Cohort studies Flashcards
Cohort study
Observational study - we don’t intervene - just observe
Features of cohort study
- Cohort that is group at risk of outcome (disease-free)
- Cohort is classified according to exposure
- Observe subjects longitudinally
- Compare risk of outcome in those with/without exposure
Types of cohort study
Retrospective or Historical
Prospective or concurrent
Differ in the timing of collecting exposure and disease information
prospective
- Today we identify and measure our exposure
- Follow them up into the future
- At some point enough outcomes occur -> then do analysis
At the start of the study none of the outcomes have occured yet
retrospective
- Today we identify population and assess exposure in the past
- then do analysis of outcomes
- ALL the outcomes have already occured at the time we start the study
Pros and cons of prospective cohort studies
Exposure and outcome have not yet taken place when the study is initiated
- Can determine which variables to measure and how, including changes over time
- Can reduce selection bias as outcome is yet to occur
Long time and expensive
May not be ethical to wait for answer
Studies of rare disease or diseases with long latency may be difficult
Retrospective
Exposure and outcome have taken place before the study is initiated
- Variables may not be available and little control of method or consistency of measurement
Missing data - selection bias more likely as exposure and outcome have already occured
- can be done relatively quickly and may be cheap
- good for rare diseases with long latency
Biobank studies
- Vast range of different variables collected prospectively at baseline and subsequently
- Blood and urine samples
- No specific research question asked at the time of data collection
- Studies nested with the Biobank described as prospective
1. Data collected prospectively
2. Although specific question asked retrospectively - Contrast with database studies - retrospective
Designing a cohort study
- Research question (PICO)
- Is a cohort study the right decision?
Stages: 1. DEsign Selection of study population 2. Data collection 3. Analysis
Issues:
1. Bias, confounding, random error
Non-participation or non-response bias
Affects power to detect an association
Certain groups may be over or under represented
Non-response bias (only if non-response is related to exposure and outcome)
Type of selection bias
Losses to follow-up and bias
Losses inevitable in most cohort studies
Loss of participants in follow-up reduces the power and precision of the study
Differential loss to follow-up, with respect to baseline characteristics and outcome may lead to bias
High loss to follow-up doesn’t mean bias is introduced but makes it more likely
Rules of thumb: <5% loss leads to little bias, >20% serious threats to validity
Lossed to follow-up
-> how to avoid?
Maintaining a high follow-up rate
- choice of study population (where contact can be easily maintained)
- incentives
- linkage/flagging (patient registers, patient records)
- shorter follow-up
- can be labour intensive
BEST to avoid loss but you can compare baseline characteristics of those lost vs. not
Survivor bias (prevalent cohort bias)
Survival bias is a type of sampling error or selection bias that occurs when the selection process of a trial favours certain individuals who made it past a certain obstacle or point in time and ignores the individuals who did not (and are generally less visible).
Outcome data collection in cohort studies
- Some people leave a study before the end
- SOmething might happen that prevents the outcome of interest from happening, e.g.
1. Outcome=heart attack (some may have preventive surgery)
2. Outcome=recurrence of cancer (some may die from another cause first)
3. Some people may still not have experienced the outcome by the end of the study
Called CENSORED
- must take into account for analysis
Risk ratio
-risk of disease in exposed individuals relative to unexposed
