The statistical analysis plan ensures an a priori defined analysis plan, transparency for all collaborators and a optimal workflow.
The statistical analysis plan should contain the following:
Outcome
Define the outcome including type of data (continuous, binary, categorical, ordinal, time-to-event)
Covariates
Specify if any covariates should be accounting for, preferably with a directed acyclic graph (DAG)
Clustering
Specify if there are clustered observations, if yes specify the structure (e.g. breasts within patients, samples within patients)
Analysis hierarchy
Specify the order of analyses. Specify the role of each analysis
(primary, secondary, subgroup, sensitivity)
- Primary analysis: Risk of all complications (hematoma, explantation
and infection) at 5 years
- Secondary analysis: Same as above for each of the complications
- Subgroup: Same as the primary analysis but for patients between
2018-2024
Data dictionary
A table with four columns specifying
1) The variable name in the dataset 2) Very short description of the
variable 3) type of data.
- If continuous specify the distribution (0 to 1, non-negative,
exponential)
- If categorical specify the levels and if there is any ordering such as
(low, moderate, severe)
4) The use of the data point (e.g. outcome, exposure, covariate,
cluster)
name |
description |
type |
role |
|---|---|---|---|
hema |
whether hematoma occured |
binary, 0 = censuring, 1=event |
outcome |
t_hema |
time to hematoma |
continuous, non-negative |
outcome |
imp_plane2 |
implant plane |
binary, sg/sm |
exposure |
age |
age at surgery |
ordinal, binned 5-year age groups |
covariate |