2009 Livsmedelsverket 5 - Proficiency Testing - Trace Elements in Foood Round T-18
2009 Livsmedelsverket 5 - Proficiency Testing - Trace Elements in Foood Round T-18
2009 Livsmedelsverket 5 - Proficiency Testing - Trace Elements in Foood Round T-18
LIVSMEDELS
VERKET
NATIONAL FOOD
ADMINISTRATION, Sweden
�
Produktion: Livsmedelsverkets rapportserie är avsedd för publicering
Livsmedelsverket, Box 622 av projektrapporter, metodprövningar, utredningar m m.
SE-751 26 Uppsala, Sweden I serien ingår även reserapporter och konferensmaterial.
För innehållet svarar författarna själva.
Teknisk redaktör: Rapporter som trycks utges i varierande upplagor och till-
Merethe Andersen trycks i mån av efterfrågan. De kan rekvireras från Livsmedels-
Uppsala 2009-05-20 verkets kundtjänst tel 018-17 55 06, fax 018-17 55 11 eller via
webbplatsen www.livsmedelsverket.se
SWEDISH NATIONAL FOOD ADMINISTRATION
Food Chemistry
Laboratory code:
TABLE OF CONTENTS
1. Summary 5
2. Introduction 5
3. Test material 5
3.1 Test for homogeneity 6
3.2 Tables 1-2: Results of the homogeneity test 7
4. Instructions to participants 7
5. Statistical evaluation 7
5.1 Z-scores 9
5.2 Horrat-value 9
6. Results
Table 3: Results and z-scores 10
Table 4: Summary of statistical parameters 11
Table 5: Percent satisfactory z-scores 11
8. References 18
The test material for Round T-18 consisted of fish soup in which Pb, Cd, Cr, Cu,
and Fe were available for analysis. The participants were instructed to report the
results on fresh weight basis.
The assigned values (xa) were derived from Robust statistics, with the exception
of Fe for which Kernel density was used. The target value (σp) for the standard
deviation was based on the Horwitz equation
2. Introduction
3. Test material
The homogeneity testing was carried out at the NFA. After stratifying, ten
sample containers were randomly selected, one from each stratum and
analysed in duplicate. The samples were microwave-digested under pressure
and the metals determined by atomic absorption spectrophotometry. Lead and
cadmium was checked for homogeneity according to the procedure described
in [2]. Chromium, copper and iron were checked using the same method, but
which is not accredited for these elements. The data was checked for extreme
differences within bottle by Cochran outlier test. If an outlier was detected it
was marked with a C in Tables 1 - 2 and not used in further calculations. The
analytical and sampling variances were estimated by one-way ANOVA. The
following parameters were calculated:
The results of the homogeneity test indicated that the test material was
sufficiently homogeneous for the elements to be analysed in this round.
Table 1. Pb, Results of the homogeneity test Table 2. Cd, Results of the homogeneity test
Bottle A B Bottle A B
1 0,056 0,053 1 0,0108 0,0111
2 0,059 0,063 2 0,0118 0,0111
3 0,066 0,053 3 0,0117 0,0114
4 0,052 0,051 4 0,0120 0,0108
5 0,052 0,055 5 0,0119 0,0124
6 0,049 0,058 6 0,0114 0,0106
7 0,052 0,074 7 0,0131 0,0121
8 0,050 0,050 8 0,0114 0,0122
9 0,049 0,050 9 0,0116 0,0122
10 0,049 0,046 10 0,0116 0,0112
Mean, n 0,054 20 Mean, n 0,012 20
san 0,006 san 0,0005
ssam2 0,000 ssam2 0,000
sall2 0,000 sall2 0,000
Critical 0,000 Critical 0,000
Suff. Homog. Yes Suff. Homog. Yes
Target sp 0,013 Target sp 0,0036
Ratio san/σp 0,21 Ratio san/σp 0,02
4 Instructions to participants
The participants were recommended to use their standard method, and were
asked to report one result per metal and to give the results with three
significant figures on fresh weight basis. Furthermore, the participants were
instructed to report the actual found value, even when being below the Limit
of. Detection (LOD).
The participants were also asked to provide information about the analytical
methods used. Method parameters are reported in appendix 1. Methods that
are accompanied by a reference are are listed in appendix 2.
