 So, Hounet, as discussed previously, has three main components. The interesting one is data analysis. But before doing that, you must do either data entry manually or backlink to get your data into Hounet. But of course, to start everything, you start with laboratory configuration and installation. So we've already done all that. So now we're ready finally to move on to data analysis. So a lot of people, when they think of data analysis, they immediately want to know percent resistance and outbreaks and emerging trends and new things. All of those things are important. However, when I start, my first point is really to understand the laboratory's abilities. What is their data volume? Do they have samples that make sense? Are they making obvious mistakes? For example, in one month, how many blood cultures do they do? What percentage are positive? What percentage are negative? Are they testing the correct antibiotics? Do they have the ability to do difficult organisms? I'll choose some common ones like strep pneumonia, haemophilus, or do they do anaerobes? Do they do mycology? Do they do fungi? So these are some of the things I look for. So in all of the slides I will discuss with you, I will eventually get to epidemiology issues, surveillance of resistance and infections. But before that, I will always start off with those two other elements about laboratory capacity as well as quality assurance. Well, now there's two ways of running analyses. One is through the screen that I'm showing you here. It is the normal data analysis module where you play around with the data. I want to try this. Let me try these organisms. Let me try these specimen types where you interactively work with the database to do the analyses that you want. Of course we will start with that. The second way of running Hoonet is what's called quick analysis, where you batch things together. If you do the same analyses every Monday or the first day of the month, if you do the same analyses every quarter, then you would like to just hit the button and have Hoonet do a series of five, 20, 40 analyses one after the next. And so in Hoonet we call the quick analysis. It takes advantage of Hoonet's macro features. Features can be combined into reports, and you will see all of that. One nice thing about quick analysis is that it can also be automated. So in my own hospital, we download the data. My hospital exports the data to us every morning at 1 a.m. We use backlink automatically at 1.15 to convert it into a Hoonet file. And at 1.30 Hoonet runs a series of these quick analysis reports to generate a series of Excel files or AXIS files. For our infection control team, people every morning look at the infection control, look at the Excel outputs. So on a daily basis, nobody is looking at backlink, nobody is looking at Hoonet, we've automated all the steps. So Hoonet is simply looking at, the infection control staff is simply looking at the Excel outputs from Hoonet. So we'll eventually get there, but the place to start is in normal interactive data analysis. So in Hoonet, you go to the open Hoonet, choose your laboratory, go to data analysis, and you will see two options, normal data analysis and quick analysis. And when you choose the normal data analysis, this is the screen that you will get. So on the screen, you see a series of questions. There are three required questions. They appear on the left. Question number one, what analysis do you want to do? Do you want to run percent resistance? Isid alerts, outbreak detection, do you want to count the number of E. coli, do you want to count the number of blood? So you simply tell Hoonet the kind of analysis you would like to do. Then you can tell Hoonet which organisms you want. You can be very specific like staph aureus and E. coli, or you can say all organisms or all gram negatives or all salmonella or all intravactory EACA. As a reminder, Hoonet can include fungi and parasites and viruses. Hoonet does not do virus antibodies, but if you find a virus like through an antigen test or a molecular test, you can also manage those data with Hoonet. So you can do individual organisms or groups of organisms such as all parasites, all fungi, etc. That's question number two. Question number three, which data files do you want? Do you want the January data, 2019 data, do you want the data from hospital five, the data from hospital 20, do you want five hospitals all grouped together? So with those three questions, that's kind of analysis, which organisms, which data files, you tell Hoonet what analysis, with which species, with which data set, then you are ready to begin your analysis and then Hoonet will give you the results. But there are other questions that you may choose, and I'm trying to remind myself to go more slowly, thank you for your patience, so I am ready to begin the analysis. Or there are some other optional features that are also useful. For example, maybe I do not want all of the staff and E. coli, maybe I only want the urine isolates, or the ICU isolates, or the imipendium resistant isolates, or the female outpatient urine isolates for typical outpatient urine or tract infection. So those are options, useful options, not required, but useful. Hoonet also has some features such as one per patient. If you have a patient with 20 isolates of staph aureus, I want all 20 isolates in the database. That way I can look at the patient moving around the hospital and the different dates and the different specimen types. I can also see if the resistance pattern was consistent. Did the patient start with a sensitive strain and keep that sensitive strain? Or did the sensitive strain eventually become resistant? Or did the patient pick up a second unrelated strain can also happen? So I want all the isolates in the database for some analyses. But for other analyses, I just want to count each person once. First isolate per year, per month, per resistance pattern. There are a variety of ways that Hoonet permits you to find one per patient. We will talk about those later. Then Hoonet also has this one called options. Some people want to see the percent resistant. Some people want to see the number resistant. Some people want to see the number of isolates. Some people want to see the number of patients. So these are just small features, small adjustments to allow you to tweak the output. So now that I've answered the three questions on the left and answered my possible options on the right, then I am ready to begin the analysis. The lower left hand portion of the screen, it says macros. Very important, but we will talk about that later. So now I will show you a series of output slides, highlighting and discussing the purpose of each. So Hoonet, the first analysis at the top of the screen in analysis type, is called isolate listing and summary. So the list is obviously a list. We see a list here. On the next screen, we see a summary of the list. So first we will start by discussing the list. In this example, in the upper right-hand corner, slides in Spanish at some point, I will update that, microorganism is staph aureus. There were 880 isolates. And I requested the ones that were oxicillin resistant. So these are isolates of MRSA. And here you see many isolates from the same person. So if I go down to this person here, there's a patient with the number starting 3007, 824221. You see this patient at MRSA six times. The patient at MRSA in room 77 on October 21st. And then the patient moved to room 67. So this patient arrived in the hospital and moved to one room of the hospital. And then they were moved to a different room. In fact, I do know this hospital is actually the intensive care unit. So the patient started off with organ and drain and bronchial and sputum. So this patient at MRSA in different parts of their body. I can see the amicasin zone diameter is 12, 14, 14, 16, 13. So I can see a number of things here with the repeats. I can see the patient moving around, the different dates, the different specimen types. I can also see how consistent they are with each of the days. And you see, look at the amicasin. On this day, they had the zone diameter was 12. And then a 14 and a 16 and a 13 and a 14. It's very reproducible. There is sort of an inbuilt two millimeters plus or minus variability and susceptibility testing, just diffusion approximately. So the true number is probably like 14. So it might be somewhere between 12 and 16 if you add in, you know, plus or minus two millimeters. So it's very consistent. So if a patient has a strange result, if they have that strange result four times, it's probably true, reproducible. On the other hand, if the patient has a vancomycin resistant staphylococcus aureus once, it's probably just a mistake because it's commonly a mistake. But if I see repetitions, that's more likely to suggest it's true. So these are some of the things that I can do with a listing. Even though it's a very, very simple analysis, it's one of the most common things that people ask. John, please give us a list of the positive blood cultures from the nursery. John, give us a list of the imipenem resistant enterobacteriaceae. Please give us a list of the diphylobothrium latum, a rare parasite. So like bug of the week. So it's hardly an analysis. It's simply a filter, a list. But we do give these lists in Excel to our orthopedic staff and our transplant staff, our infection control staff, because they have their work to do. They don't really need me to analyze the data. They do want to see their patients. Once they have their patients, they can do their investigation. The orthopedic team, a hip infection or a knee infection, they just want the list of the people and then they can go investigate the risk factors and if there are any issues. So isolating is extremely simplistic, but it is one of the most common things people request, especially for high priority patients or high priority resistant results. Let's see, a number of people also use HUNET for public health reporting. For example, in the United States, in almost all hospitals, in almost all countries, people have a list of notifiable diseases. Salmonella typhi, vibrio cholera, bacillus anthrax, of course now COVID-19. If you have one of these reportable conditions, HUNET, you can use the isolate listing feature to make a list of those people and then you can send that list once a week to the National Health Department, giving the patients details. So I'll finish down with the list, move on to the next slide, which is a summary of the list. The summary is counting. You can count the number of staff or E. coli or pseudomonas or you can count the number of bloods or the number of urine or the number of ICU or the number of results from laboratory one. So you can summarize the data in many different ways. In this example, I have asked a summary by Sala, that's the Spanish word for room. So we've sorted by room by date. So I can see that in room 38, room 39, I can see how many. I mentioned room 67 is the intensive care unit. There were seven patients with MRSA in January, 13 February, 8 March, 8 April. So I can see that this is the room with the most MRSA. So room 67, as you would imagine, has a lot of MRSA. It's the intensive care. If I draw your attention to the top of the list, you'll see room 41. Room 41 has an MRSA patient in January and February and in May. And three in July and two in August and one in September. So in the first half of the year, they had three people with MRSA in this room, room 41. In July, they had three people in one month in that room and two in August. This is making me suspect possibly an outbreak. This hospital, the room does not have a lot of MRSA, but now all of a sudden, they have five in a short period of time. Is this an outbreak? I do not know. It's not Hoonet's job. Hoonet's job is to tell you this is unusual. This is different. This might be an outbreak. Or it might be a quality control problem. Or it might just be a coincidence. So that's your job. But Hoonet will help to show you these things. What I have done here is I've circled in red something to me that looks a little bit unusual out of the ordinary, five isolates in a two month period. This is what I would call manual outbreak detection or visual outbreak detection. By looking at the table, by looking at the graph, it kind of looks more than usual. I can do this by species, for example, salmonella MRSA with resistance. Or I can do this by room, as you see here. I can do it by resistance pattern. I can do it by many factors. So in this example, my summary is room by month. But you can summarize the data in a variety of other ways, and we will see how to do that. This is visual outbreak detection. Or instead of I don't need the months, I can do by year. I can do the organism by year. Or I can do organism by ward or organism by specimen type. Or very valuable from a national data management perspective, I can see organism by laboratory. How many E. coli did laboratory one and laboratory two and laboratory three have? So by doing