International Journal of Food Microbiology 37 (1997) 215–219

Short Communication

Risk factors for contamination of smoked salmon with Listeriamonocytogenes during processing

a , a b a*Liv Marit Rørvik , Eystein Skjerve , Bjørn Røthe Knudsen , Magne YndestadaNorwegian College of Veterinary Medicine, Department of Pharmacology, Microbiology and Food Hygiene, 0033 Oslo, Norway

bDirectorate of Fisheries, Department of Quality Control, Trondheim, Norway

Received 24 February 1997; accepted 9 May 1997

Abstract

Forty smoked salmon processing plants were examined for the occurrence of Listeria monocytogenes and other Listeriaspp. in the smoked salmon and the drains. L. monocytogenes was detected in smoked salmon from 13 (33%) and in thedrains samples from 25 (63%) of the plants. Other Listeria spp. were found in smoked salmon samples from 16 (40%) andin the drains of 30 (75%) of the plants. Multivariate analyses of data on hygiene, management, production facilities of theplants and bacteriological results showed that job rotation was the strongest expressed risk factor for isolation of L.monocytogenes from the smoked salmon (hazard ratio, HR 5 11.0, p 5 0.002). Well-maintained facilities (HR 5 0.31,p 5 0.064) and use of vats for salting of the fillets (HR 5 0.33, p 5 0.109), showed a preventive effect. L. monocytogenes inthe drains was found to be a sensitive predictor for the presence of L. monocytogenes in the smoked salmon. In general,detection of other Listeria spp. in the smoked salmon or the drains could not be demonstrated to have any association withdetection of L. monocytogenes. 1997 Elsevier Science B.V.

Keywords: Listeria monocytogenes; Risk factors; Smoked salmon; Processing plants

1. Introduction source of one sporadic case, and smoked cod roe ofanother case (Facinelli et al., 1989; Fredriksen,

After the connection between Listeria monocyto- 1991). Recently, products of rainbow trout weregenes contaminated foods and human listeriosis was assumed to be the source of an outbreak of listeriosis

¨established in the 1980s, a range of food items have in Sweden (Eklow et al., 1995).been linked to outbreaks and sporadic cases of L. monocytogenes has been isolated from severallisteriosis (Facinelli et al., 1989; Schuchat et al., seafood products (Dillon and Patel, 1992), and the1991). Raw fish and shellfish were epidemiologically bacteria have been found sporadically in smokedassociated with an outbreak of listeriosis (Lennon et salmon (Jemmi, 1993; Rørvik and Yndestad, 1991).al., 1984), undercooked fish was assumed to be the L. monocytogenes can grow well on vacuum-packed,

smoked salmon during storage at 48C (Guyer and*Corresponding author. Fax: 47 22964850. Jemmi, 1991; Rørvik et al., 1991), and contaminated

0168-1605/97/$17.00 1997 Elsevier Science B.V. All rights reserved.PII S0168-1605( 97 )00057-3

216 L.M. Rørvik et al. / International Journal of Food Microbiology 37 (1997) 215 –219

smoked salmon may therefore represent a health accordance with Nordic Committee on Food Analy-hazard. An earlier investigation in a smoked salmon ses (1990), using the two-step enrichment method.processing plant demonstrated that one L. monocyto- A quantitative determination of L. monocytogenesgenes strain had colonized the smokehouse, and in the vacuum-packed, smoked salmon samples wascontaminated the smoked salmon during the manu- performed by spreading 0.1 ml of the 1:10 smokedfacturing process (Rørvik et al., 1995). The contami- salmon/LEB 1 (Listeria Enrichment Broth, CM863nation routes for L. monocytogenes were not de- and supplement, SR142E, Oxoid, Basingstoke, UK)termined, but slaughtered salmon and seawater did mixture, before incubation, directly on Oxford agarnot seem to be important sources of the bacteria. (Listeria Selective Agar base and supplement,

The main aim of this study was to identify risk SR14OE, Oxoid). Incubation and identification offactors for contamination with L. monocytogenes suspected colonies were as described in the methodduring the production of smoked salmon. The study mentioned above.also aimed at investigating whether L. monocyto- Serotyping of L. monocytogenes was performedgenes in the drains, or other Listeria spp. in smoked with Bacto-Listeria-O antisera (Difco Laboratories,salmon or the drains, had any predictive value for the Detroit, MI, USA).occurrence of L. monocytogenes in the smoked In order to obtain an indication of the presence ofsalmon. Listeria spp. in the processing environment, two

cotton plugs were placed in the drains in themanufacturing rooms for two hours on one occasion.

