568 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO.
3, 2003
MICROBIOLOGICAL METHODS
Enumeration of Probiotic Bacilli Spores in Animal Feed:
Interlaboratory Study
RENATA G.K. LEUSCHNER and JAN BeW
Central Science Laboratory, Department for Environment, Food and Rural Affairs, Sand Hutton, York YO41 1LZ, UK
ARMANDO CRUZ
Gaiker, Parque Tecnologico, Edificio 202, 48170 Zamudio (Bizkaia), Spain
Collaborators: A. Adler; E. Auclair; G. Bertin; M. Braconnier; K. Domig; P. Jones; W. Kneifel; M. Krause; S. Marmo;
J. Michard; H. Mietke; A. O’Briain; A.S. Olofson; W. Ruppitsch; A. Thalmann; A. Voets; C. Warmerdam
Fourteen out of 17 laboratories completed an
interlaboratory study comparing 2 pretreatment
protocols of feed samples containing authorized
probiotic bacilli spores. Both methods used
tryptone soy agar for enumeration. Pretreatment A
involved preparation of a suspension of the feed
sample in 50% ethanol. For pretreatment B, the
sample was suspended in peptone salt solution
and heated at 80°C for 10 min. Each laboratory an-
alyzed 12 samples (6 per pretreatment), which rep-
resented duplicates of a high (109 colony-forming
units [CFU]/g) and low (105 CFU/g) level of bacilli
spores or a blank that contained vegetative
probiotic bacteria only. For pretreatment A, the re-
peatability relative standard deviation (RSDr) was
2.9% for the low level and 2.5% for the high. The
reproducibility relative standard deviation (RSDR)
values were 7.8 and 5.9%, respectively. Pretreat-
ment B revealed RSDr values of 1.1 and 1.0%, and
RSDR values of 5.8 and 3.4%, respectively. The
heat treatment (pretreatment B) of feed samples
had better precision data, resulted in higher viable
bacilli counts, and was more effective in deactivat-
ing vegetative background flora. It is therefore rec-
ommended for adoption for official control pur-
poses and for CEN and ISO standards.
robiotics are viable microorganisms used in animal nu-
P trition as feed supplements to produce beneficial effects
in the host animal (1). Probiotic bacilli have effects on a
variety of animals and are therefore added to animal feed-
stuffs (2–9). Methods for the official control of probiotic ba-
cilli used as feed additives were developed within a European
Community project (SMT4 CT98-2235) to enable independ-
ent quality control of these products. For acceptance of stan-
dards by the Comité Européen de Normalisation (CEN),
Received August 26, 2002. Accepted by AH December 2, 2002.
Corresponding author’s e-mail: r.leuschner@csl.gov.uk.
methods must be validated by means of an interlaboratory
study. The method developed in this study was based on re-
ports in the literature and on available standard methods that
suggest heat treatment to inactivate vegetative cells before
bacterial spore enumeration in feedstuffs (10, 11). The heat
pretreatment is routinely used in industrial laboratories and in-
volves heating the initial suspension of the feed sample at
80°C for 10 min. The project partner of the European Union
project with responsibility for the development of a probiotic
bacilli enumeration method suggested an ethanol pretreat-
ment. This required initial suspension of the feedstuff sample
in 50% ethanol and incubation at room temperature on a rotary
shaker for 1 h. The ethanol treatment has been described as an
alternative for spores of Clostridium botulinum, which can be
either heat-sensitive or -resistant (12). Both methods used
nonselective tryptone soy agar (TSA) and incubated at 37°C
for 16–24 h under aerobic conditions.
Method precision data have not been available for either pre-
treatment enumeration methods except for Bacillus cereus (13).
The 2 pretreatment methods for enumeration of bacilli spores
were therefore validated according to a protocol conforming to
the International Harmonized Protocol for Collaborative
Studies (14). Data from the interlaboratry study were used to
calculate the repeatability (r) and reproducibility (R) of the
methods in relation to feed and premixtures for both
pretreatments using nonselective TSA (15). The precision data
of both methods were compared and, according to their perfor-
mance in the trial, one method was selected for Official Control
Method, and is intended for submission to CEN for inclusion in
the corresponding CEN methods and subsequent acceptance by
the International Organization for Standardization (ISO) via the
Vienna Agreement.
Interlaboratory Study
Study Design
Seventeen laboratories from 12 European countries partici-
pated in a interlaboratry study, the design of which was de-
scribed in a detailed trial chronology and standard operating
procedure (SOP). A test report was included, on which results
were to be recorded and returned to the trial leader for analy-
 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003 569

Table 1. Homogeneity data of 10 randomly selected validation samples of types I and J for probiotic bacilli spore
enumeration analyzed by pretreatment method A and a separate sample set by method Ba

