Back to the Top
Dear all,
I am so sorry to bother everyone on this very basic concept yet giving
me problems. In our instituition we only practise 3rd dose TDM based
on the concept of steady state. It is known to assume that by the 3rd
dose, drugs would have reach steady state thus when you take peak and
trough levels after 3rd dose, you will get the estimate peak and
trough at steady state.
3rd dose TDM can delay treatment outcomes as it takes a longer time
before we draw levels. I have heard and seen others using first dose
kinetics to extrapolate the levels to calculate the Cmax and Cmin at
steady state. I believe this will help patients to achieve therapeutic
levels faster. However I am not sure how to explain this concept to
the physicians and thus change the practise.
Can anyone teach me about first dose kinetics?
Thanks
Angelinia
Pharmacist
Singapore
Back to the Top
Angelina,
I don't know where you heard about the '3rd dose TDM' idea but it has
little if any scientific credibility. What you call 'first dose
kinetics' has been advocated for decades.
There is no reason to wait for steady state in order to learn about an
individual's pharmacokinetics. What you can learn depends on how many
samples you are prepared to take but two samples are typically enough
e.g. for aminoglycosides taking sample about 30 min after the end of
an infusion of the first dose and a sample about 8 hours later is a
simple way to get enough information for dose individualisation.
A Bayesian forecasting tool such as TCIWorks is a convenient way to
interpret concentrations and also build up a locally relevant
pharmacokinetic database.
http://www.tciworks.info/download.html
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.at.auckland.ac.nz
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford
Back to the Top
Angelinia,
You can contact Prof. Roger W. Jelliffe at. His
working team is steady developing nonlinear PK tools for PK/PD
analysis for different clinical scenarios including first dose TDM.
Regards,
Dimiter Terziivanov
--
Dimiter Terziivanov, MD,PhD,DSc, Professor and Head, Clinic of
Clinical Pharmacology and Pharmacokinetics, Univ Hosp "St. Ivan Rilski",
15 Acad. Ivan Geshov st,
1431 Sofia, Bulgaria
e-mail: dterziivanov.aaa.rilski.com; terziiv.aaa.yahoo.com
Back to the Top
The following message was posted to: PharmPK
Dear Angelina,
Building on Nick's remarks, one way of explaining the need to change
to your medical staff is that, they learned to do 3rd dose kinetics at
"steady state" because not too many years ago (appoximately 20-25
years) most readily available means of computation (i.e., first
generation hand held calculators) were not sufficiently powerful to
handle none steady state equations and subsequent calculations. The
programs then readily available at most hospital wards used steady
state calculations to fit and/or interpret serum drug levels. Now,
powerful computers are available within the hospital environment, and
indeed, most PDA's possess sufficient computational power to curve fit
serum drug concentrations after any dose--"pre-steady state, steady
state, whatever." So there is no need to wait to get levels with a
consequent delay of therapy. It's time to move this aspect of patient
care into the 21st century.
Ron Floyd, PharmD, BCPS, FCCP
Sharp Mary Birch Hospital for Women
San Diego, CA 92123
Back to the Top
Ron,
You wrote:
'Building on Nick's remarks, one way of explaining the need to change
to your medical staff is that, they learned to do 3rd dose kinetics at
"steady state" because not too many years ago (appoximately 20-25
years) most readily available means of computation (i.e., first
generation hand held calculators) were not sufficiently powerful to
handle none steady state equations and subsequent calculations. '
Your history is not quite accurate. Programmable calculators (e.g.
TI-59) appeared about 1975 (33 years ago) when I arrived at UCSF as a
clinical pharmacology fellow. Several people there learned the crude
programming language and implemented PK tools for interpreting 'first
dose kinetics' in order to used TDM for dose forecasting (e.g. Keith
Muir and Mike Winter). Carl Peck implemented a non-linear least
squares estimation first on on a Commodore PET and then on an HP-85.
Lewis Sheiner built on that pioneering step and implemented Bayesian
estimation on the HP-85 in the early 1980s which led to the commercial
realisation by Abbott of the DrugCalc Bayesian forecasting software
for use with their TDM concentration measurement kits. Around the same
time Roger Jelliffe masterminded the development of the USC-PAK
software including, among other great ideas, the clinical
interpretation of drug concentrations based on similar concepts.
