Epidemiology of Campylobacter jejuni infections in Sweden, November 2011–October 2012: is the severity of infection associated with C. jejuni sequence type?

Abstract

Background

Campylobacter jejuni is among the most frequent causes of bacterial gastroenteritis in Europe. Over 8,000 C. jejuni multilocus sequence typing sequence types (STs) have been described; ST-21 and ST-45 have been identified as the most frequent types in all human studies so far. In contrast to other STs, ST-22 has been associated with the Guillain–Barré syndrome and ST-677 was recently linked to severe systemic infections in Finland. We investigated risk factors associated with hospitalisation in individuals with C. jejuni infections acquired in Sweden.

Methods

A total of 1,075 individuals with domestically acquired C. jejuni infection diagnosed between November 2011 and October 2012 in Sweden were included in this retrospective cohort study. Typing data for the isolates as well as clinical data including hospitalisation dates and diagnosis codes for individuals with C. jejuni infection were obtained. Factors associated with hospitalisation and length of hospitalisation were investigated by multivariable analysis.

Results

A total of 289 individuals were hospitalised due to C. jejuni infection (26.8%); those with co-morbidities were over 14 times more likely to become hospitalised than those without (odds ratio [OR]: 14.39, 95% confidence interval [CI]: 6.84–30.26). Those with underlying co-morbidities were also hospitalised longer than those without (4.22 days vs. 2.86 days), although this was not statistically significant. C. jejuni ST-257 (OR: 2.38; CI: 1.08–5.23), but not ST-22 or ST-677, was significantly associated with hospitalisation.

Conclusion

ST-677 was not associated with increased hospitalisation or a longer hospital stay in our study whilst ST-257 was. However, individuals with C. jejuni infections were generally more frequently hospitalised than previously demonstrated; this requires further consideration including possible targeted interventions.

• gastrointestinal pathogens
• epidemiology
• sequence type
• Campylobacter

Campylobacter is the most common zoonotic bacterial cause of gastroenteritis in Europe with over 200,000 laboratory-confirmed cases occurring yearly (1). Although the prognosis of Campylobacter gastroenteritis is typically good, infections have been linked to systemic invasive disease and post-infectious complications such as reactive arthritis and Guillain–Barré syndrome (GBS) (2, 3). Data on campylobacteremia are rare, but it is known to occur mostly in the elderly and immunocompromised individuals (4, 5). In Sweden, more than 90% of the human Campylobacter gastroenteritis cases are due to C. jejuni (6).

Genotyping has shown C. jejuni to be genetically diverse (7). Multilocus sequence typing (MLST) is used to characterise C. jejuni into arbitrary sequence types (STs), which can be further grouped into clonal complexes (CCs) (7, 8). MLST measures the DNA sequence variations in a set of housekeeping genes and characterises strains according to their sequencing profile; sequences differing at even a single nucleotide are assigned as different alleles and thus as different ST. The genotyping data are stored in an online database facilitating international comparisons (www.pubmlst.org/campylobacter/) (8). To date, more than 8,000 C. jejuni STs have been identified. Some genotypes are more frequently isolated from human than the others; ST-21 and ST-45 have been identified as the most frequent types in all human studies so far (9–14).

As most Campylobacter infections are thought to be sporadic in industrialised countries, the sources of infections remain mostly uninvestigated (15). Based on epidemiological studies, eating improperly cooked meat, handling raw chicken meat, contact with pets, drinking untreated water, and swimming in contaminated natural water have been associated with acquiring Campylobacter infection (9, 15–19). Previous MLST-based source attribution studies have shown that a majority of human C. jejuni isolates are attributable to chicken (20–23), whereas the minority of infections have also been linked to cattle, water, and wild birds (21, 24). Although very little is known about risk factors associated with the severity of infection (25), certain C. jejuni types have been reported to be associated with severe infections. A link between ST-22 and GBS has been proposed (7, 26), and ST-677 was recently linked to severe systemic infections (27). Furthermore, the infections caused by ST-677 C. jejuni have been previously shown to be associated with more frequent hospitalisation and a longer hospital stay than infections caused by other STs (13). Old age has also been generally associated with a longer hospital stay (28).

