Cohort outcomes
Compare a binary outcome between groups and analyze time-to-event, with effect sizes and explicit censoring reported alongside significance.
Research question
Do outcomes differ between two cohorts, and how do survival or time-to-clearance curves compare — as associations, not causal effects?
Who this is for
- Infectious-disease researchers comparing treatment, exposure, or pathogen groups
- Hospital epidemiology teams evaluating intervention-associated outcomes
- Analysts needing risk ratios, odds ratios, and Kaplan-Meier curves
Data requirements
| Data | Required | Purpose |
|---|---|---|
| Episodes with a grouping attribute | Yes | Defines the two cohorts to contrast |
Outcomes with event_observed | Yes (comparison) | Binary risk, risk ratio, odds ratio, tests |
Outcomes with time_to_event_days | Yes (survival) | Kaplan-Meier and log-rank |
Workflow
Model episodes and outcomes → Cohort comparison (RR/OR + test)
→ Outcome analysis (Kaplan-Meier + log-rank)
→ AI interpretation
Step 1 — Cohort comparison
The Cohorts page compares a binary outcome between groups defined by a participant or episode attribute, reporting per-group risk, risk ratio, odds ratio, and a chi-square or Fisher test. Effect estimates are associational, never causal.
Step 2 — Time-to-event
The outcome analysis fits Kaplan-Meier survival with explicit censoring and, when a stratifying attribute is supplied, a log-rank test between strata.
Expected outputs
- Per-group risk, risk ratio, and odds ratio with a p-value
- Kaplan-Meier survival curves, overall and by stratum
- Log-rank statistic and p-value between the first two strata