Skip directly to site content Skip directly to page options Skip directly to A-Z link Skip directly to A-Z link Skip directly to A-Z link
Volume 27, Number 9—September 2021
Research

Human and Porcine Transmission of Clostridioides difficile Ribotype 078, Europe

Geraldine Moloney1Comments to Author , David W. Eyre1, Micheál Mac Aogáin, Máire C. McElroy, Alison Vaughan, Tim E.A. Peto, Derrick W. Crook, and Thomas R. Rogers
Author affiliations: Trinity College Dublin, Dublin, Ireland (G. Moloney, M. Mac Aogáin, T.R. Rogers); University of Oxford, Oxford, UK (D.W. Eyre, A. Vaughan, T.E.A. Peto, D.W. Crook); Central Veterinary Research Laboratory, Celbridge, Ireland (M.C. McElroy); St. James’s Hospital, Dublin (T.R. Rogers)

Cite This Article

Abstract

Genomic analysis of a diverse collection of Clostridioides difficile ribotype 078 isolates from Ireland and 9 countries in Europe provided evidence for complex regional and international patterns of dissemination that are not restricted to humans. These isolates are associated with C. difficile colonization and clinical illness in humans and pigs.

Clostridioides (formerly Clostridium) difficile was considered to be a predominantly nosocomial pathogen until findings of several whole-genome sequencing studies suggested a more complex epidemiology. For example, Eyre et al. reported that only 35% of nosocomial C. difficile infections (CDIs) were potentially attributable to other cases on the basis of genomic data, and only 19% were additionally linked through sharing possible hospital-based contact (1). This finding suggests that a major proportion of C. difficile from CDI cases occurring in healthcare institutions originates from other sources, including the community (2).

Community-associated CDI (CA-CDI) is now well recognized, accounting for ≈25% of cases in Australia, <25% of cases in Europe, and 33% of cases in the United States (3,4). There is increasing recognition that C. difficile is a near ubiquitous environmental organism and that humans have widespread environmental exposure to it. C. difficile has been detected in samples from parks (24.6%); water sources, including rivers, lakes, and sea water; homes (17.1%); commercial stores; and other premises (6.5%–8.1%), in addition to hospitals (16.5%) (5,6). Isolates of C. difficile from these studies underwent ribotype analysis. Overall, ribotype 027 isolates were most commonly identified in hospital samples, and ribotype 014–020 isolates predominated in other environmental samples. Isolates of the most common ribotypes were not restricted to any particular location (5). These findings support the possibility that there are different sources for exposure to each C. difficile ribotype.

Occurrence of CDI caused by C. difficile ribotype 027 has been greatly reduced in the United Kingdom, most likely the result of the combination of antimicrobial stewardship and hospital infection prevention and control measures. However, these interventions have not reduced the incidence of infections caused by other ribotypes, including ribotype 078 (7).

Findings of genomic analysis of isolates from the European, Multi-Center, Prospective, Biannual, Point-Prevalence Study of Clostridium difficile Infection in Hospitalized Patients with Diarrhea (EUCLID) showed that specific C. difficile ribotypes were associated with healthcare clusters, and other ribotypes had an international distribution across Europe (8). For example, ribotype 078 isolates did not cluster by their country of origin, indicating a complex distribution unrelated to nosocomial transmission. The mechanisms of transmission have not been identified, but might be related to the movement of food, other animal-derived products, or persons across Europe (8).

C. difficile carriage and infection has been well described in livestock and other animals (3); certain ribotypes of C. difficile are considered to be major ribotypes from a One Health perspective. These ribotypes include ribotype 078, carriage of which has been reported in 9%–100% of piglets from North America, Europe, Asia, and Australia (3). Carriage rates in calves (56%) and cows (13%) have been lower. Although many studies did not identify any major carriage in adult pigs, 1 study in the Netherlands reported a rate ranging from 6.6% to 100% (3).

We have reported C. difficile ribotype 078 in cases of typhlocolitis in neonatal piglets in Ireland (9), and Knetsch et al. found that ribotype 078 isolates carried by farmers in the Netherlands and their pigs were identical by whole-genome sequence analysis (10). These findings suggest that C. difficile isolates might be shared between humans and pigs when in close proximity. However, the mechanisms and directions of transmission are not known.