In Sweden, it is mandatory to report all laboratory-confirmed Campylobacter infections to the Public Health Agency (www.folkhalsomyndigheten.se). Currently more than 8,000 cases are notified per year and 40% of those are of domestic origin. Campylobacter in broilers has been monitored since 1991 in Sweden (29), and a significant decrease in prevalence of Campylobacter-positive broiler flocks has been seen since the 1990s. However, the incidence of human campylobacteriosis has not followed that trend (30). The objective of our study was to determine the risk factors associated with the hospitalisation linked to domestically acquired C. jejuni infections in Sweden in order to better target the future Campylobacter interventions. From that, we have also estimated the total number of hospital bed days due to domestic Campylobacter infections in Sweden.

Material and methods

Study design

We performed a retrospective study of individuals with domestically acquired C. jejuni infection in order to assess factors associated with related hospitalisations in Sweden between November 2011 and October 2012.

Description of cohort

A total of 8,585 notifications of cases with laboratory-confirmed Campylobacter infection were received at the Public Health Agency of Sweden during the study period, and 3,434 from those were classified as having been acquired in Sweden. Clinical microbiological laboratories in Sweden were asked to submit all isolates obtained from domestically acquired Campylobacter cases to the National Veterinary Institute (SVA) in Uppsala for MLST typing during the study period; a total of 1,914 samples were received from 25 laboratories. From these samples, 1,139 were selected for typing by stratified random sampling to generate an equal sampling fraction through the year and the region.

Microbiological data

A total of 1,139 C. jejuni stool isolates obtained from domestically acquired cases between November 2011 and October 2012 were included in our study. These isolates had already been typed at the SVA according to the previously published MLST protocol (7). Data collected included Swedish personal identification number, age, sex, and name of the laboratory where primary diagnostic testing was performed as well as the date of sampling and MLST type.

Clinical data

The personal identification numbers from all 1,139 C. jejuni cases were used to match the microbiological data to the registry of hospitalisation from the Swedish Board of Health and Welfare. Clinical data obtained included all hospitalisation events for these patients from January 2011 to December 2012, including admission and discharge dates, primary diagnosis, and other diagnosis encoded with the ICD-10 system. Hospitalisation events occurring within 2 weeks from the C. jejuni isolation were considered potentially related to the Campylobacter infection; these diagnostic codes were checked individually and if unlikely to be related to Campylobacter infection (i.e. road traffic accident or burns), hospitalisation was considered as non-related. In addition, patients without matching hospitalisation records or hospitalisation outside the window period were considered not having been hospitalised due to Campylobacter infection.

Definitions

Campylobacter infection was defined as domestic if the individual had not travelled abroad in the 2-week period before the laboratory diagnosis. Co-morbidity was defined as having a chronic underlying health condition, including chronic heart, lung, kidney, and liver disease, or being immunocompromised (any form of neoplasm or acquired immunodeficiency) based on provided ICD-10 codes. Typed isolates were excluded from this study if it was not possible to match the sample identifier to a record from registry of hospitalisations from the Swedish Board of Health and Welfare.

Statistical analysis

The factors associated with two separate outcome measures, hospitalisation (yes/no) and length of hospitalisation (in days), were investigated. The factors included C. jejuni ST, county of residence, age, sex, and co-morbidity. Variables which were present in less than 33 cases (i.e. detection frequency <3%) were omitted from the analysis; P-values under 0.05 were considered statistically significant.

1. Hospitalisation. Univariable analyses to investigate associations between the risk factors and hospitalisation were performed using the 2×2 chi-square test. Risk ratios (RRs) with 95% confidence intervals [CI] were calculated. To adjust the RRs for potential confounders such as age, sex, and co-morbidity, a multivariable analysis using a logistic regression model was performed. Risk factors with P<0.20 in the univariable analysis were included in a multivariable model built through stepwise forward selection using the likelihood ratio test.

2. Length of hospitalisation. Univariable analysis to assess the association between the risk factors and length of hospitalisation (i.e. time to discharge) was performed using a Cox proportional hazard model. Hazard ratios with a 95% CI were calculated. Risk factors with P<0.20 in the univariable analysis were included in a multivariable model built through stepwise forward selection using the likelihood ratio test.

