Introduction into survival

analysis and basic

demography
2. Survival analysis
Vladimir Kozlov
IOS (Regensburg)
24. – 29.05.2024
ITMO, Biostatistics and Demography
[email protected]
 Main definitions
• In many demographic and epidemiologic studies, the
main outcome under assessment is the time to an event
of interest.
• The generic name for the time is survival time, although
it may be applied to the time ‘survived’ from complete
remission to relapse or progression as equally as to the
time from diagnosis to death.
• However in demography the survival analysis in more
cases ís used to the time to the birth or marriage
 Censoring
• The specific difficulties relating to survival analysis arise largely from the
fact that only some individuals have experienced the event and,
subsequently, survival times will be unknown for a subset of the study
group.
This phenomenon is called censoring and it may arise in the following
ways:
• (a) a patient has not (yet) experienced the relevant outcome, such as
relapse or death, by the time of the close of the study;
• (b) a patient is lost to follow-up during the study period;
• (c) a patient experiences a different event that makes further follow-up
impossible. Such censored survival times underestimate the true (but
unknown) time to event.
 Right
• Visualizing the survival process of an individual as a time-line, their
event (assuming it were to occur) is beyond the end of the follow-up
period. This situation is often called right censoring.
 Left
• Censoring can also occur (be left censored) if we observe the presence
of a state or condition but do not know where it began. For example,
consider a study investigating the time to recurrence of a cancer following
surgical removal of the primary tumour. If the patients were examined 3
months after surgery to determine recurrence, then those who had a
recurrence would have a survival time that was left censored because the
actual time of recurrence occurred less than 3 months after surgery.
 What to do with censoring
• In general, the feature of censoring means that special methods of
analysis are needed, and standard graphical methods of data exploration
and presentation, notably scatter diagrams, cannot be used
 Survival data (2 probabilities)

• Survival data are generally described and modelled in terms of two

related probabilities, namely survival and hazard.
 Survival data (2 probabilities)
• The survival probability (which is also called the survivor function) S(t) is the
probability that an individual survives from the time origin (e.g. diagnosis of
cancer) to a specified future time t. It is fundamental to a survival analysis
because survival probabilities for different values of t provide crucial summary
information from time to event data. These values describe directly the survival
experience of a study cohort.
 Survival data (2 probabilities)
• The hazard is usually denoted by h(t) or λ(t) and is the probability that an
individual who is under observation at a time t has an event at that time.
• Another way, it represents the instantaneous event rate for an individual who has
already survived to time t. Note that, in contrast to the survivor function, which
focuses on not having an event, the hazard function focuses on the event
occurring. It is of interest because it provides insight into the conditional failure
rates and provides a vehicle for specifying a survival model. In summary, the
hazard relates to the incident (current) event rate, while survival reflects the
cumulative non-occurrence.
 Kaplan–Meier (KM) survival estimate
• The survival probability can be estimated nonparametrically from observed
survival times, both censored and uncensored

• – is a probability of being alive at time tj

• nj the number of patients alive just before tj, and dj the number of events at t
• = 0 and S(0) = 1
 KM visualization
• The KM survival curve, a plot of the KM survival probability against
time, provides a useful summary of the data that can be used to
estimate measures such as median survival time. The large skew
encountered in the distribution of most survival data is the reason
that the mean is not often used.

• Let us follow the examples from

Clark, T. G., Bradburn, M. J., Love, S. B., & Altman, D. G. (2003). Survival analysis part I: basic
concepts and first analyses. British journal of cancer, 89(2), 232-238.
Example of the table
Example of the graph
• Instead of S(t), the incidence of death curve, or 1-S(t), may be
presented. The cumulative incidence at a time point is simply one
minus the survival probability.
Surviving vs. Incidence
 Hazard and cumulative hazard
• There is a clearly defined relationship between S(t) and h(t), which is given by the
formula:

• Cumulative hazard - H(t)

• This is defined as the integral of the hazard, or the area under the hazard function
between times 0 and t, and differs from the log-survivor curve only by sign, that is
H(t) = -log(S(t)).
• The interpretation of H(t) is close to the situation that H(t) is as the cumulative
force of mortality, or the number of events that would be expected for each
individual by time t if the event were a repeatable process. H(t) is used an
intermediary measure for estimating h(t) and as a diagnostic tool in assessing
model validity
Cumulative hazard
 Hazard ratios
• Assume, that we have total expected number of events in each group, say Ei for
group i. And the total observed number of events, say Oi for treatment group i
• The hazard ratio (HR) is a measure of the relative survival experience in the two
groups and may be estimated by