In this study, we investigated the genomic relationships between C. difficile ribotype 078 isolates of human and porcine origin collected from Ireland and compared these with international ribotype 078 isolates. We also investigated the extent to which geographic proximity could explain clusters of clonal isolates.

Methods

Samples and Settings

Clinical isolates of C. difficile ribotype 078 were collected prospectively as part of an investigation of consecutive episodes of CDI conducted at St. James’s Hospital (Dublin, Ireland), a 900-bed tertiary referral center, during 2013–2016. Stool samples, sent from patients with diarrhea, had the C. difficile toxin B gene identified by using the EntericBio PCR Kit (Serosep, https://www.serosep.com). We reviewed medical notes of inpatients to obtain relevant clinical data, including antimicrobial drugs and proton pump inhibitors prescribed before the onset of diarrhea, features indicative of severe CDI with or without complications, and the antimicrobial drugs used for management of CDI. These data were pseudonymized and stored in a dedicated database.

We retrieved an additional 9 C. difficile 078 isolates from a study of recurrent CDI at St. James’s Hospital during 2012–2013 (11). Five additional C. difficile ribotype 078 isolates were provided from those submitted to a national surveillance study of CA-CDI in Ireland conducted during 2015. Isolates of C. difficile were recovered from pigs that had been referred for autopsy at the Central Veterinary Research Laboratory (CVRL; Backweston, Ireland) during 2014–2015, irrespective of the suspected cause of death, by sampling colonic contents or feces that had positive results for C. difficile toxins A/B by using the Premier Elisa Kit (Meridian BioScience Inc., https://www.meridianbioscience.com). We treated human fecal and porcine colonic/fecal samples with ethanol shock before anaerobic incubation on cycloserine cefoxitin egg yolk medium. DNA was extracted from resulting colonies for PCR ribotype analysis and Illumina (https://www.illumina.com) genomic library preparation as described (11).

Whole-Genome Sequencing

Whole-genome sequencing was performed either on an Illumina MiSeq or MiniSeq platform at Trinity College (Dublin, Ireland) or on the Illumina HiSeq platform at the Wellcome Centre for Human Genetics, University of Oxford (Oxford, UK). Sequence data generated have been deposited in the National Center for Biotechnology Information Short Read Archive (https://www.ncbi.nlm.nih.gov/sra) under BioProject PRJNA692997.

We mapped sequence reads to the ribotype 078 reference genome M120 (GenBank accession no. FN665653.1), and identified high-quality variants by using an approach developed and calibrated for C. difficile (1) with later refinements (12) (Appendix). We obtained published comparison sequences from the EUCLID pan-European cross-sectional survey conducted during in 2012–2013 (8) and from farm animal and human isolates from the Netherlands (2002–2011) described by Knetsch et al. (10).

Sequence Comparisons

We compared sequences by using single-nucleotide polymorphisms (SNPs) and obtained differences between sequences from maximum-likelihood phylogenies corrected for recombination (Appendix). We reviewed phylogenetic analysis of closely related genomes in conjunction with available epidemiologic data. Within the clinical database, CDI recurrence was defined as identification of 2 isolates within 10 SNPs from 1 patient (1) for which that patient had clearly documented clinical resolution of symptoms after their first episode. On the basis of rates of C. difficile evolution and within-host diversity (1), we defined plausible, short-term, transmission/mutual exposure as isolates differing by 0–2 SNPs.

We made epidemiologic matches between patients who had in-patient admissions and demonstrable links with respect to time, location, or healthcare staff, where their C. difficile isolates were within 0–2 SNPs. Because epidemiologic details were not available for either the CA-CDI investigation in Ireland or the EUCLID isolates, we analyzed linkage between cases on the basis of genetic similarity alone. These genomic pairs were named by the isolate sources in chronologic order of identification.

Ethics

Investigation of hospital-associated CDI (HA-CDI) cases at St James’s Hospital was conducted after obtaining approval from the St. James’s Hospital/Tallaght Research Ethics Committee. Porcine isolates were exempt from requiring ethics approval.

Results

A total of 171 C. difficile ribotype 078 isolates were included in the analysis: 53 isolates from CDI episodes in 44 inpatients at St. James’s Hospital, including 5 community-associated isolates; 20 porcine isolates from Ireland; 67 clinical, farmer, and porcine isolates from the Netherlands; and 31 clinical EUCLID isolates. We provide details of their country of origin, source, and date of isolation (Table 1). The EUCLID isolates were obtained from 9 countries in Europe. Six countries, including Ireland, submitted >2 isolates.