All this was before 1983 when I moved to New Zealand - more than 25
years ago. Hospital computers became faster and more flexible and
moved steadily into the clinical workspace. However as far as I know
they all ignored this well established science and effectively the
practice of TDM became dummer and dummer as evidenced by the '3rd dose
kinetics' question that started this thread. As I am completely out of
touch with what happens in hospitals these days I would be very
interested to hear if there are hospitals which currently provide
Bayesian forecasting methods linked directly to drug concentrations
(which are no doubt reported in the electronic patient record).
It is interesting to note that a Google search for TDM and UCSF brings
up many more hits for TDM as 'traffic demand management' than it does
for 'therapeutic drug monitoring'.
A comprehensive review of the almost lost science of Bayesian
forecasting can be found in Mungall et al. 1995.
In recent times the technology has been reincarnated in the freely
available TCIWorks system (http://www.tciworks.info/download.html).
TCI=Target Concentration Intervention is a deliberate renaming of TDM
("TDM is tedium")- see Holford 1999.
IMHO doctors should never learn anything about pharmacokinetics and
all quantitative dose decisions should be done by computers. Doctors
dont need to know how a car engine works or what street to take in
order to visit their patients -- the car engine is controlled by
computer and the route is determined by the GPS. Similar technology
can help them write the optimal dose on the prescription.
Best wishes,
Nick
Holford NHG. Target Concentration Intervention: Beyond Y2K. Br J Clin
Pharmacol. 1999;48:9-13.
Mungall D, Heissler J, Kaltenbach M. Computer applications in clinical
pharmacokinetics and pharmacodynamics. In: Derendorf H, Hochhaus G,
editors. Handbook of Pharmacokinetic/Pharmacodynamic Correlation By
CRC Press; 1995. p. 415-63.
http://books.google.co.nz/books?hl=en&lr=&id=DOe43pEklIEC&oi=fnd&pg=PA415&dq=Mungall+D,+Heissler+J,+Kaltenbach+M.+&ots=0iaLHJtQvI&sig=mBbeQXlTUorI5CWRzl7cWPIBj14
[URL - one line - db]
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.aaa.auckland.ac.nz tel:+64(9)923-6730 fax:+64(9)373-7090
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford
Back to the Top
The following message was posted to: PharmPK
Dear Nick,
While I was a pharmacokinetics fellow at USC in 1975 (with Jelliffe's
USC Pak available for first dose kinetics) I moonlighted as a
pharmacist at a typical small community hospital in a nearby suburb.
No (as in not any) computing facilities were available there. The
only pharmacokinetics support was what I could accomplish on my HP-25--
an amazing scientific calculator with RED LED's (highly visible in the
dark of a patient's room) but which lost the program whenever it was
turned off. I agree that other early programmable calculators were
available at the time and many improvements were made quickly
thereafter but these largely did not trickle down very quickly into
the community hospitals which made up then and which continue to make
up the largest proportion of patient care provision in the US.
While there are many brilliant individual scientists and physicians
working in the field of PK/PD and clinical pharmacology, most seem to
limit their endeavours to research and/or teaching within university
settings. And this practice of teaching medicine seems to me to be
quite conservative, in the sense of changing what is communicated very
slowly. Thus IMO we have the persistence of "3rd dose" serum levels
because 25-30 (or 33) years ago, this is what could readily be
accomplished at most medical centers. This is what the medical
students at the time learned was practical (for them) and so when they
became teachers of medicine, this is what they passed on as wisdom to
their students.
(Both you and I were very fortunate in that we received out training
at that time at institutions that were very much at the forefront of
the development of PK. At the time I did not appreciate just how
advanced the few "PK centers" in the US were, over most other medical
and pharmacy teaching centers).
I quite agree that physicians generally should not be concerned with
the mechanics of pharmacokinetics and/or pharmacodynamics. This
should be transparent to their practice. I think an argument could be
made (and you could make it far better than I) that physicians should
not even be concerned with the mechanics of pharmacology. Once a
working diagnosis has been made, let the clinical pharmacologist (or
computer algorithm) select the drug, tailor the regimen to the
individual and adjust.
