speaker-0: Hi, this is Frank Giannis, host of Smarter Food Safety. Thank you for joining us. Food companies and regulators test food products every day to help protect consumers and determine whether food is safe. But here's an inconvenient truth. What if many of the food tests being conducted or the negative test results being reported do not actually guarantee the absence of pathogens or even that the food is safe? Sound disturbing? Well, it should be. In this episode of Smarter Food Safety, Recorded while I was away in Greece, Dr. Marcel Sweetering, professor and chair of the International Commission on Microbiological Specifications for Foods, joins us to discuss one of the most misunderstood concepts in food safety: the role of testing. We'll explore what test results mean and just as importantly, what they don't, and how sampling plans can be strengthened to improve food safety outcomes. Why does it matter? Because a consuming public will always be more sensitive than any sampling plan. Stay tuned, you won't want to miss this episode. The Smarter Food Safety Podcast is sponsored by Ecolab, a trusted partner throughout my career. They're dedicated to protecting what's vital by advancing food safety, ensuring clean and safe environments, and enhancing water quality around the world. Marcel, it's so good to see you and thank you for joining us on our podcast, Smarter Food Safety. What a treat and what a pleasure. Thanks again. Looking forward to today's conversation.

speaker-1: Thank you for the invitation. Glad to be here.

speaker-0: Yeah, it's ⁓ or what I call early days Barcel for our podcast. Actually, this is the eighth episode, so we're fairly new and we've been really intentional and selective on what kinds of topics we wanted to hear about or what we thought the audience might want to hear. And also very intentional, not to flatter you on ⁓ what speakers that we invite. And so when I talk about this topic of microbiological sampling and sampling plans, I thought it's so important, I thought. Well, who should we get to be on the podcast? And your name, not the flattery, immediately rose to the top of the list because the important role you play. Now, I know ⁓ who Marcel is. Dr. Marcel, I want to make sure I pronounce your last name right, Swicktering. Because I followed your work and I know you're involved with ICMSF. I'm not going to read your bio. I generally don't do this on this podcast, but what I've seen over the years is you've been very involved with. ICMSF, which stands for the International Mission Microbiological Specifications for Foods. I know there's a lot of books, Marcel. It was up to eight of them. Some of the early ones certainly they had a place on my bookshelf my entire career. So I encourage readers to take a look at them. I know you're a professor at the University of Bach and Hagen. And I also know ⁓ you've published, I think, over two hundred research papers or more. Started off as a predictive microbiologist, are still teaching students. But more importantly, as we get to back to ICMSF, you're the chair of ICMSF. We're going spend a little bit more time on them. But rather than reading your bio or tell the audience, you know, more about you, why don't you tell us a little bit about your current role, what you're doing these days, and specifically a little bit about ICMSF and the role you play as chair, because that's a very important role. You ch you share that entire organization.

speaker-1: So, like you mentioned, my name is Marcel Zwietering and I'm a professor of food microbiology at Wageningen University in the Netherlands. And in that role, of course, I do teaching, I do research, I supervise PhD students, I do the management of the group. But also we work a lot together with the food industry, with all kinds of consumer organizations, with the government. And so that is also a task that I do. And I do that both in the Netherlands, in Europe and also international. yeah, the IZMZF, the Eye of IZMZF International, so that already shows that it's international commission for Margeable Specifications for Foods. And that commission exists already for ⁓ 60... two years, think, because it started in 62 or so. In the beginning, the ISMSF was mostly into microbiological methods and really developing methods to test microorganisms in food. And there it started off with, but soon it was considered that it is not only testing, but it's also really prevention, which is important. hazard analysis critical control points came up and that was then also became a very strong. tension point of the commission. So that was the second stage where they went through. Then it also went into sampling plans and the ICMSF came with 15 different cases of sampling plans, which was a nice milestone. Then the commission went more into food safety management and developed the concept of food safety objectives. We also did do a lot of work and there I was more involved in the statistics of sampling plants. And currently the commission is working, you could say, towards the future of food safety. And we do that in the commission of about 20 members. They are from all kinds of different regions in the world, but also having different competences going from ⁓ epidemiology, microbiology, people knowledgeable very much into certain commodities. people more into quantitative aspects. And that gives a very rich, you could say, symbiosis between all these different people in that we can exchange ideas. And what we then do as commission is that we, ⁓ you could say on the short term, sometimes write position papers, we write scientific articles, that's really what is going on today. We also have some sort of missionary task where we... disseminate all the information that we have gathered, for instance, about sampling plan, about food safety management. And we do that by doing workshops, by doing symposia at conferences, when we have meetings around the world, we also disseminate locally or knowledge. So that is the second thing. And thirdly, we try to, on the long term, develop concepts like the testing, the HACCP, the sampling plans, the food safety objectives. So really doing the long term issues. that is the work that we do.

