Evaluating clinical tests

Question 1

Positive Predictive Value

Answer 1

PPV = TP/(TP+FP)

Falls when prevalence falls

Question 2

Negative Predictive Value

Answer 2

NPV = TN/(TN+FN)

Increases if prevalence falls

Question 3

Sensitivity (true positive rate)

Answer 3

Sensitivity = TP/(TP+FN)

Unaffected by prevalence

Question 4

Specificity (true negative rate)

Answer 4

Specificity = TN/(FP+TN)

False positive rate = 1 - specificity

Unaffected by prevalence

Question 5

Accuracy

Answer 5

(TP + TN) = (TP + FP + TN + FN)

Question 6

Likelihood ratio for a positive test result (LR+)

Answer 6

LR+ = TP/FP

LR+ = sensitivity / (1 - specificity)

LR+ > 1 - positive test more likely in a person with a disease than without

LR+ <1 - positive test less likely to occur in a person with disease than without

LR+ 10 or more - significantly increase the probability of a disease

Question 7

Likelihood ration for a negative test result (LR-)

Answer 7

LR- = FN/TN

LR- = (1 - sensitivity)/specificity

LR- > 1 = negative test is more likely to occur in people with the disease than without

LR- <1 = negative test less likely to occur in people with the disease than without

LR- 10 or more - significantly increase the probability of disease, below 0.1 - virtually rules out disease

Question 8

Pre test odds

Answer 8

Pre test probability/[1-pre test probability])

Question 9

Post test odds

Answer 9

Pre test odds x likelihood ratio

Question 10

Pre test probability

Answer 10

Point prevalence or period prevalence

Question 11

Post test probability

Answer 11

Post test odds / [1 + post test odds]

Question 12

Absolute risk reduction (ARR) aka attributable risk / ABI

Answer 12

CER - EER

Question 13

Number needed to treat (NNT) / Number needed to harm (NNH)

Answer 13

1 / ARR

Question 14

Relative Risk Reduction (RRR)

Answer 14

ARR / CER

Question 15

Confidence interval

Answer 15

CI = mean +- (ZxSE)

Z = 1.96 for a confidence interval of 95%

SE = standard error

Question 16

Type 1 Error

Answer 16

Alpha error, error of the first kind, false positive

When a true null hypothesis is incorrectly rejected

Question 17

Type 2 Error

Answer 17

Beta error, error of the second kind, false negative

Question 18

Population Attributable Risk

Answer 18

ARR x proportion exposed

Question 19

Range of values

Answer 19

Probabilities: 0-1
Absolute risks: 0-1
Attributable risks: -1-+1
Relative risk: 0 to infinity
Odds: 0 to infinity
Odds ratio: 0 to infinity

Question 20

Relative Benefit Increase (RBI)

Answer 20

RR - 1

Question 21

Patient Expected Event Rate (PEER)

Answer 21

Control event rate in local population in which the intervention is planned

ABI (new) = RBI x PEER

NNT (new) = 1/ABI(new) or NNT (new) = NNT (old) / F value

(will be given PEER value or F value to make calculations)

Question 22

Variance

Answer 22

Sum of squared differences of individual observations from mean / (number of observations -1)

Question 23

SD

Answer 23

SD = square root of variance