2. Materials and methods The plugs were collected and incubated in 225 ml ofLEB 1. Further analysis followed the method de-

2.1. Processing plants scribed above, except for the quantitative determi-nation.

Forty-three cold-smoked salmon processing plantssituated on the coast of Norway were selected for the 2.3. Hygienic and management factorsstudy. Inclusion criteria were as follows. Processingplants situated in the middle and southern part of Information on various factors related to hygiene,Norway, and inspected and approved by the Directo- management, and production facilities, was collectedrate of Fisheries in Norway. They produced smoked during one inspection and interview visit at eachsalmon continuously enough to make several sam- processing plant. Experienced inspectors from thepling periods possible. Drop-outs were processing Directorate of Fisheries used a detailed questionnaireplants in which production ceased during the study, to standardize the information. A manager from thebefore at least two smoked salmon samples and the factory assisted during the collection of data. Neitherenvironment sample had been collected. Thus, the the inspector nor the manager knew the results of thecompleted study included 40 processing plants. The bacteriological analyses when the questionnaire wasinvestigation was carried out over a period of seven filled out.months (June to December). The questionnaire covered (1) general information

concerning geographic location, size, types of pro-2.2. Bacteriological examinations duction, water supplies, and application of HACCP

(Hazard Analysis Critical Control Points) systems;A 50 to 100 g mass of pooled samples of cuts of (2) production facilities, possibilities for cross con-

smoked salmon from the slicer were collected on tamination, state of repair; (3) information about thethree occasions at each processing plant, and ex- employees, including hygiene training, use of work-amined for L. monocytogenes and other Listeria spp. ing clothes, use of gloves etc.; (4) manufacturingIn processing plants which did not normally slice the routines, such as rotation of assigned duties, saltingsmoked salmon, cuts from five to ten manufactured procedures, smoking conditions, elimination of smallsmoked salmon sides were collected for each sample bones from the fillets, use of slicing, routines for(50 to 100 g). On each occasion, 25 g of the hand hygiene and footwear disinfection; (5) origincollected cuts were examined for Listeria spp. in and handling of raw materials; and (6) washing and

L.M. Rørvik et al. / International Journal of Food Microbiology 37 (1997) 215 –219 217

disinfection routines, such as the use of foam, high (63%) of the processing plants. The bacteria werepressure, manual cleaning, types of detergent and found exclusively in the drains in thirteen (33%) ofdisinfectant, frequency of disinfection, and elimina- the processing plants, while only the smoked salmontion of water after cleaning. samples were contaminated in two plants. Totally, L.

monocytogenes was found in one or more samplesfrom 27 of the processing plants (67%).

2.4. Statistical analysesFrom seventeen (63%) of the L. monocytogenes-

positive processing plants only serotype 1 wasThe information from the bacteriological inves-

isolated, from four (15%) only serotype 4, whiletigations and the questionnaires was coded and made

both serotypes were found in six (22%) of theavailable for computerized analyses using the Excel

contaminated plants.spread sheet (Microsoft Corp., Redmond, WA, USA).

Other Listeria spp., including L. innocua and L.Data were analysed using the statistical packages

seeligeri, were isolated from 32 of the processingEpi-Info (version 6.0, CDC/WHO, Atlanta, GA,

plants (80%). The smoked salmon samples fromUSA) and EGRET (SERC, Seattle, WA, USA).

sixteen (40%) of the plants, and the samples fromUnivariate analysis of all risk variables was done

the drains of 30 (75%) of the plants contained theusing Epi-Info. Variables with a p-value of , 0.15

bacteria. L. seeligeri was detected almost as often aswere brought into EGRET for multivariate analysis.

L. innocua.The final models were built using the Cox regression

The main results from univariate analysis of riskand multiple logistic regression approach (Selvin,

factors are shown in Table 1. Only variables demon-1996).

strating a p-value , 0.15 are shown. A relative riskThe predictive power of bacteriological analyses

(RR) . 1 indicates a positive association, while afor other Listeria spp. and for L. monocytogenes in

RR , 1 indicates a preventive effect.the drains, regarding the occurrence of L. monocyto-

Some of the risk factors listed in Table 1 deservegenes in the smoked salmon, was tested using Epi-

to be more fully explained. Job rotation refers toInfo.

rotation of assigned duties in the smoked salmonprocessing facilities, i.e., between departments suchas filleting department, packaging department and

3. Results and discussion others. In contrast, rotation within each department,or to other productions, did not appear as risk

The 40 cold-smoked salmon processing plants factors.included in the study produced from two to 650 tons Cleaning during production includes variousof smoked salmon each year (median 52.5 tons). routines, but usually washing without detergent orOnly about half of the production facilities were disinfectant before the breaks. Facilities built fororiginally built for the production of smoked salmon. smoked salmon processing means that they wereThirty-nine of the plants produced other marine food originally built for this purpose, while most of theitems as well, like fresh fish, fish fillets, other types plants were rebuilt for smoked salmon production.of smoked fish, and gravad salmon, often in the same Thawing and washing in separate vats mean thatproduction facilities. after thawing the unfilleted salmon was moved to a

