Freezing as an intervention to reduce the numbers of campylobacters isolated from chicken livers



Mike Hutchison, School of Veterinary Sciences, University of Bristol, Langford, Bristol BS40 5DU, UK. E-mail:


The aims of this study were (i) to determine the prevalence and numbers of campylobacters in 63 samples of raw livers purchased at retail across the UK and (ii) to investigate whether the freezing of chicken livers contaminated with Campylobacter was a reliable method for decontamination. Chicken livers naturally contaminated with campylobacters were subjected to freezing at −15 and −25°C for one day and 7 days. Numbers of campylobacters on the livers were determined immediately before and after a 24-h or 7-days freeze treatment and daily during 3 days post-thaw refrigerated storage. Freezing for 24 h at −25°C can reduce numbers of Campylobacter by up to 2 log10 CFU g−1. Freezing the livers for 24 h at −25°C, thawing overnight in a fridge set to 4°C and refreezing for another 24 h at −25°C reduced the numbers of campylobacters by up to three logs. Reduction in the numbers of campylobacters was significantly greater following a second freeze treatment compared with a single freeze treatment.

Significance and Impact of the Study

Freezing chicken livers can reduce, but not eliminate, campylobacters. If poultry processors were to freeze livers destined for human consumption as part of routine processing, there is a potential for a reduction in campylobacteriosis associated with the consumption of imperfectly cooked chicken livers and derivatives, such as pâté.


Outbreaks of infectious intestinal disease from foodborne sources have fallen overall in England and Wales over a period of roughly two decades prior to 2008 (Gormley et al. 2011). However, over the same period, there was an overall increase in the incidence of human campylobacteriosis largely from poultry-derived foods (Gormley et al. 2011). Recently, there has been growing evidence from source attribution studies (Strachan et al. 2012), investigations of outbreaks of campylobacteriosis (Little et al. 2010; Merritt et al. 2011) and retail surveys (Fernandez and Pison 1996), which implicates imperfectly cooked chicken livers as a likely source of illness. In particular, chicken livers prepared in catering premises and consumed as ‘pan-fried’ or in the form of pâté or parfait appear to be making a disproportionately large, and increasing, contribution to human illness (Little et al. 2010).

In Switzerland, Campylobacter has often been isolated from chicken livers in numbers up to 5 × 104 CFU g−1 with a prevalence varying between 10 and 100% depending on season (Baumgartner and Felleisen 2011). Although no account was taken of season, an earlier US survey of Campylobacter in chicken livers reported an overall prevalence of 48% (Barot et al. 1983) with a more recent US study reporting a 77% prevalence (Noormohamed and Fakhr 2012). A Polish study determined fresh liver Campylobacter prevalence to be 31% (Mackiw et al. 2011), whereas in Chile, the prevalence in frozen livers was 93% (Baumgartner et al. 1995; Fernandez and Pison 1996). In addition to apparent widespread contamination, there is good evidence that Campylobacter contamination is not restricted to the surface of liver, with reports of low numbers of Campylobacter inside (Barot et al. 1983) and in ‘hotspots’ such as the bile ducts (Baumgartner et al. 1995).

Collectively, these studies provide evidence that chicken livers from campylobacter-colonized chickens are routinely contaminated with campylobacters. Consequently, the cooking condition required to eliminate Campylobacter from chicken livers has been defined as an internal temperature that exceeds 70°C for at least 2 min (Whyte et al. 2006). A key observation made by Whyte and colleagues was that undercooked livers were pink in colour which appeals to consumers. If the livers were cooked for too long, they became an unappealing grey. Both O'Leary et al. (2009) and Little et al. (2010) suggested that the desire of the caterers to keep livers pink and appealing contributed to UK outbreaks of campylobacteriosis associated with the consumption of chicken liver dishes.

Although adequate cooking kills campylobacters, there are other potential interventions that could help control this bacterium that does not significantly impact on the organoleptic properties of the livers. It has been reported previously that freezing can reduce numbers of viable campylobacters on chicken skin and muscle (Garenaux et al. 2009; Sampers et al. 2010) and that typically freezing chicken meat or skin for 24 h causes around a one-log-cycle reduction in the numbers of campylobacters (Oyarzabal et al. 2010; Sampers et al. 2010). However, we were unable to find information describing the effect of freezing chicken livers contaminated with campylobacters. Consequently, this study assesses the effect on numbers of Campylobacter if chicken livers are frozen for up to one week at temperatures chosen to mimic domestic and catering freezers before thawing for 24 h in a refrigerator. In addition, the effect of freezing, thawing and then refreezing the livers is reported.

Results and discussion

Thirty-three fresh and 30 frozen samples of livers were purchased from 51 different retail outlets throughout Great Britain. From the identification codes, it was concluded that the 63 samples originated from 14 different EU-registered slaughterhouses. Samples were collected from supermarket chains (n = 48), independent butchers (n = 10), delicatessens (n = 1), commercial catering suppliers (n = 2) and convenience stores (n = 2). Quantitative results reporting the presence of Campylobacter in these samples are summarized in Fig. 1. Of the 63 samples, 55 contained countable numbers of Campylobacter and eight contained fewer than the limit of detection of the test method (<1 CFU g−1 liver). Livers purchased unfrozen tended to have higher counts than livers purchased frozen, and this difference was statistically significant (t-test, P < 0·02).

Figure 1.

The distributions of the concentrations of campylobacters isolated from unfrozen (a, n = 33) and frozen (b, n = 30) chicken liver samples purchased at retail in GB between October 2012 and February 2013.

Further analyses revealed that there was no relationship between the temperatures of the entire 63 samples at the time of purchase and the numbers of campylobacters within those samples. The same was true when only those test results above the limit of detection were compared (i.e. the positive samples). Furthermore, there was no correlation between the numbers of campylobacters counted and the number of days of shelf life remaining when all 63 results were compared. When only the results from either the fresh or the frozen samples were compared, there were also no correlations between the numbers of campylobacters and the remaining shelf life (results not shown).

When the numbers of campylobacters on fresh livers from the retail survey were compared with the slaughterhouse-sourced livers, no significant differences between the counts were found (t-test). Two freeze temperatures were selected to represent best and worst cases in terms of freezing effectiveness. The worst-case freezer, set to −15°C, was unable to freeze the liver masses to the target temperature within 24 h (Fig. 2a) achieving only −11°C. After two days, the minimum temperature achieved was −14°C (Fig. 2b). In contrast, the freezer set to −25°C lowered the temperature in the centre of the liver masses to −26·2°C within 6 h (Fig. 2c) with a further 1°C temperature reduction over the course of several days (Fig. 2d). It was observed that in both freezers, the rate at which the temperature decreased slowed down when the samples reached −1°C, as the water contained within the livers turned to ice. The reduced rate of freezing at the phase transition temperature was 6 h longer in the freezer set to −15°C compared with the freezer set to −25°C and was considered to be the primary reason that the less effective freezer failed to lower the temperature of the livers to −15°C within 24 h.

Figure 2.

Typical temperatures achieved inside a 1·5-kg bag of chicken livers placed in freezers set to −15 (a and c) and −25°C (b and d) for 1 day (a and b) or 7 days (c and d) followed by 3-day refrigerated storage.

Livers frozen for 24 h in the freezer set to −15°C had significantly reduced (t-test; P < 0·01) numbers of campylobacters compared with the unfrozen controls (Fig. 3a). A similar observation was made for livers frozen for 24 h in the freezer set to −25°C (Fig. 3b). The colder freezer reduced the numbers of CFU by approximately 1·5 logs, which was significantly greater (t-test; P < 0·01) than the 0·8-log reduction observed for the livers in the freezer at −15°C.

Figure 3.

The effect of freezing to target temperatures of −15 (a) and −25°C (b) on numbers of campylobacters in naturally contaminated chicken livers. Data are the result of three independent trials with 10 replicates per trial (n = 30).

There were significantly fewer campylobacters after freezing at −15°C for 7 days, compared with the numbers after 24 h at this temperature. Overall, freezing at −15°C for 7 days caused a 1·5-log reduction in numbers of campylobacters, which was broadly comparable with the reduction observed at −25°C for 24 h. There was no significant difference between the reductions obtained after 24 h and 7 days in the −25°C freezer (t-test; P > 0·05). Individual samples contained similar numbers of CFU g−1 immediately after thawing and after chilled storage for three more days (no significant differences by anova). Thus, we found no evidence to support a hypothesis that freezing causes sublethal injury to Campylobacter with a potential for recovery during subsequent thawing and extended chilled storage. Equally, there was no evidence of continued die-off if livers were stored chilled after freezing and thawing.

When contaminated livers were frozen twice to −25°C for 24 h with thawing for 24 h in-between (Fig. 4), reductions in the numbers of campylobacters were observed after each freeze/thaw treatment (Fig. 5). In broad agreement with the results of the single treatments reported above, the first treatment caused a significant reduction of about two logs (t-test; P < 0·01). The second freeze/thaw treatment caused a further decrease of around 0·75 logs (Fig. 5). The increased effect of two vs one freeze/thaw treatment was significant (t-test; P < 0·01). Overall, two freeze/thaw treatments reduced the numbers of campylobacters by about three logs. As before, comparison of individual samples tested immediately after thawing with those tested on each day of a two days’ refrigerated storage showed no significant differences by anova.

Figure 4.

The temperatures at the centre of three separate 2·5-kg bags of chicken livers. The numbers of campylobacters were determined before freezing (not shown on figure) and after each bag of livers had been frozen and thawed (A), refrozen and rethawed (B) and during refrigerated storage post-thaw for 2 days (C and D).

Figure 5.

The effect of freezing chicken livers on two occasions on the numbers of CFU g−1 Campylobacters. The numbers of campylobacters were determined prior to freezing, post-thawing (PT) after freezing to −25°C for 24 h, PT from a second freeze to −25°C for 24 h and after refrigerated storage for one and 2 days. Data are the result of three trials with 10 replicates per trial (n = 30).

The observation of a decrease in Campylobacter counts of up to 2·5 logs when the livers were frozen for only 24 h is consistent with findings reported for fresh chicken meat (Georgsson et al. 2006; Maziero and de Oliveira 2010). It has been previously reported that the length of time taken to freeze is important for bacterial survival in meat (Gill 2002; Archer 2004). For the rates of cooling used in this study, the basis of any kill was that freezing initially happened in small isolated pockets of liquid (Archer 2004). The freezing of these pockets caused dissolved molecules to be displaced into the surrounding unfrozen fluid, thereby increasing the osmotic potential of that liquid (Dumont et al. 2004). As the extracellular fluid became more concentrated, it began to remove water from the cytoplasm of the suspended cells by osmosis (Dumont et al. 2004). Ice crystals, formed from the water remaining inside the cytoplasm, are the primary method of cellular damage during freezing (Toner and Cravalho 1990). However, when cooling rates of liquids are low (a few degrees per min), it is possible that all of the intracellular water can be removed from the cell before ice crystal formation (Dumont et al. 2004). In addition, there is evidence that superoxide radicals form during freezing, which are similarly concentrated in unfrozen pockets of water and consequently contribute towards the death of campylobacters (Stead and Park 2000). This study demonstrated that the freezer set to −15°C caused less bacterial death after 24 h compared with the freezer set to −25°C. The livers placed in the freezer set to −15°C were soft-frozen with a slushy nonrigid consistency. In contrast, the livers placed in the freezer set to −25°C were hard-frozen. Figure 2 shows that it took around 12 h for freezing in the least effective freezer compared with only 4 h in the freezer set to −25°C. Thus, it seems likely that the longer time required for freezing in the least effective freezer allowed the majority of the intracellular water to be removed from the campylobacters before ice crystal formation caused significant cellular damage. It has been previously reported that chicken meat exudate can upregulate stress response genes in campylobacters and help protect them from low temperature damage (Ligowska et al. 2011). It is also possible that the longer time taken to freeze the livers placed in the freezer set to −15°C enabled the campylobacters to upregulate any cold temperature defences.

Test results for fresh livers at retail (Fig. 1) indicated that it is rare for numbers of campylobacters to exceed 10 000 CFU g−1 fresh liver, a finding in keeping with previous reports (Baumgartner and Felleisen 2011). Thus, a reduction of 2 to 3 logs would shift the numbers shown in Fig. 1 so that the majority of liver samples would contain <100 detectable CFU g−1. A literature search did not reveal any risk assessments specific for the consumption of chicken livers, although risk assessment models for Campylobacter in broiler meat generally were identified. A comparison of six independently generated models concluded that the most effective intervention measures for human illness were those that reduced the Campylobacter concentrations on chicken meat (Nauta et al. 2009).

In summary, our findings show significant reductions in the numbers of campylobacters when contaminated livers were frozen. There are some chicken processors in the UK, which freeze all of the livers collected in their plants, and thus, it seems plausible to conclude that livers could be frozen as part of routine processing from all plants without significant economic impact. Advice could also be issued to consumers that undertaking a second freeze would be beneficial in terms of a reduced likelihood of illness were the livers to be cooked inadequately.

Materials and methods

Microbiological examination

Each sample of livers was finely chopped into small (0·5 × 0·5 × 0·5 cm) pieces using sterile scissors. A 25-g subsample was generated by removing randomly selected pieces using a sterile spoon to ensure that the meat exudate was also tested. An equal volume of maximum recovery diluent (MRD, Oxoid, Basingstoke, UK) was added to each 25 g liver sample before homogenization for 1 min using a stomacher 400 (Seward, UK). Campylobacter were enumerated using the ISO 10272 part 2 direct-plating method (International Standards Organisation 2006) with minor modifications. In brief, 2 ml of the initial 1 in 2 (1 : 1) dilution was spread onto six plates of modified charcoal cefoperazone deoxycholate agar (mCCDA, Oxoid CM0739 plus SR0155). Two millilitre of the initial suspension was then mixed with 8 ml of MRD to yield a 1 in 10 dilution. All subsequent dilutions were decimal as described by ISO 10272 and made using MRD. One hundred microlitre volume of the decimal dilutions was plated in duplicate onto mCCDA. Incubation was under microaerobic conditions (CampyGen, Oxoid) at 41·5°C for 48 h. Colonies of Campylobacter spp. were confirmed by a positive oxidase reaction and inability to grow in aerobic atmosphere at 41·5°C on Columbia blood agar (Oxoid). In addition, a single colony from each plate was tested using Dryspot Campylobacter (Oxoid) according to the manufacturer's instructions.

Sample collection at retail

In total, samples (380 g to 500 g) of fresh (n = 33) or frozen (n = 30) chicken liver from England, Wales and Scotland were purchased from national and small independent retailers between October 2012 and February 2013. The samples were examined to determine numbers of campylobacters using the test protocol described above. Before purchase, the surface temperatures of the liver samples were measured with an infrared thermometer (model 59 mini, Fluke, Norwich, UK). Samples were purchased on randomly selected days (including weekends), labelled and transported in insulated Styrofoam boxes (Biotherm 45, DGP Intelsius, York, UK) using bubble-wrapped retail packs of frozen peas as refrigerant. Temperature loggers (Tinytag Plus 2; Gemini Data Loggers, Chichester, UK) were placed beside the samples to monitor the transit temperatures. Only samples with temperatures during transit between 0 and 8°C were examined. Samples were stored at 4°C in the laboratory prior to examination. All microbiological examinations commenced within 24 h of purchase.

Single freezing treatments

Broiler chickens are more likely to be colonized by campylobacters if they come from flocks that have previously been thinned, that is, taken from rearing sheds where some of the birds had been removed some days previously (Allen et al. 2008). Therefore, to maximize the chance of sourcing campylobacter-positive flocks, livers for freezing experiments were collected from a slaughterhouse where thinned flocks could be identified. Additionally, it was only possible at slaughter to collect large quantities of livers from the same flock as those most likely to contain the same strains of Campylobacter at constant concentration. Direct collection also meant confidence that the livers had not previously been frozen and had been processed only to remove gall bladders and bile ducts before cleaning and chilling in slush ice. 7–8 kg of livers was collected into a single bag and transported to the laboratory packed in ice. At the laboratory, four polythene bags, each containing 1·5 kg (roughly 90) randomly selected livers, were prepared, and a temperature logger (Tinytag Plus 2) was placed in the centre of each bag. For each experiment, two bags of livers were placed in a freezer with a set temperature of −15°C, and two were placed in a freezer at −25°C. For each temperature, one bag of livers was removed after 24 h and the other on the 7th day after freezing. After removal from the freezer, the livers were thawed overnight in a fridge set to 4°C. Thirty thawed livers were examined in batches of three pooled samples (n = 10) immediately after thawing to determine numbers of campylobacters per g. The remaining sixty livers were returned to the fridge, and 10 samples (each composed of three pooled livers) were examined after 24 h and 48 h. Each freezing trial was undertaken three times over a period of 3 months and using livers from birds raised on different farms.

Freeze, thaw and refreezing treatments

Experiments to determine the effect of freezing livers more than once were carried out as described above for single freeze treatments, with minor differences. A polythene bag containing 2·5 kg livers was prepared. The livers were tested to determine prefreeze numbers of campylobacters, frozen at −25°C for 24 h, thawed for 24 h at 4°C, tested to determine numbers of campylobacters and then immediately refrozen for a second 24 h before a second 24 h thaw at 4°C and re-examination. As before, numbers of campylobacters were determined each day for a further 2 days of refrigerated storage at 4°C.

Statistical analyses

Paired and homoscedastic t-tests, and analysis of variance (Excel 2010; Microsoft, Redmond, WA, USA) were used to compare the log10 cfu recovered from the various pre- and postfreeze liver samples and the samples collected at retail as appropriate. For those results that were reported as below the limit of detection (<1 CFU g−1), a value of half of the limit of detection (0·5 CFU g−1) was substituted to allow log10 transformations. For all tests, a P value of <0·05 was used to determine any significance between treatments. Pearson product moment correlation coefficients (Excel) were used to presumptively determine any relationships between the retail purchase temperatures and numbers of campylobacters. The least squares algorithm was used to determine the strengths of any presumptive relationships.


This study was funded by the UK Food Standards Agency as project FS101025. The authors acknowledge the donation of 40 kg of fresh chicken livers from the British poultry processing industry.