something by lab, then it allows me more easily to do national benchmarking. Any questions so far? I've shown you the main screen for HUNA data analysis. And then the two screens for the first analysis called isolate listing and summary. I'll wait a few moments, and if no questions, I will continue with the next analysis. I have a question. Yes. Yeah, this is Gabri. Is this also helpful to epidemiological reporting, the one you explained now? This probably kills the report. Yes, if you can answer that in two ways. HUNA does have a predefined list of 190 alerts. That list includes important resistance, important species, and quality control alerts. So a number of the rules that I give everybody in the world include things like anthrax and cholera, somenolotypy, paripenem-resistant interectoreaceae. So HUNA gives you a list of things that potentially could be reportable. And I will show you how to do that. It's just called HUNA's isolate alerts feature. In addition, you can make your own list of organisms. I'm going to go back to slides. Here on the left, I selected step four, SND-Coli. But you could predefine a list, and you can make your own list. I want cholera. I want different viruses. I want different parasites. So you can make your own list of different reportable species or different reportable resistance phenotypes. We can use the macros feature to help facilitate that. So yes, you can make a national reportable list and implement those either through organism selection, as I've done here, or with HUNA's isolate alert features. HUNA has a predefined list of 190 alerts, but you can also create your own alerts. And also just to re-emphasize that point, HUNA is not all, yes, HUNA, it can be used for surveillance of antimicrobial resistance. That's why we made it. But HUNA can also be used for other organisms where resistance is not a big issue. Like anthrax. Anthrax is important. It doesn't matter if it's sensitive or resistant. It's the organism's important. Cholera is important. Salmonella type is important, even if you don't have susceptibilities. Or fungi. Aspergillus fumigatus can cause important hospital outbreaks, even in the absence of susceptibilities. Invasive group A streptococcus, streptococcus pyogenes, can cause necrotizing fasciitis, the so-called flesh-hitting disease. This is an important organism, even in the absence of susceptibility test results. So HUNA can be useful for you for studies of surveillance, for studies of resistance, of course. But it can also be used simply for reportable conditions, looking at epidemiology and trends in any organism with or without susceptibilities. Other questions? Okay, I will proceed. So that's isolate listing and summary. The next analysis in HUNA is called percent RIS and test measurements, or the histograms. And in this example, we're doing pseudomonas aeruginosa. So in this analysis, I really am focusing, of course, on the antibiotics. You see there's one antibiotic on each row, in the case, industry, and emcephapine. At the top, first of all, I'll start pseudomonas aeruginosa. You know, so many laboratories do not call it pseudomonas aeruginosa. They call it pseudomonas species, because they don't always have the reagents to know the species. So the first thing, same thing, Klebsiella pneumonia. Some labs call it Klebsiella pneumonia. Other labs call it Klebsiella species. It tells you a bit about their ability by how detailed they go. For example, a machine like a Vitec or a Marcuskin would never say Klebsiella species. It will always say specifically which species it is, as all the biochemicals are there. But when you do the test manually, you don't always speciate. A very good example of this is Quaggules negative staph lecococcus. A Vitec will never say Quaggules negative staph. They'll know which one it is. Staph epidermidis, staph sephrophidicus, et cetera. So by looking at the degree of speciation, it already tells me something about the laboratory's ability. Are they using a machine? Are they doing this manually? Do they have all of the reagents needed for common identification? Or do they have important gaps? That's the first thing I look at. Second thing I look at, in one year, this is one year of data, they had 356 isolates. Well, a year has 365 days. So on average, this laboratory is very consistent. They seem to have one pseudomonas every day on average. So it tells me a bit about the volume of the data. Okay. What do I look at next? I look at the list of antibiotics. Are they testing antibiotics that make sense? Does amic acid make sense for pseudomonas? Yes, it does. Psephotaxim, yeah, kind of. Psephotazidim, yes. Psephotazidim is a very good drug for pseudomonas. Psephotaxim, sure, why not? It's not as good as Psephotazidim, but as long as you're testing a certain set panel for your gram negatives, there's nothing wrong with Psephotaxim. Gentamicin and epenem. So my comment here is, yes, these are appropriate drugs, but very commonly people do not test the appropriate drugs. There are some obvious examples. You do not test vancomycin on E. coli. Why? It's resistant. Of course it's resistant. It's always resistant. You could test it. There's nothing wrong with testing it, except it's a waste of time and money. When you test E. coli with a vancomycin disc, it will be resistant. So why waste your time and money to test it? That's an obvious mistake. There are other mistakes that are not obvious. A lot of laboratories test E. coli with amoxicillin. Why did they do that? Because amoxicillin is commonly used to treat E. coli infections. But the CLSI and UCAST do not make amoxicillin breakpoints. So they say test ampicillin, and with the ampicillin, you can predict the amoxicillin result. So many laboratories do test amoxicillin because they think they are supposed to. The clinicians say please test amoxicillin because we use it for our patients. But when you look at the CLSI guidelines or the UCAST guidelines, there are no breakpoints, meaning it's an invalid test. So even though the drug sort of makes sense, it is not validated and not recommended for testing. That's an obvious mistake. It's not an obvious mistake because you do need to know CLSI or UCAST, but it is still a mistake. Another mistake, streptococcus pneumonia. Penicillin is a very important drug for treatment, meningitis from pneumonia. So a lot of people test the penicillin disk, and that's a mistake. It is not a reliable test. If you wanna test streptococcus pneumonia, you're supposed to use a penicillin MIC or the oxicillin disk. And if you know CLSI, you know UCAST, of course you know that, but a lot of people don't. So one of the things I'm looking at here are the testing drugs that make no sense. So I'm looking at the list of antibiotics. And here, I'm happy with this. Pseudomonas, these are all appropriate drugs. I do want you to draw your attention to colistin. So colistin, percent resistance is zero. Percent intermediate is zero. Percent sensitive is zero. Percent question mark is 100%. The reason for that is there are no breakpoints. So this is kind of an invalid test, but it is an important drug for multi-resistant organisms. The problem is that it's a hard to do drug. It's a not reliable drug. So CLSI had breakpoints and then they got rid of the breakpoints because they were unreliable and now they see do MIC testing. So here you see the three categories, RIS. You also see the category question mark. If you get results question mark, it means the test is not approved or it is not valid. It means you're testing the wrong drug or maybe you chose the wrong dispotency. Maybe you made a configuration issue. So I'm looking at percent question mark because percent question mark means they didn't do something right. They didn't do it right either because they don't know exactly what they're doing or because they do know what they're doing and they have their own opinion. For example, Callistan does not have official distribution breakpoints, but if you look at the medical literature, there are some published breakpoints. So people sometimes use unofficial breakpoints, but you should only do that if you know what you're doing and you do know that this is not official, CLSI, not official, you cast. So again, these are the quality issues. Are they testing the right drugs for this organism? Also, I do recommend you can see. Oh, yes, go ahead. Yeah, thank you. This is a lot of, you know, you clearly explain which antihist should be used for a given microorganism. But, you know, this one requires a theoretical background, right? So if I am a theoretical background. What background? You know, if I'm not a microbiologist, I will not understand which antibiotic should be used for a given microorganism. Of course. Yeah, in that case, maybe I don't know how can we do this one, which antibiotic, by, you know, using Hoonit, will it be possible to do? In a few months, yes. So, but I'm going to take a little detour. I'm going to internet right now. And I'm going to do a search for CLSI-free, or CLSI-free. I do not recall. In Ethiopia, are you doing CLSI or you cast? CLSI, where are you doing CLSI? So if I just do a search for CLSI-free, you see free resources, CLSI. And you see the M100 and M60-free. The M100 is the annual update breakpoint tables and the M60 for routine bacteriology. The M60 is the same thing for yeast, fungi. So welcome to CLSI, M100 and M60. Click here to use guest access. And you see there are three documents. The M23 is for quality control, M60 for fungi for yeast, M100 is the normal one. But I'm now in the CLSI document and on the left side of the screen, although I can't see it because they go to mid-end controls. Okay, there is an icon here that says TOC. That is the table of contents. If I go to the table of contents, I do have to move some things on my screen. Okay, good. You see table one, table two, table two has all of the breakpoints, table three, you see a lot of tables here. Table one A, let's look at that. Suggested groups of antibiotics for testing for normal bacteria, not fastidious. Table one B, recommended testing for fastidious organisms. Table one C for anaerobes and that's it. And then it moves on to the two A for the breakpoints. So table one A, okay. So they have divided antibiotic testing into four categories and you should not follow this list explicitly. This list is just a general guidance for the world. But of course, you know, you always want to customize it depending on your local antibiotic needs, your local resistance patterns. So please do not do exactly what they recommend here but please do something like what they do here. So what they are recommending in group A, so group A is always tests, always report. And it's only a few high priority first line drugs. Group B, include antibiotics that may warrant primary testing and always test, but selectively report. So in other words, they recommend that you test everything in group A and everything in group B. But if the bacteria is sensitive to everything, don't tell the doctor the group B drugs, just tell the doctor the group A drugs. That's the idea here is called selective reporting. There's selective testing and selective reporting. So the CLSI's general recommendation is test everything in group A, test everything in group B but report the group B antibiotics if it's important, if it's a resistant organism. Group C is basically what we call supplemental or selective testing. Please consider testing these drugs especially. So maybe on day one, you will test 12 drugs and then if it's resistant to many of those, go ahead and test some of these drugs on day two. Don't test everything every day but please test enough drugs on day one to satisfy most of the clinical needs. If there's a lot of resistance to the day one drugs, then you consider testing things in group C. But here you're seeing they're putting enterococcus, gentamicin, high-level resistance in group C. We disagree, we do that in group A. We feel this is a very important antibiotic for this organism. So that's why I'm saying these principles are important but don't do exactly what they see do here. You do need to adapt it to the national needs. The final category is group U and that's for urine, these are urine drugs, nitroferantoin, phosphomycin, sulfazoxazol. So this table one A, these are for non-fustidious organisms and to bacteriallis, Pseudomonas, stefinenorococcus. And then it continues, same exact thing but as an adiabacter, Berkle, Darius, denitrophomonas and other non-bacteriallis, group A, group B, group C, group U and I think that's it, that is correct. So this is table one A is giving you recommendations for testing the non-fustidious organisms. If I go to table of contents and I go to table one B and I wait, there's always a delay here. There's not that much of a delay. There it is, okay. These are the fustidious organisms. So for example, for homophilus, they recommend testing and reporting ampicillin for first line. They recommend testing ampicillobactam, cephotaxi, and cipro. We