Summary-measures analysis
Reduce each patient to one number, then use a standard two-group test
Clinical example
Blood pressure measured at ten visits in two treatment arms. For each patient you compute the change from baseline to the last visit, and compare that change between the arms.
When to use it
- The outcome is a measurement taken three or more times per patient.
- You can name one clinically meaningful feature per patient in advance: change from baseline, peak, area under the curve or slope.
- You want to compare groups with a standard two-group test on that number.
Assumptions to check
Choose the summary before collecting data
Choosing it after seeing the data makes false positives more likely.
One number per patient
After summarising, each patient is one independent observation, so a standard two-group test applies.
Check first
- Pick the summary before collecting data
What to report
- The chosen summary per group, with a CI for the difference
Common mistakes
A separate t-test at every time point
Fitting one regression line through all measurements as if independent
Alternatives
| If your situation is… | Use |
|---|---|
| Model the whole trajectory, with missed or irregular visits | Linear mixed model |
| Only two time points | Paired t-test |
In your software
# R (long format: one row per patient and visit)
library(dplyr)
per_patient <- d_long %>%
arrange(patient_id, visit) %>%
group_by(patient_id, arm) %>%
summarise(change = last(sbp) - first(sbp), .groups = "drop")
t.test(change ~ arm, data = per_patient)
* Wide format: one variable per visit.
COMPUTE change = sbp_last - sbp_first.
EXECUTE.
T-TEST GROUPS=arm(1 2)
/VARIABLES=change.