Prevalence of extended-spectrum β-lactamase-producing Enterobacteriaceae in humans living in municipalities with high and low broiler density

Authors

  • P. M. C. Huijbers,

    1. Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    2. Quantitative Veterinary Epidemiology Group, Wageningen Institute of Animal Sciences, Wageningen University, Wageningen, the Netherlands
    Search for more papers by this author
  • M. de Kraker,

    1. Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    Search for more papers by this author
  • E. A. M. Graat,

    Corresponding author
    1. Quantitative Veterinary Epidemiology Group, Wageningen Institute of Animal Sciences, Wageningen University, Wageningen, the Netherlands
    • Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    Search for more papers by this author
  • A. H. A. M. van Hoek,

    1. Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    Search for more papers by this author
  • M. G. van Santen,

    1. Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    Search for more papers by this author
  • M. C. M. de Jong,

    1. Quantitative Veterinary Epidemiology Group, Wageningen Institute of Animal Sciences, Wageningen University, Wageningen, the Netherlands
    Search for more papers by this author
  • E. van Duijkeren,

    1. Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    Search for more papers by this author
  • S. C. de Greeff

    1. Centre for Infectious Disease Control, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
    Search for more papers by this author

Corresponding author: E. A. M. Graat, Quantitative Veterinary Epidemiology Group, Wageningen Institute of Animal Sciences, Wageningen University, P.O. Box 338, 6700 AH Wageningen, the Netherlands

E-mail: Lisette.Graat@wur.nl

Abstract

Prevalence of, and risk factors for, carriage of extended-spectrum β-lactamase (ESBL) -producing Enterobacteriaceae were determined for 1025 Dutch adults in municipalities with either high or low broiler densities. Overall prevalence of ESBL carriage was 5.1%. The hypothesis that individuals in areas with high broiler densities are at greater risk for ESBL carriage was rejected, as the risk was lower (OR = 0.45; p 0.009) for these individuals. Owning a horse increased the risk (OR = 4.69; p ≤0.0001), but horse owners often owned multiple species of companion animals. Routes of transmission from animals to humans in the community, and the role of poultry in this process, remain to be elucidated.

Extended-spectrum β-lactamase (ESBL) -mediated resistance presents a threat for treatment of bacterial infections [1]. ESBL-producing Enterobacteriaceae have been detected in patients [2, 3], individuals from the community [4, 5], meat [3, 6], livestock [7, 8] and companion animals [9]. Transmission between humans and animals [10, 11] might occur through the food chain [2, 3], contact with livestock [12], or the environment [8]. An ESBL prevalence of 4.9% has been found in Dutch hospitalized patients in the province with the highest number of broiler chickens [3], however little information is available about the prevalence of ESBL carriage in the Dutch community. ESBL-producing bacteria are present on all Dutch broiler farms, and 33% of farmers were ESBL carriers [12]. Individuals in areas with high broiler densities might therefore have an increased risk for ESBL carriage. The aim of this cross-sectional study was to determine prevalence of, and identify risk factors for, carriage of ESBL-producing Enterobacteriaceae in individuals living in municipalities with either high or low broiler densities.

A random sample of adults (≥18 years), stratified according to age and gender (Table 1), was taken from eight Dutch municipalities across four provinces. Per province, the municipality with the highest number of commercial broilers per km2 and a municipality with a similar number of inhabitants, but with no commercial broilers, were selected. It was calculated that a sample size of 1800 individuals in both areas was required [13] based on an ESBL prevalence of 4.9% [3] in areas with high broiler densities, to show a difference of 3% (one-tailed, α-error = 0.05, β-error = 0.20, assumed response = 25%). In total, 3949 individuals were asked to return a rectal swab and a questionnaire on demographics, contact with animals, lifestyle, medical history, eating habits and travel. For each respondent, distance to the nearest broiler farm was obtained using geographic data. Exclusion criteria were living or working on a commercial broiler farm, and moving outside the study area. The study was approved by the Medical Ethics Committee of University Medical Centre Utrecht, the Netherlands (protocol number 11-277).

