Introduction

Listeria monocytogenes is widespread in the environment and is capable of infecting both animals and humans. Although human listeriosis is rare in many countries [1–7], it is one of the main causes of food-related deaths [8, 9]. Several outbreaks of listeriosis caused by consumption of contaminated foods such as milk [10], butter [11], soft cheese [12–14], ready-to-eat foods (mainly fish or meat products) [15–20], and produce [21] have been reported [22]. Infections in previously healthy individuals may cause febrile gastroenteritis and are usually mild and self-limiting [23, 24]. However, among elderly individuals, immunocompromised patients, and patients with malignancies, diabetes mellitus, chronic liver disease, or chronic renal disease, listeriosis may lead to severe illnesses (such as meningitis and septicaemia) [25–28]. Furthermore, the risk of infection is increased during pregnancy. Although infections in pregnant women usually remain asymptomatic or develop to a mild influenza-like illness, perinatal infections often lead to abortion, stillbirth, premature birth, or neonatal disease [7, 20, 25, 29–31]. If the central nervous system is affected, infections may cause neurological sequelae [12, 31].

Thirteen serotypes of L. monocytogenes have been identified, but serotypes 1/2a, 1/2b, and 4b are responsible for most cases of human disease. Although serotype 1/2a is more often found in foods, 4b is the most frequent serotype in human listeriosis [32–34].

In the Netherlands, listeriosis is not a notifiable disease, so information about the incidence of diseases due to infection with L. monocytogenes is scarce. Therefore, we reviewed the data on hospital patients diagnosed with listeriosis in the period 1995–2003. We also performed a retrospective study of patients whose Listeria isolates were received by the Netherlands Reference Laboratory for Bacterial Meningitis (NRLBM) from 1999 to 2003.

Materials and methods

Review of data on hospital patients with listeriosis, 1995–2003 (Study 1)

Data about hospital admissions from January 1995 to December 2003 in which listeriosis was the primary or secondary diagnosis (code 27.0 in the International Classification of Diseases, revision 9 [ICD-9]) were obtained from the National Medical Registration (study 1). The data included the age and sex of the patient, ICD-9 codes of the primary and secondary diagnoses, duration of hospitalization, and destination after discharge.

Retrospective study of data on NRLBM patients with Listeria, 1999–2003 (Study 2)

Clinical microbiological laboratories in the Netherlands are requested to send bacterial isolates from patients with proven meningitis or with septicaemia with suspected meningitis to the NRLBM. The NRLBM receives about 85% of the isolates from Dutch meningitis patients [35]. The coverage for patients with septicaemia is probably lower. Patients whose isolates of L. monocytogenes were submitted to the NRLBM from January 1999 to December 2003 were included in the retrospective study (study 2). Standard microbiological methods were used to identify L. monocytogenes [36]. The isolates were sent to the National Institute for Public Health and the Environment for pulsed-field gel electrophoresis (PFGE) analysis and serotyping. Serotyping was performed as described elsewhere [37]. DNA fingerprints were generated with PFGE, using AscI as the restriction enzyme. Cluster analysis of fingerprints was performed with Bionumerics (dendrogram type=UPGMA, similarity coefficient=dice). Isolates with more than 85% of corresponding fragments were regarded as “related”. The isolates were considered “indistinguishable” when 100% of the fragments were identical.

Patient information such as age, sex, origin of isolate, L. monocytogenes serotype, clinical symptoms (if known), and death (if known) were obtained from the NRLBM. We obtained the discharge letters for these patients from the hospitals to which they had been admitted. If available, we retrieved data about the clinical manifestation, underlying disease, duration of hospitalization, occurrence of neurological sequelae, recurrence of infection, readmissions to hospital, and mortality. Conditions considered to predispose to listeriosis included haematological and solid malignancies, chronic liver disease, chronic renal disease, connective tissue disease, transplantation, diabetes mellitus, alcoholism, use of immunosuppressive therapy (corticosteroids or cytostatics), pregnancy, and birth.

The information for patients whose hospital discharge letters could be obtained is likely also included in the information obtained from the National Medical Registration of hospital patients (study 1). However, the exact overlap in study populations in both studies could not be established because detailed patient information enabling identification of duplicate patients was not available.