Sample type I, method A 10–3 Sample type I, method B 10–3 Sample type J, method A 10–6 Sample type J, method B 10–7
dilution of initial suspension dilution of initial suspension dilution of initial suspension dilution of initial suspension

Replicate 1, Replicate 1, Replicate 1, Replicate 1,

Sample No. CFU/mL Rep. 2 CFU/mL Rep. 2 CFU/mL Rep. 2 CFU/mL Rep. 2

1 123 126 271 228 348 362 109 100

2 88 79 266 216 338 347 111 129
3 145 92 251 234 399 390 122 138
4 103 99 240 232 400 412 135 148
5 102 90 273 226 380 416 126 156
6 124 104 292 265 351 343 127 138
7 134 142 234 238 344 323 127 124
8 90 96 266 229 390 324 134 105
9 101 84 261 242 304 376 96 133
10 104 114 226 253 374 345 121 92
Mean 107 247.2 363.3 123.6
F 3.01 0.56 2.02 1.34
F critical 3.02 3.02 3.02 3.02
F < critical Pass Pass Pass Pass

a
Individual samples were enumerated from 2 separate serial dilutions (replicate 1, 2), each on duplicate plates (counts were added).

sis. The SOP was sent to participants 2 weeks before the trial
started, and other documentation was sent along with the vali-
dation samples. The SOP was written in ISO document format
and contained supplementary information and suggestions to
ensure that all samples were treated in the same way in each
laboratory. The test report contained all supplementary infor-
mation that could have influenced the results. The participants
were requested to record the following information: shipment
conditions, media formulations that were used in the trial
(manufacturer, code number, batch numbers, expiration
dates), temperature in the laboratory during analysis, incuba-
tion conditions, number of each colony type, total count per
plate, confirmation of results, and all deviations from the SOP.
Before the interlaboratory study, a pretrial was organized be-
tween the 5 contractors of the project to verify whether the
SOP as written contained any potential ambiguities and to es-
tablish that the test materials were fit for purpose.
Each participant received 2 sets of 6 blind samples, one for
method A and one for method B. Each of the 3 different sam-
ple types (I–K) were represented in duplicate in each set. All
test materials were extensively tested before the pretrial and
interlaboratory study to ensure that they could be maintained
in a stable and homogeneous state for the duration of the dis-
tribution and trial period.
The test materials were shipped to the participants by cou-
rier in polystyrene insulated boxes. An ice pack was included
to protect samples from high temperatures during transport.
Availability of media needed for enumeration was the respon-
sibility of the laboratories. Sample codes were randomly se-
lected for each laboratory to ensure confidentiality of each
participant’s results. The examination of the samples had to
take place within 1 month. A validation report was distributed
to all participating laboratories after the trial. It contained the
homogeneity data of the test materials; all raw data as received
by the organizing laboratory, including any supplementary in-
formation in the test reports; and the statistical analysis with
the method performance data and recommendation of
one method for Official Control purposes.
Methodology for Analysis of Samples
The procedure for enumerating bacilli in feedstuffs in-
volved the following steps: reconstitution, pretreatment, enu-
meration, incubation, confirmation, and result presentation.
(a) Reconstitution of test material and preparation of ini-
tial suspensions.—Three different categories (a–c) of feed-
stuffs are currently available commercially. These are (a) ad-
ditives containing 1010 colony-forming units (CFU)/g, and
(b) premixtures containing 108 CFU/g. The third category
(c) are feeds, meal or pellets, containing 106 CFU/g, compris-
ing complete feedstuffs, mineral feedstuffs, and milk replac-
ers. For enumeration analysis, a 20 g sample was taken for the
first 2 categories (a, b) and a 50 g sample for the third (c).
(b) Pretreatments.—Two different pretreatments were in-
vestigated. For pretreatment A, the sample was suspended in
50% ethanol and shaken for 1 h at room temperature. Decimal
dilutions were made from the initial suspension in peptone salt
diluent. For pretreatment B, additives (a) were homogenized
in phosphate-buffered saline (PBS) and premixtures (b) or
570 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003

Table 2. Interlaboratory study results for bacilli spores (CFU/g) in feedstuffs using method A with ethanol treatment
to inactivate vegetative background floraa,b

Laboratory Sample type I Sample type I Sample type J Sample type J Sample type K Sample type K

