Chapter 1. Design and Methods

Antimicrobial Resistance

What's New

• Bacterial culture and antimicrobial susceptibility testing of Salmonella, Escherichia coli and Campylobacter isolates from retail turkey was started in January 2012.
• Bacterial culture and antimicrobial susceptibility testing of Campylobacter isolates from pigs at the abattoir was started in January 2012.
• CIPARS adopted a new Enterobacteriaceae plate CMV2AGNF for Salmonella and E. coli in 2012. The changes to the Enterobacteriacae plate were: addition of azithromycin (Category II) and removal of amikacin (Category II).
• CIPARS adopted the new Clinical and Laboratory Standards Institute (CLSI)Footnote 1 resistance breakpoint of ≥ 1 μg/mL for ciprofloxacin in Salmonella and E. coli isolates in 2012. The decision by CIPARS to expand the breakpoint change to all Salmonella serovars and E. coli was based upon the desire to keep the breakpoints harmonized across the Enterobacteriaceae we monitor, due to their close biological similarities and ease of sharing of resistance genes. Specifically for E. coli, using the same breakpoint reinforces its role as a commensal indicator of the pool of resistance genes available for exchange with more pathogenic organisms. Furthermore, the ciprofloxacin resistance breakpoints were the same in the past for both genera (at ≥ 4 μg/mL); keeping the breakpoints the same maintains the precedent previously set. In this report, all ciprofloxacin data have been recalculated retrospectively using this new breakpoint. The impact of this change was minimal considering that most Salmonella and E. coli isolates tested have minimum inhibitory concentrations (MIC) for ciprofloxacin that are many dilutions below the resistance breakpoint.

Human Surveillance

Objective(s)

The objective of the Surveillance of Human Clinical Isolates component of CIPARS is to provide a representative and methodologically unified approach to monitor temporal variations in the prevalence of antimicrobial resistance in Salmonella isolated from humans.

Surveillance Design

Hospital-based and private clinical laboratories culture human Salmonella isolates in Canada. Although reporting is mandatory through laboratory notification of reportable diseases to the National Notifiable Disease Reporting System, forwarding of Salmonella isolates to provincial reference laboratories is voluntary and passive. A high proportion (84% in 2001)^{Footnote 2} of Salmonella isolates are forwarded to Provincial Public Health Laboratories (PPHLs), but this proportion may vary among laboratories. The Yukon, Northwest Territories, and Nunavut, which do not have a PPHL counterpart, forward their isolates to one of the PPHLs.

Prior to 2002, PPHLs forwarded Salmonella isolates to the Enteric Diseases Program, National Microbiology Laboratory (NML), Public Health Agency of Canada (PHAC), Winnipeg, Manitoba for confirmation and subtype characterization. A letter of agreement by which provinces agreed to forward all or a subset of their Salmonella isolates to NML for CIPARS was signed in 2002 by the PPHLs and PHAC. This agreement officially launched the surveillance program.

To ensure a statistically valid sampling plan, all human Salmonella isolates (outbreak-associated and non-outbreak-associated) received passively by PPHLs in Saskatchewan, Manitoba, New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador were forwarded to the NML. The PPHLs in more heavily populated provinces (British Columbia, Alberta, Ontario, and Québec) forwarded only the isolates received from the 1st to the 15th of each month. However, all human S. Newport and S. Typhi isolates were forwarded to the NML because of concerns of multidrug resistance and clinical importance, respectively.

The PPHLs were also asked to provide a defined set of data for each forwarded isolate, including serovar name, date collected, and patient age, sex, and province of residence.

Retail Meat Surveillance

Objectives(s)

The objectives of CIPARS Retail Meat Surveillance component are to provide data on the prevalence of antimicrobial resistance and to monitor temporal variations in selected bacteria found in raw meat at the provincial/region level.

Surveillance Design

Retail surveillance provides a measure of human exposure to antimicrobial-resistant bacteria via the consumption of undercooked meat. Retail food represents a logical sampling point for surveillance of antimicrobial resistance because it is the endpoint of food animal production. Through meat sample collection and testing, the retail surveillance provides a measure of human exposure to antimicrobial resistant bacteria through the consumption of meat products available for purchase by Canadian consumers. The scope of the surveillance framework can be modified as necessary (e.g. to evaluate different food commodities, bacteria, or geographic regions) and functions as a research platform for investigation of specific questions regarding antimicrobial resistance in the agri-food sector.

The unit of concern in Retail Meat Surveillance in 2012 was the bacterial isolate cultured from one of the commodities of interest. In this situation, the commodities were raw meat products commonly consumed by Canadians, which originated from the 3 animal species sampled in the Abattoir Surveillance component as well as turkey beginning in 2012. These raw meat products consisted of chicken (legs or wings [skin on]), turkey (ground), pork (chops), and beef (ground).

For ground beef, a systematic collection of extra-lean, lean, medium, and regular ground beef was performed to ensure representation of the heterogeneity of ground beef with respect to its origins (e.g. domestic vs. imported beef or raised beef cattle vs. culled dairy cattle. The meat cuts "legs or wings with skin on", "ground turkey", "pork chops", and "ground beef" were chosen on the basis of suspected high prevalences of the targeted bacterial species within and the low purchase prices of these commodities^{Footnote 3} and for comparability to other international retail surveillance programs .