Table 1. Distribution of respondents over age and gender in areas with high and low broiler density
Age classHigh broiler densityLow broiler density
No. of men (% of total)No. of women (% of total)TotalNo. of men (% of total)No. of women (% of total)Total
18–40 years36 (7)72 (13)108 (20)32 (7)68 (14)100 (20)
41–65 years122 (23)187 (35)309 (58)114 (23)155 (31)269 (55)
≥66 years65 (12)51 (10)116 (22)66 (13)57 (12)123 (25)
Total223 (42)310 (58)533 (100)212 (43)280 (57)492 (100)

Rectal swabs were analysed within 3 days of collection, and not frozen before processing. Isolation of Enterobacteriaceae occurred by selective pre-enrichment with(out) 1 mg/L cefotaxime (Sigma, St Louis, MO, USA), followed by screening on MacConkey agar no. 3 (Oxoid, Basingstoke, UK) supplemented with 1 mg/L cefotaxime. All oxidase-negative bacteria (BBL Dryslide Oxidase; Becton Dickinson, Franklin Lakes, NJ, USA) were tested phenotypically for ESBL production by a combination disc-diffusion test using cefotaxime and ceftazidime, with and without clavulanic acid, according to CLSI guidelines [14]. A cefoxitin disc was used to detect combined ESBL/AmpC phenotypes. Individuals were classified as ESBL-positive when at least one isolate was identified phenotypically as an ESBL-producing gram-negative bacterium (BBL Crystal E/NF test; Becton Dickinson). Prevalences and their exact 95% CI were calculated based on the binomial distribution. Univariable and multivariable logistic regression analyses were performed for the variables in Table 2 to assess the relation between ESBL carriage and possible risk factors, according to the method of Hosmer and Lemeshow [15].

Table 2. Frequency (n and%), prevalence (%), and overall p values in univariable logistic regression of possible risk factors (= 1025). Overall prevalence was 5.1%
VariableCategoryFrequencyPrev. (%)Overall p value (-2LL)a
n (%)
  1. a

    Variables with p value in bold (p <0.25) were included in the multivariable modelling procedure.

  2. b

    Linearity of the logits was assessed before inclusion as continuous explanatory variables.

  3. c

    Exact logistic regression.

  4. d

    Living/working on livestock farms include farms with laying hens, veal calves, dairy cattle, breeding pigs, finishing pigs, sheep/goats; people living/working on broiler farms were excluded.

  5. e

    Visiting livestock farms include broiler farms, in addition to the farms mentioned in footnote d.

Research question
Broiler densityLow49248.06.7 0.02
High53352.03.6
Distance to nearest broiler farm (per km; continuous)bESBL-negatives4.1 ± 3.0 km 0.04
ESBL-positives5.0 ± 3.2 km
General characteristics
ProvinceNoord-Brabant28027.33.6 0.05
Gelderland21921.43.7
Overijssel26826.24.5
Friesland25825.28.5
SexMale43542.45.30.79
Female59057.64.9
Age (per year; continuous variable)bESBL-negatives53.6 ± 15.6 years 0.58
ESBL-positives52.4 ± 14.1 years
Country of birthcAbroad272.700.27
the Netherlands99197.35.3
Contact with animals
Owning/contact with companion animalNo49648.54.0 0.14
Yes52751.56.1
CatNo86084.14.5 0.08
Yes16315.98.0
DogNo72170.54.4 0.16
Yes30229.56.6
RodentNo87985.94.90.50
Yes14414.16.3
BirdNo94091.95.10.91
Yes838.14.8
Hobby chickenNo92890.74.7 0.15
Yes959.38.4
Hobby sheep/goatNo99196.94.8 0.10
Yes323.112.5
HorseNo95993.74.4 0.001
Yes646.315.6
No. of companion animal species049648.54.00.31
130029.45.3
213112.85.3
3525.17.7
≥4434.211.6
Companion animal received antibioticsNo94593.54.6 0.13
Yes666.59.1
Living on a livestock farmdNo98096.34.8 0.16
Yes383.710.5
Working on a livestock farmdNo94593.24.90.74
Yes696.85.8
Visiting a livestock farmeNo71970.55.30.67
Yes30129.54.7
Lifestyle
No. of individuals in residence≤263061.54.3 0.15
>239538.56.3
Working in health careNo85383.45.5 0.14
Yes17016.62.9
Having a vegetable gardenNo85183.45.20.81
Yes16916.64.7
Swimming in river, lake, or pondNo94092.15.00.65
Yes817.96.2
Medical history (during last 6 months)
Visiting general practitionerNo45644.96.6 0.06
Yes56055.13.9
Cases of urinary tract infection No95493.65.00.70
Yes656.46.2
Hospital admissionNo94192.15.00.65
Yes817.96.2
Visiting polyclinicNo66665.55.60.37
Yes35134.54.3
Eating habits
Eating meatcNo131.300.39
Yes101198.75.1
Eating chicken meatNo565.53.60.58
Yes96794.55.2
Eating meat purchased from farmNo97495.04.8 0.16
Yes515.09.8
Eating vegetables from gardenNo81179.15.6 0.16
Yes21420.93.3
Travel
Travelling abroadInside Europe95593.44.8 0.18
Outside Europe686.78.8

