Campylobacter spp. are a leading cause of bacterial gastroenteritis infections worldwide (1) and represent the most common cause of foodborne infections in the U.S. since 2013 (2). While C. jejuni causes a vast majority of human infections, other species including C. coli, C. upsaliensis, C. lari, C. fetus, C. insulaeingrae, and C. hyointestinalis, are also important (3). Collectively, these pathogens were estimated to cause 1.5 million infections in the U.S. each year, contributing to 13,000 hospitalizations and 120 deaths (4). In 2018, the Centers for Disease Control and Prevention (CDC) estimated the incidence of campylobacteriosis to be 19.6 cases per 100,000 individuals, which had increased from 12.0 cases per 100,000 in 2015–2017, among sites participating in the Foodborne Diseases Active Surveillance Network (FoodNet) (2).

Clinical manifestations of campylobacteriosis include fever, abdominal pain, vomiting, weight loss, chills, fatigue, myalgia, malaise, and acute watery or bloody diarrhea (1). The incubation period is typically 1–4 days after exposure, and the severity of symptoms tends to vary by bacterial density and strain (5). Post-infectious immune sequalae such as Guillain-Barré Syndrome, Miller-Fisher syndrome, and reactive arthritis, have been linked to Campylobacter infection as have inflammatory bowel disease, esophageal and colo-rectal cancers, and extra-intestinal infections like bacteremia and meningitis (6). Although most infections are self-limiting, antibiotics are often needed for immunocompromised patients or those with more severe or persistent infections (7).

Water, poultry and livestock are common reservoirs for C. jejuni (8). Transmission to humans typically occurs via consumption of contaminated food products, and direct contact with animal or environmental reservoirs (9). According to a meta-analysis of 72 studies, the key risk factor for campylobacteriosis was international travel, yet consumption of undercooked chicken and direct exposure to Campylobacter from the environment or farm animals were also important (10). Regardless, it is important to note that risk factors often vary by geographic location even across the U.S., with different FoodNet sites reporting considerable variation in the frequency of infections (11). In addition, the FoodNet sites were not selected to be representative of the U.S. population and were shown to have an unequal representation of all racial and ethnic groups and contained fewer individuals living below the poverty level (12, 13).

Campylobacter jejuni has also been designated a serious antibiotic resistant threat resulting in 448,400 resistant infections and 70 deaths each year (14). Resistance to ciprofloxacin, a fluoroquinolone used to treat more severe human infections, increased in the U.S. from 13% in 1997 to 25.3% in 2015 (14, 15). Campylobacter jejuni resistance to multiple drug classes has also increased over time (16) and resistant isolates have been linked to more severe infections requiring lengthier hospitalizations (17). Because NARMS does not utilize data from each state and the Midwest region only receives a subset of Campylobacter isolates from the Minnesota FoodNet site for testing (18), these resistance frequencies and trends may not be representative of those in other locations. Additionally, many clinical laboratories have shifted to the use of culture-independent tests to detect Campylobacter infections, which can obscure actual rates of resistance circulating within patient populations and prevent the identification of risk factors for resistant infections. Indeed, a 2019 FoodNet report noted that 42% of Campylobacter infections were detected using a culture-independent test (2). This shift is concerning and highlights the need for more culture-based studies to better define the epidemiology of and resistance phenotypes in this common foodborne pathogen.

Herein, we sought to describe the susceptibility profiles for 214 C. jejuni isolates cultured from patients with campylobacteriosis during surveillance activities in Michigan (2011–2014) and to identify risk factors for both susceptible and resistant infections. We also sought to make comparisons to national data available through NARMS since Campylobacter resistance is not monitored in Michigan via NARMS (19). Studies such as these highlight the importance of using culture-based diagnostic tests to more accurately monitor resistance phenotypes and frequencies in distinct geographic locations to identify potential exposures and risk factors that may be state and/or region specific.

Through this study we have detected important trends in the prevalence of campylobacteriosis and antibiotic resistant C. jejuni isolated from Michigan patients between 2011 and 2014, further highlighting the importance of pathogen surveillance efforts using culture-based methods. Since campylobacteriosis was not classified as a notifiable infection until 2015 (25), data about disease frequencies and resistance profiles have been limited, particularly for states like Michigan that are not participating in FoodNet or NARMS. In addition, the widespread adoption of culture-independent tests has hampered the ability to routinely monitor important phenotypes such as antibiotic susceptibility profiles. Indeed, it was estimated that 42% of campylobacteriosis cases identified via FoodNet in 2019 were diagnosed by culture-independent tests and among these, culture for Campylobacter was only attempted for 63% of the positive samples (2).

