Good day everyone. In this lecture, we are going to discuss sampling at farms and slaughterhouses. Sampling is one of the essential parts of a well-defined metagenomics study and should be carefully considered. So, today we will discuss an epidemiologic study design shortly, and thereafter focus on data collection on farms and slaughterhouses. A metagenomics epidemiologic study design consists of several design phases. First, you have the theoretical design, in which you design your occurrence relation. So, you define your outcome, your determinants, your study domain and then you have to design your data collection. You can choose between an experimental or an observational study, you can choose between census or sampling, and you have to define at which time you will sample. Due to the costs of metagenomics sampling, most metagenomics observational studies will probably be cross-sectional, in which the outcome of your study and the study determinants are measured at one specific point in time. And then you have the design of your data analysis, so the statistical techniques you want to use. Remember to define all parts of your study design beforehand. So today, we will focus on the design of the data collection only. If you want to sample farms and slaughterhouses you can ask yourself multiple questions. For instance, where, what type of farms and slaughterhouses am I going to select, or what type of data am I going to collect and from whom? From animals or from farmers? Slaughterhouse employees. And how? Do I include a questionnaire or do I only perform sampling? Do I sample myself or do I hire personnel to do the sampling for me or, for instance, a farmer? What time points am I going to select? Will I sample just one time or do I perform a follow-up study? And of course, there are some practicalities, like the budget you have to think of. All points on these slides will be discussed in this lecture. So during this lecture, we will use the example of the EFFORT- study from time to time. EFFORT is a large European study on AMR in the food chain. Part of the study includes metagenomics sampling. For instance, faecal samples from animals were collected and analyzed in nine European countries. Also, faecal samples from slaughterhouse workers and carcass samples were collected at a slaughterhouse and analyzed metagenomically. Metagenomics sampling within effort was mainly performed in a cross-sectional manner. So which farms and slaughterhouses are you going to sample? Before recruiting farms or slaughterhouses you have to define your selection criteria for these farms and slaughterhouses. These criteria will define to which type of farms and slaughterhouses you will be able to generalize the study results to and it will define the external validity of your study. For instance, which type of species do these farms breed? Do I sample biological or conventional farms? Closed or open farms? Farrow-to-finish farms? In the ideal situation, in order to avoid selection bias, you should randomly select farms and slaughterhouses from a list of all the available farms and slaughterhouses within your study area. For instance, within EFFORT we sampled 9 different countries and used regional stratification by including 20 farms per country in a random fashion. However, lists can be quite pricey to obtain and it can take a long time in order to convince people from these lists to participate. In that case, as has been done within EFFORT, you can perform a two-step approach. You can first obtain a list of veterinary practices or cooperation of farmers and slaughterhouses, and then you can ask them to perform a selection for you. Veterinarians, for instance, are in closer contact with the farmers and might be better able to convince farmers to participate. So let us now go to the type of data you can collect. Before sampling can take place, first of all, you need to consider if ethics approval is required for human but also for animal sampling. It can take months to get this approval, so plan ahead. The approval could also involve for instance sampling protocols, sample size, questionnaires and how to provide feedback to your participants. So a lot to consider beforehand. So, what type of samples are there? You can think of faecal samples, animal feed, carcass samples taken within a slaughterhouse, environmental samples like for instance dust or air sampling or gloves worn by slaughterhouse workers. For each type of sample, you need to create a field form in order to record the meta-data. Each sample also needs to have a unique sample identifier. So animal faecal sampling. When taking faecal samples from animals, you have to take into account seasonality, which means that you have to divide your samples across seasons to avoid seasonal differences affecting your samples. Also, you have to consider how old the animals need to be in order to be sampled. Within EFFORT, we sampled animals as close to slaughter as possible, since we wanted to get an idea of consumer exposure to resistance. However, when sampling veal calves for your own safety reasons it is recommended to only sample calves up to 20 weeks of age and not as close to slaughter as possible. Above 20 weeks of age, sampling can become quite dangerous. So, should you sample fresh rectal or floor samples? According to a metagenomics study performed in 2017, pooled floor samples contain the same DNA as pooled rectal faecal samples, with the exception of a few contaminating taxa that might proliferate in the extra-intestinal environment. Depending on your study design, this means that it might be easier to just collect floor samples instead of rectal samples, which are more difficult to obtain and also could provide discomfort to the animals. We have come to environmental sampling. One way of sampling the environment of a farm is to use electrostatic dust fall collectors (EDCs). They consist of two dust clothes and can give you an idea of the resistance transmitted through air. Here again, you have to take into consideration to divide sampling across seasons. Furthermore, EDCs needs to be placed horizontally, this could be, for instance in a cardboard box hanging from the ceiling, from feed tubes or on a windows sill. EDCs have to lay inside the stable for a minimum of two and a maximum of seven days, and afterwards farmers can send them back by mail which is actually quite practical. Be aware though of ventilation systems, heating systems or sprinklers, which can interfere with your measurements. Also, do not forget to include blank EDCs, in order to test your method. Another way to sample dust or air is by using pumps with filter heads attached to it. This way you analyze