Version of Record: https://www.sciencedirect.

com/science/article/pii/S0168160518301259
Manuscript_ba54c8c23441f663e05ecc22c9982079

2 Title:

3 Validation of EN ISO 6579-1 - Microbiology of the food chain - Horizontal method for

4 the detection, enumeration and serotyping of Salmonella - Part 1 Detection of

5 Salmonella spp.

8 Authors:

9 Kirsten A. Mooijmana,*, Annemarie Pielaata,1, Angelina F.A. Kuijpersa

10
a
11 National Institute for Public Health and the Environment (RIVM), Centre for Zoonoses and

12 Environmental Microbiology (Z&O), EURL-Salmonella, P.O. Box 1, 3720 BA Bilthoven, the

13 Netherlands

14

15
*
16 Corresponding author, email: [email protected]

17
1
18 Current address: Unilever Research and Development, Olivier van Noortlaan 120, 3133 AT

19 Vlaardingen, the Netherlands

20

21

22

23

24 Keywords: EN ISO 6579-1; Salmonella; detection method; food chain; validation;

25 performance characteristics

26

27

28

1
© 2018 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
29 Abstract

30 The European and International Standard method for the detection of Salmonella spp. in

31 samples from the primary production stage, EN ISO 6579:2002/Amd.1:2007, was validated

32 by an interlaboratory study in the frame of Mandate M/381, ordered by the European

33 Commission and accepted by the European Standardisation Organisation (CEN). In addition

34 to this study, results from two interlaboratory studies organised earlier by the European

35 Union Reference Laboratory (EURL) for Salmonella were used for determination of the

36 performance characteristics. Parallel to the performance evaluation for the Mandate, the

37 revision of EN ISO 6579:2002 started. Part of this revision was the incorporation of the

38 standardised method for detection of Salmonella in samples from the primary production

39 stage (EN ISO 6579:2002/Amd.1:2007) and its performance characteristics in the new part 1

40 of EN ISO 6579. The 2002 version of EN ISO 6579 already contained performance

41 characteristics for the detection of Salmonella in food samples, but LOD50 values

42 (contamination level at which 50% of the samples are found positive) were not yet included.

43 To be in line with the performance characteristics determined for detection of Salmonella

44 spp. in samples from the primary production stage, LOD50 values for detection of Salmonella

45 in food samples were calculated from the raw data of the validation studies performed in

46 2000. In this paper, the performance characteristics of EN ISO 6579-1:2017 are determined

47 based not only on the results of the interlaboratory study carried out in 2013 under the

48 Mandate, but also on several other interlaboratory studies. These performance

49 characteristics consist of specificity, sensitivity and LOD50.

50

51

52 1. Introduction

53 At International and at regional level in Europe, respectively ISO and CEN are active in

54 developing harmonised methods. For microbiology of the food chain several harmonised (EN

55 ISO) standard reference methods exist, which are important tools for the internationally

56 uniform analysis of (food) samples. The majority of the microbiological EN ISO methods

2
57 concern ‘traditional’ culture methods, which are considered as the reference methods. Any

58 new/alternative method has to be validated against the reference method to test if this new

59 method performs at least equally well as the reference method. To know how a reference or

60 alternative method performs, it is important to have information on the performance

61 characteristics. However, up to 2010, the majority of microbiological EN ISO methods have

62 been published without performance characteristics. To overcome this omission, the

63 European Commission (EC) issued Mandate M/381 in 2010, which was accepted by the

64 European Standardisation Organisation (CEN), to determine performance characteristics of

65 15 microbiological standard reference methods related to food hygiene legislation. For this a

66 harmonised experimental design was drafted, based on EN ISO 16140 (Anonymous, 2003a)

67 and its revised version, EN ISO 16140-2 (Anonymous, 2016b), which was still a draft in

68 2010. For the determination of the performance characteristics of each of the 15 reference

69 methods, one or more interlaboratory studies (ILS) had to be organised testing one (for so

70 called ‘vertical methods’) to five (for ‘horizontal’ methods) different matrices. To obtain

71 sufficient data for statistical analysis, at least 10 (for quantitative methods) or 13 (for

72 qualitative methods) collaborators were needed for each ILS.

73 Part of Mandate M/381 was the determination of the performance characteristics for

74 EN ISO 6579:2002/Amd.1: 2007 for ‘Detection of Salmonella spp. in animal faeces and in

75 environmental samples from the primary production stage’ (Anonymous, 2007). As this

76 concerned a so-called ‘vertical’ method, only one matrix (samples from the primary

77 production stage) needed to be analysed in an interlaboratory study.

78 To judge whether the performance characteristics determined from the results of the ILS

79 performed under the Mandate were representative for samples from the primary production

80 stage, they were compared to the performance characteristics derived from interlaboratory

81 studies organised earlier. This concerned two interlaboratory studies organised by the

82 European Union Reference Laboratory (EURL) for Salmonella in 2008 and 2012 (Kuijpers et

83 al., 2008; Kuijpers and Mooijman, 2013). The set-up of these earlier studies deviated from

84 the set-up of the ILS as agreed upon for the Mandate with respect to the number of samples

3
 85 tested. For the study under the Mandate, each participant had to test 8 samples for each

86 contamination level (blank, low and high). In the earlier studies each participant tested (only)

87 5 samples for each contamination level. On the other hand, in these earlier studies two sets

88 of low and high contaminated samples were included, artificially contaminated with two

89 different Salmonella serovars and thus resulting in 10 low and 10 high contaminated samples

90 analysed by each participant. Additionally, the total number of participants in the earlier

91 EURL studies was (much) larger than the prescribed number for the study of the Mandate

92 (approximately 30 participants in the earlier EURL studies, against 13 participants prescribed

93 for the ILS of the Mandate), resulting in a sufficiently large dataset to determine the

94 performance characteristics.

95 Parallel to the performance evaluation for the Mandate, the revision of EN ISO 6579:2002

96 started. Part of this revision was the incorporation of the standardised method for detection of

97 Salmonella in samples from the primary production stage (EN ISO 6579:2002/Amd.1:2007)

98 and its performance characteristics in the new part 1 of EN ISO 6579 (Anonymous, 2017;

99 Mooijman in press). The 2002 version of EN ISO 6579 (Anonymous, 2002) already contained

100 performance characteristics for the detection of Salmonella in food samples. However, these

101 consisted of specificity, sensitivity, accordance and concordance, while under the Mandate

102 M/381 accordance and concordance were replaced by LOD50 (contamination level at which

103 50% of the samples are found positive; Anonymous 2016a). Although not part of the

104 Mandate, but still part of the revision of EN ISO 6579, an attempt was made to calculate the

105 LOD50 values for detection of Salmonella in food samples from the raw data of the original

106 validation studies performed in 2000 for determining the performance characteristics of the

107 2002 version of EN ISO 6579 (Feldsine et al., 2003).

108 In this paper, the performance characteristics of EN ISO 6579-1:2017 are determined

109 based not only on the results of the interlaboratory study carried out in 2013 under the

110 Mandate, but also on several other interlaboratory studies. These performance

111 characteristics consist of specificity, sensitivity and LOD50.

112

4
113

114 2. Materials and methods

115

116 2.1 Design of the studies

117 Additional to the interlaboratory study (ILS) organised under the Mandate in 2013, data

118 from interlaboratory studies (ILSs) organised by the EURL-Salmonella in 2008 and 2012

119 were used. In all studies, National Reference Laboratories (NRLs) for Salmonella of the (by

120 then) 27 different EU Member States participated, as well as some additional NRLs from

121 non-EU Member States. The total number of participants per study is shown in Table 1. All

122 laboratories were accredited or in the process of being accredited according to EN ISO/IEC

123 17025 (Anonymous, 2005).

124 The design of each ILS was described in a detailed protocol, a Standard Operating

125 Procedure (SOP) and a test report in which results were recorded and returned to the

126 organiser of the study for analysis. In the test report, participants were requested to detail all

127 supplementary information that could have influenced their results. The documents of the

128 three interlaboratory studies described here, are available at the EURL-Salmonella website

129 (EURL-Salmonella, 2017).

