 Today, I'd like to focus on two things. One is partially a review and one is partially new discussions. The part that's a review is it's so important to understand what your laboratories are testing in terms of which antibiotics. Are they testing appropriate antibiotics? Are they missing important antibiotics? Are they testing antibiotics that they really shouldn't test? And also, are they testing the same antibiotics across all laboratories? The laboratories do not have to test exactly everything exactly the same way, of course. But the more similar are the test practices, the easier it's going to be for you to do comparisons in national statistics. For example, if only one laboratory in the country reports amiccason, then you can have the statistics for amiccason, but they don't really represent the country. They represent that one laboratory. So it's good to have some degree of standardization of at least the core minimal set. There are also examples where some laboratories will test seftraxone, some laboratories will test sephotaxym. Both of them are perfectly fine. They're nearly identical. But for purposes of national statistics, it's just easier if people agree on one or the other. I don't want to see the national Ethiopia statistic for sephotaxym on one row and the national statistics for sephotaxym on a different row, because it's basically the same drug microbiologically. So we've discussed this before, especially one facility at a time. Once I did a comparison of the test practices between facilities, it was at the end of a call. I went through it quickly. So I'm going to start with that, to see what are the different antibiotics tested by the different facilities. And I'm going to start with the gram-positive staph aureus. Because people on the gram-positive side, there are not a lot of options. People are testing penicillin and sephoxidin for staph aureus, sephrocochromoxazole. There are not a lot of different options. Some people test sephro, some test levo. So there are a few options. But most of the time, people pretty much agree on most of the important things. That's the old review. The new part is the HUNET analysis called resistance profiles. It is one of HUNET's most valuable analyses, most interesting analyses in a number of ways, for outbreak detection, for tracking individual strains and their epidemiology and their movements. I mentioned on an earlier call between human, animal, and food isolates. The best way to know if the human, animal, and food isolates are the same is with molecular typing. Whole genome sequencing, pulse field gel electrophoresis, MLST. But in practice on a day-to-day, week-to-week basis, it's not practical. It takes time, money, equipment, expertise. And even if you have all of that, you're not going to sequence everything. People often use these molecular technologies to confirm suspected outbreaks, but they don't use molecular technologies, usually, to find possible outbreaks. It's used more in the confirmation stage than in the detection stage. Of course, in salmonella, this is now changing. A lot of public health labs do start whole genome sequencing from the very beginning. But in a general sense, most people use the molecular method to confirm outbreaks, not to find the outbreaks. Okay. So first, we'll go with test practices, and then we will do the multi-drug-resistant profiles. Okay. So let's see. I will go to HUNA, and I'm going to go to Ethiopia, all hospitals. As I've mentioned on a previous call, that can be all hospitals, all antibiotics, or it can be all hospitals, a subset of antibiotics that you have more interest in. Let me just go to normal data analysis, so go to data analysis, data analysis. These two buttons in the upper right, we move them up a little just to give us a bit more space, so you see these small incremental changes in HUNA. I'm going to go to analysis type, and I want to know what they are testing for staph aureus. The easiest way to do that is to do an analysis of staph aureus and see what they are testing. I can do a detailed format. I can do a summary format. Either one will serve this purpose. I'll start with the detailed format, because that's what we have done more often. I will choose all the antibiotics, as you can see at the below. Percent RIS, details, all antibiotics, okay. Organisms, I just wanted my staph aureus to keep it nice and simple. Later I can do E. coli or club C. elli, choose an example of a gram negative. I click okay. Data files, and there's my Ethiopia, all data files. I'm going to move my go-to meeting controls out of the way, if I can. It's a little bit annoying that I'm glad that you cannot see it, but I can see it, so good. Now I can see my screen. Good. Ethiopia all, I click okay. I click on begin analysis, and here we've done this many times. This is the percent resistance at the national