preclinical research

/Tag:preclinical research

The Use of Gatekeeping Procedures in the Statistical Planning of Animal Experiments

Benjamin Mayer, Vicky Stahl and Martina Kron

Statistical sample size calculation is essential when planning animal experiments in basic medical research. Usually, such trials involve the testing of multiple hypotheses, and interpreting them in a confirmative manner would require the appropriate adjustment of the Type 1 error. This has to be taken into account as early as possible during sample size estimation — otherwise, all the results obtained would be
exploratory, i.e. without cogency. In this paper, the concept of gatekeeping is introduced, along with alternative approaches for Type 1 error adjustment. The application of gatekeeping to the calculation of sample size is demonstrated by using data sets from case studies. Overall, the evaluation of these examples showed that gatekeeping is able to keep the required number of animals comparatively small. In contrast to exploratory planning, which led to the lowest sample sizes, gatekeeping suggested a mean increase of 12% in sample size, while conservative Bonferroni adjustment raised the sample size by 34% on average. Gatekeeping is a prominent strategy for handling the multiple testing problem, and has been proven to keep the required sample sizes in animal studies comparatively low. Therefore, it is a suitable approach to a compromise between the Three Rs principle of reduction and the appropriate handling of the multiplicity issue in animal trials with a confirmative focus.

This article is currently only available in full to paid subscribers. Click here to subscribe, or you will need to log in/register to buy and download this article

Sample Size Estimation for Pilot Animal Experiments by Using a Markov Chain Monte Carlo Approach

Andreas Allgoewer and Benjamin Mayer

The statistical determination of sample size is mandatory when planning animal experiments, but it is usually difficult to implement appropriately. The main reason is that prior information is hardly ever available, so the assumptions made cannot be verified reliably. This is especially true for pilot experiments. Statistical simulation might help in these situations. We used a Markov Chain Monte Carlo (MCMC) approach to verify the pragmatic assumptions made on different distribution parameters used for power and sample size calculations in animal experiments. Binomial and normal distributions, which are the most frequent distributions in practice, were simulated for categorical and continuous endpoints, respectively. The simulations showed that the common practice of using five or six animals per group for continuous endpoints is reasonable. Even in the case of small effect sizes, the statistical power would be sufficiently
large (≥ 80%). For categorical outcomes, group sizes should never be under eight animals, otherwise a sufficient statistical power cannot be guaranteed. This applies even in the case of large effects. The MCMC approach demonstrated to be a useful method for calculating sample size in animal studies that lack prior data. Of course, the simulation results particularly depend on the assumptions made with regard to the distributional properties and effects to be detected, but the same also holds in situations where prior data are available. MCMC is therefore a promising approach toward the more informed planning of pilot research experiments involving the use of animals.

This article is currently only available in full to paid subscribers. Click here to subscribe, or you will need to log in/register to buy and download this article