speaker-0: A great overview. ⁓ so the listeners should know, and a lot of listeners will be familiar, but the International Commission on Microbiological Specification for Foods is critically important. If you're involved in testing foods, designing sampling plans, and that I love the way the organization's has kind of pushed advanced concepts, you know, like food safety objectives and HACCP. In this series of books, I think they're the last I I remember there were up to eight, maybe more, but you should take a look at their books. Marcel, we've titled this. Podcasts, false confidence when food testing gets sampling wrong. And why so early in the series of the Smarter Food Safety Podcast episodes did I want to put this one? Because one of the things I've seen over the course of my career, Marcel, I don't know if you're going to agree or not, is that this is a misunderstood and misapplied concept. In the field of food safety, you know, a lot of people are doing food testing, both in the public and private sector. Lots of testing going on. But I I just remain struck for, you know, I've been doing this for a long time now, 35 plus years, how misunderstood the concept is. And I'll just give you some examples and then get you to react. You know, I've reviewed hundreds of microbiological specifications. And in most of the specifications that companies have s submitted to me or that I've reviewed, I don't see anything stated about the sampling plan, just the test. You know, not the sampling pan. I think about my time in retail, I think you know, Marcel, I wrote worked in retail for a long time and I can do specifically not to pick on one commodity, but beef. ⁓ in the early days there was N sixty sampling plans for E. coli one five seven H seven and beef. And every single shipment of beef that I received when I worked for a retailer had an N sixty negative certificate of analysis. Yet occasionally, quite frequently, somebody would test it or there would be an outbreak or illnesses and we have to recall. So I had I've recalled dozens of lots of raw beef that had provided me with a negative certificate analysis. Powdered infant formula complaints. You know, I see how some of the powdered infant formula companies respond to the complaints. They pick a couple of cans if they can get a couple of retains and then determine or make these crazies, you know, statements that the testing is negative. That means this the protection batch is negative. And even most recently, I don't know if you saw this one. Marcelo, there's a company involved in an E. coli outbreak, a company that produces raw milk, but cheeses from raw milk, clearly strong epidemiological evidence that said they were the source of the illnesses and they were testing their cheeses and said, we don't find it. So we're not going to recall it. And so this concept of misunderstanding what a test means and what it doesn't mean, I feel has been around for so long. So the question, Marcel, is let's begin with the high-level question. Why do you think Food safety professionals continue to get this concept, ⁓ true understanding of what testing tells you and what it doesn't tell you wrong. Why has this been such a s stubborn issue in the profession?

speaker-1: Yes, so maybe in the profession, at least there we do a lot of effort and there I hope that many people do understand that habitat are sometimes the people a little bit outside the real profession of food safety management. Also, of course, sometimes it happens with quality assurance managers and so on. But hopefully most of them do get this point. But outside of that, often it is largely misunderstood. If you calculate how much material is sampled, the E. coli and beef, or also with the powdered-in-hand formula, or with any other commodity. If you look at how much percentage of the total batch is sampled, and you would calculate that it is generally 0.0002 % of the all the material which is sampled. And the only thing what you then can conclude was that that 0.0002 % of all that food did not contain the microorganism, but all the other 99.99998 % is going to the market and can make people ill. Of course, then you can say, okay, why do we do the sampling? That's not at all useful. That is also of course not true. It's good to do the sampling to verify and to find gross errors and really things that do go wrong. It also gives really a push towards better hygiene. So I'm not against sampling. But we should not misinterpret the sampling. If you find the organism, it is there. maybe sometimes your test will not work perfectly. But if you find it, it is there. And for that it is sometimes helpful because then you can recall batches or can prevent things from going to the market. But if you find negative test results, it's not a proof that the whole batch is free of the organism. And that is really a wrong interpretation.

