National Microbiological Baseline Study in Broiler Chicken
December 2012 – December 2013

Study design and sampling methods

A farm-to-retail design was developed to provide baseline prevalence estimates at the farm, processing and retail levels. No farm visits were conducted or environmental samples collected from broiler chicken barns. The farm component was assessed through the collection and testing of caecal contents of chicken carcasses at slaughter to reflect the contamination status of the flock. The selection of federally-registered poultry slaughter establishments was based on the weight class and slaughter volume of chickens from active establishments between January and December 2011. In this study, establishments processing broiler chickens having a live weight greater than 1.4 kg and less than 2.7kgkg with an annual slaughter volume greater than 100,000 birds were included in the sampling frame. This study population constitutes 92.9% of total chicken production and excludes all small Rock Cornish game hens and roasters. Any establishments producing less than 100,000 birds annually of the selected weight class were excluded from the sampling frame. From a total of 45 federally-registered chicken slaughtering establishments listed in 2011, 37 met these criteria, accounted for 93.9% of total chicken slaughter, and were distributed across nine provinces.

The sample size was determined to estimate the prevalence of Campylobacter and Salmonella in the target population and products with a precision of ± 5% of the true value with 95% confidence and based on the probability that 50% of broiler chicken and chicken products would be contaminated. For farm prevalence, the sample size was corrected for the finite population of broiler chicken farms present in each province. The sample size of the target population and products was then adjusted to allow comparison of prevalence data by season. A season is defined as a quarter of the year such as winter corresponds to the months of December, January and February and so on. Within each season, a sampling date from Monday to Thursday (excluding all statutory holidays and two weeks during Christmas) was randomly assigned to each sample. For logistical reasons, the number of samples in a given sampling day was adjusted to allow a maximum of three caeca samples, one whole carcass, one chicken part and one weep for a total of six samples per abattoir.


To estimate the prevalence of Campylobacter and Salmonella in the entire population of Canadian flocks and farms, broiler chicken lots slaughtered in federally-registered establishments were sampled using a multistage sampling method. In each establishment, a sample of the primary units or broiler lots listed in a given sampling day or kill day was randomly selected and then a fixed number of secondary units or viscera packs from that lot were selected. As such, a lot or a truck load of broiler chickens of 1.4 to 2.7 kg live weight was first selected over the entire production day (includes night shifts in multiple-shift plants) from which a set of 20 caeca from individual birds was pooled and tested at the laboratory to confirm the microbial status of the lot. The number of lots sampled in each establishment was calculated based on the total number of farms within the province and the establishment's slaughter volume.


Whole carcasses and carcass parts produced in federally-registered establishments were randomly selected at different times within a production day to evaluate the prevalence of pathogens and indicator organisms on raw chicken products prior to distribution to the retail market. The selected types of raw chicken meat products sampled at the abattoir were similar to those collected at retail. Whole carcasses were selected at post-chill, while skinless and boneless (SLBL) breasts and skin-on and bone-in (SOBI) thighs packaged in traypacks were preferably collected immediately after packing, or if not available, directly from bulk-pack containers or at the last readily accessible point prior to packing. The number of whole carcass and carcass part samples per establishment were allocated proportionally to their slaughter volume.

In addition, weep fluid from a limited number of bulk packs containing multiple carcasses was sampled to primarily estimate the concentration of selected pathogens in these fluids that might be potential sources of cross-contamination in HRIs. The number of allocated samples in each establishment was proportional to their slaughter volume.


The retail component was designed to reflect the retail market share with respect to store type and types of chicken products purchased by the Canadian population. It was previously estimated that the main supermarket chains and their affiliates possess 78% of the market share of chicken products in Canada (National Farmers Union, 2005). Retail samples were thus collected from supermarket chains and independent grocers at a 4:1 ratio in the urban area of 33 census metropolitan areas (CMA) across Canada. These CMAs or major cities are formed by one or more adjacent municipalities centred on a large urban area. The total urban area of these CMAs allows the coverage of 62% of the Canadian population based on the 2006 census, while restricting the sampling territory to a manageable level. The number of retail samples allocated in each CMA was based on the population size of each urban area. The random selection of a supermarket chain store or an independent grocery store (including butcher shops) was performed by provincial inspectors (one CMA was sampled by a CFIA inspector) from each CMA using existing provincial or regional databases. Three popular types of raw chicken meat products purchased by Canadian consumers were selected based on a food consumption study (Nesbitt et al., 2008) and were collected at the following frequencies: 50% SLBL breasts, 25% SOBI thighs, and 25% whole chicken carcasses. To limit the number of store visits to a manageable level, the maximum number of samples collected within the same store during a given sampling day ranged from one to three samples depending on the sample size of each CMA.

Date modified: