Fine-Gray Competing Risk Model

The Fine-Gray model is a competing risks regression that models the subdistribution hazard of a specific event so covariate effects translate directly into effects on that event's cumulative incidence.

Fine-Gray 모형은 특정 사건의 하위분포 위험을 모형화하여 공변량의 효과가 해당 사건의 누적발생률에 대한 효과로 직접 해석되도록 하는 경쟁위험 회귀모형이다.

How do I analyze time to an event when another event can prevent it from happening?

다른 사건이 발생하면 관심 사건이 일어날 수 없을 때 생존시간을 어떻게 분석하나요?

The Fine-Gray subdistribution hazard model estimates how covariates relate to the cumulative incidence of the event of interest when competing events, such as death from another cause, preclude it. It fits prediction or prognosis questions about the absolute risk of one event type in the presence of competing risks.

When to use Fine-Gray Competing Risk Model

• Time to relapse where death before relapse is a competing event
• Estimating cumulative incidence of a specific cause over time
• Risk prediction models where competing events are common

When to use something else

• The aim is etiology, i.e., the effect on the rate of the event among those still at risk: use cause-specific Cox models
• There are no competing events, only censoring: use the standard Cox proportional hazards model

Assumptions and requirements

• Proportional subdistribution hazards
• Independent censoring
• Event types are clearly defined and mutually exclusive
• Correct coding of event, competing event, and censoring

What the analysis reports

• Subdistribution hazard ratios (sHR) with confidence intervals
• Cumulative incidence function curves by group
• Gray's test comparing cumulative incidence between groups
• Number of events of interest and competing events

How to report it

In the Fine-Gray model, [predictor] was associated with a higher cumulative incidence of [event], sHR = [value], 95% CI [[value], [value]], p = [value].

Related methods

• Cox Proportional Hazards Model
• Cause-Specific Hazards Model
• Cumulative Incidence Function

Run Fine-Gray Competing Risk Model on Quorant