DTA

Question 1

What type of study do you use for evaluating diagnosis test accuracy?

Answer 1

Special cross-sectional study

Question 2

What is meant by the evaluation bypass?

Answer 2

unevaluated procedure bypasses evaluation and becomes taken up by healthcare services

Question 3

What can cause an evaluation bypass?

Answer 3

enthusiasm, convictions, commercial pressures

Question 4

What makes a test useful?

Answer 4

If it changes our ability to predict whether a person has a condition or not. (guides treatment options)

Question 5

What should a test result do?

Answer 5

Alter the probability of a condition being the suspected condition

Question 6

Define test accuracy

Answer 6

The usefulness of a test result in diagnosing a patient. How good it the test at spotting who does/doesn’t have the disease?

Question 7

Define diagnostic yield

Answer 7

How much information a diagnostic test can give - informs diagnosis

Question 8

Define therapeutic yield

Answer 8

The amount it can impact the treatment decision

Question 9

What do you need to consider beyond accuracy?

Answer 9

Harms - e.g. radiation, anxiety
Also is it reproducible? Would someone else doing the test get the same result?
More accurate may not improve outcomes - could be too late to intervene, may not change therapeutic input.

Question 10

Tests for screening are…

Answer 10

Used for early identification of disease in asymptomatic patients (e.g. mammography)

Question 11

Tests for monitoring are..

Answer 11

Used to monitor treatment response (e.g. blood glucose)

Question 12

What do you compare for test accuracy?

Answer 12

Index Test and Reference Standard

Question 13

What is the index test?

Answer 13

The disease state estimated by the test of interest

Question 14

What is the reference standard?

Answer 14

The best estimate of true disease state (most accurate existing test)

Question 15

What is blinded-cross classification and when is it used?

Answer 15

Comparison of the index test and reference standard. Compare the results blinded (without knowing)

Question 16

What are the components of a test accuracy question?

Answer 16

Participants, index test, target disorder, reference standard

Question 17

What are the components of participants in a test accuracy question?

Answer 17

1. Presentation - presence, duration and severity of symptoms
2. Prior test - have they been examined or self diagnosis, any bloods before

Question 18

What components of the index test are important in a test accuracy question?

Answer 18

• Could be more than 1
• Could be a new test or unevaluated test
• Assumed to be less accurate than the reference standard
• Availability in primary versus secondary care
• Interpretation by the operator

Question 19

What is the reference standard?

Answer 19

Most accurate and feasible method of detecting a target disorder, ‘gold standard’
Could be multiple tests

Question 20

What are the parts of critical appraisal of a DTA study?

Answer 20

1. Internal validity (bias)
2. What are the results (numerical expression of test accuracy)
3. Can I apply the results to my patient? (generalisability)
4. What would be the impact of using the index tests in my population?

Question 21

Define spectrum bias

Answer 21

Specific groups of patients are inappropriately excluded (e.g. difficult to diagnose patients). Makes the index test appear more accurate (selection bias)

Question 22

Define review bias

Answer 22

Interpretation of the index test not independent to the reference standard. More common with subjective tests. Index test not blinded on interpretation.

Question 23

Define verification bias

Answer 23

Not all participants get the index test and the reference standard. Often seen if the index test is negative

Question 24

Define sensitivity

Answer 24

What percentage of people with the disease are picked up by the index test? Good at ruling people out.

Question 25

Interpret a sensitivity of 0.88 or 88%

Answer 25

88% of people who have the disease will be correctly identified by the index test.
12% of people will have a false negative - told they don’t have the disease when they do

Question 26

Define specificty

Answer 26

What percentage of people who don’t have the disease are correctly identified as not having the disease. For ruling in disease

Question 27

Interpret specificity of 41%

Answer 27

41% of people who don’t have the disease are correctly told they don’t have the disease
59% of people who are given a negative result actually have the disease

Question 28

What is the calculation for sensitivity?

Answer 28

True positive / (true positive + false negative)

Question 29

What is the calculation for specificity?

Answer 29

`True negative / (true negative + false positive)

Question 30

What does a high sensitivity mean?

Answer 30

Lower false negative rate

Question 31

What does high specificity mean?

Answer 31

Lower false positive rate

Question 32

How do you calculate false positives?

Answer 32

False positive / (false positive + true negative)

Question 33

How do you calculate false negative rate?

Answer 33

False negative / (false negative + true positive)

Question 34

What is precision?

Answer 34

How narrow the confidence intervals are (usually 95%)

Question 35

Define confidence intervals

Answer 35

95% of the time the result would fall within this limit.

Remember to consider what the worst would be!

Question 36

Define applicability/generalisability

Answer 36

Applicable to the population? Can the index test be applied in your healthcare setting?

Question 37

How do you know if a result is comparable different?

Answer 37

Confidence intervals don’t overlap