In the four Michigan hospitals examined herein, we observed a significant increase in C. jejuni infections over time, which is similar to national trends (2) and could partly be due to improved sampling and detection capacity. Seasonal differences were also observed with a greater proportion (73.8%) of Michigan cases occurring during the summer and fall. Seasonal variation has been reported previously with several studies showing a peak incidence of C. jejuni infections during warmer months; climate, temperature, increased shedding from animal reservoirs, and/or seasonal-specific behaviors have all been suggested to contribute to seasonality (26–28).

Extracting epidemiological data from case records has also facilitated the identification of factors that increase risk of campylobacteriosis. Similar to our prior study of 7,182 campylobacteriosis cases reported in Michigan between 2004 and 2013 (29) and those from the FoodNet sites (25, 30), most infections affected children <10 (30.8%) or adults between 19 and 65 (53.1%) years of age. Despite this bimodal distribution, the likelihood of hospitalization increased with increasing age. Cases between 41 and 65 years were significantly more likely to be hospitalized than those between 19 and 40 years of age as were cases over 65. The link between older age and more severe disease has been reported for the FoodNet sites and in our prior population-based study (29, 31).

Although males and rural residents represented a greater proportion of the 7,182 campylobacteriosis cases in Michigan (29), similar distributions were not observed among the cases at the four hospitals. For instance, no difference was observed by sex and significantly more cases (62.0%) were from urban areas, suggesting that the four hospitals may not be entirely representative of the Michigan population of campylobacteriosis cases. Such differences are likely due to the structure of the surveillance system since we utilized four of the largest health care systems. Despite having wide catchment areas, each hospital is in a metropolitan location that can result in differences in access to health care, particularly for rural residents, supporting the suggestion that geography as well as patient-specific and cultural factors can impact care seeking behaviors (11). Indeed, we observed a lower likelihood of hospitalization among rural residents in this and our prior study (29), although this association was not significant after controlling for race, sex, and age.

Because the likelihood of hospitalization was significantly greater for cases self-identifying as non-White, urban areas should be an important focus for reducing disparities in infections caused by C. jejuni and other enteric pathogens. Certainly, neighborhood and geographic barriers have previously been suggested to be important for the acquisition of foodborne disease (13). Although race, ethnicity, and other socially constructed categorizations such as socio-economic status, are not typically collected for foodborne disease surveillance systems, prior studies have document increased frequencies of gastroenteritis in minority and low-socioeconomic populations globally (32–34). Additional studies are needed, however, to identify specific risk factors, exposures and causal factors within urban environments that may explain these relationships. Use of previously reported proxies and markers of poverty such as urban residence, and social constructs like self-reported race as we have used, have complex interactions with social determinants of health (35, 36). We therefore cannot describe causal factors for hospitalization of C. jejuni without addressing these shortcomings. We also cannot rule out the possibility that different strain populations with distinct pathogenic traits are circulating in the different areas and are partly responsible for the differences observed.

Since most hospital laboratories in Michigan have switched to the use of culture-independent tests to detect C. jejuni, viable isolates are not typically recovered for characterizing important phenotypic or genotypic traits. Hence, our assessment of resistance frequencies and trends in the four hospitals over this 4-year period yielded notable results. Resistance was detected in 63.1% of the 214 isolates and to all nine antibiotics comprising 11 distinct resistance profiles. The overall predominance of tetracycline (56.1%) and ciprofloxacin (22.9%) resistance was similar in our prior study of 94 isolates recovered in 2011 and 2012 (22). The inclusion of 120 additional isolates recovered from the same hospitals in 2013–2014, however, allowed for the detection of several important changes over time, including an increase in the frequency of fluoroquinolone resistance. This gradual increase is concerning given that fluoroquinolones are commonly used to treat human infections and the Food and Drug Administration (FDA) banned use of these drugs in poultry in 2005 (37). Point mutations in chromosomal genes such as gyrA, which is critical for DNA replication and transcription, have been linked to fluoroquinolone resistance (38, 39). Given that these mutations do not halt transcription, there is no impact on bacterial survival and hence, these resistant bacterial populations can persist in the absence of antibiotic selection (40, 41). This increasing frequency of ciprofloxacin resistance in C. jejuni is consistent with national trends for older strain sets recovered via culture-based detection methods (25). Because of the increased use of culture-independent methods to detect Campylobacter, however, actual rates of ciprofloxacin resistance among clinical isolates in different parts of the U.S. are not well-established. Additionally, despite the critical role that FoodNet and NARMS have played in the detection of resistant foodborne pathogens, neither system is entirely representative of the U.S. population (12). To represent the entire Midwest (Region 5) that includes Michigan, for instance, NARMS only receives a subset of isolates from Minnesota for testing (18). It is therefore important to note that the frequencies and trends reported by NARMS may not accurately reflect those observed in other locations with distinct geographic features and population traits.