disagree with that. A lot of what they put in group B, we put in group A. We're a university hospital. The doctor wants a lot of choices because the laboratory doesn't know how sick the people are. So we do not do a lot of group B. If we test it, we usually report it. So that we have Nicerogonorrhea, strepneumonia, group B strep, I'm sorry, not group B strep but beta hemolytic strep and viridin strep. That is at the top of this table. If I go down further, that might be the end of the table. And then if I go to table of contents, table one C, and I wait, this is what's tested for the anaerobes and they divide it into gram negative anaerobes and gram positive anaerobes. So basically, Hoonet is not smart enough to know this. So if you wanna recommend to laboratories what they should be testing, use these official recommendations. Do not learn microbiology from Hoonet. Learn microbiology from the selicide materials such as that I am showing you here. Having said that, we do plan in the next few months to take these tables and putting them into Hoonet because we would like to make it easier to guide people in the right direction. So right now, Hoonet doesn't know anything about table one A, B or C. Hoonet doesn't know all the break points. Hoonet knows table two but Hoonet at the present time does not know table one, the recommended test practices. But we will put it in and then Hoonet can be smarter and Hoonet can help guide you towards an appropriate set of drugs. So in short, don't learn microbiology from Hoonet. Learn microbiology from the experts. But we will make Hoonet smarter so we can also help you a bit further. Does that help? Yeah, thank you. Maybe another question. Sure. Can you show me the slide? The slide you were presenting. Sure. Before I leave this screen, I do highly recommend to all of the countries that they do establish these kinds of lists in terms of national standardization that has a number of benefits. Maybe there's some simple things. Some labs test ME-Penem, some labs test Meropenem. There's nothing wrong with that. Both are proper drugs. But at the national level, it's just easier if everybody tests the same thing, either ME-Penem or Meropenem, because then you can easily compare the data. That's just a convenience factor. But also I recommend that like E. coli, in my hospital, we test like 12 or 15 drugs. But when people do the test manually, some people do, these small diffusion plates are 90 millimeters. They can usually fit around six drugs. People usually put five to seven drugs. Some people test one plate, that would be six drugs. Other people test two plates, that would be 12 drugs. 12 drugs is also how many you can fit onto a large dis-diffusion plate, that's 150 millimeters. I do recommend you come up with a recommended minimal set for all of your laboratories to be testing. For example, for E. coli, or for staph aureus. For staph aureus, I recommend tests of oxidant disc, test erythromycin disc, test Cypro, test Cochromoxazol, maybe test penicillin. A lot of people don't test penicillin because resistance is so high, but it's certainly a valid drug. So I would make a list of maybe five or six minimum drugs that everybody should test. There are other drugs like Daptomycin, and Daptomycin and Linesalid, the vancomycin MIC, not the vancomycin disc, that's invalid. So I would recommend that you make a national list of minimum testing recommendations, as well as supplemental testing. They might do the supplemental routinely, the more data the better, but I do have to recognize there are costs and convenience aspects, but I think I would recommend you come up with a list of five or six minimal drugs that everybody should be testing, as well as supplemental useful drugs that people might do routinely, or they might do in a supplemental way. This is gonna have two big benefits for you. One is it's really gonna organize laboratories into trying to do things in a common way. So they are testing the appropriate drugs in a way that's gonna be comparable around the country. That's one reason why I think this is a good idea. It also helps you with purchasing. Just purchase a ton of imipenem discs and distribute them to everybody. Or same thing, cephotaxymceftoraxone, they're more or less equivalent. You don't test both. You test cephotaxymceftoraxone. Just easier if everybody is testing the same drug. So these are some benefits to having a minimum agreed national list and an agreed national supplemental list. That's one advantage. The other advantage is I think it's gonna be valuable for you if we think of strategies for scoring. We would like to tell the laboratory, you have entered, you've done a great job of data entry. Data birth, 90% completeness, age, gender, 95% completeness. Data of admission, 10% completeness. So you can help to judge them on how complete their data entry is. In addition, you can also give them a score about how complete the minimum testing is. For example, for Staph aureus, if you tell them, please test these five drugs, at least these five drugs. Then if you have 100 Staph aureus, they should have 500 results. 100 Staph aureus times five antibiotics that should be 500 results. And from those 500 possible results, maybe they have 400, meaning they have 80% completeness in recommended minimal testing. So by coming up with scores, it's gonna be easier for you to give feedback to the labs about how well they are doing. And it will be easier to compare lab one to lab two to lab three. And it's also gonna be easier to monitor over time that maybe in January, they were doing 70% minimum testing completeness. And maybe they get that to 80% to 90%. I want to score them on the minimal testing requirement, erythromycin, suffoxidin, SIPRO. If they're doing Linesilid and Venkamycin-MIC and Daptamycin, great. But I don't wanna penalize them because it's not recommended routine testing. So again, if you come up with a minimal list, this is gonna lead to better quality microbiology, better purchasing, better teaching. And it's also going to allow you to score the labs to give them more appropriate and targeted feedback, okay? How do we people come up with recommended national lists? Well, as a small joke, whenever they discuss that, I leave the room. In Argentina, they have an annual in-person meeting every year and they get in the room to discuss, they do two plates of this diffusion. So they do 12 drugs. So what 12 drugs, and they do them routinely. So what 12 drugs should they do routinely? And then what supplemental drugs could they also do? And every year they get in an argument. I want Livo, I want SIPRO, I want this, I want that. I just leave the room. These are all reasonable choices. I just want them to come to consensus. So, and that's how we do recommend that as the national coordinators, you come up with your recommendations. We think this is a good national list. But then when you get the laboratories involved, they'll often have very good ideas. Did you think about this? Or we're having trouble buying that disk. Can you get, instead of doing several taxing, can you just have Pharaxone? Because we just have a better experience with the vendor. Or there's a new drug. The problem with the new drug is everybody wants to test the new drug, but then you have to get rid of one of the drugs or there's not enough room on the panel. So I do recommend that you sort of discussed nationally what recommended minimal testing would be, but then open it up for discussion with the labs is that also improves the idea of ownership. If they have contributed to it, if they agreed as a group to test these six drugs or to test these 12 drugs, it's easier to say, you said you were gonna do this, but you're not doing it. Instead of just mandating, you must do this. This is especially important when some of the recommendations are not practical, like doing a vancomycin-MIC on every staph aureus. You know, it's just in the United States we do, but it's often not realistic because of the expense and they don't have the machines, they don't have the E-test. Now I'm forgetting what the question, oh, that's right, that's right. I wanted to finish this discussion about the appropriate testing. So I recommend you take the idea of this list, but then you customize it for Ethiopia. And I do recommend that a place like Black Lion Hospital, Black Lion Hospital have a much larger list of routine testing because you have sicker people at the university hospital, whereas a community hospital, six drugs routinely might be enough, but a university hospital maybe 12. So these are kind of customizations and adaptations. So if you can agree on what you would recommend is minimal testing, that would be a good start. And you can also make recommendations for supplemental testing that some people will do supplementally, but other people might do routinely. So that was the end of that discussion about CLSI. I will close the CLSI document and then you asked me about my slide. Let me bring the, I have to move this around and then I move, I bring this back and go back to the slide. Yes, so what is your question? Yeah, thank you for the nice response for my question. Another question maybe regarding this output, the one you shown us. You know, sometimes people, they want to present non-susceptible figures or maybe the percent of non-susceptible with regard to particular microorganisms. So is there any way of merging intermediate with, I mean intermediate with RADISA? Okay, I'm going to leave the PowerPoint for a moment and I'm going to jump ahead to the Hoonet software. Going to the Devoto test hospital that we use for teaching purposes. Here you see the two versions of data analysis, data analysis and quick analysis. I will go to data analysis. You see these questions. I'm going, let me just put in, you know, for example, Staph aureus and data files. Okay. Well, for this one, let me just do the E. coli instead. I'll come back to that later. Okay. And so analysis type. So we've already talked about my slides for isolate listing and summary. We showed you the listing. We discussed the summary. In fact, we did both. The summary that Hoonet is suggesting is organism by month. But the slide that I showed you was not organism by month. The slide that I showed you was location by month. But you can see you can change it however you want. Organism by month, by day, by quarter, by year. So we'll come back to that later. You asked me a question about percent resistance. So the percent resistance analysis is two formats. Percent RIS and test measurements. Okay. And I just lost my mouse. I'm hoping my finger works fine. This happened last time as well. Some problem with my mouse pad. But anyway, I can use my finger. So here, percent RIS and test measurements. You see two options here. There's what we call the detailed report. Percent R, I, and S. There's also the summary report. Okay. And as with your question, the RIS detailed report does not merge R and I. It is what it is. R is R, R, I is I. So the detailed RIS analysis will not merge them. It just presents them in detail as they are. But there's the second option called summary option. Summary option, you have a choice on the right-hand side by default, it's percent susceptible. But you can change it to percent resistant or percent non-susceptible. And that's what you asked about. Or percent non-resistant. The issue, of course, is the intermediate range. And non-susceptible, we group the I's with the R's. And non-resistant, we group the I's with the S's. So does that answer your question? Yes, exactly. So the detailed report does not. But the summary report does. Okay. So let's see, why did we give people different options here? There's a very good reason. The traditional way for clinicians and pharmacy people to look at antibiotic resistance is by percent susceptible. Because they wanna choose a drug with high efficacy. They are looking for a drug that's 95% susceptible, 98% susceptible, 90% susceptible. Some people use 90% as a cutoff, some people use 85% as a cutoff. Of course, there are issues of bias. We could talk about bias later. But for a pharmacist looking for a good drug, the traditional way is to look at percent susceptible. But for epidemiologists and microbiologists looking at emerging resistance, they wanna look at emerging resistance. They don't wanna look at decreasing susceptibility. They wanna look directly at percent resistance. They wanna see is percent resistance emerging over time or percent non-susceptible to lump the I's with the R's. So clinicians and pharmacy people usually wanna look at the percent sensitive. But microbiologists and epidemiologists often wanna look at the percent R or the percent non-susceptible. And because all of these are valid, interesting ways to look at the data, we offer that as an option inside of HUNET. Okay. I'm not gonna go back to the slide. In the upper left-hand corner, it says copy table. So in fact, you can also copy and paste the data to Excel. And then of course in Excel, you could also combine