Bacteria of interest in chicken and turkey were Campylobacter, Salmonella, and generic E. coli. In pork both Salmonella and E. coli were cultured, but only isolates of E. coli underwent antimicrobial susceptibility testing for routine surveillance. Salmonella was isolated from pork mainly to provide recovery estimates from this commodity for other Public Health Agency of Canada programs. Because the prevalence of Salmonella in pork is low, antimicrobial susceptibility results are not presented on an annual basis but are pooled and presented over a multi-year period in the interest of precision. Recovery of Campylobacter from pork was not attempted because of the low prevalence observed in the initial stages of Retail Meat Surveillance. In beef, only E. coli was cultured and then tested for antimicrobial susceptibility given the low prevalence of Campylobacter and Salmonella in these commodities at the retail level, as determined during the early phase of the program. In turkey, Campylobacter, Salmonella, and E. coli were isolated from retail samples.

Sampling Methods

Generally, the sampling protocol was designed to evaluate antimicrobial resistance in certain bacterial species that contaminate retail meat and to which Canadian consumers may subsequently be exposed. In 2012, it primarily involved continuous weekly submission of samples of retail meat from randomly selected geographic areas (i.e. census divisions defined by Statistics Canada), weighted by population, in each participating province.

Retail meat samples were collected in British Columbia, Saskatchewan, Ontario, and Québec. In past years retail data have been presented for the Maritimes (a region including the provinces of New Brunswick, Nova Scotia, and Prince Edward Island). In 2012, due to unforeseeable delays with respect to resuming sampling, very few retail samples were collected and thus, data from the Maritimes region are not presented in the 2012 Annual Report. Retail data for this region will be presented again in the 2013 Annual Report.

Data from Statistics Canada were used to define strata. This was done by using cumulative population quartiles (or thirdtiles) from a list of census divisions in a province, sorted by population in ascending order. Generally, between 15 and 18 census divisions per province/region were then chosen by means of stratified random selection and weighted by population within each stratum. The number of sampling days allocated to each stratum was also weighted by population and is summarized as follows:

Ontario and Québec

• Stratum One: 10 divisions selected, with 2 sampling days per division per year
• Stratum Two: 4 divisions selected, with 5 sampling days per division per year
• Stratum Three: 2 divisions selected, with 10 sampling days per division per year
• Stratum Four 1 division selected, with 20 sampling days per year

Saskatchewan

• Stratum One: 9 divisions selected, with 2 sampling days per division per year
• Stratum Two: 5 divisions selected, with 3 sampling days per division per year
• Stratum Three: 2 divisions selected, with 5 sampling days per division per year
• Stratum Four: 1 division selected, with 7 sampling days per year

British Columbia

• Stratum One: 10 divisions selected, with 1 sampling day per division per year
• Stratum Two: 4 divisions selected, with 3 sampling days per division per year
• Stratum Three: 1 division selected, with 20 sampling days per year

Maritimes Provinces

• For the 3 Maritimes provinces, results are aggregated and presented at the Maritimes region level; however, sampling activities for this region were proportional to the population within each province as indicated below. Furthermore, as with the other provinces sampled in the retail component, sampling within each province was proportional to the census division subpopulations and is summarized as follows:
• Nova Scotia
  • Stratum One: 5 divisions selected, with 1 sampling day per division per year (on average)
  • Stratum Two: 4 divisions selected, with 2 sampling days per division per year
  • Stratum Three: 1 division selected, with 10 sampling days per division per year
• New Brunswick
  • Stratum One: 5 divisions selected, with 1 sampling day per division per year (on average)
  • Stratum Two: 4 divisions selected, with 2 sampling days per division per year
  • Stratum Three: 2 divisions selected, with 4 sampling days per division per year (on average)
• Prince Edward Island
  • Stratum One: 1 division selected, with 1 sampling day per division per year
  • Stratum Two: 1 division selected, with 2 sampling days per division per year

Generally, field workers in Ontario and Québec conducted sampling on a weekly basis, and those in British Columbia, Saskatchewan, and Maritimes region (no retail data presented for this region in 2012) conducted sampling every other week. Sampling was less frequent in British Columbia, Saskatchewan, and the Maritimes region (very sparse number of samples for this region in 2012) because of funding constraints, limited laboratory capacity, and a desire to avoid over-sampling at particular stores. Samples were collected on Mondays or Tuesdays for submission to the laboratory by Wednesday. Samples submitted from outside Québec (with the exception of samples from the Maritimes region) were sent to the same laboratory via 24-hour courier. In the rare sampling weeks for the Maritimes region in 2012, samples from the whole Maritimes region were collected on Mondays or Tuesdays and submitted to a laboratory in Prince Edward Island within 24 hours.

In each province, 2 census divisions were sampled each sampling week. In each census division, 4 stores were selected prior to the sampling day, based on store type. Generally, 3 chain stores and 1 independent market or butcher shop were selected. An exception to this protocol was made in densely populated urban census divisions (e.g. Toronto or Montréal), where 2 chain stores and 2 independent markets or butcher shops were sampled to reflect the presumed shopping behaviour of that subpopulation. From each store type, 1 sample of each commodity of interest was attempted, for a desired total of 15 meat samples (4 chicken, 4 turkey, 4 pork, and 3 beef samples) per division per sampling day^{Footnote 4}. When possible, specific stores were sampled only once per sampling year. In some cases due to reduced availability of certain meats and store closures etc., the desired sample yield was not achieved.

Prevalence estimates were used to determine the numbers of samples to be collected, which were based on an expected yield of 100 isolates per commodity per province per year, plus 20% to account for lost or damaged samples. Because sampling was less frequent in British Columbia, Saskatchewan, and the Maritimes region than in Ontario and Québec, the target of 100 isolates per year may not have always been met in those provinces/region.