speaker-0: Strong interpretation. That's that's great. It's ⁓ Marcel. One of the things I've often stated, I don't think it's an opal, it's what you stated, saying it a little differently, which is a positive test result means about something. You know, it's there. A negative test result oftentimes doesn't mean much, and you have to take to that with caution. And so no one is suggesting in this podcast that we shouldn't be doing verification and validation testing. The message I really want to drive home is, you know, how how do you use the results and what how much Do you read into the results, you know, interpreting it correctly? And so that allows us to pivot Marcel to the concept of sampling plans itself. And you know, I see MSF as some been so useful for me to help me think about this. In fact, when I would write microbiological specifications, ⁓ or I advise companies now, and they say, Hey, let's Giannis, can you help us review our specs, especially the microbiological component? Because I'm a microbiologist by training. I always have companies Design not only the spec, but the sampling plan to cite that in into your specification or your criteria itself. And that I got from you guys at ICMSF. But let's begin with this concept that, you know, if you want to test foods, you have to design a sampling plan. And I know, you know, you you know this better than anyone. You chair ICMSF. Yeah, I know your background, you understand statistics. What are some of the core concepts if people are designing testing programs? and specifications, what are some of the core concepts they have to think about in the sampling plan? I know there's things about, you know, probability detection, how is the microbe dispersed, et cetera. But maybe give us, you know, a reader's digest overview. What are the core concepts they should consider and why they're important?

speaker-1: There are a lot of different dimensions which are of great importance. The biggest one to start with is the severity. If the severity is larger, have to take more attention. You mentioned for instance e-hack and also powdered-in-hut formula. If you have a coronabacter infection in premature babies, it's... in many cases has a very high mortality. of course that is a big severity. The same with with e-hack in raw beef. If a little children gets an e-hack infection, can get kidney failure, dialysis for the rest of their life or also in high mortality. So that is a big severity issue. While you have other also other hazards, for instance, certain toxins that you only need to vomit a couple of times. That is of course, it's not pleasant and But it's of course a much less severe hazard. So this is one very important dimension for sampling the severity of the hazard. The other thing is also if at the moment of sampling, from that moment on the hazard will remain stable or if it will go up or if it will go down. that's if you have for instance EHAQ in beef, in principle you should do cooking of the beef and therefore the hazard will go down. But of course in the middle of the beef it can maybe survive. So it's not a total destruction. It can also be for instance Listeria and ice cream that will generally remain stable because it's minus 18 degrees that does not inactivate the Listeria but it will not be able to grow. But it can also be in the case of the E. coli and beef if you store that at too high temperatures the level might increase. So you have to consider will the risk go up? Will the risk go down or will it remain stable? That also has an impact on your sampling plan. But also the overall control that you do have. Because if, for instance, you have a sterilized food product, then you can also do sampling. But I would say that the control is totally in the process. So you can start testing Clostridium botulinum in canned foods. Also Clostridium botulinum is a very severe hazard. aspect, should maybe be important to sample, but this is not the way to control this hazard. You have very good process control. it would be, and the probability that you will detect Clostridium botulinum is one in a thousand billion. So there is really no need in sampling in that case. So, and on the other side, you have certain raw products and that is totally different there. is raw products with not much control in the food chain. There You have to trust more on the sampling, although you cannot really trust in it. But it's also how much control you do have in your food chain if you need to have more effort on the sampling or not. So that are several dimensions that are ⁓ relevant for sampling. You also mentioned, for instance, heterogeneity. That can also be relevant. for that, is very relevant to always take a lot of samples. You mentioned N equals 60 for B, instance. You take a lot of different samples at different locations. And in that manner, you really sample on several places. And in that manner, you can find also local contamination, which can in certain cases, of course, be very relevant if you have heterogeneous distribution. And generally in microbiology, you could say you will have heterogeneous contamination. That generally will not be very homogeneous in food products, except maybe in very well mixed fluids. But in most of the cases, it's heterogeneous and therefore it's good to have a lot of different samples that you all do analyze.

