Study design Flashcards
What are the types of studies?
Descriptive Migrant Analytical Epidmiology - cross-sectional, case-control, cohort Experimental In vitro lab studies
Define bias
estimate of systematic measurement error
What are the types of bias?
Selection bias
Information bias
Prevalence bias
Define selection bias
bias because of errors in selection of study population or selective loss-to-follow up
Define information bias
bias because of errors in the measurement of exposure or the disease outcome
Define prevalence bias
slowly progressing cases are more likely to be sampled in a cross-sectional survey than those with severe/faster disease progression
Define counding
Mixing of the effects of exposure on the outcome with the effects of a third variable
When is a factor a confounder?
Affects risk of the studied health outcome
Associated with the exposure of interest
Not an intermediate in the causal chain between the exposure of interest and the health outcome
What is mediation?
factor that is associated with the exposure of interest and the disease outcome and lies on the causal pathway
Give an example of exposure, outcome and confounder
Exposure = Controlling mothers
Outcome = High fat/ Low fat
Confounder = Academic focus will lead to more controlling mothers and it lead to high fat/ low fat intake cuz kids are stressed
Not a confounder = gene ->high fat/low fat but no association with exposure
Give an example of mediation
Exposure = Dairy
Outcome = Hip fractures
Mediation = Calcium
Dairy causes hip fractures through calcium
What is the bias in this example?
Case: persons with pancreatic cancer identified through hospitals
Controls: persons without pancreatic cancers identified through the same physicians as the cases
Result: coffee consumption associated with pancreatic cancer
Selection bias. Controls did not have coffee consumption habits representative of the source population / diet not reflective of population as a whole
What is the bias in this example?
Case: women with asthma
Controls: women without asthma with the same family doctor
Result: lower fruit consumption is associated with asthma
Selection bias. Controls that responded may have been a very motivated, health conscious selection with higher fruit consumption than source population
What is the bias in this example?
Case: women with breast cancer
Controls: women without breast cancer
Result: higher fat intake was associated with higher risk breast cancer
Recall bias. Women with breast cancer may have over-estimated their past fat- intake. They might think why they have breast cancer and not others? then they think its their consumption of fat
What are case reports?
description of rare clinical events
Highly detailed, can use sophisticated methods
Unable to estimate frequency or the role of bias or chance
What are case series?
study of a larger group of patients witha particular health condition
limitation: absence of comparison group
What is an ecological study?
Observational studies in which the unit of analysis is a group rather than an individual. (usually used for making hypothesis, rather than causal effect)
What are the limitations of an ecological study?
Many differences that are difficult to disentangle (confounders)
Ecological fallacy - inferences about individuals based on group level data
Device a randomised controlled trial
population - group 1 - treatment - outcome
-group 2 - control - outcome
difference is just treatment given or not given
What are the potential limitations of randomised controlled trial?
cannot blind people cuz nuts is not a type of drug that can be disguised
Compliance
Ethical issues
Generalisability
High cost and long duration
Device a cohort study
Population - disease free - unexposed (D or ND)
-exposed (D or ND)
look at population through time (forward)
Evaluate a cohort study
exposure assessed before disease diagnosis which reduces selection and recall bias
Limitations: residual confounding and loss-to-follow up can be potential problems
May not be feasible with rare diseases and exposure-diseaase relationships with long induction period
what is the difference between case-control and cohort study?
cohort study: selection of disease-free population and then categorising them into exposure and non-exposure group them do a follow-up to see who developed the disease and who did not
case-ctrl: selection based on outcome (case vs controls) and then look into past and compare exposure
What is a case-control study?
comparison of the frequency of exposure between cases and controls
selection of incident cases within a specific period of recruitment
selection controls: from comparable population without the disease (eg community, other hospitals)
What are the limitations of case-control study?
prone to recall and selection bias as comparedd to cohort study
What are the advantages of cross-sectional studies?
Advantages:
fairly simple, rapid and inexpensive
can study a wide range of factors
Disadvantages:
not good for rare diseases (prevalence of chicken pox)
reverse causation: “chicken or egg”
length bias
Interpretation of results?
was the finding due to bias?
was the finding due to chance?
was the finding due to confounding?