In 2012, personal digital assistants (PDAs) were used to capture the following store and sample data:

• Type of store
• Number of cash registers (surrogate measure of store volume)
• "Sell-by" or packaging date
• "May contain previously frozen meat" label - yes or no
• Final processing in store - yes, no, or unknown
• Air chilled - yes, no, or unknown (applied to chicken samples only)
• Organic - yes, no, or unknown
• Antimicrobial free - yes, no, or unknown
• Price per kilogram

Individual samples were packaged in sealed zipper-type bags and placed in 16-L thermal coolers for transport. The ambient environmental temperature was used to determine the number of ice packs placed in each cooler (i.e. 1 ice pack for temperatures below 20°C and 2 ice packs for temperatures 20°C or higher). In 1 or 2 coolers per sampling day, instruments for recording temperature data^{Footnote 5} were used to monitor temperatures to which samples were exposed.

Abattoir Surveillance

Objective(s)

The objectives of the CIPARS Abattoir Surveillance component are to provide nationally representative, annual antimicrobial resistance data for bacteria isolated from animals entering the food chain, and to monitor temporal variations in the prevalence of antimicrobial resistance in these bacteria.

Surveillance Design

Abattoir Surveillance only includes animals that originated from premises within Canada. Established in September 2002, this component initially targeted generic Escherichia coli and Salmonella within the food animal commodities associated with the highest per capita meat consumption: beef cattle, broiler chickens, and pigs. In 2003, the component was refined to discontinue Salmonella isolation from beef cattle because of the low prevalence of Salmonella in that population. Campylobacter surveillance was initiated in beef cattle in late 2005 in order to include a pathogen in beef cattle surveillance and to provide data on fluoroquinolone resistance, following the approval of a fluoroquinolone for use in cattle. Campylobacter surveillance was also initiated in chickens in 2010 and pigs in 2012.

In the Abattoir Surveillance component, the unit of concern (i.e. the subject of interest) was the bacterial isolate. The bacteria of interest were isolated from the caecal contents (not carcasses) of slaughtered food animals to avoid misinterpretation related to cross-contamination and to better reflect antimicrobial resistance in bacteria that originated on the farm.

Over 90% of all food-producing animals in Canada are slaughtered in federally inspected abattoirs annually^{Footnote 6}. The program is based on the voluntary participation of federally inspected slaughter plants from across Canada. The sampling method was designed with the goal that, across Canada, 150 isolates of each targeted bacterial species would be recovered from each of the 3 animal species over a 12-month period. The exception was Campylobacter in beef cattle, for which it was estimated that 100 isolates would be recovered over the same period. These numbers represented a balance between acceptable statistical precision and affordability^{Footnote 7}. The actual number of samples collected was determined for each food animal species on the basis of the expected caecal prevalence of the bacteria in that animal species. For example, if the expected bacterial prevalence was 10%, then 1,500 samples would need to be collected and submitted for bacterial isolation.

The sampling design was based on a 2-stage sampling plan, with each commodity handled separately. The first stage consisted of random selection of federally inspected slaughterhouses. The probability of an abattoir being selected was proportional to its annual slaughter volume. The second stage involved systematic selection of animals on the slaughter line. The annual number of caecal samples collected at each abattoir was proportional to its slaughter volume.

Sampling Methods

To minimize shipping costs and allow each abattoir to maintain efficiency, the annual total number of samples to be collected in each abattoir was divided by 5, resulting in the number of collection periods. For each collection period, 5 caecal samples were collected within 5 days, at the convenience of the slaughterhouse staff, provided the 5 animals and associated samples originated from different groups. Sampling from different groups of animals was important to maximize diversity and avoid bias attributable to overrepresentation of particular producers. The largest plants were scheduled to sample up to 7 animals from different groups over the 5 day collection period in order to achieve the required number of samples annually. Collection periods were uniformly distributed throughout the year, leading to an abattoir-specific schedule for collection of caecal contents. The uniform distribution of the collection periods helped to avoid any bias that may have resulted from seasonal variation in bacterial prevalence and antimicrobial susceptibility test results.

Thirty-seven federally inspected slaughter plants (4 beef cattle plants, 22 poultry plants, and 11 swine plants) from across Canada participated in the 2012 CIPARS Abattoir Surveillance component. Samples were obtained according to a predetermined protocol, with modifications to accommodate various production-line configurations in the different plants. Protocols were designed to avoid conflict with carcass inspection methods, plant-specific Food Safety Enhancement Programs, and Health and Safety requirements. They were also designed to avoid situations of potential cross-contamination. All samples were collected by industry personnel under the oversight of the Veterinarian-in-Charge of the Canadian Food Inspection Agency.

Farm Surveillance

Objective(s)

The objectives of the CIPARS Farm Surveillance component are to provide data on antimicrobial use and resistance, to monitor temporal trends in the prevalence of antimicrobial resistance, to investigate associations between antimicrobial use and resistance on grower-finisher pig farms, and to provide data for human health risk assessments.

Surveillance Design

The Farm Surveillance component was the third active surveillance component implemented by CIPARS. Taken together, with the Abattoir and Retail Surveillance components, these data validate the information collected at key points along the farm-to-fork food production chain. This initiative is built on a sentinel farm framework. Questionnaires are used to collect data on farm demographics, animal health and antimicrobial use. Composite pen fecal samples are collected and submitted to laboratories for bacterial isolation and antimicrobial susceptibility testing. The CIPARS Farm Surveillance component is administered and coordinated by the Laboratory for Foodborne Zoonoses.

CIPARS Farm Surveillance component was initiated in 2006 in the 5 major pork-producing provinces in Canada (Alberta, Saskatchewan, Manitoba, Ontario, and Québec). The swine industry was selected as the pilot commodity for development of the farm surveillance infrastructure because the Canadian Quality Assurance (CQA®) program had been extensively implemented by the industry and because, in 2006, unlike in the other major livestock commodities, there had not been a recent outbreak of foreign animal disease in pigs.

