Human listeriosis in England and Wales, 1981–2015

The aim of this report was to review data on the investigations of foodborne listeriosis in England and Wales between 1981 and 2015 which includes reviewing data in the peer reviewed literature together with hitherto unpublished data. Between 1981 and 2015, a total of 5252 cases of human listeriosis in England and Wales were reported. This report highlights the importance of submission of L. monocytogenes isolates from clinical cases of listeriosis for characterisation which is essential to establish foodborne links. We estimate that isolates are received by the national reference laboratory from >95% of cases reported (PHE unpublished data).

Listeriosis cases were considered over the five time periods: 1981–1986, 1987–1989, 1990–2001, 2002–2006 and 2007–2015 (Table 1). Prior to 1987, there were between 58 and 136 cases reported per year and from 1987–1989 this increased to 237–278 cases per year (Table 1) with just over half the number of cases associated with a large outbreak (outbreak O1 [Reference McLauchlin11]). There then followed a decline in reported cases between 1990–2001, with annual totals ranging from 90 to 146 cases. The number of cases per year increased to between 139 and 233 between 2002 and 2015 (Table 1).

Amongst all the 5252 cases, there were 28 incidents (480 cases) where consumption of specific foods was identified as associated with transmission of listeriosis, and these comprised 9% of the total reported cases. The 28 incidents included 11 sporadic cases and 17 outbreaks. Ten incidents occurred in hospitals (nine outbreaks and one sporadic case, total 37 cases) and the remaining 18 incidents (10 sporadic cases and eight outbreaks, total 443 cases) occurred in the community. The large outbreak occurring between 1987 and 1989 resulted in 51% of reported cases attributed to a specific food during this period. However, if the cases from this outbreak are excluded, 0.8% of all the cases reported between 1987 and 1989 were identified as linked to specific foods. Therefore, between 1981 and 2001 and excluding the large outbreak, ⩽0.8% of cases were linked to the consumption of specific foods. The proportion of the total numbers of cases linked to specific foods increased to 3% between 2002 and 2006 with the implementation of AFLP and then to 5% between 2007 and 2015 with fAFLP being the primary discriminatory typing tool (Table 1).

During the 1981 to 2015 study period, a variety of techniques were used to characterise L. monocytogenes, and these techniques provided sequential improvements in reproducibility, discrimination and typability. With these improvements in the characterisation of this bacterium, there was increased attribution of cases to specific foods. Despite the efforts to investigate cases, it was not until the advent of more robust and discriminatory typing systems for L. monocytogenes (particularly AFLP and fAFLP) that the proportion of cases attributed to specific foods associated with transmission increased to over 1%: this highlights the need to apply optimal typing techniques not only to isolates from clinical cases, but also to those from food (see below).

In December 2015, PHE introduced WGS for routine analysis of all isolates submitted to the reference laboratory. At the time of writing in 2019, out of approximately 500 listeriosis cases, WGS increased the percentage of cases linked to specific foods to ~10% (PHE, unpublished data).

This review is timely since the improvement in typing from WGS will further continue to increase the identification of similarities between clinical and food/environmental isolates within England and Wales, and this is likely to be enhanced as other countries adopt the use of WGS. The possibilities of international linkage were illustrated by the outbreak of listeriosis occurring between 2015 and 2018 and associated with frozen sweetcorn produced in Hungary [34]. Related cases were identified by the use of WGS not only in the UK, but also in Austria, Australia, Denmark, Finland, Hungary and Sweden. However, the utilisation of this advance in technology by analysis of WGS data will only provide maximal public health benefits to surveillance when combined with other data inputs, hence the emphasis in this review to re-examine data from historic incidents.

Incidents in hospitals

The 10 incidents occurring in hospitals (H1–H10; Table 2), comprised a single sporadic case (H5), and nine outbreaks involving between two and nine cases (Table 2). Seven of these outbreaks have previously been described in the literature [Reference Graham12, Reference Dawson13, Reference Shetty15–Reference Little17]. All 10 incidents were associated with the consumption of pre-prepared sandwiches served in this setting: one of the outbreaks also implicated a salad product (H8, [Reference Coetzee16]).