Figure

Recombination-adjusted maximum-likelihood phylogenetic tree of sequences from human and porcine Clostridioides difficile isolates from Ireland and 9 other countries in Europe. Isolates are shown as triangles for healthcare-associated C. difficile cases and circles for community-associated C. difficile cases. Isolates from pigs are shown as crosses and those from farmers as squares. The color at each tip indicates the country of origin of the isolate. The tree was based on 4,861 variable sites before correction for recombination, based on a median (interquartile ranges) of 93.4% (93.0%–93.8%) and (83.1%–96.2%) of the reference genome being called. Scale bar indicates single-nucleotide polymorphisms.

Figure. Recombination-adjusted maximum-likelihood phylogenetic tree of sequences from human and porcine Clostridioides difficile isolates from Ireland and 9 other countries in Europe. Isolates are shown as triangles for healthcare-associated ...

Of the 53 isolates causing CDI in Ireland, 9 were from recurrent CDI episodes in 7 patients (7 subsequent isolates were 0 SNPs different from, the baseline isolate, 1 was 1 SNP different, and 1 was 8 SNPs different). Only the first isolate from each patient was considered in subsequent analyses. We provide genomic relationships between the remaining 162 ribotype 078 isolates (Figure). Despite the diverse sampling frame, only limited diversity was seen; the greatest root-to-tip distance in the phylogenetic tree was 48 SNPs.

Isolates from Ireland were found throughout the tree, but specific clusters of these isolates were seen, including, as shown at the ≈240° (≈8 o’clock) position (Figure), a cluster of cases that included isolates from HA-CDI and CA-CDI cases as well as cases from pigs. Within this cluster, several porcine isolates were directly ancestral to 1 HA-CDI case. Another 5 CDI cases, including 1 CA-CDI, had another porcine isolate directly ancestral. This finding suggests a porcine origin for these cases, either directly or by >1 or more intermediate (unsampled) transmission routes. This same cluster also contained an isolate from a pig and a farmer from the Netherlands. Several other clinical isolates from the Netherlands were closely related to porcine isolates (Figure).

We provide epidemiologic links between genetically related isolates within 0–2 SNPs (Table 2). Although nearly all genomic pairs occurred among isolates with the same country of origin, the epidemiologic information available can explain only a small proportion of transmissions/mutual exposures.

Discussion

Our findings support a complex regional and international distribution of C. difficile ribotype 078 isolates. In contrast to the EUCLID study, which obtained samples on single days in winter and summer, more dense sampling was undertaken in our study. In the EUCLID study, no evidence of clustering of ribotype 078 within countries was seen, which is consistent with a complex pattern of dissemination in Europe over timescales spanning years (Figure). However, our study showed evidence of sublineages of ribotype 078 that are predominantly found in isolates from the Netherlands and others predominantly found in isolates from Ireland (Figure). It is likely that this denser sampling has enabled recent, local, onward transmission to be better captured. We also identify a EUCLID isolate from Italy (2013) and a CA-CDI isolate from Dublin, Ireland (2014), that are within 2 SNPs, which is consistent with a temporally related transmission. However, we do not know of any epidemiologic link between these 2 cases.

For 10 pairs of isolates within 2 SNPs from inpatients who had HA-CDI, possible healthcare-based epidemiologic links could be made for 6 of these pairs but not the other 4. Plausible ward-based transmission only accounted for 3 pairs. For other genetically related isolates pertaining to inpatients in our study, there was a median of 559 days between their associated CDI episodes (range 147–651 days) without overlapping hospital admissions or appointments. Overall, nosocomial transmission accounted for 15% of closely genetically related (<2 SNPs) C. difficile ribotype 078 cases in this study, and equal proportions were attributable to farms and unknown transmission routes. In a study in Leeds, UK, which had comparable phylogenetic analysis, hospital ward-based epidemiologic linkage was reported as 11% for ribotype 078 cases versus 64% for ribotype 027 cases (13).