With all respect, my best wishes to you!
Ron Floyd
Back to the Top
Dear Angelina:
I think the 3rd dose approach came about because of the old
obsolete method of doing linear regression on the logs of the levels.
The only way this method would work was when the samples were drawn
at the steady state and after the drug had distributed itself in the
body after the last dose. Neither of these preconditions is useful any
more.
Forget this, as Nick says. Almost any decent software package
now easily permits some sort of Bayesian analysis of the data obtained
at any time. In view of this, there is no need, as you say, to wait
for such conditions before doing TDM. It leaves the patient at risk
by not knowing how he/she is handling the drug. Do your TDM with the
first dose. Usually a peak level just after the end of the IV
infusion, and a second one drawn when you think the levels have fallen
to about 1/3 of the peak value (often about 8 hrs for aminoglycosides)
is not far from a D-optimal strategy, and will give you close to
optimal Bayesian posterior model parameter estimates.
Another question is how you choose to weight the assay data
according to its credibillity. A good way is to determine the
relationship between the assay concentration and the SD with which it
is measured. Take a blank sample, another at a low level, another at a
middle level., another at a high level, and another at the highest
level you can. Measure each of them in at least quintuplicate (some
say 10 replicates for each sample). You need more samples to get good
estimates of SD's and variances than you do just for means. That is
why so many samples are needed. Fit the relationship between the mean
concentration and the SD to a polynomial of order 2 or 3. Then you can
enter this in the software as a good estimate of the assay error for
any single measured sample, so it can be correctly weighted by its
Fisher information (the reciprocal of its variance - see any
statistics book under Fisher information).This is much more useful
than describing it using CV%. The CV% cannot be used at all for
fitting such TDM data, and data has to be censored because of the
illusion of imprecision with low measurements.
The other errors in the environment are the errors with which
the doses are prepared and given. Usually about a 5-10% SD is
realistic for the doses (most software does not consider these
errors). The other remaining source of error is essentially that of
the errors in recording the times at which doses are given and samples
are drawn.
All these sources of error can be specifi ally entered into
the MM-USCPACK clinical software for multiple model (MM) Bayesian
adaptive control of drug dosage regimens. The software uses
nonparametric population models because they are consistent, because
they make no assumptions about the shape of the model parameter
distributions, and because the likelihoods they compute are exact, and
because they permit prediction of the weighted squared error with
which a selected therapeutic target goal will be achieved.. These
models are therefore naturally linked to the MM method of dosage
design, which produces the regimen which specifically minimizes the
weighted squared error with which a clinically selected target goal is
achieved. This is the great strength between the nonparametric models
and the maximally precise MM designed drug dosage regimens.
The data of doses and levels can be entered into the
software, and two (soon three) methods of Bayesian analysis can be
done. One is the MM
method of getting a Bayesian posterior, which also can use data of
rapidly changing creatinine clearance estimation in patients with
rapidly changing renal function to fit the data and get a good model
for an individual patient. The other is the interacting MM sequential
Bayesian approach which tracks parameter values for which there are no
good descriptors or covariates. This IMM approach has tracked the
behavior of Gentamicin and Vancomycin better than any other current
method in post cardiac surgical patients. Both MM and IMM are much
better than the usual parametric MAP Bayesian method, which cannot
estimate the predicted errors or precision with which any dosage
regimen is likely to be achieved.
It is because of the use of nonparametric pop models,
estimation of CCr based of a pair of changing serum creatinine
specimens, and the use of MM dosage design that the above software
permits mathematically optimal use of all information available up to
the present time in developing the next regimen. This is now available
as a demonstration version of the MM-USCPACK software. You can
download it from our web site http://lapk2.hsc.usc.edu/beta. User name
is lapkbeta, password is uscpack. It is easy to use, and is highly
graphical.
References:
1. Bayard D, Jelliffe RW, Schumitzky A, Milman M, and Van Guilder
M: Precision Drug Dosage Regimens Using Multiple Model Adaptive
Control: Theory, and Application to Stimulated Vancomycin Therapy.
Vasudevan Memorial Volume on Theoretical Physics, Stochastic
Processes, and Biological Sciences. ed. by Rao KS and Sridhar R,
Allied Publishers Ltd., Madras, India, pp. 407-426, 1995.