The Farm Surveillance component concentrates on grower-finisher hogs. Pigs in this stage of production were chosen because of their proximity to the consumer.

Sampling Methods

Swine veterinarians recruited sentinel herds to participate in this voluntary national surveillance program. The number of sentinel herds allocated to each of the 5 participating provinces was proportional to the national total of grower-finisher units, except in Alberta, where 10 additional sentinel herds were included. Support for the 10 extra herds, and laboratory testing for all samples collected from the CIPARS sentinel herds in Alberta was provided by the Alberta Agriculture and Rural Development Agri-Food laboratory.

To preserve the anonymity of participating producers, herd veterinarians collected the samples and data and submitted coded information to the Public Health Agency of Canada. In the case of corporate herds, 2 noncorporate supervisory veterinarians ensured confidentiality by holding the key to corporate herd codes. This step was taken because knowing a corporate veterinarian's name could have identified the corporation associated with the herd, thereby breaking anonymity.

Veterinarians were purposively selected from the list of veterinarians practicing swine medicine in each province. Each veterinarian selected a predetermined number of sentinel farm sites by use of specific inclusion and exclusion criteria. To be included, herds were required to be CQA® validated, produce more than 2,000 market pigs per year, and be representative of the characteristics (i.e. similar production volumes and types of production systems) and geographic distribution of herds in the veterinarian's swine practice. Herds were excluded when they were regarded as organic with respect to animal husbandry, were fed edible residual material, or were raised on pasture. These criteria helped ensure that the herds enrolled were representative of most grower-finisher swine herds in Canada.

Sentinel grower-finisher herds were visited once per year for sample and data collection. Pooled fecal samples were collected from 6 pens of pigs that were close to market weight (i.e. more than 80 kg [175 lb]).

Surveillance of Animal Clinical Isolates

Objective(s)

The objective of Surveillance of Animal Clinical Isolates is to detect emerging antimicrobial resistance patterns as well as new serovar/resistance pattern combinations in Salmonella.

Surveillance Design

This component of CIPARS relies on samples that are typically collected and submitted to veterinary diagnostic laboratories by veterinarians and/or producers. Consequently, sample collection and submission as well as Salmonella isolation techniques varied among laboratories over the year.

Salmonella isolates were sent by provincial and private animal health laboratories from across the country to the Salmonella Typing Laboratory (STL) at the Laboratory for Foodborne Zoonoses, Guelph, Ontario (LFZ-Guelph) with the exception of Québec, where isolates from animal health laboratories were sent to the Laboratoire d’épidémiosurveillance animale du Québec, du ministère de l’Agriculture, des Pêcheries et de l’Alimentation du Québec for serotyping.

Samples may also have been collected from animal feed, the animal's environment, or non-diseased animals from the same herd or flock. Reported here are results from chicken, turkey, cattle, pigs, and horses. Cattle isolates could have originated from dairy cattle, milk-fed or grain-fed veal, or beef cattle. Chicken isolates were largely from layer hens or broiler chickens, but could also have been from primary layer breeders or broiler breeder birds. A proportion of the turkey isolates might have been recovered from turkey-related environmental samples.

Feed and Feed Ingredients

Sampling Design

Data from the Feed and Feed Ingredients component of CIPARS were obtained from various sources, including monitoring programs of the Canadian Food Inspection Agency (CFIA) and a few isolates from provincial authorities. Information on specimen collection methods was only available for the CFIA monitoring programs.

The CFIA collects samples of animal feed under 2 different programs: Program 15A (Monitoring Inspection - Salmonella) and Program 15E (Directed Inspection - Salmonella). Under Program 15A, feeds produced at feed mills, rendering facilities, ingredient manufacturers, and on-farm facilities are sampled and tested for Salmonella. Although this program makes use of a random sampling process, extra attention is paid to feeds that are more likely to have a higher degree of Salmonella contamination, such as those that contain rendered animal products, oilseed meals, fish meals, grains, and mashes. Program 15E targets feeds or ingredients from establishments that (i) produce rendered animal products, other feeds containing ingredients in which Salmonella could be a concern (e.g. oilseed meal or fishmeal), or a significant volume of poultry feed; (ii) are known to have repeated problems with Salmonella contamination; or (iii) have identified a Salmonella serovar that is highly pathogenic (e.g. Typhimurium, Enteritidis, or Newport). Program 15E is a targeted program; samples are not randomly selected.

Bacterial Isolation Methods

All samples were cultured by use of standard protocols as described below. All primary isolation of human Salmonella isolates was conducted by hospital-based or private clinical laboratories in participating provinces. Most primary isolation of Escherichia coli , Salmonella , and Campylobacter from agri-food samples was conducted at the Laboratory for Foodborne Zoonoses, Saint-Hyacinthe. Primary isolation for Retail Meat Surveillance in Prince Edward Island was conducted at the Atlantic Veterinary College, University of Prince Edward Island. Part of the primary isolation for Farm Surveillance was conducted at the Agri-Food Laboratory of the Alberta Agriculture and Rural Development. Samples from the CIPARS Animal Clinical Isolates component were cultured by various participating laboratories. Most primary bacterial isolation from Feed and Feed Ingredients samples was conducted by the CFIA - Laboratory Services Division (Calgary or Ottawa).