A EUCLID isolate from Ireland (2013) forms a genomic cluster with 1 CA-CDI isolate (2015) and 2 HA-CDI isolates (July 2015 and December 2015). These 4 isolates were from patients in 3 Dublin healthcare facilities and from 1 case of CA-CDI that had been collected within a 3-year period. This finding suggests shared exposure across the greater Dublin area, and that nosocomial transmission is not the dominant route of acquisition of C. difficile ribotype 078. This observation is consistent with the EUCLID study findings (8).

It is not clearly understood how persons who have CA-CDI acquired their infection because they do not have the risk factors for HA-CDI (14). Anderson et al. described proximity to livestock farms, agricultural industry, and nursing home facilities as risk factors for CA-CDI in North Carolina, USA, but they did not include analysis of C. difficile molecular data in their models (15). In contrast, Van Dorp et al. found no evidence of either localized point sources or livestock exposure as risk factors for C. difficile acquisition in the Netherlands (16). They included ribotype detail in their analysis, but found no evidence of geographic clustering of ribotype 078 CDI cases (16). This finding is consistent with that of Knetsch et al., who reported clonal isolates of farm and clinical origin without a geographic basis for those clusters (10).

Knetsch et al. identified another genomic cluster of C. difficile ribotype 078 isolates, which included an isolate of animal origin from Canada (2004) and 8 isolates of clinical origin from the United Kingdom (2008–2012) (17). We also identified a cluster of clinical and porcine 078 isolates from Ireland, where there was no known occupational exposure of the affected patients who lived in urban locations far from relevant pig farms. Knight et al. reported clonal ribotype 014 isolates from Australia that were considerable geographic distances from each other, which is suggestive of long-range transmission and major community reservoirs (18). They concluded that this transmission was unlikely to have been caused by direct contact between the humans and animals involved, and suggested that by-products, such as manure or compost, could enable indirect transmission from animals and humans (18). In a study from the United States, biosolid-based compost had the highest rate of C. difficile recovery that included ribotype 078 isolates (19), which was also the most common ribotype in an investigation of manure from Japan (20).

Findings based on ribotype analysis alone are insufficient for clear identification of transmission events pertaining to community reservoirs (21). Moradigaravand et al. identified ≈90% of their collection of clinical and wastewater isolates as clade 1 (231/256), and only 10 (3.9%) as clade 5/ribotype 078 (22). When their ribotype 078 isolates were compared with the same isolates from the Netherlands included in our analysis, they found divergence of ≈20 years between the isolates from the United Kingdom and the Netherlands. This finding suggests that water is not the primary reservoir or route for dissemination of C. difficile ribotype 078 isolates. It is still considered possible that dissemination of ribotype 078 isolates occurs by the food chain, the environment, or both (23,24). This view is supported by the presence and distribution of tetracycline-resistant determinants in C. difficile genomes, reflecting the antimicrobial drug selection pressure from tetracycline use in agriculture or veterinary practice, and thereby facilitating emergence and spread of ribotype 078 bacteria (24).

It is not completely understood how some livestock might have asymptomatic C. difficile colonization, whereas others show development of infection (25). The porcine isolates from Ireland in this analysis were from available samples processed at the CVRL. These isolates included samples from neonatal piglets that had typhlocolitis (9). We have identified genomic similarities among isolates causing human and veterinary infections. This finding augments the need for a One Health approach for C. difficile ribotype 078.

The strengths of this analysis include the large number of C. difficile ribotype 078 isolates included, from different sources including humans and animal species, and geographic origin. The limitations of this study include the lack of epidemiologic data available to the investigators for CA-CDI and the limited number of porcine strains from samples available at the CVRL. In conclusion, our analysis of C. difficile ribotypes 078 isolates from Ireland and 9 other countries in Europe showed close overlap between isolates from humans and pigs, including the occurrence of plausible transmission, either directly or by an unknown intermediate source.

This study was supported by the Health Research Board, Ireland.

G.M. received support from the Health Research Board, Ireland, as a Research Training Fellowship for Healthcare Professionals and M.M.A. is the recipient of an Irish Research Council fellowship (EPSPD/2015/32). D.W.E. has received lecture fees from Gilead outside this study and is a Robertson Foundation Fellow and a National Institute for Health Research Oxford Biomedical Research Centre Senior Fellow.

Dr. Moloney is an infectious diseases physician at Cork University Hospital, Cork, Ireland. Her primary research interest is infections with Clostridioides difficile.