the R's and the I's together in Excel. So that's a different way. So in HUNET, you can combine them together using the susceptible summary feature. But if you just copy and paste the detailed report to Excel, you can add the columns with percent R and the percent I together. So you can manually do the detailed report and then just add in the percent non-susceptible as an Excel calculation. Sure, let me just show that. Let me go here and I'm doing begin analysis. Oh, I did the detailed report. I don't want the detailed report. I'm sorry, the summary report. The summary report, we have one row per organism. The detailed report, we have one row per antibiotic. So obviously there's more detail when you're doing one row per antibiotic. So here I'm in Excel with the detailed report. And I'm just gonna insert a new column here, column percent non-susceptible. I'm just gonna say equals this, plus this. You wanna copy this all the way down. So here you do have the percent non-susceptible. So if you do want the detailed report, you can calculate the percent non-susceptible yourself. I have one other comment about non-susceptible. The reality is non-susceptible means different things in different contexts. The way I just described non-susceptible to you was percent R plus percent I. I have already closed my CLSI document, but there was also a different category called non-susceptible. For certain bacteria like homophilus influenza and ceptraxone, resistance is extremely rare. When resistance is extremely rare, CLSI makes what we call a susceptible breakpoint. For example, 20 millimeters and larger is sensitive. But if it's smaller than 20, there's not enough clinical data. There are no isolates to say if it is R or if it is I or low-susceptible. So one meaning of non-susceptible is to add R and I together. That's one meaning. A different meaning for non-susceptible is for some drug bug combinations where resistant isolates are so rare that CLSI does make S breakpoints, but they say if it's not S, we really don't know what it is. We're just gonna call it non-susceptible. So if you see homophilus influenza with ceptraxone, you will never see, at least with this diffusion. No, it's anything with MIC. With homophilus influenza and ceptraxone, you will not see the percent resistant. You will see percent truly susceptible and the percent non-susceptible, meaning anything which is not definitely susceptible. So NS actually has two different meanings depending on which organism and what context you are looking at. Okay, great. Okay, so let's see. I'm leaving the live demonstration. I'm going back to the PowerPoint. So let's see, I discussed here. The last thing I discussed here was the Calistin, about 100% question mark because there are no breakpoints. If you want breakpoints, you have to manually put it in yourself, hopefully from a respectable publication or from the vendor. Sometimes the vendors provide information. There are, so in the United States, a lot of people don't know this, but we have two sets of breakpoints. We have the CLSI breakpoints. We also have FDA breakpoints. They try to keep them aligned, but their speed is different. So there are a few cases where FDA has breakpoints, but CLSI does not. It usually means CLSI will in the future, but sometimes the FDA makes some breakpoints, preliminary breakpoints first. So what else am I looking at? Again, I'm focusing not yet on epidemiology. I'm focusing on test practices. So they have 356 pseudomonas aeruginosa. How many times did they test on the case? They tested 336 times. As Trinium, 336, Calystin, 336. They're very consistent. If they have a pseudomonas, they test certain drugs all of the time, but they only test itself at PIM twice. At the bottom of the screen, they tested, I think that's ampicillin. They tested it once. So there are certain drugs that they always test. That's basically the CLSI category A category B, always test. But they only tested stuff at PIM twice. Why did they do that? One, it could be a typing mistake. They accidentally put stuff at PIM when they didn't test it. Somebody just made a typing mistake. There's one reason. Another reason is that this was so resistant to the normal drugs, they did some supplemental testing. They said, you know, this is a, staff or MRSA resistant to almost everything. Let me test Daptomycin. I'm not gonna test Daptomycin routinely. It's not usually needed, but sometimes it is needed. So I do supplemental testing. That's what CLSI calls category C, supplemental testing. Or like a vancomycin MIC on a staph aureus. Or sometimes a vendor comes along, they give you a couple of free discs. You test them out, you run out. So when I'm looking at, as I'm looking at how systematic they are, what do they test and do they test them regularly? Here I see two from CFP, I'm kind of guessing it was a multi-resistant organism that needed a couple of more choices for two patients in the ICU maybe. Sometimes people also do urine and non-urine. So for example, maybe they test nitroferentron in every urine, but they don't test it in blood. So again, this is selective testing for certain specimen types you do it and certain other specimen types you don't. So by looking at the number tested column, I get a sense, do they always test the same drugs or do they test certain drugs some of the times, but not other times? Unfortunately in a lot of low-resourced countries, sometimes those number differ because they run out of discs. Amikation 336, as Trinam 336, Cipro 200. And I asked them, why do you have Cipro only 200? Ah, we ran out of discs. It's a reason, it's an unfortunate reason. Or sometimes they'll say, well for the first half of the year we did Cipro, the second half of the year we used Levo, we changed in the middle of the year. So there's a lot you can learn by that column called number tested. How often do they test these different antibiotics? Do they run out of discs? Do they do first line testing, second line testing? Good, that's how often they test. So it's allowing me to critique their ability to do testing. I can go back to them, please test more drugs, please be more systematic, please order the discs on time, you're not running out of discs. Then I start looking at the percent RIS results from a quality perspective. For example, for Staph aureus, what percentage are vancomycin resistant? Because if the number is not zero, somebody made a mistake. In the world, there are almost no cases of VRSA. So I'm looking for mistakes, other mistakes. 15 years ago, imipenem resistant E. coli was very rare. But some people had 5%, 10% resistant, but it was because of a mistake. Imipenem discs are not stable in a hot tropical humid environment. So even though the disc says imipenem, there's no imipenem in it because the imipenem has degraded. So I'm looking for certain drug bug combinations that don't look right. Or other things for quality, Clepsiola pneumonia. I saw one lab that Clepsiola pneumonia was ampicillin sensitive 20%. That doesn't make sense. Clepsiola pneumonia should be intrinsically resistant. So Clepsiola pneumonia is usually 90% resistant, 95% resistant, 98% resistant. It's usually 95 or 98 or 100% resistant. So when I saw a lab that was Clepsiola pneumonia, 80% resistant, there's some mistake there. Either it is not Clepsiola pneumonia or it is not ampicillin resistant. Or it is ampicillin resistant, I mean to say. So again, I'm always focusing on these quality issues first because there's no sense in talking about epidemiology if you can't believe the data. Finally, we get to epidemiology and that's where I look for percent MRSA, percent CRE, percent ESBL possible producers. I'm looking for my key drug bug combinations. I'm looking at my reserve agents. Amication resistance should be very low. In the example I'm showing you here, 3% are resistant, but 4% are intermediate and that's important to monitor. So I'm looking at this from a quality perspective. One comment about the intermediate range. Intermediate range is usually very small, 2%, 3%, 6%. That's because the intermediate range is usually only two or three millimeters. If I see a lot of bacteria in the intermediate range, it suggests one of two things. There's some mistake. Sometimes if the disk diffusion, if the medium is not the right thickness, a lot of sensitive bacteria will be incorrectly called intermediate. And what you will see is a high percent intermediate. But sometimes it is true, especially if you have a wide range. So look here at this tree in NAM, but look at sephotaxim. Well, look at amication. The intermediate range is 15, 16. It's only two millimeters. Sceptazidine, 15, 16, 17 is three millimeters. Levofloxacin, imipenem, each of these intermediate ranges is only two or three millimeters. When the intermediate range is only two or three millimeters, the percent I is gonna be small in most cases. But look at the intermediate range for sephotaxim. It's 15 to 22. That's very wide. So my percent I is very high. And these bacteria probably are ESBL producers. Let's see, Cipro is 16 to 20. So that's also a wide range. So the percent I is 6, 8, 6%. Whenever the, an astrionam, the intermediate range is 16 to 21. So whenever the intermediate range is above 5%, I just kind of question whether it's expected or unexpected. For example, when the intermediate range is wide, it's kind of expected. But when the intermediate range is very narrow, then the, if a high percent I is there, I do worry about quality issues. This is sort of refinement. I do not expect everybody to do the same level of checking I do. But since I'm talking, I tell you the way I look at data. I don't recommend you follow everything that I do. It's just too many rules. But I want to put in these rules into Hunan. I'm sorry, what was the question? Yes? Okay. Yeah. You know, most of the antibiotics are tested equal number of times. Again, it's heteronomous originesis. But in our case, because of some issues like the stock issue or any other issues, we are not testing equal number of times. So in that case, interpret. Sorry? That's exactly right. Yes. If you don't have equal number of times that the drug is tested against a particular microorganism, is it good to interpret the person resistant or person susceptible? Because here you have equal number of times. Yes. Yes. That's an excellent question. And I'm going to get into a broader discussion about bias. You're familiar with the idea of bias. So let's see, but I'll come back to that. So the answer to this simple answer to that question is that if they always test the same drugs the same number of times, the interpretation of the data is easier. If they do not test the data the same number of times, the simple answer to your question is that the interpretation is problematic. In short, it kind of depends on why they did it. And I'll get some examples. So for example, cephapim, they only tested it twice. And they only tested it twice. My guess is that they tested it in a multi-resistant strain. So there are two problems with the cephapim result here. It's only two times. The CLSA recommendation is at least 30. So one problem is that the number here is very small. That's not the only problem. The problem is it's very biased. They did not test two random isolates for cephapim. They probably tested two multi-resistant strains. So the answer you get is not representative. Okay, in this graph, it's 100% sensitive representing those two isolates. But I don't wanna say it represents everybody. It only represented two strange isolates. I'm going to give an extreme, and this is because of selective testing of multi-resistant strains. I'm gonna give an example for my hospital in 1990 with imipenem and Klepsiella. At that time, imipenem was a new drug. And the pharmacy asked me, John, how are we doing with imipenem? It's a new drug and Klepsiella pneumonia. And I looked at the data and I said, oh, yeah, it's 20% resistant. And there's a 20% resistant. How can it be 20% resistant? It's a new drug. I said, hold on a second. I didn't tell you the full story. We had about 800 Klepsiellas, but we don't routinely test imipenem. It's a new drug. It's not on the panel. We only tested against multi-resistant strains. So even though we had 800 Klepsiellas, we only tested imipenem five times. Out of those five times, it was resistant once. So one resistant out of five is 20%. So the number was correct, but it was not representative. It was not meaningful. Resistant one time out of five is 20%. But it wasn't resistant one time out of five. It was really resistant one time out of 800. But we didn't test it the 800 times. So when the number is small, like I told you here, two or five, if it's because you are testing the multi-resistant strains, you have incorporated a very significant bias and the results are not reliable. But I mentioned a different example. Maybe you tested, let's take E. coli. Maybe you tested ampicillin 100 times and Cipro 100 times and Cochromoxazole 100 times. I'm happy. Maybe you only tested nitroferantoin 