S2 Measurement and Study Designs Flashcards
What is the purpose of statistics?
To generalise to/infer about the population
How do we making inferences about the entire population?
Take a sample which is representative, unbiased and precise
What are the two types of error that can occur in a study that may influence results?
- Chance/random error - due to sampling variation, will reduce as sample size increases
- Bias/systematic error - difference between the true value and the expected value, doesn’t reduce as sample size increase
What are sources of bias?
- Selection bias
2. Information bias
What are some types of selection bias?
- Study sample not representative of whole population of interest
- Group selection within a study - groups in a study not comparable
- Healthy worker effect - people who work have a lower overall mortality than the general population
What are some types of information bias?
- Recall error - differences in recollection
- Observer/interviewer error - preconceived expectations/knowledge
- Measurement error - differences in measurement
- Misclassification - participants put into wrong group usually arises from observational or measurement error
What is prevalence?
The proportion of people who have a disease at a given point in time
Often reported as a proportion
What is incidence?
The number of new cases of a disease within a given timeframe
Often reported as a rate
When is someone no longer counted in prevalence?
If they die or are cured
If they are cured and have a regression, they are classed as a new case (incidence)
What is the incidence rate ratio?
Compares the incidence rate in one group to another
It is a relative measure between two groups (or more groups)
How do you calculate incidence rate ratio?
IRR = incidence rate in group A / incidence rate in group B
How do you calculate incidence rate?
Incidence rate = number of new cases / (sum of) patient time at risk = events per person per year
Why do you measure incidence rate as a rate (e.g. patient time at risk)?
Not all patients are at risk throughout the whole study/whole life e.g. if someone is dead they are no longer at risk
How do you calculate prevalence?
Prevalence = number of people with the disease / total population
Do ‘relative risk and risk ratio’ and
‘relative rate and rate ratio’ have the same meanings?
Yes
If the probability of an event is p, how do you calculate the odds of that event?
Odds = p / (1-p)
How do you calculate odds ratio?
Odds ratio = odds of group A / odds of group B
This is a relative comparison
In cohort study, with consistent follow-up, we can calculate odds ratio/relative risk estimates directly, what is the calculation?
Relative risk = absolute risk for group A / absolute risk for group B
What would be the ratio if there was no difference between the groups?
1
What is the calculation for risk difference?
Risk difference = absolute risk in group A - absolute risk of group B
What is an issue to be aware of when comparing groups? What do you do about it?
Confounding factors (things you can’t control)
Adjust for differences in known confounding factors - standardisation
What is risk difference?
The absolute difference in risk between one group compared to another
What is relative risk/risk ratio?
Describes the comparison of the probability of an Evernote occurring in one group compared to another group
What are confounding factors?
Factors that, when comparing groups, the association/effect between an exposure and outcome is distorted by the presence of another variable
What are the four epidemiological study types?
- Ecological studies (descriptive)
- Cross-sections surveys (descriptive)
- Case-controlled studies (analytical)
- Cohort studies (analytical)
What is a descriptive study trying to do?
Sampling so can infer back to population
What is a analytical study trying to do?
Want to compare ‘like for like’ samples
How do you carry out an ecological study?
- Identify groups to study
- Define the characteristics to be studied - exposure and outcome
- Decide whether analysis is - counting categorical data (nominal/ordinal) or measuring continuous data (interval/ratio scale)
- Gather data on the group-level characteristics
What are the issues with ecological studies?
- definition of characteristics
- measurement variation
- confounding
- chance
How do you carry out a cross-sectional survey?
- What population do you want to generalise to?
- What population can you get access to?
- How can you get access to them?
- Who is in your study?
What are the issues with a cross-sectional survey?
- sampling bias
- responder/participant bias
- chance
How do you conduct a case-control study?
- Identify group of cases
- Identify suitable group of non-cases/controls
- Find out everyone’s previous exposure status
- Compare level of exposure in cases and controls
Are case-control studies retrospective?
Yes
What are the issues with case-control studies?
- Selection bias
- Information bias (e.g. exposure)
- Confounding
- Chance
How do you carry out cohort study?
- Can be historical/retrospective (looking back) or concurrent/prospective (find people before they have the disease and follow them up to when they got the diease) - identify outcome free individuals
- Group individuals according to level of exposure
- Ascertain outcomes for everyone
- Compare incidence rates for each exposure group
What are the issues with cohort studies?
- loss of follow-up
- information bias (e.g. for outcome)
- confounding
- chance
What can you use for the analysis of a cohort study?
Odds ratio or rates ratio (as you have a time scale)
What can you use for the analysis of a case-control study?
Odds ratio
