 But one comment, John, if you, what we, last time with a major comment and the challenges, actually the major comment given from different angles, the report needs to include more epidemiological also reports. It has to show some epidemiological indications. Can you really support on this to give us some highlight to include epidemiological aspects of the report? In this moment, I won't have a lot of comments. Sometimes when people ask for epidemiology comments, there are two kinds. There are things that are realistic with existing data and data sources. And the things that are not realistic, maybe not in the short term, but perhaps in the longer term. Of course, people would love to have things like incidents. What is the incidence of MRT? And the problem with things like incidents is their problems with the numerator and with the denominator. In the numerator, you only have, you know, if people want to know the incidence of positive blood cultures, the problem is in the numerator. Let me take that back. If you want to know the incidence of bacteremia, you need two things. You need a patient with bacteremia and a positive blood culture. But if you don't take the blood culture, the patient is bacteremia, but you don't have any evidence of that. So in a lot of low-research countries, it's trouble to come up with the correct case count because of under-sampling. People have blood cultures. Most things are treated empirically. The problems with the numerator are coming up with a meaningful case count of disease. You can come up with a case count with culture results, but the culture results are simply a subset of everybody who has the disease. So there are problems with the numerator. There are also problems with the denominator. You know, what population is served by my laboratory? That's not an easy question. I'm in the city of Boston. The metropolitan Boston area has over 2 million people, but we have about 20 hospitals. And so they all overlap. I can tell you the population of Boston, but I can't tell you the population served by my hospital. And also we have to keep in mind, my hospital serves the local community of Boston, but my hospital also serves Massachusetts for some things. And the sixth is the surrounding states of New England. So we provide primary care to local communities, but we provide secondary and tertiary care to a much larger area. So it doesn't mean we don't do this, but we do need to keep in mind that there are significant challenges coming up with meaningful numbers. The risk is if you simply publish the numbers, you end up usually underestimating the disease because we are not taking cultures to a very large degree. You end up underestimating the disease incidents, but you often overestimate the resistance proportion because when we do take cultures, it is often the sickest people, the treatment failures, the ICU patients, patients with complicated medical histories. So often what you may see is that in a biased dataset, when you're only looking at treatment failures, you might find a dataset that thinks that in this database, in this non-representative database, maybe 40% of the isolates are resistant to a particular antibiotic. But if you really had a true unbiased sample, it might only be 10% resistance. So these ideas of incidents are extremely valuable. But with routine data, you have to be very careful with a lot of caveats and a lot of understanding of the biases. Otherwise, we underestimate the disease, we overestimate the resistance. We may recommend imipenem when it's not really needed. If we are overestimating resistance to appropriate drugs, we may switch to second-line drugs because of these biases. One of the things we're doing in Vietnam is we are trying to do some simple things. We are doing things like per hospital bed size. This hospital has maybe 100 beds. This other hospital has maybe 100 beds. The first hospital takes maybe 50 blood cultures in a month. And the second hospital maybe only takes 10 blood cultures in a month. So this can be very valuable to look at culturing practices. The hospital one maybe does a very good job of taking blood cultures. Maybe they have the resources, the trust, the materials, the clinical training to take blood cultures. So number two might have exactly the same disease epidemiology, but they are not taking a lot of blood cultures. So this is one element of epidemiology with meaningful denominators to look at culturing rates, try to estimate incidence rates. One nice thing that they do have in the Vietnam hospitals are two variables that the CDC uses a lot. One is number of admissions per year. For example, in this hospital last year, we had 10,000 admissions. And then you can do an incidence of, for example, MRSA spectraemia per admission. I think of 10,000 admissions with 100 MRSA, you can calculate an incidence based on the admission rates. In Vietnam, the hospitals also have what we call patient days. Patient days, if you have one person staying in the hospital for 30 days, that is 30 patient days. Alternatively, you may have 30 different people and they each stay one night in the hospital. That is also 30 patient days. Or if you had 25 people and on average, they each stayed for two days. That is also 50 patient days, if I said that right. 25 people, two days each person. That is 50 days. So it just averages out over the course of the year that we had patients occupying beds for like, you know, 500 nights. So this is another incidence measure. You can say, for example, MRSA per patient day of hospitalization. This is very useful for hospital infections and hospital incidents. So in the hospital incidence measures and epidemiology, it's some degree easier because you do know how many patients were hospitalized. You do know if you have access to the data, how long they stayed. So it's easier to come up with incidence measures in a hospital setting because you present your results in terms of hospital days and hospital visits. It's problematic, as I said, because I don't know how many patients were hospitalized in a hospital setting. It's problematic, as I said, because I don't know how large the, what is the population of autosavava served by facility one, by facility two. But it's useful to think about epidemiology of hospital infections and the incidence of the epidemiology of community infections in a different way because the denominators are different. The biases are different. So that's some comments I have on that. And I think it's useful, as I said, and be a calm, that they do have the patient, the hospital number of beds, the hospital days of admission, the hospital number of admissions, and the hospital patient days. So those are some comments. But if you send me the document and we compare with some other documents, we could look at those as well. Another useful thing when you're looking at epidemiology, is to try to distinguish between hospital infections and community infections. And this is not easy. There are two potential ways you can try to distinguish between hospital infections and community infections. And the US CDC has a national program called NHSN, the National Health Care Safety Network. And they offer two different definitions for reporting to the CDC of hospital infections. The first approach is called clinical reporting and the second approach is called laboratory reporting. The clinical reporting tries to come up with the truth. And they take every patient and they evaluate the patient. Is this blood culture from this patient, from a hospital infection or a community infection? And they look at the patient's history and the patient's risk factors that may interview the patient, that may try to find out where was the patient, was the patient in person home, did the patient come from another hospital, did the patient have surgery three weeks ago? So this clinical reporting tries to answer the question, did this patient I am looking at now have a hospital infection or a community infection? The problem with this clinical approach, two problems with this clinical approach is it takes a lot of time and effort and information that's not easily available. If you want to do this for everybody, every positive blood culture in the hospital, you need to look at the medical records for every patient with a positive blood culture to see and spend maybe half an hour or an hour on the patient's history and say, I see that this patient was hospitalized last week. I see that this patient had a surgery three weeks ago. I see that this patient was in a nursing home. I see that this patient was simply at home and they came in with pneumonia. So you can do that and say, this is a true hospital infection or a true community infection. One problem with that, as I said, is it's a lot of work to do this correctly. And the second problem with that approach is that different people will still come up with different answers. There's certain people where it's easy to say this was a community infection. There are other people that's easy to say this is a hospital infection. They had a hip surgery and then they had hip surgery with MRSA infection and then two days later they had a positive blood culture. So there are many people where it's easy. This is a hospital infection, this is a community infection. But there are a lot of patients in the middle where it really is not clear. Patient was exposed to the hospital but they didn't get high risk exposure. By the way, if someone could turn off their, if someone could turn on their mute, I hear some noise in the background. Thank you. So two problems with the clinical approach to determining hospital infections is one, it's a lot of work and you need the patient's medical record in detail. Second problem is that even two different people may come up with a different answer. They'll say, yes, this one is, this one isn't, this one I'm not sure. And I'll say yes it is and the next person may say no it isn't. So that's a problem with the clinical approach. The other approach is a purely a data driven approach. It's not always correct, but it is simple to apply if you have the relevant details and it is consistent. So every hospital in the United States around the world, if they apply a consistent definition, it might not tell you the truth for every person, but on average it will be approximately true. It's much easier to do and it's also easier to standardize. If you have new staff, you don't have to, they don't have to be experts in reviewing charts. So I'll now explain to you the lab, it's called the laboratory identified event. It's the laboratory approach to saying if this was possibly a hospital infection. It's what we call a proxy definition. It's what we call a surrogate definition. So the proxy definitions on average are usually right, but there will be exceptions to this. So what do I mean by this? So the simple proxy definition is if you have a patient with an outpatient sample, we are going to call that an outpatient infection. It's easy to do. The patient came to the emergency room, the patient went to their private doctor, the doctor, the patient went to a local clinical outpatient laboratory. The fact that they went to an outpatient location in the simple approach, we're going to simply say it's a community infection. That's very easy to do, but it's not always going to be correct. If the patient was in the hospital for two months and then went home and then a week later, it goes to their private doctor out in the community that could easily be a hospital infection after the patient went home. The patient has a hip surgery, they go home and then they have an infection while they're at home that is still considered a healthcare associated infection. So this simplified definition of outpatient isolates or outpatient infections is very simple to do, but it will not always be correct. Obviously, just because you're in the outpatient setting, it's probably an outpatient infection, but it might not be. In the other direction, if you have somebody in the hospital and they have a positive blood culture with MRSA, is that a hospital infection or a community infection, the simple thing to say, well, the patient was in the hospital, let's call it a hospital infection, but that itself is there are a lot of cases will not be true depending on the day of hospitalization. If the patient goes to the emergency room, they're very sick, they have a high fever, they go immediately to the intensive care unit. Once they're in the intensive care unit, the doctor takes a blood culture. If that blood culture is on hospital day one or hospital day two, the patient probably brought the bacteria in from the community. So this is an example of a community infection requiring hospitalization. So just to repeat myself, if you have a positive blood culture in a hospital sample, if that hospital sample was taken on hospital day one and two, it's probably a community infection. If the sample was taken on hospital days three, four and five, it might be a community infection and they took a while for them to diagnose it, but we're just going to call it a hospital infection. So this is, now I'm going to make a recommendation for you in the future in Ethiopia to apply this simple definition of the WHO glass definition of inpatient infection or