6. Clin Epi Diagnosis II

Question 1

A new test comes out to test for prediction of COPD. The total patient population was 2000 people. There are 700 patients with the disease. The new test had 800 postive test 675 were true.What is the positive likelihood ratio?

A. 6.7
B.  7.3
C >.1
D. 9.7
E. 8.4

RFA

Answer 1

Ans. D 9.7

.96/(1-.90)=9.7

Sen= 96%= 675/700

Spec= 90%=1,175/1300

Question 2

Prostate cancer screening is done the total patient population was 2,000. 100 people in the population have prostate cancer. 88 people tested positive. 40 were true positives. What is the positive predictive value of this test?

A. 40%
B. 99%
C.97%
D. 45%

RFA

Answer 2

Ans. D 45%
40/(40+48)=45

A. is the Sensitivity 40/(40+60)=40%

B. is the Specificity
1852/(1852+48)=98.8%

C. Negative predictive value
1852/(1852+60)=96.8%

Question 3

What does a negative likelihood ratio of <0.1 mean?

A. Strong evidence to rule in the disease
B. Weak evidence to rule in the disease
C. No change in likelihood of disease
D. Weak evidence to rule out the disease
E. Strong evidence to rule out the disease

SZ

Answer 3

Ans E. <0.1 negative likelihood means there is strong evidence to rule out the disease.

> 10 means there is strong evidence to rule in the disease

1.0 means this is no change in likelihood of the disease

Question 4

On a ROC (Receiver Operating Characteristics) Curve sensitivity is plotted on the y-axis and this is plotted on the X-axis?

A. Accuracy
B. Positive Likelihood Ratio
C. 1-Specificity
D. Specificity
E. Negative Likelihood Ratio

SZ

Answer 4

Ans C. (1-specificity)

(A) Accuracy is the area under the curve
(B) Pos Likelihood Ratio are tangential lines to curves

Question 5

A student comes into the SGU clinic and tells you about their wild night at brewery. After hearing the story you think its best to give them a STD test. They ask you what are the chances the test says they have a disease. Which formula gives you the answer for them?

A) Tp/ Tp+Fn
B) Tn/ Tn+Fp
C) Tp/ Tp+Fp
D) Tn/ Tn+Fn
E) Tp + Tn / Population
F) Sensitivity/ 1-Specicifity 

MM

Answer 5

Ans A

If they ask what are the chances the test will say they have a disease you give them the sensitivity

Question 6

You are a school nurse and the entire school is filled with a bunch of chubbos. You think its best to screen for diabetes. What level of prevention are you performing?

A) Primary
B) Secondary
C) Tertiary
D) Quaternary

mm

Answer 6

Ans B

D - this is made up

Question 7

In an ideal scenario, you would not have a trade-off between specificity and sensitivity. How would that be possible? (MG)

A. The disease and normal population curves do not overlap
B. The disease and normal population curves overlap perfectly
C. The peak of the disease population curve is significantly higher than that of the normal population
D. The peak of the normal population curve is significantly higher than that of the disease population

Answer 7

A. The disease and normal population curves do not overlap

This completely prevents false negatives and false positives. This is not realistic, but they could give us graphs with separate peaks since one was in the notes.

Question 8

When studying receiver operating characteristics, how can you tell which data set has the greatest accuracy? (MG)

A. The set where the tangent is the steepest
B. The set with the highest X-axis value
C. The set with the highest Y-axis value
D. The set where the tangent is the least steep

Answer 8

C. The set with the highest Y-axis value

ROC curves always start at 0, so if a dataset has the highest point on the Y-axis, it will level out there and you will get the greatest area under the curve, which is the greatest accuracy