Salmonella

Surveillance of Human Clinical Isolates

Hospital-based and private clinical laboratories isolated and identified Salmonella from human samples according to approved methods^{Footnote 8,Footnote 9,Footnote 10,Footnote 11.}

Surveillance of Agri-food Isolates (Retail Meat Surveillance, Abattoir Surveillance and Farm Surveillance)

The method used to isolate Salmonella was a modification of the MFLP-75 method^{Footnote 12}. This method allowed isolation of viable and motile Salmonella from fecal (Farm Surveillance), caecal (Abattoir Surveillance) content, and meat (Retail Meat Surveillance) from agri-food samples. It is based on the ability of Salmonella to multiply and be motile in modified semi-solid Rappaport Vassiliadis (MSRV) medium at 42°C.

Retail Meat Surveillance: Depending on the sample type either 1 chicken leg^{Footnote 13}, 1 pork chop or 25 g of ground turkey was added to 225 mL of Buffered Peptone Water (BPW). One hundred milliliters of the peptone rinse were kept for Campylobacter and/or E. coli isolation. Chicken and turkey samples were left in the remaining volume of peptone rinse and incubated at 35 ± 1°C for 24 hours. Afterward, a MSRV plate was inoculated with 0.1 mL of the rinse and incubated at 42 ± 1°C for 24 to 72 hours. Suspect colonies were screened for purity and used to inoculate triple-sugar-iron and urea agar slants. Presumptive Salmonella isolates were assessed using the indole test, and their identities were verified by means of slide agglutination with Salmonella Poly A-I and Vi antiserum.

Abattoir Surveillance and Farm Surveillance: A 10-g portion of each pig cecal or fecal sample was mixed with 90 mL of BPW. Chicken caecal contents were weighed and mixed with BPW at a ratio of 1:10. Samples were incubated at 35 ± 1°C for 24 hours. Afterward, the method used was the same as the one described in the Salmonella - Retail Meat Surveillance section.

Surveillance of Animal Clinical Isolates

Salmonella was isolated according to standard procedures, which varied among laboratories. Most methods for detecting Salmonella in animal clinical isolates were similar in principle and involved pre-enrichment, selective enrichment, differential and selective plating, isolation, and biochemical and serological confirmation of the selected isolates.

Feed and Feed Ingredients

Under both Canadian Food Inspection Agency programs (15A and 15E), all samples were collected aseptically and submitted for bacterial culture and isolation. For Salmonella isolation, MSRV medium was used.

Escherichia Coli

Retail Meat Surveillance

Fifty milliliters of the peptone rinse prepared as stated in the Salmonella - Retail Surveillance section were mixed with 50 mL of double strength EC Broth and incubated at 45 ± 1°C for 24 hours. One loopful of the mixture was then streaked onto Eosin Methylene Blue agar and incubated at 35 ± 1°C for 24 hours. Suspect colonies were screened for purity and transferred onto trypticase soy agar with 5% sheep blood. Presumptive E. coli colonies were assessed using Simmons citrate and indole tests. The E. coli isolates with negative indole test results were confirmed using a bacterial identification test kit^{Footnote 14}.

Abattoir and Farm Surveillance

One drop of the peptone mixture prepared as stated in the Surveillance of Agri-Food Isolates/Salmonella - Abattoir and Farm Surveillance section was streaked onto MacConkey agar and incubated at 35°C for 18 to 24 hours. Suspect lactose-fermenting colonies were screened for purity and transferred onto Luria-Bertani agar. Presumptive E. coli colonies were assessed as in the Retail Meat Surveillance for E. coli.

Campylobacter

Retail Meat Surveillance

Fifty milliliters of the peptone rinse prepared as stated in the Salmonella - Retail Surveillance section were mixed with 50 mL of double-strength Bolton broth and incubated in a microaerophilic atmosphere at 42 ± 1°C for 44 to 48 hours. A loopful of broth was then streaked onto a modified Charcoal Cefoperazone Deoxycholate Agar (mCCDA) plate and incubated in a microaerophilic atmosphere at 42 ± 1°C for 24 to 72 hours. Suspect colonies were streaked onto a second mCCDA and on a Mueller Hinton agar plate. Both plates were incubated in a microaerophilic atmosphere at 42 ± 1°C for 24 to 48 hours. Presumptive Campylobacter colonies were identified using the following tests: Gram stain, oxidase, and catalase. A multiplex PCR (mPCR)^{Footnote 15} was used to speciate colonies. Specific genomic targets (hippuricase in C. jejuni and aspartokinase in C. coli) were amplified by mPCR from bacterial lysates. Products were visualized on agarose gel and identified based on their specific molecular size. An internal universal control (16s rRNA) was incorporated into the PCR method. The priming oligonucleotides used in the PCR were highly specific for C. jejuni or C. coli and will not amplify DNA present in any other Campylobacter spp. or non-Campylobacter organisms. Unidentified species of Campylobacter are collectively referred to in the CIPARS reports as "other Campylobacter spp." However, when used alone, the term "Campylobacter" refers to all Campylobacter species.

Abattoir Surveillance

One milliliter of BPW mixture prepared as stated in the Salmonella - Abattoir and Farm section was mixed with 9 mL of Hunt's enrichment broth (HEB) and incubated in a microaerophilic atmosphere at 35 ± 1°C for 4 hours. After this first incubation, 36 μL of sterile cefoperazone were added to the HEB tubes which were then sent back to microaerophilic incubation, this time at 42 ± 1°C for 20 to 24 hours. A loopful of HEB was then used to inoculate a mCCDA plate which was incubated at 42 ± 1°C in microaerophilic conditions for 24-72 hours. Suspect colonies were assessed as in the Campylobacter - Retail Meat Surveillance section.

