The Influence of Breastfeeding, Cesarean Section, Pet Animals, and Urbanization on the Development of Inflammatory Bowel Disease: Data from the Swiss IBD Cohort Study

Inflammatory bowel diseases (IBD) are chronic immune-mediated diseases of the gastrointestinal tract, which can be subclassified into Crohn’s disease (CD), ulcerative colitis (UC), and indeterminate colitis (IC).

In population-based studies, the incidence and prevalence of IBD in Europe and the USA have been on the rise since the beginning of the 20th century [1, 2]. Moreover, in Asian regions, for example, in countries like China, India, or Indonesia, both prevalence and incidence of CD and UC also increased continuously over the last 30 years [3].

Current concepts of IBD pathogenesis imply deregulation of host immune responses and changes in intestinal microbiota in genetically susceptible individuals [4, 5]. More than 240 single nucleotide polymorphisms associated with IBD risk have been identified, but even in combination, they only partially explain IBD risk [6-9]. Further, genetic susceptibility does not explain the increase of IBD in the last 100 years, because genetic risk factors have been present in humans since thousands of years.

A number of environmental factors (EFs) have been implicated in the pathogenesis of IBD [10]. There is an association between Westernized lifestyle and the increase in CD and UC in the Western world and in Asian countries [2, 3]. However, which factors of modern lifestyle really contribute to IBD risk remain to be elucidated. A frequently discussed concept, the “hygiene hypothesis,” postulates that improvements in hygiene result in reduced microbial exposure, thus reducing the natural training of the immune system and enhancing the risk for immune-mediated diseases [11-14]. A non-exclusive concept, the “microflora hypothesis,” suggests that changes in the composition of the microbiota, for instance, due to diet or use of antibiotics, disrupt microbial-mediated mechanisms of immunological tolerance [15].

Knowledge about environmental triggers of IBD might enable preventive efforts to reduce the incidence of IBD and enhance treatment of this disease. However, identifying causative EFs that influence the onset of IBD turned out to be challenging because potential risk factors may affect patients for years or decades before the diagnosis of the disease. In such a scenario, an early environmental insult would lead to a subclinical intestinal inflammation, which can be present long before occurrence of the first IBD symptoms [16]. Long-term prospective observational cohorts (e.g., the Swiss IBD Cohort Study, SIBDCS [17]) are well suited to identify or confirm potential environmental triggers.

Breastfeeding has been associated with protection from IBD, but some controversy remains in the literature [18-22]. A recently published review from Ananthakrishnan et al. [23] reported a protective effect of breastfeeding for both CD and UC. As Cesarean section (C-section) strongly affects the infant microbiota, an association between C-section and IBD risk has been postulated [24-26]. On the contrary, a population-based study of Bernstein et al. [27] reported no association between the mode of delivery and IBD. Further, having pets, exposure to farm animals, >2 siblings, and a shared bedroom are inversely associated with the risk of IBD [28].

Interestingly, an increase in occurrence of CD and UC in closely inhabited areas is well described in a recent publication [29]. Urbanization describes a process of a population shift from a rural to an urban way of life and therefore expansions of cities and alterations in lifestyle, which constitute a predisposition to immune-mediated diseases, such as IBD. In this study, we analyzed SIBDCS data to test whether urbanism, breastfeeding, C-section, or pet animals are associated with the risk of developing IBD in Swiss patients.

We used data from the SIBDCS and 2 related cohorts of patients’ friends and mothers to evaluate the impact of various EFs on IBD risk. SIBDCS is a nationwide prospective cohort study established in 2006 and has included patients diagnosed with IBD from all parts of Switzerland.

Since the beginning of data collection, clinical and treatment data have been prospectively captured once a year and entered into a database. SIBDCS is funded by the Swiss National Science Foundation. Purposes and methodology of SIBDCS have been described by Pittet et al. [17].