speaker-0: Just that's a great great concept to close on. And the knowing a little bit of assumptions on the dispersion of the microids in the in the product that you're testing. So I think about, you know, in the US it's common to do field sampling for stec and E. colion one five seven. And you know, the data suggests that the contamination is very sporadic. So they take large samples and they do Z pattern samples, et cetera. I think about powdered infant formula, for example. ⁓ you know, the literature suggests, and I think you'll agree with this, that contamination is sporadic. And they even use a term, Marcel, you probably agree with this or can collaborate, clumpy, you know, that it's it clumps together and it's not homogeneous. ⁓ so the understanding a little bit of the dispersion ⁓ of it. And what I'm hearing from you is when we know it's not homogenous, the larger the end, the number of samples you take, the greater the probability that you find it. Okay. Let's pivot to this concept of just because people are testing, and oftentimes I feel Marcel, they're not testing large samples and the probability of detection often is low if they're not designing their sampling plan. We tend to see these negative certificates of analysis and people place a lot of confidence in it. Right now I'm working with Michigan State University, Massey University, University of Otega, and New Zealand. To try to study this concept of what causes companies to become complacent in food safety. And we think there's probably a few factors, but Frank Giannis personally believes one reason companies become complacent in food safety because be they're overconfident. And we've seen companies, for example, that just became overconfident. You know, we had an ice cream manufacturer here that had been in business 108 years, never had a problem, and then they had a Listeria outbreak. Or you can see people that have negative certificates of analysis repeatedly, and then they're surprised when they Find that their product is linked to illnesses. I think that this concept of repeatedly getting negative test results with low probability sampling plans could be leading people to become complacent. I see in one of ICMSF's books, it says something you've already stated: microbiological testing of ⁓ products, in and of itself, does not ensure the safety of food. What do you think? Do you think companies or individuals do become over? complacent just because they continue to get negative test results. And if that's the case, what advice do you have for them?

speaker-1: Yes, indeed, I mentioned already the 0.0002 % of sampling. That does not say that the rest is negative. But that other part is all consumed by consumers. Of course, not everything, but the big majority of all the other material is consumed and can cause illnesses. I made some calculation examples where you have a very low, very well controlled, for instance, salmonella and chocolate. And if you then look at a very low contamination level, what the probability of finding a positive is, then you can for instance, that you find a positive once in five years. And then you think, okay, everything is fine with my one in five years. But if you then calculate how many people you make ill at that same level, then you make for instance, 10 people ill every year. And that can nowadays also sometimes be linked to your company with whole genome sequencing. That is then, you can get caught. But also you have also the other side, course. You make 10 people ill with your food product. And at a certain moment, you find a positive. But also five years you don't find the positive. And then the risk is also that if you don't find these positives, that at a certain moment you get a little bit of more negligence. That you say, why are we doing all this effort? It costs a lot of money. The testing costs a lot of money. All the other... activities that we do cost a lot of money. And slowly, we are going to pay less attention. And if even that you do, then that risk that I was just mentioning and the amount of people that you make ill is even increasing. And so that is indeed very dangerous to do. So my advice would be to remind people of these outbreaks that they can occur and that sometimes you see those that also in time. hitting back. If we look at salmonella and chocolate, that are outbreaks already since 1960. And if you look at that, then they do appear and then people pay a lot of attention to it. And then slowly the attention seems to drop. And then after 10 years, outbreaks do occur. And we have had the same with coronabacterium, powdered infant formula and around 2000. That was really, you could say a hot topic. And then slowly attention drops a little bit and then it's... it comes back again. this shows that really we need to show attention to the pathogens and to these issues and not assume that if we don't see it anymore that the risk is not existing.

speaker-0: Yeah. Yeah, that's that's in great examples. And I hear you. I mean, just 'cause you're getting the this is what I mean, the negative results, you assume overconfidence and maybe let your guard down and you're saying we can't do that because it's a probability game. And then you see these outbreaks repeatedly occurring. So I love that. This might be a very tough question and you might think it's unfair. But I thought I'd ask you anyways. If you had to choose just one, what do you think is the greater danger? Marcella, you can only choose one of these two options. What do you think is the greater danger? Do you think not testing enough or believing too much in your testing?

speaker-1: I think I would go for the first, not testing enough. I should also say that it is testing of end product, but not only that. It's testing of end product, also of your environment, of your raw materials. So it is more the precaution where I would put my effort on than that last part. If that last part of course results in the fact that you are going to reduce... the amount of sampling, you come to the same point. But I think really not testing enough, that is the biggest danger. Additionally, that is your second point. I'm not allowed to vote also for that one. But that has some sort of negative effect that you are going to do the first thing. But the most important, enough testing. And not only of the end product, but also of other aspects.

