1 / 30

Dichotomous Tests

Dichotomous Tests. Thomas B. Newman, MD, MPH September 27, 2012 Thanks to Josh Galanter and Michael Shlipak. Overview. Clarifications, chapter 1, chapter 2 material Definitions: sensitivity, specificity, prior and posterior probability, predictive value, accuracy 2 x 2 table method

rangle
Download Presentation

Dichotomous Tests

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Dichotomous Tests Thomas B. Newman, MD, MPH September 27, 2012 Thanks to Josh Galanter and Michael Shlipak

  2. Overview • Clarifications, chapter 1, chapter 2 material • Definitions: sensitivity, specificity, prior and posterior probability, predictive value, accuracy • 2 x 2 table method • Likelihood ratios - WOWO • Probability and odds • FP/FN confusion • Test/treat thresholds

  3. Clarifications • EBD errata on book website • SLUBI= Self limited undiagnosed benign illness – not a term I use with parents

  4. Definitions: Sensitivity and Specificity Sensitivity = A/ (A+C) Specificity = D/ (B+D) P.I.D. = Positive in Disease N.I.H.= Negative in Health

  5. Accuracy = (A + D)/(A + B + C + D) = (A+D)/N Accuracy demonstration: screening for brain tumors Definitions: Positive and Negative Predictive value, Accuracy PPV=A/(A+B) NPV=D/(C+D)

  6. Definitions: Pretest (prior) and post-test (posterior) probability Pretest probability = (A+C)/(A+B+C+D) Posttest probability = A/(A+B) or C/(C+D) ONLY IF SAMPLING IS “CROSS-SECTIONAL”!

  7. Subjects are sampled randomly or consecutively, so that the proportion with disease (pretest probability, prevalence) is clinically meaningful “Cross-sectional” sampling PPV=A/(A+B) NPV=D/(C+D)

  8. Subjects with and without disease are sampled separately Proportion with disease is determined by investigator “Case-control” sampling PPV=A/(A+B) NPV=D/(C+D)

  9. Prevalence vs Pretest probability • Pretest probability is the more general term • For screening tests, pretest probability = prevalence • For diagnostic tests, pretest probability incorporates history and physical exam items

  10. Post-test probability vs. Predictive value • Posttest probability after a + test is the same as positive predictive value • Posttest probability after a – test is 1– negative predictive value

  11. 2  2 Table Method • Research vignette “Tom, you need to call this mother. She’s really upset.”

  12. Choroid Plexus Cyst

  13. Pretest probability 0.0003 Sensitivity 33% Specificity 98.5% Fill in table

  14. Likelihood Ratios

  15. Likelihood ratios • A ratio of likelihoods:P(Result|Disease)P(Result|No Disease) • WOWO = With Over WithOut • Pretest odds x LR = Posttest odds • (Prior odds x LR = Posterior odds)

  16. Negative test Positive test Reasurance Order a Test Treatment What Tests Do • Their results change the probability of disease HIV- HIV+ 0% 100% • A good test moves us across action thresholds.

  17. Likelihood of Disease Depends on 2 Things • Where you started from (low, medium, high risk) • Length and direction of the “arrow” • Basic paradigm: • What we thought before  test result  what we think now

  18. Likelihood ratio Effect of test result Very small (0.01) Greatly decreases P(disease) Less than 1 (0.5) Decreases P(disease) One No effect on P(disease) More than 1 (2) Increases P(disease) Very big (100) Greatly increases P(disease)

  19. Likelihood Ratios • Advantages • Calculation of post-test probability easier (especially when disease is rare) • Capture information for multi-level and continuous tests (next week) • Disadvantages • If either pretest or posttest probability is high (~> 10%) you need to use odds (or a slide rule or calculator)

  20. Switch to board • LR for the choroid plexus cyst example • Dichotomous test def of LR • Probability and odds

  21. Can Use Slide Rule

  22. False-negative confusion Sensitivity of rapid strep test is 85% Therefore, false negative rate is 15% 15% is too high, so always culture to confirm negative rapid strep tests

  23. What’s wrong? 2 definitions of “false negative rate” Def #1: 1-sensitivity = FN/(TP+FN). This one is easier because it’s (assumed to be) constant. Def #2: 1 - negative predictive value = FN/(FN+TN). This one is harder because it depends on prior probability, but it is the one that should determine clinical decisions. Strep No Strep Total Rapid Test + TP FP TP+FP Rapid Test - FN TN TN+FN TP+FN FP+TN

  24. If prior probability of strep = 20% and specificity is 98% False negative rate (def #2) = 15/407 = 3.7% NNC (number needed to culture) = 1/.037 = 27 to identify 1 false negative rapid test. (Pre-test probability of 20%) At some prior probability of strep, culture after negative quick test is not indicated.

  25. Sensitivity 85% Specificity 98% Prior probability = 20% Rapid test is NEGATIVE LR = Try it with slide rule!

  26. Similar examples: Sensitivity of UA for UTI is only 80%, therefore always culture after a negative UA Sensitivity of CT scan for subarachnoid hemorrhage is only 90%, therefore always do LP after a negative CT False positive confusion is similar: 1-specificity vs. 1-positive predictive value

  27. No test Test Treat Test/Treat Thresholds

  28. “X-Graph”

  29. New “X-Graph”

  30. Questions?

More Related