2 8.3 E+5 9.5 E+5 1.6 E+9 5.4 E+9 4.1 E+6 2.0 E+4
3 3.8 E+5 5.1 E+5 2.1 E+9 2.9 E+9 <1 E+4 <1 E+4
4 7.5 E+5 8.5 E+5 3.0 E+13 3.8 E+12 1.5 E+4 <1 E+3
5 <1 E+4 <1 E+4 1.5 E+9 6.1 E+8 <1 E+4 <1 E+4
6 8.0 E+4 3.3 E+4 7.4 E+8 4.8 E+8 <1 E+3 <1 E+3
9 5.6 E+5 1.5 E+6 6.9 E+8 1.8 E+9 1.2 E+3 9.0 E+2
11 2.9 E+5 3.5 E+5 1.0 E+9 4.2 E+8 <1 E+6 <1 E+6
12 3.2 E+5 3.2 E+5 4.7 E+9 4.2 E+9 <1 E+4 <1 E+4
16 1.0 E+5 2.5 E+5 1.8 E+9 1.4 E+9 <1 E+5 <1 E+5
17 9.5 E+6 2.9 E+7 1.6 E+9 8.0 E+9 1.3 E+8 6.4 E+7
18 5.1 E+5 5.1 E+5 1.0 E+9 1.0 E+9 <1 E+3 <1 E+3
20 7.5 E+4 7.6 E+4 8.4 E+7 9.0 E+7 5.0 E+2 3.0 E+2
22 2.2 E+5 1.2 E+5 1.9 E+9 2.6 E+9 <1 E+4 <1 E+4
23 5.0 E+5 4.4 E+5 5.1 E+9 1.2 E+10 3.0 E+9 8.2 E+7

a
Laboratories 1, 10, and 13 were unable to perform the analysis because of financial and time constraints. Laboratories 7, 8, 14, 15, 19, and
21 did not participate in this trial.
b
E+2 = 102, E+3 = 103, E+4 = 104, E+5 = 105, E+6 = 106, E+7 = 107, E+8 = 108, E+9 = 109, E+10 = 1010, E+11 = 1011, E+12 = 1012, E+13 =
1013 .

feeds (c) in sodium hydroxide diluent in a Waring blender. Ad-
ditives and premixtures were blended for 3 min at high speed.
Feeds (c) were blended for 1 min at high speed (starting with
low speed and turned to high speed to avoid splashing); they
were left to stand for 30 min at 5°C and then blended for another
2 min at high speed. Immediately after blending, 1 mL of the
suspension was transferred into 9 mL peptone salt diluent and
heated in a water bath at 80 ± 1°C for 10 min. The tube was agi-
tated manually for 2 min in a water bath at 10°C to cool the
sample to room temperature, and then further decimally diluted.
Where appropriate, the existing standard for preparation of ini-
tial suspension and decimal dilutions was followed (16).
(c) Enumeration.—Duplicate TSA plates were inoculated
with 0.1 mL each of appropriate decimal dilutions.
(d) Incubation.—TSA plates were incubated at 37 ± 1°C
for 16–24 h.
(e) Confirmation.—To confirm presumptive colonies, a
wet preparation of a colony from the agar plate was examined
by phase-contrast light microscopy for the presence of
phase-bright spores. The species of bacilli can be confirmed
by an API 50 CHB test.
(f) Result presentation.—Typical colonies and results were
recorded in the test report as CFU/g feedstuffs. The data were
reported to the organizing laboratory for statistical analysis.
Preparation of Samples
Samples for the trial consisted of feed and premixtures
which contained single microorganisms or mixtures. All sam-
ples were numbered chronologically. Three different types of
samples (I–K) were prepared. They contained bacilli spores
only, mixtures of bacilli spores and other vegetative probiotic
microorganisms, or only vegetative probiotic microorganisms
without bacilli spores (blanks).
Samples of type I consisted of 50 g feed sample containing
spores of Bacillus subtilis and B. licheniformis (Chr. Hansen
Biosystems, Hørsholm, Denmark). The product was milled
for homogeneity according to the following protocol: 3 kg
were milled with a Retsch Mill (Glen Creston Ltd., Stanmore,
UK; Ultracentrifugal mill type ZM100) and sieved with a
1.5 mesh sieve. Samples of type J contained bacilli spores
(B. subtilis and B. licheniformis) for enumeration and
probiotic pediococci and yeast as background contaminants.
The sample weighed 20 g and contained 17.5 g Provita
Biogrow (Provita, Omagh, N. Ireland), 1.5 g Dengie Yea-Sacc
by Alltech (Dengie Crops Ltd., Southminster, UK), and 1 g
Bactocell pig (Wyreside Products Ltd., Preston, UK). Sam-
ples of type K were blanks, containing no spores but only veg-
etative cells of probiotic lactobacilli, pediococci, yeast, and
enterococci. The sample weighed 20 g and was composed of
17.5 g Probiotics International Protexin Supermix for horses
(Probiotics Int. Ltd., Somerset, UK), 1 g Dengie Yea-Sacc, 1 g
Bactocell pig, and 1 Quest Non-Dairy Acidophilus Plus cap-
sule (Quest Vitamins Ltd., Birmingham, UK). The spore con-
centrations in all sample types were analyzed by methods A
and B, and the vegetative cell concentration was further quan-
tified by using MRS agar for lactic acid bacteria and
chloramphenicol glucose yeast extract (CGYE) agar for yeast.
All constituents were weighed on a 4-decimal-place bal-
ance, corrected to 2 decimal places, and placed into sterile
plastic containers with metal screw-cap lids. Containers of
 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003 571