Table 2.

Characteristics of listeriosis in hospitals associated with pre-prepared sandwich consumption: characteristics of nine outbreaks and a sporadic case which occurred in England and Wales, 1981–2015

Regions: L, London; E, East; NW, North West; WM, West Midlands; NE, North East; SE, South East; W, Wales

The duration of all outbreaks ranged from 4 days (H2) to 32 months (H7). All cases occurred within immunocompromised adults who were >60 years of age except for outbreak H3 which affected five pregnant women and their unborn infants who ate sandwiches from a retail outlet in hospital whilst attending antenatal clinics [Reference Dawson13]. Eight of the nine outbreaks occurred within single regions: three (H1, H6 and H9) in more than one hospital within the same regions. The final outbreak (H9) occurred in hospitals across two adjacent regions and involved four cases.

For all 10 incidents, there was descriptive epidemiological evidence linking the cases to sandwich consumption: in one outbreak, a salad product was also implicated (H8, [Reference Coetzee16]). In nine of these incidents, L. monocytogenes of the same type was recovered from patients' specimens as well as from the implicated sandwiches collected either within the hospital, and or from the point of manufacture (sandwiches or environmental sites) and therefore were categorised as having both strong microbiological and epidemiological evidence for the consumption of this specific food. For the 10th incident (H8), there was strong descriptive epidemiological evidence for sandwiches and a salad product but the implicated type of L. monocytogenes was not isolated from these foods [Reference Coetzee16]. However, in this incident, all three cases were in the same hospital with a common exposure to sandwiches and salad from a single manufacturer: strong evidence was provided by product-tracing investigation based on patient menu choice records and food supplier records.

There was no information on levels of L. monocytogenes in sandwiches or fillings in one of the outbreaks (H1). The bacterium was detected in all samples at less than 10² cfu/g in products collected at various points in the food chain, including in finished sandwiches collected in hospitals ready for consumption in seven of the incidents (H2, H3, H6, H7, H8, H9 and H10). In the final two incidents (H4 and H5), L. monocytogenes was detected at >10² cfu/g in complete sandwiches and fillings collected at the point of manufacture.

For one of the incidents (H8) there was no information on the ingredients used in the sandwiches, but for the remaining, sandwiches with various types of fillings were contaminated with the implicated strains: seven contained various cooked meats, two eggs, five cheese, six salad or other plant based materials and five with fish or crustaceans. In eight of the incidents, the implicated strain was recovered from environmental sites, utensils or equipment within the sandwich production environments and, as previously identified [Reference Little17], provided evidence for cross-contamination at these factories. Although contamination of an ingredient or ingredients from suppliers to the sandwich manufacturers cannot always be excluded, the most likely reservoirs of contamination were from sites within the sandwich production environments. There was evidence of poor temperature control (>8 °C) of the sandwiches at the hospitals in five of the incidents (H2, H4, H5, H7 and H8).

In six of these nosocomial outbreaks, local, or (in H3) national increases of cases in one or more hospitals were detected and, following patient interviews or review of hospital food consumption records, preliminary information linked these clusters to pre-prepared sandwich consumption (Table 3). For three of the outbreaks (H5, H9 and H10), the source of infection was indicated by reviewing L. monocytogenes typing data of isolates obtained from unrelated microbiological testing. For the single sporadic case (H5), unrelated testing identified high (>10² cfu/g) levels of L. monocytogenes in pre-packed sandwiches which were notified by the producer and this led to a review and identification of a case infected with the same strain in a hospital supplied by the producer.

Table 3.

Initial observations that provided evidence to link outbreaks and incident of foodborne listeriosis to specific foods

As previously reported [Reference Little17], the most common food type associated with transmission of listeriosis in England and Wales during 1981 to 2015 was pre-prepared sandwiches served in hospitals. The duration of these nosocomial outbreaks ranged from a few days to 32 months and two thirds occurred within a single hospital. Of the three outbreaks that occurred in more than one hospital, the affected patients were in hospitals within the same or in adjacent regions which were supplied by the same manufacturer and this reflects distribution patterns for this short shelf life (usually <5 days) food product. At the time of writing (2019), similar incidents with pre-prepared sandwiches served in hospitals have continued to occur [Reference Zwizwai35]. Although all the incidents in England and Wales described here were either sporadic or small outbreaks generally affecting less than five cases, there is the potential to affect greater numbers of patients, and the incidents described here show many similarities to an outbreak that occurred in Canada in 2008 where 57 cases and 24 deaths occurred [Reference Currie36].

