Lecture #10 - Hunting Associations P2 Flashcards
1
Q
Measures of association:
- Quantify how much the……
- Relative Risk
A
- Exposure increases or decreases the likelihood of the outcome.
- Ratio of incidences: RR
- How many times likely the exposed group is to develop the outcome than the comp group
2
Q
Risk difference
- Also called
- Difference in what?
- “How many extra/few…..”
- What is the null value?
- What if Ie > Ic?
- What if 5 other way around?
A
- Attributal risk
- Difference in incidences (Ie - Ic) - can use CI or IR but both needa be same
- How many cases of outcome in the exposed group are attributed to the exposure
- 0
- Risk factor - excess amount go disease/cases that can be attributed to the exposure
- Protective factor`
3
Q
Interpreting risk diff
A
There were VALUE extra/fewer cases of OUTCOME in EXPOSED GROUP compared to COMP GROUP buuuuuuut reporting is different for CI and IR
- CI = extra/fewer cases per 100 ppl over 10 years
- IR = extra cases per 100 person-years
4
Q
Which one to use - RR or RD?
A
RR:
- clues to aetiology (causes) - further away from null = stronger ass = more likely a cause
- tells you about strength of ass
RD:
-impact of exposure - how many additional cases likely to be due to exposure (what happens if you take incidence away?)
5
Q
What are the three measures of impact?
A
Pop attributed risk
Pop attributable fraction
Number needed to treat
6
Q
Pop attributable to risk (PAR)
- Formula etc
- How many extra/fewer……
- Interpretation
A
- (incidence in total pop) - (incidence in comp group)
- excess cases of outcome in total pop (not just exposed group like RD) attributable to risk exposure. Can use both incidences. - How many extra/fewer cases of the outcome in the total population are attributable to the exposure? (if we eliminated that exposure from pop, how many cases could we prevent of outcome?)
- There were VALUE extra/fewer cases of OUTCOME in TOTAL POP attributable to EXPOSURE
7
Q
Pop attributable fraction (PAF)
1. Formula and what it means
A
- PAR/(incidence in total population)
- What proportion of the total population incidence is attributable to the exposure e.g. 33% of cancer cases in the total pop can be attributed to to smoking (so can prioritise interventions)
8
Q
PAR/PAF important points
- About prevlance
- PAF can be more than…
- PAR and PAF represent what?
- Both assume what relationship between exposure and outcome
A
- Depends on prevalence of exposure in total pop so specific to that population (if high prev + strongly associated with outcome = more outcomes around). Something may have huge RD but not that prevalent so not worth prioritising. So increase pre = increase PAR = increase PAF
- if exposure has high prev, would expect removing it to have a greater impact on total pop incidence than if had a low prevalence - PAF can sum to more than 100%
- PAR and PAF represent the max amount of disease that can be prevented (max amount of incidence you can get rid of if took away exposure but not the case because other factors too)
- Both assume a causal relationship between exposure and outcome
9
Q
Clinical impact
- RD can help…..
- Can use it to determine the number of people you need to.,…..
A
- RD can help assess clinical impact of an intervention
- Can use it to determine the no. of people you need to treat w/ an intervention to prevent one case of the outcome = NNT
10
Q
Number needed to treat
- Formula for both CI and IR
- Interpretation
A
- CI: 1/RD and IR: (person-years)/RD
- E.g. 20 = NNT - we would need to treat 20 ppl permanently w/ medicine x to prevent ONE heart attack (always round up to nearest whole number person)
- Comparing across studies - need PECOT to be same. E.g. if one study used placebo as comparison group and other uses current medication then RD much greater in placebo so NNT much smaller in placebo
- Better alternatives? -medicines that are more pleasant/more convenient
- Adverse effects? - could be effective but cause other problems
- Cost-effective