3. Estimation of hospitalisation linked to Campylobacter infections in Sweden. Based on the reported numbers of Campylobacter infections and the hospitalisation rate obtained in this study, we have estimated the total number of hospitalisations linked to domestic Campylobacter infections in Sweden during the 1-year study period.

Ethics

The confidentiality of study subjects was protected via anonymisation of the data at the SVA. The study was approved by the Uppsala Regional Ethics Committee (2011/306).

Results

A total of 1,139 C. jejuni isolates were included in this study, but 64 of these were excluded as it was not possible to match their sample identifiers to the data from the Swedish Board of Health and Welfare. The remaining 1,075 C. jejuni isolates consisted of 119 distinct C. jejuni STs and 26 CCs. The most frequent STs identified and included in statistical analyses were ST-21 (139 isolates; 12.9% [belonging to the CC-21]), ST-50 (116; 10.8% [CC-21]), ST-19 (104; 9.7% [CC-21]), ST-45 (105; 9.8% [CC-45]), ST-677 (90; 8.4% [CC-677]), ST-48 (88; 8.2% [CC-48]), and ST-257 (41; 3.8% [CC-257]); all the remaining STs were identified in less than 33 cases and were clustered into their own group named ‘other’ (392 isolates; 36.5%) (Table 1). Although STs were further grouped into CCs, analyses have been performed based on STs as genetic diversity is less within STs than within the CCs (and results based on the CC were in line with the results based on STs; data not shown). Overall, 76 STs were represented by only one or two isolates in the collection. Samples were obtained from all 21 Swedish counties; those counties which had less than 33 Campylobacter cases in total (i.e. frequency <3% regardless of ST) were further clustered into their own group named ‘other’ for multivariable analysis. The sample was geographically and temporally representative of all domestic cases reported in Sweden during the study period (data not shown).

Table 1. The frequency of detection and hospital admission, length of hospital admission and mean age of individuals infected with C. jejuni MLST sequence types (ST) in Sweden, between November 2011 and October 2012

	Identified in this study			Admitted to hospital
C. jejuni type	Number	Percentage	Mean age (range)	Number	Percentage	Mean age (range)	Length of hospitalisation (days)
ST-21	139	12.9	37.3 (0–87)	28	9.7	45.2 (0–87)	3.04
ST-50	116	10.8	39.1 (1–88)	37	12.8	42.7 (2–85)	2.51
ST-45	105	9.8	48.1 (1–87)	34	11.8	57.3 (1–87)	3.41
ST-19	104	9.7	41.6 (2–73)	20	6.9	48.7 (16–73)	3.15
ST-677	90	8.4	50.5 (1–86)	25	8.7	57.9 (1–86)	3.28
ST-48	88	8.2	40.6 (0–89)	24	8.3	44.9 (1–89)	4.21
ST-257	41	3.8	39.8 (0–77)	17	5.9	45.2 (10–77)	2.65
Other^a	392	36.5	43.0 (0–95)	104	36.0	51.9 (2–95)	3.28
All	1,075		42.5 (0–95)	289		50.0 (0–95)	3.20

^aSequence types which were identified in les s than 33 isolates have been combined into this ‘other’ group.

A total of 289 individuals with C. jejuni infection were hospitalised (26.9%; Table 1): the smallest proportion of hospitalisation was noted for those infected with C. jejuni ST-19 (19.4%) and highest for those infected with ST-257 (40%). Most individuals (98%; 292/298) were hospitalised within 7 days (from 2 days before to 5 days after) from microbiological diagnosis of C. jejuni; the remaining six were hospitalised either 4 days (three individuals) or 5 days (three individuals) before the diagnosis. Most individuals hospitalised with C. jejuni infection were admitted either with the diagnosis of enteritis or colitis (n=265) or with symptoms of diarrhoea and abdominal pain (n=6), and the reason for admission was not given for five individuals. However, Campylobacter-specific ICD-10 diagnostic code was used for 148 hospitalised individuals (A405; 51%). A total of seven individuals with colitis were known to have either ulcerative colitis (n=4), diverticular disease (n=1), or Crohn's disease (n=2). Furthermore, one individual was admitted with GBS and one with defined Campylobacter sepsis. Isolates from gastroenteritis were represented in all STs, whereas most isolates obtained from individuals with inflammatory bowel disease were found in ST-45 (4/6), GBS in ST-22 (1/1), and sepsis in ST-677 (1/1).

