Measures of effect Flashcards
Which two processes are involved in epidemiology?
- Estimating the frequency and distribution of diseases in populations
- Comparing the effects of exposures on the frequency of health outcomes in a population
Recap what a risk is
Risk relates the number of new cases to the size of the population at risk at the beginning of the period of study
- the probability of the occurrence of disease
Number of New cases/Number of people at risk at the beginning of the study
Recap what a rate is
Rates relate the number of new cases to the Person-time at risk which is a measure that takes into account any changes to the size of the population at risk during the follow up period
Recap what an odds is
The ratio of people who get the disease to the people who do not get the disease
Number of new cases in a time period / Number of people who do not get the diseases in the time period
“Lecture theatre of 200 students for 1 hour, outcome of interest is sneezing at least once, in the hour, 50 people sneezed at least once”
what is the risk in this situation
What is the odds in this situation
Risk = 50/200 = 25% Odds = 50/150 = 0.33
What does a measure of effect compare?
The frequency of disease in populations which are exposed compared to that in populations which are not exposed to a particular risk factor
What does a measure of effect measure?
Measures the association between a disease and a risk factor
What is used to measure effects?
Measuring Ratios or Differences
Which Ratios can be be used to measure effect?
The rates, or risks, or odds of the exposed compared to the unexposed.
What are the names of the Ratios used to measure effect?
Rate Ratio
Risk Ratio
Odds ratio
Give the formula for rate Ratio
Rate in exposed (R1) : Rate unexposed (R0)
RR = R1/R0
Give the formula for the Risk Ratio
Risk In Exposed (R1) : Risk in unexposed (R0)
RR = R1/R0
Give the formula for Odds ratio
Odds in exposed (O1) : Odds in unexposed (O0)
Confusingly, what might all the ratios measuring effect be referred to as?
Relative risks
In chad, people who are HIV positive have 31 times the rate of Pneumonia compared to people who are HIV negative
Comprehend this
The rate Ratio here is 31
The rate of Pneumonia cases occurring in the exposed group is 31 times greater than in the unexposed group
In Brazil, the risk of low birthweight was 2 times higher in babies born to low income families compared to high income families.
Comprehend this
The Risk Ratio is 2
Babies from low income families are twice at risk of being born at a low birthweight compared to those born from a high income family
Among people who had direct contact with someone infected with Ebola, those exposed to bodily fluids had 3.6 times the risk of infection compared to those who were not exposed.
Comprehend this
The risk ratio is 3.6
Those exposed to bodily fluids had a 3.6 times greater risk of becoming infected compared to those that were not exposed
What is a protective effect and explain this
An exposure is known to have a protective effect when the ratio effect measures less than 1 as this indicates that this exposure is associated with a lower risk of the outcome
How do we always express odds and odds ratios?
How do we not express them?
We always express them as a number between 0-1
They are proportions so we never express them as a %
What do ratio measures measure in epidemiology?
What are they not a measure of?
They are a measure of the strength of the association between a risk factor and a disease
They are not a measure of how much disease is present in the population or how much of this was caused by a certain risk factor
What does the impact of a factor depend on?
How common it is in the population
How strong of an effect it has
Give an example of a strong association with little effect
There is a strong association between repeated x rays during pregnancy and leukaemia in childhood, however, as many people are discouraged from these x rays due to education, there are fewer occurring so this means that fewer of these leukaemia cases are from this exposure and therefore the impact of this exposure is reduced in the population
What do measures of effect measure, and when is the exception to this?
They always measure the strength of an association UNLESS this particular association is ENTIRELY causal
People with grey hair have 100 times the risk of getting cancer compared with people who don’t have grey hair, the risk ratio is 100.
Does Grey hair cause cancer?
In epidemiology, what is this an example of?
No.
Just because the association is strong does not mean that this is a causal relationship
There has been a confounding effect here, age is the confounder and is the true explanation for the observed association ( in other words age is the true causal factor here)
Using measures of effect, how do we generally go about measuring the effect of more than one level of exposure?
Choose a baseline group which will be used for comparison
This is usually the lowest exposure or no exposure
This is R0
Then we can group other exposures into their progressing levels of exposure
So the next lowest exposure after no exposure will be R1
Then the highest level of exposure will be R2
Compare R1 to R0 for the risk ratio
Compare R2 to R0 for the risk ratio
How may we select the baseline groups for comparison when comparing multiple exposures?
- By Natural choice - eg never having smoked is the natural exposure
- By convention - the group with the lowest risk - but not if it is a study of protective measures
- For particular statistical purposes - The group with the largest number of subjects
What is a difference measure?
A way of estimating the excess risk caused by exposure in the exposed group of a study
What are the two types of difference measures?
- Rate difference (RD) AKA risk difference
2. Rate Difference Percent (RD%) aka risk difference percent
What is the Rate Difference and how is it calculated?
It is the absolute or actual difference between two rates
Rate difference = Rate in the exposed group(R1) - Rate in the unexposed group (R0)
Rate Difference = (R1-R0)
Why don’t we often use Odds Differences?
They are too difficult to calculate and not possible to interpret
How do we interpret Rate or Risk differences?
Known as the attributable risk
Assuming causality, it is the percentage or number of cases which we can attribute to the exposure we are studying
What is the Rate Difference Percent?
The proportion of cases in the exposed group which were caused by the exposure
How do we calculate the Rate Difference Percent
Divide the Rate Difference by the Rate among the unexposed
RD% = (R1-R0)/R1
What can the Rate Difference Percent also be known as?
Attributable Fraction for exposed
Etiologic Fraction
Which two things do we always have to note with Difference measures?
- They do not take into consideration the rate in the whole population
- They are independent of the frequency of the exposure in the entire population
Briefly, which measures do take into account the rates in the entire population?
Measures of impact
What is a topical measure which is very similar to Rate Difference Percent and what is the definition of it?
Vaccine Efficacy
- The percentage reduction in incidence among vaccinated attributable to vaccination
What is the formula for Vaccine Efficacy
r0 = Unvaccinated R1 = Vaccinated
VE = (R0-R1)/R0 or VE - 1-(R1/R0)
How do we present Vaccine Efficacy when we calculate them?
“We estimate that Vaccine X appears to have reduced the incidence of X by X%”
What are the most important things to remember about Measures of Effect?
- They are actually measures of association, unless the association is known to be entirely causal
- Association may or may not be causal