Aalen Additive Hazards Model

The Aalen additive hazards model lets each covariate add a time-varying amount to the hazard and estimates cumulative regression functions by least squares at each event time.

알렌 가법위험모형은 각 공변량이 위험에 시간에 따라 변하는 양을 더한다고 보고, 사건 시점마다 최소제곱으로 누적회귀함수를 추정하는 모형입니다.

Does the effect of my covariates on the event risk change over follow-up time, measured as an excess risk?

공변량이 사건 위험에 주는 효과가 추적기간 동안 달라지는지 위험 차이로 보고 싶어요.

The Aalen additive model writes the hazard as a baseline plus covariate effects that may change freely over time, and estimates their cumulative effects B(t) by least squares at each event time. The slope of each B(t) shows when a covariate adds or removes risk, and a Wald statistic at a chosen time tests whether its cumulative effect differs from zero.

When to use Aalen Additive Hazards Model

• Covariate effects are expected to change over time
• You want absolute (excess) risk differences rather than hazard ratios
• The proportional hazards assumption is clearly violated

When to use something else

• Effects are constant on the relative scale; use a Cox proportional hazards model
• The sample has few events relative to covariates; use a penalized Cox model such as Lasso-Cox

Assumptions and requirements

• Covariates are measured at baseline
• Censoring is independent
• Enough subjects remain at risk for the least-squares increments to be stable

What the analysis reports

• Cumulative regression coefficients B(t) at a chosen time
• Standard errors, z statistics and p values
• Plots of B(t) over time
• Average excess rate per unit time

How to report it

In an Aalen additive hazards model, the cumulative regression coefficient for [covariate] at t = [value] was [value] (SE = [value], z = [value], p = [value]).

Related methods

• Cox Proportional Hazards Model
• Time Dependent Cox Hazards Model

Run Aalen Additive Hazards Model on Quorant