Table 3. Interlaboratory study results for bacilli spores (CFU/g) in feedstuffs using method B with a heat treatment at
80°C for 10 min to inactivate vegetative background floraa,b

Laboratory Sample type I Sample type I Sample type J Sample type J Sample type K Sample type K

2 9.2 E+5 1.7 E+6 2.6 E+9 1.9 E+9 <1 E+4 <1 E+4
3 9.6 E+5 9.2 E+5 4.1 E+9 6.3 E+9 <1 E+4 <1 E+4
4 1.3 E+6 1.3 E+6 2.1 E+12 >1 E+10 <1 E+3 6 E+5
5 <1 E+4 1.5 E+6 6.9 E+9 5.7 E+9 <1 E+4 <1 E+4
6 1.1 E+6 1.3 E+6 3.4 E+9 3.8 E+9 2.0 E+3 <1 E+3
9 1.1 E+6 1.2 E+6 2.4 E+9 2.1 E+9 3.2 E+3 1 E+3
11 1.7 E+6 1.6 E+6 3.7 E+9 3.1 E+9 <1 E+6 <1 E+6
12 1.6 E+6 1.3 E+6 5.4 E+9 9.1 E+9 <1 E+4 <1 E+4
16 1.5 E+6 1.3 E+6 6.8 E+9 5.9 E+9 1.4 E+5 1.0 E+4
17 4.8 E+8 2.5 E+7 1.0 E+10 2.9 E+9 1.8 E+9 3.0 E+8
18 1.6 E+5 1.6 E+5 3.4 E+9 3.3 E+9 1.2 E+4 1.5 E+4
20 2.1 E+5 2.4 E+5 4.2 E+8 7.6 E+8 3.0 E+3 2.0 E+3
22 1.2 E+6 1.3 E+6 3.1 E+9 3.3 E+9 1.4 E+4 2.4 E+4
23 9.8 E+7 7.2 E+6 6.9 E+9 8.5 E+9 5.0 E+10 3.0 E+6

a
Laboratories 1, 10, and 13 were unable to perform the analysis because of financial and time constraints. Laboratories 7, 8, 14, 15, 19, and
21 did not participate in this trial.
b
E+2 = 102, E+3 = 103, E+4 = 104, E+5 = 105, E+6 = 106, E+7 = 107, E+8 = 108, E+9 = 109, E+10 = 1010, E+11 = 1011, E+12 = 1012.

100 or 60 mL were used, depending on sample weight. The
lids were closed tightly and the containers were stored in air-
tight plastic bags at 4°C until dispatch. Materials used to vali-
date the methods were tested extensively before the pretrial
and the interlaboratory study to ensure their homogeneity. Ho-
mogeneity of individual sample constituents was tested on the
minimum amount to be used in sample mixtures and on
10 randomly selected samples (total sample number of each
type was 70) in duplicate by both methods. For experimental
analysis, the procedure was followed as outlined in the SOP.
Principle of Methods A and B
The principle of the methods is the inactivation of all vege-
tative background flora by treatment in 50% ethanol for 1 h
(method A) or by heat treatment for 10 min at 80°C
(method B). For both methods, bacilli spores that are resistant
to both pretreatments are enumerated on TSA. TSA is
nonselective and requires a selective pretreatment for bacilli
spores in feeds.
Apparatus and Reagents
(a) Incubator.—Capable of maintaining incubation tem-
perature of 37 ± 1°C.
(b) Water bath.—Capable of maintaining temperature of
48 ± 1°C, 80 ± 1°C, and 10 ± 1°C.
(c) Balance.—Capable of weighing to the nearest 0.01 g.
(d) Rotary shaker.—Operating at 100 rpm.
(e) Erlenmeyer flasks or screw-cap bottles.—Sterilizable
bottles or Erlenmeyer flasks of appropriate volume sealed
with screw caps or cotton caps, respectively.
(f) Sterile glass containers.—Screw-cap bottles of appro-
priate capacities, 25 mL universals, bottles, flasks, and
1000 mL Duran bottles.
(g) Blending equipment.—Two-speed Waring blender
(18 000 and 22 000 rpm) with 1 L bowls sterilized in oven for
1 h at 170–180°C, or equivalent.
(h) Graduated pipets.—Nominal capacity 1, 0.2, and
0.1 mL.
(i) Spatulas.—Sterile, stainless steel.
(j) Sterile Petri dishes.—Diameter 90 mm, and triple vent.
(k) Test tubes.—Sterile test tubes for decimal dilutions.
(l) Mechanical stirrer.—Vortex mixer.
(m) Sterile spreaders.—Glass or disposable plastic.
(n) Ethanol solution.—Solution of 50% (v/v) ethanol pre-
pared by mixing 110 mL 96% ethanol and 100 mL sterile dis-
tilled water. The solution was filter-sterilized by using a bottle
top filter with 0.2 mm membrane.
(o) Peptone salt diluent.—Composed of 1 g/L peptone
and 8.5 g/L NaCl, Polysorbate (Tween 80) 0.01% (v/v), pH
7.0 ± 0.2, at 25 ± 1°C. Autoclaved at 121 ± 1°C for 15 min.
(p) TSA.—Containing 15 g tryptone, 5.0 g NaCl, 5.0 g
soya peptone, 15 g agar in 1 L distilled water, with final pH 7.3
± 0.2. Autoclaved at 121 ± 1°C for 15 min.
(q) PBS.—Containing 8.0 g NaCl, 0.2 g KCl, 1.15 g
Na2HPO4, 0.2 g KH2PO4, pH 7.3 ± 0.2, in 1 L distilled water.
PBS was autoclaved at 115 ± 1°C for 10 min.
(r) Sodium hydroxide diluent.—Containing 2.0 g NaOH
in 1 L distilled water. Autoclaved at 121 ± 1°C for 15 min.
(s) API 50 CHB identification strips.—Commercially
available from BioMérieux (Marcy-l’Etoile, France).
572 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003