40 times. That's okay, because that's on the urine. So they did not test it on the blood. They did not test it on the cerebrospinal fluid. They tested it on the urine. So in that case, they only tested it 40% of the time, but they tested on 100% of the urine. So in this case, it doesn't bother me because it is representative for what they did it for. So in this case, they did not always test it, but they did test it for every urine, meaning that it is reliable and useful and meaningful and representative for the urines. So in short, even though they did not always test it, in this example, it doesn't bother me because I don't want it for everything. I want it for the urines. So this is an example where there is not a bias. It's just a selective testing or it's selective comprehensive testing for certain subset. Okay, I'll give a third example. If in the first half of the year, they test Cipro and the second half of the year, they test Levo, there's not necessarily a bias. If resistance did not change, if you find that Cipro resistance is the beginning of the year is 40% and the end of the year, the Levo resistant is 38%, it's probably just the same thing. The two drugs are so similar for an E. coli that so the numbers are not the same, but they just switch from something to something very similar. Same thing, imipenem to meropenem. So there's no bias, the two drugs are very similar. If you see that Cipro resistant at the beginning of the year is 20% and Levo resistant at the end of the year is 30%, I don't think that's a bias. I think that resistance really did grow. So that's not a bias. The two drugs are so similar. So if they're continuing to test the Cipro, if they continue to test the Levo in the same way that they tested the Cipro, the two drugs are so similar, I can still look at the trends. So in answer to your question, it's easiest if you always test the same drugs because then you don't have to worry about these testing issues. On the other hand, sometimes the numbers are not the same and then it's problematic, but it doesn't mean it's wrong. You just wanna understand why are the numbers inconsistent? If it's inconsistent because of stock issues, don't introduce a bias. It just means you have missing data. If I did Cipro in January and February and June and July and November, December, I'm gonna get the correct national average for the year even though I'm missing data. So this is an example where you have missing data which is not introduced a bias. But if you only test the ICU multi-resistant strains with second line testing, that's the class C I described earlier, then this is gonna introduce a bias. I see this commonly with amiccation. A lot of people do not test amiccation routinely. They only test it on the multi-resistant strain. Generally, amiccation is a much better drug than genomycin. So usually the percent resistant for genomycin is much higher than the percent resistant for amiccation. I'll show you that here. So here at genomycin, it is 18% resistant and amiccation is 3% resistant. That makes sense to me. But if I see genomycin resistant is 18% but amiccation resistant is 30%, I wanna check the number tested because if they only tested amiccation a small number of times, there's probably just a bias. It's not that the number is wrong, but the number is not representative. So you ask me a simple question and the answer is it kind of depends. You want consistent testing because then you'll get the easiest data to interpret. But if the numbers are not consistent, you wanna understand why are the numbers not consistent. If the numbers are not consistent because of stock issues, that does not introduce a bias. It's just that you have missing data, okay? Thank you. Sure, okay, great. We've spent a lot of time now talking about the top of the screen. Now let's talk about the bottom of the screen. The top of the screen is about the percent R-I-S. The bottom of the screen is about the zone diameter zones of inhibition or the MIC values. It's about the measurements. So if you record the measurements, Hoonet will allow you to do the bottom of the screen. However, if you just type R-I-N-S into Hoonet, the top table is fine. Hoonet is no trouble with R-I-N-S data, but it doesn't allow you to do the zone diameter distribution because it does not have the zone diameters. So almost all of the Hoonet analyses work with R's, I's and S's, but there are a few where it does use the zone diameters or the MIC values. And that's what we see here. So at the bottom, you see the red lines in the middle of the screen. Those are the break points. So if I look at the table, it says gentamicin intermediate 13 to 14. And that's where I see the red lines in the graph. Everything with the zone diameter larger than 14 is susceptible. Everything with the zone diameter lower than 13 is resistant. Everything between 13 and 14 is intermediate. So this allows me to see graphically, separately the resistant to the left, intermediate in the middle, and the sensitive bacteria to the right in the graph. Why do I like the zone diameters? I'm going to give you five reasons I like the zone diameters. Reason number one and the most important reason is it's the correct way to do the tests. A lot of people, if they do not take out a ruler and they do not measure, they use what we call the eyeball method. They take the disc in their hand, they look at their eye and they say, oh, I think that's resistant. Or they show it to the person next to them. Do you think this is resistant? No, no, I think it's sensitive, but they don't get out a ruler. So unfortunately, this eyeball method is very common. If the zone diameter is less than 10, they're probably going to be correct. If the zone diameter is above 30, they're probably going to be correct. But if the zone diameter is 18, you don't know. 18 can be resistant, can be intermediate, can be sensitive, and nobody knows all the zones and the organisms and antibiotics, so nobody can eyeball reliably all these zone diameters. For high-level resistance, you will generally be correct, but a lot of important resistance is low-level resistance, and there is no way to measure low-level resistance accurately with your eye. So the most important reason to measure is because there's a sick person and a doctor waiting for results, and you want to give that person the correct answer. The patient had a sample taken, a simple sample like a urine or a blood or a CSF. If they went through the time and energy and pain and effort and the doctor sent it to the lab and somebody's going to have to pay for this test, it's the laboratory's job to give the doctor the correct result. Do good quality testing and part of that is measure and use the breakpoint tables to give the doctor the correct interpretation with reason number one. Reason number two, sometimes these breakpoints change. Doesn't happen often, but this happened with some important examples. The imipendent breakpoints changed in June of 2010. Why June? Because they used to just do it in January. They changed them in June of 2010 because CLSR realized they were wrong and they were afraid people were dying because they were afraid the labs were telling the doctor sensitive when the bacteria was really resistant. So breakpoint did not change very often, but when they change, it's an important change. Some examples, imipendent resistance, sephotaxymes have traction resistance change for streptococcus pneumonia and for ESPL also. Vencomycin breakpoints have changed a number of times. Recently, all of the quinolone breakpoints, the fluoroquinolone change breakpoint change for E. coli. So if you have the measurements, it allows you to compare your old data and your new data. So if I have data from 2010 with zone diameters, I can just repeat my analysis using today's breakpoints. So with the measurements, it allows you to compare your old data and your new data using the new breakpoints. It's not that the test method changed, but the test method is the same method, but the interpretation has changed. So one CLSR recommendation is if you analyze 2010 data today, do not use the 2010 breakpoints. Use the 2020 breakpoints because the 2020 breakpoints are more accurate than the 2010 breakpoints. So if you have the measurements, you have the flexibility to do this. But if somebody simply goes to who in a data entry and types R's, I's, and S's, you cannot reinterpret the data. You'll just have to say, here are my data from 2010, here are my data from today, but I can't strictly compare them because the breakpoints have changed. So reason number one, I like breakpoints. It's the correct way to get the doctor the correct result. Reason number two is to compare data over time if the breakpoints change. Reason number three is from a quality perspective. You know, the zone diameters for the sensitive bacteria in China and Argentina and Greece and Ethiopia should all be the same. You know, the sensitive bacteria around the world, the resistant bacteria are different. The sensitive bacteria are basically the same in terms of their phenotypes. So if I see the zone diameter shifting, it suggests a quality issue. If I see a lot of intermediates, it might mean that the sensitive bacteria are incorrectly in the intermediate and resistant range. This might happen if the medium is too thick. It might happen if the discotency is too great. It might happen if the inoculum is too heavy. So reason number three, I like measurements is just as a quality control check. If you look at my sensitive range there, it's in between 16 and 22. It is a nice normal distribution. That is what good quality data look like. But if you have media and reagents and inoculum and reading that are all over the charts, you're not gonna have a nice, beautiful normal distribution like this. So reason number three, I like measurements is it allows us to a better job of assessing data quality. Reason number four, I like measurements is for epidemiology reasons. So here in this graph, in the far left, you see bacteria that are highly resistant. Those have a zone diameter of six. And then you see a few bacteria zone diameters of 10 and 12. So those are low-level resistant. So here I have high-level resistance and low-level resistance. My first comment is the doctor doesn't really care. If it's high-level resistant or low-level resistant, I'm not gonna use the drug. I'm gonna choose a different drug where it's sensitive. So from a clinical perspective, it doesn't really matter if it's high-level, low-level resistance, we're gonna avoid the drug. But it can be very valuable from an infection control perspective. If I have two patients in the same room with pseudomonas, with genomicin resistance strain, I'm a little bit worried that patient A infected patient B, or a healthcare worker or the environment infected both of them. So if I have two patients in the same room, both patients of pseudomonas, genomicin resistant, the zone diameters can help to reassure me whether or not they might be linked. So if I see that one of the patients, the zone diameter was six, and the other patient had a zone diameter of 12, it suggests to me it's just a coincidence. They seem to be two different bacteria. On the other hand, if one of the bacteria has 10 millimeters, the other patient has 12 millimeters, I'm very worried that patient A infected patient B. So the measurements are a very good way to get a sense of who has the same strain. The best way to know if two patients have the same strain is with molecular typing, do sequencing, PCR, MLST, and that'll tell you definitively if this is the same bacteria, but that takes time, money, equipment, expertise, and it's just not worth the effort for routine use, but you do have the zone diameters. So if you have two patients with pseudomonas and the zone diameters are very, very similar, it suggests a transmission, possible small outbreak. But if the zone diameters are different, it's probably just an unrelated bacteria. So that's reason number four, I like the measurements. It allows me to distinguish high-level resistance and low-level resistance, very susceptible, decrease susceptible. So it helps you to look at the molecular epidemiology, clone A, clone B, clone C. These are not defined using sequence typing, but they're definitive clones. They look like the same thing, and that's very valuable for outbreak detection. The fifth reason I like the measurements is a new reason, or new, not new, but it's new in importance. There's this very important issue of one health. Integrated human, animal, food, environmental sampling. And there are different breakpoints for humans and for cats and for horses. In fact, there are different breakpoints for young horses and adult horses. They're different breakpoints for fish. They all have different breakpoints. So if you want to compare data across systems, you kind of need the measurements, because I can't really compare the dog breakpoints with a human, I can't compare statistics prepared with my dog breakpoints to the statistics