Epidemiology: Measuring & Describing Disease 3

Question 1

What determines the most appropriate study design

Answer 1

• There are different types of observational study design. The hypothesis and variables
(exposures and outcomes) being collected will determine which design is most
appropriate.

Question 2

What are the usual statistic outputs in observational studies? What is the difference in these outputs

Answer 2

• In observational studies relative risk (RR) and / or an odds ratio (OR) are the usual
statistical outputs. Where there is zero association RR and OR are 1.0. However, OR
increases more rapidly and some observational study designs (retrospective) do not allow for RR. The OR and RR are broadly similar where a disease outcome is rare
(say… less than 10%).

Question 3

What are the types of errors and their probabilities

Answer 3

• Type I errors detect false positives and Type II errors miss true findings / detect false
negatives. Alpha is the probability of making a type I error and beta is the probability
of make a type II error.

Question 4

What is statistical power of a study linked to?

Answer 4

• The statistical power of a study is intrinsically linked to sample size, difference of
interest/clinical significance, type I and type II error. Prior to commencing a study an
adequate sample size for the desired power is calculated but with consideration of the
financial implications. Interpretation of results should be mindful of the study’s power.