Mean age of individuals infected with C. jejuni was 42.5 years, whereas individuals hospitalised with C. jejuni infections were generally older with a mean age of 50 years (Table 1). The mean age of individuals infected and hospitalised with certain C. jejuni ST varied. The individuals infected and hospitalised with C. jejuni ST-677 were the oldest (mean age of 50.5 and 57.9 years, respectively). Furthermore, individuals infected with C. jejuni ST-21 (37.3 years) and individuals hospitalised with C. jejuni ST-50 (mean age of 42.7 years) were the youngest.

Risk factors associated with hospitalisation and the length of hospitalisation

Infection with C. jejuni ST-257 (RR: 1.57; 95% CI: 1.08–2.30), an age of 60 years or older (RR: 2.13; 95% CI: 1.76–2.57), and underlying co-morbidity (RR: 4.49; 95% CI: 2.91–6.92) were factors significantly associated with hospitalisation among patients who had C. jejuni infection in Sweden based on univariable analysis (Table 2). Most of the cases with reported co-morbidities, including being immunocompromised and those with chronic underlying conditions admitted to hospital, also had a laboratory-confirmed C. jejuni infection (83%; 80/97), whereas a smaller proportion of individuals without these co-morbidities were hospitalised with C. jejuni infection (22%; 209/978). Furthermore, age under 20 years (RR: 0.69; 95% CI: 0.50–0.95) or between 40 and 60 years (RR: 0.69; 95% CI: 0.55–0.87) as well as being diagnosed in Halland and Skåne counties (RR: 0.45; 95% CI: 0.21–0.97 and RR: 0.67; 95% CI: 0.46–0.97, respectively) were identified as statistically significant protective factors for hospitalisation coinciding with C. jejuni infection. No significant differences in proportion of individuals with co-morbidities or in age distribution of individuals were observed between counties.

Table 2 Potential covariates and their association with hospitalisation evaluated in univariable analysis using a 2×2 chi-square test in laboratory-confirmed domestic cases of C. jejuni in Sweden, November 2011–October 2012

Variable	Hospitalised/total	Percent	RR	95% CI	P
C. jejuni sequence type
ST-21	28/139	20.1	0.72	0.51–1.02	0.07
ST-50	37/116	31.9	1.21	0.91–1.61	0.24
ST-19	20/104	19.2	0.69	0.46–1.04	0.08
ST-45	34/105	32.4	1.23	0.02–1.66	0.22
ST-677	25/90	27.8	1.04	0.73–1.47	0.92
ST-48	24/88	27.3	1.02	0.71–1.45	1.00
ST-257	17/41	41.5	1.57	1.08–2.30	0.049
Other^a	104/392	26.5	0.98	0.79–1.20	0.89
County
Södermanland	10/34	29.4	1.09	0.65–1.87	0.89
Dalarna	12/38	31.6	1.18	0.73–1.91	0.63
Gävleborg	13/35	37.1	1.40	0.90–2.18	0.23
Västra Götaland	46/214	21.5	0.76	0.57–1.01	0.06
Halland	6/48	12.5	0.45	0.21–0.97	0.03
Jönköping	6/28	21.4	ND^b
Kalmar	8/22	36.4	ND^b
Blekinge	5/16	31.3	ND^b
Värmland	6/33	18.2	0.67	0.32–1.39	0.35
Östergötland	11/35	31.4	1.18	0.71–1.94	0.67
Norrbotten	6/18	33.3	ND^b
Skåne	24/128	18.8	0.67	0.46–0.97	0.04
Stockholm	67/238	28.2	1.06	0.84–1.34	0.68
Västernorrland	5/13	38.5	ND^b
Västerbotten	13/36	36.1	1.36	0.87–2.12	0.28
Uppsala	9/29	31.0	ND^b
Gotland	7/12	58.3	ND^b
Västmanland	7/25	28.0	ND^b
Kronoberg	7/19	36.8	ND^b
Örebro	17/46	37.0	1.40	0.95–2.07	0.16
Jämtland	4/8	50.0	ND^b
Male gender
No	169/638	26.5	–	–	–
Yes	120/437	27.5	1.04	0.85–1.27	0.78
Age (years)
<20	33/169	19.5	0.69	0.50–0.95	0.02
20–39	66/292	22.6	0.80	0.63–1.02	0.07
40–60	74/358	20.7	0.69	0.55–0.87	0.0015
>60	116/256	45.3	2.13	1.76–2.57	<0.0001
Co-morbidity
No	209/978	21.4	–	–	–
Yes	80/97	82.5	4.49	2.91–6.92	<0.0001

Variables with more than two categories (ST, county, and age group) were dichotomised by category, using all the remaining observations were as the reference group (e.g. ST-21 vs all other STs). Results for variables shown to be statistically significantly associated with hospitalisation (P>0.05) are shown in bold face.