5. Statistical evaluation
The assigned value (xa) is the best available estimate of the measurand. This may
be established in two ways: using the mode of kernel density, or using the robust
mean after removal of non-valid values according to [2]. The procedure described
below will generally be used. In some cases other statistical methods may be used
if considered necessary.
The kernel density plot is examined and classified according to the different cases
below:
Case A. If the kernel density plot shows a near normal distribution the robust
mean is used, see step 2.
Case B. If the distribution is tailing the mode of the kernel density is used as
assigned value.
In this round the distribution for Pb, Cd, Cr and Cu were considered to be Case A,
and the robust mean, which was calculated after outlier elimination, was used as
the assigned value (xa). Iron was considered to be Case B. The target value for the
standard deviation (σp) was calculated using the Horwitz equation.
Results reported as being below LOD (<-values) were not used in any
calculations and no z-score was calculated.
z = (x-xa) / σ where:
x = reported value of analyte concentration in the test material
xa = assigned value (”true” value)
σ = target value for the standard deviation
A z-score of |z| ≤2 implies that 95% of the results (i.e. 19 out of 20) should fall
within this limit and 5% should be outside of this value. This is not always
absolutely true. If the Z-score is based on a population of results in which several
result have been eliminated as outliers before the assigned value is defined, it is
probable that more than 5% of the total number of results fall outside of 2 Z-
scores, i.e. those identified as outliers. For the remaining results the first statement
still applies.
5.2 Horrat-value
The Horrat (Horwitz ratio) value is the ratio between the experimental relative
standard deviation from the current PT-round and the RSD estimated from the
Horwitz equation. This equation describes the expected relative standard deviation
at a given concentration. The Horrat indicates if the result of the interlaboratory
study has a satisfactory precision. If the ratio exceeds 2 the precision is
questionable, but not necessarily unacceptable. The Horrat is given for the
different elements in Table 4 as additional information.
Reported results and z-scores are found in table 3. Table 4 shows a summary
of the statistical parameters, and Table 5 gives a presentation of z-score
distribution
Table 3. Found results as reported, expressed as mg/kg and z-scores. Round T-18,
2009, Fish soup.
Assigned
value 0,057 0,010 0,099 8,056 12,00
Lab Pb Cd Cr Cu Fe
no Found Z-score Found Z-score Found Z-score Found Z-score Found Z-score
1 0,0695 1,0 0,0115 0,5 0,0990 0,0 8,34 0,3 11,7 -0,2
2 0,055 -0,2 0,011 0,2 0,125 1,2 12,2 4,4 17,0 3,8
3 0,0533 -0,3 0,0106 0,1 0,0770 -1,0 8,47 0,4 12,1 0,1
4 0,072 1,2 0,009 -0,6 0,120 1,0 8,71 0,7 12,35 0,3
5 0,068 0,8 0,012 0,7 0,138 1,8 8,83 0,8 13,9 1,4
6 0,015 -3,4 0,007 -1,5 0,034 -3,0 4,86 -3,4 17,1 3,9
7 0,089 2,5 0,011 0,2 - - -
8 0,00 -4,5 0,11 43,4 - 7,69 -0,4 -
9 0,047 -0,8 0,010 -0,2 0,137 1,8 10,4 2,5 13,0 0,8
10 0,046 -0,9 0,008 -1,1 0,091 -0,3 7,34 -0,8 10,5 -1,1
