Cumulative incidence (Aalen-Johansen)
The Aalen-Johansen estimator gives the chance of a specific first event by a stated time when other events can prevent it.
Fit
When to use it, and when not to
Use it when
- Death or another first event makes your outcome impossible.
- You want the observed absolute probability of each event type over follow-up.
Not when
- Death from any cause is the sole outcome, with no competing cause. Use Kaplan-Meier survival. Read: Kaplan-Meier estimation
- You need covariate-adjusted associations. Choose a competing-risks regression that matches your question. Read: Choosing an analysis for competing risks
Before you start
What you need
- Mutually exclusive first eventscompetingEventsCode the outcome of interest, competing events and ordinary censoring separately.
- A common start and horizonDefine when risk starts and the time at which probability matters.
- Observation and entry timesIf people enter after time zero, record their entry times and use risk sets allowing delayed entry.
- Within each reported group, loss to follow-up is unrelated to future event times.How to check: Record why follow-up ended. Assess how results change if those lost have a worse outlook.
- Different people supply independent observations.How to check: For patients grouped within clinics or families, use inference that accounts for those groups.
- Each first event has a reliable type and date.How to check: Check outcome adjudication. A competing death is an observed event, not missing follow-up.
Plain language
How it works
At each event time, find the chance of still being free of every event just before that time. Multiply it by the share of the current risk set having the event of interest. Add these contributions across time. Deaths reduce later event-free survival, so they also reduce later chances of the outcome. This is the competing-risks form of Aalen-Johansen.
Separate cumulative incidence curves can be drawn for each first event. Their sum is the chance of any first event. Korn and Dorey illustrate these crude incidence curves. Allignol and colleagues derive variance results for Aalen-Johansen, including delayed entry. Use a compatible interval method and show the numbers still at risk. Do not interpret the curve as risk after eliminating death.
Worked example
A worked example
- Setting
- Ten patients are followed for first relapse or death, with no loss to follow-up.
- Question
- What is relapse probability by month 6?
- Approach
- Two relapse at month 2. Three die at month 4. One relapses at month 6; four remain event-free.
- Result
- Relapse cumulative incidence is 2/10 + (5/10) × (1/5) = 0.30, or 30 in 100 (30%). Censoring deaths gives 36 in 100 (36%) instead.
- What it means
- The incorrect Kaplan-Meier complement is higher because it treats competing deaths as if relapse remained possible.
In practice
Recent studies PubMed search · not vetted
Watch out
Common mistakes
The one to remember
Cumulative incidence includes the chance of remaining free of every first event. A Kaplan-Meier curve that censors a competing death or another event overstates cumulative incidence when read as one minus survival.
- Removing competing deaths before estimating risk.Instead: Keep them as a separate event type in the Aalen-Johansen calculation.
- Calling a curve without a horizon a patient probability.Instead: State the event, time zero, horizon, estimate and confidence interval.
Inside Mallard
How Mallard uses this
Learn search lists Cumulative incidence (Aalen-Johansen). The method registry links this method to its single-method article and code.
competing-risks-cumulative-incidence).analysis.aalen_johansen, the id a plan cue records.Related Learn pages
Related designs and methods
Evidence
Sources & further reading
- A note on variance estimation of the Aalen-Johansen estimator of the cumulative incidence function in competing risks, with a view towards left-truncated data. (opens in a new tab) · Biom J · doi 10.1002/bimj.200900039
- Applications of crude incidence curves. (opens in a new tab) · Stat Med · doi 10.1002/sim.4780110611