speaker-0: Okay, like environmental monitoring and we'll get into that. But that's ⁓ I I'm fascinated with the way you answer that question. It's a tough question and you want to answer them both, right? They're both important. We want you to test more and we don't want you to become overconfident. But if you had to choose one, the importance of testing more, and this speaks to trying to create statistically valid to the extent that you can, sampling plans that you can have confidence in. Okay, so that's good. I I want to pivot to the regulators' role. And this is not to be critical of regulators. But what I think I've seen over the course of my career that sometimes regulatory policies, it's not only regulators, sometimes it's policymakers. They want the private sector to do testing. And by the way, ⁓ we didn't mention in your bio, Marcel, I know you have private sector experience. I think you worked at Denone, right? And Unilever, if I'm not mistaken. So you got private sector experience. You know this. Everybody wants the private sector to do testing. Regulators, policymakers, and They actually ask you and I'll give you some examples. You know, when I was at FDA, the FDA required powdered infraform manufacturers to test powdered infraformula using an M30 sampling plan. 30 samples per production run. And we'll talk a little bit about these production runs. They can be un enormous. 10 grams per each. So you're testing 300 grams. Sometimes in a production run, that could be 44,000 pounds. So you do the math there. We s we're seeing it right now, state lawmakers saying we want you to test these products for heavy metals, but they don't they don't ever talk about sampling plans. There was an outbreak here maybe about fifteen years ago of salmonella Heidelberg and raw poultry, Marcel. And it was a company that had done very good work reducing rates of contamination of salmonella on raw chicken carcasses 'cause the US Yay, the United States Department of Agriculture had a performance standard on whole birds. They didn't have any performance standards on chicken parts. And it was the chicken parts because after you get salmonella down to low levels on a herb whole bird and you start cutting or dicing it, contamination rebounds. And so I think you know the regulators kind of had them focused on the wrong testing regime, right? You're testing whole carcasses as opposed to chicken parts. So I've I personally feel like sometimes Unintentionally, and it's not to point fingers or cast blame. Sometimes the wrong pressure is put on by regulators to l to to kind of cause the private sector to become overconfident in their testing. Have you seen that? You have you seen any examples of that in your in your work?

speaker-1: Yeah, globally what you sometimes see to say it a little bit in a negative way that if there is somewhere a big outbreak, if something is going wrong, then the reaction sometimes of the regulators is, okay, the solution is let's do testing. We have had a big EHAQ outbreak in sprouts in Germany and in Europe. then the reaction is we make legislation for EHAQ and sprouts, but E-hack of course is a problem in beef and sprouts, also in leafy greens and in all kinds of every ready to eat commodity that one way or another can get in contact with the fetus of animals can result in a problem with E-hack. There was an outbreak with sprouts and then, okay, let's make legislation for sprouts. So that is not ⁓ a very strong reaction, but it is just okay. We have then covered our issue. or testing, so the problem is over. And this is again that misconception. So it is an easy way to get ⁓ rid of such a problem, which in reality you don't. But this is the negative side. But on the other side, when that legislation came out, there are also some very positive aspects in what they have done in that ⁓ Sprout legislation and the European legislation is because... It's ⁓ a ⁓ multi-dimensional legislation, could say, because they say every batch of seeds for sprouts, you first need to test. And that is really proactive. So, already you take a little bit of a sample of these sprouts and you sprout them and you look if the organism is already present in your seeds. So that is really a very positive side of this legislation. And also, you have to test the irrigation water of your sprouts. And the nice thing of that irrigation water is that it gets into contact with all the different parts of your batch. And therefore it collects, you could say, those clumps of cells that you earlier mentioned. You collect also from that material. So the heterogeneous contamination in that manner, you can tackle a little bit. And also there is a part on end product testing. So it's a nice three-dimensional way. So sample your seeds. sample the irrigation water and the end product. the negative side is it's closing the stable door while the horse has bolted. the problem did occur and we are now hiding behind this legislation you could say. But on the other side, the legislation is also multi-dimensional, a little bit proactive for the seeds. It's also tackling that heterogeneity and also doing some end product testing. Given that example, ⁓ it's two-sided and there are certain positive aspects and certain negative aspects to it.

