Survival - Notes (Lecture 1)

Lecture 1: Survival Models and Analysis
Introduction
Ms. Beryl Ang’iro
August 21, 2020
Ms. Beryl Ang’iro August 21, 2020 1 / 17

Introduction
Introduction
Survival analysis is the phrase used to describe the analysis of data that
correspond to the time from a well-defined time origin until the occurrence
of some particular event or end-point.
In many biomedical applications the primary end-point of interest is time

to a certain event. e.g;
• time to death;
• time it takes for a patient to respond to a therapy;
• time from response until disease relapse (i.e., disease returns); etc.

Introduction
Introduction

• time to death;

Introduction
Introduction

• time to death;

Introduction
Introduction

• time to death;

Introduction
Introduction

• time to death;

Introduction
Some Introducing Examples
Example: Times to Weaning of Breast-fed Newborns

(National Longitudinal Survey of Youth (NLSY, 1995))
Youths, aged 14 to 21 in 1979, were interviewed yearly through 1988.

Beginning in 1983, females in the survey were asked about any pregnancies
that have occurred since they were last interviewed (pregnancies before
1983 were also documented). This data set consists of the information
from 927 first-born children to mothers who chose to breast feed their
children.
The response variable in the data set is duration of breast feeding in

weeks, followed by an indicator of whether the breast feeding was
completed (i.e., the infant is weaned).

Introduction
Example: Times to Weaning of Breast-fed Newborns

(National Longitudinal Survey of Youth (NLSY, 1995))
Youths, aged 14 to 21 in 1979, were interviewed yearly through 1988.

Beginning in 1983, females in the survey were asked about any pregnancies
that have occurred since they were last interviewed (pregnancies before
1983 were also documented). This data set consists of the information
from 927 first-born children to mothers who chose to breast feed their
children.
The response variable in the data set is duration of breast feeding in

weeks, followed by an indicator of whether the breast feeding was
completed (i.e., the infant is weaned).

Introduction
Example: Leukemia
(Klein and Moeschberger (2003))
Bone marrow transplants are a standard treatment for acute leukemia. In

this multi center trial, patients were followed from the moment they had a
bone marrow transplant until the leukemia returns. The researchers
investigated the time until relapse and looked for factors that influenced
this time.
Some patients did not have a relapse during the study period or died
without relapsing.

Introduction
Example: Leukemia
(Klein and Moeschberger (2003))
Bone marrow transplants are a standard treatment for acute leukemia. In

this multi center trial, patients were followed from the moment they had a
bone marrow transplant until the leukemia returns. The researchers
investigated the time until relapse and looked for factors that influenced
this time.
Some patients did not have a relapse during the study period or died
without relapsing.

Introduction
Example: Time to First Use of Marijuana
Turnbull and Weiss (1978) report part of a study conducted at the

Stanford-Palo Alto Peer Counseling Program (see Hamburg et al. [1975]
for details of the study). In this study, 191 California high school boys
were asked, “When did you first use marijuana?” The answers were;
• the exact ages (uncensored observations);

• “I never used it,” which are right-censored observations at the boys’
current ages; or
• “I have used it but can not recall just when the first time was,” which
is a left-censored observation.

Introduction


current ages; or

Introduction


current ages; or

Introduction


current ages; or

Introduction
Censoring
Censoring
In each of these examples, we cannot fully observe the time until a certain
event. Due to different practical reasons, we only observe in the examples
a lower bound of the true time.This is called right censoring.
There are generally three reasons why censoring might occur:
• A subject does not experience the event before the study ends.
• A person is lost to follow-up during the study period.
• A person withdraws from the study
These are all examples of right-censoring.
In general, any situation in which you cannot fully observe a time until an
event but only observe some boundaries for this time is called censoring.
In the above examples, there is another non-negative random variable
C (C ≥ 0) which we call the censoring variable and which obscures the
observation of T .
Introduction
Censoring
Censoring
observation of T .
Introduction
Censoring
Censoring
observation of T .
Introduction
Censoring
Censoring
observation of T .
Introduction
Censoring
Censoring
observation of T .
Introduction
Censoring
Censoring
observation of T .
Introduction
Censoring
Censoring
observation of T .
Introduction
Censoring
Right Censoring:
Let C1 ,...,Cn be a sample of censoring times.

