Definition of sampling procedures for collective-eating establishments based on the distribution of environmental microbiological contamination on food handlers, utensils and surfaces

Accepted Manuscript Definition of sampling procedures for collective-eating establishments based on the distribution of environmental microbiological contamination on food handlers, utensils and surfaces Antonio Valero, Juan Carlos Ortiz, Gislaine Fongaro, Marta Hernández, David Rodríguez-Lázaro PII:

S0956-7135(17)30023-3

DOI:

10.1016/j.foodcont.2017.01.013

Reference:

JFCO 5420

To appear in:

Food Control

Received Date: 7 December 2016 Revised Date:

8 January 2017

Accepted Date: 18 January 2017

Please cite this article as: Valero A., Ortiz J.C., Fongaro G., Hernández M. & Rodríguez-Lázaro D., Definition of sampling procedures for collective-eating establishments based on the distribution of environmental microbiological contamination on food handlers, utensils and surfaces, Food Control (2017), doi: 10.1016/j.foodcont.2017.01.013. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 8th January 2017

Food Control

based

on

the

distribution

of

RI PT

Definition of sampling procedures for collective-eating establishments environmental

microbiological

SC

contamination on food handlers, utensils and surfaces

Antonio Valeroa*, Juan Carlos Ortizb, Gislaine Fongaroc, Marta Hernándezd,e, David

a

M AN U

Rodríguez-Lázarof*

Department of Food Science and Technology, University of Cordoba, Campus de Rabanales, Edificio Darwin,14014 Córdoba, Spain

b

Ayuntamiento de Madrid, Madrid, Spain

c

Laboratório de Virologia Aplicada, Departamento de Microbiologia, Imunologia e Parasitologia, UFSC, Florianópolis, Brazil. d

e

TE D

Laboratory of Molecular Biology and Microbiology, Instituto Tecnológico Agrario de Castilla y León, Valladolid, Spain. Departamento de Ingeniería Agrícola y Forestal, Tecnología de los Alimentos, E.T.S. Ingenierías Agrarias, Universidad de Valladolid, 34004 Palencia, Spain f

AC C

EP

Division of Microbiology, Department of Biotechnology and Food Science, Universidad de Burgos, Burgos, Spain.

*

Corresponding authors: [email protected] (Antonio Valero); [email protected];

[email protected] (David Rodríguez-Lázaro)

1

ACCEPTED MANUSCRIPT ABSTRACT

2

Environmental sampling has been identified as an effective procedure to verify correct

3

implementation of food safety control systems in catering establishments. At the same

4

time characterization of microbial distribution of environmental contamination could

5

potentially address effective fit-for-purpose sampling procedures. In this study 1,202

6

environmental samples from three types of food catering establishments located in

7

Madrid, Spain were monitored for presence of mesophilic bacteria, Enterobacteriaceae,

8

Staphylococcus aureus and Escherichia coli. Samples corresponded to food-contact

9

utensils, handlers’-contact utensils and food handlers, using 3M™ Petrifilm™ count

10

plates. Contamination routes were identified through the calculation of Spearman

11

correlation coefficients. Further, characterization of statistical distributions of microbial

12

contamination and suggestion of sampling procedures were also performed. Results

13

showed that 53.0% of the samples were positive for at least one of the bacterial group

14

studied and 328 among those (27.1%) with counts between 1-15 CFU/plate.

15

Enterobacteriaceae were present in 62.1% of food handlers’ samples as well as E. coli

16

and S. aureus (7.5% and 26.6%, respectively). Contamination routes from food handlers

17

to handlers’-utensils was identified in a bidirectional way, being it subsequently spread

18

to utensils in contact with foods. Finally, it was shown that the selection of the

19

microbial distribution significantly affected significantly the number of samples needed

20

to detect positives above a certain microbial level. As expected, when negative results

21

are present (high zero counts or left censored data), Poisson-log normal distributions

22

can describe properly the distribution of microbial contamination. However, log normal

23

distributions presented better fit for samples with higher microbial counts and right-

24

censored data (mesophilic bacteria) so that they can be used to describe contamination

25

at high levels. The data and results generated in this study could be of high relevance to

AC C

EP

TE D

M AN U

SC

RI PT

1

2

ACCEPTED MANUSCRIPT 26

food safety authorities to appropriately address environmental sampling procedures in

27

catering establishments.

28

Keywords: food-contact surface, Poisson- log normal distribution, environmental

29

sampling, Enterobacteriaceae, catering establishments

AC C

EP

TE D

M AN U

SC

RI PT

30

3

ACCEPTED MANUSCRIPT 1. Introduction

32

In recent years, the catering sector has been experiencing an increase in technological

33

innovation in correspondence with changes in consumer habits lifestyles, demographic

34

trends, etc., which have increased consumer preferences for healthy, safe, and

35

convenient foods. Legislation in food hygiene at EU level prioritizes control measures

36

to protect public health, making food operators responsible to assure product safety (EC

37

No. 852/2004, EC No. 178/2002 and EC. No. 2073/2005). Regarding catering

38

establishments, important aspects such as the size of establishments and heterogeneity

39

of foods served justify the implementation of prerequisite programs and HACCP

40

systems in food service operations as a part of the food safety management system

41

(Codex Alimentarius Commission, 2003; Jacxsens et al., 2009). However, given the

42

complexity of the food chain and variety of menus and meals prepared, simplified and

43

flexible self-control measures must be required in most cases to increase efficiency and

44

homogeneity of such systems. One useful tool that serves to verify that the system is

45

working properly is the establishment of fit-for-purpose sampling procedures

46

throughout the incoming raw materials, intermediate and end products as well as the

47

result of processing environment monitoring programs (Oses et al., 2012; Lahou et al.,

48

2014).

49

Additionally, previous studies have highlighted the relevance of considering

50

environmental sampling during processing steps as an effective option to control

51

pathogenic contamination sources (Hedberg et al., 2006; Muhterem-Uyar et al., 2015).

52

It includes evaluation of food handlers, utensils and food-contact surfaces which may

53

help to identify contamination sources (cross contaminations via raw materials or

54

biofilms, hygiene failures, etc.). Such contamination can be an intermediate step in

AC C

EP

TE D

M AN U

SC

RI PT

31

4

ACCEPTED MANUSCRIPT transmission of pathogens from their original habitat in the environment (in biofilms,

56

water and organic soil residues) to food contact surfaces and food under processing

57

(Reij et al., 2004; Da Silva & De Martinis, 2013). The macroscopic visual approach is

58

the common procedure for the evaluation of the efficiency of cleaning (Tebutt, 1991;

59

Tebutt et al., 2007), and evaluation of the disinfection methods have been reported in

60

several international organizations’ recommendations (Codex Alimentarius, 1993; EC

61

Reg 1441/2007; Hedberg et al., 2006; Rutula et al., 2008; Sagoo et al., 2003).

62

Nevertheless, in those regulations’ definitions only microbiological and hygienic

63

criteria were established, but no limit values or recommendations were indicated.