Results

Listeriosis was diagnosed in 283 hospitalized patients from 1995 to 2003 (Fig. 1), which indicates an average annual incidence of 2.0 per million inhabitants. In this period, the NRLBM received L. monocytogenes isolates from 241 patients (Fig. 2). Although the number of hospital patients fluctuated over the years, the number of Listeria isolates at the NRLBM increased steadily from 1995 onward, and it even exceeded the number of hospitalized patients in 2002 and 2003.

Fig. 1
figure 1

Number of hospitalized patients with listeriosis as the primary or secondary diagnosis in the period 1995–2003 (study 1)

Full size image
Fig. 2
figure 2

Origin of Listeria isolates from 241 patients, received at the Netherlands Reference Laboratory for Bacterial Meningitis in the period 1995–2003 (study 2). The origins of isolates are not available for the years 1995–1998

Full size image

Review of data for hospital patients with listeriosis, 1995–2003 (Study 1)

Of the 283 hospitalized patients with listeriosis, 149 (53%) were men. The hospitalized patients were mainly adults, with 40% of the patients aged 65–79 years. Nine neonates and five young children (<5 years old) were hospitalized. The median age was 66 years (25th percentile [P25]: 53 years; 75th percentile [P75]: 75 years). The median duration of hospitalization was 18 days (P25: 10 days; P75: 29 days). Following discharge, 104 (60%) patients returned home, 24 (14%) patients went to an institution, and 44 (26%) died in hospital. The nonsurviving patients were all ≥50 years of age.

Listeriosis was the primary diagnosis for 172 of the 283 (61%) patients (Table 1). Ninety-five (55%) of these patients were male. The age distribution of patients with a primary diagnosis of listeriosis did not differ from that of all 283 hospitalized patients, except for patients aged 19–34 years, in whom listeriosis often was a secondary diagnosis. Twelve of the 172 (7%) patients with a primary diagnosis of listeriosis did not have a secondary diagnosis. Most patients (n=128, 74%) had one to four secondary diagnoses, and 32 (19%) patients had five or more secondary diagnoses. Among the 160 patients with a primary diagnosis of listeriosis and at least one secondary diagnosis, bacterial meningitis was the most common secondary diagnosis (Table 2).

Table 1 Primary diagnoses of hospitalized patients with a primary or secondary diagnosis of listeriosis in the period 1995–2003 (study 1)
Full size table
Table 2 Frequency of secondary diagnoses among 160 hospitalized patients with a primary diagnosis of listeriosis in the period 1995–2003 (study 1)
Full size table

One hundred eleven of the 283 (39%) patients had a primary diagnosis other than listeriosis (with listeriosis as a secondary diagnosis; Table 1). Among these patients, malignant neoplasm was the most common primary diagnosis. Ten patients had meningitis, septicaemia, or bacteraemia as the primary diagnosis.

One or more predisposing condition was registered for 134 (47%) patients: for 51 of the 111 (46%) patients with primary diagnoses other than listeriosis, the primary diagnosis was a predisposing condition (Table 1), and for 17 (15%) patients, one of the secondary diagnoses was a predisposing condition (data not shown). Of the 172 patients with a primary diagnosis of listeriosis, 66 (38%) had predisposing conditions among the secondary diagnoses (Table 2). A malignant neoplasm or a history of malignant neoplasm was the most frequent predisposing condition (n=68, 24% of all hospitalized patients).

One hundred twenty-eight patients had meningitis or encephalitis and 12 patients had septicaemia or bacteraemia, which corresponds to average annual incidences of 0.9 and 0.08 per million inhabitants, respectively. Seven patients had both septicaemia and meningitis and were included in both incidence estimates. Furthermore, 24 cases were pregnancy related: there were 15 pregnant women and nine neonates, resulting in an average annual incidence of 0.17 per million inhabitants or 1.3 per 100,000 pregnancies over 24 weeks of gestation (including both live births and stillbirths as pregnancy outcomes).

Of the nine neonates with listeriosis, two were premature (one died) and three had meningitis (one with concomitant septicaemia). Listeriosis was the sole diagnosis for two neonates, and two others had diagnoses other than preterm/full-term birth, septicaemia/bacteraemia, and meningitis/encephalitis. The outcome of pregnancy was recorded for 12 of the 15 women: five gave birth to a premature, live baby, three had early onset of delivery and stillbirth, and two gave birth to a full-term, live baby. One pregnancy resulted in a live baby, but no information about the duration of gestation was available, and one pregnancy resulted in an early delivery, but no information about the vital status of the baby was available.

Retrospective study of NRLBM patients with Listeria isolates, 1999–2003 (Study 2)

The NRLBM received L. monocytogenes isolates from 159 patients in the period 1999–2003, which indicated an average annual incidence of 2.0 per million inhabitants. The number of Listeria isolates, especially those from blood, increased steadily from 1999 to 2003 (Fig. 2). Eighty-three of the 159 (52%) patients were men. The patients were mainly adults, and 42% were aged 65–79 years. Fourteen patients were children: 12 neonates and two 1-year-olds. The median age was 67 years (P25=52 years; P75=74 years). The predominant serotype was 4b (44%), followed by 1/2a (36%), and 1/2b (16%). Serotypes 1/2c, 3a, 3b, and 4e were found only sporadically (6 patients in total). The median age of the patients with serotype 1/2a was 67 years (P25=52 years; P75=78 years), with serotype 4b, 65 years (P25=38 years; P75=73 years), and with serotype 1/2b, 71 years (P25=62 years; P75=74 years). These differences were marginally significant (p=0.056; Kruskal–Wallis test).

Isolates of 134 patients were PFGE typed. Culture of stored isolates from 25 patients was no longer possible. The PFGE analysis divided the isolates into 58 genotypes. These genotypes were divided into 26 main groups, each comprising “related” genotypes (Fig. 3). In general, genotypes within a main group were related, but not exclusively, to a specific serotype. For example, strains of main group 1 were mostly of serotype 1/2a (70%), whereas those in groups 15 and 16 were mainly of serotype 4b (88 and 83%, respectively), and strains in group 19 were predominantly of serotype 1/2b (63%).

Fig. 3
figure 3

Pulsed-field gel electrophoresis (PFGE) patterns of 134 of the total 159 isolates of L. monocytogenes received at the Netherlands Reference Laboratory for Bacterial Meningitis in the period 1999–2003a (study 2). aThe right column is the continuation of the left column; the numbering of main groups and subgroups is not consecutive because other isolates (e.g. food isolates) were included in the original PFGE typing

Full size image

The 26 main groups were further divided into subgroups of “indistinguishable” genotypes. We found 28 clusters of isolates with “indistinguishable” genotypes. However, within 12 of these clusters, there was a heterogeneous serotype distribution among strains. By combining PFGE and serotyping, 25 clusters of isolates with both “indistinguishable” genotypes and identical serotypes were identified, comprising 86 isolates. The number of isolates within each of these clusters varied from 2 to 19. Analysis of the date of isolate submission to the NRLBM and the place of residence of the patients showed either large time intervals or large geographical distances between patients within clusters, or both. In only three clusters were the time intervals and geographical distances relatively small (Fig. 3). One cluster of three isolates consisted of a mother and her newborn son and a 78-year-old man living in the same province. The isolate submission date for the man was 1 month earlier than that of the mother and child. The second cluster consisted of isolates of two men who were admitted to the same hospital within a 2-month period. Finally, within a cluster of 19 isolates, a neonate and a 67-year-old woman lived in the same residence, and the period between isolate submissions was 3 months.

Discharge letters were received for 107 (67%) patients. Letters were available for 13 of the 14 children (response 93%). The response did not differ for other age groups, sex, or serotype. The main clinical manifestations of listeriosis were septicaemia (36%), meningitis (34%), and a combination of both (16%). Thirty-one (29%) patients died. The primary causes of death were listeriosis or its complications in ten patients, listeriosis in combination with underlying illnesses in nine patients, and an illness other than listeriosis in 11 patients. The cause of death of one patient was unknown. Thus, listeriosis was a factor in the deaths of 19 of the 107 (18%) patients. The mortality rate due to listeriosis was lower among the older patients infected with serotype 1/2b than among patients infected with another serotype (6 vs. 20%), although this difference was not statistically significant (p=0.14, Fisher’s exact test).

Seventy-six patients (71%) had at least one predisposing condition (Table 3). The most frequent predisposing factor was immunosuppressive therapy (29% of all listeriosis patients with a discharge letter), but this particular factor was often present in combination with other predisposing conditions. Other frequent predisposing factors were haematological malignancies (12%), solid malignancies (15%), pregnancy (17%), and diabetes (10%). Neither pregnancy nor birth occurred in combination with other predisposing conditions. Predisposing conditions other than pregnancy were only present in patients aged 35 years or older. Information about underlying conditions was not available for seven patients. The serotype distributions among patients with and without predisposing conditions were similar.