L. monocytogenes was detected in smoked salmon new vat for washing.samples from thirteen (33%) of the plants. From one The final model from Cox regression containedof the L. monocytogenes-positive processing plants, only three predictor values (Table 1): job rotation inall three smoked salmon samples were contaminated, the smoked salmon processing facilities, well-main-from four plants two samples were contaminated, tained facilities, and salting in vats. A hazard ratioand from the rest (eight) only one sample contained (HR) . 1 implies a positive association, while athe bacteria. L. monocytogenes was found only after HR , 1 represents a preventive effect. All three wereenrichment, i.e., the concentration of the bacteria was in the group showing the lowest p-values in uni-less than 100 colony forming units /g. variate analysis. A different analytical approach,

L. monocytogenes was detected in the drains in 25 using multiple logistic regression, gave almost the

218 L.M. Rørvik et al. / International Journal of Food Microbiology 37 (1997) 215 –219

Table 1Risk factors for the occurrence of Listeria monocytogenes in smoked salmon, results from univariate analysis in Epi-Info and frommultivariate Cox regression

Variable Univariate Multivariate

Relative risk (95% C.I.) p Hazard ratio (95% C.I.) p

Well-maintained facilities 0.24 (0.09–0.64) 0.0035 0.31 (0.09–1.07) 0.064Job rotation 4.50 (1.14–17.749) 0.0099 11.0 (1.4–85.5) 0.002Salting fillets in vats 0.27 (0.09–0.84) 0.011 0.33 (0.086–1.28) 0.109Cleaning of equipment during production 0.32 (0.16–0.66) 0.031 – – –Thawing and washing of salmon in separate vats 3.11 (1.52–6.35) 0.031 – – –Salting fillets in small vessels 2.72 (1.16–6.4) 0.032 – – –Purpose-built facilities 0.36 (0.13–0.99) 0.034 – – –Footwear disinfection, general 2.57 (1.13–5.85) 0.052 – – –Footwear disinfection at the packaging area 2.52 (1.14–5.58) 0.07 – – –Refrigeration of product directly after smoking 0.40 (0.18–0.88) 0.075 – – –Manual vacuum-packaging 3.3 (0.84–12.91) 0.08 – – –Plant situated in rural area 2.7 (1.25–5.85) 0.09 – – –Cleaning routines also for the washing equipment 0.46 (0.20–1.07) 0.12 – – –Use of disposable protective clothing 4.0 (0.59–27.02) 0.12 – – –

Results are expressed as relative risk (95% confidence interval, C.I.) with Mantel–Haenschel or Fisher p-values for dichotomous variables,and as hazard ratio (95% C.I.) and p-values.

same model, but also with cleaning during pro- plants. The predominance of serotype 1 is in accord-duction as a variable in the model. Cox regression ance with other investigations on smoked salmongives more easily communicable results, and is (Jemmi, 1993).considered to be a more conservative approach than The hygienic and management practices revealedlogistic regression, and thus the main focus in the by the multivariate analysis to be risk factors arefollowing will be on the Cox regression results. biologically relevant, but their interpretation should

The finding of L. monocytogenes in the drains be carried out with caution. The final models presentappeared as a good predictor for finding L. mono- the risk factors, attention to which is most likely tocytogenes in smoked salmon, with a relative risk of reduce the level of contamination in general, while3.3 (95% C.I. 0.84–12.9). The finding of other additional specific advice might be given at eachListeria spp. in the drains (RR 5 0.74, C.I. 0.29– processing plant.1.91) or in smoked salmon (RR 5 0.94, C.I. 0.19– Rotation of assigned duties among departments in4.29) demonstrated no association with the presence the smoked salmon processing facilities was the mostof L. monocytogenes in the smoked salmon itself. strongly expressed risk factor for isolating L. mono-

L. monocytogenes contamination of the smoked cytogenes from the smoked salmon (HR 5 11.0, p 5

salmon from 33% of the smoked salmon processing 0.002). This finding can be easily connected with theplants appears to be a high level compared with the hazards involved when the same workers move fromresults of earlier screenings of Norwegian smoked one department to another in the smoked salmonsalmon, where samples from about 10% of the processing plant, often with limited or no precautionsinvestigated plants contained the bacteria (Rørvik taken to avoid spread of bacteria.and Yndestad, 1991; Rørvik, unpublished data). Processing plants in which production facilitiesThese results were, however, based on a single were in a good state of repair had a lower frequencysample of vacuum-packed, smoked salmon from of L. monocytogenes contamination than those witheach processing plant, while the present study in- moderate or heavy wear and tear. This may becluded three pooled samples. attributed to easier and more effective cleaning and