130 The number and type of samples, as well as the matrices varied per ILS. Details per

131 study are given in Table 1. In each study ‘real’ matrix samples were used, artificially

132 contaminated with two different levels of a Salmonella serovar: low and high. Additionally,

133 blank matrix samples were used. For the low level samples, the contamination level should

134 be at or slightly above the detection limit of the method. The objective was to obtain 25% to

135 75% positive results when testing the samples with the low inoculation level (where the

136 negative results would be mostly being due to the absence of Salmonella in these samples).

137 The contamination level of the high level samples was aimed to be 5-10 times above the

138 detection limit of the method. For the ILS under the Mandate, the matrix samples were

139 artificially contaminated with a diluted culture, while for the earlier ILSs reference materials

140 were used to artificially contaminate the samples (for more details on the test materials, see

5
141 2.3). For the two earlier interlaboratory studies, the participating laboratories received the

142 matrix (chicken faeces or pig faeces) and reference materials separately packed. At the day

143 of analyses, the participants had to prepare the samples by combining 10 g chicken faeces

144 (ILS 2008) or 25 g pig faeces (ILS 2012) with one reference material according to a detailed

145 protocol.

146 For all studies, all samples (or reference materials) were individually packed and

147 randomly labelled and shipped to the participants by courier service. The samples were

148 packed in double containers with ice packs, and sent in accordance to the International Air

149 Transport Association Regulations as ‘Biological Substance category B’, classified as UN

150 3373 (Anonymous, 2013). To obtain more information on the temperature during transport,

151 each parcel contained a small electronic temperature device which measured and registered

152 the temperature every hour. After receipt of the samples, the electronic device had to be

153 returned to the EURL-Salmonella after which it was read with the computer. The samples

154 had to be stored at 5 °C (boot socks samples and faeces) or at -20 °C (reference materials),

155 until the day of analysis. Participants used their own media and reagents for each ILS,

156 following the prescriptions of EN ISO 6579:2002/Amd.1:2007 (Anonymous, 2007). The

157 examination of the test materials was initiated on the same day by all participants and started

158 one week (ILS 2013 and 2012) or two weeks (ILS 2008) after dispatch of the samples.

159

160 2.2 Method under collaborative trial

161 In all studies, EN ISO 6579:2002/Amd.1:2007 (Anonymous, 2007) was followed for the

162 detection of Salmonella in samples from the primary production stage. The method required

163 the following successive stages:

164 a) Preparation of the initial suspension, by adding each test portion to a quantity of non-

165 selective broth (Buffered Peptone Water; BPW) to yield a tenfold dilution.

166 b) Pre-enrichment of the (non-selective) initial suspension a) at 37 °C ± 1 °C for 18 h ± 2 h.

6
167 c) Selective enrichment of a culture obtained from b) on a selective semi-solid agar

168 medium, Modified semi-solid Rappaport-Vassiliadis (MSRV) agar, at 41.5 °C ± 1 °C for

169 24 h ± 3 h and if negative for an additional 24 h ± 3 h.

170 d) Plating-out of suspect cultures on MSRV agar plates obtained from c) on two selective

171 isolation agar media: Xylose Lysine Deoxycholate (XLD) agar and a second isolation

172 medium for choice. The XLD agar plates were incubated at 37 °C ± 1 °C for 24 h ± 3 h,

173 the second agar plates according to the manufacturer instructions.

174 e) Confirmation of at least 1 suspect colony obtained from d) by biochemical and

175 serological testing. If negative, 4 additionally suspect colonies had to be confirmed.

176

177 For calculation of the LOD50 values for detection of Salmonella spp. in food samples, raw

178 data were used of two validation studies performed in the frame of another EU project in

179 2000 (Feldsine et al., 2003). In these studies, EN ISO 6579:2002 was followed which

180 comprises of the same successive culture steps as described above, but with the use of two

181 selective enrichment broths instead of MSRV agar for step c): Rappaport Vassiliadis broth

182 with soya (RVS), incubated at 41.5 °C ± 1 °C for 24 h ± 3 h, and Muller-Kauffmann

183 tetrathionate-novobiocin (MKTTn) broth incubated at 37 °C ± 1 °C for 24 h ± 3 h.

184

185 2.3 Preparation of test materials

186

187 2.3.1 Test materials for interlaboratory study organised under the Mandate in 2013

188 For the interlaboratory study organised under the Mandate, boot sock samples were used.

189 Boot socks are frequently used for sampling of chicken stables. The socks are worn over the

190 boots and by walking through the stable, environmental material from the stable is collected.

191 To make sure that the environmental material sticks well, the socks are moistened before

192 use. Instead of walking through stables, the samples for the interlaboratory study were

193 prepared in a more controlled way to make sure that each sample contained the same

194 amount of ‘environmental material’.

7
195 Each sample consisted of one pair of boot socks in a plastic bag (Sodibox, Nevez,

196 France), moistened with 15 ml peptone saline solution (1.0 g enzymatic digest of casein, 8.5

197 g sodium chloride, in 1 L distilled water), to which 10 g chicken faeces from a pathogen-free

198 laying hen flock was added. The absence of Salmonella in the chicken faeces was previously

199 checked following EN ISO 6579:2002/Amd.1:2007 (Anonymous, 2007). The amount of

200 (natural) background flora in the chicken faeces was determined after receipt of the faeces,

201 as well as after storage at 5 °C, at the start of the interlaboratory study. For this, the total

202 bacterial count as well as the number of Enterobacteriaceae were determined, by following

203 respectively EN ISO 4833 (Anonymous, 2003b) and EN ISO 21528-2 (Anonymous, 2004).

204 The boot socks (with faeces) were artificially contaminated at the laboratory of the EURL-

205 Salmonella. For this, Salmonella Typhimurium (STM) ATCC 14028 (American Type Culture

206 Collection, USA) was cultured in BPW at 37 °C overnight. Next, tenfold dilutions were made

207 in peptone saline solution, and the number of colony forming units (cfu) per dilution were

208 counted by plating on XLD-agar and incubating the plates at 37 °C for 20-24 h. Meanwhile,

209 the dilutions were stored at 5 °C. Based on the plate counts, a sufficient amount of two (final)

210 dilutions were prepared from the stored dilutions and used to inoculate individual boot sock

211 samples with approximately 0.1 ml, resulting in samples with 5-10 cfu/pair of boot socks (low

212 level samples), or with 50-100 cfu/pair of boot socks (high level samples). For each level, the

213 boot sock samples were inoculated with the same volume from the same (mixed) dilution of

214 Salmonella Typhimurium. In this way, the variation in the contamination level between the

215 samples of one batch was kept as low as possible. The control boot sock samples (without

216 faeces), were inoculated in the same way. The concentration of the inoculums used to

217 contaminate the boot sock samples was confirmed by plating the relevant dilution on XLD

218 agar plates. A sub-set of the boot sock samples were not inoculated with Salmonella, to

219 become the blank samples.

220 A few days before dispatch of the samples to the participants of the interlaboratory study,

221 a total of 1200 boot sock samples were prepared and stored at 5 °C until the day of mailing.

222

8
223 2.3.2 Test materials for interlaboratory study organised by EURL-Salmonella in 2008

224 Five batches of reference materials were prepared, a few weeks before the interlaboratory

225 study. For this purpose milk, artificially contaminated with a Salmonella strain from the

226 National Institute for Public Health and the Environment (RIVM) culture collection, was spray-

227 dried (Veld et al., 1996). The obtained highly contaminated milk powder was mixed with

228 sterile (γ-irradiated) milk powder (Carnation, Nestlé, the Netherlands) to obtain the desired

229 contamination level. The mixed powder was filled in gelatine capsules resulting in the final

230 reference materials (RMs), with the following mean contamination levels (Kuijpers et al.,

231 2008):

232 • 5 and 44 cfu/capsule for Salmonella Typhimurium (STM5 and STM44): ALM40 (RIVM

233 Collection, the Netherlands);

234 7 and 91 cfu/capsule for Salmonella Enteritidis (SE7 and SE91): LBA88-8993 (RIVM

235 Collection, the Netherlands) ;

236 • Blank capsules containing no microorganisms.