Furthermore, fluoroquinolone resistant C. jejuni infections have also been reported to increase the duration of illness (42, 43). Mutations in gyrA have been tied to changes in DNA supercoiling, which can lead to enhanced colonization of the chicken gut and an increase in virulence properties such as motility, biofilm formation and invasion of intestinal epithelial cells in vitro (44–47). These studies establish a mechanism by which fluoroquinolone resistant mutations enhance virulence and support prior associations between resistance and a lengthier illness duration. Additional support comes from our finding that patients with ciprofloxacin resistant infections were twice as likely to be hospitalized than patients with tetracycline resistant infections, suggesting that the former may be more severe. It is not clear, however, if a unique patient population, differential treatment regimens, or distinct bacterial factors account for the difference in the hospitalization rates observed.

Significant differences in the frequency of tetracycline resistance, which was highest in this population of Michigan patients despite the gradual decrease in tetracycline resistance over time, were also observed. These data indicate that unique regional factors may impact resistance rates, yet those factors that contribute to variation across locations are not clear. The tetracyclines have been used to treat zoonotic and rickettsial diseases in human medicine (48) and have been used extensively in livestock and poultry production worldwide. In the U.S., the FDA reported that tetracyclines were the predominant drug class used in food-producing animals at the time of this study (2009–2014), representing an average of 42% of all antibiotics used (49). Continuous use of tetracycline has selected for resistant strains and resistance genes that can persist in reservoir hosts and the environment. For example, TetO has been shown to mediate resistance to tetracycline in C. jejuni by offering ribosomal protection by binding to an unoccupied site (39). This protein is encoded by tet(O), which is commonly carried on the pTet plasmid but has also been detected in the chromosome (50, 51). Given the high transmissibility rates of these resistance plasmids within bacterial populations even in the absence of tetracycline use (52, 53), it is clear that C. jejuni serves as an important reservoir for these and other resistance genes. In our prior study, we demonstrated that tetracycline resistance was more common in strains belonging to multilocus sequence type (ST)-982, a lineage that was also common in Michigan cattle (22, 54) and has been linked to livestock in other locations (55, 56). Together, these data show the importance of clonal expansion of resistant lineages and highlight the role that mobile genetic elements play in dispersion and maintenance of tetracycline resistance.

Although we observed a significant association between livestock contact and tetracycline resistance, the number of cases (n = 13) reporting this exposure was low. It is noteworthy, however, that only one of the cases reporting contact with livestock had a ciprofloxacin resistant infection compared to 11 (84.6%) with tetracycline resistant infections. While increasing frequencies of ciprofloxacin resistant C. jejuni have been recovered from feedlot cattle throughout the U.S. (57), our data suggest that different factors are important for the acquisition of ciprofloxacin vs. tetracycline resistant infections. Consistent with prior studies (25, 30, 42, 58), we have demonstrated that international travel in the month prior to infection is the strongest predictor of ciprofloxacin resistance in this sample of Michigan patients. Infection during the summer or fall months was also independently associated with ciprofloxacin resistance, but we did not observe an association with poultry consumption as was described in other studies (42, 58, 59). This difference could be due to the low number of cases reporting no poultry consumption a week before symptom onset or the high frequency of missing data since many patients failed to answer the food history questions, a common problem with long-term epidemiological studies (60). In general, however, the identification of risk factors that have also been described in other studies is encouraging and indicates that these factors are likely important regardless of the geographic location.

Collectively, the data presented herein demonstrate the importance of monitoring antibiotic resistance phenotypes and frequencies using culture-based methods in multiple geographic locations. The significant difference that we observed in NARMS Region 5 relative to other regions after including our Michigan data with those from Minnesota, illustrates the need for more comprehensive testing and highlights the variation across different geographic locations. Future studies are still needed, however, to link resistance profiles and patient data to epidemiological data to identify those exposures and risk factors that are unique to specific states or regions.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

The studies involving human participants were reviewed and approved by Institutional Review Boards at Michigan State University (10-735SM), the Michigan Dept. of Health and Human Services (842-PHALAB) and each of the four participating hospitals. Written informed consent from the participants' legal guardian/next of kin was not required to participate in this study in accordance with the national legislation and the institutional requirements.

SDM, WC, and JAR designed the study. JTR, DWN, HS, PL, and WK organized sample collection at each site. REM and WC isolated pathogens and extracted epidemiological data. JAR and WC performed the experiments. JAR, SM, and SDM managed the data and conducted analyses. JAR developed the first manuscript draft. All authors contributed and approved the manuscript content.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.