We observe a sample of couples (Y1 ,δ1 ) ,..., (Yn ,δn ) where for i=1,...,n;
(
Ti , if Ti ≤ Ci
Yi = Min(Ti , Ci ) =
Ci , if Ti > Ci
(
1, if Ti ≤ Ci
δi = I (Ti ≤ Ci ) =
0, if Ti > Ci
In general we assume that for i=1,...,n, Ti and Ci are independent. The
observed value is the minimum of the censoring and failure times; subjects
whose failure time is greater than their censoring time are right-censored.

Introduction
Censoring
Right Censoring:

(
Ti , if Ti ≤ Ci
Ci , if Ti > Ci
(
1, if Ti ≤ Ci
δi = I (Ti ≤ Ci ) =
0, if Ti > Ci

Introduction
Censoring
Right Censoring:

(
Ti , if Ti ≤ Ci
Ci , if Ti > Ci
(
1, if Ti ≤ Ci
δi = I (Ti ≤ Ci ) =
0, if Ti > Ci

Introduction
Censoring
Right Censoring:

(
Ti , if Ti ≤ Ci
Ci , if Ti > Ci
(
1, if Ti ≤ Ci
δi = I (Ti ≤ Ci ) =
0, if Ti > Ci

Introduction
Other censoring schemes
Left censoring:
This occurs when the event of interest has already occurred before
enrolment. This is very rarely encountered.
We observe a sample (Y1 ,δ1 ) ,..., (Yn ,δn ) where for i=1,...,n;
(
Ti , if Ti ≥ Ci
Yi = Max(Ti , Ci ) =
Ci , if Ti < Ci
(
1, if Ti ≥ Ci
δi = I (Ti ≥ Ci ) =
0, if Ti < Ci
Some Examples:
• Development of children behavior: At which age can a child perform
a certain task. Some children can perform the task when they enter
the study.
• Detection limits: A measuring device cannot give a correct value
below a fixed limit.
Introduction
Left censoring:
(
Ti , if Ti ≥ Ci
Ci , if Ti < Ci
(
1, if Ti ≥ Ci
δi = I (Ti ≥ Ci ) =
0, if Ti < Ci
Some Examples:
the study.
Introduction
Left censoring:
(
Ti , if Ti ≥ Ci
Ci , if Ti < Ci
(
1, if Ti ≥ Ci
δi = I (Ti ≥ Ci ) =
0, if Ti < Ci
Some Examples:
the study.
Introduction
Left censoring:
(
Ti , if Ti ≥ Ci
Ci , if Ti < Ci
(
1, if Ti ≥ Ci
δi = I (Ti ≥ Ci ) =
0, if Ti < Ci
Some Examples:
the study.
Introduction
Left censoring:
(
Ti , if Ti ≥ Ci
Ci , if Ti < Ci
(
1, if Ti ≥ Ci
δi = I (Ti ≥ Ci ) =
0, if Ti < Ci
Some Examples:
the study.
Introduction
Interval Censoring
A data point is somewhere on an interval between two values. Instead of a
sample of lifetimes T1 ,...,Tn we get for each individual an interval in which
the event occurred. Hence we get (L1 ,R1 ], ..., (Ln ,Rn ]
Example:
In a clinical trial on breast cancer patients, the researchers were interested
whether there was a difference in cosmetic effects for early breast cancer
patients when they were treated with radiotherapy only or with
radiotherapy and chemotherapy. At each visit, every 4 till 6 month, a
clinician recorded a measure for breast retraction. Of interest was the time
until moderate or severe breast retraction appeared.
Regardless of the type of censoring, we must assume that it is

non-informative about the event; that is, the censoring is caused by
something other than the impending failure.
Introduction
Interval Censoring
Example:

Introduction
Interval Censoring
Example:

Introduction
Censoring example
Imagine a study of the survival of heart-lung transplant patients who are

followed up after surgery for a period of 52 weeks. The event of interest is
death. Not all subjects will die during the 52 - week follow-up period, but
all will die eventually. The figure below depicts the survival histories of six
subjects in the study, and illustrates several kinds of censoring (as well as
uncensored data):
Introduction
Censoring example
Imagine a study of the survival of heart-lung transplant patients who are

followed up after surgery for a period of 52 weeks. The event of interest is
death. Not all subjects will die during the 52 - week follow-up period, but
all will die eventually. The figure below depicts the survival histories of six
subjects in the study, and illustrates several kinds of censoring (as well as
uncensored data):
Introduction
Censoring example
• Subject 1 is enrolled in the study at the date of transplant and dies