level, well, by the national level, I have here data from four hospitals. What I want to look at now is number tested, and what I can see here are two things. I can see what antibiotics are they testing, and how often do they test those drugs. I see they have a lot of data for SXT, clindamycin, and erythromycin. So three of the antibiotics have a lot of data. And then four antibiotics, Penn, OXA, Fox, Cipro, have a good amount of data, followed by genomycin and tetracycline, which have a lot of data, but it's still the minority. And then we have antibiotics that are very infrequently either tested or reported, and also they could be typing mistakes. Like meri-pennum is an inappropriate drug. Pyrprisilin is two things. It's inappropriate. It's also a very old drug that nobody's testing anymore. Norflex is said, daptomycin, that's a good drug. It's a reserve agent. Most people do not have, in low-research countries, most people do not have access to this drug, or they do, but it's an expensive drug that they don't routinely test. Here we see azithromycin twice. One of those would be the azithromycin disc. One would be the azithromycin MIC. Clarithromycin, nitroferantoin for urinacellates, Linnezolid, there's nothing wrong with Linnezolid. It's a perfectly valid drug for stephorius. It is a reserve agent, so a lot of places do not test it routinely. We tested it routinely because it's on our Vitek MIC panel plate, but if you're doing manual distribution, a lot of people do not test it as part of first-line testing. Venkomycin, keep in mind that Venkomycin should be tested by MIC and not by disc. Ferrumvenical doxycycline. So what we see here are basically, I'm going to say seven, one, two, three, four, five, six, seven antibiotics that we have data for over half of the isolates. I'm very happy with these three, SXT, clindar, erythro, because that seems to be almost every isolate. And then we have these four others where over half the isolates are tested. So those are my seven drugs I want to focus on. Good, so that's the two things. I can see what they are testing, and I can see how often they are testing them. Good, good, I'm going to click on, if any questions, just interrupt me. Okay, I'm going to click on continue. Analysis type, I'm going to repeat that exact same analysis, but just the summary. Well, actually, let me just show you one more thing on that particular screen. On that particular screen, you do see here a column that says number. I'm going to click on the column that says number once to sort it, and number again to sort it backwards. So this is just re-emphasizing what the graph already told me. These three organisms are tested basically all, these three antibiotics are tested basically all the time. We have 229 erythros, 228 SXTs, 223 clindus, followed by these four, 183, 179. So there's a drop off. So what I would like to see going forward in the future are fewer antibiotics, but more consistent testing. I would like to see the same seven antibiotics always tested to get the best information for the physicians and the most complete information for the physicians, and also the most, the best information for epidemiology and tracking. I'm not going to report statistics on in this example. I'm not going to report any statistics on like when as a lid because, you know, they're almost never tested. And then you see the tetracycline 99 times gentamicin 64 times. And then finally at the bottom, you see, oops, I'm going in the wrong direction. Sorry to sorry and to let me move back to the left. I'm going to go down to the bottom of the table. And here you see noreflexicin 1s, meropenem 1s, and it could have simply been a typing mistake. Tobromycin twice. So there's nothing wrong with Tobromycin. It's a perfectly valid drug, but they only tested it twice. So it's just not interesting epidemiologically. And as you can see, it's either 100% resistant or 100% susceptible because they only tested it once, you're really tested it twice. So it could have been 50, 50, but, you know, it's 100%. OK, so those drugs are valid. Some of them are valid, but it's just not enough data to be meaningful. Good. So I'm just demonstrating here how the graph and the table have exactly the same information, but sometimes it's just easier to talk about the graph and sometimes it's easier to talk about the table. Continue. That's a detailed report. Let me now change that to the summary report. OK, begin analysis. And I'm just going to immediately copy this over to Excel, because Excel will be easier for us to discuss. Excel. Let me just save this as long as I'm here. Let me save this on the desktop and desktop. The reason I'm doing that is I can send this to you after the call in case you want to look at this in more detail later. So I'm going to call this stuff aureus. Antibiotic test practices. And let me change the columns. Good. Good. So I see it's deforious. I see in total they have 270s deforious. 