Table 3 Predisposing conditions in 76 of 107 listeriosis patients with a discharge letter (study 2)
Full size table

Of the 18 patients with pregnancy-related predisposing conditions, six were pregnant women and 12 were neonates. Ten pregnancies (56%) resulted in the premature birth of a baby with meningitis, septicaemia, or a combination of both. One of these babies died of progressive hydrocephalus and cerebral injury with mental and physical retardation after Listeria meningitis and septicaemia, and one died of Listeria septicaemia combined with respiratory and circulatory insufficiency. One baby developed severe mental retardation, and one had imminent hydrocephalus. Four pregnancies (22%) resulted in intrauterine death, three (17%) resulted in a full-term baby with meningitis, and one resulted in the premature birth of a healthy baby. All premature babies were born in the third trimester of pregnancy. Two of the four intrauterine deaths occurred in the second trimester of pregnancy (at 22 and 23 weeks gestation), one occurred in the third trimester (at 32 weeks), and for the fourth one, the duration of pregnancy was unknown.

Only serotypes 4b and 1/2a were isolated from pregnant women and neonates, with 4b predominating (72%). For 13 (72%) mothers, information about symptoms was available. Nine (69%) women had symptoms, mostly fever. Information about the recurrence of infection, readmission to hospital, and development of neurological sequelae was sketchy, because discharge letters were the only source of this information. Therefore, the follow-up time of the patients was limited. Still, one recurrent Listeria infection and nine readmissions to hospital were recorded: one patient was hospitalized for meningitis due to Listeria, two patients for pneumonia, one patient for progressive hydrocephalus, and four patients for underlying diseases. Eight patients (10%) developed neurological sequelae: three had mild or incomplete paresis, two developed hydrocephaly, one had mental retardation, and one had vision problems and twitching legs. One patient showed a loss of cognitive functions due to a tumor of the central nervous system or Listeria meningitis.

Assuming that the 107 patients with available discharge letters were representative of all 159 NRLBM patients, the average annual incidences of meningitis and septicaemia would both be estimated as 1.0 per million inhabitants. Cases with a combination of meningitis and septicaemia were included in both estimates. The mean annual incidence of pregnancy-related listeriosis was 0.34 per million inhabitants, or 2.4 per 100,000 pregnancies over 24 weeks of gestation (including both live births and stillbirths). In this calculation, the two intrauterine deaths at weeks 22 and 23 were excluded from the nominator.

Discussion

The current studies have provided more detailed information on the incidence of invasive L. monocytogenes infections in the Netherlands. The estimated average annual incidence of 2.0 per million inhabitants is low compared with the reported mean annual incidence of 3.4 per million within the European Union. However, it is difficult to compare the rates because the sensitivity of surveillance systems differs between countries [2].

The study populations in the review of hospital patients (study 1) and the retrospective study of patients with an isolate at the NRLBM (study 2) probably overlap considerably. The extent of this overlap depends on the extent to which laboratories send isolates to the NRLBM and on the clinical manifestation of the infection (only isolates from patients with septicaemia or meningitis are sent to the NRLBM). Unfortunately, the databases lack the detailed patient information necessary for the identification of duplicate patients. Therefore, results of both studies are considered separately and are compared with each other.