^aContains all other C. jejuni sequence types identified in this study.

^bND = not determined for variables with prevalence <3%.

After correction for potential confounders, co-morbidity (OR: 13.89; 95% CI: 7.88–24.46), an age of 60 years or older (OR: 1.98; 95% CI: 1.22–3.25), and C. jejuni ST-257 (OR: 2.35; 95% CI: 1.15–4.81) were found to be independently associated with hospitalisation (Table 3). In addition, Västra Götaland (OR: 0.59; 95% CI: 0.36–0.96), Halland (OR: 0.30; 95% CI: 0.21–0.97), and Skåne (OR: 0.43; 95% CI: 0.46–0.97) counties were identified as statistically significant protecting factors for hospitalisation with C. jejuni infection in multivariable analysis.

Table 3 Result of multivariable logistic regression analysis of variables associated with hospitalisation of laboratory-confirmed domestic cases of C. jejuni in Sweden, November 2011–October 2012

Variable	OR	95% CI	P
C. jejuni sequence type	0.98	0.93–1.05	0.62
ST-21	0.81	0.48–1.37	0.43
ST-50	1.59	0.96–2.62	0.07
ST-19	0.71	0.39–1.28	0.25
ST-45	1.14	0.67–1.94	0.63
ST-677	0.99	0.56–1.77	0.98
ST-48	1.28	0.72–2.29	0.40
ST-257	2.35	1.15–4.81	0.017
County	1.02	1.00–1.04	0.12
Södermanland	0.89	0.65–1.87	0.81
Dalarna	0.74	0.73–1.91	0.50
Gävleborg	1.09	0.90–2.18	0.85
Västra Götaland	0.59	0.36–0.96	0.033
Halland	0.30	0.21–0.97	0.018
Värmland	0.59	0.32–1.39	0.30
Östergötland	0.86	0.71–1.94	0.72
Skåne	0.43	0.46–0.97	0.007
Stockholm	0.76	0.84–1.34	0.24
Västerbotten	1.14	0.87–2.12	0.75
Örebro	1.31	0.95–2.07	0.48
Age (years)	1.22	1.05–1.43	0.012
20–39	0.99	0.61–1.62	0.61
40–60	0.79	0.48–1.27	0.33
>60	1.98	1.22–3.25	0.006
Co-morbidity (yes)	13.89	7.88–24.46	<0.0001

The ‘other’ categories were used as a reference groups for ST and county variables. Results for variables shown to be statistically significantly associated with hospitalisation (P>0.05) are shown in bold face.

None of the factors investigated were significantly associated with duration of hospitalisation based on univariable analysis (Table 4) or multivariable analysis (data not shown). Duration of hospitalisation was not statistically different between cases infected with ST-677 or any other C. jejuni STs. It varied from 2.51 days (ST-50) to 4.21 days (ST-48), with the mean hospital stay of 3.20 days. Duration of hospitalisation was not statistically significant between different age groups, but those older than 60 years were hospitalised longer than those under 60 years of age (4.07 days vs. 2.53 days [<20 years], 2.14 days [20–39 years], and 3.12 days [40–59 years]). Furthermore, those infected with C. jejuni who had underlying co-morbidities were also hospitalised for longer than those without (4.16 days vs. 2.84 days).

