Hazard Function

The hazard function at any time tj is the number of deaths at that time divided by the number of subjects at risk, i.e.

image036x

If dj > 1, we can assume that at exactly at time tj only one subject dies, in which case, an alternative value is

image037x

We assume that the hazard function is constant in the interval [tj, tj+1), which produces a step function. The standard error of h(t) is

image038x

By Property 1 of Basic Concepts of Survival Analysis, the cumulative hazard function is represented by the formula

image039x

As shown in the proof of Property 1 of Kaplan-Meier Theory, it follows that the standard error of H(t) is

image040xReference

NCSS (2015) Kaplan-Meier curves (Logrank tests)
https://www.ncss.com/wp-content/themes/ncss/pdf/Procedures/NCSS/Kaplan-Meier_Curves-Logrank_Tests.pdf

5 thoughts on “Hazard Function”

  1. Hello Dr. Charles Zaiontz, how are you?

    I would be so grateful, if you tell me how can I add a Survival distribution function S(t) overtime in a Cox Proportional Hazard method? I’d like to add the same chart available in the Kaplan-Meier approach.

    Best Regards,
    Raphael

    Reply
  2. Dear Charles,

    Need your help for creating
    a) Kaplan Meier Hazard curve
    b) Nelson Aalen Hazard curve

    Am not from the Mathematical field and honestly do not understand if they are two different things
    Need it for a clinical trial, related to risk for an event with time on X axis

    Please suggest which would be better if the two are different

    Reply
      • Yes I have used it from your web page and it was very useful

        The problem in using is as follows

        I am writing about patients landing up in more and more severe levels of kidney failure over time
        When I use survival curve – It conveys how many of them did not get into that stage
        When I use hazard curve – it conveys how many of them got into those severe stages

        That is the reason I was looking for hazard curve

        Regards
        Dr Anitha

        Reply

Leave a Comment