Clinical evidence that stands up to committee review
Biostatistics for academic and pragmatic trials, real-world evidence and biomarker analysis.
In clinical evidence, the analysis does not start when the database is locked. It starts when you decide which question you want to answer, which outcome represents it and which comparison will support the conclusion. A trial can be held back by a poorly justified sample size; a study in routine clinical practice, by a comparison that does not adequately control for confounding; a biomarker panel, by working only in the cohort in which it was discovered.
We work with clinical research groups, biotech companies at the preclinical and early clinical stages, and medical affairs teams. The aim is the same in all three settings: every claim can be traced back to a design that supports it, an analysis defined with sound judgement, and data whose uncertainty and limitations are on the table.
We regularly work with
Clinical research groups sponsoring academic or pragmatic trials who need to turn a clinical question into a defensible design, sample size and analysis plan.
Biotech companies with clinical, molecular or omics signals who need to know whether a biomarker holds up beyond the cohort in which it appeared.
Medical affairs teams who want to compare effectiveness in routine clinical practice without treating observational data as if they came from random allocation.
Researchers who want to review the design, the analysis or the interpretation before those decisions reach a reviewer.
Where we usually come into a clinical evidence project
The need changes as the project moves forward. Before recruitment, the issue usually lies in the design and pre-specification; once the data are available, it becomes a question of which comparison can be sustained, how to model the outcome and how to tell a reproducible signal from a finding that depends on the sample. If you would rather start from what you need to do than from your sector, each block leads to the full path for that need.
Design a study
A study is won or lost before the first data point is collected.
I have data and don't know where to start
You already have the data. The question is what to do with them.
Identification and validation of diagnostic, prognostic and predictive biomarkers
View service →Time-to-event analysis for clinical prognosis
View service →From repeated measurements over time to an understanding of how things change
View service →From sequencing data to integration with the rest of your study
View service →Evidence of effectiveness in routine clinical practice
View service →The rest of the catalogue, organised by what you need to do
A clinical study rarely fits within a single sector label. It may need a database that was built for patient care to be prepared, a causal comparison to be set up, an analysis that has already been run to be reviewed, or the result to be turned into a reproducible report. Those services are organised by the need they address, not by the discipline the project comes from.
What we are usually asked in this sector
Our protocol isn't finalised yet. Is it too early to talk?
No. That is exactly when there is still room to correct decisions that data collection will later lock in. Before recruitment, you can check whether the outcome answers the clinical question, whether the difference used to size the sample is genuinely relevant, whether the measurement schedule will allow the target effect to be estimated, and which analysis needs to be pre-specified. When these issues come to light after database lock, the analysis can document them and adapt, but it cannot always repair the design that caused them.
Can you analyse data from routine clinical practice to support an effectiveness claim?
Yes, as long as the design and the data allow a defensible comparison to be built. Without randomisation, the work consists of clearly defining the population, exposure, outcome and time zero, identifying the sources of confounding and applying the appropriate method (propensity score matching or weighting, among other approaches, and instrumental variables when the context genuinely justifies them). The conclusion is bounded by the assumptions of the design and comes with sensitivity analyses. The aim is not to make an observational study look like a trial, but to make clear which effect can be estimated and under what conditions.
What do we receive at the end of the statistical work on a trial?
It depends on the agreed scope, but the work can cover anything from the statistical analysis plan before database lock to the executed analysis with traceable code, the tables and figures, and the statistical write-up of the methods and results. The common thread is traceability: the link between protocol, analysis populations, decisions taken, code and final result can be followed. If a later review calls for an additional analysis, that structure allows what has already been done to be reproduced and makes it clear what changes and why.
Does your work fit better elsewhere?
Shall we talk?
Tell us where your study stands and we'll give you a straight answer on how we can help.