Diagnostic test accuracy Flashcards
How are DTA’s carried out
P (presentation and prior testing)
I (index test)
T (target condition)
R (reference standard- the current best available test- everyone who had the index test should have this)
Spectrum bias
Occurs when some people are purposefully excluded from the study
Review bias
Occurs when interpretation of a test is not blind
Verification bias
Occurs when some people don’t get the reference standard
Sensitivity
High sensitivity mean the test is good at ruling out disease (low FN rate so if people test negative then you can be fairly sure it is true, therefore they are likely to not have the condition).
=TP/ (TP+FN)
Specificity
High specificity means the test is good at ruling in disease (low FP rate, so if they test positive you can be fairly sure it is true, therefore likely to have the condition)
= TN/ (TN+FP)
(Sensitivity and FP rate)
Sensitivity tells us nothing about the false positive rate which could potentially be high i.e. the test could incorrectly identify healthy patients as having the disease.
If the false positive rate is high, the test will be poor at ruling in a condition, because although few disease cases may be missed, lots of healthy people will be incorrectly labelled as having the disease.
What is a good test
A good test should change the probability of a condition after it is done
When are DTAs used?
Used in screening, diagnosis and surveillance
Consequences of a FP
Anxiety and worry for the patient
Labelling a patient with a condition they do not have
Will lead to unnecessary further testing and potentially treatment, which could have adverse effects
These effects may lead to delay in patient going for another test- may lead to missed disease in the future
When these consequences are severe, it is important to have a high specificity e.g. Down syndrome testing
Consequences of a FN
False reassurance
May lead to missed disease- which could have potentially very dangerous consequences for the patient
When these consequences are severe, it is important to have high sensitivity e.g. screening blood for HIV
Evaluation bypass
When an un-evaluated test is used in clinical practice, can be due to enthusiasm, convictions but usually is due to commercial pressure
Why are DTAs carried out
Good to evaluate the tests we use in clinical practice
They are much faster and less expensive than RCTs
Aspects of a good test
Fewer errors
Less costly (to time, money and resources)
Easier to interpret
Less invasive
Safer
Higher adherence
How does prevalence affect sensitivity and specificity
As sensitivity and specificity are proportions, prevalence affects them.
At very low prevalence, PPV likely to be very low, NPV high. NPV will remain high until you go over a prevalence of about 90%, after this it will be very small and PPV will be higher. PPV increases up to its maximum (then continues) at about 35%.