community infection depends on knowing three things. Is the sample taken in the community or is the sample taken in the hospital? So that's what we call the location type. So if you know the location type, you know if it's a community sample or a hospital sample, I'm not saying community infection, I'm not saying hospital infection, I'm describing the sample. If it was taken in the community, we're calling it community sample and we will also call it a community infection for WHO reporting. That's simple to do. It's not always correct, but it's simple to do and that's the WHO recommendation. On the other hand, if it is a hospital sample, then we need to know the date of admission and the specimen date. Using the date of admission and the specimen date, Hoonat can calculate the hospital day number and Hoonat says, well, if the sample is from hospital day one or hospital day two, we're going to call it a community infection and if the patient has the sample in hospitals three, four or later, we're going to call it a hospital infection. So this is a simple element of epidemiology that I would recommend you introduce in the future. As I said, a lot of people say, well, if it's a hospital sample, it must be a hospital infection. But in my own database, for depending on the organism, depending on the specimen type, a lot of times like 60, 70% of the hospital samples are taken on hospital day one and two. In other words, these are community infections, severe community infections requiring hospitalization. So I've given you a lot of comments, but I just want to simplify it to this with regards to hospital infections and community infections. I do recommend that in your database, you try to capture systematically in the future, now when possible, try to capture the location type. Is it inpatient or outpatient? And try to capture where possible date of admission and specimen date. Well, specimen date is not a problem. Everybody keeps track of that. But date of admission, for many hospitals is very easy. For many hospitals, it's difficult because they don't put, the doctors don't put the date of admission on the report sheet. Now, so a lot of places do, is they tell them that we cannot, we do not know the date of admission this year in most of the hospitals. Next year, we're going to try to introduce this. So I would recommend that over time, you try to include on the request forms, you try to include in your data management systems, the date of admission. This will allow you a useful way of a quick, simple separation of hospital infections and community infections. It's not perfect, but it is standardized and it is easy to apply if you have the location type and the date of admission and the specimen date. I'll leave it at that. Any other questions on that point or other similar points? What about patient-level data? Can you also reflect on that just a minute? Patient-level data? That's another element of... When I first started, I started this project in 1989, so 31 years ago, and at the beginning, a lot of dermatorepidemiologists, I was telling them what I was doing. I'm saying we're collecting routine data. And they say, you can't do that. The data are so biased and there are all these quality issues and you're not getting all the detail. And they had elements of truth, but on the other hand, it's good now that people have switched. They say, oh, yes, let's do that. Let's use the routine data. But then sometimes people also forget the limitations of it, the biases. The way I view this is there are two kinds of data needs. One is what you want is routine, comprehensive data collection as close to real time as you can. All organisms, all specimens, all antibiotics. If you believe that microbiology data quality, then you can use these analyses for epidemiology and statistics and outbreak detection. If you don't believe the microbiology data quality, I still want to collect the data. I think it is a good thing to collect bad quality data, but don't publish it. Use the bad quality data to identify the problems in data quality and to improve them. For example, if I get data from 20 laboratories and two of the laboratories have very strange results, staff resistant to vancomycin, a lot of Klebsiella sensitive to ampicillin, I say, I don't believe the data from your laboratory. I'm not going to tell them don't send me your data because that doesn't help anybody. That means they don't know their problems. We're sort of hiding the problems. I say, please continue to send me your data. I'm not going to include your data in the national report. I'm not going to use your data for treatment guidelines. I'm not going to use your data for looking at epidemiology, but I am going to use your data to help you to improve your data quality. So good data, bad data, bias data, unbiased data, all of these have a value and you need to understand the value in order to appropriately utilize it. One of the purposes of the data is to improve the data system. More complete data entry, more accurate data entry, better microbiology, better testing, normal results, not strange results. Better data collection to improve data quality. Good. I just got an email that I read quickly. So bad quality is still collected, but do not share it with anybody except for that hospital and just in the network coordinators. The good quality microbiology data still has problems of biases and sampling. Let's see where was I going with that? Okay, so patient level data. So one kind of surveillance that I recommend is just collect the routine data. If you can use it for epidemiology and treatment guidelines and other things, if you can't use it for data collection, use it for data quality improvement and feedback. Use the data, the patient data that are routinely available. There are certain basic things about patients that everybody should get or should try to get. A reliable patient identifier that is a challenge if you don't have reliable patient identification numbers. Patient gender and patient age. As I said in patient outpatient, that's not really a characteristic of the patient, but it is a characteristic of the patient's medical visit, the patient's medical interaction. So at a minimum for routine surveillance, I would suggest a patient identity failure, perhaps the age if you're going to, I'm sorry, perhaps the patient's name if you're doing the clinical reporting. Age and gender and date of birth if you have it. Another advantage of date of birth, as you can tell you the age, another advantage of date of birth is you can use it to find out if these two people with the same name in fact are the same person. So this is the minimum I would require for routine data collection for patient level data. But as I said, a lot of the epidemiologists were very critical, but we want to know what were the patient's risk factors? What were their travel history, depending if it's an entire pathogen? Did the patient in the hospital, did the patient have surgery? Did the patient require antibiotic therapy? What antibiotics did they get? What was the patient's outcome? Were they discharged in five days? Did the patient die? Was the patient transferred to another hospital? All of this is extremely valuable information. And in my mind I distinguish between what I call public health surveillance and public health research. A lot of these things I just mentioned, clinical outcome, diagnosis, therapy, risk factors, medical interventions are extremely valuable. But most normal data information systems cannot easily generate these data in real time. They're in different databases or they're on paper. So I would recommend for routine data surveillance, come up with a minimal set like age and gender, maybe date of birth, because it's realistic and it has value, epidemiological value, male versus female, young versus old, what are the pediatric pathogens, what are the elderly pathogens. If you want to go beyond that to the risk factors and the all those things that I mentioned, that's valuable, but it's often not realistic to do this routinely, forever, in all of your hospitals. There's an excellent hospital in Thailand that does exactly this kind of work. So it's what I consider a sentinel site. Some sentinel sites just do like a sort of a research protocol one month a year or for six months, or they'll do it for blood cultures and sepsis. So when you start doing these protocols, you often be more specific. You often say, well, I'm only interested in the blood cultures for this study. There is a very good group in Thailand at one of the hospitals. They have a web application and every day the infection control nurses go around the hospital and with the smartphone application they put in the risk factors. Did the patient come from home? Did the patient come from a nursing home? Was the patient transferred from another hospital? They put the therapy decisions. They put the patient outcome death, discharge, transfer. So it's extremely valuable and they feel this is important enough that they do it all the time. But it's just a lot of time and effort and training that's not realistic for a lot of other places. So one recommendation I've made for a lot of people for a long time is I recommend focusing on routine basic comprehensive surveillance because of its value for laboratory quality improvement, data management, expertise, outbreak detection. Even if your data are extremely biased, if there's an outbreak you'll still see that in the biased data. If normally you have three pseudomonas in a hospital in a month, in this month they have 20 pseudomonas, even if there's a big bias something has happened. So this approach is minimal resources. It's using the routine. It's comprehensive and a new pathogen if you have sorority, rubidiae. It's not an organism on most people's surveillance lists, priority lists, it's not a priority but if a hospital normally has zero and now they have 10, that's still important. That's still something epidemiologically has changed. In fact, there's a group I'm working now in Asia-Pacific that's again coming over the protocol that says we only want blood, we only want this, we only want that. That might be fine at the global level for Geneva but as a national surveillance program you really do want all the pathogens, all the specimen types, all the locations for outbreak detection so I recommend using the routine data. There are certain important things where you can improve the routine data. I would start with the simplest things, age, gender and slowly over time date of admission. So use the routine data and for simple things try to include those in all hospitals in a routine way and then on certain priority issues, certain priority decision needs, you're making a national treatment guidance for example, for blood cultures and you want to do a special study. The special study often does need extra resources and you might want to do the special study for three months in a few of the hospitals and then maybe repeat the study a few years later. I think I'll just leave it at that but I hope those comments were useful. There's routine surveillance where you sometimes what people want is not realistic, WHO included, they said what we want to know is not a patient outcome. Because they want to know mortality. They said yeah, I know you want to know that but the laboratory doesn't know that. They have a sample today and the patient dies three weeks later. The laboratory is not going to know that. It's valuable information but it's sort of a different paradigm for what kind of surveillance do you want to do, what is sustainable and acceptable for existing resources. You can't do that for everything. You have to choose some priorities. Did I answer the question enough or do you have a follow-up question on that? Thank you, John. Just I am satisfied. Thank you. I mentioned this because some of the clinicians already asked us this kind of question. Whenever we present the annual surveillance report most physicians from the facilities ask us where is the patient level data? You don't have patient level data or you have to improve this and this kind of comment is always given us from the clinical side. Yes. Whenever people ask me a question like that because I want to say that both are needed. I do recommend strengthening the routine because it is sustainable and doable has great value but it does have its limitations. One additional advantage that I mentioned of this comprehensive approach is if you have a comprehensive approach with 20 hospitals you are making a network of 20 hospitals with a long-term commitment to collaborate over the next 5, 10, 20 years. Now that you have a platform for collaboration that platform is suitable for doing special studies. The special study might be for the hospitals and it might be for six months. The problem with a lot of these special wonderful research projects is these things start and end. You do a wonderful project for six months and then the project ends. And then three years later you do a different project. The problem with that approach is you end up starting from the beginning. The people have changed, the data quality has changed, all that stuff has to be redone or if you don't have a good routine base you just can't come into a special research project because there's a lot of basic capacity building that has to be done. So one problem that I'm recommending the routine approach is the core of an ongoing national surveillance strategy but some people do that and then they stop there. They never do anything to try to address the limitations. One of the things they do in Argentina is they have the routine surveillance that involves about 90 hospitals but they also have a special project on respiratory pathogens. There's a Latin American project called Sireva. It's for respiratory infections where they do serotyping and serotyping is extremely value for pneumatoccus to know if you have the right vaccine. So this is an example where they have the ongoing surveillance with 90 hospitals. They don't have all 90 hospitals do serotyping. There's a subset of the hospitals that are given extra funds and training to do serotyping following more of a protocol and for these people they also have some follow up questions if a patient is positive pneumatoccus was the patient vaccinated? So they have a national platform inside of which they do some of these special projects. So when the physicians ask you that kind of question you need to be ready to answer the question because sometimes they say well I don't trust what you're doing you're not giving us what we need and my first thought is can you please tell me in more detail what you want sometimes what they want is valuable but it's not realistic comprehensively they say well I see what you're saying but we don't have the research that information is not available or it could be available let's collaborate and let's do a study on that and then we'll publish the study let's try to apply for a grant let's try to get money to do that study we will publish it well for any one of the most annoying one of the most unhelpful comments is when people say well where's the epidemiology so what do you mean can you please be more specific when people say well we need more quote unquote clinical data when some people say they want more clinical data all they want is the patients demographics age and gender some people they want to know the risk factors and the treatments and the outlines and the outcomes so when people say I want more epidemiology data is there they're coming to so vague I say well can you please be more specific some things that you might want we have some things that you want we don't have but they are realistic let's try to put them in sometimes what they want is the patient outcome for example in Ireland they have one of the questions they do is 30 day mortality for positive blood cultures for hospital infections rodeo for hospitalized patients so what they will do is they will find they will as do as much as they can looking at the hospital's electronic records I don't know exactly how they do it I just know they do it somehow I can tell you the details of how they do this but basically like in my hospital if I want to I easily have access to the positive blood cultures if I go to my hospital's computer system I can look up these patients and see whether they're still alive at least according to the hospital records you know if the patient went someplace else and died in my hospital system immediately so something like mortality it's not as if it's not realistic but it's not easy to do I would have to look them up manually because I don't have a database on mortality we're working with a United State Nebraska and they have a very nice system they have a comprehensive system for national collection of the microbiology results on the one side on the other side of course they have the statewide death so we need to link those two databases together what patients had a positive blood culture and which of those patients died within 30 days after the positive blood culture so the next time somebody tells you about well where's the epidemiology and they're critiquing you need to ask them can you please be a bit more specific some of what you want we have some of what you want we don't have it now but yes we could try to do that sometimes what they want I can see the value in it it's something really that needs extra time and money and resources and maybe best suited by a research project for example is we're trying to with the Fleming fund and WHO and other projects we want every country in Africa to have a strong national surveillance program if WHO is funding a special project on sepsis we don't need every country in Africa to be part of that project we could take a few from west Africa east Africa so the idea here is to take it to have every country to have a good strong national ongoing program that serves many objectives when there's a certain need pick some of the countries some of the facilities and do a special project I hope that is helpful John thank you very much yes when they critique it sometimes their critiques I said they're valid critiques but what you want is not realistic with the money that we have available so the way I view this is they say well that's not valid it is valuable it's not what you need what you need we need more money but what we have tells us about data quality it tells us about trends it tells us about outbreaks all of those are valuable things we're doing those laboratory testing what can we use those data for maximize that what can we not use the data for well that's when you need to think about these special questions and Sentinel sites and protocols other questions would you have no more questions I'm going to go to the to the outline that you see here but other questions okay no obvious questions obviously if you think of one just interrupt me so I'm going to start here with this number one also I'll start here tomorrow is tomorrow is our last remote session well that's regarding you know Mikael and Fern I don't plan to do this kind of writing with you every week every other week but I'm still available I am a WQ collaborating center to do this kind of work so feel free to reach out to me anytime I'm happy to do more sessions but I am busy I can't do them every week every other week we will try to put more and more videos on the web short things I'm thinking of like three to ten minute videos so that you can learn in other ways especially the data managers the data managers should feel free to reach out to me so I'm going to start by email if we can't address these issues by email we can set up one of these sessions for an hour an hour and a half so I'm not disappearing I've been doing this for 31 years I hope to do at least 31 more years if not more I work with Tom O'Brien who is now 91 and he is still very active in our work so I'm still around so just write to me by email and then if needed we can set up another kind of training session with a small group or a larger group so I'm going to start at this point second point or review of anti-biograms okay anti-biograms okay let's see let me go to do a Google search anti-biograms in other words cumulative antibiotic statistics that hospitals or countries typically do on an annual basis I'll show you a couple of resources I'm going to look first so I hope all of you are familiar with CLSIM100 if you are not familiar with this document there's just a free version of the document the M100 is the document on how to do how to interpret antibiotic measurements zone diameters, MIC values this document is updated every year it's the breakpoint tables for just diffusion in MIC M100 is the tables for routine bacteriology the M60 is the routine tables for yeast M61 is for mold but M60 is for yeast those two documents are available to read you cannot download them, you cannot print them but you can read them so routine bacteriology, routine yeast you can read online skipping down to vet 08 vet 08 the veterinary breakpoints these documents are free the M100, the M60 and the vet 08 for you to read online there are other documents M61 for mold M62 for nocardia and ectinomycetes another one for tuberculosis M45 is an important one CLSI M45 is for infrequently isolated or fastidious bacteria most of them you've probably never heard of you may not have ever isolated them a lot of them have to do with bioterrorism anthrax some of the pastorella so this is it's not routine bacteriology I think it does include campylobacter which is important but it is a fastidious organism so the M45 is not a free document and it's also another one on the veterinary side the vet 08 the vet 08 I showed you that's routine bacteriology the vet 06 along the same lines it's infrequent or rare fastidious veterinary pathogens there's also the CLSI vet 03 vet 03 vet 04 vet 03, vet 04 is about aquatic animals fish shellfish, shrimp, lobster mostly fish salmon, things like that so these are not free documents but most people don't need these documents this is very specialized documents so these are CLSI documents about how to interpret antibiotic testing and perform antibiotic testing there's a different document called the CLSI M39 I'm one of the lead authors of this document this document is analysis and presentation of cumulative antimicrobial susceptibility test data as you can see the price ranges between $54 if you are a member $153 or not it only actually costs about $80 to become a member so a lot of times if you want to buy a bunch of documents pay the $80 and then the rest of the documents usually go for like half price also I'm talking about for high resource countries in a low resource country I would hope that you can get discounts or somebody to sponsor to purchase some of these key documents the M100 is updated annually most of these other documents like the veterinary documents are updated infrequently the M39 we updated about every five or six years this is the fourth edition we're now getting ready to almost finalize the fifth edition so I wouldn't recommend buying this document because hopefully next year we'll have the fifth edition and if I view the sample pages the key person who really was the parent who did all this was Janet Henner wonderful personal friend I went to her wedding I first met her I think I first met her at Kemri in Nairobi giving a training course and she was showing people how to bleed sheep she's one of the US lead microbiologists and she's the lead of this and I'm the last one because I'm the guy who actually did all the statistics to support the recommendations here so this document analysis and presentation so this guidance is official CLSI guidance on how to do here's the table of contents how do you design information system data analysis data verification the last many, many weeks is covered here somehow calculations, bias, limitation section number nine, culturing practices small numbers, comparing results and then confidence intervals so this is official guidance on CLSI and how to do facility anti-biograms in the fifth edition we are now finally getting to network by anti-biograms including national anti-biograms so the initial focus of this was for specific hospitals to make local treatment recommendations in this fifth version we're moving a bit more towards a surveillance objective so the focus is still on treatment guidelines but we try to introduce elements that are not only treatment guidelines but general surveillance and benchmarking comparisons so that's what we're trying to put into the fifth version and then build on that eventually in the sixth version okay that's the M39 document there's another one that I'm also a co-author on they keep on changing the number CLSI vet zero five yeah vet zero five generation presentation application of susceptibility test yeah, the veterinary antibiotic surveillance program this document it has the letter R, the R means it's a report it's a one-time thing a lot of things starters report the first time and then later they get made into guidelines so there's not a guideline, this is a report some of these reports are simple report published once some of the reports eventually are transformed into guidelines that are updated periodically so that's that on veterinary documents there's also an FAO document FAO document on well I don't want to get too detoured on this the FAO also has recommendations not for data analysis but for surveillance in veterinary pathogens I've discussed some of that on a previous call okay so that's the CLSI M39 if I go to PubMed and if I look for Stelling and Hindler we have two publications together the first one is about definitions for multi-drug resistant MDR, XDR, PDR this is an important document commonly referenced by many people when they're looking for definitions of MDR, XDR and PDR the 2012 document really needs to be updated now that will happen eventually it was coordinated by Dominique Monet at the ECDC in Stockholm so Janet was on that committee but Janet and I together just the two of us we did this publication in 2007 where we because the M39 is a commercial document we made basically a free version of it as an article and it also accompanied the the M39 so we want to reduce something like this with more authors on it so these are some resources from some guidance on how to do anti-biograms some other resources I'm going to go to do another Google search and I'm going to search for Philippines NMCOBU resistance RITM the Research Institute for Tropical Medicine I do encourage you and