Top

Acknowledgment

D.W.E. has received lecture fees from Gilead outside this study and is a Robertson Foundation Fellow and a National Institute for Health Research Oxford Biomedical Research Centre Senior Fellow.

Top

References

  1. Eyre  DW, Cule  ML, Wilson  DJ, Griffiths  D, Vaughan  A, O’Connor  L, et al. Diverse sources of C. difficile infection identified on whole-genome sequencing. N Engl J Med. 2013;369:1195205. DOIPubMedGoogle Scholar
  2. Moloney  G, Mac Aogáin  M, Kelleghan  M, O’Connell  B, Hurley  C, Montague  E, et al. Possible interplay between hospital and community transmission of a novel Clostridium difficile sequence type 295 recognized by next-generation sequencing. Infect Control Hosp Epidemiol. 2016;37:6804. DOIPubMedGoogle Scholar
  3. Knight  DR, Riley  TV. Genomic delineation of zoonotic origins of Clostridium difficile. Front Public Health. 2019;7:164. DOIPubMedGoogle Scholar
  4. European Centre for Disease Prevention and Control. Healthcare-associated infections: Clostridium difficile infections. In: Annual epidemiological report for 2016. Stockholm: The Centre, 2018 [cited 2021 Jun 5]. https://www.ecdc.europa.eu/en/publications-data/healthcare-associated-infections-clostridium-difficile-infections-annual
  5. Alam  MJ, Walk  ST, Endres  BT, Basseres  E, Khaleduzzaman  M, Amadio  J, et al. Community environmental contamination of toxigenic Clostridium difficile. Open Forum Infect Dis. 2017;4:ofx018. DOIPubMedGoogle Scholar
  6. al Saif  N, Brazier  JS. The distribution of Clostridium difficile in the environment of South Wales. J Med Microbiol. 1996;45:1337. DOIPubMedGoogle Scholar
  7. Dingle  KE, Didelot  X, Quan  TP, Eyre  DW, Stoesser  N, Golubchik  T, et al.; Modernising Medical Microbiology Informatics Group. Effects of control interventions on Clostridium difficile infection in England: an observational study. Lancet Infect Dis. 2017;17:41121. DOIPubMedGoogle Scholar
  8. Eyre  DW, Davies  KA, Davis  G, Fawley  WN, Dingle  KE, De Maio  N, et al.; EUCLID Study Group. Two distinct patterns of Clostridium difficile diversity across Europe indicating contrasting routes of spread. Clin Infect Dis. 2018;67:103544. DOIPubMedGoogle Scholar
  9. McElroy  MC, Hill  M, Moloney  G, Mac Aogáin  M, McGettrick  S, O’Doherty  Á, et al. Typhlocolitis associated with Clostridium difficile ribotypes 078 and 110 in neonatal piglets from a commercial Irish pig herd. Ir Vet J. 2016;69:10. DOIPubMedGoogle Scholar
  10. Knetsch  CW, Connor  TR, Mutreja  A, van Dorp  SM, Sanders  IM, Browne  HP, et al. Whole genome sequencing reveals potential spread of Clostridium difficile between humans and farm animals in the Netherlands, 2002 to 2011. Euro Surveill. 2014;19:20954. DOIPubMedGoogle Scholar
  11. Mac Aogáin  M, Moloney  G, Kilkenny  S, Kelleher  M, Kelleghan  M, Boyle  B, et al. Whole-genome sequencing improves discrimination of relapse from reinfection and identifies transmission events among patients with recurrent Clostridium difficile infections. J Hosp Infect. 2015;90:10816. DOIPubMedGoogle Scholar
  12. De Silva  D, Peters  J, Cole  K, Cole  MJ, Cresswell  F, Dean  G, et al. Whole-genome sequencing to determine transmission of Neisseria gonorrhoeae: an observational study. Lancet Infect Dis. 2016;16:1295303. DOIPubMedGoogle Scholar
  13. Martin  JSH, Eyre  DW, Fawley  WN, Griffiths  D, Davies  K, Mawer  DPC, et al. Patient and strain characteristics associated with Clostridium difficile transmission and adverse outcomes. Clin Infect Dis. 2018;67:137987. DOIPubMedGoogle Scholar
  14. Khanna  S, Pardi  DS, Aronson  SL, Kammer  PP, Baddour  LM. Outcomes in community-acquired Clostridium difficile infection. Aliment Pharmacol Ther. 2012;35:6138. DOIPubMedGoogle Scholar
