Study Design in Pharmacoepidemiology

							Study Design in
Pharmacoepidemiology
   Study design               result
    of study
   Example 20 subject vs 40 subjects
Type of association error in study

   None
   Artifactual (chance and bias)
   Indirect (confounded)
   Causal (direct or true)
Bias

   Selection
   Information (recall and interviewer)
How to reduce the confounding?

   Random allocation
   Subject selection (exclusion and
    matching)
   Analysis (stratification and
    mathematic model)