Table 4 Variables investigated for their association with length of hospitalisation evaluated in univariable analysis using Cox regression in laboratory-confirmed domestic cases of C. jejuni in Sweden, November 2011–October 2012

Variable		Me an stay (median)	HR	95% CI	P
C. jejuni sequence type	Other	3.28 days (3 days)	–	–	–
	ST-21	3.04 days (2 days)	1.12	0.74–1.71	0.56
	ST-50	2.51 days (2 days)	1.34	0.92–1.95	0.13
	ST-19	3.15 days (2 days)	1.01	0.63–1.63	0.97
	ST-45	3.41 days (3 days)	0.85	0.58–1.26	0.42
	ST-667	3.28 days (3 days)	0.98	0.63–1.51	0.91
	ST-48	4.21 days (4 days)	0.96	0.62–1.49	0.86
	ST-257	2.65 days (2 days)	1.49	0.89–2.50	0.13
County	Other	3.25 days (2.5 days)	–	–	–
	Eskilstuna	2.30 days (2 days)	1.31	0.67–2.56	0.42
	Falun	2.92 days (3 days)	0.97	0.53–1.81	0.94
	Gävle	2.77 days (2 days)	1.07	0.59–1.95	0.82
	Göteborg	3.28 days (3 days)	0.89	0.61–1.30	0.56
	Halmstad	3.83 days (4 days)	0.89	0.39–2.06	0.79
	Karlstad	2.67 days (2.5 days)	0.84	0.36–1.93	0.68
	Linköping	4.18 days (3 days)	0.77	0.40–1.45	0.42
	Malmö	3.54 days (2.5 days)	0.85	0.52–1.34	0.46
	Stockholm	2.81 days (2 days)	1.30	0.92–1.84	0.13
	Umeå	2.69 days (2 days)	1.22	0.68–2.23	0.50
	Örebro	4.71 days (3 days)	1.05	0.62–1.81	0.84
Male gender	No	3.11 days (2 days)	–	–	–
	Yes	3.34 days (3 days)	1.01	0.80–1.27	0.95
Age (years)	<20	2.53 days (2 days)	–	–	–
	20–39	2.14 days (2 days)	1.02	0.67–1.54	0.92
	40–60	3.12 days (2 days)	0.83	0.56–1.25	0.38
	>60	4.07 days (3 days)	0.73	0.50–1.07	0.11
Co-morbidity	No	2.84 days (2 days)	–	–	–
	Yes	4.16 days (3 days)	0.92	0.73–1.17	0.51

Estimation of hospitalisation linked to the domestic C. jejuni infections in Sweden

As the mean length of hospitalisation was 3.2 days, and 289 individuals with domestically acquired C. jejuni infections were hospitalised, it can be calculated that a total of 925 hospital bed days were used (95% CI: 879–971). As only a proportion of domestic C. jejuni infections were included in our study (1,075 from 3,434), the true numbers would be bigger (estimated around 2,954 bed days). Furthermore, one case of GBS among our study population would translate into three cases among all hospitalised individuals due to domestic C. jejuni infections in Sweden during the study period.

Discussion

This is the first study to determine the risk factors associated with the hospitalisation and the length of hospitalisation of individuals with domestically acquired C. jejuni infections in Sweden. Our study included 1,075 C. jejuni–infected individuals, representing approximately 30% of all reported domestic C. jejuni cases in Sweden during the 1-year study period. A high diversity of MLSTs among human C. jejuni isolates with a total of 119 distinct STs were observed, from which the seven frequent STs (ST-21, ST-50, ST-19, ST-45, ST-677, ST-48, and ST-257) covered 64% of all isolates in accordance with previously published studies from the United Kingdom, Finland, Denmark, and Canada (9–14, 24) (26). These STs were recently shown to have been responsible for a large proportion of domestic C. jejuni infections in Sweden already in 2000, and have also been isolated from Swedish broilers (23). It has been suggested that the poultry is likely the most important source of domestic C. jejuni infections (19).

Although Campylobacter infections are usually considered mild and self-limiting, they can also lead to more severe infections requiring hospitalisation (3–5, 31). Up to 26.9% of individuals included in this study were admitted to hospital with the domestically acquired C. jejuni infection. We have matched the data on Campylobacter infections obtained from the Public Health Agency of Sweden and from the registry of hospitalisations from the Swedish Board of Health and Welfare. Based on our data, only 50% of hospitalisations were registered with Campylobacter-specific ICD-10 coding (A04.5), and the remaining were registered with non-specific gastroenteritis and colitis coding. The numbers on the hospitalisations related to Campylobacter infections are likely underestimated in studies where only specific ICD-10 codes are used without matching (31–34). Only 5% of reported Campylobacter cases were hospitalised according to the review of three national databases over a 4-year period in Canada (32), whereas an annual hospitalisation rate of 12.3% was obtained in a Danish registry-based study (33) and 10% in US surveillance study (34). As all the studies have used different methods to obtain the hospitalisation estimates, we cannot state if the hospitalisation rates are truly increasing, if they are higher since a smaller proportion of milder infections have been sampled, or simply more accurately estimated than recorded previously. This needs to be investigated further.