11 0,0438 -1,1 0,0108 0,2 0,0966 -0,1 9,09 1,1 14,1 1,6
12 0,0565 -0,1 0,011 0,2 0,103 0,2 - -
13 0,0711 1,1 0,0141 1,6 - 8,88 0,9 12 0,0
14 0,052 -0,4 0,010 -0,2 - 8,51 0,5 11,9 -0,1
15 ND 0,005 -2,4 ND 7,63 -0,5 13,7 1,3
16 - - - 7,39 -0,7 11,6 -0,3
17 0,254 15,6 0,0089 -0,7 0,09 -0,4 6,46 -1,7 11,25 -0,6
18 0,054 -0,3 0,0093 -0,5 0,051 -2,2 6,43 -1,7 11,6 -0,3
19 0,029 -2,2 0,013 1,1 - 13,1 5,4 12,8 0,6
20 < 0,08 < 0,05 0,054 -2,1 7,1 -1,0 10,3 -1,3
21 0,081 1,9 0,057 20,3 - - -
Pb Cd Cr Cu Fe
All Without All Without All Without All Without All Without
results outliers results outliers results outliers results outliers results outliers
Assigned value 0,057 0,010 0,099 8,06 12,0
Target sigma 0,013 0,002 0,022 0,94 1,3
n 18 14 19 16 13 12 18 17 17 17
Mean 0,064 0,057 0,018 0,010 0,094 0,098 8,41 8,14 12,8 12,8
Median 0,055 0,055 0,011 0,011 0,097 0,098 8,41 8,34 12,1 12,1
Robust Mean 0,057 0,057 0,011 0,010 0,094 0,099 8,23 8,06 12,5 12,5
SD 0,052 0,014 0,025 0,002 0,033 0,029 1,98 1,65 1,9 1,9
Robust SD 0,023 0,014 0,003 0,002 0,036 0,032 1,55 1,32 1,5 1,5
HORRAT 1,1 0,8 1,5 1,4 1,2
Min 0,000 0,029 0,005 0,007 0,034 0,051 4,86 4,86 10,3 10,3
Max 0,254 0,081 0,110 0,014 0,138 0,138 13,1 12,2 17,1 17,1
Pb Cd Cr Cu Fe
n, all results 18 19 13 18 17
Satisfactory 13 16 10 14 15
Questionable 2 1 2 1 0
Unsatisfactory 3 2 1 3 2
Satisfactory,% 72% 84% 77% 78% 88%
1
Z-score
-1
-2
-3
-4
-5
8 6 19 11 10 9 14 3 18 2 12 5 1 13 4 21 7 17
Laboratory Number
1
Z-score
-1
-2
-3
-4
-5
15 6 10 17 4 18 9 14 3 11 2 7 12 1 5 19 13 21 8
Laboratory Number
1
Z-score
-1
-2
-3
-4
-5
6 18 20 3 17 10 11 1 12 4 2 9 5
Laboratory Number
1
Z-score
-1
-2
-3
-4
-5
6 18 17 20 10 16 15 8 1 3 14 4 5 13 11 9 2 19
Laboratory Number
0 2 4 6 8 10 12 14 16
1
Z-score
-1
-2
-3
-4
-5
20 10 17 16 18 1 14 13 3 4 19 9 15 5 11 2 6
Laboratory Number
0 5 10 15 20 25
* Digestion technique
DA=Dry Ashing
MDC=Microwawe Digestion, Closed
MDO=Microwawe Digestion, Open
OWD=Open Wet Digestion
PBD=Pressure Bomb Digestion
Reference method
Lab no Metal*
19 Pb, Cd, Cu, Fe
*Metals in Bold have z-score outside ±2
Institute/Laboratory Country
Ministry of Healts, Department of Scientific Services Brunei Darusalam
Grain Research Laboratory Canada
Instituto de Salud Pública de Chile Chile
Laboratorio de Diagnóstico y Análisis de la Región de Magallanes y
Antartica Chile Chile
Universidad Técnica Frederico Santa Maria Chile
Instituto National de Pesca Ecuador
Matís Iceland
Central Food Laboratory India
Istituto Superiore di Sanità Italy
Republic Institute of Health Protection Macedonia
Bufete Quimico SA de CV Mexico
Centro Nacional de Servicios de Constatación en Salud Animal Mexico
Veterinærinstituttet Norway
INASSA International Analytical Services SAC Peru
Instituto Tecnologico Pesquero del Perú Peru
Sociedad de Asesoramiento Técnico Peru
Center for Food Analysis Serbia
Laboratorio Arbitral Agroalimentario Spain
Industrial Technology Institute Sri lanka
ALS Analytica AB Sweden
Eurofins Food / Agro Sweden AB Sweden
Livsmedelsverket Sweden
SVA, avd för kemi Sweden
Ministry of Public Healts, Department of Medical Schiences Thailand
LIVSMEDELS
VERKET
NATIONAL FOOD ISSN 1104-7089
ADMINISTRATION, Sweden
