The problem
A statistician brought in at the end can only describe what happened.
The decisions that determine whether a study can demonstrate anything are made at design: the endpoint, the effect size assumed, the power, the population definition. By the time the data exists, those are fixed.
Under-powering is the expensive version of this. A study runs to completion, costs what it costs, and produces a result nobody can act on. The statistical mistake was made two years earlier in a conversation the statistician was not part of.
So we push to be involved at protocol design. It is the cheapest hour of statistical time you will ever buy.
Talk to a specialist
Tell us the problem. A specialist replies, not a form robot.
What we do
What we deliver.
Study design input
Endpoint selection, effect size assumptions and power, argued rather than assumed.
Statistical analysis plans
Pre-specified so regulators accept the analysis rather than question it.
Randomisation
Schedules and stratification designed for the study, not copied from the last one.
Interim analyses
Adaptive designs and DSMB support with the alpha spending planned.
CDISC deliverables
SDTM and ADaM datasets, define files and reviewer guides.
TLFs and reporting
Tables, listings and figures built for the CSR and the submission.
For consultants
Roles we place in biostatistics.
Statisticians and statistical programmers are among the scarcest profiles in the sector and are almost always working. Being known to us matters more here than anywhere.
- Biostatisticians, all levels
- Statistical Programmers (SAS and R)
- CDISC and Standards Specialists
- Principal and Lead Statisticians
- DSMB support statisticians
- Statistical Analysis Plan authors
Proof
Design
The cheapest hour of statistical time you will ever buy.
Client testimonials for this discipline are being collected and will appear here once approved in writing by the clients concerned.
Questions
Questions we get asked.
Can you support a study already underway?
Yes, though we will be candid about what can still be changed. Analyses can be tightened; a design decision usually cannot be undone.
SAS or R?
Both. SAS remains the submission default in most cases; R is increasingly used alongside it. We screen for the one you actually need.
Do you handle regulatory statistical questions?
Yes, and it goes better when the people answering wrote the original plan.
Is this work remote?
Almost always, which is why these roles attract strong candidates nationally.
How do you price statistics work?
SAP authoring and analysis packages work as fixed scope. Ongoing programming support is time and materials.
Related
Where to go next.
Talk to someone who has done this.
Hiring for it, or looking for the work. Either conversation is welcome.