Calculating the number of study participants is an essential part of clinical research. If the number of study participants is too small, study objectives cannot be fulfilled. On the other hand, an overly large number of study participants can result in a waste of resources. But how many participants does it take for a trial to draw meaningful and conclusive results?
There are two methods to calculate this number, which is also called the sample size:

• sample size calculation based on hypothesis testing (traditionally used)
• precision-based sample size calculation.

The precision-based sample size calculation method is more in line with the current trends towards presenting study results in the form of confidence intervals and away from p-values (Bland 2009). However, even if most statistical software packages contain some sample size calculation algorithms, they are mostly based on hypothesis testing.

To close this gap, statisticians at the CTU Bern have developed the presize R package, which provides a range of functions for performing precision-based sample size calculations. It includes methods for descriptive statistics, absolute and relative differences, various measures of correlation, and diagnostic measures.

The tool was developed on behalf of the SCTO’s Statistics & Methodology Platform and is one result of the platform’s grant for 2018. Read more about the two award winners below under “Grants to foster open-access statistical tools”.

R is a statistical computing programme and language commonly used in research. presize is now available on R’s main package repository CRAN. There is also a user-friendly web app that supports almost the entire functionality of presize for those who do not know R.

Relevant links:

• presize R-Package via CRAN
• presize Web-App
• Bug reports, feedback, and feature requests are welcome at https://github.com/CTU-Bern/presize/issues.

Show more

Hide

Data sharing has become a requirement of many funding bodies and is becoming a scientific standard in many disciplines. In medical research, however, data sharing can conflict with clinicians’ obligation to protect patients’ privacy.

In June 2020, the Statistics & Methodology Platform, in collaboration with the Data Management Platform, has developed guidelines for sharing clinical research data, with particular reference to the national context and applicable laws. Included in the guidelines are details of aspects to be considered, decision-making criteria, examples, and checklists – all of which aim to support clinical research data sharing in practice.

The guidelines have been published as a concept paper in progress, which is accessible on a public repository as a preprint for external review. A second version will incorporate external comments.

Read the concept paper and leave a comment at www.preprints.org.

Show more

Hide

As of 2018, members of the CTU Network can compete annually for a grant to develop statistical codes or programmes. Applicants must demonstrate that their codes are geared at improving clinical research and bringing fresh solutions to persisting difficulties.

Each will be awarded CHF 12,000 to develop, test, refine, and share their code and each winner is obliged to collaborate with another CTU to crosscheck and user-test their code. These tools are then ideally published in open-access development repositories, such as The Comprehensive R Archive Network CRAN and Github.

Announcing the 2019 winners

In September 2019, two projects from four applicants were selected by the members of the Statistics & Methodology Platform as 2019 award-winners:

• Dr Lukas Bütikofer & Dr Alan Haynes, CTU Bern, for "Summary tables in R”
• Dr Thomas Zumbrunn, CTU Basel, for “Export tables and metadata to achieve the intended layout in any output format, R"

Their products should be available for the global community of statisticians, early 2021. Congratulations to the winners! You’ll find their products announced here.
 

2018 winners’ outputs published

As of October 2019, the two 2018 co-winners published the fruits of their labour, so statisticians worldwide can draw upon them:

• Dr Andreas Limacher and Dr Armando Lenz, CTU Bern, developed Precision-based sample size calculation, available on Github.

• Dr Thomas Fabbro, CTU Basel, developed a tool to represent visually assumptions about underlying sample size calculation, available on CRAN. A report about the project has been published in German in the November 2019 issue of DKForum, by the University Hospital of Basel.

Show more

Hide

As a first ever for Switzerland, a competition was run for statisticians in the CTU Network environment in 2018. The statisticians populating this network were challenged to show their prowess – competing for a grant to develop code or programs that will ultimately help improve clinical research and bring fresh solutions to persisting difficulties.

The Statistics & Methodology Platform of the SCTO received six applications from both the French and German parts of the country. Precisely half of the applicants proposed tackling the tricky, pertinent matter of calculating sample size of participants for clinical trials: How many participants does it take for a trial to draw meaningful and conclusive results? Those who design studies need to work closely with statisticians on this question, which can be the make or break of a study. The SCTO established its network of platforms for this very purpose – so platform members can not only provide leadership and share resources, but also inspire one another, revitalise their fields, and keep apace with trends and best practices.

Double winners 

A number of statisticians met for a day to put their heads together. The teams, each one based at a CTU, pitched their presentations. Then altogether, they debated and rated the proposals according to their applicability to clinical research. Double winners emerged – two teams. Each was awarded CHF 12,000 to develop, test, refine, and share their code: the team of Dr Andreas Limacher, CTU Bern, for “Precision-based sample size calculation” and that led by Dr Thomas Fabbro, CTU Basel, for a tool to represent visually assumptions about underlying sample size calculation, so doctors can assess these assumptions more easily. Each winner is obliged to collaborate with another CTU to crosscheck and user-test their code.

Dynamic community

Dr Brigitta Gahl, Platform Coordinator, commented: “Once these winning teams have finalised their product, it will go public, in the interests of research worldwide. They will post the open-access code on online forums visited by a lively community of professionals who can immediately put it to practice internationally. Statistics is an extremely vibrant field. Statistics has developed tremendously in recent years, with established methods being scrutinised and refined and new ones gaining ground. It’s all about community. We statisticians need to keep sharing and networking, so we can get the most out of this dynamic time and really shape it.”

Show more

Hide