The correlation between co-morbidities and hospitalisation due to domestic C. jejuni infection is not surprising, given that most co-morbidities result in decreased immune function and thus more likely will lead to symptomatic C. jejuni enterocolitis. Co-morbidity category also included all known immunocompromised individuals (i.e. those with known malignancy). Furthermore, individuals with co-morbidities are more likely to seek health care with gastrointestinal symptoms and are more likely to be tested for Campylobacter than those without co-morbidities. There was also a trend towards more severe infections among the individuals with co-morbidities; they were hospitalised for longer than those without (4.16 days vs. 2.84 days). Although these differences were not statistically significant, they suggest more severe infections in those with co-morbidities. In a previous study, a high hospitalisation rate of 47.6% was described in patients with ulcerative colitis (35). However, the main focus of the Arora study (35) was to determine the risk factors for C. jejuni infection and not to investigate the risk factors associated with hospital admissions due to Campylobacter infections. Furthermore, previous studies have shown that those aged over 60 years are three times more likely to be hospitalised with Campylobacter infection than younger individuals (34), and that age >60 years can be a risk factor for a longer hospital stay (28). It is most likely that the older individuals have more diagnosed co-morbidities than the younger ones; the factor not considered in those studies. We clearly demonstrate that co-morbidity is the most significant risk factor for hospitalisation while having C. jejuni infection, not old age. Indeed, over 60% of all those hospitalised were younger than 60 years of age.

It is not fully understood why the hospitalisation rates were lower in the southwestern parts of Sweden (Västra Götaland, Halland, and Skåne counties) than elsewhere. None of the Swedish microbiology laboratories were using molecular methods for Campylobacter detection during the study period. A potential explanation could be differences in study population, such as overrepresentation of the elderly or those with co-morbidities, but no significant differences were observed between counties. The lower hospitalisation rates in these counties could also reflect differences between counties in primary health care policies with regard to testing patients with gastrointestinal illness for Campylobacter, where counties with a more liberal testing policy would be identified as ‘protective’ in this literature due to the relatively larger number of cases detected, and not hospitalised. The finding could also be due to differences in hospitalisation policy (i.e. stricter criteria for hospital admissions).

Only one study has previously investigated the association between C. jejuni STs and hospitalisation; based on that it was expected that C. jejuni ST-677 would be associated with more frequent hospitalisations and also with increased length of hospitalisation than other Campylobacter types (13). This hypothesis was also supported with more recent evidence demonstrating that C. jejuni ST-677 is specifically associated with invasive infections (i.e. campylobacteremia) (27). In our study, the association between C. jejuni ST-677 and the increased hospitalisations or a longer hospital stay could not be demonstrated. However, it should be noted that the Feodoroff study investigated blood culture isolates of C. jejuni collected over a 10-year study period, whereas the number of invasive C. jejuni infections during our 1-year study period would have been minimal (36). Interestingly, individuals infected with another less common C. jejuni type, ST-257, were shown to be at increased risk of hospital admissions in our study. Although ST-257 was recently isolated from domestic dogs in Sweden (37), only 4% of dogs were positive for C. jejuni and thus the importance of dogs as a source of human infections needs to be clarified further. In addition, ST-50 was linked to the highest number of Campylobacter–related hospitalisations in Sweden (37/289). Both of these types, ST-257 and ST-50, have been previously identified from Swedish broilers (23).

