Observational studies

Generating preliminary hypothesis of exposure/disease relationship is

a) descriptive reports/studies

b) analytical studies

a) descriptive reports/studies

testing hypotheses and looking for relationships is

a) descriptive reports/studies

b) analytical studies

b) analytical studies

T/F: observational studies allow the investigators to intervene if necessary

• false

  • NO INTERVENTION

  • only observe

what is selective bias? 

• When a study fails to select a representative sample

  • unequal chance of getting the outcome

what is information bias ?

• unequal measurement of exposures or outcomes in two study groups could occur

  • more likely to record outcomes in one study group

  • one group = seen more

  • one group = better able to recall/report exposres/outcomes

what is confounding? 

• distortion of the association between an exposure/intervention and outcome because a third variable = independently associated with both 

what are the 3 factors to consider a variable as a confounder? 

1. must be associated with exposure of interest (equally distributed between exposed/unexposed 

2. must be a risk factor for the outcome 

3. must NOT be a result of exposure (not part of the causal pathway) 

T/F: to control confounding in an observational study, one can employ randomization 

• FALSE

  • use restriction or matching

you want to find out how alcohol consumption affects mortality. what is the exposure? what is the outcome? what might be some confounding factors? 

• exposure = alcohol consumption 

• mortality = outcome 

• age, sex, ethnicity, etc = confounding variable

what are some limitations of observational studies?

• selective bias

• information bias 

• potential for confounding variables

T/F: only associations can be established in observational studies, not cause-effect

true

what is restriction? 

• exclude subjects with a certain confounding characteristic from the stud

• may limit generalizability of study (external validity) 

  • can be used in observational + experimental studies 

what is matching? 

• match subjects by certain characteristics between study groups

• common in case-control study design

what is propensity score?

• probability of being in the exposed (or case) group conditional on observed baseline characteristics(covaries/confounders) 

what is matching by propensity score? 

• summarizes many confounders into a single score (0-1) 

  • makles groups more comparable

• PS = estimated using regression techniques

• mainly used in cohort study design

what is the limitation of PS matching?

• matching does not address bias form unmeasured confounders 

how do you control confounding in statistical analysis?

1. stratification

2. regression analysis 

what is stratification?

• stratify subjects based on confounder 

  • create homogenous groups for comparison 

• useful with few confounders 

what is regression analysis?

• useful for many potential confounders

• confounders = predictor variables 

• results = “adjusted” based on confounders 

  • large difference = suggests confounding 

T/F: Observational studies, such as cohort studies, are able to show a cause-effect relationship between the exposure and outcome.

• false 

In a hypothetical observational study, investigators wanted to assess the risk of birth defects from exposure to drug A. They recruited mothers of newborns from several hospitals and asked them to complete a questionnaire about their medication use during pregnancy, including drug A. Mothers of infants with birth defects were more motivated to recall and report all possible exposures, while mothers of healthy infants often could not remember or left some responses blank. Which of the following threat to validity would be a potential concern in this study?

a) Information bias 

b) selection bias 

c) confounding

a) Information bias 

A group of researchers conducts an observational study to investigate whether a new cholesterol-lowering drug reduces the risk of heart attacks. They recruit participants from a cardiology specialty clinic who are already motivated to take preventive medications and regularly attend follow-up visits. The results show a substantial benefit of the drug in reducing heart attacks. Which of the following threat to validity would be a potential concern in this study?

a) Information bias

b) Selection bias

c) Confounding

b) Selection bias

Investigators conduct an observational cohort study to evaluate whether use of aspirin is associated with a reduced risk of myocardial infarction (MI) in adults. They find that patients taking aspirin have a significantly lower risk of MI compared with those not taking aspirin.

On further review, the investigators realize that patients taking aspirin were also more likely to engage in physical activity.

Which of the following best describes the role of physical activity in this study?

a) It is part of the causal pathway between aspirin use and MI

b) It is a confounder because it is associated with both aspirin use and risk of MI

c) It is a source of selection bias because patients who exercised are more likely to be included in the study

d) It is a source of information bias because physical activity was self-reported by patients

b) It is a confounder because it is associated with both aspirin use and risk of MI

Which of the following best describes propensity score matching in an observational study? 

a) Restricting enrollment to patients with certain characteristics to eliminate confounding

b) Randomly assigning patients to treatment groups to ensure balance of baseline characteristics

c) Matching patients in different treatment groups based on their probability of receiving a treatment, given baseline characteristics

d) Stratifying the analysis by levels of a potential confounder (e.g., age groups)

c) Matching patients in different treatment groups based on their probability of receiving a treatment, given baseline characteristics

Investigators conduct a cohort study to evaluate whether beta-blocker use reduces the risk of hospitalization for heart failure. They are concerned that age may act as a confounder, since older patients are both more likely to receive beta-blockers and more likely to be hospitalized for heart failure.

To address this, the investigators divide participants into two groups: those younger than 65 years and those 65 years or older, then analyze the association between beta-blocker use and heart failure hospitalization within each age group separately.

Which method of controlling for confounding is being used?

a) Randomization

b) Restriction

c) Stratification

d) Regression analysis

c) Stratification