speaker-0: That one's a good example because ⁓ it doesn't rely just on end product testing. I think testing the raw materials and the way you can do it with the sprout seeds by testing the the rinse water, you know, you increase the chances of finding the, you know, contamination. I think that's a good one. There was a recent example here as you're speaking, it made me think of a recent report that just came out a couple of weeks ago. And I'm not picking on any type of products, it just I'll bring powdered infant formula again because I just saw when I say I see sometimes sometimes regulators get this wrong. There was a recent report here where the US FDA and I have a lot of admiration for them. I spent almost four and a half years of my career there. Great people, Marcel, but they they were looking for heavy metals and powdered infant formula. They were looking for PFOS and some pesticides. They took a total of I think three hundred and twelve samples, if I'm not mistaken. And they put out a report that said, you know, most of the samples met very low levels of ⁓ contamination. And I'll quote what it said, that the results affirmed the safety of the U.S. infant formula supply. 312 samples. I know because I was at the agency when we had the infant formula crisis here in the United States, and this is in the public domain, the White House published it. Back then, Marcel. The United States was selling, consumers were buying about four million pounds per week. So you can do the math. Four million pounds of powdered infant formula salt per week in the United States. And we test three hundred and twelve samples and make statements that say it assures the safety. What's your reaction to that as a as a you know, somebody that looks at sampling plants?

speaker-1: I would say I would not dare to make the conclusion that infant formula is safe for two reasons, even for three reasons. One reason is what you say, there are only 300 samples, where there are an incredible amount of infant formula is every week or every day or every year consumed. So it's only a very limited amount of material that is investigated. this is reason one. Also, what is very important is they only looked at the chemical risks and they said powdered infant formula is safe, but there can also be macrological risks that they did not test for. that, I mean, a couple of days after this report came out, there was also a recall of powdered infant formula of a certain brand because there were serolites found in that food product. that is already a proof that they didn't look at that. even if they would have tested also for serolite, maybe they would not have found it in those 300 samples and the 16 brands that they did look at, because there are also more than 16 brands of infant formula that are ⁓ distributed in the United States. it is a false conclusion to say that it is then safe. What you should say is, did do this big investigation, we evaluated it, we looked at the chemical risks and in the samples that we analyzed, no dangerous levels about the limits were found. That's factual. But to extrapolate that the product is safe, is a wrong conclusion and we should always be aware not to extrapolate that and say that a certain thing is safe. We should only say we did test the samples and it was negative. And in principle, that's also if you get your ⁓ beef in with the certificate, it is not saying that the beef does not contain E. coli. It only says, we tested 60 samples and they were negative. And that is the only thing that you can conclude. We tested it and the samples were tested negative. And that is the important thing. I hope that it is not really the scientists that... FDA that extrapolated that. what is sometimes happening here is that also the communication departments of certain companies or the communication department of the FDA or sometimes also of universities. Because we have the same that sometimes are read things in on websites of universities. And then I think that is not possible. And then you really go to the papers and then you see, what what they say in the scientific paper is much more nuanced. But then sometimes it's translated sometimes by the communication department and sometimes by the media. So that is also the risk that a certain message is slowly on changing, being nuanced from the lab to what's less nuanced in the communication department and even less so in the media and even more so in the title that the media.

speaker-0: Right, right, right, right, right. So ⁓ a great point on the importance of risk communication. And you know, it's it's really important because mislead consumers and the public. But we do have to scientists make sure that we interpret the data correctly. Hey, ⁓ we're I I know we can go on for hours. We're gonna be having to wrap up here. I have three questions that I want to get through still, believe it or not. Marcel, since I've had the privilege of having you on, but one is do you because you mentioned there's other forms of testing, not just end product testing, and you alluded to that in some of your responses. I always like to draw the distinction between testing products versus testing the process. So and you know, we've been focused a lot today's conversation about end product testing, but there's validation. It's what I call verification and validation. You know this as well as I do, two very critical concepts of HACCP. Maybe explain a little bit on how you say testing end product versus testing the process because I I sometimes see people not, in my view, doing enough of testing or validating their processes. I've been involved in some outbreaks where I actually went out and visited suppliers that had an intervention. ⁓ I remember very early in my career with ⁓ diced tomatoes. So we got very concerned the company I work for at that time. We'll call them Theme Park X. And we started going out and see, and the companies didn't have validations of their wash process. So we started doing the validations for them. And I saw Marcel, this was the extreme, companies that had very effective sanitation and washing process of their tomatoes, and they could get a two-log reduction. And other companies had process, they thought they had interventions, but those interventions were resulting in a two-log increase. And so maybe speak about the importance of testing not only the process, ⁓ the product, but also the process.