Serotyping and Phage Typing Methods for Salmonella

Surveillance of Clinical Human Isolates

In general, clinical laboratories forwarded their Salmonella isolates to their Provincial Public Health Laboratory (PPHL) for identification and serotyping. The PPHL further forwarded Salmonella isolates to the National Microbiology Laboratory (NML) according to the predefined testing protocol. Isolate identities were confirmed by the NML when isolates received did not have a serovar name^{Footnote 16} or when inconclusive results arose during phage typing. The O or somatic antigens of the Salmonella isolates were serotyped by use of a slide agglutination method^{Footnote 17}. At the NML, Salmonella H or flagellar antigens were detected via slide and confirmatory tube agglutination methods. Salmonella isolates were maintained at room temperature (25° to 35°C) until typed.

Phage typing was performed at the NML for isolates of the following Salmonella serovars: Enteritidis, Heidelberg, Typhimurium, Hadar, Newport, Typhi, Paratyphi A, Paratyphi B^{Footnote 18}, Paratyphi B var. L(+) tartrate (+), Infantis, Thompson, Oranienburg, Panama, I 4,[5],12:b:-, and I 4,[5],12:i:-. For phage typing the standard technique described by Anderson and Williams^{Footnote 19} was followed. Isolates were streaked onto nutrient agar plates and incubated at 37°C for 18 hours. One smooth colony was selected and used to inoculate 4.5 mL of phage broth^{Footnote 20}, which was then incubated for 1.5 to 2 hours in a shaking water bath at 37°C to attain bacterial growth with a turbidity equivalent to 0.5-McFarland standard. Phage agar plates^{Footnote 21} were flooded with approximately 2 mL of culture medium, and the excess liquid was removed with a Pasteur pipette. Flooded plates were allowed to dry for 15 minutes at room temperature. Afterward, approximately 20 μL of each serovar-specific typing phage was used to inoculate the bacterial lawn by means of a multiple inoculating syringe method^{Footnote 22}. The plates were incubated at 37°C overnight, and lytic patterns were subsequently interpreted^{Footnote 23}.

Salmonella Enteritidis strains were phage typed with typing phages obtained from the International Centre for Enteric Phage Typing (ICEPT), Central Public Health Laboratory, Colindale, UK^{Footnote 24}. The phage-typing protocol and phages for Salmonella Typhimurium, developed by Callow^{Footnote 25} and further extended by Anderson^{Footnote 26} and Anderson and colleagues^{Footnote 27} were obtained from the ICEPT. The Salmonella Heidelberg phage typing protocol and phages were supplied by the NML^{Footnote 28}. Isolates that reacted with the phages but did not conform to any recognized phage type were designated as atypical. Strains that did not react with any of the typing phages were designated as untypable.

The Identification and Serotyping and the Phage Typing units at the NML have attained International Standards Organization (ISO) 17025 accreditation by the Standards Council of Canada. These identification and Serotyping, Phage Typing, and Antimicrobial Resistance units participate in the annual Global Food-borne Infections Network (WHO-GFN), External Quality Assurance System of the World Health Organization, the Enter-net (a European network for the surveillance of human gastrointestinal infections) proficiency program for Salmonella, and a strain exchange with the Laboratory for Foodborne Zoonoses (Salmonella and Escherichia coli). The NML and the Centre for Foodborne, Environmental and Zoonotic Infectious Diseases have been a strategic planning members of the WHO-GFN program since 2002.

Surveillance of Agri-food, Animal Clinical and Feed Isolates

Animal clinical Salmonella isolates from Québec were serotyped at the Laboratoire d’épidémiosurveillance animale du Québec, du ministère de l’Agriculture, des Pêcheries et de l’Alimentation du Québec and were sent to the STL^{Footnote 29} for phage typing.

All Salmonella isolates from other provinces were submitted to the STL for serotyping and phage typing. The serotyping method detects O or somatic antigens of the Salmonella isolates via slide agglutination^{Footnote 30}. The H or flagellar antigens were identified with a microtitre plate well precipitation method^{Footnote 31}. The antigenic formulae of the Salmonella serovars as reported by Grimont and Weill^{Footnote 32} were used to identify and name the serovars.

For phage typing, the standard technique by Anderson and Williams^{Footnote 33} and described above was followed. The sources of the typing phages for Salmonella Enteritidis, Typhimurium and Heidelberg were the same as described above for Surveillance of Human Clinical Isolates.

Since 1995, the STL has participated in annual inter-laboratory exchange of serotyping panels with up to 3 other laboratories. The STL began external proficiency testing of the accuracy of phage typing in 2003. Every year, the STL participates successfully in phage typing proficiency panels from the Central Public Health Laboratory, Colindale, United Kingdom.

Antimicrobial Susceptibility Testing Methods

All Salmonella isolates of human origin were tested for antimicrobial susceptibility at the National Microbiology Laboratory (NML) and all isolates of agri-food or feed origin were tested for antimicrobial susceptibility at the Laboratory for Foodborne Zoonoses, Guelph, Ontario (LFZ-Guelph). The majority of Campylobacter and Escherichia coli isolates from all agri-food components were tested at the Laboratory for Foodborne Zoonoses, Saint-Hyacinthe, Québec (LFZ-Saint-Hyacinthe). In most instances, only 1 isolate per positive sample was submitted for antimicrobial susceptibility testing. In the case of Farm Surveillance, antimicrobial susceptibility testing was performed on 3 E. coli isolates, and 1 Salmonella isolate per sample. All E. coli isolates from Retail Meat Surveillance in Prince Edward Island were processed at the Atlantic Veterinary College, University of Prince Edward Island. Whereas a portion of E. coli isolates from Farm Surveillance in Alberta and Saskatchewan were processed by the Agri-Food Laboratory Branch, Alberta Agriculture and Rural Development.

