Clinical Research & Evidence-Based Medicine Unit Aristotle University of Thessaloniki Μελέτες Διαγνωστικής Ακρίβειας Πασχάλης Πάσχος MD, MSc Γαστρεντερολόγος Μονάδα Κλινικής Έρευνας και Τεκμηριωμένης Ιατρικής Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης
Τι σημαίνει διάγνωση; Our task is not to attain certainty, but rather to reduce the level of diagnostic uncertainty enough to make optimal therapeutic decisions Kassirer, 1989, p.1489
Τι σημαίνει διάγνωση; Διαγνωστικές δοκιμασίες δεν νοούνται μόνο αυτές που παραγγέλνονται στα εργαστήρια αλλά συμπεριλαμβάνουν: Κλινικές πληροφορίες (ιστορικό κλινική εξέταση) Πληροφορίες από απεικονιστικές εξετάσεις Συνδυασμός ευρημάτων που οδηγούν σε διάγνωση Σκοπός: Αύξηση της βεβαιότητας για την παρουσία/ απουσία μιας νόσου Εκτίμηση βαρύτητας νόσου Παρακολούθηση κλινικής πορείας Εκτίμηση πρόγνωσης κινδύνου/σταδίου
Τι είναι φυσιολογικό... Gaussian: ο με σος ο ρος ± 2 τυπικε ς αποκλι σεις. Αυτο πρου ποθε τει κανονικη κατανομη σε ο λες τις εξετα σεις, και συνεπα γεται ο τι ο λες οι μη φυσιολογικε ς τιμε ς συμβαι νουν με την ι δια συχνο τητα. Ποσοστιαίος: εντο ς του ευ ρους, ας που με 5-95%. Πολιτισμικα επιθυμητός: ο ταν το «φυσιολογικο» ει ναι αυτο που προτιμα ται απο την κοινωνι α, ο ρο λος της ιατρικη ς γι νεται ασαφη ς. Παρα γοντας κινδυ νου: που δεν επιφε ρει επιπρο σθετο κι νδυνο της νο σου. Στιγματι ζει/βα ζει ετικε τες σε ασθενει ς ανεξα ρτητα απο το αν μπορου με να παρε μβουμε για να ελαττω σουμε τον κι νδυνο Διαγνωστικός: ευ ρος αποτελεσμα των πε ρα απο το οποι ο η υπο διερευ νηση διαταραχη ει ναι πολυ πιθανη. Θεραπευτικός: ευ ρος τιμω ν πε ρα απο το οποι ο η θεραπει α κα νει περισσο τερο καλο παρα κακο.
Diagnostic test accuracy (DTA) study Provides evidence on how well a test correctly identifies or rules out a target condition. It compares one or more tests of interest (the index test ) to an existing diagnostic test (the Reference standard ), which is the best available method for establishing the presence or absence of the target condition Series of patients Index test Reference standard Compare the results of the index test with the reference standard, blinded
Diagnostic test accuracy (DTA) study
Study Population Reflect the population that will undergo the diagnostic test in clinical practice. Subjects with symptoms commonly associated with the target condition and in whom we are uncertain of the diagnosis should be included Optimal design: cross-sectional study Individuals are selected randomly or consecutively on the presence of predefined symptoms or signs of the target condition. Alternative design: case -control studies or "two-gate" studies disease group and control group are selected with different sets of criteria. potential bias depending how the two groups were selected
Study Population
Study Population
Reference Standard Test, series of tests, or set of procedures that is used to determine the presence or absence of the target condition Gold standards with 100% sensitivity and specificity are rare A standard as close as possible to the theoretical gold standard. The best available, clinically accepted, error-free procedure Applied for all included subjects. Blinding New diagnostic tests superior to prevailing reference standards may be developed. Even biopsies are not infallible, as they rely on subjective assessment from human observers with varying expertise.
Reference Standard
Reference Standard For many diseases, a perfect diagnostic test may not exist or cannot be applied owing to costs or ethical concerns. Dendukuri N. et al BMJ 2018;360:j5779
Bias Bias is systematic errors in design or conduct of the study that result in estimates of test performance which differ from true performance of the test. The Ugly 5.
Spectrum bias study uses only highly selected patients.perhaps those in whom you would really suspect have the diagnosis Ideally, test should be performed on a group of patients in whom it will be applied in the real world clinical setting
Partial verification bias only some patients get the reference standard..probably the ones in whom you really suspect have the disease Series of patients Reference standard Β ΝΟ Reference standard
Differential verification bias the diagnosis of target condition is verified with two or three different reference standards Reference standard Α Reference standard Β Compare the results of the index test with the reference standard, blinded
Incorporation Bias when the index test is incorporated in the reference standard. Reference standard.. includes parts of Index test
Observer bias test is very subjective, or done by person who knows something about the patient or samples Ideally, the reference standard is independent, blind and objective Reference standard Unblinded cross-classification
Standards for Reporting Diagnostic Accuracy (STARD) statement.allows readers to detect the potential for bias in a study (internal validity) and to assess the generalisability and applicability of results (external validity).
