Epidemiology for Boards Flashcards
Sort criteria A
Based on meta-analysis or high quality RCT
Sort criteria B
Observational studies
Sort criteria C
Expert opinion or anecdote
Relative Risk - how to calculate
- Calculated on cohort study or RCT
- ratio of incidence: outcome of exposed vs unexposed
= risk exp/risk unexp = EER/CER = [a/(a+c)]/[b/(b+d)]
Absolute risk reduction - how to calculate
- difference between control event rate and experimental event rate
= control event rate - experimental event rate
= CER - EER = [b/(b+d)] - [a/(a+c)]
NNT or NNH - how to calculate
= 1/ARR or 100/%ARR
Cohort study
Follow the unexposed over time to see if they develop disease or not
Relative risk - how to interpret
- A RR of 1.0 indicates probability of exposed and unexposed groups are identical
- 1.33 = 33% more likely to develop the disease if exposed
How do I set up my contingency table to calculate Sensitivity and Specificity?
- X axis is disease
- Y axis is test result
How do I construct a contingency table for RR and ARR
- x axis is exposure
- y axis is disease
Sensitivity - how to calculate
= TP/TP + FN
= A/(A+C)
= looking for positivity in people with disease
SNOUT to rule out, sensitive test with negative result will rule out disease
Specificity - how to calculate
= TN/FP + TN
= D/(B+D)
= probability that test will be negative in patients without a disease
SPIN to rule in, specific test with positive result will confirm disease
Positive predictive value
Proportion of positive results that are true positives
- likelihood that a person with a + test actually has disease
= TP/TP+FP x 100
= a/(a+b) x 100
Negative predictive value
proportion of negative results that are true negatives
- likelihood that a person with a negative test truly does not have disease
= TN/FN+TN
= d/(c+d)
