Week 10 Flashcards
Diagnosis:
determining the nature of a disorder by considering the patients signs and symptoms, medical background, etc.
Differential diagnosis:
Diagnosis of a condition whole signs and/or symptoms are shared by various other conditions. For example, abdominal pain may be due to multiple different disorders, which need to be ruled out before finding correct diagnosis.
There is not always a one-to-one link between a sign or symptom and underlying disorder…..
- One symptoms could indicate multiple disorders
- One disorder could have multiple symptoms
Reference test of known high accuracy—>
Gold standard, make sure its valid
Diagnostic accuracy study design involving two tests:
Reference and index test
Index test - evaluated by the diagnostic accuracy study —>
if index test results agree with the reference test results = diagnostic accuracy of index test
In the reference test results: Case = _____ and control = _____
Case = person with health condition Control = person without health condition
In the index test results: positive = _____ and negative = _____
Positive = the person classified as a case Negative = the person classified as a control
Case, control, positive, negative… Which ones line up?
Case and positive, control and negative
Diagnostic study design: Level ____ - Diagnostic accuracy study with blinded, independent testing with random or consecutive patients; “consecutive cohort”
2
Diagnostic study design: Level _____ - Diagnostic accuracy with non-consecutive patients
III-1
Level ___ - Diagnostic case-control study - cases identified from reference test then compared on index test
III-2
- Overestimates accuracy (biased) because case status is already known
Level IV diagnostic yield study, has no ……?
Reference list
Diagnostic accuracy studies tend to be cross-sectional
- about “___ and ___”, not longitudinal or tracking
here and now
Diagnostic tests SHOULD distinguish people who have the condition from others who don’t have it —- that’s ____ discrimination
valid
If an index test is valid then:
- All of its positive results will be cases - people who really do have the health condition
- All of its negative results will be controls - people who really don’t have the health condition
Biases in diagnostic studies–> incorporation bias…?
When reference tests include the index tests, an obvious design flaw because any measures of diagnostic accuracy will include the agreement of the index test with itself, a measure of reliability rather than validity
Biases in diagnostic studies–> verification bias…?
Can happen because of sampling problems. Where sensitivity is over estimated, means that more people are testing positive than should test positive. Or underestimated, where fewer controls are testing negative than should test negative. All together—> too many positive results from the index test because of the bias
Continuous , ordered data measured on a scale examples? (must be categorised into cases and non-cases)
height, weight, pain, degrees of flexion, grip strength, etc
Categorical, unordered [nominally scaled] data …
- Patients categorised naturally into…
- Cases and controls on reference list
- Positive and negatives on index test
Linking reference and index test categories
- Statistical measures of association
- Tests of diagnostic accuracy
Sensitivity (Sn):
- Index test ability to detect people with the condition
- Percentage of cases with a positive result
Specificity (Sp):
- Index test to ability to identify people without condition
- Percentage of controls with negative result
Sensitivity and specificity usually ___ for the same test
differ
