Factors associated with increasing campylobacteriosis incidence
in Michigan, 2004–2013
W. CHA
1
,T.HENDERSON
2
,J.COLLINS
2
AND S. D. MANNING
1
*
1
Department of Microbiology and Molecular Genetics, Michigan State University, East Lansing, MI, USA
2
Michigan Department of Health and Human Services, Lansing, MI, USA
Received 26 May 2016; Final revision 20 June 2016; Accepted 27 June 2016;
first published online 4 August 2016
SUMMARY
This study was conducted to examine the incidence trend of campylobacteriosis in Michigan over
a 10-year period and to investigate risk factors and clinical outcomes associated with infection.
Campylobacter case data from 2004 to 2013 was obtained from the Michigan Disease
Surveillance System. We conducted statistical and spatial analyses to examine trends and identify
factors linked to campylobacteriosis as well as ecological associations using animal density data
from the National Agricultural Statistics Service. An increasing trend of Campylobacter incidence
and hospitalization was observed, which was linked to specific age groups and rural residence.
Cases reporting ruminant contact and well water as the primary drinking source had a higher
risk of campylobacteriosis, while higher cattle density was associated with an increased risk at the
county level. Additional studies are needed to identify age-specific risk factors and examine
prevalence and transmission dynamics in ruminants and the environment to aid in the
development of more effective preventive strategies.
Key words: Campylobacter, epidemiology, foodborne infections, surveillance.
INTRODUCTION
Campylobacter, a Gram-negative zoonotic pathogen,
is one of the most widespread infectious agents in
the world [1]. It is not only the leading cause of gastro-
enteritis in humans, but it can also cause serious long-
term sequelae like Guillain–Barré syndrome, reactive
arthritis, and inflammatory bowel disease [2]. The an-
nual incidence of campylobacteriosis varies between
countries, but the numbers of reported cases have
been increasing worldwide in the past decade [1].
Importantly, estimated burden of disease is consider-
able, as it is estimated to cost US$1.7 billion/year in
the United States alone [3].
Chicken consumption and handling have been iden-
tified as major risk factors for campylobacteriosis [4]
as broiler chickens are frequently colonized with C.
jejuni. Recent studies have also reported attributable
rates for cattle in up to 19·3% of Campylobacter
cases [5, 6], suggesting that cattle serve as another im-
portant source for human infections. Indeed, raw milk
and cheese have been implicated in several outbreaks
[7]. Campylobacter is also widespread in the environ-
ment including water and soil, where it has been
found to survive for several months [8, 9]. Water, es-
pecially, has been identi fied as an important source
for Campylobacter infections and has been linked to
outbreaks as well [10, 11]. Human-to-human trans-
mission via the faecal–oral route has been reported;
however, zoonotic or foodborne transmission is the
predominant mode. With the high prevalence of cam-
pylobacteriosis reported throughout the world, foreign
* Author for correspondence: S. D. Manning, PhD, MPH,
Michigan State University, 1129 Farm Lane, Room 194,
E. Lansing, MI 48824, USA.
(Email: mannin71@msu.edu)
Epidemiol. Infect. (2016), 144, 3316–3325. © Cambridge University Press 2016
doi:10.1017/S095026881600159X
https://doi.org/10.1017/S095026881600159X Published online by Cambridge University Press
travel has also emerged as an important risk factor
[12].
Seasonal variation has been described for Campylo-
bacter infections. A significantly higher incidence of
campylobacteriosis has been reported in warmer sea-
sons in different countries and from different sources
such as animals and water [13, 14]. The reason behind
this seasonality is not known, but has been suggested
to be the result of multiple factors including longer
survival of Campylobacter sp. in the environment,
increased shedding levels in animal reservoirs [15],
and changes in human behaviour [16]. Spatial deter-
minants such as urban vs. rural settings, have also
been linked to campylobacteriosis incidence [17, 18],
suggesting the need to assess environmental factors
when conducting risk-factor analyses.