237 The contamination levels of the capsules were determined following the procedure as

238 described by Schulten et al. (2000). In short the procedure is as follows:

239 • reconstitution of each capsule in 5 ml peptone saline solution in a Petri dish at

240 38.5 °C ± 1 °C for 45 min ± 5 min;

241 • repair of Salmonella by the addition of 5 ml molten double concentrated plate count agar

242 to the reconstituted capsule solution, and after solidification incubation at 37 °C ± 1 °C

243 for 4 h ± 0.5 h;

244 • after incubation, 10 ml of molten double concentrated Violet Red Bile Glucose agar was

245 added as an over layer and after solidification the plates were incubated at 37 °C ± 1 °C

246 for 20 h ± 2 h.

247

248 For the interlaboratory study, each participant received a batch of pathogen-free chicken

249 faeces separately packed from the reference materials. At the day of analyses, the

9
250 participants had to prepare the samples by combining 10 g chicken faeces with one

251 reference material according to a detailed protocol.

252

253 2.3.3 Test materials for interlaboratory study organised by EURL-Salmonella in 2012

254 Five batches of reference materials were prepared by the Health Protection Agency (HPA,

255 currently named Public Health England), Newcastle, United Kingdom, 1-4 months before the

256 interlaboratory study. These reference materials consisted of so-called lenticule discs, being

257 plano-convex (lens-shaped) discs containing microorganisms at a defined number in a solid

258 water-soluble matrix (Boyd et al., 2006; Desai et al., 2006). Lenticules with the following

259 mean contamination levels were used for the ILS of 2012:

260 • Salmonella Typhimurium at a level of 10 cfu/lenticule disc (STM10): NCTC 12023

261 (National Collection of Type Cultures, UK) batch 323-111025;

262 • Salmonella Typhimurium at a level of 58 cfu/lenticule disc (STM58): NCTC 12023 batch

263 523-100927R;

264 • Salmonella Derby at a level of 6 cfu/lenticule disc (SD6): NCTC 5722 batch 624-111215;

265 • Salmonella Derby at a level of 37 cfu/lenticule disc (SD37): NCTC 5722 batch 634-

266 111214;

267 • Blank lenticule discs containing no microorganisms: batch 000-110825.

268 The mean number of organisms of each batch was counted by HPA before the lenticule

269 discs were sent to the EURL-Salmonella. For this, the HPA tested 30 lenticules per batch.

270 For the current study, the contamination level of each batch of lenticule discs was verified at

271 the EURL-Salmonella by testing 2 lenticule discs (containing STM) or 5 lenticule discs

272 (containing SD) per batch after receipt and storage at -20 °C. For the counting of the lenticule

273 discs, each lenticule disc was placed onto Colombia agar plates with sheep blood (Oxoid

274 PB5008A, Germany). After 10 minutes of rehydration of the lenticule disc, the resultant ‘drop’

275 was spread over the plate and incubated at 37 °C for 20-24 hours (Kuijpers and Mooijman,

276 2013).

277
10
278 For the interlaboratory study, each participant received a batch of pathogen-free pig

279 faeces separately packed from the reference materials. At the day of analyses, the

280 participants had to prepare the samples by combining 25 g pig faeces with one reference

281 material according to a detailed protocol.

282

283 a. Homogeneity and stability of the test materials

284

285 2.4.1 Test materials for interlaboratory study organised under the Mandate in 2013

286 Acceptance of the production batches for use in the interlaboratory study was made on

287 the basis of achieving satisfactory homogeneity and stability according to the laboratories’

288 own criteria.

289 For an indication on the variation between samples in the number of Salmonella, and to

290 gain information on the contamination levels in the final samples, a five-tube most probable

291 number (MPN) technique was used. For this, tenfold dilutions of five boot sock samples of

292 each contamination level were tested, representing 10 g, 1 g and 0.1 g of the original

293 sample. The presence of Salmonella was determined in each dilution by following EN ISO

294 6579:2002/Amd.1:2007 (Anonymous, 2007). From the number of confirmed positive

295 dilutions, the MPN of Salmonella in the original sample was calculated, by using an MPN

296 program in Excel, freely available on the Internet (Jarvis et al., 2010).

297 The stability of the artificially contaminated boot sock samples was tested prior to the

298 interlaboratory study in three experiments with specially prepared batches with different

299 contamination levels. For this, low and high contaminated test samples were stored at +5 °C,

300 the normal storage temperature of the samples, and at elevated temperatures to test the

301 influence of abusive transport and/or storage temperatures (+10 °C and +15 °C). The

302 presence of Salmonella was tested in 5 samples at each storage temperature at day 0, 7, 14

303 and 21, following EN ISO 6579:2002/Amd.1:2007 (Anonymous, 2007). The set-up of the

304 three experiments was as follows:

11
305 • Experiment 1: boot sock samples artificially contaminated with Salmonella Typhimurium

306 at 7 cfu/sample stored at 5 °C and at 15 °C;

307 • Experiment 2: boot sock samples artificially contaminated with Salmonella Typhimurium

308 at 55 cfu/sample stored at 5 °C and at 15 °C;

309 • Experiment 3: boot sock samples artificially contaminated with Salmonella Typhimurium

310 at 14 cfu/sample stored at 5 °C and at 10 °C.

311 The stability of the (natural) background flora was tested in blank boot sock samples (with

312 faecal material, but without the addition of Salmonella) for the same storage conditions. For

313 this purpose the EN ISO procedures for establishing the total number of aerobic bacteria (EN

314 ISO 4833; Anonymous, 2003b) and for analysing the number of Enterobacteriaceae (EN

315 ISO21528-2; Anonymous, 2004) were followed.

316

317 2.4.2 Test materials for interlaboratory study organised by EURL-Salmonella in 2008

318 Before filling all mixed contaminated milk powders into gelatine capsules, test batches of

319 60 capsules were prepared of each mixture to determine the mean number of cfu per

320 capsule and the homogeneity of the mixture. The remaining mixed powders were stored at

321 -20 °C. If the test batches fulfilled the pre-set criteria for contamination level and

322 homogeneity, the relevant mixed powders were completely filled into gelatine capsules and

323 stored at -20 °C.

324 Information on (long-term) stability of the capsules was obtained from studies performed

325 earlier (2001-2006) on the same highly contaminated milk powders as used for preparing the

326 batches of capsules for the interlaboratory study of 2008. Additionally, also short term

327 stability of the capsules was tested when stored at elevated (abusive) temperatures. Details

328 on the homogeneity and stability studies can be found in Kuijpers et al., 2008.

329

330 2.4.3 Test materials for interlaboratory study organised by EURL-Salmonella in 2012

331 Each batch of lenticule discs were delivered with an insert, giving information on the

332 contamination level and homogeneity (tested by HPA).

12
333 Tests on the long-term stability at storage temperature (-20 °C) and short-term tests for

334 stability at abusive storage temperatures on lenticule discs containing S. Enteritidis (SE) and

335 S. Typhimurium (STM) were performed earlier by the EURL-Salmonella, and showed good

336 results (Kuijpers and Mooijman, 2011 and 2012).

337 For the interlaboratory study of 2012, some additional stability tests were performed for

338 storage of the lenticule discs at storage temperature (-20 °C) and at elevated temperatures.

339 Details on the homogeneity and stability studies can be found in Kuijpers and Mooijman,

340 2013.

341

342 2.5. Statistical analysis of the interlaboratory studies data

343 For the performance of a qualitative method, specificity, sensitivity and LOD50

344 (contamination level at which 50% of the samples are found positive) were calculated. LOD50

345 was calculated with the model described in EN ISO 16140-2, using a program in Excel, freely

346 available on the Internet (Anonymous, 2016b). In this program, the contamination level of

347 each batch of test samples has to be entered in cfu/g or cfu/ml. For this, the inoculum levels

348 of each batch of samples were used for the ILSs of 2008, 2012 and 2013. For the earlier

349 EURL-Salmonella studies this concerned the mean level of each batch of reference materials

350 per 10 g faeces (ILS 2008) or 25 g faeces (ILS 2012). For ILS 2013, this concerned the level

351 of Salmonella in the inoculum to contaminate the low level and high level samples per 10 g

352 chicken faeces (the weight of the pair of boot socks was not taken into account). Of the food

353 samples used in the validation studies in 2000, information on the inoculum levels was no

354 longer available. However, MPN results of the artificially contaminated food samples were

355 available and these were used for calculation of the LOD50 values.