with the human breakpoints. So if you want to do one health surveillance, you kind of need the measurements. I want to get to an important issue that it's not obvious. A lot of people think that the purpose of CLSI in UCAST is to find resistance. Ironically, that is not exactly true. You think we do this, we want to do this because we're looking for resistance. That's not exactly true. The reason CLSI makes these breakpoints is to predict clinical outcome, which is a slightly different thing. There are many examples of bacteria that are a little bit resistant, but the patient is still going to get better. For example, if it's a urine isophthalate and antibiotics are concentrated in the urine, if it's in respiratory. So there are many, many cases where a little bit of resistant doesn't mean the patient will fail therapy. It's what we distinguish, we call it on the one hand clinical resistance and microbiological resistance. So when people say, why do you have different breakpoints for horses and swine and cattle? And the answer is not because the bacteria have different levels of resistance, it's the same bacteria. They have the same degree of resistance. The question is whether or not the animal would get better or the human would get better. So let's see, this might be a confusing concept, but I'll mention it if there's questions, I'm happy to repeat it. But the reason that an adult horse has different breakpoints than a young horse, which is a strange situation, but it is true, is that a young horse, a little resistance is more likely to kill the horse. For an adult horse, a little bit of resistance, the horse is probably gonna be fine, because the adult horse has better immune system, antibodies, previous exposures. So the goal of these breakpoints, so I'm gonna take the example here of amiccasin. So here you see 16 millimeters is intermediate, 17 millimeters, precisely speaking, means the patient would probably get better. I wanna draw your attention to the graph. So here we have genomycin, you do see there are some bacteria down at the measurement 15. This is a fourth group of bacteria. I see very sensitive resistant and commuted. There are some bacteria at 15. My guess is that these bacteria are a little bit resistant. According to CLSI, they're still gonna get better. So this graph is a good example. I have three gray circles. The gray circle to the far right is truly susceptible. The gray circle just to the left of it is decreased susceptibility. It's kind of susceptible, the patient should get better, but it is a little bit resistant. And that's why cattle and fish and horses that have different breakpoints because we're trying to predict whether the patient would get better or not. So if you wanna do one health surveillance, you kinda need the measurements for comparability. If you're taking care of a sick cow, use the cow breakpoints. But if you wanna compare the salmonella from a cow to the salmonella from a human, in that case, it makes more sense to use the human breakpoints. So I'm gonna repeat. I said that I like measurements for five reasons. Number one, it's the correct way to do the tests. Number two, these breakpoints can and do change over time. And I wanna compare my old and my new results. Three, it assists us with data quality assessment. Four, it allows us to look at molecular epidemiology, high-level resistance, low-level resistance, outbreak possibility. And number five, in a one health context, you kinda need the measurements so that you can compare data across the different systems. This is a long presentation about this one issue. Are there any questions? Question, John. In one health surveillance, how can we see the trend of transmission from human to animal by just interpreting the zone diameter? That's the one you are discussing now. Okay, sure. Well, okay, there are a few ways to answer that. For example, one way is just to, if I do have this graph here, randomized and for humans, I could do that exact same graph for cows and sheep and dogs and fish. And I just wanna see how similar and how different they are. What I suspect is that the sensitive bacteria are gonna be in the same place. Sensitive people or sensitive people in Manila are basically the same all over the world. You know, these bacteria have had millions of years to spread around and be sensitive. So sensitive bacteria are very similar. The sensitive bacteria might be very common in cows and sheep and very rare in humans. So the percent sensitive might be very different, but the location of the sensitive bacteria, the millimeter should be basically the same. For humans, my sensitive bacteria are in this example, you know, there was kind of between 16 and 23. For cattle and sheep, the sensitive bacteria should all be between 16 and 23. So basically, I suspect the sensitive bacteria will be very similar, irrespective of the animal or human. But the resistant bacteria might be very different. So here we have high level resistance in the human, but maybe you do not have high level resistance in the animals. So by using the zone diameter, it helps you to look at, do they have the same resistance or a different resistance? I can do that using one drug. Later, we're going to talk about an analysis called the resistance profile analysis. The resistance profile analysis is looking at multi-drug resistance. And this is a very good way to look at the phenotypic relationship. Of course, the best way to know if humans and animals have the same salmonella is molecular typing. PFGE, pulse field gel electrophoresis, or increasingly whole genome sequencing or MLST. So the best way to know is molecular testing, but we do not routinely have those data, but we do have the phenotypic data. So if you see a salmonella, let's assume we have an outbreak in humans, and the salmonella in the outbreak is salmonella-derbin, resistant to ampicillin, cotrimoxazole, and Cipro. I can then look at salmonella-derbin from cows and sheep and pigs, and I can see who has the same kind of salmonella. I've looked for salmonella-derbin resistant to these three drugs, but sensitive to the other ones. So the human one is the one that has the outbreak. And I'm trying to figure out which food, which meat, which animal has the same thing as the human does. If I have the time and money, I'll do molecular typing. If I don't have the time and the money, or even if I do, I still want to start. I mean, even if I have the time and money, it still takes time and money, but I can still start with the phenotype. So on the human case, if I see salmonella-derbin, not derbin, but I see salmonella-derbin resistant to three drugs, I want to see which part of the country, which food, which animals have salmonella-derbin that phenotypically look the same. That's going to give me a quick heads up as to the most likely culprit. Once I've identified the most likely culprit, then I can do my molecular typing to do that confirmation. So the best answer to your question is molecular typing, but it's not a practical answer routinely and quickly. So by using these own diameters and the multi-drug resistance phenotype, that's going to do a good job of helping to find possible outbreak connections. We do the same thing in the human thing in the hospital. If there's an outbreak of pseudomonas-resistant to five drugs, I look for other people with pseudomonas-resistant to five drugs. I ignore the pseudomonas-resistant to three, I ignore the pseudomonas-resistant to seven. So if we look at the multi-drug resistant phenotype, it helps you to draw these connections. And you will see, we'll probably spend a lot of time talking about the analysis in a few more slides. Other questions? Just doing a quick time check, okay? We have about half an hour more. Okay, we've spent a lot of time on this slide and I'm going to spend more time on it. We talked about this issue of biases in testing. If you only test imipenem on multi-resistant strains, you'll get reliable results for those strains, but they're not representative of everybody. That's a bias because of selective testing. There's a much bigger source of bias, which is biases in sampling. So this is a very big issue, especially in low-resourced countries. I'm going to take a very common example, the most common example of women without patient urinary tract infections. It's the most common infection, routine bacterial infection around. So if you can imagine 100 women with a urinary tract infection, do all women get a microbiology sample? The answer to that is no. Even in a high-resourced place, not every woman with a UTI, your urinary tract infection, gets a microbiology sample. Why? Because some women will get over their urinary tract infection without treatment. You know, sometimes they'll take cranberry juice, they'll do other things. They'll just treat themselves without an antibiotic. Or they'll treat themselves with some antibiotic they have left over from previous treatment. Or they'll get some antibiotic from another family member. So a lot of women will get over their urinary tract infection because they treat themselves at home. Other women will go to the pharmacy or they'll go to the hospital, to the medical clinic, to the emergency room. And in many cases, the doctor will give them an antibiotic, the pharmacy will give them an antibiotic and the patient will get better, but nobody took a sample. Why didn't they take a sample? Well, who's going to pay for it? And is the woman going to come back? It's going to take two or three days to get the result. So if you have a hundred women with urinary tract infection, maybe half of them just treat themselves at home. Maybe 30 of them go to a pharmacy and a clinic and are given an antibiotic without a sample. Other women will take an antibiotic and they won't get better. If they don't get better, they might go back to the hospital a second time. And maybe the second time the doctor will give them a different antibiotic. And then maybe eventually somebody's going to take a sample. So if you can imagine, in Ethiopia, if 100 women have a urinary tract infection, how many do you think are actually going to have a microbiology sample? And I suspect it's going to be a very small number because people do not have access to the labs, they can't afford the tests, or the doctor is there, but the doctor says, you know, I think you're going to be fine with Cochrane Moxizal. If this doesn't work, come back and when you come back, we'll take a sample. So if you have 100 women with a urine sample, well, I'm sorry. So 100 women with a urinary tract infection, maybe only five or 10 are going to have a sample. Who are the women who are going to get the sample? They're going to be women with a urinary tract treatment failures. They took ampicillin, they didn't get better. They took Cochrane Moxizal, they didn't get better. They took something else, they didn't get better. Finally, somebody takes a sample. Or maybe it's a patient with a complicated medical history. It's a patient with oncology, a patient who was discharged last week, or a patient who has money. They say, I've got the money, please do the test. They go to a private clinic and they said, please do the test for me. So the people who get sample provide valuable information for those people, but that result is not representative of everybody. And this can lead to very extreme biases. In India, they did one example like this. They just looked at urinary tract infections. They looked at women, they went door to door to door. They asked, do you have the urinary tract infection? So they got a purely random sample. That's great, but it's very time consuming. It's not sustainable. They only did it because it was a special research study. So they did get a completely unbiased sample. And then they compared that to the women who were coming to the clinic and the women who were coming to the clinic who actually had a urine sample. And the resistance rates were much, much higher in the biased samples of women coming to the clinic to be sampled. So this issue of bias is a very big concern for epidemiologists. They said, we can't use the data. We can't use the data because they're not representative. The reality is it's the data that we got. You gotta use these data because we don't have other data. We don't have time and money and resources to be doing random community samples all of the time. We have the data in front of us, but the data in front of us still have a lot of values for a number of reasons. If resistance is low, if they come to the clinic and resistance is only 5%, that 5% might be very biased, but the true resistance is even lower than that. So the bias is still allowing you to see that good antibiotics are still good antibiotics or even better in an unbiased sample. It allows you to look at trends. If resistance went from five to 10 to 20% resistant, it's getting worse. It might be biased, but it's still getting whatever the biases are, the problem is getting worse. There's another consideration is I don't need to, if