2. Jelliffe R, Schumitzky A, Bayard D, Milman M, Van Guilder M,
Wang X, Jiang F, Barbaut X, and Maire P: Model-Based, Goal-Oriented,
Individualized Drug Therapy: Linkage of Population Modeling, New
"Multiple Model" Dosage Design, Bayesian Feedback, and Individualized
Target Goals. Clin. Pharmacokinet. 34: 57-77, 1998.
3. Jelliffe R, Schumitzky A, and Van Guilder M: Population
Pharmacokinetic / Pharmacodynamic Modeling: Parametric and
Nonparametric Methods. Therap. Drug Monit. 22: 354-365, 2000.
4. Jelliffe R: Estimation of Creatinine Clearance in Patients
with Unstable Renal Function, without a Urine Specimen. Am. J.
Nephrology, 22: 320-324, 2002.
5. Bustad A, Terziivanov D, Leary R, Port R, Schumitzky A, and
Jelliffe R: Parametric and Nonparametric Population Methods: Their
Comparative Performance in Analysing a Clinical Data Set and Two Monte
Carlo Simulation Studies. Clin. Pharmacokinet., 45: 365-383, 2006.
6. Bayard D, and Jelliffe R: A Bayesian Approach to Tracking
Patients having Changing Pharmacokinetic Parameters. J. Pharmacokin.
Pharmacodyn. 31 (1): 75-107, 2004.
7. Macdonald I, Staatz C, Jelliffe R, and Thomson A: Evaluation
and Comparison of Simple Multiple Model, Richer Data Multiple Model,
and Sequential Interacting Multiple Model (IMM) Bayesian Analyses of
Gentamicin and Vancomycin Data Collected From Patients Undergoing
Cardiothoracic Surgery. Ther. Drug Monit. 30:67-74, 2008.
8. Neely M, and Jelliffe R: Practical Therapeutic Drug Management
in HIV-Infected Patients: Use of Population Pharmacokinetic Models
Supplemented by Individualized Bayesian Dose Optimization. J Clin
Pharmacol. 48: 1081-1091, 2008.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.at.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dear Nick:
There is a difference between Bayesian forecasting, which
does not really have a specific target goal in mind, but rather an
implied range, and Bayesian Adaptive Control, which has a specific
target and aims to hit it most precisely. This is the difference
between UCSF and USC. If you ave a window in mind, then you need to
quantify and compare the risks and probabilities of being below it,
and the same for being above it. This makes the decision making
process extremely complicated and generally insoluble. In contrast,
when you have a specific target, the risks of being just a little bit
below it are hardly different from those of being just a little bit
above it. This is the main reason we select specific target goals and
try to hit them with greatest precision. It makes the process of
decision analysis MUCH easier. This is why we now have moved from
using parametric models to nonparametric models and multiple model
(MM) dosage design, calculations of CCr suited for rapid changes in
renal function, and, if needed, the Interacting MM (IMM) sequential
Bayesian approach to track drug behavior best in extremely unstable
patients. We are now also studying new strategies to learn about the
patient optimally while having to treat him/her at the same time. We
used to use time shared computers back in 1966-7 to do the job, even
before assays became available, using pop models for digitoxin and
digoxin, and comparing the patient's clinical behavior with the
computed concentration of drug in the patient's body.
IMHO, doctors need to learn enough PK and enough about target
goals for patients so they can select, and take the responsibility for
selecting, the target goals, adjusted up or down for each patient
according to their clinical judgment and the patients perceived need
for the drug. They they or a pharmacist can go about the job, using a
PC laptop at the nurses' station, for example, of seeing what regimen
achieves the target goal most precisely. Failure to teach such
approaches is what has made us all such a world of therapeutic
cripples at the present time. Achieving reasonably selected target
goals improves care, reduces complications, reduces hospital stay, and
reduces costs. Most pharmacists are not trained to take the clinical
responsibility to set anything more than commonly accepted target
goals, and this limits their ability to truly individualize therapy.