Although the pathogenesis of C. jejuni remains poorly understood, some C. jejuni types have been associated with severe infections. In our study, one individual was diagnosed with GBS and C. jejuni ST-22 infection, whereas another individual with systemic invasive infection had C. jejuni ST-677, consistent with the previous literature (7, 26) (27). Furthermore, we identified another C. jejuni type, ST-257, as a risk factor significantly associated with hospitalisation. This ST has not been linked to a specific disease outcome and thus the reason(s) for the observed increased rate of hospitalisation is not currently known. However, C. jejuni ST-257 was shown to be the most virulent type measured as larval survival in the insect Galleria mellonella model (38). It is likely that the genetic differences between C. jejuni types will determine their pathogenicity, as previously demonstrated by combining the data from whole-genome sequencing and phenotypical characterisation of C. jejuni ST-677 (27, 39).

Although most Campylobacter infections are self-limiting and do not require treatment, antibiotics are often given to immunocompromised individuals (25), and thus it is important to consider possibility of resistance. In that context, it is important to note that a previous study from Slovakia demonstrated equally high frequency of antibiotic resistance among isolates obtained from humans, animals, and food. These included universal persistence of ciprofloxacin-resistance among C. jejuni ST-50 isolates (40); the second most common C. jejuni type in Sweden. High levels of ciprofloxacin resistance were also previously identified not only among C. jejuni ST-50 isolates but also among ST-257 isolates in Belgium (41). However, much lower levels of ciprofloxacin-resistance was observed among these types in the United Kingdom (10). Data on antibiotic resistance in human Campylobacter isolates is not routinely collected in Sweden, but a high level of ciprofloxacin resistance for untyped human Campylobacter isolates was reported in 2002–2011 (49%; (42)). For these reasons, further systematic monitoring of antimicrobial resistance in Sweden for human Campylobacter should also be considered.

Limitations

The clinical data obtained in this study were based on the diagnostic codes used and reported. As they have been assigned by individual doctors, they might differ according to the doctor and hospital practice. In addition, since only the diagnostic codes were used, we do lack other information on disease severity (i.e. data on ITU admission, need for IV fluids, antibiotic treatment, and resistance). It should also be noted that only individuals who have been diagnosed with certain disease entities will have been given the diagnostic codes (i.e. young individuals with newly diagnosed neoplasm might still be under ongoing investigations and thus definite diagnosis may not have been reported as yet). However, these limitations are likely to result in underestimation of associations and cannot be minimised within this study design. In addition, only Campylobacter cases who sought medical care could be included in this study (prerequisite for sampling) and thus milder infections that did not require medical attention were not included in this study. This may result in an under- or over-estimation of the effect of genotype on hospitalisation since cases of mild illness may not be uniformly distributed among genotypes of C. jejuni, as well as overestimation of the population proportion of hospitalised cases.

Recommendations

We have demonstrated that over a quarter of all individuals diagnosed with domestic C. jejuni infection are hospitalised in Sweden, and that those infected with ST-50 or ST-257 are slightly more likely to be subject to hospital care than those infected with other STs. Although over 50% of all hospitalised individuals were less than 60 years of age and did not have any co-morbidities and were not known to be immunocompromised in this study, the biggest risk factors for hospitalisation identified included co-morbidities and old age. We have shown that individuals with co-morbidities are at a 14-time higher risk of becoming hospitalised with a domestic C. jejuni infection than those without. As these co-morbidities are often seen in older people, the burden of domestic C. jejuni infections is likely to increase even further in the future, considering the ageing population trend. Targeted public health messages including strict advice on kitchen hygiene and safe drinking water sources for those with co-morbidities as well as further investigations on antimicrobial resistance in Campylobacter in humans in Sweden are required.

Conflict of interest and funding

The authors have not received any funding or benefits from industry or elsewhere to conduct this study.

Acknowledgements

We thank Boel Harbom for laboratory coordination during the collection of the isolates at the Swedish National Veterinary Institute and Mattias Myrenås, at the same institute, for molecular typing. The Campylobacter Source Attribution Study Group (www.sva.se/en/research/researches/attributing-human-cases-of-campylobacteriosis-to-their-sources-as-a-tool-for-targeted-interventions) consists of 25 clinical microbiological laboratories throughout Sweden, who assisted with the collection of Campylobacter isolates from domestic human cases. This data collection was conducted for a study funded by the Swedish research council, investigating the sources of human campylobacteriosis in Sweden. We also thank Henrik Källberg at the Public Health Agency of Sweden for statistical assistance and Aftab Jasir at European Centre for Disease Control for continuing support through this project and reviewing this manuscript.