70 percent sensitive, 79 percent sensitive, 89 percent sensitive. For right now, I actually don't care about this stuff. I'm going to delete all of those columns. So those columns are now gone. For today's call, I'm not really interested in percent sensitive or percent resistant. I'm interested in how often are they tested? So I can see they tested out of 270s deforious. They tested. Clinton, my son, 223 times 223 and divided by 270 times 100. So about 83 percent of the time they tested Clinton, my son. Nineteen percent of the time they didn't test it. I suspect they just didn't do susceptibility testing at all. You know, just because you have a staph aureus, usually, of course, you do the organism and the antibiotic results. But sometimes you do the organism, but you don't do the antibiotic results. For example, if it's a wound, if it's a urine with a low colony count, if the patient already had staph aureus five times, you don't have to repeat the susceptibility every time. Or maybe they just run out of discs. So they have 223 Clinton, my sense. But they have 270 staff. So 270 minus 223. So 47 times 47 divided by 270 times 100. So 17 percent of the time they didn't do a susceptibility test for this particular drug. In contrast, if I look at azithromycin nine divided by 270 times 100. So they only tested azithromycin three percent of the time. There's another one with the drug. It's a perfectly good drug for staph aureus, I think. I'm not an expert, but I think it's a good drug for staph aureus. But they don't test it very often. They tested a very, very similar erythromycin. So erythromycin, if I copy this four, well, yeah, if I copy. So equals, I'm sorry, equals 229 divided by 270. So they tested Clinton, my son, 83 percent of the time. They tested erythromycin 84 percent of the time. Any questions so far on what I've done? We've done this kind of thing before of reemphasizing myself. OK. And so what I'm going to do now is to do all the antibiotics as a percentage. So I'm going to show some Excel formulas. Let me just make the font a little bit bigger. I'm going to do a quick search and replace. I'm going to search for the word number. And I'm going to replace it by nothing. So you can see I only did is I just simplified the heading. I simplified the heading just because I want to make the columns closer together. So 23 times nine times 179 times. So I see the good drugs, good drugs. Clinton, my son, erythromycin, suffoxidin. OK, it's not great. It's 179 divided by 270 times 100. So about 66 percent of the time they tested the suffoxidin. I if I recall correctly, I think maybe at the beginning they were doing oxicillin and then they switched to suffoxidin, which would explain why the numbers are different. OK. So now what I'm going to do is figure out the percentage for everything. Equals. The number 23, which is column D, wrote to divided by 270. And I'm going to copy this all the way across. And on top of that, I'm going to change this to a percentage with zero decimal points. So you can see here, they tested Clinton, my son, 83 percent of the time. Erythromycin, 85 percent of the time. Tetracycline, 37 percent of the time. Venkamycin, 6 percent of the time. So this is just so so basically what I would like to see here, the drugs with the most data and as a percentage, it's easy to do that 83 percent, 85 percent. Any questions on what I've just done here? OK. There is a nice feature in Excel. I'm going to highlight all of these numbers. I'm going to highlight all of my percentages. Excel is a nice formatting option. You see where it says conditional formatting? I'm clicking on conditional formatting. Highlight cells. That are greater than. Let me say. Let me just choose a number. OK, 50 percent. And I don't want to fill it in red. I want to fill it in green. And I'm saying, OK. So this is just a visual. It's just visually showing me the drugs that were tested at least half of the time. So it's just a visual aid to help me with my work. And this allows me to see immediately that I have those seven drugs. Fox, Cipro, Clinda, Erythro, Auxa, Penn, is SXT that we have at least half of the isolates. Any questions? If you don't see, if you don't recall exactly the steps that I did, that's not so important. But if you want to do it yourself later, just remember what I did and then I can tell you how I did it later. In this particular example, I did calculations. I divide everything by the number 270. And then I use conditional formatting to highlight things that were greater than 50 percent. The truth is I really don't want greater than 50 percent. I like to choose at least 80 percent of the time. But if I do 80 percent of the time, that only gives me three drugs, the 83 percent, the 85 percent and whatever the other one was, the SXT. Okay, good. I'm gonna continue with this a little bit more. I have a column for number of isolates, number tested max. As you saw, there were about 47 isolates that had no susceptibility tests. Probably nothing was tested. So I don't really care about those. I care about the isolates that had some susceptibility testing. There's a nice formula in Excel called max. And if I highlight all of this, the biggest number is 229. So even though there were 270 isolates in total, the biggest number of antibiotic results was this one here, erythromycin, 229. So the formula allowed me to see that very quickly. So 229 is the biggest number there. So one column 270 is how many isolates. 