64

Evaluation of microbial indicators is crucial for determining the food safety of prepared

65

meals and the study and enumeration of microbial indicators in foods represents the

66

major areas of microbiological analysis in food laboratories (Rodríguez-Lázaro and

67

Hernández, 2015). Indeed, some recent studies used environmental monitoring control

68

to search for potential correlations between microbial indicators and the hygienic-

69

sanitary conditions of several food commodities (Milios et al., 2014; Tomasevic et al.,

70

2016; Zoellner et al., 2016).

71

Additionally, to effectively establish environmental monitoring procedures, prior

72

characterization of the distribution of microbial contamination is needed. There are

73

well-known

74

contamination is distributed in a specific food, in accordance to its composition, nature

75

or contamination level. Statistical distributions can be either continuous (i.e. Log

76

normal), or discrete (i.e. Poisson-log normal, [zero-inflated] Poisson, Poisson-Gamma)

77

being able to reflect microbial concentration in food matrices (Gonzáles-Barrón &

78

Butler, 2011; Gonzáles-Barrón et al., 2010; 2012). The use of log normal has been

79

extensively described to deal with homogenous matrices and usually high concentration

AC C

EP

TE D

M AN U

SC

RI PT

55

statistical

approaches

for deriving distributions

5

describing how

ACCEPTED MANUSCRIPT levels, where bacteria can be described as “continuous” entities. However, in case of

81

censored data (when the observed microbial concentration is only partially known; i.e.

82

concentration values are within a defined range but the true value is unknown), high

83

proportion of negative results or clustering contamination, the use of discrete

84

distributions is more appropriate since log normal distribution does not account for

85

zeros and it can underestimate the proportion of non-defective units in a food lot. The

86

Poisson-log normal distribution considers variability within lots, which is characterized

87

by a Poisson sampling process combined with variability between lots through the

88

assumption that concentration is log-normally distributed (Jongenburger et al., 2012).

89

There are relatively few published data on environmental microbial contamination in

90

food service operations. Characterization of distributions of microbial contamination

91

would help to implement effective sampling procedures which could be used as

92

verification tools of correct implementation of food safety management systems. The

93

present study aimed at evaluating the microbiological contamination on food handlers,

94

food-contact utensils and handlers’-contact utensils during food preparation for

95

collective meals in Spain, as well as to determine contamination routes and their

96

relationships

97

Enterobacteriaceae,

98

characterization of statistical distributions of microbial contamination and suggestion of

99

sampling procedures were also performed.

TE D

M AN U

SC

RI PT

80

microbial

EP

between

coli

and

(aerobic

mesophilic

Staphylococcus

aureus).

bacteria, Further,

AC C

Escherichia

indicators

100

2. Material and methods

101

2.1. Study design and collection of samples from catering establishments

102

Seventy-six catering premises were assayed in this study in 31 primary schools, 29

103

nurseries and 16 nursing homes in Hortaleza Area, Madrid, Spain. Menus were prepared 6

ACCEPTED MANUSCRIPT in situ in 51 centres, while food was prepared in a central kitchen and served by a

105

catering company in 25 centres (Supplementary Table 1). Environmental samples were

106

taken in 183 routine official health veterinary inspections, during one-year period, from

107

three type of samples: food handlers (both hands); utensils in contact with food handlers

108

(10 types) and utensils in contact with food (21 types) (Supplementary Table 2).

109

2.2. Microbiological analyses

110

Bacterial counts were determined using 3M™ Petrifilm™ count plates (3M-UK,

111

Bracknell, Berkshire, UK): 3M™ Petrifilm™ E. coli/Coliform Count Plates E. coli

112

counts; 3M™ Petrifilm™ Enterobacteriaceae Count Plates for Enterobacteriaceae;

113

3M™ Petrifilm™ Aerobic Count Plates for aerobic mesophilic bacterial counts, and

114

3M™ Petrifilm™ Staph Express Count Plates for Staphylococcus aureus. Plates were

115

prepared following the manufacturer’s instructions. Sampling areas corresponded to 20

116

cm2 for Enterobacteriaceae, mesophilic bacteria, and E. coli; and 30 cm2 for

117

enumeration of S. aureus. Briefly, plates were hydrated with 1 mL of 0.1% peptone

118

water, and the top film was carefully lifted avoiding touching the circular growth area.

119

Then the circular gel portion of the top film was put in direct contact with the surface

120

being tested and finally the top and bottom films were re-joined. Plates were

121

individually identified, transported at 4ºC, and incubated at 37ºC during 24 hours for

122

enumeration of Enterobacteriaceae; or 48 hours for mesophilic bacteria, E. coli and S.

123

aureus. In total 1,212 microbiological determinations were done.

124

2.3. Characterization of statistical distributions for describing microbial

125

contamination in catering establishments

126

In the present study, microbial contamination was described through statistical

127

distributions.

AC C

EP

TE D

M AN U

SC

RI PT

104

Seven

actual

datasets

corresponding 7

to

food-contact

utensils

ACCEPTED MANUSCRIPT 128

(Enterobacteriaceae,

mesophilic

129

(Enterobacteriaceae,

130

(Enterobacteriaceae, E. coli and S. aureus) were considered for the distribution fitting.

131

For the sake of comparison between the evaluated distributions, the results from all

132

premises for each microbial group and type of sample were pooled together and models

133

were fitted to each of the seven data sets.

bacteria)

handlers’-contact and

food

handlers’

utensils counts

RI PT

mesophilic

bacteria),

To this aim, semi-quantitative censored data obtained regarding microbial

135

concentration expressed in CFU/plate (no dilutions were performed) were grouped into

136

concentration intervals and number of samples analysed falling into each interval was

137

calculated. To start data processing intervals assigned were estimated as follows

138

(CFU/plate): absence of microorganisms in the sampled area; (i.e. left-censored data <1

139

CFU/plate); interval censored data: 1 – 15; 16-30; 31-50; 51-70; 71-90; 91-110; 111-

140

150; 151-175; 176-200 CFU/plate and right censored data >200 CFU/plate.