356 The calculation of specificity and sensitivity (in %) is described below.

357

358 Specificity (SP):

359 = ∗ 100%

13
360 Where N- is the number of negative results at blank level and N is the total number of

361 analysed samples at this level.

362

363 Sensitivity (SE):

364 = ∗ 100%

365 Where N+ is the number of positive results at a low/high contamination level and N is the

366 total number of analysed samples at this level.

367

368

369 8 Results

370

371 3.1 Stability and homogeneity of the test samples of the interlaboratory study organised

372 under the Mandate in 2013 (boot sock samples)

373 Three batches of boot sock samples were subjected to the stability tests prior to the

374 interlaboratory study. These batches were artificially contaminated with Salmonella

375 Typhimurium at two low levels (7 cfu/sample and 14 cfu/sample) and one high level

376 (55 cfu/sample). The results of the stability studies are shown in Figure 1. The artificially

377 contaminated boot sock samples were relatively stable when stored at 5 °C and at 10 °C.

378 After 14 days, 4 to 5 of the 5 samples (high contaminated as well as low contaminated) were

379 still tested positive for Salmonella. A longer storage time (21 days), especially in combination

380 with a higher storage temperature (15 °C) resulted in a lower number of positive samples.

381 This was most clear for the low contaminated samples containing 7 cfu Salmonella/sample

382 and stored at 15 °C. An explanation could be that some die off occurred of Salmonella in the

383 samples and/or that due to the increase of the background flora during storage at 10-15 °C

384 (see below), the growth of Salmonella was suppressed.

385 The amount of natural background flora in the chicken faeces added to the boot socks

386 showed to be stable when stored at 5 °C for 3 weeks (Figure 2). When stored at higher

14
387 temperatures (10 °C and 15 °C) the total aerobic count and the number of Enterobactericeae

388 increased with 1 to 1.5 log10 cfu/g (Figure 2).

389 A stability of 14 to 21 days of the artificially contaminated boot sock samples was

390 considered sufficient for the test samples of the interlaboratory study, when taking into

391 account the following: preparation of the test samples shortly before the performance of the

392 interlaboratory study, contamination level of the low contaminated samples between 5 and

393 10 cfu/sample, contamination level of the high contaminated samples approx. 10 times

394 higher than the low contaminated samples, storage of the samples at 5 °C and short

395 transport time (1-2 days) at 5-10 °C.

396 The final batches of boot sock samples for the interlaboratory study were inoculated with

397 Salmonella Typhimurium at a low level of 9 cfu/sample and at a high level of 81 cfu/sample.

398 After 10 days of storage at 5 °C, the contamination level in the inoculated test samples were

399 enumerated with an MPN technique, showing an MPN of 3.3 S. Typhimurium per sample

400 (95% confidence interval 1.1 -10) for the low contaminated samples, and an MPN of 160 (53

401 – 490) S. Typhimurium per sample for the high contaminated samples, which was

402 considered acceptable for use in the interlaboratory study. The amount of background flora

403 at the date of the study was approximately 107 cfu/g for the aerobic bacteria and

404 approximately 104 cfu/g for the number of Enterobacteriaceae (Kuijpers and Mooijman,

405 2014).

406 Details on the results of the homogeneity and stability studies of the test materials used in

407 the EURL-Salmonella interlaboratory studies of 2008 and 2012 can be found in respectively

408 Kuijpers et al., 2008 and in Kuijpers and Mooijman, 2013. For these latter studies, the matrix

409 samples were artificially contaminated by the participating laboratories by adding a reference

410 material to a prescribed amount of matrix at the day of analysis. For both type of reference

411 materials used in the two studies (capsules and lenticules), it was shown that the

412 homogeneity fulfilled the pre-set criteria and that the reference materials were stable when

413 stored at -20 °C for at least 1 year. Additionally, the reference materials were shown to be

15
414 stable when stored at +5 °C for approx. 1 week, which was sufficient for mailing of the

415 samples.

416

417 3.2 General results of the interlaboratory study organised under the Mandate in 2013

418 All participants received the boot sock samples within 48 hours after dispatch. For the

419 majority of the parcels, the temperature did not exceed 5 °C during transport and storage at

420 the participating laboratory. Parcels of four laboratories were subjected to temperatures of

421 5 °C - 10 °C for only a few hours and the parcel of one laboratory was transported and/or

422 stored at 10 °C - 16 °C for two days. Still, no negative effects were seen on the results of the

423 laboratories concerned as the laboratory with the longest transport time in combination with

424 the highest temperatures still tested all contaminated samples positive for Salmonella. All

425 participants used the prescribed method (EN ISO 6579:2002/Amd.1:2007) for analysis of the

426 test samples. No problems were encountered with the analysis of the test samples, neither

427 with the analysis of the control samples.

428

429 3.3 Results excluded from further analysis

430 On forehand it was agreed that results from laboratories would not be excluded unless

431 they reported (major) technical deviations which might have influenced the results. For that

432 reason, data of some laboratories have not been used for calculation of the performance

433 characteristics. The number of datasets which were excluded from further analysis and the

434 reasons for exclusion are given below.

435 • Interlaboratory study organised under the Mandate in 2013: the results of 3 laboratories

436 were excluded because these laboratories used deviating concentrations of novobiocin

437 in MSRV agar. The prescribed concentration is 0.01 g/L, and the three laboratories used

438 a higher concentration of novobiocin (two laboratories used 0.02 g/L and one laboratory

439 used 0.05 g/L). Earlier studies have shown that the concentration of novobiocin can

440 affect the bacterial motility (Soutourina et al., 2001; Veenman et al., 2007) and as the

16
441 specific growth of Salmonella on MSRV agar is based on motility, it is essential that the

442 prescribed concentration novobiocin in MSRV agar is used.

443 • Interlaboratory study organised by EURL-Salmonella in 2008: the results of 13

444 laboratories were excluded for various reasons. One laboratory used a longer time to

445 reconstitute the reference materials (60 min instead of 45 min), which may have resulted

446 in possible growth of Salmonella during reconstitution and thus in more positive samples

447 than expected. One laboratory incubated the pre-enrichment broth (BPW) for 24 h

448 instead of 16-20 h. A longer incubation time of the BPW may cause overgrowth of

449 background flora and suppression of the growth of Salmonella and therefore possibly

450 more negative samples than expected. One laboratory incubated the MSRV agar plates

451 at 37 °C instead at 41.5 °C. Incubation temperature is a selective factor for the growth on

452 MSRV agar. Using a lower incubation temperature than prescribed may result in more

453 growth of the background flora and may negatively affect the growth of Salmonella. Ten

454 laboratories used deviating concentrations of novobiocin in MSRV agar of which four did

455 not report the concentration or reported the concentration to be 0 g/L and six

456 laboratories reported a higher concentration than 0.01 g/L.

457 • Interlaboratory study organised by EURL-Salmonella in 2012: the results of 7

458 laboratories were excluded because of a longer incubation time of the BPW (23 h) than

459 prescribed (one laboratory) and because of using deviating concentrations of novobiocin

460 in MSRV agar (six laboratories).

461

462 3.4 Statistical analysis of data

463

464 3.4.1 Performance characteristics detection of Salmonella in samples from the primary

465 production stage

466 The results of the data analysis of the interlaboratory study organised under the Mandate

467 in 2013 are summarised in Table 2. After exclusion of the data of three laboratories (see 3.3),

468 a total of 264 data per contamination level retained for the analysis of the performance

17
469 characteristics. Only one laboratory tested one blank boot sock sample (false) positive for

470 Salmonella, resulting in an overall specificity of 99.6%. The sensitivity rates for the low

471 contaminated boot sock samples as well as for the high contaminated samples were 94.7%

472 and 98.1%, respectively. Thirteen laboratories did not detect Salmonella in 14 low

473 contaminated samples and 5 laboratories did not detect Salmonella in 5 high contaminated

474 samples. The overall LOD50 was approximately 4 cfu/sample.

