Cohort and Cross Sectional Studies Flashcards
<p>How are participants chosen in a cohort study?</p>
<p>-two groups are chosen based on their exposure status
- follow these groups over a period of time to see if they develop the outcome
- measure exposure before disease has happened
- at baseline no one has experienced outcome </p>
<p>How does cohort study design impact temporal relationship?</p>
<p>-measure exposure before we know what happens to person at end of the study
-can make temporal judgment that exposure preceded the outcome (unlike case control study)</p>
<p>When are cohort studies useful?</p>
<p>-when time between exposure and disease is relatively short (look at chronic disease like diabetes or heart disease)
- low risk that people who enroll will leave the study (if we enroll lots of people but we can't follow most for at least a year you don't get much info)
- when adequate funding to continue study is available</p>
<p>What is a retrospective cohort study?</p>
<p>-all of the data have already been collected
- look back in the past to recreate what exposure histories may be
- 2 groups
- look for exposure in past then look for disease after baseline measure was taken </p>
<p>When do we use retrospective cohort studies?</p>
<p>-no losses to follow up
- temporality is still preserved (info on exposure is collected before anyone becomes diseased)</p>
- when extremely long follow up times are required to assess the relation between exposure and outcome
<p>Compare case control, prospective and retrospective cohort studies</p>
<p>case control: researches select cases and controls based on the presence or absence of the outcome respectively and look back in time for disproportionate exposures
prospective studies: follow groups defined by different exposures and observe the incidence of the outcomes
retrospective: researcher starts the study at the time follow up has already been completed </p>
<p>Where to look for data for retrospective cohort studies?</p>
<p>-databases, drug utilization databases, hospital records, death certificates
-problems: no control over completion or accuracy of records and historical effect (advances in medicine may render some data obsolete)</p>
<p>What is proper conduct with the use of a database?</p>
<p>-form hypothesis, then look for variables in dataset and do analyses
-compared to improper conduct of randomly checking for associations between variables and write paper based on statistically significant findings (data mining, data dredging)</p>
<p>How do we select people to be in cohort study?</p>
<p>-select people who are likely to be at risk of developing the outcome and are reasonable to generalize our results to
- group of people that is as representative of the people we want to study as we can</p>
- from a well defined population
<p>What is random sampling?</p>
<p>-each person who is a member of the population has an equal probability of being selected to be a part of the study
- increases generalizability
- ensure no bias in the way we are selecting people so we get a good representation of population we want to study</p>
<p>What is selection bias?</p>
<p>-systematic difference in the way people stay in or drop out of the study or the way people are selected to stay in or leave the study
-when we lose people to follow up, we have to make sure it didn't have anything to do with their exposures they have (if they left because of that it would be significant to the study)</p>
<p>Importance of having equal follow up in both groups</p>
<p>-if we follow up with one group more than the other, we will have more data
- information bias (observational/misclassification bias): different quality of information on exposed as we do unexposed
- participants who have high levels of exposure may give you better information than people who don't so it's important to keep similar tabs on both groups to avoid getting more info from one group to another</p>
<p>What is ascertainment bias?</p>
<p>-assessor of disease status knows subjects' exposure staus</p>
-form of information bias
<p>What is analytic bias?</p>
<p>-people that are responsible for analyzing data
- they usually have no contact with participants and usually blind them to case and control status</p>
- beliefs of investigators influence data analysis and interpretation
<p>What are effects of biases?</p>
<p>-make associations appear weaker or stronger than they really are
-make associations not applicable to the population from which participants were drawn </p>
What is relative risk?
- aka risk ratio
- association between exposure and outcome in cohort study
- (risk in exposed)/(risk in unexposed)
- if relative risk