Table 4. Statistical analysis of raw data (log10 CFU/g) of bacilli spores in feedstuffs using method A with ethanol
treatment to inactivate vegetative background floraa

Laboratory Sample type Ib Outlier Sample type Jb Outlier Sample type Kb Outlier

2 5.92 5.98 9.21 9.73 6.61 4.30 C

3 5.58 5.71 9.32 9.46 4.00 4.00
4 5.88 5.93 13.48 12.57 C 4.18 3.00 C
c c
5 — — NC 9.18 8.79 4.00 4.00
6 4.90 4.52 8.87 8.68 3.00 3.00
9 5.75 6.18 8.84 9.26 3.08 2.95
11 5.46 5.54 9.00 8.62 6.00 6.00
12 5.50 5.50 9.67 9.62 4.00 4.00
16 5.00 5.40 9.26 9.15 5.00 5.00
17 6.98 7.46 GS 9.20 9.90 8.11 7.81
18 5.71 5.71 9.00 9.00 3.00 3.00
20 4.88 4.88 7.92 7.95 2.70 2.48
22 5.34 5.08 9.28 9.41 4.00 4.00
23 5.70 5.64 9.71 10.08 9.48 7.91 C

a
Laboratories 1, 10, and 13 were unable to perform the analysis because of financial and time constraints. NC = noncompliant; C = outlier
according to Cochran’s test; GS = outlier according to single Grubbs test.
b
Samples of type I: lower concentration; samples of type J: higher concentration; samples of type K: contained no spores, only vegetative
background flora. For statistical analysis of sample type K, “less than” was replaced with the given value.
c
These data were reported as a “less than” value and, hence, could not be converted to a log10 value. The term ‘NC’ means that data were
excluded before any statistical evaluation.