I'm trying to come up with a treatment guideline, well, before I say that, it also allows you to do comparative testing. If resistance to drug one is 20%, resistance to drug two is 10%, drug two is better. Even if it's biased, it's still a better drug. So these are things that you can do with biased data. I also wanna make another important distinction is that it is not the doctor's job to treat all women out in the community. It's a lot of those women, they're not going to come in, they'll treat themselves successfully. It is the doctor's job to successfully treat the women coming to the clinic. The women coming to the clinic are not representative of all women, but the doctor's not responsible for them. The doctor's responsible for treating the women coming to see the doctor in the clinic. So even though the women coming to the clinic are not representative of everybody, they are representative of the people that the doctor is immediately responsible for. Same thing with hospital infections. If you do percent resistance in your hospital population, it's not representative of everybody in the community, but it is representative of your hospital population and that's important. Because the people in the hospital setting are the people that you are seeing, you're making decisions for them. They're sicker, more complicated medical histories. So even, so I would not necessarily want to use my treatment guideline for the hospital patients to be the same treatment guideline I give for outpatient urine infections because I know the hospital data are biased, but the biased hospital data are very valuable for hospital treatment decisions. But I didn't necessarily want to use that as my national treatment guideline. You may be in the hospital, it is appropriate to use gender miscellaneous first-line therapy, but it's not appropriate under the community because it's more expensive, it's more toxic, it's an intravenous drug. So we're going to come back to lots of this issue of bias. There are certain problems in laboratory test quality. There's a straight way forward. We're going to do good quality testing. So bad quality testing, bad quality microbiology, there's a clear way forward. We do quality control testing, but for sampling biases, there's no easy solution for that except just doing a random study, a random sample. You need to be aware of these biases. So I'm looking at these data. My first row, amication is about 93% sensitive. I think the true percent sensitive is even higher than that because in most cases, the biases lead to higher levels of resistance. I'm sorry, I got that. The percent sensitive here is 93%. The percentage is 93% sensitive. I think the true percent sensitive is a bit lower than that. Maybe 90%, maybe 85%. I don't know how much lower because I don't know how biased the data are. So everything you do, keep in mind that the biases are hard to avoid, but you do need to include them when you interpret the data. Even when you have biases, you can still find outbreaks. If normally you have five Klebsiellas in a month and now you have 20 Klebsiellas in a month, something has changed. So even with biased data, it's still extremely valuable for outbreak detection. Okay. WTO has the program called Glass. I don't really worry about the biases for Glass. You know, Glass, it's about, it's what I call surveillance for advocacy. Geneva is not looking for outbreaks. Geneva is not making treatment guidelines. Geneva is not trying to make definitive statements about resistance in Ethiopia. You know, Geneva is trying to make some general global statements. Resistance is bad. Countries have different problems. The problems are getting worse over time. So if you are doing surveillance for advocacy, I don't worry so much about the biases because you're not making, whether resistance is 25% resistant or 38% resistant, it's not gonna change for advocacy purposes what you do with the data. The goals of advocacy are awareness raising, education, fundraising, gap identification. So for surveillance for advocacy, I don't really worry about the biases. This was a big issue. This is a big issue that there was one case in Europe where there were more cases of MRSA in Sweden than in Romania. That's not true. But the thing is that Sweden has a very active dynamic program for blood cultures. They had a program for blood, but Romania was doing very few blood cultures. So their numbers were artificially low. So for advocacy, I don't really care so much, but I do care about the biases a lot if I'm trying to make treatment recommendations. I do not want to tell the doctor to use any pen them. If like, if Cipro is 20% resistant, that sounds bad, but 20% resistance in a biased sample might only be 10% resistant in an unbiased sample. So if you are going to use these data to make treatment guidelines, these biases are very important. Just because the labs is 20% resistant, it is not really 20% of what's in the lab is resistant, but 20% of what's in the community, it might not be 10% in the community. So the biases are important to remember. They do impact treatment recommendations. They don't impact advocacy so much. They don't impact outbreak detection so much. So the biases are often important and often they're less important. So everything you look at here is only what reached the lab. You do have to think about what didn't make it to the lab. Okay, any more questions? Otherwise, I will finally continue with the next slide. Yeah, I have one question, John. Yes. So one of our major problem in our current MR surveillance is we only have sampling from hospital community. Yes. So many people advised us to include also from community population. How do you advise us to improve these bias by adding some either under sampling or whatever you suggest to improve the biases? Still, this is one of our, you know, bias in our MR surveillance reporting. Yes, that's a very important question. I'll answer it in a few ways. One is what is sustainable and what requires extra money and resources. So I recommend make maximum use of the routine data because every day of the next 20 years you will have your routine data. So let's try to maximize the benefit and use an interpretation of the routine data keeping in mind its limitations. But sometimes the routine data just had these big problems and that's where you wanna do a study. But you can't do a study all the time on every organism and every clinical scenario. So if money does become available, you wanna be selective about it in terms of the priority needs. Maybe focus on things like sepsis, this is particularly severe diseases. Or focus on urinary tract infections not because of its severity, but because of how common it is. So I don't wanna make recommendations for treating outpatient urinary tract infections in Ethiopia using the hospital data. I'm gonna make a statement that's a little, it's not controversial, but it's not what you would expect. Generally you would think that doing surveillance of resistance is a good thing. However, there is a risk. And the risk comes in if you do not keep track of these biases because if you find that resistance, let's just take the case of outpatient urinary tract infections. If Cipro resistance in the lab data is 20%, you might say, oh, that's terrible. I'm not gonna use Cipro for my outpatients. And instead I'm gonna use imipenem because imipenem is 98% sensitive. And this is a bad thing because Cipro probably is effective. In the unbiased sample, it's probably only five or 10% resistant. So one risk of using resistance data, if you do not keep track of the biases is you may go to imipenem and genomicin and imication earlier than you should. So the resistance in these first line agents like Cotrimoxazole and Cipro and erythro are going to be artifactually high in the hospital data but that does not mean that they are inappropriate drugs in the community setting. So there is the risk to AMR surveillance is that you may incorrectly tell people to use more expensive, more toxic in a few cases like genomicin, more expensive reserve agents before you have to. So there is a risk in doing AMR surveillance is that you may make inappropriate recommendations. You're telling people don't use Cipro because resistance is too high but Cipro resistance is not that high in the general community. So that is a risk when you use resistance data to guide treatment recommendations. And I do recommend in all of these settings where people are starting these resistance surveillance activities to be very careful about using these data for treatment guidelines. Use it for outbreak detection, use it for quality improvement, use it for the hospital infections, use it for advocacy. If you're going to use the data for treatment guidelines which of course everybody wants to do you need to be specially attentive to these bias issues. So those are some initial comments. So what can you do? Well, I mentioned two kinds of surveys that you could do. One survey is go knocking door to door to door. That's just not realist. It's nice to do that as a study. Maybe do that once and then maybe repeat it every five years. But I did mention something else which is more relevant to the clinical decisions. You like if a hundred women come to the medical clinic under normal sacrum stances you don't take a urine sample from all hundred women with UTIs. You will take a urine sample if the patient has failed previous therapy if they're recently discharged, if they're a high-risk patient like oncology. So if a hundred women come to your hospital clinic most doctors do not take samples from all women. If it's an uncomplicated UTI they'll just give a normal antibiotic. And you'll sort of the hundred women normally you'll take maybe 10 samples. But one thing you can do is you can do a random, what you can do is a random sample maybe one month a year. Maybe this month we're gonna take a sample from every one of those hundred women. Or maybe not all hundred women we're gonna take like a quarter of them we'll just do a random sample take 25%. So these women do not, don't represent the general community infection but they do represent the women coming to the hospital and these are the patients that your doctors are responsible for. So this is a simpler way to do a random study is maybe like one month a year or maybe every other year or maybe every couple of months take a sample from every woman coming to the clinic for that week or that month. Or if you can't do all of them maybe just take a random sample. So in that case, you will have a random sample of a certain subset of people coming to the clinic. You'll have a random sample of them and those results will represent the women coming to the clinic. They do not represent the general community so that needs its own study but in terms of routine diagnostics the doctor is just trying to take care of the people who come to walk in the door. So is that helpful? So use your routine data as much as you can because that's the only thing that's gonna be there year in and year out but you can also do once in a while a special study. Choose a few of the hospitals don't choose all the hospitals choose a few of the hospitals a few specific issues like urine, attract infection and blood and maybe this year we'll do urine next year we'll do blood and you can rotate these special studies and you can do a random sample or a comprehensive sample of everybody coming to the clinic. That's very realistic because the people are already coming to you. You just need to take more samples. If you really wanna be a big research burden of disease group because there are these burden of disease studies they do hire people and they do go knock on doors. I'm involved in one project in Uganda, Tanzania and Kenya called Hattua in Swahili it means well the acronym is holistic something or other antibiotic therapy use whatever but they are going to knock on doors. They're going to three districts they're knocking on doors. Have you had they're asking the women or have you had a recent history of UTI do you currently have the UTI? And if so they take a sample these kinds of special landmark studies are very valuable to have in the literature but they're just they're expensive and they're just not reproducible and relevant for everybody to do it all the time. So does that help to answer the question? Yes. Bias data have a lot of value but the biggest trouble with bias data is in treatment guidelines and of course that's one of the most important reasons. So if you're using them for treatment guidelines you must be aware of these biases. Mr. Pete you can look at if resistance is low in a biosample the true resistance is even lower. You can compare drug one to drug two in a biosample you can see if the problem is getting worse and that's important like if the standard treatment guideline if the treatment the national standard treatment guidelines as UCIPRO but UCIPRO increasingly then something's getting worse and you need to reevaluate it. There is something called there's