Questionnaires in Swiss national languages were distributed to adult IBD patients (n = 2,294) addressing mode of delivery, duration of breastfeeding, pets in household, and places of residence since birth. Identical questionnaires were distributed to up to 3 matched childhood friends who grew up in a similar environment (i.e., “friends cohort”), which serves as a control group. Furthermore, questionnaires addressing the patient’s early childhood were filled by the mothers of the patients.

In case of discrepancies between mothers’ and patients’ answers, the input from the mothers was used. Questionnaires were sent out between December 2015 and October 2016. We included all questionnaires returned until January 2018 in the present analysis.

Patients were asked in the questionnaire if they currently possess pet animals or owned pets in the past. As well, we asked the patients to declare in which timeframe they were obtaining animals and tried to pick those out who were in contact with pet animals before the age of 20. Thus, with an interquartile range of age of 20–38 years at the time of diagnosis for CD and 24–41 years for UC/IC, we get an insight into the period before the outbreak of the disease of most patients and try to identify if different sorts of domesticated animals influence the pathogenesis of IBD.

To investigate if duration of breastfeeding has an impact on the onset of IBD, we created 2 subgroups of >6 months and <6 months being breastfed. For our analysis, we defined “urban” as living for ≥10 years in a city with at least 100,000 inhabitants, allowing the 6 biggest cities in Switzerland to be included. If not indicated otherwise, the time period before the age of 20 was considered.

In the final analysis, patients and friends were matched for age and sex. However, other potential confounders like socioeconomic status or place of residence have not been matched in the analysis. Further, for the analysis of a specific parameter, urban or rural living, breastfeeding, C-section or pet animals in household were also not matched for the remaining parameters as this would have decreased the sample size too much.

All statistical analyses were performed using the Stata Software (v. 14.2; Statacorp, College Station, TX, USA). Baseline demographics and clinical data were described. Continuous data distribution was assessed using Normal QQ-plot. Gaussian distributed data were summarized as mean, SD, and range, while non-Gaussian distributed data were summarized as median, interquartile range, and range. Categorical data were summarized as raw frequencies and relative percentages. Differences in means for Gaussian distributed data were assessed using the Student’s t test. Differences in distribution for non-Gaussian distributed data were assessed using the Mann-Whitney-Wilcoxon rank sum test. Differences in distributions for categorical variables were assessed using the χ² test, or the Fisher exact test in case of smaller sample size.

To study the association between EFs and the incidence of IBD, a case-control analysis was done. IBD patients and controls were matched according to age and sex, and a logistic regression model was used. Odds ratios (ORs) and 95% confidence intervals (CIs) were reported. A p value <0.05 was considered as statistically significant.

Out of 2,294 questionnaires sent to SIBDC patients, 1,111 questionnaires were returned (response rate: 48.4%). In addition, for 305 out of 1,111 responding patients, mother-filled questionnaires were available (response rate: 27.5%). Further, from 225 out of 1,111 responding patients, we received questionnaires from at least 1 patients’ friend (response rate: 20.3%). As we accepted multiple friend-questionnaires per patient, 352 returned questionnaires from friends could be included.

In total, we received questionnaires from 466 patients with UC, 28 with IC, and 617 with CD. Epidemiological parameters and data for the clinical history of UC/IC and CD patients are depicted in Table 1 and are consistent with a mixed cohort of IBD patients with mild, moderate, and severe disease.

Breastfeeding was significantly lower for UC/IC patients than for CD patients (UC/IC: 298/494, 72.3% vs. CD: 400/617, 78.1%; p = 0.042; Table 2). However, breastfeeding rates for controls (238/352 individuals, 74.8%) was similar to those for IBD patients. While duration of breastfeeding did not differ between UC and CD patients, more controls reported a shorter duration of breastfeeding (<6 months) than IBD patients (p = 0.014).