I don't know examples in Africa if you know any good examples around Africa for anti-biograms it's not that I there are none it's just that I have not worked much in Africa until the Fleming Fund came along so if you know good examples from Africa I would love to know about those so let's see this is RITM they have the anti-biogram resistance reference slide but be good for you to look at these to see the scope of their activities their training courses, their quality assurance strategy what I would like to show you if I can find it well I think they have a more recent one this is ARSP is the National Anti-biogram Resistance Surveillance Program oh to find out more about ARSP let me just go to the ARSP website let's go to that link instead that's going a bit slow so while that's coming up I'm going to go to Thailand Thailand and I'm going to look for NARST well I know what NARST is let me see if I can find it well I'll just go ahead this is the National Anti-biogram Resistance Surveillance Thailand Center of Thailand so here you can see their reports here's a funny way to spell Hoonat I think they've made a mistake there so everything that they have there is based on Hoonat so they have an option for changing this to English well okay here's anti-biograms so let me look at their all anti-biograms for 2019 so this is a simple drug bug combination with colors so this is not a report it's a simple anti-biogram some of the things like here you see Staph aureus and they have separated into MRSA, MSSA, ICU, inpatient, outpatient so these are a simple epidemiology things that you can do so when they say epidemiology that's too vague, epidemiology is so many components so some elements of epidemiology we can put in easily and green means it's a good drug yellow means it's a medium drug red is a bad drug I think gray means intrinsic resistance meaning it would never work there are a lot in Europe I'm going to show you the one from the United States NARMS, FDA Database it's a collaboration between the US CDC the US Department of Agriculture and the FDA the Food and Drug Administration for Human, Food and Animal Islets where are we where's the database database data resistance genes there's a lot of resources here resistance genes about NARMS let me check resources so they have a nice so I recommend that you just look at what other people have done so I did that with Vietnam they were making the first annual report and I gave them some of these examples and they copied a lot of the ideas out of the Philippines NARMS methodology interpretive criteria some are in here so I don't I'm not going to look for it now but they have an interactive database with graphs and charts on the animal side you can choose chicken, cattle they've done a nice job with that so I'm now going to show you what we've done in Vietnam and a lot of this was done with CDC input if I go to here and I go to countries and countries and I go to Vietnam and I go to where am I going and your report and let me just choose which is the most September okay this is there they're getting close to distribution I won't give this, oh I'm sorry this is a this is an existing publication it's a 2019 publication which is nice so the Vinaris is you know Vietnam and AmeriCorps resistance and and you can see who that is mentioned in their methodology for how they collected their data collection that's not what I meant to show you this is the draft report from September 9th and they are really in the hopefully next few weeks they're going to finalize it and eventually make it available so I'm going to go over this quickly because you're not supposed to look at their data and obviously you're not going to read this so quickly but you can see here data collection they describe what they've done key results pathogen distribution organisms of interest it's a little hard to read let me make this a little bit bigger and able editing back to the top so I'm just going to focus on the table of contents overview of the Vietnam program what are their objectives, what are their audience what is their process for data collection validation cleaning, evaluation of results limitations very important to include that key results, data submission how many facilities, how many beds the patient days I mentioned summary of how much data was contributed and then a special thing subset about things relevant for glass plus it's a small subset of what they do the organism distribution is part two the antibiotic characteristics of priority pathogens is part three different for each one, Papsiola the new doctor, next steps so that's that, I'm just going to go quickly down to the bottom just to show you some of the graphs so they're showing their country and the maps describe what kind of hospital each one is north south central that's the glass table I'm trying to find their where's their what do they use Hoonet or something else, do they include the positive do they include the negatives do they include the positives, yes do they include the negatives data on isolates is being submitted all of it is subset of data do they have this, data completeness comments not only about the epidemiology and here you see average daily patient census number of admissions rate of samples per 1,000 admissions and you see some of the numbers don't make sense, they said that doesn't make sense, that's number too high this number is too small, this one is 4 million so they're also finding problems in their underlying quality that they will not include in the report in this first year but allows them to make improvements data volume by facility if I pause, 10 most common pathogen seen in blood staff horse distribution by inpatient outpatient, I'm just kind of going this is strep pneumonia with its antibiotic resistance, I'm going to get out of this but I really encourage you to look at some of the reports because these will give you good ideas and bad ideas about things you want to do and things you don't want to do one of the things in Vietnam is the initial report was way too long they were doing everything by blood and urine and it was just too many tables no one's going to read all of that it's nice to have it on an interactive website so if somebody wants to do it they can generate themselves but you want to document that people can read you also don't want to focus purely on the tables the table of course is a core but people don't read the tables they read the executive summary, they read the highlights so and a lot of the highlights in the first year the most interesting findings are often because of mistakes or biases and different kinds of errors and that's part of the highlights of what we learned in this first year in the first year you talk about the limitations as the years go on you can more confidently talk about the epidemiological findings so this is the Philippines page did come up and here you see serotype on salmonella so it's a combination of things that they've done and this is actually a PDF file so it's not interactive but this is describing some of the work that they have accomplished okay good so for instance, six documents three of the documents are about anti-biograms I'm going to start first of all with this Word document a cumulative anti-biogram this is from the United Arab Emirates one of our best beta testers or evaluators a person's recommending new features is Jens Thompson Abu Dhabi, well he's from Germany but he's been in Abu Dhabi for about 15 years about 13 years so here's just a nice example of what can be done and I showed you another example for Thailand where they put the colors in he also does that for facility so he's also done that for other examples I didn't send that so this is just nice short and simple this first one is gram-negative bacteria the second one is gram-positive bacteria and they separate stephoris by MSSA and MRSA and the third one, what is this one this one is Canada, so this one is fungi and this last one is tuberculosis so that's one document that for instance a second document is two presentations so the first one is this power point and I think this one is from United Arab Emirates cumulative anti-biograms good things, bad things, basics caveats, biases what is this so it's a 61 slides so you see here it's got the colors so if you want to go through this on your own you might learn some useful things or you might want to develop some of your own training materials why do you do these informed clinicians guide informed, so it's not one thing there are multiple objectives for different audiences that's a presentation he gives then Janet Henler gave a presentation in 2019 in Mexico so again it's simplifying it's giving examples of precisely per patient separating things by inpatient, outpatient maybe age, maybe gender for example in urine infections it's very important to separate male and female females have a ton of routine outpatient normal uncomplicated urine infections women also have complicated catheter and other associate urine infections the men it's a much different set so the men and the women share the same risk factors for catheter infections but the men when they have a urine infection in the outpatient setting it's often related to prostatitis so for many things like blood I wouldn't necessarily separate male and female but for urine infections it does make sense or for E. coli because often 70-80% of your E. coli are from the urine sometimes you want to bring in the epidemiology about the patients because it is relevant sometimes you don't male and female for wound infections probably pretty similar for hip infections probably pretty similar but it would differ for E. coli especially for urine it would differ by age group for the pediatrics for the elderly so these are simple epidemiology things that you can incorporate into your analyses so these are some examples of what an anti-biogram can look like and presentations on teaching how to incorporate into biograms the other three documents are certain reports I'm going to show you the Vietnam one because the Vietnam one they took Hoonet standard reports I showed you Hoonet is the standard reports they copied the ideas of the standard reports and put them into their web-based platform and then they put more things that I want to put back into Hoonet so what you see here in Vietnam is what Hoonet does and what this does that Hoonet doesn't do we will do it we are going to add some of these things in fact I have this version I'm showing you from July they have a newer version but I mean it's close enough so this is a sheet this is an automatically generated Excel file on this first sheet it tells you about data completeness 100% complete, 99% complete age, gender, 0% complete next year, that's sheet number one data completeness sheet number two different things missing by patients that we would recommend patient IDs do you have unique IDs or not intrinsic resistance they did simplify this because I told them it's good to simplify it this is about intrinsic resistance so this is a sorority of more resistance resistant to ampicillin that's normal it's a routine resistance that's to be expected sometimes you'll see some of these in red like here you see this is in red that's unexpected EnterBacter is usually resistant to this organism but this isolate is sensitive so it's helping you to find unexpected results Hoonet gives you this Hoonet has these microbiology isodalerts but unusual things basically what I'm showing you here is what I will call a data feedback report it's not an epidemiology report it's basically data completeness strange, unusual or important results next one suggested in a microbial they're saying you should have tested this we would have recommended we would have recommended this in this but you only did this in this it's what I referred to earlier as a core set of antibiotics I recommend you pick a set try to get everybody to do that set if they want to do more than that set that's fine this sort of minimal set of value reporting and statistics so this is about that and that's the end of their data check feedback report from July they have now taken my recommendations and incorporated them they haven't incorporated all of them into the Excel document that's more programming but they have updated it into their interactive web interface so when you go to the web interface this is very nice the facility can see their own results so we're not ready for a web interface yet in Ethiopia but you are ready for Hoonet standard report I have shown you the Hoonet standard report in the past I go to Hoonet 2020 I go to WHO test hospital I go to data analysis, quick analysis you see here four standard reports the first one is the one that I made 20 years ago Vietnam used that to inspire theirs these other three ones they each have their advantages and disadvantages we're going to start integrating them there are things I like about number one there are things I like about the other so we're going to start taking the best of both the old and the new approach what I like about the old approach is it's very concise it's just the minimum high level in a way that you can put into a Word document in fact you can choose your word and it'll go to Word the other ones have more detail and more graphs and you want both sometimes