Diagnostic Test Accuracy Flashcards
What are DTAs?
Studies used to tell us about the accuracy of medical tests
When is a test useful?
When it changes the ability to predict whether a patient has a condition or not
What effect should a test result have on diagnosis?
It should alter the probability of diagnosing the condition
Why can’t one assume that the evaluated test accuracy is the same in different contexts, even if the same diagnostic test is used?
A change of context results in a change of accuracy and change of probability of outcome of the diagnostic test
Why should diagnostic tests be evaluated?
To avoid adoption of non-useful, non-beneficial or harmful diagnostic tests
To choose the most cost-effective of the useful diagnostic tests
What are the two criteria that diagnostic tests have to meet to be implemented?
To be proven as safe
To be proven to measure what they say they measure
What is important factor is not required to be proven when a diagnostic test is implemented?
The usefulness of the test in diagnosis
What evidence is not required when a diagnostic test is implemented?
The overall benefit of the diagnostic test
How the diagnostic test is best used
What are some consequences of using inappropriate diagnostics?
Anxiety, discomfort and pain
Unnecessary further investigations
Waste of money
False positive results, which cause unnecessary stress and investigation
False negative results, which causes false reassurance and allows the disease to progress unnoticed
Other than different accuracy, give some positive ways in which diagnostic tests can impact on patient outcomes
Safer, easier to perform, faster results, less invasive,
What is required for a diagnostic test to have potential value?
It must be evaluated as accurate
Give three reasons why DTA studies are useful
They are readily available, unlike RCTs
They do not required prohibitively big sample sizes
Answers are obtained quickly
What are the two dimensions of test accuracy?
Sensitivity and specificity
What does sensitivity measure?
The proportion of individuals with the disease
What does specificity measure?
The proportion of individuals without the disease who have negative results (true negative)
What is the inverse of sensitivity?
False negative rate
What is the false negative rate?
The proportion of individuals with the disease but a negative test
What is the inverse of specificity?
False positive rate
What is the false positive rate?
The proportion of individuals without the disease but with a positive test
What acronym gives the DTA evaluation components?
PITR
What does the P stand for in P?
Participants or population
Give two examples of things which would fall under the participants/population component of DTA evaluation
Presentation
Prior tests
What does the I stand for in PITR?
Index test
What is the index test?
The new diagnostic test which is being evaluated
What must be considered when evaluating the index test?
Whether the test is affected by the conductor or skill of the officer
Whether the test is affected by technology
What does the T stand for in PITR?
Target disorder
What is the Target disorder?
The condition that we want the index test to indentify
What does the R stand for in PITR?
Reference standard
What is a reference standard?
The most accurate method available to detect the target disorder
What statistical check must be done is different reference standards are used for index positive and index negative patients?
It must be assessed whether there is a statistically significant difference between the two reference standard tests
What role does the reference standard play?
It is a control
What is Blinded Cross Classification?
A comparison of the index test and the reference standard test results for detection of a target disorder
What are the three source of bias which can arise in DTA evaluation?
Spectrum bias, review bias, verification bias
What is spectrum bias?
Purposefully excluding ‘difficult to diagnose’ patients
What effect does spectrum bias have on the DTA evaluation?
The index test appears more accurate at diagnosing than it is
How can spectrum bias be avoided?
Clear descriptions of the characteristics of the population
Avoidance of inappropriate exclusions
What is review bias?
When the interpretation of an index test is not independent and blind to the reference standard
What effect does review bias have on the DTA evaluation?
It makes the index test seem more accurate than it is
How can review bias be avoided?
Ensure interpretation of the index test was independent and blinded to interpretation of the reference standard, and vice versa
What is verification bias?
The tendency not to give the reference standard test to patients who were negative on the index test, because the reference
What is verification bias?
The tendency not to give the reference standard test to patients who were negative on the index test, because the reference standard is more invasive
How can verification bias be avoided?
Ensure all patients have both the index test and reference standard test, regardless of result
What is the effect of verification bias on the DTA evaluation?
Over/underestimation of the index test accuracy
What four questions need to be asked to apply the results of a DTA evaluation/what factors contribute to the impact of test use on patient outcomes?
Was there a clear study question?
Is the study population similar to patients in my practice?
Can the index test be applied in the same way in my healthcare setting?
Does the definition of the target condition match the condition I want to identify in my practice?
What are the four steps to critically appraising a DTA study?
Construct a diagnostic 2*2 table
Assess sensitivity and specificity
Assess precision for the estimates of accuracy
Evaluate consequences for patients
What is sensitivity (in a DTA)?
The proportion of true positive results in individuals with the target condition
How is sensitivity calculated?
True positive results/True positive results + False negative results
What does sensitivity indicate in terms of usefulness of the test?
How good a test is at ruling out a condition
What does sensitivity indicate in terms of false results?
The magnitude of false negatives
How can sensitivity be expressed in terms of probability?
The probability that a patient has a positive test, given that they have the disease
A negative test with high sensitivity will be better at doing what?
Ruling out a condition
When is a highly sensitive test less reliable?
When the test result is positive
What does a highly sensitive test have fewer of?
False negatives
If a patient is negative on an index test with high sensitivity, what does this mean statistically?
They are statistically more likely to be a true negative
What is specificity (in a DTA)?
The proportion of true negative results in those without the target condition
How is specificity calculated?
True negative results/True negative results + False positive results
What does specificity indicate in terms of usefulness of a test?
How good a test is at ruling in a condition
What does specificity indicate in terms of false results?
The magnitude of false positives
How can specificity be expressed in terms of probability?
The probability that the result is negative, given that the patient is well
A positive test with high specificity will be better at doing what?
Ruling in a condition
When is a highly specific test less reliable?
When the result is negative
What does a highly specific test have fewer of?
False positives
If a patient is positive on an index test with high specificity, what does this mean statistically?
They are more likely to be a true positive
When is a test considered uninformative?
When specificity or sensitivity are below 50%
What is used to assess the precision of an estimate of accuracy of a diagnostic test?
Confidence intervals
Give 3 consequences of false positives for patients
Unnecessary intervention with associated risks
Unnecessary stress and anxiety
Waste of money and time
Give 3 consequences of false negatives for patients
Delayed diagnosis
Disease progression
Need for more invasive intervention later