M AN U

SC

134

Firstly, concentration data were assumed to follow a log normal distribution. This

142

allows to calculate the mean of the microbial concentration expressed in log counts, and

143

thus describing the variability in concentration between lots (Busschaert et al., 2010;

144

Pouillot and Delignette-Müller, 2010). The number of CFU (Y) present in the Petrifilm

145

plate from sample i and microbial group k (Yik in CFU) can be defined as Yik ~ f (Ø) ~

146

Lognormal (µ i, σi) or log (Yik) ~ Normal (µ i, σi) with mean µ and standard deviation σ

147

(log CFU/plate). The log of the concentration of microorganisms Y can be described

148

through the probability density function (pdf):

149

AC C

EP

TE D

141

1  x−µ  σ 

2

−  1 f (Y ) = e 2 σ 2π

Eq. 1

150

For the sake of comparison, as microorganisms are discrete particles that are very

151

small relative to the size of analytical units typically employed, the use of discrete 8

ACCEPTED MANUSCRIPT distributions to characterize cell concentration in the samples would be suitable. Among

153

them, the Poisson distribution describes the probability of detecting cells, by randomly

154

sampling from a well-mixed system. In this case, it was assumed that microbial

155

contamination is evenly distributed in collected samples and samples are independent

156

from each other. The Poisson-log normal distribution predicts the probability of

157

detecting a positive within a contaminated lot where microbial concentration follows a

158

lognormal distribution. The probability of detecting a positive sample is dependent on

159

the sample size. As counts are represented by CFU/ plate, this distribution will estimate

160

the probability of finding a positive plate, above a certain concentration limit. The use

161

of the Poisson- log normal distribution assumes that there are no losses in the transfer of

162

cells from the surfaces to the Petrifilm plates; the bacterial cells extracted from the

163

surfaces are randomly distributed in the sampled area; and each of the plated cells will

164

become a colony after incubation. As such, the number of bacterial cells present in the

165

aliquot or bacterial count on the Petrifilm from sample i and microbial group k (Yik in

166

CFU) can follow a Poisson distribution (Gonzáles- Barrón & Butler, 2011) Yik ~ f (Ø) ~

167

Poisson (λi). The parameter λ to describe the average concentration per gram in a single

168

sample, so that p(Y|λ) describes the probability of observing Y organisms in a sample

169

with average concentration λ. The integer value of the microbial concentration in the

170

sample (λi) varies according to a lognormal distribution f (Ø) (µ i, σi) (Izsak, 2008;

171

Williams & Ebel, 2012). In the natural log scale, the Poisson- log normal distribution

172

calculates the probability of observing Y organisms in a sample as:

AC C

EP

TE D

M AN U

SC

RI PT

152

1

2πσ x ! ∫

∞

e λ −λ

Y −1

−

( ln λ − µ )2

dλ

173

p (Y ) =

174

Maximum Likelihood Estimates (MLE) were used to estimate parameters of both

175

lognormal and Poisson-log normal distributions. Derivation of these parameters consists

0

e

2σ 2

Eq. 2

9

ACCEPTED MANUSCRIPT 176

on maximizing likehood function ∂ , which results from the product of the individual

177

probability density functions. N

178

∂Y = Π p(Yi )

179

Distributions were fitted to observed data in R v3.2.3 (cran.r-project.org) using

180

the fitdistrplus for censored data (Pouillot & Delignette-Müller, 2010) and poilog

181

(Grøtan V. & Engen S. 2008) packages.

Eq. 3

RI PT

i =1

Mean estimated parameters together with goodness-of-fit indices were obtained.

183

The latter corresponded to the log likelihood (logL), Akaike Information Criterion (AIC)

184

and Bayesian Information Criterion (BIC). The AIC and BIC criteria are measures of

185

the relative quality of statistical models for a given set of data. Given a collection of

186

models for the data, AIC and BIC estimate the quality of each model, relative to each of

187

the other models. Given a set of candidate models for the data, the preferred model is

188

the one with the minimum AIC value.

189

2.3. Estimation of the number of samples needed to detect targeted microbial

190

counts in food service centers

191

Once output distributions were characterized, cumulative distributions were built in MS

192

Excel. Then, the probability of detecting samples above a certain microbial count was

193

derived (Paccp). Finally, the number of samples was estimated at each microbial count,

194

following the methodology stated by Whiting et al. (2006).

M AN U

TE D

EP

AC C

195

SC

182

For the performance of log normal and Poisson-lognormal cumulative

196

distributions, different detection limits were considered for observed negative samples

197

as continuous and discrete counts are assumed, respectively. For the Log normal

198

distribution, the theoretical detection limit was calculated through the area of Petrifilm

199

plate (20 cm2 for Enterobacteriaceae, mesophilic bacteria and E. coli; and 30 cm2 for S. 10

ACCEPTED MANUSCRIPT 200

aureus). The limit of detection was estimated as 1/20 and 1/30 CFU/plate, respectively.

201

To determine the number of samples needed to reject the lot (n) at 95% CL for a two-

202

class sampling plan with sample size n and c=0, the probability of rejection (r) followed

203

a binomial distribution and can be calculated as follows: r = 1 − (1 – Paccp )

205

Rearranging this formula, the required number of samples would be:

206

n =

n

Eq. 4

log (1 – r )

Eq. 5

SC

log (1 – p )

RI PT

204

2.4. Correlation tests between microbial groups and surfaces analysed

208

To find out potential relationships between contamination sources and microbial groups

209

the Spearman correlation coefficient was applied for the interval censored data in the

210

catering establishments. Datasets consisted on microbial counts in food handlers;

211

(H_Ent [Enterobacteriaceae]; H_E coli [E. coli]; H_S. aureus [S. aureus]); food-contact

212

utensils (UF_mesophilic [mesophilic bacteria]; UF_Ent [Enterobacteriaceae]); and in

213

handlers’-contact utensils (UH_mesophilic [mesophilic bacteria]; and UF_Ent

214

[Enterobacteriaceae]). Two significant values of p were considered for the correlation

215

coefficient (r): p≤0.05 and p≤0.01.

216

3. Results and discussion

217

3.1. Microbial contamination of food handlers, handlers’-contact and food-contact

218

utensils in catering establishments

219

The use of microbial indicators can assess the level of hygiene of utensils and surfaces

220

and the quality of disinfection procedures applied. The presence of microbial indicators

221

in high numbers in environmental monitoring samples can highlight potential

AC C

EP

TE D

M AN U

207

11

ACCEPTED MANUSCRIPT deficiencies in the hygienic and sanitary food quality and a consequent reduction of

223

food shelf life. From the total of 1,212 microbiological determinations performed in this

224

study, 650 (53.6%) resulted positive for at least one of the bacterial group studied

225

(mesophilic bacteria, Enterobacteriaceae, E. coli and S. aureus) (Table 1), and 328

226

among those (27.1%) with counts between 1-15 CFU/plate. Regarding mesophilic

227

bacteria, the percentage of high contaminated samples (> 201 CFU/plate) were higher in

228

handlers’-contact than in food-contact utensils (46.2% and 27.0%, respectively).

