EBD Sensitivity and Specificity
Another name for specificity
True negative
How do you calculate the probability that a patient with a positive test result is actually disease negative?
1-PPV
5 Examples of Diagnostic Tests in Dentistry
1) Caries -visual, tactile, radiography, DIFOTI 2)Pulpal necrosis -electrical, thermal, percussion 3)Soft tissue lesions -Biopsy, dye 4)Periodontitis -Future attachment loss, PSR, BOP 5)Malocclusion -Index, study models, ceph
5 Main Reasons for Diagnostic Tests
1) Establish Diagnosis (symptomatic patients) -ECG -Serology (viral infection) -Biopsy (cancer) -Thermal (tooth/pulpal pain) 2)Screen for Disease -PSA -Bitewing 3)Prognosis -CD4 (HIV) -BOP (perio) 4)Monitor Therapy -PA (root canal) 5)Confirm free of disease
1)What test result types change with disease prevalence? 2)What test results don't change with disease prevalence?
1)PPV and NPV 2)Sensitivity and Specificity
1)How are PPV and NPV inversely related? 2)How are Sensitivity and Specificity inversely related?
1)PPV and NPV vary inversely with one another with changes in prevalence 2) Sensitivity and Specificity will vary inversely with the cut point
How do you calculate the probability that a patient with a negative test result actually has the disease?
1-NPV
When do you use a sensitive test?
A Sensitive test helps rule out disease (when the result is negative). Sensitivity rule out or "Snout"
When do you use a specific test?
A very specific test rules in disease with a high degree of confidence Specificity rule in or "Spin".
What is the cut point? Is it fixed? What is the relationship between sensitivity and specificity?
It is a decision point that a clinician sets between the overlap of sensitivity and specificity. The cut point is set to help the clinician make diagnoses and treatment decisions based on test results. The cut point is arbitrary and may be changed. Sensitivity and specificity are inversely associated to one another and vary with the cut point.
Predictive value for a negative result NPV Definition and formula
NPV - the probability that a negative (normal) test result is correct PV-= true negatives/(true negatives +false negatives)
Is diagnosis a perfect process?
No, it's an imperfect process all tests have some inherent inaccuracies
Predictive value for a positive result (PV+): Definition and formula
PPV - the probability that a positive (abnormal) test result is correct PV+= true positive/(true positive + false positive)
Types of Diagnostic Tests Include
Physical Exam Laboratory tests Radiography
What does an ROC Curve show?
Plots the true positive rate against the true negative rate The true positive rate is also known as the sensitivity index or d-prime
Disease Prevalence
Ratio showing the number of people that have the disease within a total population
Reduction of Diagnostic Information
Scale or Index Value Eamples: BP, Overjet, Pocket depth, Pain duration, BMI leads to... Decision Cut Point which lets you decide from 4 options.... -Treat -Don't treat -Watch -More tests
What type of test would you use for an aggressive disease where the consequences of missed diagnosis are high? Ex Cancer
Sensitive Test Think SN-out (snout) - a sensitive test will accurately rule out people who don't have the disease
What is the relationship between sensitivity and specificity for a given diagnostic test if the cut point is varied?
Sensitivity and Specificity for a given diagnostic test tend to vary inversely when the cut point if varied
Sensitivity Formula
Sensitivity= true positives/(true positive + false negative)
What type of test would you use to confirm a diagnosis that you're leaning towards as the result of other tests?
Specific Test Think SP-in (spin) - a specific test helps you rule in a patient for a disease you think they have
Specificity Formula
Specificity=true negatives/(true negative + false positives)
Test accuracy
The combination of sensitivity and specificity (Sensitivity+Specificity)/2 = Accuracy
What is the gold standard?
The definitive technique Can be: -expensive, elaborate, or difficult to perform Thus we look for: -faster, cheaper and better ways to diagnose Note: sensitivity and specificity reference the gold standard, in the "real world" clinical setting we don't have this standard, must judge how well the newest tests will get the "right" answer for a patient
What does the optimal cut point on a ROC curve represent?
The highest True Positives with the lowest False Positives
Sensitivity definition
The number of people with disease who have a positive test result. (True positive rate) -Relates Gold Standard to New Test -A sensitive test rarely misses people with disease *- Sensitive tests should be selected when there is a serious penalty for missing disease (i.e. cancer diagnosis)
Specificity definition
The number of people without disease who test negative (true negative rate)? A specific test will rarely misclassify people without the disease as diseased Specific tests are used to "rule in" a diagnosis that has been suggested by other tests
What is decision rule or cut point? (What three things does it let you decide?)
The test value above which you will treat The test value below which you will not treat The test value between which you will do more tests
What is d-prime?
The true positive rate is also known as the sensitivity index or d-prime in an ROC curve
What do ROC curves help us visualize? What does the area under the curve tell you about a test?
They relate the changes in Sn and Sp visually at different cut points Provides overall utility of tests Suggests "optimal" cut point The larger the area under a curve the better the test
Another name for sensitivity
True positive
Diagnostic Tests: Valid and Reliable -Define these terms
Valid = True Reliability = Repeatable Can't be valid unless reliable
Accuracy definition and formula
the proportion of true results among the total number of cases studied. A combination of sensitivity and specificity Accuracy= True positives + True Negatives/ (all positive and negative test results)
