4.3.2 Designing surveillance components
There are some essential surveillance design elements that must be considered when designing surveillance of AMR in animal systems:
Target (reference) population
This is the population to which the results will be extrapolated; for example, to all commercial broiler farms in a country.
To identify the target population, countries should prioritise the populations that will best serve the purpose, considering the available resources. Countries should consider which sources are likely to contribute most to a potential risk to animal and human health. It would be helpful to consider the main food-producing species and the consumption level, foodborne infections and AMC.
For example, in the protocol for active AMR surveillance in commercial broilers and layers chicken population (McKenzie et al., 2019), poultry species were selected due to the high global consumption level, foodborne infections and reported AMC.
Sources of samples
As illustrated throughout this module (and specifically outlined in Section 3.1), samples should be taken at different points of the food system, depending on the information needed, based on the surveillance objective and purpose.
For example, if the aim to establish baseline data about the prevalence of resistant foodborne pathogens, sampling the food chain might be a priority. If the interest is in an indicator bacterium, such as E. coli, data from different points in the system, including farms, food chain, hospitals and environment, might be advisable to subsequently compare and integrate the results across sectors.
You can find illustrations of examples to reflect on the sources of samples in Section 2.2.
Sampling frame and sample calculation
(This is more extensively developed and illustrated in the module Sampling (animal health).)
The
Each country can choose to develop or identify an appropriate sampling frame for its identified target population. This can be a list of farms, slaughterhouses or other sources suitable for sampling. For example, in a poultry farm, the epidemiological unit could be the flock, as an enclosed group of the same health status; in a fish farm it could be the pond, but there should be consideration if ponds are connected and could act as a sole unit, as bacteria could be transmitted through the water.
- The sample should avoid
bias , which is a systematic error due to the design, implementation or analysis of the surveillance system. Bias is reduced as representativeness is increased. - The sample should be
representative of the animal population, process, product or other unit of interest whilst taking into account the expected prevalence of the bacteria in the sample type, the expected prevalence of the resistance phenotype and the desired level of precision and confidence. - The sample size calculation should be based on independent samples. However, if there is any
clustering at the establishment or animal level, the sample size (and analysis) should be adjusted accordingly. For example, if samples are taken from the same farm or market vendors, this would result in a bias, and thus it should be adjusted in analysis and interpretation. - At low levels of expected prevalence, exact methods of sample size calculation should be preferred to approximate methods.
- Samples from which bacteria were not isolated cannot be used in the calculation of prevalence of the resistance phenotype within that bacterial population.
The FAO recommends that ‘80% of the total target population should be included in the sampling frame from which the actual samples will be drawn’ (FAO, 2019), which is something that can be challenging to achieve in LMICs. The recommendations are that:
- findings and conclusions should be limited to the nature of samples and the sampling strategies taken
- statistically valid surveys achieving appropriate representativeness and fulfilling the desired objectives of the AMR surveillance as set out in the plan should continue to be pursued progressively by each and every country.
Hazard selection refers to bacterial species to be monitored, and it could also refer to specific ARGs and antimicrobial residues. (See the module Antimicrobial resistance in animals for more information on resistant bacteria, ARGs and antimicrobial residues.)
A
- Escherichia coli
- Salmonella spp.
- Enterococcus faecium and E. faecalis
- Campylobacter spp.
These bacteria represent a combination of commensals and foodborne zoonotic pathogens, and are carried by all animals. (See the module Antimicrobial resistance in animals for more information.)
In addition, in the design of a surveillance component and subsequent documentation, there are other elements to consider that will be covered in depth in other modules. These include the following:
- Bacterial storage, where isolates should be preserved at least until reporting is completed. Preferably, appropriate isolates should be permanently stored. Bacterial strain collections, established by storage of all isolates from certain years, will provide the possibility of conducting retrospective studies.
- AST – this is discussed in more detail in the module Antimicrobial susceptibility testing.
- Recording, storage and interpretation.
If you are interested in learning more about how to design surveillance activities and practical examples, you might like to read the following sources at some later date:
- The full criteria to be considered in OIE codes Harmonisation of National Antimicrobial Resistance Surveillance and Monitoring Programmes (OIE, 2019) in the terrestrial animal health code [Tip: hold Ctrl and click a link to open it in a new tab. (Hide tip)] and the aquatic health code.
- Applied guidance in an FAO report on monitoring and surveillance of AMR (FAO, 2019).
- Examples of One Health surveillance to understand how to apply the main points explained in this section (McKenzie et al., 2019).
- ‘A 12-point checklist for surveillance of diseases of aquatic organisms: a novel approach to assist multidisciplinary teams in developing countries’ (Bondad-Reantaso et al., 2021).
4.3 Designing a surveillance system