It is intriguing that the 10 incidents that occurred in hospitals all had an association with pre-prepared sandwiches, and that this food type was not recognised in any of the outbreaks or sporadic cases or in the community (see later text). Although foods associated with transmission are only identified in a small proportion of cases overall, this may reflect differences in testing between products served in hospitals and those on retail sale. However, it may be important that the microbiological testing of sandwiches in hospitals detected L. monocytogenes at <10² cfu/g in the majority of instances: higher levels (>10² cfu/g) of contamination were detected in almost all listeriosis incidents in the community (see later text). Sandwiches are complex foods with multiple components with differing durability parameters. Although components within sandwiches are likely to support the growth of the bacterium, the shorter shelf life as compared to some of the other foods associated with the community incidents is consistent with exposure to L. monocytogenes at low levels being of greater risk for infection to immunocompromised patients in hospitals than the general population in the community. Evidence from the incidents described here indicates there are opportunities for improvements in temperature control at hospitals. This may be difficult in these types of environment and the most effective interventions (as with the incidents in the community) are to eliminate and reduce contamination at the point of production.

Incidents in the community

There were 18 incidents in the community; eight outbreaks (Table 4) and 10 sporadic cases (Table 5).

Table 4.

Characteristics of foodborne listeriosis outbreaks in the community: England and Wales 1981–2015

L, London Region; E, East Region; NW, North West Region; Y&H, Yorkshire and Humberside Region; WM, West Midlands Region; NE, North East Region. CSF, cerebrospinal fluid; AFLP, amplified fragment polymorphism; fAFLP, fluorescent amplified fragment polymorphism; cfu, colony- forming units.

^a Cases also occurred in Scotland and Norther Ireland.

^b One case in Scotland.

^c Index case resident in the South East but purchased food from the East Midlands region.

Table 5.

Characteristics of sporadic foodborne listeriosis cases in the community: England and Wales 1981–2015

NK, not known; L, London Region; E, East Region; NW, North West Region; Y&H, Yorkshire and Humberside Region; CSF, cerebrospinal fluid; AFLP, amplified fragment polymorphism; fAFLP, fluorescent amplified fragment polymorphism; PFGE, pulsed field gel electrophoresis; cfu, colony-forming units.

Considerations on foodborne listeriosis in the community

The identification of incidents of foodborne illness in the community can be problematic and there are different strengths of evidence implicating suspected foods vehicle. The evidence to attribute a particular food vehicle can be epidemiological, microbiological, descriptive environmental or based on product-tracing investigations [21]. Patients related by common food exposures can be widely separated both geographically and temporally because of the features of listeriosis i.e. the long incubation period, low attack rate, prolonged colonisation at food production facilities and the complexity of the food chain. Strong epidemiological evidence includes statistical associations in well-conducted analytical epidemiological studies or convincing descriptive evidence. For listeriosis, statistical associations may be difficult to apply since outbreaks generally affect relatively small numbers of cases: in this series, all except one of the outbreaks affected 17 or fewer patients. Furthermore, descriptive evidence may be subjective and based on value judgements, e.g. convincing descriptive evidence.

Microbiological evidence includes the detection of the causative agent in the food vehicle or its components and the detection of the causative agent in the food chain or from the preparation or processing environment. For investigating listeriosis this may be problematic, for example, the recovery of indistinguishable L. monocytogenes from the patient's clinical specimens and from food which had been in direct contact with the patient (such as from their domestic refrigerator) or from opened food collected in a single setting are not considered to be of a strong association since cross-contamination between foods or from the patient to a food cannot be excluded. The recovery of indistinguishable L. monocytogenes from a patient and from food or food components at manufacture is considered as strong microbiological evidence. Since L. monocytogenes can persist in environments for considerable lengths of time, environmental studies may be inconclusive in providing evidence for specific exposures, particularly if there is cross-contamination within a complex food chain. Product-tracing (investigating the movement of a food product and its constituents through the stages of production, processing and distribution) may be difficult, complex and time consuming. Advances in the characterisation of L. monocytogenes isolated from different parts of the food chain may allow targeting of this process. Consequently, for the investigation of human listeriosis, no one data source is likely to be sufficient for public health investigations and epidemiological and microbiological evidence for the infection should be combined with data on the causative agent in the food chain. Furthermore, because of the ability of L. monocytogenes to persist in food environments for years to decades [Reference Ferreira3], it is also important to be able to integrate these data over many years on a national or even an international basis, and interpret these on the background of the evolutionary changes of the bacterium.

Amongst all the cases in the community, there were 10 sporadic cases and eight outbreaks identified as associated with foodborne transmission. The majority of cases were part of outbreaks comprised <14 cases and continued for over 1 month to 3 years. Descriptive studies were used for all outbreaks and analytical epidemiology was successfully applied to half of these. The large outbreak associated with pâté consumption [Reference McLauchlin11] occurred over all regions in the UK and reflects the national distribution of this product. Half of the outbreaks in the community were confined to a single region, and eight of the nine nosocomial outbreaks were confined to a single hospital: this is likely to reflect more local food producers with restricted (although sometimes rather complex) distribution chains.

Because of the difficulties in attributing specific foods to the transmission of listeriosis, it is important to combine multiple sources of data. Including data from sporadic cases with that from outbreaks is likely to provide additional information on risk factors including those foods of greatest risk. Using all the data here, the most common information source to initially identify the specific foods associated with the cases was typing of isolates from food or the environment either recovered as part of unrelated food testing, or from food items recovered from patient's domestic refrigerators. This may be surprising since there is a very low probability of sampling foods associated with transmission of disease at random and independently of investigations of cases of suspect foodborne listeriosis. However, sampling is not carried out at random. PHE currently manages a network of Official Control laboratories in England which test approximately 25,000 food and environmental samples for the presence of Listeria each year. This activity generates more than 700 L. monocytogenes isolates being sent for characterisation to the reference laboratory. Cultures are also received by the reference laboratory which were isolated by Official Control laboratories in Wales, as well as from commercial food microbiology laboratories. There is often a risk basis for the collection of samples, either because of direct observations of hygiene practices (samples are more likely to be collected where hygiene is poor) as well as from prior knowledge of foods previously associated with transmission of listeriosis. Consequently, the analysis of L. monocytogenes (as well as for other enteric pathogens) isolated from food and the environment provides a unique source of data for surveillance. With the advent of the more widespread application of WGS, there is an urgent need to consolidate both national and international repositories of data such that public health investigations, including outbreak and incident investigation, risk assessment, as well as strategies for disease prevention and control can be rapidly identified and interventions in the food chain implemented [39].