In the United States, the Foodborne Diseases
Active Surveillance Network (FoodNet), which moni-
tors incidence trends of foodborne pathogens,
reported an increase in the incidence of campylobac-
teriosis, from 12·3 cases/100 000 in 2009 to 14·2
cases/100 000 in 2012 [19]. This increasing trend may
be partly due to improvements in detection methods
and enhanced surveillance, although other risk factors
such as geographical location are also likely to be im-
portant. Considerable variation in campylobacteriosis
incidence was observed among FoodNet sites
throughout the surveillance period, ranging from 7
cases/100 000 in Tennessee to 34·3 cases/100 000 in
California [19]. Because no significant differences
were identified for key risk factors, medical care-
seeking behaviour, or medical practices between
sites, it is probable that site-specific risk factors like
climate or number of anima l reservoirs are also con-
tributing to the variable rates [20].
Because Michigan is not a FoodNet site, we sought
to calculate the incidence of campylobacteriosis over a
10-year period using data collected via the Michigan
Department of Health and Human Services
(MDHHS) and identify risk factors associated with
disease. We hypothesized that the incidence of
Campylobacter infections has increased temporally
and specifi c factors are associated with increasing inci-
dence. We also investigated clinical outcomes and
demographics to assess the disease burden of campy-
lobacteriosis in Michigan and identify characteristics
associated with increased risk of infection. Ongoing
surveillance efforts are vital to monitor incidence
trends and investigate risk factors for disease in differ-
ent geographical locations in order to more effectively
define preventive measures.
METHODS
Campylobacteriosis is a reportable disease in
Michigan. A total of 7182 laboratory-confirmed
cases were reported to MDHHS with an onset date
between 1 January 2004 and 31 December 2013.
Demographic, clinical, and epidemiological data
for each case were extracted from the Michigan
Disease Surveillance System (MDSS) managed by
the MDHHS. Season was categorized based on the
onset date and travel was considered only when the
travel period was within 1 week prior to the onset of
symptoms. Data involving history of food consump-
tion and animal contact were systematically collected
from 2011 and thus, only the last 3 years of data
were used in the analysis. The water source at home
was categorized into well, municipal, bottled and
other, which included various combinations of the dif-
ferent sources. Data was considered missing for all
variables lacking data or that were reported as
‘unknown’.
Age-adjusted incidence rates (cases/100 000) were
calculated using the Bridged-Race Population
Estimates 1990–2013 dataset [
21] and U.S. 2010
standard population by the U.S. Census Bureau [22].
Statistical analyses were performed using SAS v. 9·3
(SAS Institute Inc., USA). Differences in the frequen-
cies of campylobacteriosis across variables were exam-
ined using χ
2
tests; P < 0·05 was considered significant.
Additional analyses were conducted to investigate the
association between demographic characteristics and
foreign travel history, where the prevalence ratio was
compared between individuals with and without a tra-
vel history. Multivariate analyses for hospitalization
and rural vs. urban residence were performed using lo-
gistic regression while adjusting for independent vari-
ables with a P value of <0·2 and biologically plausible
variables that could represent confounders (e.g. age,
sex). The multivariate model was constructed using
a forward stepwise method with the requirement for
a significance level of P40·1 to remain in the model.
A Geographical Information System (GIS) map
was generated in ArcGIS v. 10·2·2 (ESRI, USA)
using data from the National Center for Health
Statistics (NCHS), the bridged-race population esti-
mates, and the case numbers in this study. Based on
the NCHS classification system [23], ten Michigan
counties with large metropolitan areas were classified
as urban, while the remaining 73 counties were
defined as rural. All protocols were approved by the
Institutional Review Boards at Michigan State
Epidemiology of Campylobacter in Michigan 3317
https://doi.org/10.1017/S095026881600159X Published online by Cambridge University Press
University (IRB no. 10-736SM) and MDHHS (842-
PHALAB).
Ethical standards
The authors assert that all procedures contributing to
this work comply with the ethical standards of the
relevant national and institutional committees on
human experimentation and with the Helsinki
Declaration of 1975, as revised in 2008.
RESULTS
The average age-adjusted annual incidence was 7·22
(cases)/100 000, ranging from 6·26 in 2004 to 8·35 in
2013. An increasing trend was observed over time
with average incidence rates increasing from 6·58/
100 000 in 2004–2006 to 7·37/100 000 in 2007–2010
and 7·65/100 000 in 2011–2013. The average annual
incidence was higher for men (7·89/100 000) than
women (6·56/100 000), with an incidence rate ratio
of 1·2 (P < 0·01). The highest incidence rate was
reported in children aged <5 years (14·86/100 000)
compared to other age groups (6·76/100 000,
P<0·01). When stratified by sex, boys aged <1 year
had the highest incidence (20·5/100 000), although an
overall increasing trend in incidence was observed
after 5–19 years, particularly for cases aged between
20–29 and >50 years (Fig. 1).