• where Oi/Ei is the estimated relative (excess) hazard in group i. A confidence

interval (CI) for the HR can be calculated .
• The HR has a similar interpretation of the strength of effect as a risk ratio. An HR
of 1 indicates no difference in survival. In practice, it is better to estimate HRs
using a regression modelling technique, such as Cox regression
 Multivariate data analysis in survival
analysis
• The Cox (proportional hazards or PH) model is the most commonly used multivariate
approach for analysing survival time data in medical research. It is a survival analysis
regression model, which describes the relation between the event incidence, as
expressed by the hazard function and a set of covariates:
h(t) = h0(t)*exp
• where the hazard function h(t) is dependent on (or determined by) a set of p
covariates (x1, x2, …, xp), whose impact is measured by the size of the respective
coefficients (b1, b2, …, bp). The term h0 is called the baseline hazard, and is the value
of the hazard if all the xi are equal to zero (the quantity exp(0) =1).
• The Cox model is essentially a multiple linear regression of the logarithm of the hazard
on the variables xi, with the baseline hazard being an ‘intercept’ term that varies with
time. The covariates then act multiplicatively on the hazard at any point in time, and
this provides us with the key assumption of the PH model: the hazard of the event in
any group is a constant multiple of the hazard in any other.
 Proportionality

• Proportionality implies that the quantities exp(bi) are called hazard ratios.
A value of bi greater than 0, or equivalently a hazard ratio greater than 1, indicates
that as the value of the ith covariate increases, the event hazard increases and thus
the length of survival decreases.
A hazard ratio above 1 indicates a covariate that is positively associated with the
event probability, and thus negatively associated with the length of survival.
This proportionality assumption is often appropriate for survival time data but it is
important to verify that it holds
 Examples
• Grjibovski, A. M., Dubovichenko, D., Saduakassova, S., Zhatkanbayeva, G.,
Omarova, G., Shalgumbayeva, G., ... & Valkov, M. Y. (2018). Incidence,
mortality and determinants of survival from cervical cancer in Northwest
Russia: a registry-based cohort study. International health, 10(2), 92-99.
https
://academic.oup.com/inthealth/article/10/2/92/4843995?casa_token=_ym-5jwv434AAAAA:j71rWU1o1a__
dIIkobqKl-9nucLNn5HUeHOJXhIt0f33aZXuCZgkrhZUUi0OnRwdUjWJQD2Zy3uCqyg

• Koupil, I., Plavinskaja, S., Parfenova, N., Shestov, D. B., Danziger, P. D., &
Vågerö, D. (2009). Cancer mortality in women and men who survived the
siege of Leningrad (1941–1944). International Journal of Cancer, 124(6),
1416-1421.
https://onlinelibrary.wiley.com/doi/epdf/10.1002/ijc.24093
 Parametric models
• A number of different parametric PH models may be derived by choosing
different hazard functions. As shown previously, there is a direct link between
the survival and hazard, and the choice of hazard distribution determines that
of the survival. In fact, the models commonly applied, such as the Exponential,
Weibull or Gompertz models, take their names from the distribution that the
survival times are assumed to follow, but the most distinguishing features
between them are in the hazard function.
 Interpretation of the results (additionally)
• In addition to the ratio of two hazards, it is possible to obtain other
information from a PH regression model. One simple (and possibly
underused) quantity that may be derived from a survival model is the
predicted survival proportion at any given point in time for a
particular risk group. The survival𝛾
proportion for a given risk group at
any time, S(t), is equal to

covariates are equal to zero) and 𝛾 is equal to b1x1 þ b2x2 +…+ bpxp.
• where S0(t) is the baseline survival (the survival proportion when all
 The accelerated failure time (AFT) model
Instead of HR they use TR (time ratio)

𝜑 is an ‘acceleration factor’ that depends on the covariates according to the formula :

𝜑=exp
Log(T) = b0 + +ε
ε is a measure of (residual) variability in the survival times.
Thus, the survival times can be seen to be multiplied by a constant effect under this
model specification, and the exponentiated coefficients, exp(bi), are referred to as
time ratios.
• A time ratio above 1 for the covariate implies that this ‘slows down’, or prolongs the
time to the event,
• a time ratio below 1 indicates that an earlier event is more likely.
 Interpretation of TRs
• We compare the effect of adjuvant chemotherapy (treatment) and
radiotherapy (control) alone.
• The TR is 2.05 (unadjusted) and 1.9 adjusted.
• Therefore, we can conclude that the time to recurrence was
significantly prolonged (approximately doubled) among patents given
adjuvant chemotherapy in comparison with those who were not.
• Patients allocated to receive adjuvant chemotherapy had a predicted
median survival time of approximately 16 months, as opposed to 8
months among those treated with radiotherapy alone.