In the multivariate analysis, breastfeeding was neither associated with the risk of IBD (OR: 1.145, 95% CI: 0.789–1.660, p = 0.476) nor with the risk of CD or UC/IC. There was also no significant association of long breastfeeding (>6 months) with a diagnosis of IBD, CD, or UC. Shorter breastfeeding (<6 months) was also not associated with IBD or CD but was a protective factor for the development of UC/IC (OR: 0.473, 95% CI: 0.278–0.807, p = 0.006).

A similar fraction of CD patients (47 out of 549 patients, 8.6%) and UC/IC patients (30 out of 436 patients, 6.9%) were born by C-section (p = 0.329). However, the number of control individuals born by C-section (49/335, 14.6%) was almost twice as high compared to IBD patients (p < 0.001). In a multivariate analysis, we observed no association between cesarean delivery and the development of IBD (OR: 0.878, 95% CI: 0.561–1.374, p = 0.569), UC/IC or CD (Table 2).

474 (79.7%) CD patients and 356 (74.8%) UC/IC patients reported any animal in the household. This number was even higher in the control group (IBD: 830/1,071, 77.5%; controls: 294/347, 84.7%; p = 0.004). In a multivariate analysis, the presence of any animal or specific animals like cats or dogs, pet rodents, reptiles, or fish did not increase the risk for IBD. The presence of birds was associated with an increased risk for IBD compared to matched controls in the IBD group (OR: 1.529, CI: 1.066–2.192, p = 0.021). Interestingly, nearly all trends were stronger in the subgroup with UC/IC and reached statistical significance for any animal (OR: 0.602, CI: 0.410–0.884, p = 0.01) and in the subanalysis for pet rodents (OR: 0.671, CI: 0.496–0.907, p = 0.009) and reptiles (OR: 0.442, CI: 0.219–0.891, p = 0.022). The increased risk associated with birds was also stronger in the subgroup with UC (OR: 1.778, CI: 1.247–2.536, p = 0.001). No significant association was observed in the subgroup with CD.

These results remained robust when only animals in the household in infancy, childhood, or youth of the patient (i.e., <20 years) were considered (Table 4). For the IBD group, in the multivariate analysis, birds increased IBD risk compared to matched controls (OR: 1.737, CI: 1.198–2.520, p = 0.004), but no significant associations for other animals were observed. Most trends were also stronger in the subgroup with UC/IC (e.g., for any animal: OR: 0.687, CI: 0.495–0.954, p = 0.025; cats and dogs: OR: 0.688, CI: 0.510–0.930 p = 0.015; rodents: OR: 0.598, CI: 0.437–0.817, p = 0.001; and birds: OR: 1.671, CI: 1.151–2.427, p = 0.007), but no significant association in the subgroup with CD was observed.

A similar number of CD and UC/IC patients lived for at least 10 years in a city with >100,000 inhabitants (CD: 63, 11.4%; UC: 38, 8.6%, n.s.). In contrast, a significantly lower number of individuals from the control group reported to have lived in urban places (IBD: 101, 11.2%, control: 19, 5.4%, p = 0.007). This risk for IBD for urban individuals remained significant in a multivariate analysis (OR: 1.815, CI: 1.011–3.259, p = 0.046). A similar trend was observed for the subgroups of CD and UC/IC patients, which failed to reach statistical significance (Table 3).

Urban environment in infancy, childhood, and youth of the patient (before age 20) was not significantly associated with the risk of IBD, CD, or UC (Table 4). While all ORs for urbanism in young life also pointed toward an increased risk, effect sizes were much smaller than those in the previous analysis, considering the whole life (Table 3) and none of the associations were significant.

Using data from 1,111 IBD patients, we aimed to identify EFs, which may have an impact on the risk to develop IBD. Our data confirm long-term residence in an urban environment as a risk factor for IBD. Breastfeeding for <6 months was associated with a lower risk of UC. The presence of various household animals especially within early life decreased the risk of IBD, especially in the UC subgroup, while opposite effects of household birds were observed.