The NML is a World Health Organization Collaboration Centre for Preparedness and Response to Enteric Pathogens and their Antimicrobial Resistance. The LFZ-Guelph and LFZ-Saint-Hyacinthe, the Alberta Agriculture and Rural Development, and Atlantic Veterinary College participate in external proficiency programs for antimicrobial susceptibility testing for Salmonella and E. coli. The LFZ-Guelph and LFZ-Saint-Hyacinthe participate in inter-agency proficiency programs for identification and antimicrobial susceptibility testing of Salmonella, E. coli, and Campylobacter with the National Antimicrobial Resistance Monitoring System, United States (NARMS). The LFZ-Guelph laboratory is ISO/IEC 17025-accredited for antimicrobial sensitivity testing.

Salmonella and Escherichia Coli

The minimum inhibitory concentration (MIC) values for Salmonella and E. coli were determined by means of the broth microdilution method^{Footnote 34} by use of an automated system^{Footnote 35}. This automated incubation and reading system uses microtitre plates containing various concentrations of dehydrated antimicrobials. The CMV2AGNF plate^{Footnote 36} was designed by the NARMS and contains 15 antimicrobials (Table 1).

Isolates were streaked onto a Mueller Hinton or MacConkey agar plate and incubated at 36 ± 1°C for 18 to 24 hours to obtain isolated colonies. One colony was chosen from the plate and re-streaked onto agar plates for growth. The plates were incubated at 36 ± 1°C for 18 to 24 hours. A 0.5-McFarland suspension was prepared by transferring bacterial growth from the agar plates into 5.0 mL of sterile, demineralized water. Ten microliters of the water-bacteria suspension were transferred to 10 mL of Mueller Hinton broth (MHB). This suspension was dispensed onto CMV2AGNF testing plates at 50 μL per well and the plates were sealed with adhesive plastic sheets. After an 18 hour incubation at 36 ± 1°C the plates were read automatically with fluorometric plate reading system^{Footnote 37}. In accordance with standards set by the Clinical and Laboratory Standards Institute (CLSI)^{Footnote 38}, Staphylococcus aureus ATCC 29213, Escherichia coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, and Enterococcus faecalis ATCC 29212 were used for quality assurance purposes to ensure validity of the MIC values.

Campylobacter

The MIC values for Campylobacter were determined by means of the broth microdilution method^{Footnote 39}. The CAMPY plates^{Footnote 36} designed by NARMS and containing 9 dehydrated antimicrobials were used (Table 2). Colonies were streaked onto Mueller Hinton agar plates with 5% sheep blood and incubated in a microaerophilic atmosphere at 42 ± 1°C for 24 hours. A 0.5-McFarland suspension of bacterial growth was prepared by transferring selected bacterial colonies into a tube containing 5 mL of MHB. Afterward, 10 μL of the MHB were transferred to 11 mL of MHB with laked horse blood. The mixture was dispensed onto CAMPY plates at 100 μL per well. The plates were sealed with perforated adhesive plastic sheets. After a 24 hour incubation in microaerophilic atmosphere at 42 ± 1°C, plates were read using the Sensititre Vizion System^{Footnote 40}. Campylobacter jejuni ATCC 33560 was used as quality control organism. The MIC values obtained were compared with those of CLSI standards^{Footnote 41}.

Antimicrobial Susceptibility Breakpoints

Interpretation of Minimum Inhibitory Concentrations (MICs)

The following information is important for the interpretation of tables presenting results on the distribution of MICs.

• Roman numerals I to IV indicate the ranking of antimicrobials based on importance in human medicine as outlined by the Veterinary Drugs Directorate, Health Canada.
• The unshaded fields indicate the range of concentrations tested for each antimicrobial in the test plate configuration.
• Red numbers indicate the percentage of isolates that were resistant to the antimicrobial according to the predefined resistance breakpoint.
• Numbers to the right of the highest concentration in the tested range (i.e. red numbers in shaded fields) represent the percentage of isolates with growth in all wells of the test plate within the tested range, indicating that the actual MICs were greater than the tested range of concentrations.
• Numbers at the lowest concentration in the tested range (i.e. blue numbers at the far left in unshaded fields) represent the percentage of isolates susceptible to the antimicrobial at the indicated or lower concentrations.
• Solid vertical lines represent resistance breakpoints.
• Dotted vertical lines represent susceptibility breakpoints.
• MIC 50 = MIC at which growth of 50% of isolates was inhibited by a specific antimicrobial.
• MIC 90 = MIC at which growth of 90% of isolates was inhibited by a specific antimicrobial.
• %R = Percentage of isolates that were resistant to a specific antimicrobial.

Table 3. Example on how to interpret minimum inhibitory concentration results

Enlarge Image - Table 3

Data Analysis

Human and Agri-food Surveillance Data Management

Data Management

Laboratory data from human and agri-food surveillance originated in 2 computer programs and were subsequently transferred to a central data repository using intermediary computer softwareFootnote 42. Data were then transferred to a SAS® based harmonized databaseFootnote 43 called the Data Extraction and Analysis (DEXA) application. Additional antimicrobial resistance variables used for analysis are derived within the DEXA application; this application is also used as a central data access point. For the Farm Surveillance component of CIPARS, the bacterial species, serovar, and Minimum Inhibitory Concentration (MIC) data were maintained in a relational databaseFootnote 44.

Data Analysis

Data were analyzed with statistical softwares^{Footnote 45}, and outputs were exported into a spreadsheet application^{Footnote 46}. All tables and figures were generated with the spreadsheet application^{Footnote 46}.