475 The results of the data analysis of the interlaboratory study organised by EURL-

476 Salmonella in 2008 are summarised in Table 3. After exclusion of the data of 13 laboratories

477 (see 3.3), a total of 95 data per contamination level retained for the analysis of the

478 performance characteristics. All laboratories tested the blank chicken faeces samples

479 correctly negative for Salmonella, resulting in an overall specificity of 100%. For the artificial

480 contamination of the chicken faeces samples, two Salmonella serovars were used. The

481 sensitivity rate of the highly contaminated samples with Salmonella Typhimurium as well as

482 with Salmonella Enteritidis was 100%. The overall sensitivity rate of the low contaminated

483 samples with Salmonella Enteritidis was 67.4% (14 laboratories did not detect Salmonella in

484 31 samples), while for the low contaminated samples with Salmonella Typhimurium this was

485 96.8% (3 laboratories did not detect Salmonella in 3 samples). As a result, the LOD50 values

486 also differed per serovar: approximately 1 cfu/test portion for Salmonella Typhimurium and

487 approximately 4 cfu/test portion for Salmonella Enteritidis.

488 The results of the data analysis of the interlaboratory study organised by EURL-

489 Salmonella in 2012 are summarised in Table 4. After exclusion of the data of 7 laboratories

490 (see 3.3), a total of 130 data per contamination level retained for the analysis of the

491 performance characteristics. Only one laboratory tested one blank pig faeces sample (false)

492 positive for Salmonella, resulting in an overall specificity of 99.2%. For the artificial

493 contamination of the pig faeces samples also two Salmonella serovars were used, but the

494 difference in sensitivity rates was less pronounced than for the study of 2008. The overall

495 sensitivity rates of the highly contaminated samples was 98.5% for Salmonella Typhimurium

496 (one laboratory did not detect Salmonella in two samples) and 97.7% for Salmonella Derby

18
497 (two laboratories did not detect Salmonella in three samples). With the low contaminated

498 samples with Salmonella Typhimurium a sensitivity rate of 91.5% was found (five

499 laboratories did not detect Salmonella in 11 samples) and with the low contaminated

500 samples with Salmonella Derby a sensitivity rate of 88.5% was found (six laboratories did not

501 detect Salmonella in 15 samples). The overall LOD50 for all samples was approximately

502 3 cfu/test portion.

503 Tables 2-4 are also published in Annex C of EN ISO 6579-1 (Anonymous, 2017).

504

505 Table 2 Results of data analysis of interlaboratory study 2013, detection of Salmonella in

506 boot sock samples

507

Parameter Pair of boot socks + 10 g laying hen

faecal material +

Blank STM9a STM81a

Number of participating collaborators 36 36 36

Number of samples per collaborator 8 8 8

Number of laboratories retained after 33 33 33

evaluation of the data

Number of samples retained after 264 264 264

evaluation of the data

Sample size Boot socks

Specificity, in % 99.6 - -

Sensitivity per level, in % - 94.7 98.1

LOD50 (95 % confidence interval), - 3.8 (3.2 to 4.4)

in cfu/sample
a The boot sock samples were artificially contaminated with a diluted culture of Salmonella

Typhimurium (STM) at a level of 9 cfu/sample and a level of 81 cfu/sample.

19
 -: not applicable

508

509

20
510

511 Table 3 Results of data analysis of interlaboratory study 2008, detection of Salmonella in

512 chicken faeces samples

513

Parameter Chicken faeces +

Blank STM5a STM44a SE7a SE91a

Number of participating collaborators 32 32 32 32 32

Number of samples per collaborator 5 5 5 5 5

Number of collaborators retained after 19 19 19 19 19

evaluation of the data

Number of samples retained after 95 95 95 95 95

evaluation of the data

Test portion size, in g 10 10 10 10 10

Specificity, in % 100 - - - -

Sensitivity per serovar and level, in % - 96.8 100 67.4 100

LOD50 per serovar (95 % confidence - 1.0 (0.7 to 1.4) 4.3 (3.3 to 5.6)

interval), in cfu/test portion

LOD50 overall (95 % confidence interval), - 2.5 (2.1 to 3.0)

in cfu/test portion

a Chicken faeces samples were artificially contaminated with reference materials with the following

strains and levels:

Salmonella Typhimurium (STM) at a level of 5 cfu/test portion and a level of 44 cfu/test portion;

Salmonella Enteritidis (SE) at a level of 7 cfu/test portion and a level of 91 cfu/test portion.

-: not applicable

514

515

21
516

517 Table 4 Results of data analysis of interlaboratory study 2012, detection of Salmonella in pig

518 faeces samples

519

Parameter Pig faeces +

Blank SD6a SD37a STM10a STM58a

Number of participating collaborators 33 33 33 33 33

Number of samples per collaborator 5 5 5 5 5

Number of collaborators retained after 26 26 26 26 26

evaluation of the data

Number of samples retained after 130 130 130 130 130

evaluation of the data

Test portion size, in g 25 25 25 25 25

Specificity, in % 99.2 - - - -

Sensitivity per serovar and level, in % - 88.5 97.7 91.5 98.5

LOD50 per serovar (95 % confidence - 2.8 (2.2 to 3.5) 3.8 (3.0 to 4.7)

interval), in cfu/test portion

LOD50 overall (95 % confidence - 3.2 (2.8 to 3.8)

interval), in cfu/test portion

a Pig faeces samples were artificially contaminated with reference materials with the following strains

and levels:

Salmonella Derby (SD) at a level of 6 cfu/test portion and a level of 37 cfu/test portion;

Salmonella Typhimurium (STM) at a level of 10 cfu/test portion and a level of 58 cfu/test portion.

-: not applicable

520

521

522

22
523 3.4.2 LOD50 values for detection of Salmonella in food samples

524 In EN ISO 6579 of 2002, performance characteristics were published for the detection of

525 Salmonella in three different food samples: fresh cheese curd, dried egg powder and raw

526 poultry meat. These performance characteristics were derived from interlaboratory studies

527 organised with these matrices in 2000 in the frame of the European project SMT CT 96 2098

528 (AFSSA, 2001; Feldsine et al., 2003) and consisted of specificity, sensitivity, accordance and

529 concordance. The latter two performance characteristics are no longer used because of their

530 limited use, and therefore an attempt was made to calculate the LOD50 values instead. For

531 this, the raw data of the interlaboratory studies of spring 2000 could be used as they were

532 still available at the French Agency for Food, Environmental and Occupational Health &

533 Safety (ANSES, formerly called AFSSA). Additional to the interlaboratory study performed to

534 determine the performance characteristics published in EN ISO 6579:2002 (Anonymous,

535 2002), a second study was organised in fall 2000 to obtain fractional recovery for egg powder

536 samples and raw poultry meat. The results of this second study were published by Feldsine

537 et al in 2003 and were used, additional to the results of the first study, to calculate LOD50

538 values for the three matrices mentioned above. The specificity and sensitivity rates were

539 already published in EN ISO 6579:2002 (Anonymous, 2002) and were copied unchanged

540 into the updated tables with performance characteristics of EN ISO 6579-1:2017

541 (Anonymous, 2017). In Table 5 the LOD50 values are given for the three matrices, fresh

542 cheese curd, dried egg powder and raw poultry meat. Table 5 also shows information on the

543 contamination levels of the samples and on the number of data used for calculating the

544 LOD50 values. For completeness, also the sensitivity rates are given, copied from the tables

545 in Annex C of EN ISO 6579:2002. The specificity rate was 100% for all three matrices. The

546 updated performance characteristics have also been published in Annex C of EN ISO 6579-1

547 (Anonymous, 2017).

548

549

550 9 Discussion

23
551 In the past, only a few European and International standard methods have been published

552 with information on the performance characteristics of the method. These characteristics are

553 important, e.g. to verify the performance of a laboratory when the method is newly

554 introduced, or to validate an alternative method. To enhance the number of Standard

555 methods with performance characteristics, the European Commission issued a Mandate for

556 the validation of 15 microbiological standard methods. Although EN ISO 6579 (Anonymous,

557 2002) did contain performance characteristics for the detection of Salmonella in food

558 samples, its amendment published in 2007 for the detection of Salmonella in samples from

559 the primary production stage (Anonymous, 2007) did not. For that reason, validation of

560 Amd.1 of EN ISO 6579 was part of Mandate M/381. For this so-called ‘vertical’ method only

561 one matrix needed to be analysed in an interlaboratory study. Boot sock samples with

562 (artificially contaminated) chicken faeces were chosen as this type of samples is often

563 analysed in laboratories and was considered representative for samples from the primary

564 production stage.

