R @ Cytel
Cytel Inc. is the leading provider of clinical trial design services and specialized statistical applications primarily for the biopharmaceutical, medical device, academic and government research markets. 47 out of the top 50 biopharmaceutical companies use Cytel software to design, simulate and analyze their clinical trials.
Cytel provides study sponsors with innovative tools, training and consultation to increase process efficiencies and reduce development costs.
USE OF R AT CYTEL
QA in product development
Cytel develops industry leading software like StatXact®, East®, SiZ®, Compass® and others. These are validated against similar established products, relevant R packages and in-house R codes developed independently for the purpose.
R has extensive usage throughout all stages of the algorithm validation process. Specifically in:
Generation of test cases
Flexible Data Structure
Powerful Graphical representation
Automating the testing
Internal consistency checks
Improving the development of the algorithm
Checking the robustness of the algorithm
Calling R from products
Statistical Analysis Services: Validation of clinical reporting
Safety summaries form a bulk of submissions in the data analysis services business at Cytel. A typical safety memo summarizes the parameters in the three safety datasets, namely, ECG, vital signs and laboratory tests. The routine memos involve computing the summary statistics of the parameters and plotting the corresponding graphs.
The analysis is carried out in SAS. A SAS macro is used to carry out the routine safety summaries. This macro produces the summary statistics of count, mean and the standard error for the parameter as well as for the change from baseline values.
Validation of results is carried out by writing parallel and independent programs in R. An R script is generally used to develop such a program. A function is written in R to validate the safety summaries. The R function inputs the raw SAS data, carries out the analysis and outputs the results in a .CSV file. The .CSV file is then read in SAS and compared with the summary datasets created in SAS. The same R function can be extended to input the SAS summary datasets and carry out the comparison within R itself.
PK Summary, Clinical Study Reports and ad-hoc requests
The pharmacokinetic (PK) summaries usually involve a mixed model. These are analyzed using the „mixed‟ procedure in SAS. Traditionally, the validation was also carried out in SAS by a validator using an independent SAS code. However, R packages „nlme‟ and „lme4‟ are now being used to validate the results obtained from SAS. R software is also used to validate other summaries by writing independent code on a case by case basis for CSR and non standard or ad-hoc summaries.
Reaching out to the community
Cytel is enthusiastically promoting the use of R for validation of statistical software and clinical data reporting. Cytel employees have recently made presentations on the topic at the Indian Association for Statistics in Clinical Trials (IASCT) conference at Bangalore, India in September 2011 and Pharmaceutical Users Software Exchange (PhUSE) annual conference at Brighton UK in October 2011.