Midterm - research studies on diagnostic procedures

Question 1

The extent to which an examiner can measure multiple times and achieve the same results (You X 5)

Answer 1

Intra-examiner reliability

Question 2

The extent to which multiple examiners can perform same measure on the same person and come up with same results (5 different people)

Answer 2

Inter-examiner reliability

Question 3

How often does the test yield the same result on a diseased person?

Answer 3

Test reliability

Question 4

How often does the test achieve the same outcome when multiple measures on the same patient is achieved

Answer 4

test-retest reliability

Question 5

What is a measure of dichotomous variables’ agreement above chance alone and is used to measure reliability? Used when outcome is binary (yes/no. Positive/negative)

Answer 5

Kappa (K)

Question 6

What is a reliable Kappa statistic?

Answer 6

Depends on what it is associated with. 0.61+ for manipulative therapy. 0.80 for others

Question 7

What state is used when there are categorical (e.g. grade 1, 2, 3) or continuous variables (scale of 0-20)

Answer 7

Intra-class correlation (ICC)

Question 8

What is the ability of a test to correctly identify people who have the target disorder?

Answer 8

Sensitivity

Question 9

What is the ability of a test to correctly identify people who do NOT have the target disorder?

Answer 9

Specificity

Question 10

What kind of table or graph do you calculate sensitivity and specificity?

Answer 10

Contingency table

Question 11

What are true positives?

Answer 11

Those who test positive and have the disease

Question 12

What are true negatives?

Answer 12

Those who test negative and do NOT have the disease

Question 13

What are false positives?

Answer 13

Those who test positive for the disease but do NOT have the disease

Question 14

What are false negatives?

Answer 14

Those who test negative for the disease but DO have the disease

Question 15

We want screening tests to be extremely __________ while we want confirming diagnosis tests too be highly __________

Answer 15

sensitive

specific

Question 16

How do you use +SPin and -SNout

Answer 16

SPecificity to rule something IN with + tests

SeNsitivity to rule something OUT with - tests

Question 17

How do you calculate sensitivity and specificity

Answer 17

SeNsitivity: true pos divided by (true pos + false neg)

SPecificity: true neg divided by (false pos + true neg)

Question 18

What is a derivation of sensitivity and specificity that takes into account how impactful false pos and false neg components of the test’s performance are in being able to “trust” the outcome of a test.

Answer 18

Predictive value

this is a step up from sensitivity and specificity alone

Question 19

What is a positive predictive value

Answer 19

Probability that the disease is present when the test is positive

Question 20

What is a negative predictive value

Answer 20

Probability that the disease is NOT present when the test is negative

Question 21

What indicates the change in odds that a patient has (or does not have) a particular condition based on whether a test is positive or negative

Answer 21

Likelihood ratios

Question 22

Tell me how likelihood ratios work

Answer 22

Ok, your turn is over. My turn.

It starts with a baseline probability that a person has a disease.

That place is called perfectly uncertain and it is based on established rates of that condition in society or that subset of the population.

Then you run a test.

With each new test, you move the probability ticker mark from where it was at baseline to the right or the left.

More specifically you move it up or down the probability scale toward exclusion threshold or the therapy threshold.

Question 23

What are the formulas for likelihood ratios

Answer 23

+LR = sensitivity divided by (1-specificity)

-LR = (1-sensitivity) divided by specificity

Question 24

What does a LR of 1.0 mean

Answer 24

No impact point

everything above that is a +LR and everything below is a -LR

Question 25

What is the 15/30/45 rule in regards to the magnitude of +/- LR

Answer 25

when a +LR = 2.0 you say that: a positive test with +LR = 5.0 increases the post-test probability that the patient has the disease by ↑15%

+5 = ↑30%

+10 = ↑45%