565 The performance characteristics to be determined for qualitative methods under the

566 Mandate differed slightly from the ones already published in EN ISO 6579:2002. Instead of

567 accordance and concordance, LOD50 needed to be determined. To be able to calculate

568 LOD50 values, samples with a low level should be used so that fractional recovery is obtained

569 in the ILS (ideally, 50% of the samples should be positive, with a range of 25%-75%). The

570 preparation of such samples for an interlaboratory study is quite a challenge. The

571 contamination level should not be too low as die off may occur during storage and mailing of

572 the samples, which may result in too many negative samples. Furthermore, when real

573 matrices with (natural) background flora are used, the growth of the target organism may be

574 disturbed by the background flora present in the matrix, although the amount of disturbance

575 is difficult to predict. Taking all these possible effects into account, the contamination level of

576 Salmonella in the low contaminated boot sock samples was aimed to be 5-10 cfu/pair of boot

577 sock samples, which was well achieved with 9 cfu S. Typhimurium/sample at the day of

578 preparation. These low contaminated samples showed to be stable for 2-3 weeks when

24
579 stored at 5 °C, and for approximately 1 week when stored at 10-15 °C, which was considered

580 sufficient for the purpose of the study. After 10 days of storage at 5 °C, the number of

581 Salmonella in the samples was approx. 3 cfu (with a 95% confidence interval of 1-10 cfu),

582 confirming that the initial inoculum of 9 cfu was needed to retain sufficient Salmonella in the

583 samples at the day of analysis. The total amount of background flora in the chicken faeces

584 added to the boot socks was relatively high (approx. 107 cfu/g), so that some disturbance in

585 the growth of Salmonella was expected, probably resulting in fractional recovery. However,

586 the final results showed that the sensitivity rate of the low contaminated samples was almost

587 95%, meaning that only a (very) small amount of fractional recovery was obtained. This is a

588 good result for the performance of the method, showing that this method is able to detect low

589 levels of Salmonella in the presence of high amounts of background flora, but less ideal for

590 the calculation of LOD50. As still some fractional recovery was found, it was possible to give

591 an indication on the LOD50 value for the boot sock samples.

592 To see if the performance characteristics found in the interlaboratory study with boot sock

593 samples were representative for (other) samples from the primary production stage, they

594 were compared to data obtained from interlaboratory studies organised earlier. The

595 performance characteristics derived from these earlier studies showed in general

596 comparable values to the ones of the study with boot socks, especially for the LOD50 values.

597 Thanks to the fact that the raw data of the validation studies performed in 2000 were still

598 available, the LOD50 values could relatively easily be determined from these data for the

599 detection of Salmonella in food samples. This shows the importance of storage of raw data of

600 validation studies, so that the data remain available for future use. The LOD50 values for the

601 food samples were close, or slightly higher than the values found for the samples from the

602 primary production stage: 2-6 cfu/test portion for the food samples against 2-4 cfu/test

603 portion for the samples from the primary production stage.

604

605 10 Conclusions

25
606 Thanks to the activities performed under, and in line with, Mandate M/381, the revised

607 version of the European and International standard for detection of Salmonella in samples of

608 the food chain (EN ISO 6579-1:2017) now contains (updated) performance characteristics

609 not only for detection of Salmonella in food samples, but also for detection of Salmonella in

610 samples from the primary production stage.

611 The outcome of the validation studies showed that EN ISO 6579-1 for detection of

612 Salmonella works equally well for food samples as for samples from the primary production

613 stage.

614

615 Acknowledgements

616 The validation of International Standard EN ISO 6579-1 has been carried out under the

617 framework of European Mandate No. M381 of Directorate-General for Health and Food

618 Safety (DG SANTE, European Commission) and Directorate-General for Internal Market,

619 Industry, Entrepreneurship and SMEs (DG GROW, European Commission). The earlier

620 interlaboratory studies were organised by the European Union Reference Laboratory (EURL)

621 for Salmonella, which is funded by DG-SANTE.

622 The authors wish to thank the coordination team of this Mandate, from CEN/TC275/WG6

623 ‘Microbiology of the food chain’ Alexandre Leclercq (convenor), Gwenola Hardouin

624 (secretary), from ISO/TC34/SC9 ‘Food products – Microbiology’ Bertrand Lombard (chairman

625 SC9 and convenor of WG2 ‘Statistics’), Paul in ‘t Veld (convenor of WG3 ‘Method

626 validation’).

627 The authors wish to thank the following NRL-Salmonella laboratories for their participation

628 and cooperation in the interlaboratory study organised under the Mandate:

629 Austria: Austrian Agency for Health and Food Safety (AGES IVET), Graz, Heimo Lassnig;

630 Belgium: Veterinary and Agrochemical Research Centre (VAR), General and Molecular

631 Bacteriology (CODA-CERVA), Brussels, Vicky Jasson; Bulgaria: National Diagnostic and

632 Research Veterinary Institute, National Reference Centre of Food Safety, Sofia, Hristo

633 Daskalov; Cyprus: Cyprus Veterinary Services, Laboratory for the Control of Foods of Animal

26
634 Origin (LCFAO), Nicosia, Maria Emmanuel; Croatia: Croatian Veterinary Institute Laboratory

635 for general Bacteriology and Microbiology, Zagreb, Borka Simpraga; Czech Republic: State

636 Veterinary Institute, Prague, Tomás Cerný; Denmark: Danish Veterinary and Food

637 Administration, Microbiology Laboratory, Esbjerg, Susanne Mogensen; Estonia: Estonian

638 Veterinary and Food Laboratory, Bacteriology-Pathology Department, Tartu, Age Kárssin;

639 Finland: Finnish Food Safety Authority Evira, Research Department, Veterinary Bacteriology,

640 Henry Kuronen; France: Anses Ploufragan-Plouzané, Laboratoire d’Etudes et de

641 Recherches Avicoles, Porcines et Piscicoles, Unité Hygiene et Qualité des Produits Avicoles

642 et Porcins (HQPAP), Ploufragan, Marylene Bohnert; Germany: Federal Institute for Risk

643 Assessment (BfR), National Veterinary Reference Laboratory for Salmonella, Berlin

644 Marienfelde, Reiner Helmuth; Greece: Veterinary Laboratory of Chalkida, Chalkida, Maria

645 Passiotou; Hungary: Central Agricultural Office, Food and Feed Safety Directorate,

646 Budapest, Erzsebet Adrian; Iceland: University of Iceland, Institute for Experimental

647 Pathology Keldur, Reykjavik, Vala Frioriksdottir; Ireland: Central Veterinary Research

648 Laboratory, Department of Agriculture, Fisheries and Food (CVRL/DAFF), Kildare,

649 Montserrat Gutierrez; Israel: Israel Veterinary Services, Southern Laboratory for Poultry

650 Health, Beer Tuvia Regional Poultry Diseases Laboratory, Kiryat Malachi, Elyakum Berman;

651 Italy: Instituto Zooprofilattico Sperimentale delle Venezie, Padova Legnaro, Antonia Ricci;

652 Latvia: Institute of Food Safety, Animal Health and Environment (BIOR), Animal Disease

653 Diagnostic Laboratory, Riga, Jelena Avsejenko; Lithuania: National Food and Veterinary Risk

654 Assessment Institute, Vilnius, Ruta Bubuliene; Luxembourg: Laboratoire de Médicine

655 Vétérinaire de l’Etat, Anima Zoonosis, Luxembourg, Joseph Schon; Malta: Public Health

656 Laboratory (PHL), Evans Building, Valetta, Albert Gambin; the Netherlands: National Institute

657 for Public Health and the Environment (RIVM), Laboratory for Zoonosis and Environmental

