The role of the biostatistician
Applying statistical methods in research
Chris Oldmeadow, HMRI
September 15, 2020
HMRI Research Design and Statistical Services
Biostatistics
" To call a biostatistician after the experiment is done may be no more than asking him to perform a post mortem examination: he may be able to say what the experiment died of" - Ronald Fisher ![image]()
Common causes of study deaths
Incorrect Study Design
- Selection bias
- Contamination and clustering
- Randomisation
- Pilot studies
Failure to consider / measure confounding variables
- Stratify or adjust
- Causal modelling
Insufficient sample size
- Depends on:
- Outcome types and study design
- Clinically relevant difference
- Variability in the data
- Power and Type 1 error
Using excel to collect the data
![redcap]()
simple survey through to RCT free, easy-to-use, secure, ICH-GCP compliant
Failure to pre-specify an analysis plan
- increasingly a requirement to publish
Visualisations and dashboards
- Interactive dashboards using real time data
- Display recruitment rates, missing data, subject dispositions
More than just point and click
![dog]()
- understanding the different options - how to check if the results make sense - what to do if they dont
Reproducibility
- self contained code that will
- clean data and create analysis datasets
- run all analysis
- produce the tables and graphs
- integrated into the report
- eliminates transcription errors
- ensures results reproducible in 5, 10 years
simple survey through to RCT free, easy-to-use, secure, ICH-GCP compliant
- understanding the different options - how to check if the results make sense - what to do if they dont