Study Guides (248,238)
United States (123,300)
Manderski (2)
Quiz

10:832:335 Quiz: Week 10 - Quiz 2 Review Premium

7 Pages
72 Views

School
Department
Public Health
Course
10:832:335
Professor
Manderski
Semester
Spring

Description
Week 10 Quiz 2 Review Slides Measures of Association • Relative Risk • Interpreting RR • RR = 1 – There is no association between exposure and outcome • RR = #.# – The risk for outcome among those exposed is #.# times the risk for outcome among those not exposed – Examples – RR = 3.5, The risk for outcome among those exposed is 3.5 times the risk for outcome among those not exposed – RR = 0.3, The risk for outcome among those exposed is 0.3 times (or 70% lower than) the risk for outcome among those not exposed – Measures of Association Odds Ratio • Interpreting OR • OR = 1 – There is no association between exposure and outcome • OR = #.# – The odds of outcome among those exposed are #.# times the odds of outcome among those not exposed – The odds of exposure among cases are #.# times the odds of exposure among controls – Examples – OR = 3.5, The odds of outcome among those exposed are 3.5 times the odds of outcome among those not exposed – OR = 0.3, The odds of exposure among cases are 0.3 times (or 70% lower than) the odds of exposure among controls – Measures of Association When is OR ≈ RR? • Controls are representative of the study base (population where the cases came from) • Cases are representative of all cases • The frequency of disease in the population is small (<10%) • When should we calculate OR? – When outcome is rare – When incidence can’t be determined • Case-control studies • Measures of Association • Study Designs • Features of Cohort Studies • Exposure is measured at baseline – Subjects are classified according to exposure status • Subjects are followed over time – Active (phone calls, interviews, clinical exams) – Passive (link to vital statistics records) • Incidence of outcome is measured and compared by exposure status – Relative Risk – Attributable Risk – Best for rare exposures,* common outcomes *Exposure-based selection only • Sampling and Cohort Formation Options • Population-based – The cohort includes either an entire population or a representative sample of the population – Exposure is then determined, subjects classified according to exposure • Exposure-based – Subjects selected because they are known to have the exposure • For example, occupational groups – Cohort is then compared to a comparison group • Another occupational group • General population • Biases in Cohort Studies • Features of Case-Control Studies • Outcome is measured at baseline – People with the outcome (cases) are identified – People without the outcome are selected as a comparison (controls) • Retrospective estimation of exposure – Questionnaire – Interview – Medical Records • Exposure is compared by disease status – Odds Ratio – Best for rare outcomes, common exposures • Case-Control Subject Selection • Cases – Census of all cases (registry) – Representative sample of all cases – Should be only incident cases – Controls – Should represent the population experience from which the cases arose – Sources • Representative sample of study base • Hospital or registry controls • Special groups – Biases in Case-Control Studies • Features of Intervention Studies • Exposure is ASSIGNED by the researchers • Often feature randomization – Eliminates “conscious bias” – Averages out “unconscious bias” and unknown/unmeasured confounders – Often blinded – Reduce bias (observer bias, placebo effect) – Often have strict inclusion criteria – Improves internal validity of the hypothesis test – Reduces external validity (generalizability) • Biases in Intervention Studies • Validity in Intervention Trials • Implications of Error • Effect of Sample Size • Random Error = Poor Precision • Occurs when the factor being measured is not measured sharply • Implications of Selection Bias • Cohort Studies – If unexposed group has LOWER risk than the “counterfactual” • Overestimate effect – If unexposed group has HIGHER risk than the “counterfactual” • Underestimate effect (Health Worker Bias) – Case-control Studies – If controls have LOWER exposure prevalence than study base • Overestimate effect – If controls have
More Less

Related notes for 10:832:335
Me

OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Join to view

OR

By registering, I agree to the Terms and Privacy Policies
Just a few more details

So we can recommend you notes for your school.