Caucasians comprised 85% of the total cases in
Michigan and Campylobacter incidence was signifi-
cantly higher (6·51/100 000) in this group compared
to Asians (4·47/100 000) and African Americans
(1·56/100 000) (P<0·05). Cases self-reporting as
Asian were more likely to have a history of foreign tra-
vel [prevalence ratio (PR) 4·3], while the opposite was
observed in individuals self-reporting as African
American (PR 0·7) (Table 1). Travel destinations
were also correlated with race, as 52·4% of Asian tra-
vellers visited Asia and 47·1% of African American
travellers visited Africa. By age, individuals between
20 and 39 years were more likely to develop a
Campylobacter infection following foreign travel (PR
1·6), while the opposite was observed in children
aged <5 years (PR 0·3) and in the elderly (aged 580
years) (PR 0·1) (Table 1).
Different seasonal trends were observed for domes-
tic cases (n = 5795) compared to cases with a recent
history of foreign travel (n = 830). Domestic cases
were more common in the summer months of June
and July [odds ratio (OR) 1·6, 95% confidence interval
(CI) 1·3–1·9], while cases with foreign travel history
were more common in January and February (OR
1·8, 95% CI 1·3–2·4). For domestic cases, the season-
ality in summer (June–August) was more prominent
for individuals aged 10–59 years, especially compared
to the <1 year and 580 years (OR 1·3, 95% CI
1·1–1·6) age groups (Fig. 2).
Geographical variation in the age-adjusted inci-
dence was observed in the 83 Michigan counties
(Fig. 3). Using data from the NCHS system, ten of
the counties were defined as urban and the remainder
were rural. The incidence was significantly higher
in rural (7·78/100 000) compared to urban areas
(6·70/100 000, P<0·05). Furthermore, the incidence
increased by 41·5% from 6·45/100 000 in 2004 to
9·27/100 000 in 2013 in rural cases, relative to a
26·2% increase among urban cases (6·13/100 000 in
2004 to 7·65/100 000 in 2013).
To investigate the possibility that geographical vari-
ation is linked to factors specific to rural areas, we
conducted a case-case analysis between the urban
and rural domestic cases. The age-specific incidence
was higher for rural cases compared to urban cases,
especially for groups aged 10–19 years [incidence
rate ratio (IRR) 1·6], 20–39 years (IRR 1·3), and
>80 years (IRR 1·3). Univariate analyses demon-
strated that contact with animals (i.e. ruminants,
poultry, domestic pets) was si gnifi cantly more com-
mon for rural cases compared to urban cases as well
as consumption of raw milk, ground meat, and frozen
chicken (Table 2). The most notable difference, how-
ever, was the water source at home as rural cases
were significantly more likely to drink only well
water (OR 7·3, 95% CI 5·6–9·4). Multivariate logistic
regression controlling for age and sex ident ified con-
tact with ruminants (OR 1·6, 95% CI 1·0–2·5), con-
sumption of ground meat (OR 1·4, 95% CI 1·1–1·7),
and well water at home (OR 5·4, 95% CI 3·9–7·4) to
be independently associated with Campylobacter in-
fection in rural areas.
Given the association with campylobacteriosis and
ruminants, we evaluated the ecological association be-
tween animal density (i.e. cattle, chickens, goats,
sheep, pigs) and disease incidence by county using
data from the National Agricultural Statistics
Service for 2012 [24]. Counties with <1% of the total
number of each animal species in the state were clas-
sified as low-density counties, while counties with
51% were classified as high density. Notably, 12
counties with the highest incidence rates of campylo-
bacteriosis (13·6–35·1/100 000), all of which were
3318 W. Cha and others
https://doi.org/10.1017/S095026881600159X Published online by Cambridge University Press
also defined as rural areas (Fig. 3), were more likely to
have a high cattle density (OR 2·5, 95% CI 2·1–2·9).
No significant associations were observed for other
animal species including chicken.
Clinical outcomes also varied in cases over time; the
hospitalization rate increased from 23·3% in 2004 to
29·5% in 2013. Cases aged 560 years had a signifi-
cantly higher likelihood of hospitalization than other
Fig. 1. Average annual sex-specific campylobacteriosis incidence rate by age group and the trend of age-specific incidence
rates by year, 2004–2013.