Evaluation of Diagnostic Test Accuracy of the test is not enough Value of diagnostic test ultimately lies in effect on patient outcomes. it must be examined as part of a broader management strategy.
Evaluation of Test Role of new Test
Evaluation of Test Role of new Test Replacement more accurate less invasive, easier to do, less risky less uncomfortable for patients quicker to yield results technically less challenging more easily interpreted
Evaluation of Test Role of new Test Triage Before the existing test pathway Selection of patients for further testing (or not) less accurate - other advantages (like simplicity or low cost)
Evaluation of Test Role of new Test Add-on Detect patients not identified by existing test(s) New test limited to subgroup of patients New test more accurate but higher cost, invasive
Choice of the right population Reference standard: best available, clinically accepted, error-free procedure Test accuracy is not a fixed property of a test. The value of a test is not simply measured by its accuracy
Metrics of Diagnostic Test Accuracy Studies
Measures of Diagnostic Accuracy The diagnostic accuracy of a test quantifies the ability of this test to discriminate between certain two conditions, diseased and non-diseased.
Diagnostic test accuracy (DTA) study
Measures of Diagnostic Accuracy True positive (TP) False positive (FP) False negative (FN) True negatives (TN) True positive (TP): Diseased individuals with positive test results. True negatives (TN): Non-diseased individuals with negative test results. False positive (FP): Non-diseased individuals with positive test results. False negative (FN): Diseased individuals with negative test results.
Sensitivity (Sens) The proportion of people who have the disease who have a positive test result. So, a test with 90% sensitivity.means that the test identifies 90 out of 100 people WITH the disease
Sensitivity (Sens)
Specificity (Sp) The proportion of patients who don t have the disease who have a negative test. So, a test with 75% specificity will be NEGATIVE in 75 out of 100 people WITHOUT the disease False positive rate = 1 - specificity So a specificity of 75% means that the new test is wrong (or falsely positive) in 25% of people
Specificity (Sp)
98 2 10 240 99% 96%
Sensitivity - Specificity There are no perfect tests with both 100% sensitivity and specificity. SnNout When a test has a very high Sensitivity, a Negative result effectively rules out the diagnosis High sensitivity means there are very few false negatives so if the test comes back negative it s highly unlikely the person has the disease SpPin When a test has a very high Specificity, a Positive result effectively rules in the diagnosis High specificity means there are very few false positives so if the test comes back positive it s highly likely the person has the disease
Doc, what are the chances of being pregnant? 90% 83% I don t Know
Positive Predictive Value (PPV) What is the likelihood that this patient has the disease when the test result is positive?
Negative Predictive Value (NPV) What is the likelihood that this patient does not have the disease when the test result is negative?
Measures of Diagnostic Accuracy Sensitivity Specificity NOT influenced by the disease prevalence Depend on the spectrum of the disease in the studied group. Predictive values Dependent on the disease prevalence. The same diagnostic test will have varying predictive values in different populations. A higher prevalence will increase the PPV and decrease the NPV.
Predictive Values Your father went to his doctor and was told that his test for a disease was positive. He is really worried, and comes to ask you for help! Tell me what s the chance I have this disease? After doing some reading, you find that for men of his age: The prevalence of the disease is 30% The test has sensitivity of 50% and specificity of 90%
Predictive Values Prevalence of 30%, Sensitivity of 50%, Specificity of 90% Sensitivity: 50% Disease +ve 30 15 22 people test positive 100 of whom 15 have the disease Disease -ve 70 7 False positive rate: 10% So, chance of disease is PPV: 15/22 = 68%
Predictive Values Prevalence of 4%, Sensitivity of 50%, Specificity of 90% Sensitivity: 50% Disease +ve 4 2 11.6 people test positive 100 of whom2 have the disease Disease -ve 96 9.6 False positive rate: 10% So, chance of disease is PPV: 2/11.6 = 17%
Likelihood ratios - LRs LR = Probability of the specific test result in subjects with the disease Probability of the same result in subjects without the disease LR provides a sense of how powerful a test is in influencing our pretest probability of disease. Likelihood ratios combine the sensitivity and specificity of a test
Likelihood ratios - LRs LR > 1 indicates that the test result is associated with the presence of the disease LR < 1 indicates that the test result is associated with the absence of disease. Qualitative Strength LR(+) LR(-) Excellent 10 0.1 Very good 6 0.2 Fair 2 0.5 Useless 1 1
Using Fagan s nomogram Pre-test probability / prevalence: 30% Post-test odds for disease after one test become pretest odds for next test etc. LR+ = 25 Pre-test probability / prevalence: 2%
Confidence interval - CI 95% CI is a range of values around an estimate that have a 95% probability of encompassing the true value of that estimate 95% CI indicates the imprecision of the measure. The width of the CI is inversely related to the size of the sample The wider the confidence interval, the less precise the estimate is.