you only want the short version the long version good so basically I did this Hoonet standard report Vietnam copied it extended it and put into a web interface we will be putting those back into Hoonet so I suggest that you periodically look at these standard reports because in the next few months there are a lot of improvements I want to do here to basically help people make a data check feedback report which is what I showed you from Vietnam it's annual reports of common things that many people would like to do so I have just shown you the Excel document from Vietnam so you've received that I'm going to show you this other one from Japan John could I interrupt go ahead you mentioned Vietnam was ready to do a web interface other countries are not ready to do a web interface with their results when would you say is ready to do something like that well they did it so I don't know if it's used in Ethiopia there's a very widely used public health reporting platform called DHIS2 most of the Fleming Fund countries use DHIS2 I don't know about Ethiopia are you familiar with DHIS2 in Ethiopia well I don't hear anything but it's yes they do use DHIS2 maybe not the lab but the MOA so DHIS2 is used in many countries for public health surveillance tuberculosis, HIV gonorrhea influenza, diabetes cancer doesn't have to be infectious diseases so a lot of countries use DHIS2 for data management but not for antimicrobial resistance Vietnam decided to explore DHIS2 expansion to include antibiotic resistance the problem is DHIS2 really is not well set up for it so they ended up doing a lot of custom Vietnam programming so the Vietnam platform is 3 years old now over the years they have moved more and more away from DHIS2 features to their own customized features so what they did in Vietnam is very very nice but it's not transferable it's not easily it took a lot of expertise to do what they wanted to do things like first isolate per patient DHIS2 was not ready for that doing antibiotic interpretations they were not ready for that so let's see Sri Lanka also has a web based system for public health surveillance it's not DHIS2 and he said John how can I we're very good at SQL we're very good at the web system can you help me with Hoonet and how can we integrate it so basically they already had web based systems and they incorporated and they're incorporating or incorporated between it into it so basically I would it's most practical if the expertise already exists to do this if the expertise exists to do this it's easier because you're just adding another module to it you can either load in aggregate statistics this is basically what what glass does you're not submitting isolated to glass you're submitting your percent MRSA number tested web platforms you can either submit your isolate level data and have the web platform analyze the data itself that takes a lot of expertise that's what Vietnam is doing or alternatively you can submit to the web platform the aggregate statistics and display the aggregate statistics and then what Geneva does with glass of course those data are not yet available to the public but eventually they will be so they will be able to do everything but not antibiotic they have DHS too for most public health reporting they have made a nice web platform for antimicrobial resistance it involves manual data entry from nine pilot sites this is also another good example to quote they have nine pilot sites with three years of data they're now in the process to expand into about 40 hospitals those nine sentinel sites with a volume of data I don't know exactly what they do I'm just going to make it up approximately they collect the first 10 positive blood cultures per month the first 10 positive urines the first 10 positive sputums the first 10 positive this that and the other so it's a small subset of the data but they do collect the patient risk factors and the outcome and the diagnosis the therapy so it's kind of like a research protocol so it's basically routine ongoing specialized surveillance it's a very valuable program but it's a small data volume so now when they're going up to 40 hospitals they're going to keep the nine sentinel sites continue with a special protocol but for the other 30 hospitals they're just going to have them do normal HUNED normal data entry not do all of this additional work and they're finding problems that people are not measuring zone diameters they're not doing standardized testing so there's a lot of capacity building they started with the nine excellent laboratories in the country so on the one hand when you start a national surveillance program you often start with the most excellent labs the problem with the most excellent labs is you end up with the least representative results you end up with the ICU and the referred patients and the most resistance so it's this compromise on the one hand I want representative data but more important than representative data there's no microbiology there's no sense in having representative data if you don't trust that the laboratory knows what they're doing so what they did is they started with the nine excellent facilities worked with them to improve them over the years so those nine facilities are basically doing an ongoing special protocol small data volume a lot of patient details but for their ongoing national program involving 40 laboratories they're simplifying it to what is practical and routine to get them to move from recording R, I and S to actual zone diameters so this is part of a capacity building effort clinical reporting quality issue and as the data support they will use it more and more for epidemiology I want to describe this epidemiology sort of know what's happening the trends, the comparisons there's a special element of epidemiology that it's treatment guidelines treatment guidelines are often the worst application of resistance data it's one of the most desired you take, you want the microbiology lab data and you want to apply it to treatment guidelines the problem is the biases if your data are not representative you are overstating resistance you're telling the doctor that genomycin is 40% resistant but in the general population it might not be 40% it might be 10% on the one hand there's general epidemiology what are the trends, are there outbreaks is resistance worse here or worse there what are the most common pathogens how is that changing the general epidemiology then there's specific epidemiology related to treatment guidelines and that's where you have to be extremely careful about these biases so and you say these are reports about these accurately describe the data collected in our country they do not accurately describe resistance epidemiology in the country because of the biases so you say this is what we have there's a lot of valuable information in what we have, use it for what we have but it is just not good enough for certain applications or it's good but you have to it's all those caveats we found 40% resistance but I think the true resistance is maybe like 10% okay good so this is the Japan one this term goes back 20 years and this is their monthly feedback report it's partially an epidemiology report partially a feedback report and what you see here is not real data it's what it's 10 isolates these are not real data so just ignore the fact it's a real report for non real data good number of errors number of warnings and alerts so again that data quality component this is not meant to be shared outside of the stakeholders this is meant between the national coordinators and the facility to give them a report which covers some elements of data quality data completeness and some elements of epidemiology so this month and I like this this month you submitted 10 last month you didn't submit anything 2 months ago so it allows you to see a little bit of trends is it consistent for months a month if the numbers change a lot you kind of wonder if the data entry has changed like vacation this month you did 100 before that you did 100 this month you did 5 kind of suggest somebody went on vacation that's part 1 here what you see is interesting let me just zoom in on this okay good I'll just focus on some of these so number of submitted patients 10 patients 10 different patient identification numbers it might be 12 samples from 10 people MRSA 0 I mean these are not real data so just let's see that's fine okay let me just close my inbox I do get a lot of emails and unfortunately I do have to answer most of them okay so here what you see is very nice it's what we call a box plot and this is red is for your hospital within the context of the country in the country let's focus on let's say MRSA 0 cases well okay so in the country this is 7.96 is the median value the the quartiles or the 90% quartiles are like 1.15 that's the range the range is the lowest hospital is 1.15 the highest hospital was 38 the median was 7.96 and then these are the 5 and the 90% quartiles so you get to see not only where you are you see where you are in the national picture are you higher than the average lower than the average some things are to be expected if you're the university hospital usually you have more samples more resistance more MRSA and that's normal in pediatric hospitals you usually have less so you want to see am I where I think I should be so this is about some of the high priority pathogens VRE where are you what is your number and where are you in comparison to everybody else these are about important resistances these are just about the species staph aureus fepsiola etc so just like I showed you in Vietnam they have one about the organisms and one about the resistance of those organisms and then some grass monthly change these are not real data so in January they had 0 February they had 0 but these are only MRSA so you can see that trending over time part of that is for data quality checking part of that is for epidemiology for example a possible outbreak that was by resistance this is by species staph aureus this is major pathogen resistances by ward you see the different wards here at the top this one is for respiratory this one is for urine this is for feces I find that these things go on way too long this one is for blood this is for cerebral spinal fluid this is for other and then this is a comment in Japanese on how to interpret the rest of this table so that is a nice example of what they do they have been doing this for 20 years I am going to go to the last example which is what I want to do is take the content of the Vietnam and Hunet report but it is put into the format of the Australia report the Australia report has way too much text talk talk talk talk talk talk talk talk talk talk talk explaining what they are doing talk talk talk talk great here is some data this is actually the Hunet sample database so I ran this on my Hunet data 622 isolates this is the WH test data from January 1995 622 isolates in January it is the Hunet normal data go down further 821 blood cultures Staph aureus 12 blood cultures from 11 people and then you start to see not everything you see a subset of pathogens and resistances methicillin resistance enterococcus so I am not that impressed by the content that they have here it is very basic content it is similar to content you have seen in the others but I think they have done a very nice job of putting this into a stakeholder report so it is great if you come up with the content that you want for Ethiopia but I am trying to package this in Hunet so that you could simply go to Hunet I am going to say Hunet standard report and let me go to I am in the wrong folder let me change folders and WH test database so I will just show you how far we have got and I am going to say export this to word analysis 1 analysis 2 analysis 3 since we started having our conversations since May this part of Hunet is much faster we made a lot of improvements over the summer especially for some of these descriptive analyses so now the slow part is actually just making the word document ok it is finished if you want to open the word file now yes I do word is now slowly opening there it is so here you see what you would see on the Hunet screen file summary section B percent completeness section C so you are also seeing the concepts that you saw in Vietnam number of organisms by month key antibiotic resistances this one is microbiology alerts section E section F what is section F low frequency results does not matter configuration comments etc so I like the content but I prefer the Australian formatting so in the next several months you are going to see an improvement in the content and an improvement in the formatting so eventually it would be nice if people all over the world did not have to create from scratch their own reports what we would like is that we give them a nice Hunet standard report that they customize they improve they add their own text we are trying to at least help you with the data preparation so I am going back to the email which I have already closed so I have covered the first two agenda item points the first was on the anti-biograms the second was on these template reports that we just looked at and the third thing was basically leave it up to you still a little more time so before new topics are there any questions