Epidemiology

Question 1

What do chi-square tests analyze?

Answer 1

Compare proportions of categorized outcomes –> 2x2 table

Question 2

ANOVA is used to compare what?

Answer 2

3+ variables

Question 3

Meta-analysis is used to do what?

Answer 3

Pool data from several studies to do an analysis having a relatively big statistical power

Question 4

What is hazards ratio?

Answer 4

Ratio of event occurring in the treatment group vs. non-treatment (control) group.

> 1 = event more likely to happen with TREATMENT group

Question 5

What is PPV?

How do you calculate it?

Answer 5

Proportion of subject with (+) test and ACTUALLY have the disease

TP/(TP+FP)

Question 6

What is NPV?

How do you calculate it?

Answer 6

Proportion of subjects with a (-) test that DO NOT have the disease

TN/(TN+FN)

Question 7

What does correlation coefficient (r) represent?

Answer 7

Strength and direction (positive, negative) of linear association between 2 variables

Closer to either end (-1, +1) = stronger the relationship

E.g. –> r = -0.25
As one variable increases, the other decreases

Question 8

What helps control for cofounders during the design stage of a study?

Answer 8

Randomization

Question 9

What is the p-value?

When is it significant?

Answer 9

Probability of observing a given result by chance, assuming the null hypothesis is true

If the association b/w exposure and outcome is statistically significant (e.g. NOT chance alone)

Question 10

What is relative risk (RR)?

Answer 10

(Risk of an outcome in the exposed group) / (risk of outcome in unexposed group)

RR = 1 (NO association b/w exposure and outcome)

RR > 1 (exposure associated with INCREASED risk of disease)

RR

Question 11

How are confidence interval and p-value related?

Answer 11

CI must NOT contain the null (expected) value (or RR)

When 95% CI does NOT contain the null value (RR), this gives a p-value if it does, the study is NOT statistically significant

Question 12

What is Sensitivity?

Answer 12

of patients correctly testing positive for disease / total # of patients with disease (TP / [TP + FN])

* More sensitive = more likely to be TN (helps rule OUT)
* SnOUT

*Good screening tests –> help minimize FNs

Question 13

What is Specificity?

Answer 13

of patients correctly testing negative for a disease / total # of patients without disease (TN / [TN + FP])

* More specific = (+) tests are more likely to be TP
* SpIN

*Confirmatory tests –> minimize FPs

Question 14

How do case control studies differ from retrospective cohort studies?

Answer 14

Case control = determine outcome first, then look for associated risk factors

Retrospective cohort = ascertain risk factor exposure, then determine outcome