RI PT

222

On the other hand, Enterobacteriaceae are considered as food quality indicators

230

including E. coli being mainly related to faecal contamination. Generally, the presence

231

of these microorganisms in foods is closely linked with the implementation of

232

inadequate handling practices, inefficient cooking processes, cross-contamination,

233

inadequate personal hygiene of food handlers, equipment and food-contact surfaces as

234

well as inadequate holding time and temperature conditions (<50ºC) (Rodríguez-Caturla

235

et al., 2011). Ninety out of 145 food handlers’ samples (62.1%) were positive for

236

Enterobacteriaceae (Table 1). However, Enterobacteriaceae counts were lower than 15

237

CFU/plate. The presence of Enterobacteriaceae found in this study (62.1%) was

238

remarkably higher than that reported by Djekic et al. (2016) (10.3–15.2%) and by Lues

239

and van Torder (2007) in South Africa where Enterobacteriaceae were present in 40%

240

of food handlers’ samples. Regarding Enterobacteriaceae counts found in contact

241

surfaces, handlers-contact surfaces presented higher percentage of positives (36.3%)

242

than food-contact surfaces (26.2%). Previous studies report that hand contact surfaces

243

are more susceptible to contamination than food contact surfaces (DeVita et al., 2007)

244

and that microbial transfer by hands represents a potential cross-contamination route

245

(Pérez-Rodríguez et al., 2008). The high number of positive samples found in this study

246

could be also attributed to the analytical technique used since 3M™ Petrifilm™ plates

AC C

EP

TE D

M AN U

SC

229

12

ACCEPTED MANUSCRIPT 247

usually have a lower detection limit than other techniques used for evaluation of

248

contamination of surfaces (i.e. swabbing methods) being widely accepted and approved

249

for microbiological analysis in the food and beverage industry (Hooker et al., 2011). The presence of E. coli and S. aureus were also monitored samples from food

251

handlers. Whereas a 7.5% of the samples were positive for E. coli, with counts lower

252

than 50 CFU/plate, S. aureus was detected in 34 out of 128 samples (26.6%) with

253

counts lower than 30 CFU/plate. The presence, although in not high loads of those

254

microbial groups in the hands of the food personnel, and particularly those associated to

255

faecal contamination, highlights potential failures resulting from poor hygiene. Castro et

256

al. (2016) found 11.1% of positive samples for S. aureus in hands of food handlers

257

revealing a high prevalence of antibiotic resistance and virulence determinants among

258

the isolates. In certain circumstances, the hands may represent the most important

259

vehicle of faecal and respiratory microorganisms (Todd et al., 2010). It has been shown

260

that microorganisms, such as S. aureus, E. coli, and Salmonella enterica, can survive on

261

the hands if hygiene measures are not sufficiently appropriate. These findings highlight

262

the importance of the definition of correct food hygiene standards and procedures as

263

well as the periodical evaluation of the cleaning and disinfection procedures in catering

264

establishments.

265

3.2 Correlations between surfaces and microbial groups analysed

266

Potential relationships between microbial groups and surfaces were estimated with the

267

calculation of Spearman correlation coefficient. The datasets corresponding to food

268

handlers counts (H), food- (UF) and handlers’-contact utensils (UH) were analysed and

269

correlation significances are presented in Table 2. Significant correlations between

270

Enterobacteriaceae counts (H_Ent) were found with presence of E. coli and S. aureus

271

(H_E coli and H_S aureus) in food handlers. As expected, utensils in contact with

AC C

EP

TE D

M AN U

SC

RI PT

250

13

ACCEPTED MANUSCRIPT 272

contaminated handlers were also positively correlated with high counts of mesophilic

273

bacteria and Enterobacteriaceae (UH_mesophilic and UH_Ent). Regarding food-contact utensils (UF) they were correlated with all types of

275

samples evaluated. This fact is attributed to cross-contamination events during food

276

handling and/ or inefficient cleaning and disinfection procedures. The food handler and

277

contact with contaminated surfaces are potential causes of cross contamination and,

278

consequently, food poisoning outbreaks (de Oliveira et al., 2014). Recontamination

279

routes and sources (e.g., raw materials, food contact surfaces, food handlers) were

280

revised thoroughly in other studies (Pérez-Rodríguez et al., 2008; Reij et al., 2004)

281

demonstrating their relevance to foodborne disease outbreaks. These information

282

sources should also be incorporated in Quantitative Microbiological Risk Assessments

283

(QMRAs) to perform mitigation strategies and reduce foodborne disease.

M AN U

SC

RI PT

274

According to the results obtained, sequential contamination routes are proposed

285

(Supplementary Figure 1). Microbial contamination in food handlers may be transferred

286

to the utensils in a bidirectional way. Especially, Enterobacteriaceae present in hands

287

are directly correlated to the microbial loads found in the utensils in contact with food

288

handlers (UH_mesophilic and UH_Ent). This contamination is directly correlated with

289

the microbial counts of mesophilic bacteria and presence of Enterobacteriaceae in the

290

utensils in contact with foods (UF). It is therefore concluded that training and formation

291

of food handlers could have a positive impact to avoid the spread of microbial

292

contamination in catering establishments. Previous studies demonstrated that training of

293

food handlers and knowledge acquisition in hygienic food preparation, processing, and

294

distribution of meals is crucial in the prevention of most types of foodborne diseases

295

(Aziz et al., 2013; Bessa Martins et al., 2012).

AC C

EP

TE D

284

296 14

ACCEPTED MANUSCRIPT 3.3 Statistical distributions for the description of microbial contamination in

298

catering establishments

299

Cumulative density functions (cdf) of the estimated distributions for microbial counts in

300

food handlers, food-contact and handlers’-contact utensils are represented in Figures 1 -

301

3. The results showed that distributions were left-shifted thus indicating a high

302

proportion of low microbial counts in the samples. Log normal and Poisson- log normal

303

distributions representing contamination of food handlers were obtained for

304

Enterobacteriaceae, E. coli and S. aureus (Fig. 1). Concentration units were

305

transformed to CFU/cm2 to better interpret graphical representation considering the

306

sampled area per plate (30 cm2 for S. aureus and 20 cm2 for the remaining groups). In

307

all cases the Poisson- log normal distribution fitted better to the observed data at low

308

concentrations since this distribution accounted for the high proportion of zero counts in

309

the observed data. The better adjustment was reflected by the lower values of LogL,

310

AIC and BIC values presented in Table 3. Estimated 95th percentiles (Poisson- log

311

normal / Log normal distributions) were 3.95 and 5 CFU/cm2 for Enterobacteriaceae;

312

0.05 and 0.07 for E. coli and 0.15 and 0.16 for S. aureus. This indicates that specific

313

indicators are not frequently found in routine samples, and therefore hygiene evaluation

314

should be verified testing Enterobacteriaceae. As shown in Figure 1A, most of the

315

counts for Enterobacteriaceae were below 1 CFU/cm2, though right-tailed values were

316

obtained with counts higher than 10 CFU/cm2.

AC C

EP

TE D

M AN U

SC

RI PT

297

317

Regarding food-contact utensils, counts were obtained for aerobic mesophilic

318

bacteria and Enterobacteriaceae (Figure 2A-B). A two-sided simulated distribution was

319

obtained for aerobic mesophilic bacteria, with most of data below than 10 CFU/cm2.

320

However, there were 30% of the counts higher than this concentration (Figure 2A). Log

321

normal distribution showed a better fit to the observed data than the Poisson- log normal 15

ACCEPTED MANUSCRIPT distribution though both distributions slightly overestimated mesophilic counts at low

323

levels. Lower counts were obtained for Enterobacteriaceae in food-contact utensils than

324

those for food handlers (Fig. 2B). This could be attributed to the adhesion of

325

microorganisms to utensils in contact with food and the use of the same utensil for

326

different operations which indicates that they could be a transmission vector of enteric

327

pathogens during food elaboration. As for the food handler counts, the Poisson- log

328

normal showed a better fit.

