Population and Social Science Flashcards
Name the 4 types of observational studies.
- Descriptive
- Cross-sectional
- Ecological
- Analytical
- Case-control
- Cohort
Describe how an ecological study is conducted.
- Population level (group analysis)
- Observational
- Ecological fallacy (inference made about individuals from group data collected)
(Ecological fallacy= type of confounding variable)
Describe a cross-sectional survey/study.
- Measures exosure and disease- one point in time
- Data from individuals
Define prevalence and incidence rate.
Prevalence: Existing cases as proportion of population
Incidence rate: New cases in population per year (Person-years)
What can prevalence and incidence rate be used to measure?
Absolute risk (of event happening)
What is an incidence rate ratio?
Compares incidence rate of 2 populations
Is exposure associated with condition?
What is the difference between incidence rate ratio and risk ratio?
Incidence rate ratio= incidence rate
Risk ratio= prevalence
Identify a solution to limit the effect of confounding variables in a study.
- Selection (match confounders)
- Calculations (standardisation, stratifying etc)
What is the standardised mortality ratio?
SMR: Observed/expected deaths x 100
What does a SMR> 100 suggest?
Excess mortality with confounders accounted for
When measuring mortality rate, what’s the difference between direct and indirect standardisation?
Direct= uses reference population (same population numbers)
Indirect= uses mortality rates from one population to another (observed and expected deaths)
What is the sampling frame?
List or method through which sample= obtained
(assumed to be representative of population)
Why do we use an Error Factor (EF) with epedimiological studies?
- True value of population= different to sample
- Error factor used to help infer true value
- Size of EF depends on size of population
What is the confidence interval?
(Use EF to work out CI)
Interval where the true value lies
(which we can say with 95% confidence)
What are the 2 types of validity a study must have?
- Internal validity
- Freedom from confounding, bias, random error
- External validity
- Degree to which conclusions from study can be applied to population of interest
Define bias.
Systematic error….
…measuring exposure/ covariates/outcome variables…
…leads to erroneous association
What are the 2 main types of bias?
- Selection bias
- Information bias
What is selection bias?
Selected sample not representative of patient population
Why might selection bias occur?
- Systematic error selecting subjects
- Factors influencing participation
How might information bias occur?
- Misclassification of participants
- Recall error/bias (case-control problem)
- Observer error/bias
- Measurement error/bias
Describe how a cohort study is conducted?
- Recruit disease-free individuals
- Classify according to exposure
- Follow - long periods of time
- Monitor disease progress
- Calculate incidence rates
Retrospective/prospective
What are the advantages of cohort studies?
Good for:
- rare exposures (not good for rare outcomes)
- conditions fluctuating with age
- multiple outcomes
- showing exposure precedes outcomw
What are the disadvantages of cohort studies?
- Resource intensive
- Take long time
- Survivor bias
- Difficult with unknown confounders
How are case-control studies carried out?
- Recruit disease free and diseased individuals
- Determine exposure status of each group
- Produce odds ratio (measure of relative risk)
How are controls found for case-control studies?
Found to match confounders eg age, sex
Make 2 populations as similar as possible
What can be done when conducting a case-control study to minimise the EF (error factor)?
Use more controls than cases
Give some advantages of using a case-control study:
- Good- rare disease
- Cheap
- Quick
- Can study multiple exposure for single outcome
Give some disadvantages of using a case-control study:
- Prone to selection and information bias
- Not good for rare exposures
- Confounding
Looking at the following table, how is the odds ratio calculated?
Differentiate between cohort and case-control studies.
What is the Bradford-Hill criteria and what is it used for?
- Assesses how sure we are of a causal relationship
- Criteria:
- Association
- Exposure&Outcome
- Other evidence
Why are RCTs (randomised controlled trials) carried out?
To assess:
- Efficacy (benefit of treatment under specified conditions)
- Safety (not doing harm under specified conditions)
RCTs(randomised controlled trials) must be reproducible (under same conditions). They must also be 2 other things- what are these?
- Controlled
- Comparison of intervention
- Fair
- Unbiased, no confoundin/g
What 3 measures can be taken to ensure that randomised control trials (RCTs) are reproducible, controlled and fair?
- Randomisation
- Removes confounding and allocation bias
- Blinding
- Removes measurement bias and non-treatment effects
- Placebo
- (patients must be told if standard treatment available)
