Best Practices – Leveraging the Biostatistician
By Editorial Team on February 2, 2022
We continually urge our Sponsor/CRO clients to make better use of their study Biostatistician. These professionals have more to contribute than most realize. Yes, they crunch the numbers and help with the final study submission – but they have more to offer than that. Let’s look at how to take advantage of their skills prior to study closeout.
Protocol Development – A seasoned Biostatistician understands the perspective of regulators and can develop protocol language pertaining to randomization and analysis that regulators will understand and accept. The study biostatistician will also understand the steps used from database download to the generation of the statistical analysis. Making these steps easier for regulators to understand helps with their process of confirming the results from a study.
Study Build – When Biostatisticians get involved early, they can work with study builders and offer guidance in several ways. This early involvement will ultimately save the Biostatistician time when preparing submissions – which is always desirable.
- Database Field Conventions.
- Use of consistent references (i.e., always using “subject_id” as an identifier rather than sometimes using “patient_id” and then other times using “case_id”).
- Use of consistent values in references (i.e., for a subject_id consistently using a format of “AUS-001” and avoiding use of the similar (but not identical!) “AUS001”). This is especially important when dates will be compared. All date values should be consistently formatted to facilitate comparison (i.e., don’t intermix “May 2, 2022”, “02 May 2022”, and 5/2/22).
- Use SAS-friendly Names. Field names longer than 32 characters should be avoided. Otherwise, SAS will truncate them – making it harder for the Biostatistician to understand the data. Similarly, SAS will replace special characters with an underscore – exacerbating the problem.
- Database Table Structure. Ensuring the ability to map study data to outcomes/endpoints is critically important. The Biostatistician can review the database ahead of time and identify any potential problems in performing this mapping.
Study Validation – Sponsors/CROs and study personnel see the data from the “outside in”. They are looking at the clinical data using electronic forms that (optimally) resemble those in the protocol. But the Biostatistician is more interested in the “inside out” view.
- Perform an Analysis “Dry Run”. The Biostatistician can assess how the statistical analysis will mesh with the database, finding issues with data fields and potential data entry inconsistencies. Identifying structures that pose difficulties before live data is entered will help speed the submission process.
This up-front analysis will also help identify potential gaps (missing entries explained by comments) and allow the Biostatistician to recommend areas where comments should be captured in database fields and incorporating these gaps into data collection fields where possible. For instance, in an Osteoarthritis study having Osteoarthritis is a pre-condition for enrollment and shouldn’t be considered as an Adverse Event if unrelated to the study drug. Such guidance can help avoid tedious data review and notes to file after study lock.
Study Conduct – It’s always a good idea to understand gaps or inconsistencies in data as those data are collected – rather than scrambling to get questions answered when analysis is being conducted. The Biostatistician can work in concert with study monitors and data managers to solve such issues in a timely fashion.
- Review Live Data (Masked Review). Sometimes, issues arise with real data that were not identified with test data. Encouraging the Biostatistician to perform periodic reviews will help identify – and rectify- these problem areas early.
- Configure Shared Reports. Reports can be configured to report on enrollment numbers, adverse events, and missing/incomplete data. Masked study personnel (e.g., Monitors, Project Managers, Data Managers) can use these reports throughout study conduct to track progress and spot issues as the study progresses. Such reports may point to data collection issues in the EDC or the need for more training.
- Pre-Lock Analysis Dry Run. Before the database is locked, analysis can be run in a blinded fashion to identify dataset errors (not caught by constraints). This will uncover potential issues with fields that are named dynamically, such as IDEXX results, and may end up posing problems when imported into SAS.
General Best Practices – It’s in the Sponsor’s best interest to make the Biostatistician’s job easier. Not only will this save the Biostatistician time and effort during study submission, but it may also reduce study costs.
- Annotated case report forms. As indicated above, the study biostatistician works from the “inside out”, and as such doesn’t have easy access to a glossary of coded field names, short of working back and forth between “outside view” and the “inside view”. Providing the biostatistician with annotated case report forms would provide a valuable link useful in data analysis.
- Provide Clear Deliverable Expectations. When the Biostatistician knows what the deliverables are up front, it makes his or her life easier. The earlier this information is provided, the better. If available during study build, this information will help identify areas that may be tricky for the Biostatistician to handle during submission preparation. This guidance should include:
- Summary table layout (including headers)
- Data listing table layout (including headers)
- Summary figure expectations and format
We want to thank Dr. Steven Radecki for his contributions to this article. Dr Radecki is a consulting biostatistician who has been providing statistical support to the Animal Health industry for more than 20 years. Specializing in Animal Health has strengthened his relationships with CVM and sponsors alike and has provided him with insights into the conduct of Clinical Studies needed for product registration. He has a background in Animal Science (BS, MS, and PhD in Animal Science) and a MS in Biometrics. Dr Radecki can be reached by email at email@example.com.
Give Prelude EDC a try for your next eClinical Trial!