The estimated incidences of the main clinical manifestations of listeriosis (meningitis, septicaemia, and pregnancy-related listeriosis) in the two studies did not correspond completely. For Listeria meningitis, the mean annual incidences of 0.9 and 1.0 per million inhabitants are in line with estimates of an earlier Dutch study [38]. In that study, the incidence of Listeria meningitis was 1.2 per million in the period 1981–1990, which decreased to 0.7 per million in the period 1991–1995. For Listeria septicaemia, the incidence estimates in our studies were less consistent: 0.08 per million (study 1) and 1.0 per million (study 2). It is likely that ICD codes for septicaemia and bacteraemia in hospital patients are hardly used (and are thus incomplete), because 16% of the NRLBM patients had a combination of meningitis and septicaemia, whereas hospital patients rarely had this combination. In addition, the number of Listeria isolates (especially blood isolates) received by the NRLBM has steadily increased in recent years, whereas the number of hospital patients remained constant. This may be due in part to the increased awareness of meningitis due to Neisseria meningitides serogroup C and the recently begun vaccination campaign in the Netherlands, which may have led to an increased number of isolates submitted to the NRLBM. However, a true increase in the cases of septicaemia cannot be ruled out. The incidence estimates of pregnancy-related listeriosis of 1.3 and 2.4 cases per 100,000 pregnancies over 24 weeks of gestation correspond with the incidence of 1.3 neonatal invasive infections per 100,000 live births reported in another Dutch study [31]. However, our estimates also include pregnant women with listeriosis. Seventeen percent of all cases with isolates at the NRLBM were pregnancy related, a proportion that was also found in the Listeria surveillance in England and Wales, where an incidence of 1.5–3.7 pregnancy-related cases per 100,000 conceptions was reported [3]. Although Listeria infections during pregnancy were rarely reported, our studies confirm its severe consequences: all pregnancies in study 2 and at least half in study 1 (in the other half, listeriosis was not further specified) ended in stillbirth, premature birth, or illness of the neonate. The symptoms of the pregnant women were frequently unknown, but if reported, they were relatively mild.

Both our studies show that patients with known predisposing conditions such as malignancy or use of immunosuppressive therapy are at risk of listeriosis. A greater proportion of cases in study 2 had predisposing conditions (71 vs. 47% in study 1). However, information about the use of medication was not available from the diagnoses registered for hospital patients in study 1. This was also the case for the use of immunosuppressive therapy, which was one of the main predisposing conditions in study 2.

The overall mortality rates were 26 and 29%, which are comparable with the 21% death rate found in Denmark [4], but in England and Wales, a mortality rate as high as 50% has been reported [3]. In study 2, the mortality rate due to listeriosis was 18%, which is in accordance with the earlier-reported 16% among patients with Listeria meningitis [38]. Because the follow-up time of the discharge letters was limited, the data for the eight patients with neurological sequelae (10% of the surviving patients) in study 2 give an underestimation of the occurrence of severe long-term sequelae. In a previous study with a median follow-up time of 11 months, 18% of the surviving patients developed neurological sequelae, and this proportion was 24% in an epidemic of food-borne listeriosis [12, 38].

Serotyping showed a predominance of types 4b and 1/2a. In contrast to an earlier Dutch study [38], we found that serotype distributions were similar among patients with and without predisposing conditions. However, the mortality rate due to listeriosis in our study tended to be lower among relatively older patients with serotype 1/2b, which suggests a lower virulence of serotype 1/2b. As serotype 4b was mainly found in pregnant, otherwise healthy women, this may be a more virulent strain. This finding is confirmed by literature reports that type 1/2a is more often found in foods, whereas 4b is the most frequent serotype of human listeriosis [32–34]. In addition, patients with serotype 1/2b were relatively old, and patients with serotype 4b were generally younger than those with serotype 1/2a.

In general, PFGE types were related to a specific serotype, though not exclusively. Even among strains with “indistinguishable” genotypes, the distribution of serotypes could be heterogeneous, which contrasts with the observations of others [33, 39]. Our results clearly indicate that PFGE typing has more discriminatory power than serotyping. Because combining PFGE patterns with serotyping showed a wide geographical spread and long periods between patients within clusters, even the combination of two subtyping methods may not provide indisputable evidence that cases are also epidemiologically linked. This may be due to the wide spread of Listeria in the environment, which may act as a long-term reservoir that causes human infections occasionally. Therefore, to identify sources, complementary epidemiological data are needed for confirmation. Finally, some studies show that using several subtyping methods simultaneously improves the discriminatory power [33] and makes source identification possible [16, 40].

In conclusion, human listeriosis is rare in the Netherlands. However, the clinical manifestations are severe, with meningitis and septicaemia predominating. Pregnancy-related listeriosis is rare but has severe consequences for pregnancy outcome. In addition, the mortality rate is high. Therefore, advice currently given to pregnant women to avoid high-risk foods, e.g. soft cheeses, smoked salmon, and raw milk, should be continued. As patients with other known predisposing conditions are also at increased risk, the same dietary recommendations should be given to them. In clinical practice, this advice is already given to patients with malignancies. Another way to reduce morbidity and mortality due to listeriosis is to implement preventive measures in the food industry, such as microbiologic monitoring of high-risk food products, which has proven effective in France [41].