|
5 Questions to ask your Statistician
- Are you trained in statistics?
Some “statisticians” have formal training in another discipline with only peripheral exposure to statistics via SAS® programming, short course, or self teachings. BioSTAT statisticians have advanced degrees in statistics and average more than 20 years experience as statisticians.
- Are you experienced in nonclinical toxicology studies?
Statisticians working in other disciplines such as clinical trials, manufacturing, or academia are not familiar with the unique issues associated with nonclinical toxicology studies. Since 1994 BioSTAT has been devoted exclusively to this industry and its distinctive nuances.
GLP compliance is of particular importance if statistical services are contracted from an outside source. BioSTAT statisticians are GLP trained and all GLP study activities are subject to in-house SOPs. Our office and procedures have been audited by sponsor laboratories with favorable responses and we welcome such involvement by our clients.
- What is your statistical Quality Control process?
An individual statistician working alone has no checks and balances mechanism. While Quality Assurance personnel can assure that conformance with regulations is maintained, unless they are formally trained in statistics, it is unlikely they can provide quality control over the intricacies of a complex statistical analysis. If the protocol described a repeated measures analysis of covariance with fixed effect terms for treatment group, time, sex, and interactions, and random effects for litters nested within groups and pups nested within litters*groups, could your QA personnel assure that the methodology was accurately carried out and reported?
A BioSTAT SOP requires all GLP projects be subject to our quality control process which requires 100% duplication of all results by an independent statistician. The process supplements the client’s QA review and assures the statistical integrity of every BioSTAT project.
- What statistical software do you use?
Many laboratories utilize “black box” software for statistical analysis and summary tables. Like any technology, the statistical aspects of such software are often outdated before it’s off the shelf. Is the software capable of the most current methodologies? Does it calculate exact p-values or approximations? Are users properly trained on the software? Do users understand the options and output and can they interpret protocol methods in terms of those options and output? Are protocol statistical methods dictated by the software’s capabilities or are the statistical methods customized to address the specific study design and objective of each protocol?
BioSTAT utilizes single-use SAS® programs customized to each individual project and subject to our QC process. Upon project completion, the programs are archived with the specific project material. This allows us to adopt state of the art methodology and take advantage of the most current statistical and software technologies.
|