Following the typing of isolates from unrelated food testing, the second most common method of initially identifying incriminated foods in incidents of listeriosis was through testing items recovered from patient's domestic refrigerators. The advantage of testing foods collected from patients' homes was demonstrated in an outbreak of 34 listeriosis cases which occurred in Canada in 2016 [Reference Hanson40]. In this Canadian outbreak, the incriminated food was not identified by a case control study, but following the recovery of the implicated strain from pasteurised chocolate milk collected from one of the patient's homes. After re-interviewing 12 of the patients, seven identified exposure to this product, and this led to the sampling and recovery of the implicated strain from the food production site. Although foods may no longer be available in domestic refrigerators since listeriosis patients show variable (1–90 days) incubation period [Reference Goulet4, Reference Linnan5], the conservative nature of an individuals' eating habits and the colonisation of harbourage sites in food production environments suggest that collection of samples from domestic refrigerators (or freezers) may be valuable. In the light of the generally low attribution of specific foods to patients with listeriosis, we would advocate both better linkage of data from unrelated testing of isolates as well as investigation of foods from patients' refrigerators as part of the routine investigation of listeriosis cases. The importance of contamination of foods from harbourage sites within food production facilities has long been recognised [Reference Farber and Peterkin2]. Incidents of listeriosis are almost always associated with this route of contamination for ready-to-eat foods where this bacterium may subsequently grow during storage and prior to consumption [Reference Farber and Peterkin2]. Considerations on the levels of L. monocytogenes contamination in food and the environment together with the likelihood of growth and survival of the bacterium are important to consider since risk will increase with the level of exposure. Data from the series described here implicated foods which were capable of supporting growth of this bacterium and were processed. The most common types of food contained ingredients of animal origin (cooked chicken, pâté, pork pies, sliced cooked meats, sliced ox tongue, crab meat, butter and soft cheese), although foods of non-animal origin (marinated olives and vegetable rennet) also occurred in this series. The foods described here are generally typical of those described in other countries [Reference Farber and Peterkin2] and are ready to eat and generally able to support the growth of this bacterium. The more ‘unusual’ food vehicles in this series were butter and olives. Butter has previously been implicated in an outbreak of listeriosis in Finland [Reference Maijala41]. The Finnish outbreak was associated with 7 g individual packages whereas in the outbreak described here, 2 kg packs were implicated (Table 4). In the English outbreak reported here there was anecdotal information that microwave heating was being used to soften the butter allowing easier spreading onto sandwiches. The detection of the implicated L. monocytogenes strain at 1.8 × 10² cfu/g in butter suggests that either there is gross contamination of the milk at the start of the process (this strain was present in the drain within the dairy) or that growth was occurring in the butter matrix which might have been degraded by the microwave heating. A study was subsequently carried out on butter on retail sale in the UK [Reference Lewis42] which demonstrated that although butter is regarded as a low-risk product, it may provide an environment for the persistence and growth of Listeria spp., including L. monocytogenes. Olives are generally not a permissive environment for the survival and growth of L. monocytogenes [Reference Salazar43]. In the sporadic case described here (Table 5) the olives were marinated in garlic and sweet chili oil and the implicated L. monocytogenes strain was recovered on one occasion at 40 cfu/g suggesting some growth had occurred, particularly at a market setting without refrigeration. Data on the pH (which will limit growth if sufficiently acidic) of this product is not available but this bacterium will grow in high salt environments [Reference Farber and Peterkin2] and there is a case described in Finland associated with consumption of salted mushrooms [Reference Junttila and Brander44]. The importance of food of non-animal origin in the transmission of listeriosis is increasingly being recognised [34, Reference Smith45].

For the sporadic cases and outbreaks in the community (unlike those in hospitals), the levels of the specific L. monocytogenes strains in food were generally high (>10² cfu/g) at some point in the food chain, and indicates that the specific foods supported the growth of this bacterium. In data generated by both PHE from routine testing of foods and across the EU [1], this level is rare and exceeds limits in statutory microbiological criteria [46]. In some instances, these levels were detected several years before the first cases of listeriosis were detected. Food manufacturers and food regulators should make stringent efforts to ensure that foods with elevated high levels of L. monocytogenes do not occur in the food chain.

In summary, we review here a total of 28 incidents (480 cases) of foodborne listeriosis in England and Wales between 1981 and 2015 where specific foods were identified as associated with transmission. This report highlights that for public health interventions to prevent human listeriosis, there is a need to combine, and have rapid access to, epidemiological and clinical microbiological data together with the occurrence and behaviour of L. monocytogenes in food chains and the environment. In addition, there is a need to apply discriminatory typing systems to all isolates of L. monocytogenes. L. monocytogenes recovered from unrelated testing of samples from the food chain together with the sampling of foods from patient's domestic refrigerators should be considered essential components of surveillance strategies for human listeriosis. It is also important to be able to integrate data over many years on a national as well as an international basis. The recognition between hygiene problems within food chains prior to the onset of infection represents missed opportunities for the prevention of this disease. Finally, food manufacturers and food regulators should make more stringent efforts to ensure that foods with elevated high levels of L. monocytogenes do not occur in the food chain, although greater stringency should be applied to the presence of this bacterium for foods served in hospitals as well as other healthcare institutions which serve food to the elderly and immunocompromised.