on what I just presented if you are saying anything yes yes somebody started to say something what did you advise is it good to start with excellent facilities or many weak facilities to start the surveillance data collection let's see it depends on a number of factors in part your time so in terms of the annual report for epidemiology that you want to publish data you want to submit data to glass you want to start with the best facilities also keep in mind that the best facilities themselves have trouble either in terms of data completeness like age and gender antibiotic test practices and even if they are doing good quality testing in hospital A they may be doing different drugs in hospital B we are trying to get those to standardize so I would have one special focus on the best facilities to make sure they are really doing good quality testing with good quality data management I would also recommend if your time and resources permit to work with as many laboratories as you can manage in visits and in conversation and getting their commitments if you don't trust their data don't include their data okay include their data as you saw in the case of Vietnam some of the things were purely descriptive how many bloods did they submit whether your good quality lab or bad quality lab those numbers should be accurate we did five blood cultures whether good blood cultures is another matter so for those I will refer to that concept as diagnostic stewardship are they taking samples if you find a facility that does no blood cultures there is some problem there so you could include them in those kinds of tables that are not dependent on the laboratory quality so number of bloods number of urine number of data submissions completeness of data male, female, gender, inpatient, outpatient specimen, date specimen type so all facilities irrespective of their data quality these are meaningful things to measure and to share with the network people if you believe that they were doing a relatively good job on organism identification you can also present the statistics on that they found 20 E. coli they found 100 staff they found so many staff in blood so many staff in urine and then the antibiotic results if you don't believe their antibiotic results don't include them in those tables so the answer to your question is in the ideal world we would have every laboratory in every country participating in their national program but it depends on their desire and their willingness it depends on your ability to effectively communicate with them in Argentina there are about 130 laboratories that use HUNET 96 of them have been invited to the national network the other facilities have not been invited for a number of reasons but the most common reason is the national center said we don't have the money and the time to support them because it's not only about data collection it's about our ability to visit them involved with them in the EQA program or or sometimes data ownership sometimes a private laboratory does not want to share their data so so it's an example where there are a lot of places that use HUNET and they are included in training activities so they're not in the national surveillance program because to be in the national surveillance program you have to have acceptable performance in the national EQA program and if you are not included your data are excluded for that quarter and they give you feedback and then they monitor your improvement over time so in theory you want all the labs veterinary, environmental, human, animal in the national program of course it's not realistic in terms of your time availability so think about what is available what can you accomplish with your time and I would basically sort of as I said categorizing them to three types all laboratories from which you collect data you can report on their data volume how many bloods, how many urine whether it's a good lab or a bad lab as long as their data entry is complete you can still see how much testing they did which allows you to see how many blood cultures per hospital bed for example the second one if you believe they're organisms if you believe they do a good job on organism identification then you can include them in the organism summaries and if you do not believe they're antibiotic results or if they do just bad combinations of antibiotics you don't include them in the antibiotic tables so I have not given you an answer there's always the ideal and there's always what is realistic if you want to if your focus is to put out a published report with semi-reliable if you keep in mind the excellent quality laboratories still have major problems of biases it's not the laboratories fault that people are not sending samples people do not send samples for many reasons the hospital doesn't have money to pay for it the patient doesn't have money to pay for it in many places they don't send samples because it takes the lab 2, 3, 4 days to get results back that's the nature of microbiology and cultures it takes time so my patients in front of me I need to make a decision now they're in the emergency room so I'm not going to send something to the lab because even if I get a result the patient's gone so there's one reason why people don't send samples because of those delays they'll only take the sample if the patient's being hospitalized it's going to be around for a few days some places do not send samples to the laboratory because they simply do not trust the laboratory a lot of times physicians they say the laboratory has been reporting to me vancomycin resistant staph aureus an uninformed clinician will believe that the laboratory knows what they're doing the informed clinician says this laboratory does not know what they are doing and they might just stop sending samples what you really need is a good communication to address that so keep in mind that even excellent quality microbiology laboratories do not have representative data so all data is valuable but if I don't trust the data I don't put that into a public report on a website I just put things that I believe that I think are reasonable on top of that I still always put these biases the caveats and the limitations also I view this as a work in progress this year let's just get things off the ground let's just start with the 9-pacilla like in the case of Bangladesh you know let's just try to build a good system for data management so that we know what we're doing let's try to build some trust and confidence let's get some things out that are incentive for other people to join and receive some good results and with good results it also helped to track funding and money and this year let's just try to go for 9 and then next year let's try to do a few so that's what they did in Argentina for many years for several years they tried to incorporate between 5 and 10 labs per year because that's what they felt was they needed to visit the laboratory discuss them, train them so they ended up taking a lot of the advantage of Latin America has not and unfortunately what they have accomplished in resistance containment because they're very good on the microbiology side they're not very good on the antibiotic use side but that doesn't mean nothing has been accomplished, what Latin America has accomplished is much better microbiology capacity people often blame the microbiologist but did you really make any difference in resistance, did you prevent hospital infections, did you make better treatment guidelines, do patients get the right drugs, did you decrease antibiotic you can't blame the microbiology surveillance network if those things didn't happen because you need the pharmacy involved, you need infection control involved, you need the government authorities involved, the hospital pharmacy can control antibiotic use in the hospital but it's only the government that controls antibiotic use in outpatient pharmacies so we can blame the microbiology lab for not doing this as part of the surveillance program but you can blame the microbiology laboratory if they don't reach out to these other groups, they need to reach out to the clinical societies the pediatric society the surgical society the national hospital accreditation groups because the microbiology data is only one element of a resistance containment strategy, you do not need any data from the laboratory to tell people to wash their hands you don't need any data from the laboratory to say don't take antibiotics for viral infections so the laboratory data have great value but it's only part of the solution for intervening and preventing infections and better using antibiotics while you think of more questions I'm going to go back to my email just to see some of those comments that Fern noted and if I look for if I want to search for the word Fern I can look for the word Fern it makes oh, let me go back to my inbox from Fern and if I go to the it's a nice little thing in Outlook, it just helps you to narrow down your search so if I said Fern I get all these messages if I say from Fern, it just helps to simplify the search so I'm going to open up her email about today's agenda let me maximize this if I can let me go in this and zoom in a little bit I was just doing that while you think of more questions if you have no more questions, I'll take a look at this but we have half an hour, what would you like to talk about next? John, if I might I would prefer to leave the floor to our colleagues in Ethiopia but if there's silence I'm going to ask you if you could not in very simple terms just explain again what is phenotyping, we followed your example but please just in simple terms describe phenotyping sure yeah and another before I come into that just some of these ideas and reports keep in mind that in the first one, two years of any surveillance program the focus is really to try to build a strong program, describe the data collection, the quality measures all of these things and in answer to the questions from your colleagues where's the epidemiology, where are the patient level details to tell them we're doing this, we're just getting this off the ground we're focusing on the completeness their management, laboratory test quality what you want sounds great we're not there yet but let's have this discussion as an ongoing conversation next year let's try to do these things the following year so that's just one strategy right now you're doing something important and novel and in the first two years the most common things you find is there are big problems in the data and we need to focus on that once the data, once you can say I do trust this, I don't trust this then you really have the basis for the 5, 10 year, 20 year program also I did ask on a previous call, I do forget is Ethiopia a member of glass I do recommend that you do that because it does open opportunities for training if I say WHO glass resistance countries and if I go to country participation so I'm on the WHO website so here you see a map well the map is enlarge map enlarge map so Ethiopia is enrolled has Ethiopia submitted any data to glass because enrolling doesn't mean you have to submit data a lot of countries say I wanted to join glass I'm not ready to submit data yet but glass is not only about data it's do you have a network, do you have a national strategy, do you have a national plan who are the main context for further communications, do you have an EQA strategy, all of this is glass and then on top of that there is the official data element about resistance now it has been renamed to glass AMR I think glass is incorporating a lot of new modules glass per isolate level fungal fungemia, you know candidate in blood they have another one on consumption they have another one on whole genome sequencing they will be incorporating the one health aspect of food and animals so I'm glad that you are a member of glass but it's good if you but I don't know who knows about it so WHO is putting more and more resources and development into this that you can learn about from here also if you look at the glass reports what's interesting in the glass reports is they are not putting any resistance data into the glass report because they said there are just too many problems we do not believe the data so there is really a wonderful report this is the most recent report let me just take a look at Thailand for example acknowledgment Thailand upper middle so they are categorizing here they are reviewing an example there culture positive culture positive oh this is a special project called EGASP EGASP is for gonorrhea expanded gonorrhea surveillance so that's a special project let me continue to look so here is the page on I lost it again it doesn't matter which country I look at let me just pick the next country on the list well let me see if Uganda is here well let me know Ethiopia that's glad it says Ethiopia low income level Ethiopia population 112 million you are implementing antibiotic resistance you are not yet implementing these new ones of antimicrobial consumption HIV TB yes malaria yes price cycle is the environmental one for sewage water samples EGASP is for gonorrhea you have nine surveillance sites I hope this information is correct somebody from Ethiopia reported it there were seven hospitals two outpatient facilities nine labs