something routinely done for malaria because malaria there is no routine resistance tests I mean there's special there are special tests but routine labs do not test malaria. So in malaria they set up registers for monitoring treatment failures. So if you have if you so when the patient comes back and they tell you I was already treated for malaria but I didn't get better. So a lot of countries do have a register to keep track of that. And once the number of treatment failures hits a certain threshold it triggers a response. Why did the patient not get better? It might be because of resistance. It might because they didn't take the drug. It might be because they never had malaria to begin with because you know they were just diagnosed with the fever someone said you had malaria but they didn't have malaria. So it's an incorrect diagnosis or maybe the drug was poor quality. You know I was in Nigeria and there was an example where were they giving twice as much chloroquine as usual and they said, why are you giving twice as much? I said, well because it's not good quality drug so we give twice as much to make sure the patient gets the right amount. Of course the dangerous if they do get some good drugs in. So I mentioned this example from malaria because they do not have laboratory testing but you could also think about something similar. Keep track of women who fail the treatment guideline. And if that exceeds a certain threshold that's when you may want to reevaluate the treatment guideline. This would be a good way to evaluate your priorities for special studies. You can't do a special study for everything but you can ask your people we have given you a standard treatment guidelines for syphilis, urinary tract infections, pneumonia, dysentery. Do you think that the national recommendations are appropriate? When you follow the treatment guidelines are your patients getting better? I have an example of this from Zambia. They said, we cannot do the standard treatment guideline for, for a, blah, blah, blah, blah. And I said, what can you not do this? And I said, because the national guidelines says we should use spectinomycin we have no access to it. So this was not a resistance issue. This was an issue of availability. So whoever made the national treatment guideline for gonorrhea really did not consider the practicality of what was available in country. So that's a different reason. But in another case that might say that urinary tract infections we recommend Cochrane Moxizol or SIPRO. If you start to see a lot of the clinics are saying, we stopped doing this. We no longer follow the treatment guidelines because the treatment guidelines are not working. We're getting too many treatment failures. This would be a sign that maybe you need a special study. If you have too many treatment failures for dysentery just start collecting the stool and doing resistance testing. And what you may find is that the first line drug in the treatment guidelines has high resistance. And that should suggest a switch to a different drug. I'm not telling you much about Hoonad but I've been telling you a lot about how to utilize these data. I hope it's useful. There's something called the in vivo in vitro correlation. If you have a disease like tuberculosis or gonorrhea or shigella, if the lab says resistant the patient will probably not get better, potentially die. If a patient has gonorrhea resistance to septoraxone and you give the patient septoraxone the patient is probably not gonna get better. So you can believe the lab results. On the other hand, if a patient has an ear infection with strep pneumonia ampicillin resistant and you give the patient ampicillin the patient's probably still gonna get better. Because you know the ampicillin even if it's not great it's still gonna have some impact. And a lot of people resolve their ear infections even without treatment. So there are certain, there are many cases where if the lab says resistant the patient will fail therapy. TB, gonorrhea, shigella. You must trust the laboratory. But things for like ear infections it doesn't matter as much. You know, most of these patients get better anyway. And even if it's not great it's still a little bit helpful. I mentioned that because there are certain things like meningitis and malaria and gonorrhea that there's sort of a rule of thumb and it is just a rule of thumb, it's not completely, you know, it's just a rule of thumb that if resistance to the first line agent exceeds 5% switch to a separate agent. So for example, if first line therapy for NYSERA meningitis is chloramphenicol in West Africa it's fine as a first line agent. But if resistance goes up to 8% you really should think about switching to a different first line agent. Because it's one to one, if the lab says resistant you know the patient will probably fail therapy and you do not want 8% deaths. Of course this 5% rule only applies in unbiased samples and we get back to those sampling issues. I think I am ready to move on to the next slide. Any questions? Okay, I'll move on to the next slide and I didn't see we're coming up on time. So the main things that are left are this issue about cross-resistance. So here I'm looking at two drugs, Clebsiella with gentamicin and amic acid. On the right hand side of the screen I see in the columns, gentamicin resistant, intermediate, susceptible. And in the rows I see amic acid resistant, intermediate, susceptible. In the upper right hand corner I see 86% of the bacteria. Specifically 86% of the bacteria are sensitive to both gentamicin and to amic acid. So most of the bacteria, 86% gentamicin, amic acid susceptible. Conclusion number one, these are both very good drugs 86% susceptible to both. Lower left hand corner, 0.3% are resistant to both. I wanna draw your attention to the upper left hand corner. It says 10%. So they're in the first column so they're gentamicin resistant but they're amic acid susceptible. So gentamicin resistant amic acid susceptible which is the better drug amic acid is. So 10% of the time the gentamicin is resistant but the amic acid is susceptible. So amic acid should work. So my second conclusion, first conclusion both drugs are pretty good drugs, 86% are sensitive to both. Second conclusion, amic acid is better. Specifically it's 10% better. 10% of the time the amic acid is sensitive but the gentamicin is resistant. So they're both good, amic acid is better. Some people would then make the incorrect conclusion that we should be giving amic acid to all of our patients. Why? Because amic acid is better. So let's give it to everybody. That is not the best conclusion for a few reasons. Amic acid is much more expensive and it's also a reserve agent. We don't wanna give amic acid to everybody because if we're gonna waste it, it might be a great drug this year but if we waste it, it's gonna be a bad drug next year. So a better conclusion and more nuanced conclusion from this is that for a non-life threatening community infection gentamicin is the better choice. It is very effective. The patient's life is not in danger. So for a non-life threatening community infection, gentamicin is a very effective, inexpensive option that you work at least 86% of the time. But if I have an ICU patient or a patient with sepsis, a patient with prior therapy, for those patients, amic acid is the better choice because their life is really in danger. I don't have time to wait around to see if the first antibiotic works. So with gentamicin, it's gonna work 86% of the time and if it doesn't work, the patient is just gonna come back and we'll give the patient a different drug. So this kind of presentation called a scatterplot is very valuable for looking at first line choice, second line choice. We're looking at the cross resistance between them. Because of the time, I'm gonna go directly onto the next slide and we're now looking on multi-drug resistance. Here we see at the top of the screen E. coli. What you see here are data from one U.S. state, but we're not looking at all of the E. coli. We are looking at E. coli resistant to two drugs, Sceptraxone and Fluoroquinolones, but susceptible to Sceptraxone. So on the previous slide, we looked at two drugs at once. Now we're looking at three drugs at the same time. This combination is not a common combination. The reason is Sceptraxone resistance and Sceptraxone resistance usually go together. If it's Sceptraxone resistant, it is usually Sceptraxone resistant. So the fact that it's resistant one to sensitive to the other is uncommon. We saw it here in hospital B. We saw it in hospital C. We saw it in some of the nursing homes. We saw it in hospital F explode. So we're using the multi-drug resistance as basically a way of doing phenotypic tracking. So this is a possible outbreak strain as defined by its phenotype. I'll show that in greater detail with a different example on the next slide. Oh no, I guess I removed the slide, but that's fine, we don't have to show it. Capsule and Ammonia, some Capsule and Ammonia, almost all are resistant to Amphicillin, but some Capsule and Ammonia are resistant to one drug, some are resistant to three, some are resistant to five, some are resistant to eight. So not all Capsule are the same. So we can use the multi-resistance phenotype to look for possible outbreaks. If you have two patients in the same room with Capsule and Ammonia, I'm a little bit worried that there's a little outbreak. But if one Capsule and Ammonia is very sensitive and the other Capsule and Ammonia is very resistant, I'm really not that worried. There are a lot of different Capsule and Ammonias, but if they have the same strain, actually I'm gonna demonstrate this in a different way. I'm gonna go back into Hunan. I'm gonna go to the Devachor test hospital. I'm going to go to data analysis. Analysis type, I will go to resistance profiles. And I'm gonna say organism is dephorius. And I'm gonna choose one month of sample data. I'm gonna begin the analysis. So this analysis is called the multi-drug resistant analysis. It is one of Hunan's most valuable analyses, but most, but least utilized. Here, I'm not looking at all the drugs. I'm not looking at your phantom because it's only in the urine. I'm just looking at seven drugs. Penn, erythro, clinda, oxa, genta, poachromoxazol, Sipro. Here you see a column called profile. The bacteria at the top of this list are dephorius sensitive to everything. Here we have dephorius resistant to erythromycin, one drug. Resistance of penicillin, let me just move over to another column, move it a little to the right. I'm gonna make this column called resistance profile a little bit wider, good. So here we have resistance to one drug penicillin, resistance to two, resistance to three, resistance to four. So this is very helpful for finding outbreaks. Here at the bottom, you have four isolates, resistance to every drug that I requested. These are, I was in Venezuela and they told me in Spanish, John, I got the name I was gonna say Barambo. I said, I'm sorry, I did not understand that. They said, John, here we have a Rambo strain. I said, what is a Rambo strain? They said, you know, the movie Sylvester Stallone, Rambo, can't kill it, never dies. I said, oh, that's what the Royal Rambo is. It's a Rambo, it's completely resistant, killer bug. So here you see at the bottom four patients for different people, completely resistant to all seven drugs that I requested. What's interesting is that two of them were on oncology. So these patients may have picked up a hospital infection in the oncology unit. There were also two in the outpatient area, so I'm a little bit worried that these might be outpatients in oncology or they might be family members. So we're using, we don't have molecular typing, but we do have the phenotype. So the phenotype is very valuable for looking for outbreaks. This is the detailed list. This is the summary. So this multi-resistant to everything. We see at the bottom resistant to everything, but there was one at the beginning of the month, one at the end of the month, two in the middle of the month. It doesn't look like an outbreak, it just looks like a bad organism that we sometimes have. HUNED does have a feature called include cluster alerts. I'm gonna show a different example for that one. Let me just do this by species. Include alert. Just a friendly reminder that we need to wrap up. We have one minute. Okay. Thank you. And okay. So what I'm showing you here is HUNED's value for outbreak detection. So here for the Klebsiella pneumonia, you see those bars in red. So normally Klebsiella, they have zero, one or two, but then at the end of the month, they have two, three, five. So I'm sort of giving you a little teaser as to what we can discuss in the future. How HUNED can be used. We discussed the problems and using it for treatment guidelines because of the biases. But I wanna show you the very great value for outbreak detection by looking at these multi-resistance phenotypes. So that's just a little teaser of our next session.