Poisson surrogate for an uncensored proportional-hazards likelihood

ID: poisson-surrogate-for-an-uncensored-proportional-hazards-likelihood

Consider independent event times with hazard function and baseline cumulative hazard function . For optimization over , first hold and fixed and independent of . Under ignorable independent censoring, an observation has log-likelihood contribution
Set . A Poisson regression with response and log link has log-likelihood contribution , which differs from only by terms independent of . Thus the same maximizes both criteria. The linear predictor is , with a known offset whenever .
For uncensored observations every response is one. This is an equivalence of objective functions: the event indicators need not actually be Poisson counts. A related split-interval construction handles piecewise-constant hazard functions.[ref]
If the baseline is unspecified, its integrated values are not known offsets; fitting an ordinary generalized linear model as if they were known does not estimate an arbitrary Cox proportional-hazards model. One must estimate the baseline jointly, use a suitable baseline representation, or eliminate it through Cox partial likelihood. Zero exposure also needs separate handling rather than taking the logarithm of zero.

New to topics? Read the docs here!