Diagnostic Tests Flashcards
Three Purposes of Diagnostic Tests
1) Focus Examination in a particular area of the body; leads you to the problem
2) Identify potential problems that require referral
3) Assist in classification process and ultimately GUIDE treamtment
Test Threshold (definition)
Probability below which a test will NOT be performed bc the possibility of a certain diagnosis is so unlikely
Treatment Threshold (def)
Probability above which a test will not be performed bc the possibility of a certain diagnosis is SO likely that immediate treatment should be performed
Measurement Reliability (def)
Test produces STABLE RESULTS over time
Measurement Validity (def)
Test CORRECTLY CAPTURES what it is supposed to be testing
Increased validity =
increased chance that the test results are free from bias and error and are believable
Test results are composed of: (2)
“true value” and “error”
-Want to minimize error during testing!
Error in a test may be due to: (4)
- The subject
- The observer
- The instrument
- The environment
Statistics commonly used to evaluate reliability of diagnostic tests: (5)
- Standard Error of Measurement (SEM)
- Pearson’s Product Moment Correlation (r)
- Intraclass Correlation Coefficient (ICC)
- Spearman’s Rho (p)
- Kappa (k)
Two types of validity commonly associated with Daignostic Tests
1) Face Validity
2) Criterion Related Reliability
- May be evaluated through mathematical calculations using a 2x2 CONTINGENCY TABLE
2x2 Contingency Table is the basis for calculating:
Sensitivity and Specificity
Face Validity (def)
Does the test look like it’s measuring what it’s supposed to be measuring?
Criterion-related validity (def)
Relates diagnostic test to a gold standard
Sensitivity (def)
Capability of classifying individuals with a condition of interest (a true positive)
Mathematical Equation for Sensitivity (from Cont. Table)
(True positives)/(True positives+ False Negatives)
aka a/(a+c)
SnNout
- If you have a highly sensitive test and a negative result- RULE OUT disorder!
- meaning True positives are high, so if you get a negative result it’s probably a true negative
Specificity (def)
Capability of correctly classifying individuals WITHOUT the disorder (aka true negatives)
Mathematical Equation for Specificity
(True Negatives)/(False Positives + True Negatives)
Aka d/(b+d)
SpPin
- If you have a highly specific test and a positive result- RULE IN the disorder!
- Meaning true negatives are high so likelihood of false negatives is low
Receiver Operating Curve (def)
Graphic depiction of different test scores in relation to the number of true positive and false positive test results, obtained at each cut point
What does the area under the Receiver Operating Curve Represent?
- The true and false positives for each score obtained on the diagnostic test for each cut point
- Higher true positive rate = bigger area under the curve
Positive Predictive Value (def)
Ability of a diagnostic test to correctly determine the proportion with the diease from all of the patients who tested positive on a diagnostic test
Negative Predictive Value (def)
Ability of a diagnostic test to correctly determine the proportion of patients without the disease from all of the patients who tested negative on the diagnostic test
Positive Likelihood Ratio (LR+)
- Likelihood that a positive test result was obtained in a patient with the disorder as compared to a patient without the disorder
- HIGHER numbers are better!!
Negative Likelihood Ratio (LR-)
- Likelihood that a negative test result was obtained in a patient with the disorder as compared to a patient without the disorder
- LOWER numbers are better!
Best measure of a diagnostic test is:
Likelihood Ratios
Three Advantages to using Likelihood Ratios in regards to diagnostic tests:
1) Ratios can be calculated for all levels of test results
2) Not dependent on prevalence of the condition in the population
3) Can be used with individuals and groups