As to hospitals which do Bayesian adaptive control, there are
Sander Vinks in Cincinnati, Jeff Barrett in Philadelphia, and Milan
Grundman in Ostrava, Czech Republic, for openers, and Nathalie Bleyzac
with Busulfan and Cyclosporine in children with bone marrow
transplants, and Pierre Marquet with cyclosporine also in France, and
also a group in Marseille, which has done a super job of treating
patients with testicular CA. Not USC though, as far as I know, sad to
say.
All the best,
Roger
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-a-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Roger,
I fully agree with you that it is easier to hit a target than to to
vaguely aim for a window. That is exactly why the term Target
Concentration Intervention was introduced. I dont agree there is
really any difference in this respect between the USCF and USC
methods. If you choose a target conc then all methods do much the same
thing.
Thanks for the names of some enthusisasts who use Bayesian
forecasting. But what I asked for was instances of hospitals which
have Bayesian forecasting built into their day to day electronic
patient reporting systems. Do you know of any institutions that have
got their regular systems doing something useful to help pick the
optimal dose?
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.-a-.auckland.ac.nz
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford
Back to the Top
The following message was posted to: PharmPK
Nick Holford said:
"There is no reason to wait for steady state in order to learn about
an individual's pharmacokinetics. What you can learn depends on how
many samples you are prepared to take but two samples are typically
enough e.g. for aminoglycosides taking sample about 30 min after the
end of an infusion of the first dose and a sample about 8 hours later
is a simple way to get enough information for dose individualisation."
From my experience (since ca 1988), at many hospitals in the US and
Canada, this approach is simply not tenable for practical reasons.
When instructions are given to obtain patient samples for PK
monitoring of an aminoglycoside, the laboratory sends the phlebotomy
technician to draw blood just before the the next dose is given, and
to then return and draw the peak as you instructed. Obviously, if the
regimen is not at steady-state, the "trough" from the previous dose
will not be the same as the trough from which the "peak" is obtained.
This will cause an error in the estimation of ke, and all subsequent
calculations based on that ke will multiply the error, regardless of
sophistication. I realize that if one were to use a Bayesian approach
it could adjust the ke to the population value, but to answer your
original question, I personally am not aware of any hospital that has
such software built into the patient information system.
There are many other challenges to implementation of "cutting edge" PK
techniques: inadequate documentation of dosing times by overworked
nurses, inadequate documentation of sampling times by laboratory
technicians, lack of information technology support, (it took me one
year to get a decent method for estimating CrCL implemented into a
hospital information system), lack of standardization (ie JCAHO does
not have a best practice model for PK monitoring), bureaucratic red-
tape etc. I realize these can be seen as opportunities for
improvement, and encourage practitioners to continue to push for the
best for our patients, but those in academe and research must
recognize the "state of the art" in the vast majority of clinical
practice is nowhere near to using Bayesian or any other technology
much more advanced than a calculator with linear regression.
--
William R. Wolowich, Pharm.D., R.Ph., B.Sc.Pharm.
Chair
Department of Pharmacy Practice
College of Pharmacy
Nova Southeastern University
Ft. Lauderdale, FL.
Back to the Top
The following message was posted to: PharmPK
The majority of pharmacokinetic work for patients is "3 rd dose, pre
and post levels" in hospitals. There may be some that do otherwise,
but I have not worked in them.
At times for the occasional patient, I have ordered a single dose,
with levels at 1 hour and anywhere from 8 to 18 hours after the dose
to try and estimate ke, and then estimate subsequent doses and
intervals. Not ideal, but how can we afford the software for every
hospital in Canada? I have seen demos of Dr. Jelliffe's program, and
heard about others. How many rural/urban community teaching hospitals
do sophisticated pharmacokinetic workups?
Greg Soon
Pharmacist, Critical Care
Peterborough Regional Health Centre
Peterborough, ON, Canada
Back to the Top
The following message was posted to: PharmPK
Hi,
> The majority of pharmacokinetic work for patients is "3 rd dose, pre
> and post levels" in hospitals. There may be some that do otherwise,
> but I have not worked in them. [Ontario, Canada]
> the vast majority of clinical practice is nowhere near to using
> Bayesian or any other technology much more advanced than a calculator
> with linear regression. [Florida, USA]
I must say I was a bit taken aback at these two responses from
pharmacists working on the fringes of North America. My own career in
therapeutic drug monitoring (a discipline later to become target
concentration intervention) started in the late 1970s when it seems we
could do a better job of dosing critically ill patients than seems to be
possible today.