For Farm Surveillance, statistical analyses were performed to account for clustering of antimicrobial resistance within swine herds through generalized estimating equations (GEE)^{Footnote 47}. All statistical models for pig farms included a binary outcome, logit-link function, and exchangeable correlation structure. Exact confidence intervals were computed by use of the BINOMIAL statement^{Footnote 48} and an alpha level of 0.05. When the prevalence was 0%, an alpha level of 0.1 was used instead. Null binomial response models were used to estimate the prevalence of resistance to each antimicrobial. From each null model, the intercept (ß0) and 95% confidence intervals were used to calculate population-averaged (i.e. GEE) prevalence estimates with the formula [1 + exp(-ß₀)]^-1.

Recovery Rate

For Retail Meat Surveillance, Abattoir Surveillance, and the Farm Surveillance components, recovery rate was defined as the number of positive culture results divided by the total number of samples submitted for culture.

Resistant Isolates

The percentage of isolates with resistance to antimicrobials was defined as the number of isolates resistant divided by the total number of isolates tested for each antimicrobial, multiplied by 100.

The breakpoints used for interpretation of antimicrobial susceptibility results are listed in Table 2 and Table 3. Intermediate MIC values were categorized as susceptible for all analyses. A new ceftriaxone breakpoint was officially adopted by the CLSI in January 2010. This breakpoint was applied to all data, including historical data, and was used when performing the analysis for the 2010 Annual Report. A new Enterobacteriaceae plate, CMV2AGNF, was utilized beginning in January 2011. Notable changes to the new plate included the removal of amikacin (Category II) and the inclusion of azithromycin (Category II). Additionally, in 2012, CIPARS decided to adopt a lower breakpoint (= 1 µg/mL) for ciprofloxacin than in past years (= 4 µg/mL) for both Salmonella and E. coli. Ciprofloxacin’s new breakpoint was applied to all data, including historical data, and used for subsequent analysis. Resistance to ciprofloxacin is defined as having an MIC = 1 µg/mL. All non-susceptible isolates (0.12-0.5 µg/mL) are interpreted as susceptible strains.

Resistance Patterns

The total number of antimicrobials in each resistance pattern was calculated by summing the number of antimicrobials to which each isolate was resistant. The most common resistance pattern may include patterns with only 1 antimicrobial. In this case, like for the most common patterns including 2 or more antimicrobials, the number of isolates reported includes only those resistant to this specific pattern (i.e. without any additional resistance to other antimicrobials).

Provincial Incidence Data in Humans

For the provincial human incidence data, the number of Salmonella clinical cases in which a particular serovar was detected per 100,000 inhabitant-years was calculated by dividing the total number of isolates of each serovar received by CIPARS from that province by the provincial population (Statistics Canada post-census population estimates, January 1, 2005) and then multiplying by 100,000^{Footnote 49}. The national estimates for all serovars except S. Typhi and S. Newport were calculated as follows. In more heavily populated (or larger) provinces, the number of isolates resistant and the number of isolates submitted each month were multiplied by 2 as only isolates received in the first 15 days of the month were forwarded to CIPARS for testing. This provided us with an estimated total number of isolates resistant and estimated number of submissions for the larger provinces. Numbers of isolates resistant (estimated value in larger provinces or actual value in smaller provinces) for all provinces were summed to obtain the total estimated number of isolates resistant. Total numbers of isolates submitted (estimated value in larger provinces or actual value in smaller provinces) for all provinces were summed to obtain the total estimated number of submissions. Finally, the total estimated number of isolates resistant was divided by the total estimated number of submissions for each antimicrobial tested to obtain a national estimate of resistance for each antimicrobial and each serovar.

Temporal Analysis

Temporal analyses were performed for selected antimicrobials. Only 1 antimicrobial per antimicrobial class was selected among those antimicrobials commonly used in the agri-food and/or human sectors. Some antimicrobials were excluded from the temporal analyses for the following reasons:

• Resistance to the antimicrobial was absent or at a very low prevalence, or the breakpoint was debatable and other antimicrobials could be used to provide a surrogate measure of resistance or intermediate susceptibility (e.g. nalidixic acid for ciprofloxacin).
• The isolate was cross-resistant to another selected antimicrobial (e.g. amoxicillin-clavulanic acid and ceftiofur).
• The antimicrobial has been banned for use in the agri-food sector, and resistance to this drug is maintained because of the use of another antimicrobial (e.g. chloramphenicol).

Logistic regression models (asymptotic or exact depending on prevalence of the outcome variable) were developed with year as an independent categorical variable. Data were analyzed with commercial software^{Footnote 50}. Analyses of Farm Surveillance data were adjusted for clustering at the herd level.

For all temporal analysis, the current proportion of isolates resistant to a specific antimicrobial has been compared to those observed during the first and the previous surveillance year. In a few specific instances, the first comparison year may vary to reflect the first year of surveillance as new regions were implemented (e.g. 2005 for retail data from Saskatchewan compared to 2003 for Ontario and Québec) or the implementation of new CIPARS components (e.g. 2006 for the Farm Surveillance component in pigs). For ampicillin and ceftiofur, special temporal analyses have been conducted in E. coli and Salmonella isolated from retail chicken or abattoir chickens to compare the current year's data with that of 2004 and 2006. This was due to a change in ceftiofur use practices by Québec chicken hatcheries in early 2005 and in 2007 (start and end of the voluntary period of withdrawal respectively). These special analyses were also conducted in human Salmonella Heidelberg isolates because this human serovar was suspected to originate from chicken. A value of P ≤ 0.05 was considered significant for all temporal analyses.