the dust recovered from the filters. You can place the pump on a tripod, upwind and/or downwind the farm. Another way to sample dust/air is by personal air sampling: it can be worn by a person during a shift for instance in the slaughterhouse. Measurements should take between six and eight hours. Again while sampling air/dust you have to take into account weather conditions inside a slaughterhouse or outside the farm. A temperature of 7 to 40 degrees is recommended for this method, and a wind speed of between two and four meters per seconds. Do not forget to remove obstacles between the farm and your filter head. Farms in close proximity of the sampled farm could also interfere with your measurement. Again, do not forget to include blanks (blank filters) here. And finally, some other types of slaughterhouse samples you can think of. You can take for instance (cork borer) carcass samples, or you can sample a whole chicken carcass by rinsing it in a sterile plastic bag filled with Ringer's solution. Or you can collect gloves worn by workers to get an idea of the working environment workers are exposed to. Gloves should be collected inside out, after which you can add Ringer's solution which you then can analyze. Since microbial contamination levels and consequently resistance levels on slaughterhouse machines, and other equipment might be higher at the end of the day than at the beginning of the day directly after cleaning, it is advised to sample gloves and carcasses at different time points during the day. Depending on your research question, you could think of sampling different slaughtering steps as well to get an idea of resistance levels along the line. So, when performing a metagenomics-epidemiologic study on a farm, or within a slaughterhouse, next to collecting metadata on the collected samples, you should also collect farm antibiotic use data, since it's considered a major determinant for AMR. Also, biosecurity data and farm technical data or even welfare are important to include. Make sure to score biosecurity and welfare yourself next to interviewing the farmer in order to avoid social desirability bias. In a questionnaire you should always include an anonymized farm identifier. On the slide you see an example of an effort sample identifier from a sampling field form consisting of nine digits. The first four digits are the farm or the location identifier. And then lastly, you could also include GPS data on your questionnaire, which can be very useful for spatial analysis. So, when writing a questionnaire, be aware that all these questions at some point in time need to be entered in data entry software: for instance EpiData software as is shown here or other software packages. It will take a considerable amount of time. You could consider using an iPad at the farm or slaughterhouse to skip the data entry part. So, coming back to the sampling procedure on farms. The number of samples you take at the farm should be based on a sample size calculation. This is not always possible to define beforehand. Within EFFORT, we for instance took a convenience sample of 25 animal faecal samples per farm and placed four EDCs within each farm. Ideally, one person should perform sampling. Working in a large team however, you need to standardize your sampling procedure to avoid misclassification. Depending on the animal species, farms could consist of multiple stables, compartments and multiple pens. So make sure to randomly divide your samples across farms. A theoretical example of how to divide your samples is given on the slide. So, the sampling procedure in a slaughterhouse is, depending on your research question, quite different from farm sampling. For instance, if you want to take into account different microbial levels at different slaughter steps, you should divide your sampling over these steps. Also, as mentioned before, you have to consider sampling multiple times to correct for differences in bacterial and therefore antimicrobial resistance levels during the day. Different batches of animals carry different levels of microbial contamination. That is why you have to include sampling of multiple batches in your study design. You could choose to follow a carcass along the line, or you can just sample different carcasses and pool these samples afterwards. But in any case, be aware that sampling within a slaughterhouse takes a long time since slaughtering could take place nearly around the clock. So now, we come to some last practicalities. When sampling farms take it slow, take your time for the farmer. The farmer might want to get to know you and your project better before allowing you in his stables, but also take your time for the animals, not rushing into the stables, to avoid stress. It might also be necessary for you to shower before entering the farm. When sampling a slaughterhouse, be aware that slaughter lines move fast and so should you. Employees work very fast and will have no time to chat, or for instance change gloves when you're sampling these. Also, you will have to get used to early mornings, condensation for yourself but also for equipment. A very small working space to do your sampling in, loud noise, and smell. If you like to sample the same carcass along the line, you need a very large and quickly acting sampling team. And one other thing to take into consideration at the farm for instance, farmers are experts in their fields. So for instance, they might know a lot about AMR and might have an opinion about this. Therefore, it might be rather easy to explain your research goals. Within a slaughterhouse you have to be prepared to be able to explain your research in more layman terms, and most certainly you have to take into account different nationalities, and by that different languages. For instance within EFFORT, we had multiple different nationalities in one slaughterhouse. Therefore, you will need to hire or work with translators, to translate your study information leaflets for instance, if you want to sample slaughterhouse workers or want to involve them in your research. These differences between farms and slaughterhouses impact the way you communicate results back to your study participants as well. Delivering feedback from your results to your study population is something you definitely cannot forget. So, wrapping up, within a metagenomics-epidemiologic study, data collection is just one part. You always have to remember to randomize and standardize your sampling procedure, which will define the internal validity of your study. Sampling farms or slaughterhouses warrants different approaches. And that's it. Thank you for your attention and good luck with the rest of the course.