There is a saying "You can take a horse to water but you can't make it
drink". Many of the reasons why medical practice takes a long time to
accept and implement the findings of medical research have been explored
by Berwick e.g. See Berwick 2003.
This includes the example of it taking 264 years after the first
definitive demonstration that scurvy was preventable before it became
official policy and practice in Britain. So it may be another 200 years
or so before the the fruits of PK research become applied routinely in
health care ...
Nick
Berwick DM. Disseminating innovations in health care. Journal of the
American Medical Association. 2003;289(15):1969-75.
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.at.auckland.ac.nz
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford
[I agree Nick, it is discouraging. Back when I was at the University
of Kentucky, post-BS Pharm.D. students were taking blood samples and
consulting daily with Pharmacy faculty. Samples were assayed in the
Pharmaceutics section of the College (at no cost?). It was a great
educational experience but not economically sustainable or was it? Was
there enough work done on the cost benefit relationships? - db]
Back to the Top
The following message was posted to: PharmPK
Reading through the discussions on the concept of Therapeutic Drug
Monitoring (TDM) or even Target Concentration Intervention, reminds me
that I
too started off in the late 1970s in this area and recall getting to
grips
with computer programmes. To this end, we were encouraged by Syva
(now part
of Dade Behring) who developed the EMIT system, and the production of
their
programmes and later Abbott Diagnostics with their systems. Of course
all
this was preceded by the days of the nomograms (eg Vozeh, Richens,
Mawer),
useful for the interpretation of phenytoin or the nomogram to predict
chloramphenicol and theophylline (Koup and others) from a result of one
serum sample six hours after an initial dose of the drug. All of which
enabled the pharmacist to be actively involved together with our
clinical
biochemists colleagues, to be part of the clinical team monitoring
patient
care, although I have to admit more could be done, hopefully before
another
200 years have passed.
And to answer DB's query, certainly the measurement of anticonvulsant
and
the benefit it provides has been well documented. The same is true for
digoxin, theophylline and the aminoglycosides. For others, it is
probably
not so clear cut.
Graham Mould
Graham Mould, PhD
Consultant Pharmacist
Email: gmould.-at-.gcpl.co.uk
Surrey Technology Centre
Surrey Research Park, Occam Road
Guildford, Surrey GU2 7YG
[I agree tdm (? or tci) has benefits for a number of drugs including
those listed. However, it is intesting to note that theophylline for
one is no longer a topic in the latest edition of a text on applied
pk. Is it cheaper to use easier drugs? Cost-Benefit - db]
Back to the Top
The following message was posted to: PharmPK
Nick, I agree. I wish that we were doing more...money comes into it, and
training pharmacists to do them. We have a lot of community/retail
pharmacists working for us, and these "consults" scare the heck out of
them. Most do not have the training, and sometimes the ability. Most
little
hospitals don't have the training programs required.
The hospital pharmacy specialization (residency/PharmD) trains us to be
able to do TDM (in the 80's that was what we called it), but there are
few
of us around compared to the number of hospitals across North America.
Greg
(we are not that much on the fringe, despite being in the boonies).
Back to the Top
Graham,
A small expansion on your history -- The nomogram that Sam Vozeh
published was created after Lewis Sheiner described his HP-85 method
for Bayesian forecasting. Sam was a fellow of Lewis' and got the idea
to create the phenytoin nomogram after the computer based method for
drugs with linear kinetics had been established.
I can verify that George Mawer's aminoglycoside nomogram preceded the
Bayesian era at UCSF. I was a medical student taught by George and
used his nomogram on surgical wards when I was a junior doctor in
1972. My successes with this method so impressed my surgeon colleagues
that it was one of the reasons I eventually went into a career in
clinical pharmacology and on George's advice went to train at UCSF.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.-at-.auckland.ac.nz
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford
Back to the Top
Dear Greg and all:
The thing most people do not yet realize is that by improving
care with TDM (see Sander Vinks, Chris Destache, and Nathalie Bleyzac,
for example, complications are reduced, their treatment is reduced,
hospital stay is reduced, and the overall costs are reduced. Put in a
little effort, see improved care. See Nathalie Bleyzac, I think:
Bleyzac N, Souillet G, Magron P, Janoly A, Martin P, Bertrand Y,
Galambrun C, Dai Q, Maire P, Jelliffe R, and Aulagner G: Improved
clinical outcome of paediatric marrow recipients using a test dose and
Bayesian pharmacokinetic individualization of busulfan dosage
regimens. Bone Marrow Transplantation, 28: 743-751, 2001.