RI PT

322

Finally, in Figure 3A-B the simulated counts of handlers’-contact utensils are

330

shown. Overall, higher counts of mesophilic bacteria were obtained (>10 CFU/cm2) as

331

well as for Enterobacteriaceae than in food-contact utensils samples. This fact

332

highlights that the main contamination pathway occurred between handlers and

333

handlers’-contact utensils. Despite the low numbers for microbial counts obtained in the

334

collected samples, 95th percentiles of distributions were above the recommended limits

335

for microbial hygiene (Sneed et al., 2004; Solberg et al., 1990).

336

3.4

337

establishments

338

Sampling plans can be rationally designed based on the knowledge on how

339

microorganisms are distributed in the food environment to determine the minimum

340

number of samples above a target concentration limit. This could be of high interest to

341

food authorities when performing an official inspection since most of the non-

342

conformities are based on environmental monitoring results. In Figures 4 – 6 the

343

number of samples needed to obtain one positive above a certain microbial

344

concentration limit (CFU/plate) was calculated for both Log normal and Poisson- log

345

normal distributions. In all cases, the Poisson- log normal distribution yielded a higher

346

proportion of non-defective units (since it is based on integer values that accounts for

procedures

for

assessing

microbial

hygiene

in

catering

AC C

EP

Sampling

TE D

M AN U

SC

329

16

ACCEPTED MANUSCRIPT zero microbial counts), thus, higher number of samples is estimated to detect positive

348

samples. It should be noted that, for illustrative purposes, concentration was set in

349

CFU/plate so that the outputs are related to the number of samples (plates) needed to

350

detect positive counts at different levels. Regarding the assumed distribution,

351

differences between the Poisson- log normal and log normal were remarkable when

352

microbial contamination is low. For instance, contamination of E. coli and S. aureus in

353

food handlers higher than 10 and 15 CFU/plate could yield an increase from 28% up to

354

150% in the number of samples (Fig. 4B – C). Consequently, E. coli and S. aureus were

355

found not suitable for considering them as target groups for setting sampling procedures

356

since the number of samples needed to detect positives is too unrealistic in comparison

357

with Enterobacteriaceae. Noteworthy, the positive correlation between counts in food

358

handlers could lead to establish this group as a useful indicator for assessing the

359

presence of E. coli and S. aureus in the food environment.

360

4. Conclusions

361

The results obtained in the present study on environmental microbial contamination

362

(food handlers, food-contact surfaces and handlers’-contact surfaces) can provide

363

valuable information about the efficacy of cleaning and disinfection procedures used in

364

catering establishments. It was shown that cross-contamination between food handlers

365

and contact utensils could occur in a bidirectional way. Enterobacteriaceae could be

366

used preferentially for the evaluation of cleaning and food processing conditions in

367

catering establishments, and not just for evaluation of potential faecal contamination.

368

Finally, it was shown that the selection of the microbial distribution significantly

369

affected the number of samples needed to detect positives above a certain microbial

370

level. Particularly when negative results are present (high zero counts or left censored

AC C

EP

TE D

M AN U

SC

RI PT

347

17

ACCEPTED MANUSCRIPT data), Poisson-log normal distributions can describe properly how contamination is

372

distributed in the catering establishments evaluated. However, log normal distributions

373

presented better fit for samples with higher microbial counts and right-censored data

374

(mesophilic bacteria) so that they can be used to describe contamination at high levels.

375

These results could be of high usefulness for risk assessors and managers in order to

376

appropriately address environmental sampling procedures in catering establishments.

377

ACKNOWLEDGMENTS

378

This study was supported by the RTA2014-00024-C04-01 from the Spanish Ministry of

379

Economy and Innovation.

M AN U

SC

RI PT

371

AC C

EP

TE D

380

18

ACCEPTED MANUSCRIPT References

382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430

1.

Anonymous, 2002. Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January 2002 laying down the general principles and requirements of food law, establishing the European Food Safety Authority and laying down procedures in matters of food safety. Official Journal of the European Communities, 31, 1-24.

2.

Anonymous, 2004. Regulation (EC) No 852/2004 of the European Parliament and of the Council of 29 April 2004 on the hygiene of foodstuffs. Official Journal of the European Communities, 139, 1-54.

3.

Anonymous (2005) Regulation (EC) No 2073/2005 of the European Parliament and of the Council of 15 November 2005 on microbiological criteria for foodstuffs. Official Journal of the European Communities, 338, 1-26.

4.

Anonymous. (2007). Commission Regulation (EC) No. 1441/2007 of 5 December 2007 amending Regulation (EC) No 2073/2005 on microbiological criteria for foodstuffs. Official Journal of the European Communities, L22, 12-29.

5.

Aziz, S.A.A., & Dahan, H.M. (2013). Food Handlers’ Attitude towards Safe Food Handling in School Canteens. Procedia - Social and Behavioral Sciences, 105, 220-228.

6.

Bessa Martins, R., Hogg, T., & Otero, J.G. (2012). Food handlers’ knowledge on food hygiene: The case of a catering company in Portugal. Food Control, 23, 184-190.

7.

Busschaert, P., Geeraerd, A. H., Uyttendaele, M., & Van Impe, J. F. (2010). Estimating distributions out of qualitative and (semi) quantitative microbiological contamination data for use in risk assessment. International Journal of Food Microbiology, 138, 260-269.

8.

Castro, A., Santos, C., Meireles, H., Silva, J., & Teixeira, P. (2016). Food handlers as potential sources of dissemination of virulent strains of Staphylococcus aureus in the community. Journal of Infection and Public Health, 9, 153-160.

9.

Codex Alimentarius Comission (CAC), 1993. Code of Hygienic Practice for Precooked and Cooked Foods in Mass Catering. CAC/RCP 39-1993

10.

Codex Alimentarius Commission (CAC), 2003. Recommended International Code of practice. General. Princ. Food Hyg. CAC/RCP1-1969 (Rev.. Ed. .4), 1-31.

11.

Da Silva, E.P., & De Martinis, E.C. (2013). Current knowledge and perspectives on biofilm formation: the case of Listeria monocytogenes. Applied Microbiology and Biotechnology, 97, 957968.

12.

de Oliveira, A.B.A., da Cunha, D. T., Stedefeldt, E., Capalonga, R., Tondo, E.C., & Cardoso, M. R. I. (2014). Hygiene and good practices in school meal services: organic matter on surfaces, microorganisms and health risks. Food Control, 40, 120-126.

13.

Djekic, I., Kuzmanovic, J., Andelkovic, A., Saracevic, M., Stojanovic, M.M., & Tomasevic, I. (2016). Effects of HACCP on process hygiene in different types of Serbian food establishments. Food Control, 60, 131-137.

15.

SC

M AN U

TE D

EP

AC C

14.