The association between breastfeeding and IBD risk remains controversial [18-22, 30, 31]. Two meta-analyses support the hypothesis that breastfeeding is a protective factor for IBD [31, 32]. However, other studies observed that breastfeeding has no association with IBD or is even a risk factor for the development of the disease [33, 34]. Human milk contains, among others, lactoferrin, IgA, and lactadherin, which play an essential role in the defense against early enteric infections [35]. Maternal milk also contains human milk oligosaccharides, a group of various sugars with prebiotic effects including Bifidobacteria growth, which would affect the intestinal microbiota and may influence IBD risk [36]. In our study, we did not find a general effect of breastfeeding, but our data indicate an association of breastfeeding for <6 months with UC/IC. However, in our analysis, there was no information regarding breast milk substitution in any subgroup available, and we were unable to correct for many relevant confounders. Furthermore, the number of patients who have been breastfed for >6 months is quite low, and our study would be underpowered to detect effects in this subgroup. Therefore, additional studies are warranted.

The association between C-section and IBD is also under debate. A meta-analysis by Li et al. [26] implies an increased risk for CD but not for UC after C-section. On the other hand, a large population-based study of Bernstein et al. [27] reported no association of the mode of delivery with IBD. In our investigation, Cesarean delivery was also not associated with the risk for IBD. This is remarkable in light of evidence indicating an impact of C-section on the intestinal microbiota for up to 7 years of age [23]. However, investigations showed that differences in the intestinal flora of children delivered by C-section gradually diminish after the first year of life [25, 37].

The “hygiene hypothesis,” first published by Strachan [38], implied that the number of children in the household is inversely associated with the risk of hay fever and eczema. This theory has been extended to other immune-mediated diseases such as IBD [14]. In line with the hygiene hypothesis, various risk factors related to microbial exposure such as exposure to household pets, farm animals, >2 siblings, sharing of bedrooms, or living in rural areas have been associated with protection from developing IBD [28, 39]. A higher exposure to pathobionts and microorganisms may lead to a better training of the immune system and therefore a lower predisposition to develop diseases with inappropriate immune activation [40].

Pets in the household would be one example for an EF related to increased microbial exposure and a protective effect of pets regarding IBD has been demonstrated [28, 41, 42]. Our investigation also indicates a similar inverse association of household pets and the risk for IBD, especially of the UC/IC subtype. Interestingly, possession of birds in the household has opposite effects, increasing the chances of an IBD and UC/IC diagnosis. Further, for CD, neither pet animals in general nor any subspecies were associated with the disease. When only exposure to animals in young life, before the age of 20 was considered, we found a highly similar picture: pet rodents and cats or dogs in household were inversely associated with the risk of UC/IC but not CD, with birds increasing UC/IC risk. The reasons for differential effects of birds and other animals and different susceptibility regarding the environment toward UC/IC and CD are unclear. Literature on this subject is rare, and we suggest further investigation with a bigger sample size and extended adjustment for confounders. In contrast to our findings, Castiglione et al. [43] reported no association of animals in household with CD and UC in a large case-control study. Another investigation from Han et al. [44] demonstrated even an increased risk of CD when owing pets at home during childhood period.

A positive correlation between urban environment and the risk for IBD is well established and was corroborated by a meta-analysis [39]. In an urban environment, potential risk factors for IBD, for example, smoking, intake of antibiotics, or lack of helminths exposure, are more prevalent [45-47]. Furthermore, in urban areas, infrastructure and sanitation would be better, resulting in reduced exposure to enteric pathogens or pathobionts [48, 49]. Our data support these findings, and we report a statistically significant positive association between living in urban areas and IBD. However, this association was not robust in the subanalysis regarding urban environment in young life. A possible explanation is the small sample size in this subgroup.

Our study has strengths as well as limitations. A strength is the large patient number in our cohort with 1,111 returned questionnaires from IBD patients. Furthermore, inclusion of 344 confirmatory questionnaires from patients’ mothers most likely reduced the “recall bias,” which is invariably present in any retrospective study.