Martin P, Bleyzac N, Souillet G, Galambrun C, Bertrand Y, Maire P,
Jelliffe R, and Aulagner G: Relationship between CsA trough blood
concentration and severity of acute graft-versus-host disease after
paediatric stem cell transplantation from matched sibling or unrelated
donors. Bone Marrow Transplantation 32: 777-784, 2003.
Martin P, Bleyzac N, Souillet G, Galambrun C, Bertrand Y, Maire P,
Jelliffe R, and Aulagner G: Graft Versus Host Disease: Clinical and
Pharmacolgical Risk Factors for Acute Graft-versus Host Disease after
Paediatric Bone Marrow Transplantation from matched sibling or
Unrelated Donors. Bone Marrow Transplantation 32: 881-887, 2003.
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-a-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dear Nick and all:
As long as we are expanding on history, we had plastic card
wallet nomograms for digoxin in 1970 and for gentamicin and kanamycin
dosage in 1971. We had time-shared computer software for digitalis
compounds in general in 1967.
Jelliffe RW: Nomograms for Kanamycin and Gentamicin Therapy. 11th
Interscience Conference on Antimicrobial Agents and Chemotherapy,
Atlantic City, N.J., October 19-22, 1971. Abstracts of the conference,
page 63.
Jelliffe RW and Brooker G: A Nomogram for Digoxin Therapy. Second
Annual Fall Scientific Assembly of the American College of Chest
Physicians, in Los Angeles, California, October 25-30, 1970. Chest,
58: 282, 1970.
Jelliffe R and Brooker G: A Nomogram for Digoxin Therapy. Am J Med,
57: 63-68,1974.
All the best,
Roger
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-a-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dear Nick:
Good for you. I agree very much with the idea of hitting a
target. It seems to me that Bayesian Forecasting does not specifically
evaluate the expected precision with which a dosage regimen will hit a
selected target. In our USC MM-USCPACK software, though, we do exactly
that. That is the strength of multiple model dosage design, based on
nonparametric population models. The regimen always minimizes the
weighted squared error with which the desired target(s) are achieved.
As individual patient data is obtained, the Bayesian posterior
distribution is obtained, and the maximally precise regimen developed
again. That is multiple model Bayesian adaptive control, in contrast
to forecasting. Bayesian forecasting and Bayesian adaptive control are
really not the same thing. Much the same thing is not the same thing.
As to electronic medical records, I cannot say about that yet.
All the best,
Roger
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-at-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
Roger - I still remember chatting with you at the back of the lecture
room at the PK/PD course prior to ICAAC San Diego.
I and a couple of other pharmacists on staff believe in TDM, and we
try to do what we can (with the limited resources we have). They have
come a long way with mentoring and training.
The question is how to get software like yours into hospitals in the
first place.
Greg
Peterborough
Greg Soon
Clinical Pharmacist - Critical Care
Peterborough Regional Health Centre
1 Hospital Drive
Peterborough, ON K9J 7C6
Back to the Top
The following message was posted to: PharmPK
Please see the following references:
Streetman DS, Nafziger AN, Destache CJ, Bertino JS, Jr. Individualized
Pharmacokinetic Monitoring Results in Less Aminoglycoside-Associated
Nephrotoxicity. Pharmacotherapy 2001;21:443-451.
Bertino JM, Rodvold K, Destache CJ. Cost-Benefit of a Clinical
Pharmacokinetic Service. Clin Pharmacokin 1994;26:641-651.
Destache CJ. The Use of Therapeutic Drug Monitoring in
Pharmacoeconomics. Ther Drug Monitor 1993;15:608-610.
Destache CJ. Cost-Benefit Relationships in the Use of a Pharmacokinetic
Service. PharmacoEcon 1993;3:433-436.