RI PT

381

Gonzales-Barron, U., & Butler, F. (2011). Characterisation of within-batch and between-batch variability in microbial counts in foods using Poisson-gamma and Poisson-lognormal regression models. Food Control, 22, 1268-1278. Gonzales-Barron, U., Kerr, M., Sheridan, J., & Butler, F. (2010). Count data distributions and their zero-modified equivalents as a framework for modelling microbial data with a relatively high occurrence of zero counts. International Journal of Food Microbiology, 136, 268-277.

16.

Gonzales-Barron, U., Lenahan, M., Sheridan, J., & Butler, F. (2012). Use of a Poisson-gamma model to assess the performance of the EC process hygiene criterion for Enterobacteriaceae on Irish sheep carcasses. Food Control, 25, 172-183.

17.

Grøtan, V. & Engen, S. (2008). poilog: Poisson lognormal and bivariate Poisson lognormal

18.

Hedberg, C.W., Smith, S.J., Kirkland, E., Radke, V., Jones, T.F., Selman, CA., & EHS-Net working Group. (2006). Systematic environmental evaluations to identify food safety differences between outbreak and nonoutbreak restaurants. Journal of Food Protection, 69, 2697-2702.

19

ACCEPTED MANUSCRIPT Hooker, E.A., Allen, S.D., & Gray, L.D. (2011). Comparison of Rayon-Tip Swabs and Film Plates for Use in Collecting and Quantifying Bacteria on Hospital Bed Mattresses. American Journal of Infection Control, 39, E191-E192.

20.

Izsak, R. (2008). Maximum likelihood fitting of the Poisson lognormal distribution. Environmental and Ecological Statistics, 15, 43-156.

21.

Jacxsens, L., Devlieghere, F., & Uyttendaele, M., (2009). Quality Management Systems in the Food Industry, second ed. Sofia, Bulgaria: St Kliment Ohridski University Press.

22.

Jongenburger, I., Bassett, J., Jackson, T., Zwietering, M.H., & Jewell, K. (2012). Impact of microbial distributions on food safety I. Factors influencing microbial distributions and modelling aspects. Food Control, 26, 601-609.

23.

Lahou, E., Jacxsens, L., Van Landeghem, P., & Uyttendaele, M. (2014). Microbiological sampling plan based on risk classification to verify supplier selection and production of served meals in food service operation. Food Microbiology, 41, 60-75.

24.

Lues, J. F. R., & Van Tonder, I. (2007). The occurrence of indicator bacteria on hands and aprons of food handlers in the delicatessen sections of a retail group. Food Control, 18, 326-332.

25.

Milios, K., Drosinos, E.H.; Zoiopoulos, P.E. (2014). Carcass decontamination methods in slaughterhouses: a review. Journal of the Hellenic Veterinary Medical Society, 65, 65-78.

26.

Muhterem-Uyar, M., Dalmasso, M., Bolocan, A.S., Hernandez, M., Kapetanakou, A.E., Kuchta, T., Manios, S.G., Melero, B., Minarovi, J., Nicolau, A.I., Rovira, J., Skandamis, P.N., Jordan, K., Rodríguez-Lázaro, D., Stessl, B., & Wagner, M. (2015). Environmental sampling for Listeria monocytogenes control in food processing facilities reveals three contamination scenarios. Food Control, 51, 94-107.

27.

Osés, S.M., Luning, P.A., Jacxsens, L., Santillana, S., Jaime, I., & Rovira, J. (2012). Microbial performance of food safety management systems implemented in the lamb production chain. International Journal of Food Microbiology, 75, 95-103.

28.

Pérez-Rodríguez, F., Valero, A., Carrasco, E., García-Gimeno, R.M., & Zurera, G. (2008). Understanding and modelling bacterial transfer to foods: a review. Trends in Food Science and Technology, 19, 131–144.

29.

Pouillot, R., & Delignette-Muller, M.L. (2010). Evaluating variability and uncertainty separately in microbial quantitative risk assessment using two R packages. International Journal of Food Microbiology, 142, 330-340.

30.

Reij, M.W., den Aantrekker, W.D., & ILSI Europe Risk Analysis in Microbiology Task Force. (2004). Recontamination as a source of pathogens in processed foods. International Journal of Food Microbiology, 91, 1–11.

31.

Rodríguez-Caturla, M., Valero, A., Carrasco, E., Pérez-Rodríguez, F., Posada-Izquierdo, G.D., & Zurera, G. (2011). Hygienic conditions and microbiological status of chilled ready-to-eat products served in Southern Spanish Hospitals. Food Control, 22, 874–882.

32.

33.

EP

TE D

M AN U

SC

RI PT

19.

AC C

431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480

Rodríguez-Lázaro, D., & Hernández, M. (2015). Future directions for molecular microbial diagnostic methods for the food industry. In N. Cook, M. D'Agostino & K.C. Thompson (Eds) Molecular Microbial Diagnostic Methods Pathways to Implementation for the Food and Water Industry (pp. 19-37). Boca Raton: Academic Press. Rutala, W.A., Weber, D.J., and the Healthcare Infection Control Practices Advisory Committee (HICPAC) (2008). Guideline for Disinfection and Sterilization in Healthcare Facilities. Atalanta: CDC Press.

34.

Sagoo, S.K., Little, C.L., Griffith, C.J., & Mitchell, R.T. (2003). Study of cleaning standards and practices in food premises in the United Kingdom. Communicable Disease and Public Health, 6, 6-17.

35.

Sneed, J., Strohbehn, C., Gilmore, S.A. & Mendonça, A, (2004) Microbiological evaluation of foodservice contact surfaces in Iowa assisted-living facilities. Journal of the American Dietetic Association, 104, 1722–1724.

20

ACCEPTED MANUSCRIPT Solberg, M., Buckalew, J.J., Chen, C.M., Schaffner, D.W., O’Neill, K., McDowell, J., Post, L.S., & Boderck, M. (1990). Microbial safety assurance system for foodservice facilities. Food Technology, 44, 68–73.

37.

Tebbutt, G.M. (1991) Development of standardized inspections in restaurants using visual assessments and microbiological sampling to quantify the risks. Epidemiology and Infection, 107, 393-404.

38.

Tebbutt, G., Bell, V., & Aislabie, J. (2007) Verification of clearing efficiency and its possible role in programmed hygiene inspections of food businesses undertaken by local authority officers. Journal of Applied Microbiology, 102, 1010-1017.

39.

Todd, E., Michaels, B., Greig, J., Smith, D., & Bartleson, C. (2010). Outbreaks where food workers have been implicated in the spread of foodborne disease. Part 9. Washing and drying of hands to reduce microbial contamination. Journal of Food Protection, 73, 1937–1955.

40.

Tomasevic, I., Kuzmanovic, J., Andelkovic, A. et al. (2016) The effects of mandatory HACCP implementation on microbiological indicators of process hygiene in meat processing and retail establishments in Serbia. Meat Science, 114, 54-57

41.

Whiting, R.C., Rainosek, A., Buchanan, R.L., Miliotis, M., LaBarre, D., Long, W., Ruple, A., & Schaub, S. (2006). Determining the microbiological criteria for lot rejection from the performance objective or food safety objective. International Journal of Food Microbiology, 110, 263-267.