Statistical Analysis
The homogeneity of test samples was determined from re-
sults of 10 randomly selected samples, each enumerated in du-
plicate for each method. The results of duplicate counts per
sample were analyzed for variation between duplicate counts
(T1-test) and between the average of duplicate counts between
samples. Homogeneity was determined by an F-test
(a = 0.05) on the counts. The arithmetric means, T-1, T-2, and
standard deviations (SDs) were calculated from the counts to
determine homogeneity (13, 17).
For analysis of trial results, repeatability (r) and
reproducibility (R) were used because of their wide accep-
tance for analysis of data from interlaboratory studies in mi-
crobiology (13, 14, 18, 19). Repeatability provides a measure
of variability between analyses conducted by the same techni-
cian in the same laboratory under conditions as similar as pos-
sible. Reproducibility measures variability when analyses are
conducted by different technicians in different laboratories.
The concepts were originally related to chemical analysis, but
are equally applicable to microbiological data, provided the
counts are transformed to logarithms before analysis.
Horwitz (14) uses Grubbs and Cochran tests for outlier data
analysis (20, 21). The analysis allows conclusions about the
extent that results can be repeated or reproduced in the same or
different laboratories.
Samples of type K were evaluated by creating data sets for
each method that replaced any value “less than” with a given
limit. This enabled a log10 conversion of the original counts
and an analysis of the data with the same statistical methods as
those of samples of types I and J.
Results and Discussion
All test materials were sufficiently homogeneous for the
purpose of the trial (Table 1). The counts were stable for
3 months when stored in a dark and dry place at 4°C. All sam-
ples had to be examined within a 2 months period.
Fourteen out of 17 laboratories finalized the trial and re-
turned a test report. Three laboratories were unable to perform
the analysis after receiving samples because of time and finan-
cial constraints, and did not return a test report. All raw data
are represented for methods A and B in Tables 2 and 3, respec-
tively. The enumeration results of the organizing laboratory
were, for samples of type I, a spore concentration of 5.4 ´
105 CFU/g as determined by method A (50% ethanol treat-
ment) and 1.2 ´ 106 CFU/g by method B (heat treatment). In
samples of type J, the spore concentration was 1.8 ´
109 CFU/g by method A and 6.2 ´ 109 by method B;
pediococci were present at 8 ´ 109 CFU/g and yeast at 6.4 ´
106 CFU/g. In samples of type K, no spores were present;
lactobacilli were present at 1.1 ´ 108 CFU/g, pediococci at 8 ´
109 CFU/g, enterococci at 4.5 ´ 107 CFU/g, and yeast at 4.3 ´
106 CFU/g.
Seventeen parcels were sent by courier and were received
by 9 laboratories on the following day. One laboratory re-
ceived it a day later, one after 3 days, one after 8 days, and
 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003 573

Table 5. Statistical analysis of raw data analysis (log10 CFU/g) of bacilli spores in feedstuffs using method B with
heat treatment at 80°C for 10 min to inactivate vegetative background floraa

Laboratory Sample type Ib Outlier Sample type Jb Outlier Sample type Kb Outlier

2 5.96 6.24 9.41 9.28 4.00 4.00

3 5.98 5.96 9.61 9.80 4.00 4.00
4 6.11 6.11 12.3 —c NC 3.00 5.78 C
c
5 — 6.18 NC 9.84 9.76 4.00 4.00
6 6.04 6.11 9.53 9.58 3.30 3.00
9 6.04 6.08 9.38 9.32 3.51 3.00
11 6.23 6.20 9.57 9.49 6.00 6.00
12 6.20 6.11 9.73 9.96 4.00 4.00
16 6.18 6.11 9.83 9.77 5.15 4.00
17 8.68 7.40 GD 10.00 9.46 C 9.26 8.48 GS
18 5.20 5.20 9.53 9.52 4.08 4.18
20 5.32 5.38 8.62 8.88 3.48 3.30
22 6.08 6.11 9.49 9.52 4.15 4.38
23 7.99 6.86 GD 9.84 9.93 10.70 6.48 C

a
Laboratories 1, 10, and 13 were unable to perform the analysis because of financial and time constraints. NC = noncompliant; C = outlier
according to Cochran’s test; GD = outlier according to double Grubbs test; GS = outlier according to single Grubbs test.
b
Samples of type I: lower concentration; samples of type J: higher concentration; samples of type K: contained only vegetative background
flora. For statistical analysis of sample type K, “less than” was replaced with the given value.
c
These data were reported as a “less than” value and, hence, could not be converted to a log10 value. The term ‘NC’ means that data were
excluded before any statistical evaluation.

