Diagnostic Testing & Likelihood Ratios: how we should be ordering tests!
- 1. Dr. Nicolás Pineda @nfpineda Associate Director MUE – USS @MUE_14 Diagnostic Tests & LR
- 2. The bottom line There is no 100%, there is no 0%... Get used to it!!! We need to establish a PRETEST probability of disease LR’s: consider sensitivity, specificity & prevalence of disease Utility of diagnostic test is not only determined by LR number
- 4. Patients do not have disease, only probability of disease Diagnostic tests are merely revision of probabilities Test interpretation should precede test ordering If you are not going to change managemente with the result of a test, you shouldn’t order it
- 8. There is no 0% There is no 100%
- 10. Depends on disease Institutional variability Personal variability
- 12. Need to think… system 2 Depends on Gestalt Depends on patient
- 18. Characteristics Sensitivity Capability of detecting the disease among patients with the disease TP TP+ FN
- 19. Characteristics Specificity Capability of detecting the absence of disease TN TN+ FP
- 20. 2318 407 459 4952 The truth Test Result + - Sick Healthy Lean WL et al. Rapid diagnostic tests for group A streptococcal pharyngitis: a meta-analysis Pediatrics, 2014 Oct;134(4):771-81 S: 2318/(459+23 18) 83% Sp: 4952/(407+49 52) 92%
- 21. Characteristics Predictive values (+ & -) Probability that a positive test is due to presence of disease (TP) Probability that a negative test is due to the absence of disease (TN)
- 22. The truth 2318 407 459 4952 Test Result + - Sick Healthy Lean WL et al. Rapid diagnostic tests for group A streptococcal pharyngitis: a meta-analysis Pediatrics, 2014 Oct;134(4):771-81 S: 2318/(459+23 18) 83% Sp: 4952/(407+49 52) 92% PPV 2318/(2318+4 07) NPV 4952/(459+49 52) 85% 92%
- 23. Prevalence Low prevalence - test (+): TP or FP?
- 24. How do we take in consideration sensitivity, specificity & prevalence of disease?
- 25. L.R.
- 26. L.R. Utility of a diagnostic test How likely is that the patient has the disease Ratio of probabilities
- 27. L.R. + Probability that patient test (+) because he has the disease DIVIDED by the probability that a patient without the disease tests (+) LR+ = sensitivity/ 1 - specificity Prob True(+) / prob False(+)
- 28. L.R. + The bigger the number, better LR+ LR+ = sensitivity/ 1 - specificity Prob True(+) / prob False(+)
- 29. L.R. - Probability that patient with the disease tests (-) DIVIDED by the probability that a patient without the disease tests (-) LR- = 1 - sensitivity/ specificity Prob False (-) / Prob True(-)
- 30. L.R. - The closer to 0, better LR- LR- = 1 - sensitivity/ specificity Prob False (-) / Prob True(-)
- 32. The truth 2318 407 459 4952 Test Result + - Sick Healthy Lean WL et al. Rapid diagnostic tests for group A streptococcal pharyngitis: a meta-analysis Pediatrics, 2014 Oct;134(4):771-81 S: 2318/(459+23 18) 83% Sp: 4952/(407+49 52) 92% LR+: 10.3 LR-: 0.2
- 34. How do we use a LR? LR+: 10.3 LR-: 0.2
- 37. The bottom line There is no 100%, there is no 0%... Get used to it!!! We need to establish a PRETEST probability of disease LR’s: consider sensitivity, specificity & prevalence of disease Utility of diagnostic test is not only determined by LR number