Table 1. Demographic characteristics of Campylobacter cases in Michigan (2004–2013) by foreign-travel status
Total with known
foreign-travel status*
Foreign-travel
associated cases Domestic cases
Prevalence
ratioCharacteristic No. No. % No. %
Sex†
Male 3524 489 58·9 3035 52·4 1·1
Female 3094 341 41·1 2753 47·6 0·9
Age group (years)‡
<1 181 10 1·2 171 3 0·4
1–4 628 26 3·1 602 10·4 0·3
5–19 884 90 10·9 794 13·7 0·8
20–39 1437 267 32·2 1170 20·2 1·6
40–59 2067 314 38·2 1753 30·3 1·3
60–79 1186 117 14·1 1069 18·5 0·8
580 240 5 0·6 235 4·0 0·2
Race
Caucasian 4952 566 68·3 4386 75·7 0·9
African American 196 17 2·1 179 3·1 0·7
Asian 109 42 5·1 67 1·2 4·3
Others 553 69 8·3 484 8·4 1
Unknown 815 135 16·3 680 11·7 1·4
Total 6625 830 12·5 5795 87·5
* Only the cases with travel information (n = 6625) were included.
† Sex was unknown in seven cases.
‡ Age was unknown in two cases.
Epidemiology of Campylobacter in Michigan 3319
https://doi.org/10.1017/S095026881600159X Published online by Cambridge University Press
age groups (OR 2·5, 95% CI 2·2–2·8), while bloody
diarrhoea was more common in children aged <5
years (OR 2·5, 95% CI 2·1–2·9). Based on these
prior associations, we also sought to determine
whether there were differences in hospitalization status
among cases in rural vs. urban areas and after stratify-
ing by travel history. Both urban and domestic cases
were more frequently hospitalized compared to rural
cases and those with foreign travel history (Table 3).
Multivariate analyses demonstrated that age 560
years (OR 2·4, 95% CI 2·1–2·7), lack of foreign travel
(OR 2·6, 95% CI 2·1–3·2), and urban residence (OR
1·2, 95% CI 1·1–1·4), were independently associated
with hospitalization.
DISCUSSION
This study represents the first temporal report of cam-
pylobacteriosis incide nce in Michigan and highlights
an increasing incidence trend in the last decade. Thi s
increase may be partly due to enhanced awareness
and changes in detection practices [25], although
other factors are also likely to be important. When
stratifying incidence by age, for example, different
trends were observed. Increased incidence in cases
aged between 20–29 and >50 years suggests that age-
specific risk factors such as diet or other lifestyle fac-
tors, may be important for disease. Indeed, a higher
attributable risk was observed for foreign travel in
individuals aged between 20 and 39 years compared
to other age groups, while higher disease frequencies
occurred in the summer months for those aged be-
tween 10 and 59 years, especially compared to the
<1 and 580 years age groups. This finding suggests
that seasonality could be impacted by summer-related
behaviours including camping, swimming, or outdoor
grilling, but also that these behaviors may be import-
ant risk factors for individuals aged between 10 and 59
years. Additional epidemiological studies are war-
ranted, however, to fully investigate age-specific risk
factors for the design of novel public health
interventions.
Other factors associated with Campylobacter sea-
sonality may include humidity and temperature [26]
as high humidity and temperature levels enhance
pathogen survival and proliferation, potentially in-
creasing the load in animal reservoirs [9, 15]. Insect
vectors like flies can also facilitate transmission of
Campylobacter sp. between animal reservoirs, as well
as to humans, when there is increased ventilation
and airflow during months with warmer weather
[27, 28]. In our study, only domestic cases showed a
marked increase in incidence in the summer months,
supporting the idea that factors important for season-
ality are linked to geographical location or specific
environments.
A significant difference in Campylobacter incidence
was observed by geographical location, particularly
when counties were classi
fied as urban or rural. The
higher incidence observed in rural cases in this study
is most likely due to more frequent contact with ani-
mals and environmental exposures such as well
Fig. 2. Seasonality of Campylobacter cases reported in Michigan by age group, 2004–2013.
3320 W. Cha and others
https://doi.org/10.1017/S095026881600159X Published online by Cambridge University Press