L. monocytogenes serotype 1 was the predominant disinfection.serotype found in this study (85%), although The risk of finding L. monocytogenes was di-serotype 4 was isolated from 37% of the processing minished when vats were used for storage of the

L.M. Rørvik et al. / International Journal of Food Microbiology 37 (1997) 215 –219 219

salted salmon fillets, in contrast to small vessels, or from the Directorate of Fisheries for assistance withtrays or grates in stacks. The reason for this is not the interviews. The project was supported by a grantimmediately clear, but may be a question of cleanli- from the Norwegian Research Council.ness. Cleaning of the production line once or moreduring the daily production gave a lowered risk forL. monocytogenes contamination only in the multiple Referenceslogistic regression analysis. In most of the processingplants, cleaning included washing without detergent Dillon, R., Patel, T., 1992. Listeria in seafoods: A review. J. Food

Protect. 55, 1009–1015.or disinfectant, mostly before or after the breaks.¨Eklow, A., Danielsson-Tham, M.-L., Ericsson, H., Loncarevic, S.,Some of the factors with an effect ( p , 0.15) inUnnerstad, H., Tham, W., 1995. Listeriosis in the province of

the univariate analysis, while not ending up in the ¨Varmland. Abstracts Food 95, Uppsala, October 1995.final model because of low prevalence, might also be Facinelli, B.,Varaldo, P.E., Toni, M., Casolari, C., Fabio, U., 1989.biologically important. The limited number of Ignorance about Listeria. Br. Med. J. 299, 738.

Fredriksen, W., 1991. Listeria epidemiology in Denmark 1981–facilities included in the study gives a relatively low1990. In: Proc. Int. Conf. Listeria and Food Safety. Laval,statistical power to isolate risk factors which are notFrance, pp. 48–49.

of major importance. Guyer, S., Jemmi, T., 1991. Behaviour of Listeria monocytogenesThe risk of finding L. monocytogenes in the during fabrication and storage of experimentally contaminated

smoked salmon was positively associated with the smoked salmon. Appl. Environ. Microbiol. 57, 1523–1527.Jemmi, T., 1993. Listeria monocytogenes in smoked fish: anfinding of L. monocytogenes in the drains (RR 5

overview. Arch. Lebensmitt. Hyg. 44, 10–13.3.30). L. monocytogenes was, however, often foundLennon, D., Lewis, B., Mantell, C., Becroft, D., Dove, B., Farmer,

in the only sample from the drains and not in the K., Tonkin, S., Yeates, N., Stamp, R., Mickleson, K., 1984.three smoked salmon samples. Monitoring of L. Epidemic perinatal listeriosis. Ped. Infect. Dis. 3, 30–34.monocytogenes in the drains will probably reveal a Nordic Committee on Food Analyses, 1990. Listeria monocyto-

genes. Detection in foods. Leaflet no. 136, Esbo, Finland.problem in the factory when present, but detection ofPritchard, T.J., Flanders, K.J., Donnelly, C.W., 1995. Comparisonthe bacteria in the drains does not necessarily

of the incidence of Listeria on equipment versus environmen-indicate a real L. monocytogenes problem. Pritchard tal sites within dairy processing plants. Int. J. Food Microbiol.et al. (1995) found that Listeria spp. contamination 26, 375–384.in the environment of dairy processing plants did not Rørvik, L.M., Skjerve, E., Yndestad, M., 1991. Growth of Listeria

monocytogenes in vacuum-packed, smoked salmon, duringlead to the same level of contamination of thestorage at 48C. Int. J. Food Microbiol. 14, 111–118.equipment. The presence of L. monocytogenes in the

Rørvik, L.M., Yndestad, M., 1991. Listeria monocytogenes indrains is a possible indicator or warning that L. foods in Norway. Int. J. Food Microbiol. 13, 97–104.monocytogenes could be present in the products. Rørvik, L.M., Caugant, D.A., Yndestad, M., 1995. Contamination

Other Listeria spp. were isolated from the process- pattern of Listeria monocytogenes and other Listeria spp. in asalmon slaughterhouse and smoked salmon processing plant.ing plants more frequently than L. monocytogenes.Int. J. Food Microbiol. 25, 19–27.No association was found between the presence of

Schuchat, A., Swaminathan, B., Broome, C.V., 1991. Epidemiolo-other Listeria spp. and L. monocytogenes in the gy of human listeriosis. Clin. Microbiol. Rev. 4, 169–183.smoked salmon or the processing environment. Ac- Selvin, S., 1996. Statistical Analysis of Epidemiological Data,cording to our study, other Listeria spp. are not second ed. Oxford University Press, New York.

useful as indicator bacteria for the presence of L.monocytogenes.

Acknowledgments

The authors wish to thank Eva Karin Nekstad forexcellent technical assistance, and the inspectors