658 Microbiology (LZO), Bilthoven, Alice van der Meij-Florijn, Wendy van Overbeek, Ellen

659 Delfgou, Luc Wijnands; Norway: National Veterinary Institute, Section Bacteriology, Oslo,

660 Bjarne Bergsjo; Poland: National Veterinary Research Institute (NVRI), Department of

661 Microbiology, Pulawy, Andrzej Hoszowski; Portugal: Laboratório Nacional de Investigaçao

27
662 Veterinária (LNIV), Lisbon, Teresa Albuquerque; Romania: Institute for Diagnosis and Animal

663 Health, Bacteriology, Bucharest, Ion Sandu; Slovak Republic: State Veterinary and Food

664 Institute, Bratislava, Alena Skarkova; Slovenia: University Ljubljana, Veterinary Faculty,

665 National Veterinary Institute, Ljubljana, Vojka Bole-Hribovsek; Spain: Laboratorio Central de

666 Veterinaria, Madrid Algete, Cristina Frutos Esobar; Sweden: National Veterinary Institute

667 (SVA), Department of Bacteriology, Uppsala, Lennart Melin; Switzerland: Vetsuisse Faculty

668 Bern, Institute of Veterinary Bacteriology, Centre for Zoonoses, Bacterial Animal diseases

669 and Antimicrobial Resistance (ZOBA), Bern, Gudrun Overesch; United Kingdom: Animal

670 Health and Veterinary Laboratories Agency (AHVLA) Weybridge, Bacteriology Department,

671 Addlestone, Robert Davies; United Kingdom: Agri-Food and Bioscience Institute (AFBI),

672 Veterinary Sciences Division Bacteriology, Belfast, Gintare Bagdonaite.

673 The authors wish to thank the members of CEN/TC275/WG6 TAG8 (Technical Advisory

674 Group for Salmonella):

675 France: Anses Ploufragan-Plouzané, Laboratoire d’Etudes et de Recherches Avicoles,

676 Porcines et Piscicoles, Unité Hygiene et Qualité des Produits Avicoles et Porcins (HQPAP),

677 Ploufragan, Marylene Bohnert; France: BioMerieux SA, Marcy l’Etoile, Isabelle Desforges;

678 France: BIO-RAD, Steenvoorde, Christophe Quiring; France: AFNOR Standardisation, La

679 Plaine Saint-Denis, Cedex, Gwenola Hardouin (WG6 secretariat); Germany: Merck KGaA,

680 Darmstadt, Barbara Gerten; Germany: Lehrstul für Hygiene und Technologie der Milch,

681 Oberschleißheim, Heinz Becker; Ireland: National Standards Authority of Ireland, Dublin,

682 Annemarie Crowley; Italy: Silliker, Italia Spa, Bergamo, Stefano Colombo; The Netherlands,

683 National Institute for Public Health and the Environment (RIVM), Laboratory for Zoonosis and

684 Environmental Microbiology (LZO), Bilthoven, Wilma Jacobs-Reitsma; South Korea: Seoul

685 Women’s University, Seoul, Jung Dong Sun; Spain: Laboratoire de Salud Public de

686 Gipuzkoa, San Sebastian, Belen Moreno; Switzerland: Nestlé Research Centre, Nestec Ltd,

687 Microbiological and Molecular Analytics, Lausanne, Adrianne Klijn, Benjamin Diep, David

688 Tomas Fornes; Thailand: Ministry of Public Health, Department of Medical Science,

689 Nontaburi, Pensri Rodma; Thailand: Bureau of Senior Technical Expert, Department of

28
690 Livestock Development, Bangkok, Sasitorn Kanarat; United Kingdom: Microtech Services

691 Wessex Ltd., Bournemouth, Melody Greenwood; United Kingdom: Oxoid, R&D department,

692 Thermo Fisher Schientific, Basingstoke, James Stringer; United Kingdom: Eurofins Sientific

693 Group, Molecular Biology & Immunology, Bert Popping; United Kingdom: Health Protection

694 Agency, Centre for Infections, Gastrointestinal, Emerging and Zoonotic Infections, London,

695 Elizabeth de Pinna; USA: United States Department of Agriculture, Food Safety and

696 Inspection Service, Washington, Philip Bronstein; USA: US Food and Drug Administration,

697 Silver Spring, Thomas Hammack; USA: American National Standards Institute, Washington,

698 Henry Cheung; USA: Silliker, Chicago, Wendy McMahon.

699

700 Conflict of interest

701 No conflict of interest declared.

702

703 References

704 AFSSA, 2001. Evaluation of microbiological methods for detection and for enumeration of

705 microbiological contaminants in food. Final report Contract SMT4/CT96 2098.

706 Coordination by Agence Française de Sécurité Sanitaire des Aliments. AFSSA, France,

707 February 2001.

708 Anonymous, 2002. EN ISO 6579 Microbiology of food and animal feeding stuffs – Horizontal

709 method for the detection of Salmonella spp. International Organization for

710 Standardization, Geneva, Switzerland.

711 Anonymous, 2003a. EN ISO 16140. Microbiology of food and animal feeding stuffs –

712 Protocol for the validation of alternative methods. International Organization for

713 Standardization, Geneva, Switzerland.

714 Anonymous, 2003b. EN ISO 4833. Microbiology of food and animal feeding stuffs –

715 Horizontal method for the enumeration of microorganisms – Colony-count technique at

716 30 degrees C. International Organization for Standardisation, Geneva, Switzerland.

29
717 Anonymous, 2004. EN ISO 21528-2. Microbiology of food and animal feeding stuffs –

718 Horizontal methods for the detection and enumeration of Enterobacteriaceae – Part 2:

719 Colony-count method. International Organization for Standardisation, Geneva,

720 Switzerland.

721 Anonymous, 2005. EN ISO/IEC 17025 General requirements for the competence of testing

722 and calibration laboratories. International Organization for Standardization, Geneva,

723 Switzerland.

724 Anonymous, 2007. EN ISO 6579:2002/Amd.1:2007 Microbiology of food and animal feeding

725 stuffs – Horizontal method for the detection of Salmonella spp. – Annex D: Detection of

726 Salmonella spp. in animal faeces and in environmental samples from the primary

727 production stage. International Organization for Standardization, Geneva, Switzerland.

728 Anonymous, 2013. IATA-International Air Transport Association. Dangerous Goods

729 Regulations, 54th edition, pp 135 – 139. Montreal, Geneva.

730 Anonymous, 2016a. EN ISO 16140-1 Microbiology of the food chain – Method validation -

731 Part 1: Vocabulary. International Organization for Standardization, Geneva, Switzerland.

732 Anonymous, 2016b. EN ISO 16140-2 Microbiology of the food chain – Method validation -

733 Part 2: Protocol for the validation of alternative (proprietary) methods against a reference

734 method. International Organization for Standardization, Geneva, Switzerland. Link to

735 LOD50 calculation programme: http://standards.iso.org/iso/16140 (access date 06-11-

736 2017).

737 Anonymous, 2017. EN ISO 6579-1 Microbiology of the food chain – Horizontal method for

738 the detection, enumeration and serotyping of Salmonella – Part 1: Detection of

739 Salmonella spp. International Organization for Standardization, Geneva, Switzerland.

740 Boyd, P., Eaton, S. and Magee, J., 2006. The stability of micro-organisms preserved in

741 LENTICULE discs, demonstrated by repeat sample distributions of the EQUAL Scheme

742 for Indicator Organisms (water microbiology). Journal of Applied Microbiology 100, 272–

743 278.

30
744 Desai, M., Russell, J.E. and Gharbia, S., 2006. Genetic stability of strains preserved on

745 LENTICULE discs and by freeze-drying: a comparison using fluorescent amplified

746 fragment length polymorphism analysis. FEMS Microbiol Lett 257, 158–162.

747 EURL-Salmonella, 2017. Protocols, SOPs and Test reports of EURL-Salmonella

748 interlaboratory studies for detection of Salmonella in samples from the primary

749 production stage. Link:

750 http://www.eurlsalmonella.eu/Proficiency_testing/Detection_studies_Faeces (access

751 date 13-11-2017).

