- On 20 Oct 1997 at 12:36:14, Yasuo koyama (yasuo.k.aaa.respo.or.jp) sent the message

Back to the Top

PharmPK - Discussions about Pharmacokinetics

Pharmacodynamics and related topics

Dear readers,

Would anyone teach me which parameter, mean or median, is usually used in

evaluating the typical plasma profile and PK population parameters for the

data has the large inter-individual variability.

Thanks in advance.

Yasuo Koyama

koyamay.at.anet.ne.jp - On 21 Oct 1997 at 10:08:47, David_Bourne (david.-at-.pharm.cpb.uokhsc.edu) sent the message

Back to the Top

[db - another multireply question ;-)]

From: "Bonate, Peter, HMR/US"

To: "'PharmPK.aaa.pharm.cpb.uokhsc.edu'"

Subject: RE: Median and Mean

Date: Mon, 20 Oct 1997 16:06:57 -0500

X-Priority: 3

Mime-Version: 1.0

Whether to choose the median or mean depends on the distribution of the

data. Typically, pk parameters are log-normally distributed, in which

case the most appropriate parameter is the median. Unfortunately, most

scientists mis-represent data by reporting it as though the data came

from a normal distribution. Perhaps the simplest way to determine

whether a distribution is normal or log-normal is to transform the data

using the natural log and then use a goodness of fit test for the normal

distribution. A useful reference in this regards is R.B. D'Agostino and

M.A. Stephens: Goodness of Fit Techniques. Marcell Dekker, New York,

1986. Another good reference is LF Lacey et al: J Biopharmaceutical

Statistics 7, 171-178, 1997.

Hope this helps.

Peter L. Bonate, Ph.D.

Hoechst Marion Roussel

Clinical Pharmacokinetics

P.O. Box 9627 (F4-M3112)

Kansas City, MO 64134

fax: 816-966-6999

phone: 816-966-3723

--

X-Sender: jelliffe.-a-.hsc.usc.edu

Date: Mon, 20 Oct 1997 14:07:29 -0700

To: PharmPK.-a-.pharm.cpb.uokhsc.edu

From: Roger Jelliffe

Subject: Re: Median and Mean

Mime-Version: 1.0

Dear Yasuo:

Usually the median is better, as things are usually more symmetrically

distributed about it. Examples are the Vd of a drug when its distribution

is not Gaussian, and it hardly ever is. You might look at Dodge et al,

Population PK Models - Measures of Central Tendency, Drug Invest 5:

206-211, 1993. Hope this helps.

Roger Jelliffe

************************************************

Roger W. Jelliffe, M.D.

USC Lab of Applied Pharmacokinetics

CSC 134-B, 2250 Alcazar St, Los Angeles CA 90033

Phone (213)342-1300, Fax (213)342-1302

email=jelliffe.aaa.hsc.usc.edu

************************************************

Take a look at our Web page for announcements of

new software and upcoming workshops and events!!

It is http://www.usc.edu/hsc/lab_apk/

************************************************

--

Priority: Normal

X-MSMail-Priority: Normal

X-Priority: 3

To: PharmPK.at.pharm.cpb.uokhsc.edu

MIME-Version: 1.0

From: "Steve Duffull"

Subject: Re: Median and Mean

Date: Tue, 21 Oct 97 16:55:09 PDT

It seems to me that the data will determine whether you use mean, median or

perhaps the mode to describe the central tendency of the distribution.

For the typical plasma profile, I think that it is often best just to show

just the individual's data. Often presenting the mean, or similar, will

hide interesting things in the conc-time profile such as secondary peaks

which may be due to enterohepatic cycling.

For PK population parameters, you should present the information that you

have for them. If you used a computer program that yielded the geometric

mean (and CV%) then this is what you should present. If, however, you used

a computer program that yielded the frequency histogram of the parameters

then it would seem appropriate to present the entire frequency histogram.

Presenting only the arithmetic mean and standard deviation implies that the

distribution of the parameter is Gaussian (which would need formal

testing). Presenting the median and quartiles (etc) does not describe the

distribution very well. Both parametric and nonparametric descriptions

will tend to hide interesting things in the distribution of the parameters,

eg subpopulations of patients that have different values (eg genetic

polymorphism).

I hope this helps.

Steve Duffull

=======================Steve Duffull

Christchurch

New Zealand

Ph +64 3 3815280

Fax +64 3 3640902

Email sduffull.-at-.clear.net.nz - On 23 Oct 1997 at 09:42:54, "Leon Aarons" (laarons.aaa.fs1.pa.man.ac.uk) sent the message

Back to the Top

NONMEM, because it is an maximum likelihood method, reports the mode

for the population parameters. This is one reason that the

population estimates do not equate to the mean or median of the

individual estimates.

__________________________________________________

Leon Aarons

School of Pharmacy and Pharmaceutical Sciences

University of Manchester

Manchester, M13 9PL, U.K.

tel +44-161-275-2357

fax +44-161-275-2396

email l.aarons.-a-.man.ac.uk - On 18 Nov 1997 at 13:41:23, "Vladimir Piotrovskij" (VPIOTROV.-at-.JANBELC1.SSW.JNJ.COM) sent the message

Back to the Top

Yasuo,

If you want a simple answer to your simple question, it is median.

More detailed analysis, however, is required if the inter-individual

variability is extreemly large, as in case of polymorphic metabolism.

If the distribution of CL is actually bi- or polymodal you can hardly

find a unique parameter to represent a typical value.

How to do such analysis using minimal modelling assumptions? Suppose

(a rather typical example) you have many individual plasma

concentration-time profiles obtained at the same dosing regimen, and

individual sampling times are close to each other. Calculate medians

of individual measurements belonging to each sampling time cluster.

Then take ratios of measurements to the corresponding medians and plot

those ratios versus the individual sequential numbers. You will

immediately see the 'outlying' subjects which you have to analyse

separately from others. You may also pool all individual measurements

adjusted for sampling time-specific medians to get idea about the form

of overall concentration distribution (unimodal, polymodal, normal,

lognormal, etc.). This simple approach will help you to select an

appropriate way of further summarizing your PK data.

Regards,

Vladimir

--------

Vladimir Piotrovsky, Ph.D. Fax: +32-14-605834

Janssen Research Foundation Email: vpiotrov.-a-.janbe.jnj.com

Clinical Pharmacokinetics vpiotrov.-at-.janbelc1.ssw.jnj.com

B-2340 Beerse

Belgium

______________________________ Reply Separator

_________________________________

Subject: Median and Mean

Author: PHARMPK (INTERNET.PHARMPK1) at SWIFTGATE

Date: 10/20/97 3:59 PM

PharmPK - Discussions about Pharmacokinetics

Pharmacodynamics and related topics

Dear readers,

Would anyone teach me which parameter, mean or median, is usually used in

evaluating the typical plasma profile and PK population parameters for the

data has the large inter-individual variability.

Thanks in advance.

Yasuo Koyama

koyamay.-at-.anet.ne.jp

Also see: http://www.cpb.uokhsc.edu/pkin/pkin.html

Want to post a follow-up message on this topic? If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Median and Mean" as the subject

PharmPK Discussion List Archive Index page

Copyright 1995-2010 David W. A. Bourne (david@boomer.org)