(
h
)
HorRat values
.—HorRat values never exceed 1.5
except for several analytical linesofCa (2.8and2.6),K(2.0),
Na (1.8), P (1.6), andZn (1.6, 1.7) in sterilized cream, of Ca
(1.8), P (1.9, 2.2, 1.6) in nonfat milk powder, and of K in
infant cereals (1.7).However, the reliability and applicability
of themethodisproved,asAOACcriteriaarefulfilledwithno
values higher than 3.0. If several values are between 0.2 and
0.5, thiscanbeexplainedbyexperience, trainingofanalysts, a
goodknowledge of testedmatrixes, and careful application.
(
i
)
Ruggedness
.—Accuracy (SD
rec
) and intermediate
precision (SD
iR
) results aredisplayed inTables 20–28 for the
six matrixes tested. Global statistic recovery results were
satisfactory for all elements in all matrixes whatever the
digestion systems and ICP-AES equipments used.
Unsatisfactory statistic results (
P
<0.01) were found mainly
for the infant cereals matrix: Ca for MD1 andMD3, Fe for
MD4, KforMD1,Mg forMD1 andMD3, andNa forMD4.
OtherunsatisfactoryresultsconcernCa incornbranforMD4,
Cu in nonfat milk powder for MD3, Fe in dietetic milk
powder 1 and in corn bran for MD4, K in dietetic milk
powder1forMD1,PincornbranforMD4, andZn indietetic
milk powder for MD1. Unsatisfactory statistical recovery
results mainly concern infant cereals matrix due to relative
heterogeneity of the used batch and corn branmatrix due to
whole element concentration values close to QL. These
statistical unsatisfactory results have shown that general
applicability of the method is limited by less analytical
experience and competency regarding sample preparation
(e.g., sample homogeneity and adapted digestion program)
and ICP analysis (adapted ion buffering and use of robust
conditions).
However, AOAC requirements for recovery performance
are globally fulfilled for all elements in all matrixes between
74 and 118% for element concentrations from 0.26 mg/kg
(Mn) to 16 900mg/kg (K). Only specific recovery values of
Ca (75%), Fe (72%), andP(68%) incornbranforMD4; ofK
(81 and 82%) for MD1 and MD2, respectively; recovery
valuesofCu(67and74%) innonfatmilkpowder anddietetic
milk powder forMD3 andMD4, respectively; and recovery
value of Ca (113%) in infant cereals for MD1 showed
unacceptable values regarding AOAC criteria, as already
demonstrated for statistic recovery test at 95%confidence.
AOACrequirements forRSD
r
andSD
iR
were fulfilled for
all elements in all matrixes between 0.6 and 18.9% and 0.9
and 23.3%, respectively, for concentrations from0.26mg/kg
(Mn) to16900mg/kg (K).
Theseruggedness testshaveshowngoodrobustnessof this
method applied by different operators (at least two operators
per testofeightanalyticalseries)afteropen-andclosed-vessel
digestions on different ICP-AES equipments with axial,
radial, and dual view grating configurations using Cs 0.1%
(w/v) asminimal ionizationbuffer concentration.
RingTrial
(
a
)
Calcium
.—Table 29 presents estimated performance
characteristics of the ICP-AESmethod for determination of
Ca, Cu, Fe, K, Mg, Mn, Na, P, and Zn. The RSD
r
of the
P
OITEVINETAL.
: J
OURNALOF
AOACI
NTERNATIONAL
V
OL
. 92,N
O
. 5, 2009
1501
Table 17. SLVstudy: accuracy, precision, andHorRat values found for sodiuminsixcertifiedand in-house referencematerialsusing recommendedandalternate
lines
Material description
Line
Reference
value,mg/kg
a,b
Median±U,
mg/kg
c
Recovery,%
P
-value
d
SD
r
,mg/kg
e
SD
iR
,mg/kg
f
RSD
r
,% RSD
iR
,% HorRat
g
Sterilizedcream
NA
h
505±19
554±67
110±6
0.13
2
6
3.4
10.9
1.8
Chocolatemilkpowder
NA
h
1990±42
1950±74
98±2
0.37
32
59
1.6
3.0
0.6
Infant cereals
NA
h
1050±14
1054±64
100±2
0.89
34
60
3.2
5.7
1.0
Cornbran
NA
h
430±16
424±26
99±4
0.40
6
19
1.5
4.4
0.7
Dieteticmilkpowder 1
NA
h
4920±30
4890±87
99±1
0.20
40
79
0.8
1.6
0.4
Nonfatmilkpowder
NA
h
4970±50
4910±321
99±1
0.61
65
297
1.3
6.1
1.4
a
SDexpressedashalf of stateduncertainty (NISTandLGCcertifiedmaterials) givenas95%confidence interval.
b
SDexpressedasa reproducibilitySD(in-house referencematerials) calculatedaccording to robust statistics (29).
c
Uncertainty (U) estimated throughsimplifiedapproach (30, 31).
d
Statisticevaluationof robust
t
-test: Satisfactory result (
P
>0.05); questionable result (0.01<
P
<0.05); unsatisfactory result (
P
<0.01).
e
SD
r
=SDof repeatability.
f
SD
iR
=SDof intermediate reproducibility.
g
HorRat =RatiousingRSD
iR
andHorwitzdenominator according to ref. 32.
h
Recommended line for ICP-AESanalysis.
MTE-01
FORWORKINGGROUP/ERPUSEONLY
DONOTDISTRIBUTE
