Diagnostic test accuracy studies Flashcards
Reasons for carrying out a diagnostic test?
???
- Detection/ exclusion; rule in/ rule out disease
- Make sure not missing something (reassurance)
- Medico-legal financial reasons
- Following protocol
-> Pt ultimately better off having it
What are the consequences of inappropriate testing (for false +ve and false -ve) ?
- Waste of money that could be better used elsewhere
- Anxiety, discomfort, pain, unnecessary further testing and treatment with ass. SE
- Tests make errors, even if they are correctly performed
False +VE = Distress and anxiety to pts / unn. further tests / poss. unn. treatment
False -VE = Wrongly reassure patients / result in disease being missed / may result in harm to others (infectious disease)
Definition of sensitivity
What proportion of those WITH the disease does the test detect?
Definition of specificity
What proportion of those WITHOUT the disease get negative test results?
Definition of false negative
What proportion of those WITH disease does the test miss?
Definition of false positive
What proportion of those WITHOUT disease does the test give positive test results
What are the components of a diagnostic test accuracy question?
- P - Participants
- Presentation: e.g. asymptomatic, severity and duration of symptoms
- Prior tests: before index test (affects validity if there are fewer diffs.)
- Do they reflect the general/ target popn - I - Index test
- Conduct: experience/ skill of operator
- Technology: old vs new; fixed vs mobile
- No of tests
- Assumed to be less accurate than the ref standard
- T - Target disorder: the condition(s) we want the index test to identify
- must be specific - R - Reference standard: the most accurate and feasible method available of detecting the target disorder
- How the outcome is measured
- Often comprises more than one test
–> Outcome = accuracy ; in target disorder and reference standard
What are the 3 different SOURCES OF BIAS in a test accuracy study (critical appraisal of test accuracy study)?
- Spectrum bias
- Review bias
- Verification bias
When might spectrum bias occur?
If ‘difficult to diagnose’ patients are purposefully excluded
- this will make the index test appear more accurate than it will be in practice
When might review bias occur?
- If interpretation of the index test is not independent (and BLIND) of the reference standard
- This has the potential to make the index test appear more accurate than it actually is (as it may influence interpretation)
When might verification bias occur?
- There is a tendency for patients with negative index test results not to get the reference standard
- This will result in OVER or UNDER estimation of the accuracy of the index test
- Other biases tend to overestimate accuracy
What are the limitations of test accuracy studies?
- At risk of bias (major sources of bias originate in methodological deficiencies
What are the advantages of a test accuracy study?
- Useful as an indication of the potential values of a test
- They are readily available in contrast to test treat RCTs
- They do not require prohibitively large sample sizes
- Answers can be obtained quickly
What is test accuracy?
A comparison between two things:
- The disease state estimated by a test of interest -> the index test
- The best estimate of the true disease state -> the ref standard which may be one or more tests - measured by the ref standard
(Explicit recognition that most tests make errors even if correctly performed)
What are the components of a test accuracy study critical appraisal?
- INTERNAL VALIDITY (potential for bias) - any study issues?
- What are the results? - numerical expressions of test accuracy
- Can I apply the results to my pts?? (applicability - consider external validity) - consider cost/ acceptability to both pts and HCP’s in terms of interpretation and technology/ does the target condition def match the condition you want to identify
- What would be the impact of using the index tests in my pts/ popn?