perform AST the national reference lab EQA is provided 2019 data call there were four surveillance sites two hospitals and two data I'm sorry two hospitals two outpatient facilities so what you see is data were reported on the green pathogens you reported as an age gender less than 70% because to do that you need the data of admission data on the number of people that would include the negatives and that's it so there's nothing in this report about the resistance data because Geneva knows the data are not good enough the resistance data are not comparable or complete enough to make a data analysis report hopefully in a few years they will change their mind and hopefully in a few years they'll start choosing a subset these countries we are ready to report the resistance data okay those are things I had to say and then John sorry there was a question in chat that I didn't see so let's look at that question basically it says can we use regional anti-biogram data for empirical treatment if the region has the same demographic population as is always the case the answer is well it depends let's see the answer is that all of these things are useful it is useful to know what is happening in the world what is the general trend of imipanum resistance or vancomycin resistance or ciprofloxidin resistance so it's useful to know what's happening in the world it is more relevant to know what's happening in Africa it is even more relevant to happen what's in eastern Africa so if you saw some nice reports from Uganda or Kenya there's some surrounding countries yes that is useful and yes I would find that valuable in designing my national and local treatment guidance but always with caveats first of all their data probably has biases so the same things I was telling you about using your own data how to interpret your data they have the same problem so just because they published the data for Kenya does not mean that the data are not biased they're probably also biased so all the local biases exist in any foreign data that you may want to copy so that's why you really do want your own local data but even think about it in the same city so for example I'm at the Brigham and Women's Hospital we are connected by a bridge to the Children's Hospital we are across the city for the Massachusetts General Hospital so the resistance patterns in all of these laboratories is interesting and irrelevant especially for community path engines covering the same population I would suspect resistance rates for community infection should be pretty similar if you if you live in this part of Boston with your medical insurance maybe you'll go to Mass General Hospital but maybe instead you'll go to Brigham and Women's if it's a community infection the resistance rates are probably pretty similar irrespective of which hospital you go to but I do want to qualify that we also have Boston City Hospital and Boston City Hospital is more of a low income population low income populations also often have higher levels of resistance surprisingly to many but low income populations often have more antibiotic use in part because they have more infections more bronchitis, more smoking as well as more inappropriate use of drugs you know so so the data so for outpatient infections I would hope that the resistance rate would be similar but that won't always be true if they don't serve the same patient if the different patients go to different hospitals the information is still valuable especially if they show the same resistance trends and if the trends are consistently changing over time we found in our area that penicillin resistance in Staphylococcus aureus is going down in my hospital but also going down at Mesh General Hospital suggests a community thing where resistance really is coming down it's very different when you consider hospital infections because that has a lot to do with the hospital's use of antibiotics and the hospital's hygiene issues so you may have two identical hospitals across the street from each other and one hospital has bad antibiotic use they use a lot of hemipenem, a lot of ancomycin and they have bad hygiene so a patient gets something a lot of in-hospital outbreaks the other hospital across the street might have the same patient demographics but better hygiene better infection control and lower resistance rates so to answer your question is the regional antibiotics are relevant it is good to know what's happening it's a very bad issue for that hospital it's a very bad issue for that hospital I keep on saying hospital but of course I also include community the data from that hospital laboratory includes community samples so the data from laboratory one is valuable to the people in laboratory two a lot of the findings are probably similar this exists in my country for them it's 10% for me it's 15% so the numbers won't be the same but it is relevant but you cannot make the conclusion that their data it does not mean I don't need my own data so in the short term maybe I will use the data from the other facilities and especially in a country like Ethiopia many parts of the country have no microbiology data so the communities with no microbiology data would benefit to know what resistance rates are like in Addis Ababa the resistance rates in Ababa are probably higher than in the rest of the country and as long as you say well resistance in Addis is maybe 20% maybe resistance out in a rural community maybe 30% so yes all of these data are useful but you cannot say that my situation is different if you have your own data you compare your data with the other people's data and see what is the same and what is different or the patient population is the same as antibiotic use the same or the hygiene issues is the type of care provided in Boston we are a tertiary care hospital we have other hospitals they're community hospitals the same community if the patient has chemotherapy and sepsis they'll come to my hospital if they have pneumonia and sepsis they'll probably just go to one of the community hospitals so if you have no data of your own definitely use the regional data from within your country from surrounding countries and just keep in mind that those data themselves are probably biased and those data are useful but they're not necessarily representative when you make treatment guidelines part of what you want is to look at these treatments look at the resistant data but you also need to keep in mind the biases I've mentioned that many times you want to look at the costs you know imipanum is a better drug than ampicillin of course well ampicillin resistance rates are so high let me choose a better example imipanum is usually a much better drug than sceptra axon they're both good drugs but it's also more expensive is one problem but it's also a reserve agent you want to save these reserve agents for the people who really need them so when you're trying to make a treatment guideline you're considering resistance rates you're considering costs you're considering reserve agents and the desire to preserve reserve agents for the sickest people today like the ICU patients the emergency room patients with sepsis you want to use them for future generations you want to distinguish between the complexity a simple urine infection you might want to go with genomycin it works 80% of the time probably fine but if it's a complicated urine infection with a sepsis you may want to use amicacin so you also want to consider some of these patient considerations about severity of disease if a patient is a non-complicated urine infection they're not going to die in the next 24-48 hours so give them the cheaper drug that's usually effective if the patient is not getting better if the microbiology lab results come back then switch to another agent so for non-life threatening infections the immediate goal is not saving the patient's life the immediate goal is just decreasing suffering decreasing length of disease and in a few cases preventing eventual death but if it's a non-complicated infection you do have some time on the other hand with sepsis, meningitis life and death is there and that's where you want to go up to the next level what else about treatment guidelines, there's so many things that go into it for example, if you have a wound infection wound infections are usually skin, I'm talking about skin wound infections are usually polymicrobial for most wound infections what you need is basic good hygiene soap and water surface disinfectants vasotracea and neosporin and broad-spectrum if you decide to give a systemic antibiotic and oral antibiotic give something broad-spectrum like Cephasolin if you find sputum sputum is multi-polymicrobial so often you don't want to use the laboratory data to make decisions on wound infections because usually you just want broad-spectrum coverage and good hygiene for something like pneumonia the problem is you want a good quality sputum sample without a good quality sputum sample you end up with what we call spit samples or saliva samples meaning you're not getting the infection you put a swab in somebody's mouth or you ask somebody to spit to put sputum into a cup they go a lot of times you don't get sputum, a lot of times you just get saliva from the mouth that tells you nothing about the patient's pneumonia and what you get is just basically a sample a microbiology sample with a lot of oral flora so treatment guidelines depend on a lot of things besides the resistance data so yes one of these considerations are the regional guidelines from other countries the regional guidelines from within my country but treatment guidelines are not determined solely by the resistance data you need to keep these other things in mind for something like simple watery diarrhea simple watery diarrhea is usually viral and doesn't need treatment simple watery diarrhea is very often sominella and campylobacter and that also usually doesn't need treatment what it needs is oral rehydration therapy it needs fluids, it needs food it needs to replace the fluid that is being lost and for simple uncomplicated watery diarrhea antibiotics are not recommended and that's the treatment recommendation give oral rehydration therapy for complicated diarrhea like chigella, dysentery fever, that's when you want to use antibiotics so all of these different thoughts come into play when you're making treatment guidelines WHO does have a recommendation for example in the case of meningitis gonorrhea and malaria I think what I'm saying is approximate correct if you have the country's national first line agent is Fanzadar, chloroquine or isinitis or whatever it is you have your first line agent if the first line if the resistance to the first line agent exceeds 5% then you should think about changing to a different first line agent that's how we went from chloroquine to this to that to the other if the resistance exceeds 5% that don't consider that as a good first line agent in a life-threatening disease in a non-life-threatening disease like a simple urinary tract infection 10% resistant, 20% resistance 30% resistance especially in a biased sample is reasonable because the patient is not going to die immediately you do have some time in a life-threatening disease like meningitis gonorrhea not life-threatening immediately to most people but we have what's called the in vivo in vitro correlation just because the laboratory says resistant does not mean the patient will get better having to do with dosage and a severity of disease if patient has well let's say the patient is a sensitive sample and they get an appropriate antibiotic and the patient dies a half an hour later the patient dies a half an hour later it doesn't matter if the bacteria was resistant or not you know it doesn't matter if the patient got an antibiotic or not there simply was not enough time for the antibiotic to have any impact at all so in this case it doesn't matter if the bacteria is resistant or sensitive if they got an antibiotic and they died 30 minutes later the patient was or the disease state was too far advanced there are other cases where the patient clearly has a resistant infection but the patient gets better anyway the isolate is resistant to genomycin you give the patient genomycin and the patient still gets better why? because patients get better most of the time on their own anyway if it's a non life-threatening infection eventually they'll cure their own infection even if it's resistant bacteria it means the antibiotic is not helping or it's not helping a lot so this is a case where the laboratory says resistant but the patient got better there are other cases where the laboratory says sensitive but the patient doesn't get better and that may have to do with a variant of disease or maybe it's an abscess so there are cases where the laboratory result in the clinical outcome don't agree with each other that's what we call the in vivo person in vitro laboratory correlation but there are some things where you do have to pay attention if the laboratory says gonorrhea resistant that's important it says if the laboratory says resistant to penicillin gonorrhea, penicillin will not work this patient will not get better on their own if the laboratory says sertraxone resistant gonorrhea this patient will not get better on their