In total, 1033 (26.2%) individuals agreed to participate, of which 1025 individuals were eligible. Overall, prevalence of ESBL carriage was higher in areas with low broiler densities (6.7%; 33/492; exact 95% CI 4.7–9.3%) than in areas with high broiler densities (3.6%; 19/533; exact 95% CI 2.2–5.5%), and this difference was also present when combining test probabilities across provinces (p <0.01; χ2 = 20.6 with 8 df). The hypothesis that individuals in municipalities with high broiler densities would be at greater risk for ESBL carriage was therefore rejected. Mean distance to the nearest broiler farm was smaller for individuals in municipalities with high broiler densities (2.2 km; SD 1.4, range 0.2–7.3), compared with low broiler densities (6.2 km; SD 2.9, range 0.9–11.8; p ≤0.0001). It could be argued whether this is a distance of biological relevance. Moreover, the overall high broiler density in the Netherlands (1301 broilers/km2), and mobility of individuals between municipalities and provinces, may have disturbed a possible relationship between broiler density and ESBL carriage, leading to the rejection of our hypothesis.

Sixteen variables could be included in the multivariable analysis (Table 2). Distance to closest broiler farm was also associated with probability of a person being ESBL-positive (p 0.04), but was not included in multivariable analysis as this variable was co-linear with broiler density. Risk factors for ESBL carriage in the community suggested in the literature, such as consumption of poultry meat, travelling abroad, recent hospitalization and recent antibiotic use [2, 16, 17] were not confirmed in this study. The observed prevalence of ESBL carriage and the similarity in answers to the questionnaire, which resulted in categories with <10% of available data, make it difficult to draw statistically valid conclusions on these risk factors. Two variables remained in the final model showing no lack of fit (p 0.96): broiler density and owning/contact with a horse. Confounding was not present, and interaction was not significant (p 0.67). High broiler density decreased the risk for ESBL carriage (3.6% vs 6.7%; OR = 0.45; p 0.009). Owning/contact with a horse increased the risk (15.6% vs 4.4%; OR = 4.69; p ≤0.0001). This might not be solely attributable to owning or having contact with a horse, as prevalence increased from 4% in individuals without companion animals to almost 12% in individuals who owned more than four different species of companion animals (Table 2).

After response analysis with respect to age, sex, province and broiler density (Table 1), it was assumed that a representative sample of Dutch adults was obtained. The observed overall prevalence of 5.1% (52/1025; exact 95% CI 3.8–6.6%) therefore indicates the prevalence in the community, and is comparable with other countries [4, 5, 18, 19]. Contact with multiple species of companion animals might play a role in transmission, but further research on the mutual exchange of ESBL-producing Enterobacteriaceae between companion animals and their owners is needed. Other routes of transmission from animals to humans in the community, and the role of poultry in this process remain to be elucidated.

Acknowledgements

We thank our technicians P. Hengeveld and W.M. van Overbeek for their excellent work in the laboratory, B. van Benthem and A.W. van de Giessen for their valuable advice on the study design, and B. Bom for generating the geographic data. Preliminary results of this study were presented at the 3rd ASM conference on Antimicrobial Resistance in Zoonotic Bacteria and Foodborne Pathogens held in Aix-en-Provence from 26 to 29 July 2012.

Transparency Declaration

The authors declare no conflicts of interest.

Ancillary