42.

Williams, M.S., & Ebel, E.D. (2012). Methods for fitting the Poisson-lognormal distribution to microbial testing data. Food Control, 27, 73-80.

43.

Zoellner, C., Venegas, F., Churey, J.J. (2016). Microbial dynamics of indicator microorganisms on fresh tomatoes in the supply chain from Mexico to the USA International Journal of Food Microbiology, 238, 202-207.

EP

TE D

M AN U

SC

RI PT

36.

AC C

481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504

21

ACCEPTED MANUSCRIPT 505

Supplementary Table 1. Type of premises

Premise

Primary school

Catering service Type

Number

On-site kitchen service

21

Central kitchen service

10

On-site kitchen service Nursery

Central kitchen service Central kitchen service

Nursing home

14

15 29

16

SC

On-site kitchen service

RI PT

31

Central kitchen service

0

16

AC C

EP

TE D

M AN U

506

22

ACCEPTED MANUSCRIPT Supplementary Table 2. Type of samples collected in this study. Handlers’ -contact utensils

Food-contact utensils

Hands’ surface

Toilet door knob

Inner surface pot

Reception surface

Inner surface blender

4ºC storage room door knob

Cook trowel

Door knob

RI PT

Food handlers

Mincer

Blender handle

Ladle surface

Knife handle

Working surface

Deep fryer handle

Cutting table

Microwave switch Oven door

Blender Jar

SC

507

Jar

Serving equipment

M AN U

Thermo jar button

Crockery Thermo Jar

Catering chamber Tray Blender Saucepan Strainer Skimmer Cutter blade Knife blade

AC C

508 509

EP

TE D

Internal part of the thermo jar

23

ACCEPTED MANUSCRIPT 510

Table 1. Number of samples and concentration ranges corresponding to the different

511

sample types and microbial groups analysed in the catering establishments. Microbial group

No. Samples

Concentration range (CFU/plate)

Food handlers

Enterobacteriaceae

55

<1

71

1 - 15

5

16 - 30

2 3 9 E. coli

123 8

SC

1

RI PT

Type of sample

S. aureus

TE D AC C

EP

Enterobacteriaceae

Handlers-contact utensils

Mesophilic bacteria

Enterobacteriaceae

24

> 201 <1

1 - 15

16 - 30

31 - 50

94

<1

M AN U

Mesophilic bacteria

51 - 70

1

33

Food-contact utensils

31 - 50

1 - 15

1

16 - 30

35

<1

105

1 - 15

42

16 - 30

22

31 - 50

16

51 - 70

10

71 - 90

4

91 - 110

5

111 - 150

2

151 - 175

89

>201

225

<1

69

1 - 15

8

16 - 30

2

31 - 50

1

71 - 90

6

<1

19

1 - 15

10

16 - 30

4

31 - 50

4

51 - 70

1

71 - 90

3

91 - 110

1

111 - 150

1

176 - 200

42

>201

51

<1

23

1 - 15

3

16 - 30

ACCEPTED MANUSCRIPT 1

71 - 90

2

>201

AC C

EP

TE D

M AN U

SC

RI PT

512

25

ACCEPTED MANUSCRIPT

Table 2. Spearman correlation coefficients obtained for the microbial count ranges in the catering establishments. H_ent (Enterobacteriaceae

514

handlers’ counts); H_E coli (E. coli handlers’ counts); H_S. aureus (S. aureus handlers’ counts); UF_mesophilic (mesophilic bacteria counts in

515

food-contact utensils); UF_ent (Enterobacteriaceae counts in food-contact utensils); UH_mesophilic (mesophilic bacteria counts in handlers-

516

contact utensils); UF_ent (Enterobacteriaceae counts in handlers-contact utensils). H_S aureus

H_Ent-col

1.000

0.434**

0.624**

H_E coli

0.434**

1.000

0.698**

H_S aureus

0.624**

0.698**

1.000

UF_mesophilic

0.717**

0.170*

0.436**

UF_ent-col

-

-

UH_mesophilic

0.799**

-

UH_ent-col

0.896**

-

SC

H_E coli

UF_mesophilic

UF_ent

UH_mesophilic

UH_ent

0.717**

-

0.799**

0.896**

0.170*

-

-

-

0.436**

-

-

-

1.000

0.780**

0.893**

0.654**

-

0.780**

1.000

-

-

-

0.893**

-

1.000

0.818**

0.654**

-

0.818**

1.000

EP

TE D

M AN U

H_ent

-

AC C

517 518 519 520

Microbial counts

RI PT

513

*significance was obtained at 5% level (p<0.05) **significance was obtained at 1% level (p<0.01) - Non significant correlations at 5% level (p>0.05)

26

ACCEPTED MANUSCRIPT

Table 3. Estimated parameters (mean and standard deviation, SD) and standard errors (SE) of the log normal and Poisson-lognormal distributions

522

for the microbial counts in Petrifilm plates (log CFU/cm2). Goodness-of-fit indices are represented by the log-likelihood (LogL), Akaike

523

Information Criterion (AIC) and Bayesian Information Criterion (BIC) Microbial group

Distribution

Parameters statistical distributions

SC

Sample

RI PT

521

Food-contact utensils

mean

SE mean

SD

SE SD

-0.602

0.107

0.945

0.096

-220.76 445.52

452.97

Poisson-Log normal -0.858

0.364

1.056

0.283

-52.56

107.84

Log Normal

1.451

0.072

1.232

0.069

-615.17 1234.33 1241.93

Poisson-Log normal

1.219

0.215

1.159

0.176

-260.96 525.92

535.46

Log Normal

-0.444

0.208

1.216

0.197

-77.58

159.15

163.92

Poisson-Log normal -1.395

0.936

1.560

0.352

-83.40

170.80

171.13

2.103

0.184

1.485

0.188

-156.05 316.10

321.19

2.121

0.189

1.622

0.106

-351.38 706.76

720.99

0.259

0.099

1.052

0.093

-179.59 363.19

369.14

Poisson-Log normal -0.179

0.553

1.267

0.343

-131.56 267.11

269.94

Log Normal

-2.035

0.689

1.410

0.425

-43.34

90.68

96.49

Poisson-Log normal -2.199

1.469

1.347

0.640

-24.14

52.28

49.53

Log Normal

-0.386

0.118

0.637

0.126

-78.12

160.23

165.89

Poisson-Log normal -0.277

0.696

0.535

0.424

-45.28

94.55

92.90

Log Normal

M AN U

Enterobacteriaceae

Mesophilic bacteria

TE D

Handlers’ -contact utensils Enterobacteriaceae

Aerobic mesophilic bacteria Log Normal

E. coli

S. aureus

AC C

Enterobacteriaceae

EP

Poisson-Log normal

Food handlers

Goodness-of-fit indices

Log Normal

27

LogL

AIC

109.12

BIC

ACCEPTED MANUSCRIPT

Supplementary Figure 1. Relationships found between the evaluated samples corresponding to the significance of Spearman correlation tests.

