Cohort studies Flashcards
Where in the heirachy of evidence do cohort studies lie?
Second
What is a cohort?
sample from population without the outcome of interest (exposure is measured at baseline and they are followed up to see if those exposed to the exposure had more/less of a risk to develop outcome -> risk ratios)
What are the sources of both exposure and outcome data?
Health records
Interviews
Quiestionnaires
Physical examinations
What are sources of exposure data only?
Employment records
What are sources of outcome data only?
Death certificates
What are the two types of cohort study?
Prospective (those exposed/not followed up to observe incidence of outcome)
Historical (data on exposure from pre-existing historical records -> follow subjects with records availiable to see if developed outcome of interest at any time up to the present)
Which takes less time to undertake… Historical or prospective cohort studies?
Historical -> do not need to wait for the outcome to occur
How do the two types of cohort studies compare?
- Historical = quicker
BUT
- Historical records may not be availiable = not possible
- Exposure data not collected for research purpose = possibly poor quality
What is risk?
Risk = no. new cases of outcome in time period / no. in population without outcome at outset
Calculated seperately for both exposed and unexposed group
= porbability between 0 & 1
What is risk ratio?
risk ratio = risk in exposed group/risk in unexposed group
How do you interpret risk ratio?
RR = 1 -> no difference between exposure groups
RR = 0.5 -> exposure halves risk
RR = 0.75 -> exposure reduces risk by 25%
RR < 1 exposure reduces the risk
RR = 2 -> exposure doubles the risk
RR = 10 -> exposure increases the risk ten fold
RR > 1 = exposure increases the risk
What does risk ratio indicate?
Strength of association between exposure and outcome
BUT
associations may not be causal
Which two things should be used to interpret risk ratio values calculated from cohort studies?
P values (measure strength of evidence against null hypothesis)
CI intervals (assess orecision of estimated RR)
How do you test for a trend?
Calculate RR for each category compared to baseline
RR in baseline group = 1
If there is an association/trend increase in exposure = increase in outcome
What are the two different types of trend?
Linear trend (increase risk in outcome assiciated with increase or decrease in exposure)
Threshold effect (risk of outcome increases only for those subjects whose exposure is above or below a certain level)
** INSERT PIC ***
What is risk difference?
Risk difference = risk of outcome occurence in exposed - risk of outcome occurence in unexposed
Which values can be calculated form risk differences?
CI intervals
P values
What is the difference between risk ratios and risk differences?
Risk ratio measures strength of association between exposure and outcome (relative measure = no units)
Risk difference measures impact of exposure on outcome (has units) -> depends on length of study = cannot compare different studies
Which is more preferrable… risk ration or risk differences?
Risk ratio
because it is NOT depenent on length of study
In a prospective cohort study what can measurement bias affect?
- Unlikely to affect exposure as outcome is unknown at the time of measurement
- May affect outcome as exposure of group will be known
What are the strengths of (prospective) cohort studies (5)?
- Exposure preceeds outcome
- Outcome incidence measured directly
- Can test hypotheses about >1 outcome
- Selection bias unlikely (selected as a single gorup then classified accoridng to exposure)
- Measurement bias for exposure unlikely
What are the weaknesses of a (prospective) cohort study (6)?
- Time consuming
- Expensive
- Rare outcomes require very large sample sizes
- Changes over time can affect exposure and outcome classification
- Measurement bias for outcome
- Loss to follow-up bias (rate of loss of subjects may be related to outcome)