Week 9 - validity of diagnostic measures Flashcards

1
Q

Purpose of diagnostic Accuracy studies

A
  • determine if ‘new” test (index test) is accurate based on gold standard (reference test)
  • true status = presence/absence of condition of interest
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

design of diagnostic accuracy studies

A
  • cohort studies
  • participants: people at risk for condition of interest
  • procedure: all participants administered index test and gold standard test
  • outcomes of both tests are compared
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Validity in diagnostic design

A
  • researcher taking index measurement should be blinded to result of reference test
  • testers should be evaluated for reliability
  • timing of index and reference test
  • standards for reporting of diagnostic accuracy (STARD): checklist for essential elements of design, analysis, interpretation of diagnostic test studies
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Sensitivity:

A

how correct is test for people with disorder
- True positive/(true positive+False negative)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Specificity

A

how correct is test for people without disorder
- = True negative/(false positive+true negative)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Positive predictive value

A

= probability of having the disorder if test results is positive
- PPV is analogous to clinical situation - looking at the probability the person has disorder- PPV = TP/(TP+FP)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Negative predictive value

A
  • Probability of not having disorders if index test result is negative
    = TN/(TN+FN)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What are PPV and NPV dependent on and define it

A
  • prevalence: proportion of population with given disorder at given point in time
  • values vary in different populations with different disorder prevalence
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Likelihood ratios

A
  • Combine benefits of sensitivity and specificity into one index
  • indicate how much a diagnostic test will raise of lower the pretest probability of disorder
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Pretest probability:

A
  • estimation of likelihood person has disorder based on history and interview
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Sensitivity

A

= TP/(TP+FN)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Specificity

A

TN/(TN+FP)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Positive likelihood ratio

A
  • probability that person with disorder will test positive
    = SENS/ (1-SPec)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Negative likelihood ratio

A
  • probability that person with disorder will test negative
    = (1-Sens)/Spec=FN/TN
How well did you know this?
1
Not at all
2
3
4
5
Perfectly