; Factorial design; Power Abstract: Abstract Disease occurrence, clinical manifestations, and outcomes differ between men and women. Yet, women and men are most of the time treated similarly, which is often based on experimental data over-representing one sex. Accounting for persisting sex bias in biomedical research is the misconception that the analysis of sex-specific effects would double sample size and costs. We designed an analysis to test the potential benefits of a factorial study design in the context of a study including male and female animals. We chose a 2x2 factorial design approach to study the effect of treatment, sex, and an interaction term of treatment and sex in a hypothetical situation. We calculated the sample sizes required to detect an effect of a given magnitude with sufficient power and under different experimental setups. We demonstrated that the inclusion of both sexes in experimental setups, without testing for sex effects, requires no or few additional animals in our scenarios. These experimental designs still allow for the exploration of sex effects at low cost. In a confirmatory instead of an exploratory design, we observed an increase in total sample sizes by 33%, at most. Since the complexities associated with this mathematical model require statistical expertise, we generated and provide a sample size calculator for planning factorial design experiments. For the inclusion of sex, a factorial design is advisable, and a sex-specific analysis can be performed without excessive additional effort. Our easy-to-use calculation tool provides help in designing studies with both sexes and addresses the current sex bias in preclinical studies. Key messages acents Both sexes should be included into animal studies. acents Exploratory study of sex effects can be conducted with no or small increase in animal number. acents Confirmatory analysis of sex effects requires maximum 33% more animals per study. acents Our calculation tool supports the design of studies with both sexes. Author Affiliation: (1) 0000 0004 1937 0650, grid.7400.3, Institute of Laboratory Animal Science, University of Zurich, Wagistrasse 12, 8952 Schlieren, Zurich, Switzerland (2) 0000 0000 8580 3777, grid.6190.e, Institute of Medical Statistics and Computational Biology, Faculty of Medicine, University of Cologne, Bachemer Str. 86, 50931, Cologne, Germany (3) 0000 0000 8580 3777, grid.6190.e, Center for Data and Simulation Science (CDS), University of Cologne, Cologne, Germany (4) 0000 0001 2240 3300, grid.10388.32, Bonn-Aachen International Center for IT (b-it), University of Bonn, Bonn, Germany (5) 0000 0004 1937 0650, grid.7400.3, Center for Molecular Cardiology, University of Zurich, Zurich, Switzerland Article History: Registration Date: 11/03/2019 Received Date: 01/12/2018 Accepted Date: 10/03/2019 Online Date: 13/04/2019 Article note: Electronic supplementary material The online version of this article ( https://doi.org/10.1007/s00109-019-01774-0) contains supplementary material, which is available to authorized users.
Sex ; Animal experimentation ; Factorial design ; Power
View full text in Springer (Subscribers only)