  15. Anderson  DJ, Rojas  LF, Watson  S, Knelson  LP, Pruitt  S, Lewis  SS, et al.; CDC Prevention Epicenters Program. Identification of novel risk factors for community-acquired Clostridium difficile infection using spatial statistics and geographic information system analyses. PLoS One. 2017;12:e0176285. DOIPubMedGoogle Scholar
  16. van Dorp  SM, Hensgens  MPM, Dekkers  OM, Demeulemeester  A, Buiting  A, Bloembergen  P, et al. Spatial clustering and livestock exposure as risk factor for community-acquired Clostridium difficile infection. Clin Microbiol Infect. 2019;25:60712. DOIPubMedGoogle Scholar
  17. Knetsch  CW, Kumar  N, Forster  SC, Connor  TR, Browne  HP, Harmanus  C, et al. Zoonotic transfer of Clostridium difficile harboring antimicrobial resistance between farm animals and humans. J Clin Microbiol. 2018;56:e0138417. DOIPubMedGoogle Scholar
  18. Knight  DR, Squire  MM, Collins  DA, Riley  TV. Genome analysis of Clostridium difficile PCR ribotype 014 lineage in Australian pigs and humans reveals a diverse genetic repertoire and signatures of long-range interspecies transmission. Front Microbiol. 2017;7:2138. DOIPubMedGoogle Scholar
  19. Dharmasena  M, Jiang  X. Isolation of toxigenic Clostridium difficile from animal manure and composts being used as biological soil amendments. Appl Environ Microbiol. 2018;84:e0073818. DOIPubMedGoogle Scholar
  20. Usui  M, Kawakura  M, Yoshizawa  N, San  LL, Nakajima  C, Suzuki  Y, et al. Survival and prevalence of Clostridium difficile in manure compost derived from pigs. Anaerobe. 2017;43:1520. DOIPubMedGoogle Scholar
  21. Knetsch  CW, Lawley  TD, Hensgens  MP, Corver  J, Wilcox  MW, Kuijper  EJ. Current application and future perspectives of molecular typing methods to study Clostridium difficile infections. Euro Surveill. 2013;18:20381. DOIPubMedGoogle Scholar
  22. Moradigaravand  D, Gouliouris  T, Ludden  C, Reuter  S, Jamrozy  D, Blane  B, et al. Genomic survey of Clostridium difficile reservoirs in the East of England implicates environmental contamination of wastewater treatment plants by clinical lineages. Microb Genom. 2018;4:4. DOIPubMedGoogle Scholar
  23. Hensgens  MP, Keessen  EC, Squire  MM, Riley  TV, Koene  MG, de Boer  E, et al.; European Society of Clinical Microbiology and Infectious Diseases Study Group for Clostridium difficile (ESGCD). Clostridium difficile infection in the community: a zoonotic disease? Clin Microbiol Infect. 2012;18:63545. DOIPubMedGoogle Scholar
  24. Dingle  KE, Didelot  X, Quan  TP, Eyre  DW, Stoesser  N, Marwick  CA, et al. A Role for tetracycline selection in recent Evolution of agriculture-associated Clostridium difficile PCR ribotype 078. MBio. 2019;10:e0279018. DOIPubMedGoogle Scholar
  25. Weese  JS. Clostridium (Clostridioides) difficile in animals. J Vet Diagn Invest. 2020;32:21321. DOIPubMedGoogle Scholar

Top

Figure
Tables

Top

Cite This Article

DOI: 10.3201/eid2709.203468

Original Publication Date: August 01, 2021

1These authors contributed equally to this article.

Table of Contents – Volume 27, Number 9—September 2021

Comments

Please use the form below to submit correspondence to the authors or contact them at the following address:

Geraldine Moloney, Department of Infectious Diseases, Cork University Hospital, Wilton, Cork, Ireland

Send To

10000 character(s) remaining.

Top

Page created: June 07, 2021
Page updated: August 17, 2021
Page reviewed: August 17, 2021
The conclusions, findings, and opinions expressed by authors contributing to this journal do not necessarily reflect the official position of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors' affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above.
file_external