229 is the biggest number of antibiotic tests. I'm now gonna repeat my calculations and let me divide 23 by 229. And let me change that into a percentage using zero decimal points. And let me copy all of that all the way across make it a little bit wider. Okay, good. And now I'm going to repeat my conditional formatting. I highlight everything. I click on conditional reporting, conditional reporting, not conditional reporting, conditional formatting. And I clicked on the wrong thing. Conditional formatting, highlight cell. Greater than 50%. And let's color that in green. Okay. Oops, I didn't highlight my cells first. That conditional formatting, highlight cells greater than 50%. Highlight in green. Okay. So now you can see erythromycin was attested 100% of the time. It wasn't tested 100% of the time out of the 270. It was tested 100% of the time out of the 229 isolates that had an antibiotic result. So what it simply does is it just changes my denominator. Let me just put this over here. Let's see. Number tested. No, I'll just call it number. Number of isolates is 270. Number with antibiotic tested is 229. So I've just changed my denominator to be the organism isolates that actually have at least one of the antibiotics. So now I can see erythromycin is 100% of those. Then the mycin 97%, SXT also very high 100%. Every, well approximately 228 instead of 229. Okay. So this bottom row, the numbers are similar 9% versus 10%, but it's a bit more meaningful because I'm only looking at the isolates. I'm only comparing with the isolates that have at least one of the antibiotics tested. Good. I am now going to delete. Well, I'm not gonna delete them, I'm gonna hide them. These other antibiotics I'm really not interested in because they just don't have enough data. I'm not saying they're not interesting, but at the national level, there just isn't enough data for me to make good conclusions. There might be enough data for hospital one, hospital two, hospital three, but if I take all these four hospitals grouped together, these are the seven antibiotics that I'm most interested in. Okay. And actually I put this on the wrong row. This, oops, oops, let's see. Well, I don't know, let's find the way this. Okay. So for the rest of this presentation, this part of the presentation, I'm only gonna focus on these seven drugs. So now let me go back to continue. And I'm back in Hoonette. Hoonette is a feature that we don't use very often, but it's called Selected Antibiotics. Let me click on Browse and I don't want all of the drugs. I just want those seven. What are those seven? They're Fox, Cipro, Plinda, Cifoxitin, Cipro, Plinda. Grypto, Oxa, Penn and SXT. I click okay. I click okay. I click on Begin Analysis. And now you can see everything is exactly the same as it was 10 minutes ago, except it's only seven drugs. So this is a lot nicer at the national level. I don't wanna see all those drugs I don't care about. I only wanna see the drugs that I had planned to include in my national report. In fact, that was a request for this call about the national report. One part about the national report is which antibiotics do I wanna focus on? I don't wanna focus on all 25, 30 drugs that you had because most of them just don't have enough data. These seven drugs I am more interested in. Of course, I'm particularly interested in Cifoxitin. The oxysilin, you know, they're similar, but the recommendation of course is Dist diffusion by Cifoxitin. So these are the drugs that I want and so let me copy that to Excel. I'm going to Excel, let me just go to a new sheet. So everything here is the same except it's seven percent sensitives with seven number tested denominators, 179. Okay, and let me save that. So if we're okay with that, I'm ready to move on to the next thing, which is to repeat what I just did, but by laboratory. If no questions, let me go back to Hunet, continue. And Hunet, I go to analysis type and this is where I can start using these very valuable row variables. Row one is organism. Well, that's obvious. As you can see the row, it's, there's one row for stephorias. So that you don't change. The organism is on the first row or is the first row variable. That's the first