Define incidence rate
Number of new cases, per year, per population IR = new events/(number of people x time)
Define incidence rate ratio
Comparing the risk of 2 populations (relative risk)
Define prevalence
Measure of the number of people who currently have the disease Prevalence = number of cases/population
What does a p value of
The likelihood of value occurring by chance is
What is the 95% confidence interval
95% certainty that the true value lies within confidence interval If null hypothesis value inside 95% CI then p>0.05 If null hypothesis value outside 95% CI then p
How do you calculate CIs?
Lower bound = IRR/error factor Upper bound = IRR x error factor Error factor = exp (2 x (1/d))
Describe a case control study
Start with cases of disease, then get controls for each case (up to 6) Investigate exposure of interest Compare using odds ratio
How do you calculate odds ratio?
OR = ad/bc
Define incidence
Number of new cases in a given time (absolute risk)
What are the strengths of case-control study?
Quick and relatively inexpensive
Possible to look at lots of different exposures in detail
Good for rare outcomes
What are the weaknesses of case-control studies?
- Not good for rare exposures
- Prone to information bias and selection bias
- Can be impossible to determine whether disease causes exposure or vice-versa
Describe a cohort study
- Compares risk in one group with risk of another group (e.g. Group exposed to asbestos compared to group not exposed to asbestos)
- Follow up disease free individuals over time
- Comparisons made using IRR or SMR (if using reference population)
What are the strengths of a conort study?
- Allos for detailed and prospective assessment of exposures, outcomes, confounders
- Good for rare exposures and establishing temporal sequence
What are the weaknesses of cohort studies?
- Study and reference populations may not be comparable due to selection bias - e.g. The healthy worker effect
- Often there is limited data on reference population
- Large, time consuming and resource intensive
- Survivor bias
- Not good for rare diseases
Describe a radomised control trial
- Clinical trial in which interventions are studied
- Allocate imdividuals into two groups matching sex, age, social class to minimise confounders
- Give one group intervention, give other group placebo or original drug
What can cause one treatment to be more successful in a RCT than another?
- Chance - minimised with long run trials
- Patient knows treatment - minimised by blinding
- Treating clinician knows treatment - minimised by blinding
- assessor/investigator knows treatment - minimised by blinding
- One treatment genuinely more effective
Describe the placebo effect
- Psychological benefit that derives from, being looked after, cared for more carefully or being on a new ‘special’ treatment
What is intension to treat analysis?
- All people in trial are included in analysis, regardless of whether they took treatment or not, or partially completed treatment
- Prwserves true randomisation and elimiating confounding factors
Define the Bradford Hill criteria
- S = strength of association
- S = specificity of association
- C = consistency of association
- T = temporal sequene
- D = dose response
- R = reversibility
- B = Biological plausibility
- C = coherence of theory
- A = analogy
