[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: AW: ppv and npv

From   "Lachenbruch, Peter" <>
To   <>
Subject   st: RE: AW: ppv and npv
Date   Mon, 9 Nov 2009 08:57:02 -0800

This is a handy ado file.  There is one issue that I have with NPV and PPV.  In most studies to determine sensitivity and specificity, the samples are conditional on disease status, so the prevalence of the disease in the training sample is quite different from what holds in most applications.

For example, in a AIDS clinic, the prevalence of disease is likely to be at most 0.05, while the training sample may be 50% HIV positive.
The PPV would then be p*Sensitivity/(p*Sensitivity+(1-p)(1-Specificity))  where p is the prevalence of the HIV.  If Sensitivity and Specificity are 0.95 (actually a bit low for HIV), if we assume prevalence is 0.5 we would have "PPV"=0.5*0.95/(0.5*0.95+0.5*0.05)=0.95
If we use a prevalence of 0.05, we would have 0.05*0.95/(0.05*0.95+0.95*0.05)=0.50
In a blood donation center, the prevalence is more like 0.01, the PPV would be 0.161
It is usually a good idea to compute PPV and NPV for a variety of prevalences as a test may be used in a number of different contexts. 


Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001

-----Original Message-----
From: [] On Behalf Of Martin Weiss
Sent: Monday, November 09, 2009 4:35 AM
Subject: st: AW: ppv and npv


Try Roger`s -ssc d senspec-


-----Ursprüngliche Nachricht-----
[] Im Auftrag von Celine Genty
Gesendet: Montag, 9. November 2009 13:16
Betreff: st: ppv and npv


The "roctab" command give for each cutpoint sensitivity / specificity / 
LR+ / LR-.

I need PPV and NPV for each cutpoint too... How can I do ?

Thank you.

Céline Genty

*   For searches and help try:

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index