Summary Research Semester 3
Population: cancer etiology and prognosis
Confounding = a third variable that is associated with both the risk factor and the
outcome. Because of the confounding factor, it may seem that there is a correlation
between the potential risk factor and the outcome, which does not really exist.
- For example: you want to investigate the correlation between alcohol
consumption and lung cancer. However, many people who drink alcohol also
smoke. When not correcting for smoking, it may seem like alcohol increases
the risk of lung cancer. But in reality, there is no correlation between alcohol
consumption and smoking.
Effect modification = the magnitude of the effect of the primary exposure on an
outcome differs depending on the level of a third variable. The effect modifier is not
correlated with the risk factor. You correct for it by interpreting the results separately
by the level of the other variable.
- For example: you want to investigate the correlation between asbestos and
lung cancer. Asbestos increases the risk of lung cancer. However, the
increased risk for smokers is much higher than for non-smokers. Which makes
smoking an effect modifier.
Confounding can be prevented by using randomisation, restriction (= no variation
regarding the confounder in study population) or matching.
You can also adjust for confounding using:
- Mantel-Haenszel procedure for few nominal confounders (gender, eye
colour, blood type)
- Logistic regression for multiple or continuous
confounders (height, weight, time)
o S-curve
, The crude rate is calculated by dividing the number of deaths by the total population
deaths
( ). The crude rate is not corrected for age, which is a huge confounder for
population
morbidity and death. Standardisation helps compare mortality rates between groups
with differing age profiles and is thus a way to correct for confounders.
- Direct standardisation if
the study population had the
same age distribution as the
standard population, what
would be its death rate?
1. Get age-specific death rates for population of interest
2. Use these rates to find the expected deaths in the standard population
3. Find the age-standardised mortality rate
total expected deaths
o Age standardised mortality rate =
total population
- Indirect standardisation if
the standard population had
the same age distribution as
the study population, what
would be its death rate?
Indirect standardisation is used when there is no age-specific data.
1. Find mortality rates in standard population stratified by age group
2. Use age distribution of study population to find expected deaths
3. Compare total expected deaths to actual deaths
actual deaths
o SMR (standardised mortality ratio) =
expected deaths
o The standardized mortality rate (SMR) is the ratio of the number of
deaths observed in a population over a given period to the number
that would be expected over the same period if the study population
had the same age-specific rates as the standard population.
Etiologic fractions measure the number of cases of disease due to a specific
exposure of interest.
PAF (population attributable fraction) = proportional reduction in population disease
or mortality would occur if exposure to a risk factor were reduced to an alternative
ideal exposure scenario.
p ( RR−1)
PAF =
p ( RR−1 )+1
- p = population exposed
- RR = relative risk
Survival analysis = analysis of times from a
well-defined start time until the occurrence of
some particular event of interest.
Population: cancer etiology and prognosis
Confounding = a third variable that is associated with both the risk factor and the
outcome. Because of the confounding factor, it may seem that there is a correlation
between the potential risk factor and the outcome, which does not really exist.
- For example: you want to investigate the correlation between alcohol
consumption and lung cancer. However, many people who drink alcohol also
smoke. When not correcting for smoking, it may seem like alcohol increases
the risk of lung cancer. But in reality, there is no correlation between alcohol
consumption and smoking.
Effect modification = the magnitude of the effect of the primary exposure on an
outcome differs depending on the level of a third variable. The effect modifier is not
correlated with the risk factor. You correct for it by interpreting the results separately
by the level of the other variable.
- For example: you want to investigate the correlation between asbestos and
lung cancer. Asbestos increases the risk of lung cancer. However, the
increased risk for smokers is much higher than for non-smokers. Which makes
smoking an effect modifier.
Confounding can be prevented by using randomisation, restriction (= no variation
regarding the confounder in study population) or matching.
You can also adjust for confounding using:
- Mantel-Haenszel procedure for few nominal confounders (gender, eye
colour, blood type)
- Logistic regression for multiple or continuous
confounders (height, weight, time)
o S-curve
, The crude rate is calculated by dividing the number of deaths by the total population
deaths
( ). The crude rate is not corrected for age, which is a huge confounder for
population
morbidity and death. Standardisation helps compare mortality rates between groups
with differing age profiles and is thus a way to correct for confounders.
- Direct standardisation if
the study population had the
same age distribution as the
standard population, what
would be its death rate?
1. Get age-specific death rates for population of interest
2. Use these rates to find the expected deaths in the standard population
3. Find the age-standardised mortality rate
total expected deaths
o Age standardised mortality rate =
total population
- Indirect standardisation if
the standard population had
the same age distribution as
the study population, what
would be its death rate?
Indirect standardisation is used when there is no age-specific data.
1. Find mortality rates in standard population stratified by age group
2. Use age distribution of study population to find expected deaths
3. Compare total expected deaths to actual deaths
actual deaths
o SMR (standardised mortality ratio) =
expected deaths
o The standardized mortality rate (SMR) is the ratio of the number of
deaths observed in a population over a given period to the number
that would be expected over the same period if the study population
had the same age-specific rates as the standard population.
Etiologic fractions measure the number of cases of disease due to a specific
exposure of interest.
PAF (population attributable fraction) = proportional reduction in population disease
or mortality would occur if exposure to a risk factor were reduced to an alternative
ideal exposure scenario.
p ( RR−1)
PAF =
p ( RR−1 )+1
- p = population exposed
- RR = relative risk
Survival analysis = analysis of times from a
well-defined start time until the occurrence of
some particular event of interest.