after 40 weeks; this observation is uncensored.
– The solid line represents an observed period at risk, while the solid
circle represents an observed event.
• Subject 2 is also enrolled at the date of transplant and is alive after
52 weeks; this is an example of fixed-right censoring.
– The broken line represents an unobserved period at risk ; the filled
box represents the censoring time; and the open circle represents an
unobserved event.
– The censoring is fixed (as opposed to random) because it is
determined by the procedure of the study, which dictates that
observation ceases 52 weeks after transplant.
– This subject dies after 90 weeks, but the death is unobserved and
thus cannot be taken into account in the analysis of the data from
the study.
– Fixed-right censoring can also occur at different survival times for
different subjects when a study terminates at a predetermined date.
Introduction
Censoring example

unobserved event.
the study.
Introduction
Censoring example

unobserved event.
the study.
Introduction
Censoring example

unobserved event.
the study.
Introduction
Censoring example

unobserved event.
the study.
Introduction
Censoring example

unobserved event.
the study.
Introduction
Censoring example

unobserved event.
the study.
Introduction
Censoring example
• Subject 3 is enrolled in the study at the date of transplant, but is lost

to observation after 30 weeks (because he ceases to come into
hospital for checkups); this is an example of random-right censoring.
– The censoring is random because it is determined by a mechanism
out of the control of the researcher.
– Although the subject dies within the 52-week follow-up period, this
event is unobserved.
– Right censoring — both fixed and random — is the most common
kind.
• Subject 4 joins the study 15 weeks after her transplant and dies 20
weeks later, after 35 weeks; this is an example of late entry into the
study.
– Why can’t we treat the observation as observed for the full 35-week
period? After all, we know that subject 4 survived for 35 weeks after
transplant.

Introduction
Censoring example

kind.
study.
transplant.

Introduction
Censoring example

kind.
study.
transplant.

Introduction
Censoring example

kind.
study.
transplant.

Introduction
Censoring example

kind.
study.
transplant.

Introduction
Censoring example

kind.
study.
transplant.

Introduction
Censoring example
– The problem is that other potential subjects may well have died
unobserved during the first 15 weeks after transplant, without enrolling in
the study; treating the unobserved period as observed thus biases survival
time upwards.
– That is, had this subject died before the 15th week, she would not have
had the opportunity to enroll in the study, and the death would have gone
unobserved.
• Subject 5 joins the study 30 weeks after transplant and is observed
until 52 weeks, at which point the observation is censored.
– The subject’s death after 80 weeks goes unobserved.
• Subject 6 enrolls in the study at the date of transplant and is
observed alive up to the 10th week after transplant, at which point
this subject is lost to observation until week 35; the subject is
observed thereafter until death at the 45th week.
– This is an example of multiple intervals of observation.
– We only have an opportunity to observe a death when the subject is
under observation.
Introduction
Censoring example
time upwards.
unobserved.
under observation.
Introduction
Censoring example
time upwards.
unobserved.
under observation.
Introduction
Censoring example
time upwards.
unobserved.
under observation.
Introduction
Censoring example
We may be interested in characterizing the distribution of time to event”

for a given population as well as comparing this ”time to event” among
different groups e.g;
• treatment vs. control in a clinical trial or an observational study

or
• modeling the relationship of ”time to event” to other covariates

(sometimes called prognostic factors or predictors)
Typically, in biomedical applications the data are collected over a finite

period of time and consequently the ”time to event” may not be observed
for all the individuals in our study population (sample).
This results in what is called censored data. That is, the ”time to event”
for those individuals who have not experienced the event under study is
censored (by the end of study).

Introduction
Censoring example


or



Introduction
Censoring example


or



Introduction
Censoring example


or



Introduction
Censoring example


or



Introduction
Censoring example
It is also common that the amount of follow-up for the individuals in a

sample vary from subject to subject. The combination of censoring and
differential follow-up creates some unusual dificulties in the analysis of such
data that cannot be handled properly by the standard statistical methods.
Because of this, a new research area in statistics has emerged which is

called Survival Analysis or Censored Survival Analysis.

Introduction
Censoring example
It is also common that the amount of follow-up for the individuals in a

sample vary from subject to subject. The combination of censoring and
differential follow-up creates some unusual dificulties in the analysis of such
data that cannot be handled properly by the standard statistical methods.
Because of this, a new research area in statistics has emerged which is

called Survival Analysis or Censored Survival Analysis.

