Re[2]: Film badge/TLD results

[Sandy_Perle_at_USFPL201@email.fpl.com]

     Right on Rick!!!  The easy part is validating the processor. The 
     difficult task is to ensure that system processes "haven't changed." 
     That means establishing a Quality System with process as well as 
     outcome indicators. My program processed around 60,000 TLDs last year. 
     25% of my budget address quality assurance. When you consider that 
     staff makes up about 60% of budget, you can see that quality is a very 
     large percentage of dollars spent.
     
     Since the quality assurance program is intended to determine process 
     capability is maintained within specified tolerances .. we perform:
     
     1. Daily QC runs on all readers when processing for dose
     2. Monthly blind spiking from our plants
     3. Quarterly ANSI N13.11 blind testing .. all categories
     4. Other inter-comparison testing
     5. A multitude of statistical analysis .. such a t-testing, 
     probability testing, anova analysis, shewhart control charts (updated 
     daily), etc.
     
     Quality must be maintained .. to ensure that your workers are provided 
     the most accurate and precise dose results ..taking into account the 
     cost/benefit analysis. There is a point where the process must be 
     considered stable, where further improvements are not cost effective. 
     You must first determine your system's process capability to see if 
     you can improve your system .. before you attempt to do it.
     
     For further discussion .. please e-mail me directly.
     
     Note to Rick ... missed seeing you at the meeting in June. maybe next 
     year!
     
     Sandy Perle
     Supervisor Health Physics
     
     


______________________________ Reply Separator _________________________________
Subject: Re: Film badge/TLD results
Author:  radsafe@romulus.ehs.uiuc.edu at Internet-Mail
Date:    6/28/95 8:28 AM


     As the performance evaluation program administrator for DOELAP, I feel
     an overwhelming need to respond to Sandy's comments regarding the use
     of proficiency test results and assessments to assess commercial
     processors.  I believe that those two items only provide a starting
     point in the evaluation of a commercial processor.

     (first some preaching) Your goal is to provide the best dosimetry
     service you can to your customers, the people who work at your
     facility.  The commercial processor only processes dosimeters; they do
     not estimate dose of record for you.  As such, passing DOELAP or NVLAP
     testing weeds out the acceptable performers from the unacceptable, but
     not necessarily the good from the bad (gosh, that's hard to believe,
     but it's true.)  Therefore, for each processor you evaluate, you have
     to look at other things like: the completness of information you
     receive, the timeliness of reports and the controls put on special
     processing.

     After you select a processor, your job becomes even bigger.  In the
     DOE, for those laboratories using commercial processors, we require
     three things: (1) that the laboratory can demonstrate the accuracy of
     the results received from the processor; (2) that the laboratory can
     identify and correct anomalous dosimetry results at every point of the
     dose evaluation and (3) that the dose records for individuals contain
     sufficient information to track doses back to dosimeter reports that
     include processing information.  For #1, we recommend a combination of
     a blind audit program and a close relationship with the processor so
     the user can obtain QA information (such as QC processing information,
     calibration factors, ongoing test data and assessment reports.)  For
     #2, the DOE labs are using a 2-tier approach: (1) to identify
     anomalous doses and (2) to request that the processor provide
     information regarding anomalous chip response data as it occurs.  #3
     is pretty self-explanatory, i.e., you, as the supplier, need to
     provide the documentation linking dose histories to processing
     information.

     Okay, after this lengthy diatribe, I acknowledge that sometimes, we
     are all stuck with balancing accuracy, service and costs.  So even
     though, I believe that the major portion of our effort should be
     placed in obtaining the best accuracy and service, the question
     remains: How much do you want to spend to assure that doses below 100
     mrem/year are accurate?  Finally, with a nod to Sandy, you may want to
     contact him regarding how to track and trend dosimeter results to
     improve the performance of your program and identify ways to conduct
     cost/benefit analyses.

     Happy dosimetry.

     Rick Cummings      cumminfm@inel.gov

     Usual disclaimer regarding my opinions and employer's.

     "Shun advice at any price.  Now, that's what I call good advice."
                        Piet Hein


______________________________ Reply Separator _________________________________
Subject: Film badge/TLD results
Author:  RADSAFE (INELMAIL.RADSAFE) at _EMS
Date:    6/27/95 2:55 PM


Yesterday, Sandy Perle wrote:


    > Anyone who is concerned with the accuracy and precision of the
    > dosimetry processor they use, or are considering using, need only to
    > ask them for their NVLAP Proficiency test results ANSI N13.11 (1983 or
    > 1993 version). In addition I recommend that you ask them for the
    > on-site assessment report that is left with the lab.

    > These two items will provide enough information regarding the
    > processor to make a valid assessment of their overall program.


While true in principle, one should apply a reasonable dose of salt to the
proposal "if they did well on NVLAP (or DOELAP) then your processor will give
you good results".   I'm not taking issue with the validity of the NVLAP
process or with the idea of checking your processor's results, it should just
be kept in mind that a periodic controlled test does not guarantee "good"
dosimetry results.

A couple of things that come to  mind in this vein:


        A processor may pass NVLAP but fail DOELAP (or vice versa)...
        Or they may not be able to pass in a certain category...
        There is nothing forcing you to qualify in a given category, even
                though you may have exposure in that category...
        A processor may even use a special algorithm for reading TLDs for the
                NVLAP test that's not routinely used for personnel badges...

Other problems may creep in, such as a systematic error that is introduced by a
calibration foul-up that doesn't get corrected until months after introduction.
Or their QA may simply not be good enough to give consistent results in the
"real world".

I have heard of a number of horror stories regarding dosimetry snafu's and seen
some interesting results from limited blind spikes I have performed myself.