What are competing risks, and how do they affect interpretation of incidence in a cohort?

• A. Competing risks are events that preclude the event of interest (e.g., death before disease onset); they can bias simple incidence estimates; require competing risks methods (e.g., cumulative incidence function).
• B. Competing risks are unrelated events that can be ignored in incidence calculations.
• C. Competing risks are only relevant in case-control studies.
• D. Competing risks are the same as censoring; treat with Kaplan-Meier.

Answer: (A)

Competing risks occur when another event can happen that prevents the event of interest from ever occurring, such as death before developing a disease when you’re studying disease incidence in a cohort. Because these events remove people from the risk set, they change the actual probability that the event of interest will happen by a given time.

If you use standard methods that censor people at the time of a competing event (like death) and apply a Kaplan-Meier estimate, you’re implicitly assuming those individuals could still experience the event of interest later. That assumption is not true in the presence of competing risks, so the simple incidence estimates can be biased—often overstating the risk of the event of interest.

The correct approach is to use methods that account for competing risks, most commonly the cumulative incidence function, which provides the probability of experiencing the event of interest by time t in the presence of competing events. It incorporates both the hazard of the event of interest and the hazards of the competing events, giving a real-world interpretation of risk. For examining covariate effects, you can use models related to the subdistribution hazard (Fine-Gray) that target the CIF.

In short, competing risks meaningfully alters how we interpret incidence in a cohort, so appropriate methods are needed to estimate true probabilities rather than relying on standard survival methods that treat competing events as simple censorship.