own so these are cases where patients do not clear on their own the laboratory says you have to believe them if it says resistant do not use the drug the patient will not get better the case of meningitis, the case of tuberculosis the patient of gonorrhea it's extremely important that the patient gets treated correctly the first time because otherwise the patient may die well for gonorrhea for TB and for meningitis or the patient may not die but they're going to continue to spread their infection to other people somebody with gonorrhea typically will not die obviously in women it can cause pelvic inflammatory disease it can cause death but the most common thing in gonorrhea is often asymptomatic or dissymptomatic with continued transmission so in both of these circumstances you need to trust the laboratory the laboratory says resistant let's use a drug where the laboratory said sensitive so this comes into play with regard to treatment guidelines when you have an excellent in vivo in vitro correlation where life and death or transmission is on the line you need to go with the laboratory data something like urine infections you want to be informed by the laboratory data but you also keep in mind most people will eventually get over their urine infections or you at least have some time to change their antibiotic after 2-3 days this happens very often in the United States and elsewhere women go to the doctor the doctor gives the woman an antibiotic does not take a sample and the doctor says if you're not getting better in 2-3 days come by again we'll take a sample we'll change the antibiotics and we'll take a sample so they're just being informed generally by overall statistics but you're not worried about an immediate death so as I said for meningitis, malaria, HIV you want resistance to the first line H&B under 5% because you believe the lab and you need an effective therapy however, that 5% is assuming you're talking about an unbiased sample so maybe the true level of resistance is 3% and this drug is working perfectly well but if you have a biased sample of the treatment failures and the sickest people in the hospital the laboratory may say it's 15% resistance so just because the laboratory says 15% resistance does not mean that you're really above 5% so people say when should I change my treatment recommendations at 5%, 10%, 15% for life-threatening infections or for gonorrhea because of the transmission issue people often go for 5% if resistance exceeds 5% to your first line agent change your first line agent but that statement is only true if you have an unbiased sample so sometimes what people will do especially in the case of something like malaria in malaria they're not doing laboratory testing they're looking for treatment failures so if they have 100 people with confirmed malaria and they give these 100 people first line therapy, fanzadar or testament or whatever they find out that 100 people that 20 of them fail therapy this does not change a change in the treatment recommendation immediately they want the resistance to be under 5% and if they want the treatment failure rate to be under 5% if they find the failure rate is 20% what you want to do is investigate why and there are a number of reasons that patients fail therapy one is because of resistance there are several other reasons what one is your data might be biased most patients treat themselves for malaria at home the only people who come to the clinic are the ones who already treated themselves so you may have a bias in your 20% it's not representative or they may not be taking their pills maybe they didn't get pills or they got the pills they got the prescription but they didn't get the pills or they only took 2 days of pills and then they stopped taking them or their pills that they buy are bad quality there was a very good BBC documentary from Nigeria and India where they spent half an hour in the documentary half an hour in Nigeria and half an hour in India what they found in Nigeria they said a lot of our drugs are poor quality the malaria drugs have half the potency a tenth of the potency there's no drug in it that was the first half of the documentary the second half they went to India and they went into some of these pirate companies and the journalist had a secret camera and they went into one of these pirate drug companies and the drug company and the person said I want to buy some pills and they had to have the shipment of 1000 bottles of vipanam and the company asked 2 questions how much antibiotic do you want us to put into the pill question number 1 question number 2 how much antibiotic do you want us to put on the label or those are 2 different questions on the label put 100 in the pill put 10 so this is another reason why patients with therapy so as I said with malaria if it's 20% treatment failure I'm not going to change my treatment guideline immediately, but it does prompt the need for what they'll often do like what's called a lot quality assessment program. They want to evaluate where the drugs of good quality, where the patient's representative, did they take their pills and where they compliant? So they're looking at all these factors. So if resistance in that example goes above 5%, they do a special, if treatment values go above 5%, they do a study. And then the study may suggest that the problem is resistance, or the study may suggest that the problem is just poor quality and people are not getting their drugs and they're not taking their drugs. So this whole idea of treatment guidance is it's a complicated subject unto itself. And it's unfortunately the most common thing people want to use resistance data for, but it's also the most problematic in terms of biases and the in vivo, in vitro correlation. I do talk a lot, I hope you don't mind. Thanks a lot, John. This is great. I mean, we've tapped into your experience far beyond the Hoonet platform. So we appreciate that. We're approaching the end of the two hour session. So, and since this is the last session under our coordination, under McKell and Fern's coordination, we just want to make sure you are aware, Gabriel and everyone else, that we do have certificates. So if you would like to have a certificate, you know, we would just need your name and your institution. And you can, you know, simply respond to McKell or myself and we'll send that to you. And secondly, since we do hope to continue working with John on a series of shorter targeted videos, we will probably reach out to you in very soon and just ask you, you know, if certain sessions were particular relevance or interest to you, we would like to know that for, you know, to guide our future work, hopefully with John. And now I'll turn it over to, I guess, McKell and also we should see if there's any other questions from the participants. So over to you, McKell. Yeah, thank you very much, Fern. So before we wrap up, does the Ethiopia team have any questions on what Fern just said? No? No, of course, no. I have no, no. I have, can you hear me? Yeah, yeah. I hear you, I hear you. This is actually, you don't have time for the question, but I have some reflections. One, yes, we will also raise if we need any session, specific sessions. Once we go through all the sessions and refer all the things, we will ask actually for another session for a specific session that will be a good suggestion from Fern. And the second one is if John, if you can really give us the major references links so that we can use those references for future use instead of every time asking you, it will be also good to give us major resources. Okay, Mika, I could do that. Well, as part of the notes, will you, I don't remember what I talked about. If you note down some of them, I just did a lot of Google searches I could try to remember, but in the notes, if you maybe just put down which ones you would like to me to guide you towards, I'm happy to provide those. Yeah, we will do that. Yes, and please, we do have detailed notes with the links that John mentioned, so please do access those notes. And, Miquel, you can send out the link to the notes and the recordings again. Yeah, so one comment is, of course, you are in the unique position that you're in the same city as Africa CDC. And Africa CDC in collaboration with the WHO is working on strategies for surveillance, standards, quality, and potentially, partially between IDDS and Fleming Fund, and other ASM and other activities in Africa, all of them are collaborating with Africa CDC, I believe, with a general idea to build a platform around the country. So I suggest if I don't, do you have very close connection to the resistance group at Africa CDC? Yes, we have a close relationship and also they're supporting the antimicro-resistant surveillance starting from this year. Great, because I think that you can work library together, they may have some ideas, and you can give them on site, you know, they don't need to set up a big meeting, they can just visit you and you can get your, give your input so that your experience can help to inform them, and their ideas captured from around the continent, you could also pilot in Ethiopia. I did try to visit them in my last trip to Ethiopia, but they had some very large important meeting and they were unavailable, but definitely that will be a priority for me on the next visit. I do want to take advantage of this time to thank you all for your patience. Remember how bad the sound quality was in that first video and your interest. I do wish you luck with all of your further activities in capacity building and surveillance, and it will be available, you know, not every week for two hours, but you know, start by email with any questions you have and as needed, you know, we can set up a session, you know, for specific data analysis points that you would like to review. Well, we all thank you, John, for for this session. So I think the quality of the content is it's been always really good. I personally have enjoyed them very much. I hope everyone has enjoyed the sessions as much as I have. And yeah, just a sincere thank you for helping us for going along on this right and to remind everyone that we have a YouTube channel where all 14 sessions, all 28 hours have been unedited. They're fully just posted on the website. We have the notes that we've been sending. I'm trying to get all of the notes and all of the files that John has been sharing with us on one website. And maybe I'll put the link on one of the comments on the videos on the YouTube channel. IDDS is working on an IDDS website where we want to kind of consolidate all of our content, but we're still waiting. We still need some some approvals for that. But yeah, it's been it's been really, really interesting and I'm happy we have really good participation throughout all of the sessions. And just don't forget to reach out to me or Fern. That's for the Ethiopia team. If you need anything, just reach out. You have our emails. Just let us know. And thank you everyone. Regarding the website, you have the videos there, which is wonderful. Would it be possible to put those Word document notes or PDF document notes onto that same channel? For example, a lot of people I think in theory would be interested in it. But two hours, 14 sessions, a lot of people I think would be more interested in just downloading the notes. You might want to review the notes about things you do and do not want to share with the general public. But you may want to consider in addition to these long two-hour videos, that first I would have the patience to sit through that more than once, maybe put the notes in that same channel or a link to the documents. Yeah, but you have to store them somewhere. And that's the issue. If we put them on SharePoint, that's our IDDS SharePoint. There are some logistical issues that I'm working through. Really what we do need is an IDDS website where we can host all of this content. But yeah, I'll try. I'll find the solution to put those notes. And I agree it will be really helpful. So I'll work on that. Content and then it allows people to get a lot of the value without spending two days nonstop. Absolutely. No, the idea is to work more on the videos and even create chapters so that people can go right to the areas that they're interested in. No, that's something that we're going to be working on. Wonderful. For example, on this call, I did a lot of times it wouldn't be helpful to other people because of the interruptions, the questions. But a lot of what I said on this call, I think, would be interest to a general audience. You may want to start clipping it. If I talk for this on Envivo and Vitro, you may just want to clip those out as something for future development. Okay, we'll keep that in mind. Well, thank you so much. Thank you, everyone. And I hope you all have a wonderful day. Thank you, everyone. Thank you, John. Thank you, IDDS team. And we are expecting, as John commented, those down-level documents, Word or PDF, whatever. Thank you. Nice. You see all the thanks on the chat. So thank you for those chat messages. Thank you, John. Really, it was interesting. Thank you very much. Have a great day, everyone. Bye-bye. Likewise, John. I thank you very much for your expertise in sharing that so freely. Thank you. Thank you, Martin.