One limitation of our study is the low overall return of questionnaires (response rate for patients: 48.4%), potentially resulting in a “reporting bias.” Response rates for the control group and for mothers were even lower (20.3 and 27.5%, respectively). Therefore, the size of the control group (352 persons) was rather small compared to the larger patient group. Low response rates for the control group might be explained by the difficulty for patients to find 3 childhood friends who grew up in a similar environment. Better strategies to increase sizes and coverage of friends and mothers cohorts are warranted for future studies. Probably further minimization of the effort to fill out questionnaires is needed. In addition to the printed questionnaires, the distribution of electronic questionnaires may increase the response rate and could be considered in further investigations. Obtaining a comparison between our control group and the baseline data of general Swiss population has been difficult, considering that with a median age of 35, quality data especially regarding C-section and breastfeeding are needed from the 1980s. Therefore, we were not able to receive a clear evidence whether our data from the control group were affected by bias. Another limitation regards controlling for confounders. In our analyses, cases and controls were matched for sex and age, but we were unable to adjust for other potential confounders such as socioeconomic status, place of residence, ethnicity, or education. Furthermore, we did not match for mode of delivery, breastfeeding, and urbanism in our study. This might be relevant since, for instance, rates of C-sections or household animals would be different in urban compared to rural areas.

Karim Abdelrahman, Gentiana Ademi, Patrick Aepli, Amman Thomas, Claudia Anderegg, Anca-Teodora Antonino, Eva Archanioti, Eviano Arrigoni, Diana Bakker de Jong, Bruno Balsiger, Polat Bastürk, Peter Bauerfeind, Andrea Becocci, Dominique Belli, José M. Bengoa, Luc Biedermann, Janek Binek, Mirjam Blattmann, Stephan Boehm, Tujana Boldanova, Jan Borovicka, Christian P. Braegger, Stephan Brand, Lukas Brügger, Simon Brunner, Patrick Bühr, Bernard Burnand, Sabine Burk, Emanuel Burri, Sophie Buyse, Dahlia-Thao Cao, Ove Carstens, Dominique H. Criblez, Sophie Cunningham, Fabrizia D’Angelo, Philippe de Saussure, Lukas Degen, Joakim Delarive, Christopher Doerig, Barbara Dora, Susan Drerup, Mara Egger, Ali El-Wafa, Matthias Engelmann, Jessica Ezri, Christian Felley, Markus Fliegner, Nicolas Fournier, Montserrat Fraga, Yannick Franc, Pascal Frei, Remus Frei, Michael Fried, Florian Froehlich, Raoul Ivano Furlano, Luca Garzoni, Martin Geyer, Laurent Girard, Marc Girardin, Delphine Golay, Ignaz Good, Ulrike Graf Bigler, Beat Gysi, Johannes Haarer, Marcel Halama, Janine Haldemann, Pius Heer, Benjamin Heimgartner, Beat Helbling, Peter Hengstler, Denise Herzog, Cyrill Hess, Roxane Hessler, Klaas Heyland, Thomas Hinterleitner, Claudia Hirschi, Petr Hruz, Pascal Juillerat, Carolina Khalid-de Bakker, Stephan Kayser, Céline Keller, Christina Knellwolf-Grieger, Christoph Knoblauch, Henrik Köhler, Rebekka Koller, Claudia Krieger-Grübel, Patrizia Künzler, Rachel Kusche, Frank Serge Lehmann, Andrew Macpherson, Michel H. Maillard, Michael Manz, Astrid Marot, Rémy Meier, Christa Meyenberger, Pamela Meyer, Pierre Michetti, Benjamin Misselwitz, Patrick Mosler, Christian Mottet, Christoph Müller, Beat Müllhaupt, Leilla Musso, Michaela Neagu, Cristina Nichita, Jan Niess, Andreas Nydegger, Nicole Obialo, Diana Ollo, Cassandra Oropesa, Ulrich Peter, Daniel Peternac, Laetitia Marie Petit, Valérie Pittet, Daniel Pohl, Marc Porzner, Claudia Preissler, Nadia Raschle, Ronald Rentsch, Alexandre Restellini, Sophie Restellini, Jean-Pierre Richterich, Frederic Ris, Branislav Risti, Marc Alain Ritz, Gerhard Rogler, Nina Röhrich, Jean-Benoît Rossel, Vanessa Rueger, Monica Rusticeanu, Markus Sagmeister, Gaby Saner, Bernhard Sauter, Mikael Sawatzki, Michael Scharl, Martin Schelling, Susanne Schibli, Hugo Schlauri, Dominique Schluckebier, Daniela Schmid, Sybille Schmid-Uebelhart, Jean-François Schnegg, Alain Schoepfer, Vivianne Seematter, Frank Seibold, Mariam Seirafi, Gian-Marco Semadeni, Arne Senning, Christiane Sokollik, Joachim Sommer, Johannes Spalinger, Holger Spangenberger, Philippe Stadler, Peter Staub, Dominic Staudenmann, Volker Stenz, Michael Steuerwald, Alex Straumann, Bruno Strebel, Andreas Stulz, Michael Sulz, Aurora Tatu, Michela Tempia-Caliera, Joël Thorens, Kaspar Truninger, Radu Tutuian, Patrick Urfer, Stephan Vavricka, Francesco Viani, Jürg Vögtlin, Roland Von Känel, Dominique Vouillamoz, Rachel Vulliamy, Paul Wiesel, Reiner Wiest, Stefanie Wöhrle, Samuel Zamora, Silvan Zander, Tina Wylie, Jonas Zeitz, and Dorothee Zimmermann.

The authors would like to thank all the patients and patients’ relatives for answering the questionnaires.

The SIBDCS protocol has been approved by the local ethics committees (EK13-16). All patients provided written informed consent for participation in SIBDCS and data collection. Patients’ mothers and friends signed individual informed consent forms prior to filling the respective questionnaires.

B.M. has consulted to Gilead and Novigenix; received speaking and/or traveling fees from MSD, Vifor, Takeda, and Novartis; and an unrestricted research grant from MSD. G.R. has consulted to Abbvie, Augurix, BMS, Boehringer, Calypso, Celgene, FALK, Ferring, Fisher, Genentech, Gilead, Janssen, MSD, Novartis, Pfizer, Phadia, Roche, UCB, Takeda, Tillots, Vifor, Vital Solutions, and Zeller; G.R. has received speaker’s honoraria from Astra Zeneca, Abbvie, FALK, Janssen, MSD, Pfizer, Phadia, Takeda, Tillots, UCB, Vifor, and Zeller; G.R. has received educational grants and research grants from Abbvie, Ardeypharm, Augurix, Calypso, FALK, Flamentera, MSD, Novartis, Pfizer, Roche, Takeda, Tillots, UCB, and Zeller. L.B. has consulted to Abbvie, Janssen, MSD, Pfizer, Takeda, and Vifor; L.B. has received speaker’s honoraria from Astra Zeneca, Abbvie, FALK, MSD, Takeda, and Vifor; L.B. has received educational grants and research grants from Abbvie, MSD, and Takeda. A.R.S. has no disclosures in association with the manuscript. All other authors have nothing to disclose.

This work was supported by grants from the Swiss National Science Foundation (SNSF) to G.R. (the Swiss IBD Cohort [Grant No. 3347CO-108792])

S.A.L. and A.R.S.: organization, data interpretation, and manuscript preparation; N.F.: statistical data analysis and critical review of the manuscript; A.R.S., G.R., L.B., and V.P.: study design and critical review of the manuscript; and P.S., B.M., and M.S.: manuscript preparation and critical review of the manuscript.