Introduction
Censoring example
In routine data analysis, we may first present some summary statistics such
as mean, standard error for the mean, etc. In analyzing survival data,
however, because of possible censoring, the summary statistics may not
have the desired statistical properties, such as unbiasedness.
For example, the sample mean is no longer an unbiased estimator of the

population mean (of survival time). So we need to use other methods to
present our data.
One way is to estimate the underlying true distribution. When this

distribution is estimated (either parametrically or nonparametrically), we
then can estimate other quantities of interest such as mean, median, etc.
of the survival time.

Introduction
Censoring example

present our data.


Introduction
Censoring example

present our data.


Introduction
Challenges in Survival Analysis
• The main outcome of the survival experiment is the time-to-event,

which is almost always nonnegative.
• The distribution of the time-to-event is typically skewed.
• Survival data are often right censored; that is, survival times are
known for only a portion of the individuals under study, and the
remainder of the survival times are known only to exceed certain
values.
• Typically, there are few subjects at risk in the tail of the survival curve
after sufficient follow-up, and therefore estimation of the tail of the
survival curve can be quite difficult.
• Generally, survival models are hard to fit due to the presence of
complex censoring schemes.
Introduction

values.
Introduction

values.
Introduction

values.
Introduction

values.

Survival - Notes (Lecture 1)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Survival - Notes (Lecture 1)

Uploaded by

Copyright:

Available Formats

Lecture 1: Survival Models and Analysis

Ms. Beryl Ang’iro

August 21, 2020

Ms. Beryl Ang’iro August 21, 2020 1 / 17

In many biomedical applications the primary end-point of interest is time

Ms. Beryl Ang’iro August 21, 2020 2 / 17

In many biomedical applications the primary end-point of interest is time

Ms. Beryl Ang’iro August 21, 2020 2 / 17

In many biomedical applications the primary end-point of interest is time

Ms. Beryl Ang’iro August 21, 2020 2 / 17

In many biomedical applications the primary end-point of interest is time

Ms. Beryl Ang’iro August 21, 2020 2 / 17

In many biomedical applications the primary end-point of interest is time

Ms. Beryl Ang’iro August 21, 2020 2 / 17

Example: Times to Weaning of Breast-fed Newborns

Youths, aged 14 to 21 in 1979, were interviewed yearly through 1988.

The response variable in the data set is duration of breast feeding in

Ms. Beryl Ang’iro August 21, 2020 3 / 17

Example: Times to Weaning of Breast-fed Newborns

Youths, aged 14 to 21 in 1979, were interviewed yearly through 1988.

The response variable in the data set is duration of breast feeding in

Ms. Beryl Ang’iro August 21, 2020 3 / 17

Bone marrow transplants are a standard treatment for acute leukemia. In

Ms. Beryl Ang’iro August 21, 2020 4 / 17

Bone marrow transplants are a standard treatment for acute leukemia. In

Ms. Beryl Ang’iro August 21, 2020 4 / 17

Example: Time to First Use of Marijuana

Turnbull and Weiss (1978) report part of a study conducted at the

• the exact ages (uncensored observations);

Ms. Beryl Ang’iro August 21, 2020 5 / 17

Example: Time to First Use of Marijuana

Turnbull and Weiss (1978) report part of a study conducted at the

• the exact ages (uncensored observations);

Ms. Beryl Ang’iro August 21, 2020 5 / 17

Example: Time to First Use of Marijuana

Turnbull and Weiss (1978) report part of a study conducted at the

• the exact ages (uncensored observations);

Ms. Beryl Ang’iro August 21, 2020 5 / 17

Example: Time to First Use of Marijuana

Turnbull and Weiss (1978) report part of a study conducted at the

• the exact ages (uncensored observations);

Ms. Beryl Ang’iro August 21, 2020 5 / 17

Let C1 ,...,Cn be a sample of censoring times.

Ms. Beryl Ang’iro August 21, 2020 7 / 17

Let C1 ,...,Cn be a sample of censoring times.

Ms. Beryl Ang’iro August 21, 2020 7 / 17

Let C1 ,...,Cn be a sample of censoring times.

Ms. Beryl Ang’iro August 21, 2020 7 / 17

Let C1 ,...,Cn be a sample of censoring times.

Ms. Beryl Ang’iro August 21, 2020 7 / 17

Regardless of the type of censoring, we must assume that it is

Regardless of the type of censoring, we must assume that it is

Regardless of the type of censoring, we must assume that it is

Imagine a study of the survival of heart-lung transplant patients who are

Imagine a study of the survival of heart-lung transplant patients who are

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 1 is enrolled in the study at the date of transplant and dies

• Subject 3 is enrolled in the study at the date of transplant, but is lost