row variable. The second row variable I'm going to call laboratory. Clicking okay, begin analysis. And now I see my five rows. Let me copy that over to Excel. So row number, the first row variable is the organism. The second row variable, let me just change this to a text field. Text there, we'll repaste it, there, that's better. So, good, and let me line this up. Let me move this over to here and let me make the columns wider. We're narrower, I'm just adjusting it. Okay, also I'm going to delete stephorias. You know it's stephorias, that's what I'm, I'll just put stephorias at the top, stephorias. Here I have the code and the laboratory. I don't need both, it's the same thing. And let me just line up these columns. This goes here and this goes here and then we just reset, adjust the sizes. Good. So here is the first analysis. I did the whole country, all the four laboratories together. And that's what you see at the top. That's what we've discussed for the last 20 minutes. Now you have exactly the same statistics, but now separated by laboratory. We did talk at last time about changing zero, zero, zero, one to zero, one. I think I did it for a bunch of them. I didn't do it for all of them. So three of them are still zero, zero, one. We discussed how to fix that last time. I'm just going to delete laboratory zero, zero, one, because it's only three isolates and it's really part of zero, one. So now, well, let me leave it in. So if I add these numbers up, you look here, if I highlight these five rows in the lower right-hand corner, it says 270, which is the same thing you see here. Just to re-emphasize at the bottom of the graph, as the same information you see at the top, except here it's the country averaged together. And here it's each laboratories separately. So 270 isolates, 270 isolates. Now that I've demonstrated that, let me delete again, zero, zero, one. Okay, good. Now what I want to do is do this percentages again. So let me do the insert a column here. Number max equals the maximum of these seven numbers, the maximum of 229. I'm copying that, whoops, too far. Copying that formula to here and let me copy the heading, number max. So you can see that, first of all, you see laboratory one has 80 isolates, laboratory two, 67, 54, 42. So first of all, I can see how many each facility has. I can say percent tested equals 229 divided by 270. Let me re-format that as a percentage. So 85% of the isolates have an antibiotic result, percent tested. Let me copy that formula here. So what's interesting is laboratory number two, almost always tests. They have 69 staph aureus and they tested 67 of them. Laboratory three, they had 57 staph aureus, they tested 54 of them. Laboratory four, they have 43. So these three laboratories have a staph aureus and a susceptibility test. But laboratory one has 98 isolates, but the only susceptibility tested, 80 of them. So it's a question for laboratories year one, why are you not always testing? And there's, they have a good reason. We don't think it's a good use of resources. The patient already had it. It was a urine with a low colony count or the patient already had it five times. Also keep in mind that, you know, EPHI is a natural reference laboratory. So it's, the reference laboratory doesn't always have to test everything. If they want to know the organism, they test the organism. If they want to know the organism and the antibiotic, they do both. So it's interesting that three of the laboratories basically have a staph aureus and they test the antibiotics. The laboratory zero one has a staph aureus and 80, 82% of the time, oops. Oh, in fact, I even used the wrong denominal. No, that's right. That's correct. That's through a denominator. So laboratory zero one, when they have a staph aureus, they tested 82% of the time and 18% they don't bother susceptibility testing for some reason. And that the reason you'd be discussed with the lab. Good. I want to look at the percent testing for each of the antibiotics. So what we're going to do here is we're going to just copy this down here. Well, you know, I can copy all of it. I can copy, yeah, sure. I can copy all of this. And, good. No, no, no, I won't bother. Okay. Let me just copy this first column, okay? Good. Now I want to figure out how often, let me just copy over the antibiotic names, good. So, suffoxidin, they tested it 79 times out of 80. So I'm going by the number max. So suffoxidin, they almost always test 98%. You know, let me just reformat that as a percentage. Oops. Good. And then I'm going to copy this formula all of the way. And obviously I made a mistake in the formula. I need column C to stay unchanged, good. Let me re-copy that, okay. And I'm going to highlight here the percentages that are over, let me go to the percentages that are over 80%. Conditional formatting, highlight greater than 80%. And let me just