2 laboratories did not record the arrival of the parcel in their
test reports. Thirteen laboratories reported that samples were
still cool on arrival and one did not report the temperature.
Each participant chose whether to use commercially avail-
able dehydrated media or media prepared from individual in-
gredients. If the latter was chosen, all ingredients and concen-
trations used were to be recorded in the test report. All
participating laboratories used commercially available TSA.
Eight laboratories prepared peptone salt diluent from individ-
ual ingredients and 2 laboratories prepared PBS. Each labora-
tory used the type of pipets it normally used. Decimal dilu-
tions were prepared with micropipets (4 laboratories),
graduated disposable pipets (5 laboratories), or graduated
glass pipets (4 laboratories); one laboratory did not report
what it used. For inoculation of agar plates, micropipets were
used by 10 laboratories, one used graduated disposable pipets,
2 used graduated glass pipets, and one laboratory provided no
information. Deviations from the protocol were not reported
by any participants.
All raw data from the trial were transformed to log10 counts
and were statistically evaluated (Tables 4 and 5). Noncompli-
ance and outliers were identified and excluded from the calcu-
lation of the repeatability (r) and reproducibility (R) limits of
the method. Bacilli spores were present in sample type I in a
low concentration, and in type J in a high concentration. They
were absent in samples of type K. Participants returned
12 valid data sets for method A and 11 for method B. For the
blank samples of type C, 11 valid data sets were returned for
both methods. To calculate the efficiency of both methods for
reducing vegetative background flora, “less than” values were
replaced with the given values for statistical analysis.
The statistical precision parameters of the 2 methods re-
vealed a repeatability limit of 2.9% (low concentration) and
2.5% (high concentration) for method A, and 1.1 and 1.0%, re-
spectively, for method B. The reproducibility limit for
method A was 7.8% for the low level and 5.9% for the high
concentration; corresponding values were 5.8 and 3.4% for
method B. The statistical results of the trial included the mean
of valid results, SD, RSDr and RSDR values, and limits, respec-
tively (Table 6). Statistical analysis of the blank samples of
type K showed that the mean value was slightly lower for
method B. This method had a better reproducibility than
method A. The mean spore counts were higher for method B
than for A, indicating that method B gave higher viable spore
counts and reduced vegetative background flora. The validation
study confirmed a very good reproducibility limit for the
method proposed for bacilli enumeration. The RSDR was <8%,
compared to an RSDR for standard plate counts of 21% (22).
Precision data for enumeration of B. cereus in food and for
the enumeration of probiotic enterococci in feed were compa-
rable with those reported in our present study (13, 23).
Method B performed better than method A. The heat treat-
ment in method B was preferred by laboratories that expressed
concerns about routine handling of large quantities of ethanol
as required for method A. The heat treatment in method B ac-
tivated dormant spores and therefore resulted in higher viable
574 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003

Table 6. Statistical precision of the methods for enumeration of bacilli spores in feedstuffs using methods A and B

Method A Method A Method A Method B Method B Method B

Parametera Sample type Ib Sample type Jb Sample type Kb Sample type Ib Sample type Jb Sample type Kb

Mean 5.49 9.17 4.23 5.95 9.53 4.07

n 17 17 17 17 17 17
nc 4 3 3 4 4 3
Outliers 1 1 3 2 1 3
n1 12 12 11 11 11 11
r 0.44 0.65 0.24 0.19 0.26 0.79
sr 0.16 0.23 0.09 0.07 0.09 0.28
RSDr 2.86 2.54 2.01 1.13 0.99 6.94
R 1.19 1.51 4.43 0.97 0.91 2.24
sR 0.43 0.54 1.58 0.35 0.32 0.80
RSDR 7.75 5.89 37.41 5.80 3.40 19.68

a
Mean = overall mean of valid results (log10 CFU/g); n = number of participating laboratories; nc = noncompliant; outliers = number of
laboratories where data were recognized as Grubbs or Cochran outlier; n1 = number of laboratories with valid results; r = repeatability limit (r
= 2.8 ´ sr); sr = standard deviation of repeatability; RSDr = relative standard deviation of repeatability; R = reproducibility limit (R = 2.8 ´ sR); sR
= standard deviation of reproducibility; RSDR = relative standard deviation of reproducibility.
b
Samples of type I: lower concentration; samples of type J: higher concentration; samples of type K: contained only vegetative background
flora.