speaker-1: Yeah, you could say even in the process, ⁓ you could say two ways that there is also verification if you do environmental sampling, that is also some sort of verification that you just look at. the same for the raw materials that I mentioned earlier. But then more important than verification is that you be ⁓ proactive and that things are well designed. that one of these well designed aspects is that you have your processes well designed. And if the process is well designed, you also need to validate that and to show that you have a certain reduction while cleaning or that you have a reduction while doing a certain ⁓ processing. And that you need to validate. And that validation should be based also on experimental determination. That is sometimes very difficult to do that, of course, really in a process line, because that's very big scale and it's difficult to perform, especially with pathogenic organisms. So you can sometimes do that in a pilot plant or sometimes you can do that more in a smaller experimental unit or sometimes you can base it on literature data. you can sometimes do that with not with pathogens, but surrogate microorganisms. None of these things is generally ideal. And for that reason, it is good to often make use of all these different aspects. So if you would use only literature, literature data is very rich. You can collect, find a lot of information in the scientific literature and databases. And that is an easy to get huge amount of information, but it's not always fully representative, but still very useful to do. You can have... laboratory experiments. In laboratory experiments, can very well design your experiments, collect a huge amount of data. That is against a strong set of data, but it's not totally representative of what will actually happen. Then you can go to, you could say pilot plant scale and there ⁓ validate things and do measurements, but that is already difficult to get a lot of data because it's very expensive and heavy to perform. And then the last aspect is really, you could say the proof of the pudding is to really do the validation on industrial scale. But that is for two ways, could say almost impossible because ⁓ it's impossible because of the costs that it will have. But also you will of course never put a salmonella into your process and to test if it is then inactivated during cleaning in a process line because you don't want to bring in that pathogen in your factory. You can then use surrogates, but even that you would not like to do in a process line. Maybe that you would do in a pilot plant, because sometimes in a pilot plant you also do not want to have a pathogenic microarray. Because you might use that pilot plant also sometimes for food grade experiments. And that is the difficulty. None of these methods are perfect. So the best thing is to make use of a combination of these.

speaker-0: Right. Yeah, very good. But I appreciate you saying that, hey, you know, more important than the verification is designing a process and validating the process works. And that's HACCP 101, right? And we can't forget that. So validations and to the extent that you can test and quantify it. There's challenges, but Marcel has talked about ways to do that. Looking at some studies from the literature review, if you have access to pilot plants and implant it it is very challenging. Marcel. As we start rounding the corner here at the end, I have to ask you a question because I think listeners would say, Giannis, how did you not ask this question? We're living in a time and day where AI and predictive analytics is all the rage. How do you foresee how AI and predictive analytics is going to change microbiological sampling since we're talking about this? I had a great experience at FDA. We actually We're very concerned about seafood imports. I think you know about 94% of all seafood consumed in the United States gets imported by abroad. We've been testing it for years. While I was there, we did a couple of pilots where we actually leveraged machine learning to increase our predictive capability of detect determining which sh seafood shipments were contaminated, increased our chances of finding contaminated shipments multiple fold. It was really, really productive. But what's your view on using AI? To strengthen sampling plants.

speaker-1: Yes, so I have, you could say 50-50 ⁓ positive and negative ideas about making use of AI. Because the example that you mentioned, that is great, because you can really do some sort of risk-based sampling, where you have more probability that things go wrong, you put more effort, and that makes your process much more effective, and in that manner, you much better are able to... to find the things that really do go wrong. But you're also channeled a little bit towards those type of aspects. And you might also miss certain things that you never know. The things that you do not know, you don't test and therefore there is also not put any effort on it. if I would have a certain amount of money on sampling, I would put 50 % on that aspect and make use of that risk-based sampling. But I would also do 50 % on, you could say, really unbiased sampling, because everything what you do not know, you might encounter with that unbiased sampling. And otherwise, you might be too narrow-minded going towards the thing that you already do know. And nowadays, AI is a little bit hyped. People think that in all kinds of domains that we can now do everything with ⁓ AI. I don't believe that. I think it's a

