BioStats … Ew

Question 1

Type 1&2 error

Answer 1

Hyp True/Hyp False
Reject Hyp. Type I Good
Fail to Reject. Good Type 2

Type 1: reject hypothesis even though it’s true (worse)

Type 2: fail to reject even though it’s false

Question 2

Confidence interval & odds ratio

Answer 2

If a 95% CI contains 1 for a risk ratio or odds ratio, then is is NOT significant

Question 3

Number needed to treat
Yes. No.
Treatment. A. B.
Control C. D.

Answer 3

NNT = 1/absolute risk reduction

ARR = C/(C+D) - A/(A+B). Decrease in risk of an outcome associated with an intervention

Question 4

Sensitivity and specificity

Answer 4

Gold standard
New test Yes. No.
Positive A. B.
Negative C. D.

Sensitivity A/(A+C)
proportion of those with the disease who test positive
Specificity D/(B+D)
proportion of those without the disease who have a neg test

Negative predictive value D/(C+D) proportion of those who test negative who not have the disease

Pos pred value A/(A+B)

Question 5

Cross-Sectional study

Answer 5

All measures obtained on a single occasion
-can measure prevalence not incidence

Best design to evaluate a diagnostic test

Question 6

Case Control Study

Answer 6

Identify those with and without a specific outcome
- look back to see potential predictors

Best for identifying risk factors for a rare outcome

Question 7

Cohort study

Answer 7

Identify those with potential predictor
- look forward to see how many developed the outcome
- prospective or retrospective

Question 8

Appropriate tests matched with variable type

Answer 8

Chi-square = categorical data comparison

Student T-test = normally distributed data, comparison between 2 groups

Wilcoxon-Mann-Whitney = not normally distributed, comparison between 2 groups

ANOVA = comparison of 3+ groups
Kruskal-Wallis = comparison of 3+ groups

Question 9

Likelihood ratio positive

Answer 9

LR+ = sensitivity/(1-specificity)

Question 10

Likelihood ratio negative

Answer 10

LE- = (1-sensitivity)/specificity