1. Proficiency Test Samples: What are We Really Testing?
Dr. Robert Litman, Dr. Daniel Montgomery, Dr. David Stan Morton and Mr. Robert Shannon, Environmental Management Support
John Griggs, Director, US EPA National Analytical Radiation Environmental Laboratory (NAREL)

2. The Analytical Method for a Radionuclide
The method ensures that the chemical approach for the radionuclide and matrix, based on an event, will reliably convert the sample into an appropriate sample test source and facilitate acceptable levels of measurement bias and uncertainty?

3. PT Samples – Traditional Paradigm…
One of the biggest challenges for labs: PT samples
PT Samples demonstrate methods work as expected based on:
Chemical separations are effective (when used)
Instrument calibration and performance
Analyst Training
Suitability of the method for the matrix being analyzed …

4. What is “Suitability”
To paraphrase Gilbert and Sullivan: “Let the PT fit the sample” One method for analysis of, let’s say uranium in ground water, may not be suitable for each of the following matrices: surface water, soil, concrete, vegetation, DRPs… Or paraphrasing the Isotoner commercials in the inverse: “One PT does NOT fit all”

5. How Do Laboratories Treat PT Samples?
Success: When audits note that methods are satisfactory based on PT results
For the most part, laboratories do what is asked of them
Most clients trust labs to do the analysis correctly
Over time, laboratories tailor their methods to pass PT samples
This may or may not be good for emergency response samples (although for routine programs, as well)

6. PT Samples Test Real Analytical Systems
Why then:
Do PTs utilize the same test configurations again and again?
Are radionuclides present in readily soluble form?
Are radionuclides (usually) homogenously distributed?
Does preservation not reflect regulatory/lab requirements?
Are dynamic parent-progeny relationships not usually tested?
Are DQOs/MQOs clearly defined?
Are turn-around times long and numbers of samples small?
Do labs analyze PTs they would never routinely accept?

7. Soluble, Homogenously-Distributed Radionuclides
The same (boring) radionuclides are present again and again but not necessarily the ones in real samples
PT samples are homogenously distributed
Labs routinely analyze small subsamples of solids without homogenizing and still obtain acceptable results
Radionuclides are present in readily soluble form
Real samples may contain intractable chemical forms (various redox states, refractory Pu, complexes, compounds)
Fused matrix and DRPs would be expected in many events

8. Realistic DQOs/MQOs and Program Needs Provided to Labs
DQOs/MQOs are not usually clearly defined
Labs develop methods for PTs with no relationship to routine work
Hot labs receive low-level samples and ‘environmental’ labs hot samples
Results could be reported and evaluated to take this into account. A hybrid of traditional evaluation measures would show that a method is not suitable for its primary purpose while Z-scores would allow a hot- lab to demonstrate method performance relative to false positives

9. Realistic DQOs/MQOs and Program Needs Provided to Labs
DQOs/MQOs are not usually clearly defined
Turn-around times are long and numbers of samples small
Some PT programs touch on these issues, but they need to be much more central both to the provider and the lab...

10. PT Sample Preservation
We know that:
Iodine is lost from vegetation and filters
Zirconium will precipitate even in acid solution
Decay chain may not be in equilibrium
fission products in equilibrium in PT, or 214Bi/214Pb requested
There is no practical preservation technique for all fission and activation products in real-life water …
When PT samples are preserved differently from real-life samples results may indicate that all lab processes work even though analysis of samples might fail in real-life

11. In the End PTs May Provide a Misleading Picture of Performance
Labs obtain false sense of security about the performance of their methods
Regulators and lab customers use PTs to establish a misguided understanding of data quality and capabilities they can expect from a lab’s methods

12. Problem #1: What type of matrix?
Let’s first assume a radiological event: Urban Environment
What commodities, commercial products, or materials are likely to become contaminated?
Can samples be created that
represent the manner of contamination?
If surface contamination: Can they be transported to lab without loss?
Can/should the whole sample be analyzed?
Can we get a representative part if only analyzing some of it?

13. We Need to “Right–Size” Samples
Sample characteristics should reflect the program in question
Analytes/matrix
Homogeneity
DQOs/MQOs
Possibly preservation
Other key PT sample characteristics
The lab may refuse the sample just as they would a customer’s sample
Alternatively, PT programs could offer an alternative evaluation
Accept as with alternative MQOs, e.g., Z-score – a low-activity PT would function as a false-positive test of sorts (without significant radio-interferences) for a higher level lab

14. Example 1:Scabbled Concrete - PT
PT based on: A building was contaminated with 241Am following an RDD event. The surface was scabbled to remove the surface contamination.
Chemical form of the radionuclide used to spike?
How is the PT made (e.g., surface spiked then scabbled)?
Who homogenizes the PT?
What direction is the laboratory given to decide how much sample?
Minimum aliquant size
Are MQOs clearly communicated so the lab can determine how much sample to analyze?

15. A Real-Life Sample?
YES An RDD in an urban environment would generate such a sample. The PT form should be representative of the way the surface was contaminated/removed to form the sample. Should there be separate PTs for a sample already homogenized and one that is not?

16. Problem 2: Radionuclide spike with multiple Oxidation states/Chemical forms?
Is the oxidation state important?
Can multiple oxidation states be achieved in same PT?
Pu, U, Ru, Mo, Np, Cm, I, Fe, etc.
Are different chemical constituents accounted for in the PT?
Are different matrix configurations accounted for in the PT?

17. Problem 3: Emergency Response Samples are Unique
Can QC samples be available for method validation of ER matrices?
High and low activity concentrations depending upon lab
Analyte and matrix characteristics
Assess carrier/tracer yield?
Blank samples used for detectability determination?
Measurement uncertainty determination?
Can the same data reduction process normally employed be used?

18. Problem 4: Extraneous Radionuclides Present
Should PT samples have radionuclides present that are not of “interest”?
Assess the effectiveness of separation capability?
Create interferences that must be accounted for in the data reduction process?
Which extraneous radionuclides should be added to the PT
Do the concentration ratios present realistic problems?

19. Problem 5: Unannounced Practice
Participating laboratories willing to accept samples with
Different matrices
Three different radionuclides requiring three different methods
Two day turn-around-time
Number of labs kept to 3-4 per practice

20. How Do These Issues Get Addressed?
Currently – often not addressed which creates a huge gap in the effectiveness of PT programs
Need to test:
Real sample characteristics
Capability AND capacity
How many laboratories will participate, who will determine validity of results?
How are data processed?

21. Are We Assessing the Laboratory PT results?
How do we evaluate the results?
PT sample evaluation should not be arbitrary
80-120% limits are arbitrary.
Tie the assessment on DQOs/MQOs provided to labs
Z-score, assesses bias. and uncertainty
Which analytics should be used?
Z-score, t-test, Grubbs test
False negative, or positive at LLD
Labs must adjust their programs/methods so they will meet requirements
What feedback do laboratories get about their methods/results?
More would be better

22. Proficiency Testing Samples are the crux of the issue
Multiple levels of testing needed
Non-normal radionuclides need to be part of process
Result validation and assessment must also be a proficiency test
An ink and paper process
Should be separate from actual sample analysis
How do these efforts get funded?

23. Questions?