752 Feldsine, P.T., Lienau, A.H., Leung, S.C., Mui, L.A., Humbert, F., Bohnert, M., Mooijman, K.,

753 Schulten, S, in ‘t Veld, P., Rollier, P., Leuschner, R and Capps, K., 2003. Detection of

754 Salmonella in fresh cheese, poultry products, and dried egg products by the ISO 6579

755 Salmonella culture procedure and the AOAC official method: Collaborative study.

756 Journal of AOAC International, Vol 86, no. 2, 275-295.

757 Jarvis, B., Wilrich, C., Wilrich, P.-T., 2010. Reconsideration of the derivation of most probable

758 numbers, their standard deviations, confidence bounds and rarity values. J. Appl.

759 Microbiol. 109:1660–1667. Link to MPN calculation programme:

760 http://standards.iso.org/iso/7218/ (access date 06-11-2017).

761 Kuijpers, A.F.A., Veenman, C. and Mooijman, K.A., 2008. EU Interlaboratory comparison

762 study veterinary XI (2008) - Bacteriological detection of Salmonella in chicken faeces.

763 National Institute for Public Health and the Environment, Bilthoven, the Netherlands.

764 Report no.: 330604 011/ 2008. http://www.rivm.nl/bibliotheek/rapporten/330604011.pdf

765 (access date 07-07-2017).

766 Kuijpers, A.F.A. and Mooijman, K.A., 2011. EU Interlaboratory comparison study veterinary

767 XIV (2011); Detection of Salmonella in chicken faeces. RIVM Report 330604023,

768 Bilthoven, the Netherlands. http://www.rivm.nl/bibliotheek/rapporten/330604023.pdf

769 (access date 20-07-2017).

770 Kuijpers, A.F.A. and Mooijman, K.A., 2012. EU Interlaboratory comparison study food V

771 (2011); Detection of Salmonella in minced meat. RIVM Report 330604025, Bilthoven,

31
772 the Netherlands. http://www.rivm.nl/bibliotheek/rapporten/330604025.pdf (access date

773 20-07-2017)

774 Kuijpers, A.F.A.and Mooijman, K.A., 2013. EU Interlaboratory comparison study Veterinary-

775 XV (2012) - Detection of Salmonella in pig faeces. National Institute for Public Health

776 and the Environment, Bilthoven, The Netherlands. RIVM Report no.: 330604 028/ 2013.

777 http://www.rivm.nl/bibliotheek/rapporten/330604028.pdf (access date 07-07-2017).

778 Kuijpers, A.F.A. and Mooijman, K.A., 2014. EU Interlaboratory comparison study primary

779 production XVI (2013) – Detection of Salmonella in chicken faces adhering to boot

780 socks. National Institute for Public Health and the Environment, Bilthoven, the Nether-

781 lands. RIVM Report no.: 330604 031/ 2014.

782 http://www.rivm.nl/bibliotheek/rapporten/330604031.pdf (access date 27-06-2017).

783 Mooijman, K.A., in press. The new ISO 6579-1: A real horizontal standard for detection of

784 Salmonella, at last! Food Microbiology (2017). http://dx.doi.org/10.1016/j.fm.2017.03.001

785 Schulten, S.M., In ’t Veld, P.H., Ghameshlou, Z., Schimmel, H., Linsinger, T., 2000. The

786 certification of the number of colony forming particles of Salmonella Typhimurium and

787 number fraction of negative capsules from artificially contaminated milk powder.

788 Commission of European Communities, Community Bureau of Reference, Brussels,

789 Luxembourg. CRM 507R, EUR 19622 EN.

790 Soutourina, O.A., Semenove, E.A., Parfenova, V.V., Danchi, A. and Bertin, P., 2001. Control

791 of bacterial motility by environmental factors in polarly flagellated and peritichous

792 bacteria isolated from Lake Baikal. Appl. Environ. Mircobiol., 67 (9), pp. 3852-3859

793 Veenman, C., Korver, H. and Mooijman, K.A., 2007. Improvements in the method for

794 detection of Salmonella spp. in animal faeces. National Institute for Public Health and

795 the Environment, Bilthoven, the Netherlands. RIVM report 330300 010.

796 http://www.rivm.nl/bibliotheek/rapporten/330300010.pdf (access date 07-07-2017)

797 Veld in ‘t, P.H., Strijp-Lockefeer van, N.G.W.M., Havelaar, A.H., Maier, E.A., 1996. The

798 certification of a reference material for the evaluation of the ISO method for the detection

799 of Salmonella. J.Appl.Bacteriol; 80: 496-504

32
800

801

802 Figure 1 Stability tests of boot sock samples (5 samples per batch, time and temperature

803 combination) artificially contaminated with Salmonella Typhimurium at three different levels,

804 7 cfu/sample (STM7), 14 cfu/sample (STM14) and 55 cfu/sample (STM55), and stored at

805 three different temperatures (5 °C, 10 °C and 15 °C).

806

807

808

809

810

811

33
812

813 Figure 2 Stability tests of natural background flora in ‘blank’ boot sock samples with chicken

814 faeces from a laying hen flock, stored at three different temperatures (5 °C, 10 °C and

815 15 °C), given in log10 cfu/g total aerobic count (‘aerobic’) and log10 cfu/g Enterobacteriaceae

816 (‘entero’).

817

818

819

820

821

822

34
823 Table 1 Number and type of samples analysed in three interlaboratory comparison studies (ILS) for detection of Salmonella in samples from the

824 primary production stage

825

Year No. of Matrix No. of samples with matrix and No. of control samplesb with Inoculation
ILS parti- Salmonellaa at different Salmonellaa at different of samples
cipants contamination levels contamination levels, no matrix with
type amount per
sample Blank Low High Blank Low High
2008 32 Chicken faeces 10 g 5 5 STM 5 STM 2 3 STM 1 SE Reference
5 SE 5 SE 2 SE materials
2 SPan (capsules)
2012 33 Pig faeces 25 g 5 5 STM 5 STM 2 2 STM 1 SD Reference
5 SD 5 SD 2 SD materials
(lenticules)
2013 36 Boot socks + 10 g/pair of 8 8 STM 8 STM 2 2 STM 2 STM Diluted
chicken faeces boot socks culture
826
a
827 STM: Salmonella Typhimurium; SE: Salmonella Enteritidis; SD: Salmonella Derby; SPan: Salmonella Panama
b
828 Control samples consisted in the studies of 2008 and 2012 of reference materials only (no matrix added) and in the study of 2013 of
829 moistened boot socks inoculated with a diluted culture of STM (no matrix added) or without Salmonella (blank).

35
830 Table 5 LOD50 values and sensitivity rates for detection of Salmonella in food samples, calculated from data of two interlaboratory studies (ILS-I

831 and ILS-II) organised in 2000 (Anonymous, 2002; Feldsine et al., 2003)

832
Fresh cheese curda Egg powderb Raw poultry meatc
ILS-I ILS-I ILS-II ILS-I ILS-II
Low level High level Low level High level Low level Low level High level Low level High level
Contamination level of artificially
contaminated samples, MPN per 0.7 37.2 9.6 115 0.7 3.7 5.8 0.2 1.0
25 g (95% confidence interval)d (0.2-2.4) (7.5-95) (2.2-26) (22.5-495) (0.2-2.3) (1.0-9.5) (1.0-25) (0.04-0.9) (2.2-4.5)
Number of participating 23 23 26 26 9 25 25 13 13
collaborators
Number of samples per 5 5 5 5 5 5 5 6 6
collaborator
Number of collaborators retained 21 21 21 21 8 20 20 13 13
after evaluation of the data
Number of samples retained after 105 105 105 104 40 99 100 78 78
evaluation of the data
Test portion size, in g 25 25 25 25 25 25 25 25 25

Sensitivity, in % 74.3 83.8 98.1 99 nd 98 100 nd nd

LOD50 (95% confidence interval), 5.7 6.0 nd nd 2.2

836 b Egg powder samples were artificially contaminated with Salmonella Panama.

837 c Poultry meat samples were artificially contaminated with Salmonella Typhimurium in ILS-I and were naturally contaminated with Salmonella spp. in ILS-II.

838 d Contamination levels determined using a Most Probable Number (MPN) technique.

36