[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Statistics Question (Bioassay)
Elizabeth,
As you point out, there are several ways to do it, and it depends mostly
on the type of data that you have. Unfortunately, there is probably no
single method that can be used in every case. It's going to require
some judgment, but here are some things to consider:
1. If positive samples are identified within time periods that are
relatively short compared to the biological half-life of the compartment
with the shortest half-time, then you may be able to assess the dose as
being due to a single chronic intake instead of multiple acute intakes.
ICRP 54 discusses this approach.
2. If a person has a historical intake and has established a relatively
predictable excretion pattern (i.e., he/she either fits the patterns in
NUREG/CR-4884 OR he/she has a unique pattern that can be plotted over
time and shown to be predictable with little error), then comparison of
more recent bioassay data could be accomplished by comparing the more
recent measurement and its associated statistical error with the
predicted value. In this instance, you would determine if the predicted
bioassay measurement falls within the measured value +/- a predetermined
amount, usually some multiple of the standard deviation. Of course, if
you don't have estimates of error for the measurements, then this
technique is obviously not going to work, unless you're able to estimate
the error based on knowledge of the bioassay measurement process. If
the predicted value does fall within your tolerance limits for the
measurement, then the measurement is part of the same intake. If the
predicted value does not fall within the tolerance limits of the
measurement, then you've got a new intake.
3. If a person has a historical intake and has not established a
predictable excretion pattern, then more recent data that may or not be
positive gets more tricky; it will be mostly a judgment call. In this
case, one possible method is to judge for a particular day post-intake
the predicted value and the variation you may expect (based strictly on
your modeling). Then you compare the measured value and its associated
statistical error and see if the boundaries of the errors for predicted
versus measured overlap. You would have to set some criteria in advance
for how much overlap would be interpreted as the same intake versus a
new intake.
Obviously, this topic is more complicated than a single e-mail allows
time for. Any other experience from DOE folks who have done historical
intake evaluations?
Philip
__________________________
Philip C. Fulmer, PhD, CHP
Carolina Power & Light Company
Harris Energy and Environmental Center
3932 New Hill-Holleman Road
New Hill, NC 27562
philip.fulmer@cplc.com
(919) 362-3363
>----------
>From: Elizabeth M. Brackett[SMTP:brackett@bright.net]
>Sent: Friday, September 25, 1998 12:52 PM
>To: Multiple recipients of list
>Subject: Statistics Question (Bioassay)
>
><excerpt>Several people are having a debate regarding statistics relative
>to bioassay (urine) samples. The basic question is: How do you
>determine when a person has had a new intake of radioactive material if
>they had an intake in the past and they are still excreting activity from
>that intake?
>
>
>It seems that there are several ways to look at the issue. Do you assume
>you basically have one "lump" of radioactive material that you are
>measuring and then simply subtract the expected activity from the
>measured result (assuming you have decided that there is a measurable
>amount of activity in the sample) and say that if the difference is
>greater than 0, it's due to a new intake? Or, do you assume that there
>are 2 separate parts and that there is a greater than 0 background at the
>time of the second result, due to the previous intake, and calculate a
>new detection level on which to base your decision?
>
>
>Is there some other way to approach it? We are looking for a quick,
>simple, defensible way to do this. All of the samples are historical, so
>there is limited information available, and no chance of getting new
>samples. Some thoughts that we've had regarding disregarding subsequent
>bioassay datum are when: -the adjusted bioassay activity is less than 10
>or 20 or 100% of the detection limit and the later data are reasonably
>accounted for. -the adjusted activity is less than one or two times the
>compounded error for that value. -assuming an additional intake would be
>less than 0.02 or 0.1 ALI
>
>
>As stated in HPS N13.30, we desire "a bias on the high side ... when
>uncertainties of measurement or interpretations are present in order to
>be able to state with a high degree of confidence that ... cumulative
>exposures and risks are below a certain level."
>
>
>We look forward to your thoughts on this. Thanks in advance.
>
>
************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html