speaker-0: Well

speaker-1: It's an incredible amount of nice things we can do with it, but we should not fully trust it and think that it will do everything. You mentioned also that I started my career in predictive microbiology, so that it was in the 1990s. It's a long time ago, but at that time it was a little bit the same. It was a new development. And then I always mentioned there are two people wrong, that are the people that believe in predictive microbiology. And that are the people that do not believe in predictive microbiology. Because if you fully believe predictive microbiology, that is wrong. But if you totally do not believe it and are not going to use it, that would also be a pity. And I have the same idea nowadays with AI. If you believe AI, that is not really wise. And if you do not use AI at all, while it is a very strong method to, in certain cases, improve your efficiency, That's also not very wise. But what we at least definitely always should do is make use of AI and then critically assess what is coming out of this AI. Because there is nothing wrong with making use of AI. But what is wrong is if you totally uncritically make use of the results that AI is proposing to you. It's the same as what we have always done with correlation and causation. And it's the same type of thing if you... find all kinds of correlations, what in principle AI is also doing. And if you believe that that is causation, that is wrong. However, a correlation can sometimes point you in the right direction. So it's very strong to do all kinds of correlations. But then believing that that is then causation, that is the wrong thing. And you see this coming back in predictive microbiology. And for instance, it can be also in epidemiology or in other statistical. processes where you look at correlation. And the same we see now in AI. is sometimes you could say stupid use of AI. And what we should do is ⁓ make use of AI in a very critical way.

speaker-0: Yeah, that that's great. ⁓ and I know the lost the audience is listening in, which is use it, but be cautious on how you use it. And I thought the same thing. Well, if you're g if your if your results are always negative, it's gonna tell you you can back off a sampling. And if something changes, then you're gonna you're really gonna regret that that you were sampling less and not sampling more. And so it's not the silver bullet. It's you know, not it's not gonna be you know, the magic solution. So use it wisely is what I'm hearing you. But it it does it does have ad advantages. I think you have to have large data sets too and ⁓ something that we had. Marcel, we're near the end. I want to be respectful of your time. ⁓ I I'm so grateful that I get to join you in the same time zone. I should have said that at the start, but I I'm joining you from Greece. I'm supposedly on holiday, but I didn't want to miss an episode and I didn't want to miss the opportunity of getting to talk to you. Marcel, I'm gonna close with what I think It's a challenging question. This has been a fascinating discussion. We could go on for hours. I want people to go out and look at the ICMSF books and watch some of your videos online. But ⁓ the question I have, because I've always said to the audience, which is if you're just listening to be intellectually satisfied, ⁓ that's okay. But we would really want this to lead to people to take smarter actions. If you could change one thing about how the food industry thinks about food testing, what would that one thing be? I want people to, you know, this is about action. And ⁓ you've encountered a lot of food safety professionals in the industry in particular over the years using testing. If you could change one thing about how they think about food testing, what would that one thing be?

speaker-1: That is what we have been discussing about. ⁓ Do not trust that sampling. My slogan would be, what we should do is always do your ultimate best. And then afterwards do a test. Put your effort mainly on good design of your processes, on validation. But even one level below that, that is good hygiene. Those type of ⁓ aspects, there we really should put... a lot of effort on and don't forget your basics, don't forget your design of your product and of your process, don't forget about your validation. And if you have done that all at your best, then afterwards for verification, you can do a test. That would be my slogan to end this conversation.

speaker-0: That's great. And it's it's, you know, sometimes these rhymes help people to remember. No, always do your ultimate best, which is design your food production facilities, your food processes the best that they can be. Validate them, make sure that you have interventions that are effective and are working. And then and only then should you consider the role of testing. And ⁓ I hope that you do the testing using robust sampling plans and using some of the ICMSF principles. Marcel, what a privilege and treat to get to spend some time with you. Thank you for what you're doing. Thank you for the continued role as chair of ICMSF. ⁓ I I look forward to new books. Thank you for gener you know, I know you're helping develop a develop the next generation of food safety professionals. Appreciate what you do. Thank you so much.

speaker-1: Thank you for even doing this while you're on holiday. Great.

speaker-0: Alright, great. We'll talk soon. Alright, that's another wrap. I hope you enjoyed today's discussion as much as I did and found it useful. The next episode is already in queue and it's yet another good one. If you have ideas or suggestions, we'd love to hear from you. And if you haven't already, please subscribe on YouTube, Spotify, or Apple Podcasts and join us for these important conversations. Together, we're going to learn from one another and help create a safer, Smarter and healthier food system that benefits food producers, regulators, but most importantly, all of our bosses, the consumers. Stay tuned, we'll talk soon, and until then, be safe.