counts. Because bacilli spores are not reported to be sensitive
to heat treatment, as are some spores of C. botulinum, an alter-
native ethanol treatment is not needed (12). Method B will be
recommended by the project partners for acceptance for Offi-
cial Control of probiotic bacilli used as feed additives.
Recommendation
A method to enumerate probiotic bacilli spores in feed-
stuffs, including heat treatment of the initial suspension of a
feedstuff sample which is already routinely applied by Indus-
trial and Control Laboratories, is recommended for Official
Control purposes.
Acknowledgments
We thank Merete Mørk Jensen (Chr. Hansen Biosystems,
Hørsholm, Denmark) for providing the analysis protocol of
method B, Ken Mathieson (Central Science Laboratory, York,
UK) for expert statistical analysis of the experimental data,
and the European Federation of Animal Feed Additive Manu-
facturers (FEFANA, Brussels, Belgium) for their collabora-
tion and support.
We also thank the following collaborators for their partici-
pation in the validation trial:
A. Adler, Federal Office of Agrobiology, Linz, Austria
E. Auclair, Lesaffre International, Lille, France
G. Bertin, Alltech, Goussainville, France
M. Braconnier, Laboratoires ASTA, Ettelbück, Luxembourg
K. Domig and W. Kneifel, Institut für Milchforschung und
Bakteriologie, Universität für Bodenkultur, Vienna, Austria
P. Jones, Interprise Ltd., West Glamorgan, UK
M. Krause, Danish Plant Directorate, Lyngby, Denmark
S. Marmo, Plant Production Inspection Centre, Agricul-
tural Chemistry Department, Vantaa, Finland
J. Michard, Laboratoire DGCCRF, Rennes, France
H. Mietke, Sächsische Landesanstalt für Landwirtschaft
FB, Leipzig, Germany
A. O’Briain, State Laboratory, Dublin, Ireland
A.S. Olofson, National Veterinary Institute (SVA),
Uppsala, Sweden
W. Ruppitsch, Federal Office and Research Center for Ag-
riculture, Vienna, Austria
A. Thalmann, Landwirtschaftliche Untersuchungs-und
Forschungsanstalt (LUFA), Kahlsruhe, Germany
A. Voets, Ministry of Agriculture,
Rijksontledingslaboratorium, Tervuren, Belgium
C. Warmerdam, Rikilt-DLO State Institute for Quality Con-
trol of Agricultural Products, Wageningen, The Netherlands
This validation study was financially supported by the Eu-
ropean Commission in the framework of a Standards, Mea-
surement and Testing project (SMT4CT98-2235) and by the
Animal Feed Division of the Food Standards Agency (project
code F01001).
References
(1) Simon, O., Jadamus, A., & Vahjen, W. (2001) J. Anim. Feed
Sci. 10, 51–67
(2) Jadamus, A., Vahjen, W., Schafer, K., & Simon, O. (2002) J.
Anim. Physiol. Anim. Nutr. 86, 42–54
(3) Rengpipat, S., Rukpratanporn, S., Piyatiratitivorakul, S., &
Menasaveta, P. (2000) Aquaculture 191, 271–288
 LEUSCHNER ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 3, 2003 575
(4) Fritts, C.A., Kersey, J.H., Motl, M.A., Kroger, E.C., Yan, F.,
Si, J., Jiang, Q., Campos, M.M., Waldroup, A.L., &
Waldroup, P.W. (2000) J. Appl. Poult. Res. 9, 149–155
(5) Kumprecht, I., & Zobac, P. (1998) Czech J. Anim. Sci. 43,
327–335
(6) Kreuzer, M. (1994) Agribiol. Res. – Z. Agrarbiol.
Agrikulturchem. Okolog. 47, 13–23
(7) Adami, A., Sandrucci, A., & Cavazzoni, V. (1997) Ann.
Microbiol. Enzimol. 47, 139–149
(8) Roth, F.X., Kirchgesser, M., Eidelsburger, U., & Gedek, B.
(1992) Agribiol. Res.–Z. Agrarbiol. Agrikulturchem. Okolog.
45, 294–302
(9) Gedek, B., Roth, F.X., Wiehler, S., Bott, A., Eidelsburger, U.,
& Kirchgesser, M. (1992) Agribiol. Res.–Z. Agrarbiol.
Agrikulturchem. Okolog. 45, 311–320
(10) BSI (1986) BS 4285-3.3., British Standards Institution, Lon-
don, UK
(11) Michard, J., & Levesque, A. (1989) Bul. d’Inf. Station
d’Aviculture de Ploufragan 29, 146–151
(12) Broda, D.M., De Lacy, K.M., & Bell, R.G. (1998) Int. J.
Food Microbiol. 39, 61–68
(13) Schluten, S.M., in’t Veld, P.H., Nagelkerke, N.J.D., Scotter,
S., de Buyser, M.L., Rollier, P., & Lahellec, C. (2000) Int. J.
Food Microbiol. 57, 53–61
(14) Horwitz, W. (1995) Pure Appl. Chem. 76, 331–343
(15) BSI (1993) BS ISO 3534-1, British Standards Institution,
London, UK
(16) BSI (1999) BS EN ISO 6887-1, British Standards Institution,
London, UK
(17) Heisterkamp, S.H., Hoekstra, J.A., van Strijp-Lockefeer,
N.G.W.M., Havelaar, A.H., Mooijman, K.A., in’t Veld, P.H.,
Notermans, S.H.W., Mayer, E.A., & Griepink, B. (1993) Re-
port No. EUR 15008 EN, Brussels, Belgium
(18) IDF Standard 135B (1991) International Dairy Federation,
Brussels, Belgium
(19) Szita, G., Tabajdi, V., Fabian, A., Biro, G., Reichart, O., &
Kormoczy, P.S. (1998) Int. J. Food Microbiol. 43, 123–127
(20) Grubbs, F.E. (1969) Technometrics 11, 1–21
(21) Snedecor, G.W., & Cochran, W.G. (1980) Statistical
Methods, 7th Ed., Iowa State University Press, Ames, IA
(22) Richardson, G.H., Yuan, J.T.C., Sisson, D.V., & Stokes, B.O.
(1994) J. AOAC Int. 77, 623–627
(23) Leuschner, R.G.K., Bew, J., Domig, K., & Kneifel, W.
(2002) J. Appl. Microbiol. 93, 781–786