Destache CJ, Meyer SM, Bittner MB, Hermann KG. Impact of a Clinical
Pharmacokinetic Service on Patients Treated with Aminoglycosides: A
Cost-Benefit Analysis. Ther Drug Monit 1990;12:419-426.
Destache CJ, Meyer SM, Rowley KA. Does Accepting Pharmacokinetic
Recommendations Impact Length of Hospitalization? A Cost-Benefit
Analysis. Ther Drug Monit 1990;12:427-433.
Chris
Chris Destache, Pharm. D., FCCP
Professor of Pharmacy Practice, Internal Medicine, Medical Microbiology
and Immunology
Creighton University School of Pharmacy & Health Professions
Hixson-Lied Bldg. Rm 114
2500 California Plaza
Omaha, NE 68178
Email Destache.aaa.creighton.edu
Back to the Top
Dear Greg:
The payoff is that good TDM reduces overall costs. If you
like, you can go to our web site, and download a demo version. Our web
site is below. Any trouble, pease let me know. The software is pretty
intuitive. But now we need to do a decent user manual.
All the best,
Roger J
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-at-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
Dear Roger,
You wrote:
"I think the 3rd dose approach came about because of the old
obsolete method of doing linear regression on the logs of the levels.
The only way this method would work was when the samples were drawn
at the steady state and after the drug had distributed itself in the
body after the last dose. Neither of these preconditions is useful any
more".
Although I do agree with you that there is no need to waste time to
get to steady state and dose individuation can be made right after the
first dose, I see it is important not to draw the first sample until
the distribution phase is complicated. This is especially important if
the dose is administered as a short infusion. Several hospitals around
the world are even using a slow IV push for aminoglycosides (which may
arguably provide a better dynamic response following first dose), and
thus a greater error in PK parameter estimates is expected.
Rgds,
Khalid Alkharfy
College of Pharmacy
King Saud University
Riyadh, Saudi Arabia
Back to the Top
Hi Nick et al.,
More "history"...
To expand on the pre-Bayesian, nomogram period with respect
specifically to phenytoin therapy, Richens and Dunlop published their
nomogram which was not based on Michaelis-Menten kinetics in 1975
(Richens, A. and Dunlop, A. Lancet 2, 247-248, 1975) and a revised
version which was (Rambeck, B. et al., Ther. Drug Monitoring, 1,
325-333, 1979). In between these two papers, Tom Ludden published a
graphical technique based on a linear transformation of the M-M
equation (Ludden, T.M. et al., Clin. Pharm. Therap., 21, 287-293,
1977) and I proposed using the "direct linear plot" (DLP) (Mullen,
P.W., Brit. J. Clin. Pharm., 4, 733-734, 1977 and Mullen, P.W., Clin.
Pharm. Therap., 23, 228-232, 1978).
Sam Vozeh's "orbit graph" technique was actually a Bayesian expansion
of the DLP approach (Vozeh, S. et al., J. Pharmacokin. Biopharm., 9,
131-146, 1981).
As you know, George Mawer was my mentor also.
Cheers,
Peter
Peter W. Mullen, PhD, FCSFS
KEMIC BIORESEARCH
Kentville
Nova Scotia, B4N 4H8
Canada
E-mail pmullen.at.kemic.com
Back to the Top
Dear Khalid:
Thanks for your note. Actually, there is no need to wait for
distribution to be complete after a dose. For digoxin, for example,
the D-optimal times are about 1.75 hr after an oral dose, and at the
trough. For the exchange rate constants from central to peripheral
compartment, the best times are about 0.5 hr and 7 hr after the oral
dose. You will miss a lot if you wait for full distribution after the
dose. For the aminoglycosides, even with a short IV infusion, the best
times are at the peak, of maybe 5 min after, and then when the levels
have fallen to about 1/3 the peak, of 1/k, or .36 of the true peak
value.
Look at D'Argenio D: Optimal sampling times for pharmacokinetic
experiments. J. Pharmacokin. Biopharm. 9: 739-56, 1981.
All the best,
Roger
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-a-.usc.edu
Our web site= http://www.lapk.org
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 "First dose kinetics concept" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)