Elemental Analysis Manual: Section 4.5B Cold Vapor Atomic Absorption Spectrometric Determination of Total Mercury in Seafood Using Microwave Assisted Digestion

Appendix B - Supplemental Information on Method Performance

Version 1 (June 2008)
Author: John Cheng

GLOSSARY

---

Interlaboratory performance of EAM Method 4.5 was estimated from analytical data and results from laboratory proficiency testing. Although the design and purpose of proficiency testing is different than for method performance studies, the fundamental requirements and conditions for reproducibility were met and the data and results were suitable for determining interlaboratory method performance.

4.5B.1 MERCURY QUALITY ASSURANCE PROGRAM (MQAP)

The MQAP is a proficiency testing program organized and managed by the Canadian Food Inspection Agency for assessing laboratory performance. Participation in the program is voluntary, free of charge, open to all organizations, and participant information is held confidential. Each dispatched round consists of several test materials that are sent to enrolled labs for analysis. Analytical instructions are provided along with a report form for reporting results. An extensive report summarizing results, descriptions of analytical methodologies used, and performance scores (participants are identified by code only) is distributed to all participants upon conclusion.

In 2004, four FDA laboratories participated in MQAP series 352-355. MQAP series 352-355 consisted of 4 canned seafood samples to be analyzed for total mercury. Each can was identified only by a sample number with no additional information. The accompanied paperwork instructed participants to analyze each sample in triplicate on 2 different days and to report all six total mercury values. All FDA labs followed procedures in EAM Method 4.5 draft A using identical mercury analyzers (CETAC Technologies Quick Trace model M-7500) and microwave digestion systems from CEM Corporation. FDA participation was coordinated through the Division of Field Science (DFS). A copy of the MQAP 352-355 final report1 and the 4 laboratory codes used to identify FDA participants in the report was provided by DFS.

4.5B.2 PROFICIENCY RESULTS

To compare interlaboratory results, the MQAP used assigned values based on consensus, i.e., the mean, corrected for outliers, of all reported results for each sample¹. FDA results, individually and collectively, were in good agreement with assigned values (4.5B Table 1). Individual FDA lab results and the interlaboratory mean were within 1 standard deviation of the assigned values for all 4 samples. However, the ratio of FDA mean to assign value was slightly greater than 100% for all samples. This apparent difference is likely due to a lowering influence on the consensus value from participants that used open beaker acid digestion during sample preparation, a process that is susceptible to mercury loss due to volatilization. Also, assigned values in comparison studies are not intended to be used in the same context as true or certified values (reference materials) so a small difference is not a good indicator of error. A graphical comparison of means and variances (4.5B Figure 1) between FDA interlaboratory mean and the MQAP assigned value does not appear statistically or significantly different.

4.5B Table 1. FDA Mercury Results for MQAP Samples 352-355a

	Hg (mg/kg)
FDA Lab	352	353	354	354
1	0.368 ± 0.027	0.316 ± 0.009	0.202 ± 0.013	0.610 ± 0.012
2	0.407 ± 0.012	0.333 ± 0.008	0.222 ± 0.004	0.654 ± 0.018
3	0.402 ± 0.041	0.317 ± 0.023	0.190 ± 0.039	0.638 ± 0.047
4	0.398 ± 0.007	0.316 ± 0.006	0.206 ± 0.006	0.616 ± 0.027
Mean	0.394	0.321	0.205	0.63
Assigned valueb	0.377 ± 0.048 (61)	0.302 ± 0.043 (62)	0.198 ± 0.032 (62)	0.576 ± 0.094 (62)
FDA Mean assigned value	104%	106%	104%	109%

^a FDA results are mean ± 1 standard deviation; n=6.
^b Corrected mean ± 1 standard deviation of all MQAP participants in mg/kg. Number of data points is in parenthesis.

4.5B Table 2 shows the MQAP series 352-355 z-scores for FDA participants. Individual z-scores were calculated as follows:

All FDA z-scores were acceptable, i.e., |z| ≤ 2, indicating satisfactory performance.

4.5B Table 2. FDA z-scores for MQAP 352-355

	Sample ID
FDA Lab	352	353	354	355
1	-0.188	0.336	0.133	0.365
2	0.619	0.732	0.749	0.831
3	0.509	0.359	-0.233	0.661
4	0.421	0.342	0.256	0.427

4.5B.3 INTERLABORATORY PRECISION AND HORRAT

Interlaboratory precision was estimated by adding the intra-laboratory variance and variance of the interlaboratory mean. The intra-laboratory variance was estimated from the pooled intra-laboratory standard deviations (average variance). The observed interlaboratory precision ranged from 5 to 12% relative standard deviation and the HORRAT ranged from 0.3 to 0.6 (4.5B Table 3).

4.5B Table 3.
FDA interlaboratory precision for MQAP series 352-355
Mean and stdev (n = 6) in mg/kg.

	Sample #352
FDA Lab	mean	stdev	pooled stdeva
1	0.368	0.027	0.0255
2	0.407	0.012
3	0.402	0.041
4	0.398	0.007
FDA interlab	0.394	0.018
Interlab precision
stdevb	0.031
RSD	7.9
HORRATc	0.4

	Sample #353
FDA Lab	mean	stdev	pooled stdeva
1	0.316	0.009	0.0133
2	0.333	0.008
3	0.317	0.023
4	0.316	0.006
FDA interlab	0.321	0.008
Interlab precision
stdevb	0.157
RSD	4.9
HORRATc	0.3

	Sample #354
FDA Lab	mean	stdev	pooled stdeva
1	0.202	0.013	0.0209
2	0.222	0.004
3	0.19	0.039
4	0.206	0.006
FDA interlab	0.205	0.013
Interlab precision
stdevb	0.0247
RSD	12
HORRATc	0.6

	Sample #355
FDA Lab	mean	stdev	pooled stdeva
1	0.61	0.012	0.0292
2	0.654	0.018
3	0.638	0.047
4	0.616	0.02
FDA interlab	0.394	0.018
Interlab precision
stdevb	0.0355
RSD	5.6
HORRATc	0.3

^a where s_k = standard deviation of mean for laboratory k (k = 1 to 4)
^b where s_r = pooled intra-laboratory standard deviation and s_d = standard deviation of interlaboratory mean
^c Reference 2

4.5B.4 CONCLUSION

The results from MQAP proficiency testing indicate satisfactory performance among the FDA laboratories using EAM Method 4.5 draft A. FDA interlaboratory results are in good agreement, within statistical uncertainty, with overall consensus values. The estimated interlaboratory precision ranged from 5 to 12% relative standard deviation for seafood samples containing approximately 0.2 to 0.6 mg/kg total mercury.

---

REFERENCES

1. Mercury Quality Assurance Program Series 352-355 Final Report, December 27, 2004. Canadian Food Inspection Agency. 
2. Official Methods of Analysis of AOAC INTERNATIONAL (2005) 18^th Ed., AOAC INTERNATIONAL, Gaithersburg, MD, USA, Appendix D: Guidelines for Collaborative Study Procedures To Validate Characteristics of a Method of Analysis. AOAC .