help teaching about likelihood ratios, sensitivity, positive pred values!

John Brush jebrush at ME.COM
Mon Sep 10 01:06:06 UTC 2018


	Thanks for the nice comments about my book "The Science of the Art of Medicine." The iBook version is now FREE. Please download if you have an iPad, iPhone or Mac. The print version has a modest charge, mostly to cover the production and distribution costs. 

In answer to your question below,
	People get confused with sensitivity and specificity and positive and negative predictive value. The terms are very confusing for both early learners and experienced clinicians. Sensitivity is the true positive rate and specificity is the true negative rate. I think the TPR and TNR terms are better because they do a better job in explaining what they stand for. The mnemonics SnNout and SpPin are helpful in remembering how to think about the value of tests with high sensitivity or high specificity.
	The other problem with these terms is that they are all rates with different numerators and denominators. People can’t keep track of what is in the numerator and denominator. This is why likelihood ratios are handy because likelihood ratios are dimensionless numbers. With LRs, you don’t need to keep track of what is in the numerator and denominator. You can easily use LR’s to calculate post-test probabilities (converting probability to odds, multiplying by LRs, then converting odds back to probability). But also you can use LRs to intuitively calibrate your intuition if you use the anchoring and adjusting heuristic (which we all do every day). I talk about this in my book. 
	Geoff Norman, Johathan Sherbino and I are currently doing a randomized controlled trial to test whether I can actually effectively teach this stuff (LRs) to medical students. We are close to completing the enrollment and should have results soon. We’ll see…
John

John E. Brush, Jr., M.D., FACC
Professor of Medicine
Eastern Virginia Medical School
Sentara Cardiology Specialists
844 Kempsville Road, Suite 204
Norfolk, VA 23502
757-261-0700
Cell: 757-477-1990
jebrush at me.com




On Sep 8, 2018, at 5:11 PM, Tom Benzoni <benzonit at GMAIL.COM> wrote:

John Brush, The Art of the Science of Medicine
ISBN-13: 978-0990961376
Says it better than I can.
tom benzoni

On Sat, Sep 8, 2018 at 12:29 PM Tom Yuen <0000001243181998-dmarc-request at list.improvediagnosis.org <mailto:0000001243181998-dmarc-request at list.improvediagnosis.org>> wrote:
Hello, 
 
Young faculty here just tasked with teaching new family medicine residents diagnostic reasoning. During my inaugural lecture on LR (likelihood ratios)/PPV (positive predictive value)/NPV (negative predictive value) and sens/specificity, a newly minted 2nd year resident posed this comment: 

 "I understand that mathematically and statistically these are all subtly different. But in the real world, when I'm admitting someone for chest pain- all these things are pretty much the same.  A very sensitive test that is negative means that the likelihood of the disease is very low.  Most sensitive tests also are low likelihood rations and very good NPV. It's all pretty much the same in the real world. Why learn how these are different?”

As an example another resident quoted these numbers (I have yet to confirm them, but irregardless her point was):  A negative troponin has a sensitivity of 95% and a neg LR of 0.06 and a NPV of 98% for acute myocardial infarction.  What does it matter which statistic you use, they all "say the same thing”.

Intellectually I was able to dig myself out, at least with NPV since it depends on the prevelence of the disease. But I did not have a more eloquent answer for LR and sensitivity.  And it all rang hollow to me as I tried to put myself in their shoes.

Does anyone have any suggestions, or tips? Or reading? Or are they right that these numbers are tell pretty much the same tale when being used by a on-call resident at 2am in the "real world”?

Thanks. Tom Yuen, MD

Faculty Crozer Family Medicine Residency


Address messages to: IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG <mailto:IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG>

To unsubscribe from IMPROVEDX: click the following link:
http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1 <http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1> or send email to: IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG <mailto:IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG>

Visit the searchable archives or adjust your subscription at: http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?INDEX <http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?INDEX> 
Moderator:David Meyers, Board Member, Society for Improving Diagnosis in Medicine

To learn more about SIDM visit:
http://www.improvediagnosis.org/ <http://www.improvediagnosis.org/>


To unsubscribe from IMPROVEDX: click the following link:
http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1 <http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1>

or send email to: IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG

Visit the searchable archives or adjust your subscription at: http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?INDEX


Moderator:David Meyers, Board Member, Society for Improving Diagnosis in Medicine

To learn more about SIDM visit:
http://www.improvediagnosis.org/








Moderator: David Meyers, Board Member, Society to Improve Diagnosis in Medicine


HTML Version:
URL: <../attachments/20180909/45058e77/attachment.html>


More information about the Test mailing list