525

H_Ent (Enterobacteriaceae handlers’ counts); H_E coli (E. coli handlers’ counts); H_S. aureus (S. aureus handlers’ counts); UF_mesophilic

526

(mesophilic bacteria counts in food-contact utensils); UF_Ent (Enterobacteriaceae counts in food-contact utensils); UH_mesophilic (mesophilic

527

bacteria counts in handlers-contact utensils); UF_Ent (Enterobacteriaceae counts in handlers’-contact utensils).

TE D

M AN U

SC

RI PT

524

UF_Ent

UF_mesophilic

H_Ent

AC C

H_S aureus

EP

H_E coli

UH_mesophilic

UH_Ent 28

ACCEPTED MANUSCRIPT 528

Figure 1. Fitted lognormal and Poisson-log normal distributions for Enterobacteriaceae

529

(A), E. coli (B) and S. aureus (C) handlers’ counts.

530

A) 1.00 0.90

RI PT

0.80

0.60 0.50

Observed data

0.40

Lognormal distribution

SC

Probability

0.70

Poisson-lognormal distribution

0.30 0.20

M AN U

0.10 0.00 0

531 532

1

2

3 4 5 6 7 2 Microbial concentration (CFU/cm )

B) 1.00

TE D

0.95

0.85

EP

Probability

0.90

AC C

0.80

8

9

10

Observed data Lognormal distribution Poisson-lognormal distribution

0.75 0.70

0.0

0.5

1.0 1.5 2.0 Microbial concentration (CFU/cm2)

533 534 535 536 537 538 539 540 541 542 29

2.5

3.0

ACCEPTED MANUSCRIPT 543

C) 1.00 0.90 0.80

0.60 0.50

Observed data

RI PT

Probability

0.70

Lognormal distribution

0.40

Poisson-lognormal distribution

0.30 0.20

SC

0.10 0.00 0

0.1

0.2

0.3 0.4 0.5 0.6 0.7 Microbial concentration (CFU/cm2)

AC C

EP

TE D

M AN U

544 545

30

0.8

0.9

1

ACCEPTED MANUSCRIPT 546 547 548

Figure 2. Fitted lognormal and Poisson-log normal distributions for mesophilic bacteria (A) and Enterobacteriaceae (B) food-contact utensils’ counts. A) 1.00 0.90 0.80

RI PT

Probability

0.70 0.60 0.50

Observed data

Lognormal distribution

0.40

Poisson-lognormal distribution

0.30

SC

0.20 0.10 0.00

549 550

1

2

3

4

5 6 7 8 9 10 11 Microbial concentration (CFU/cm2)

M AN U

0

B) 1.00 0.95 0.90

TE D

Probability

0.85 0.80 0.75 0.70

0.60

AC C

0.55

13

14

15

Observed data Lognormal distribution Poisson-lognormal distribution

EP

0.65

12

0.50

0.0

551 552

0.5

1.0

1.5 2.0 2.5 3.0 3.5 Microbial concentration (CFU/cm2)

31

4.0

4.5

5.0

ACCEPTED MANUSCRIPT 553 554

Figure 3. Fitted lognormal and Poisson-log normal distributions for mesophilic bacteria

555

(A) and Enterobacteriaceae (B) handlers-contact utensils’ counts.

556

A) 1.00

RI PT

0.90 0.80

0.60 0.50 0.40

SC

Probability

0.70

Observed data

0.30

Lognormal distribution

M AN U

0.20 0.10 0.00 0

557 558

1

2

3

4 5 6 7 8 Microbial concentration (CFU/cm2)

B) 1.00

0.80 0.70 0.60 0.50

10

11

12

Observed data

EP

Probability

9

TE D

0.90

Poisson-lognormal distribution

0.40

AC C

0.30

Lognormal distribution Poisson-lognormal distribution

0.20 0.10 0.00

0

1

2

3 4 5 6 7 Microbial concentration (CFU/cm2)

559 560 561

32

8

9

10

ACCEPTED MANUSCRIPT Figure 4. Number of samples needed to obtain one positive above a certain microbial

563

concentration (CFU/plate) for Enterobacteriaceae (A), E. coli (B) and S. aureus (C) in

564

handlers’ counts estimated by the fitted Log normal and Poisson-log normal

565

distributions.

566

A)

RI PT

562

160

120

SC

100 80

Poisson-log normal

60

M AN U

Number of samples

140

40 20 0 0

50

100

150

200

Log normal

250

Microbial counts (CFU/plate)

567 B)

TE D

568

800

500 400

EP

600

AC C

Number of samples

700

Poisson-log normal

300

Log normal

200 100

0

0

10

20

30

Microbial counts (CFU/plate)

569 570 571 572 33

40

ACCEPTED MANUSCRIPT 573 574

C) 1200

800

RI PT

Number of samples

1000

600

Poisson-log normal Log normal

400

0 0

5

10

15

20

M AN U

Microbial counts (CFU/plate)

SC

200

575

AC C

EP

TE D

576

34

ACCEPTED MANUSCRIPT Figure 5. Number of samples needed to obtain one positive above a certain microbial

578

concentration (CFU/plate) for mesophilic bacteria (A) and, Enterobacteriaceae (B) in

579

food-contact utensils’ counts estimated by the fitted Log normal and Poisson-log normal

580

distributions.

581

A)

RI PT

577

18 16

SC

12 10 8

Poisson-log normal Log normal

6

M AN U

Number of samples

14

4 2 0 0

50

100

150

200

250

Microbial counts (CFU/plate)

582 B)

TE D

583

800

500 400

EP

600

AC C

Number of samples

700

Poisson-log normal

300

Log normal

200 100 0

0

20

40

60

Microbial counts (CFU/plate)

584 585 586 587 35

80

ACCEPTED MANUSCRIPT Figure 6. Number of samples needed to obtain one positive above a certain microbial

589

concentration (CFU/plate) for mesophilic bacteria (A) and Enterobacteriaceae (B) in

590

handlers’-contact utensils’ counts estimated by the fitted Log normal and Poisson-log

591

normal distributions.

592

A)

RI PT

588

7

SC

5 4

Poisson-log normal

3 2 1 0 0

50

100

M AN U

Number of samples

6

150

200

Log normal

250

Microbial counts (CFU/plate)

593 B)

TE D

594

350

200 150

EP

250

Poisson-log normal

AC C

Number of samples

300

Log normal

100 50

0

0

50

100

150

200

Microbial counts (CFU/plate)

595

36

250

ACCEPTED MANUSCRIPT Highlights: •

Microbial environmental monitoring of mass catering establishments was assessed.

•

Enterobacteriaceae could potentially serve as indicators of microbial

•

RI PT

contamination. Poisson- log normal distribution could better describe contamination at low levels.

Contamination routes between food handlers and surfaces were identified.

•

Environmental sampling procedures could be implemented in mass catering.

AC C

EP

TE D

M AN U

SC

•