put it green so I'm being consistent. Conditional formatting, highlight greater than 80%. And I forgot to change the color. Highlight greater than 80% and green. So what I'd like to see are green. So if I can, if I focus, and let me just copy over the laboratory names, okay. So if I focus on laboratory number one, laboratory one, they always test suffoxidin, 99% of the time. They always test, well, not always, they test it for an 89% of the time. Clean to 100% of the time. The results are rising 81%. Panicillin is right at the border. Let me just change the formula a little bit. Conditional formatting, highlight cells greater than, I'm going to put 79.5%. Oops, then I changed, I forgot the color change. Okay, conditional formatting, highlight cells greater than 79.5 because of the round off error. So basically hospital zero one, I'm happy with. Every antibiotic is tested at least 80% of the time. I'm very happy with three drugs. Fox, 99%, Thinda 100%, S16, 95%. Hospital number two, I'm happy with these five drugs. They tested oxysilin, 85% of the time, but they only tested suffoxidin, 24% of the time. In fact, suffoxidin is the correct drug. Laboratory number three, only tested two drugs to my satisfaction. They tested Erythrum and they tested SXT. Hospital four tested everything very well, except for oxysilin, but actually I'm okay with that. You're actually not supposed to test oxysilin. What a number of people do, and this is not a wrong, is they'll test suffoxidin and they will report it as an oxysilin result. That's okay, it's called proxy reporting. I tested the oxysilin, but I told the doctor I tested oxysilin, that's fine. So laboratory four is perfectly fine. They're testing everything, they're not doing the proxy reporting, which is not necessary. In fact, there's suffoxidin, because suffoxidin really should not be tested, I'm going to remove it. So there's six drugs here that are left. I'm very happy, my assessment. I'm not gonna say great, because I really want to see like 90%, 95%, and here it's kind of three of them around 80%. So this is good, very good I will say. So laboratory two should stop testing or oxysilin, but should test box and sippro. Hospital three should start testing box, sipp, clinda, and pen. Penicillin is debatable, why? Because let's look at percent sensitive. So if we look here, penicillin it's 19% sensitive, 3% sensitive, 26% sensitive. It's not a good drug for staff or it's usually resistant. So many laboratories stopped testing penicillin simply because it is not a useful treatment option, it's just too resistant. So it is debatable, a lot of people don't test it because it's just not a good drug for treatment. But there is still value for epidemiology. The fact that three of the laboratories tested and the fourth one doesn't, I would recommend that they start testing it just to improve comparability between the laboratories. And this one, wonderful. Hospital four, 98, 100, 98, 100, 98, 93. I mean, they could do a better job at SXT, but everything is here is over 90%. So this is the kind of feedback we can give to each laboratories. You can say we now have a national panel. We recommend that everybody test, for example, these six drugs. And if we can get it up to seven arrays, we'll have more data. So any questions? I'll just stop for a moment there because we're almost ready to finish this and I'm ready to move on to discussion of the HUNET resistance profile drugs. So any questions on what I've shown here? National Antibiotics Set for Staph aureus. Based on your historical data, I would recommend SIP, Fox, SIP, Clinda, Arithrap, Penn, and SXT. Not oxacillin because oxacillin is really just a, it's a substitute for the suffoxidin. So these six drugs, you already have good data for. You have very good data for the, yes, go ahead. Yeah, just a question. Ben, would you go through the same process animal specimens, animal isolates? Well, so what I'm doing is regarding historical data. So if you have historical animal data, yes, we can do exactly the same thing so that we can try to test the human bacteria and the animal bacteria with the same antibiotics. For the, we discussed this on a previous call, for the animal isolates, it is very good to test the human antibiotics so that you can compare with the human surveillance program. But in addition, you also want to test some of the purely veterinary drugs, you know, because if you're taking care of sick animals, also having said that a lot of times the, it depends on why you're doing testing of the animal isolates. If you're taking care of animal, sick animals, you should test animal antibiotics. But if you're testing animal antibiotics in order to compare with the human surveillance program, you can test the human program, because a lot of times when the animal isolates come from the slaughterhouse, you go to the slaughterhouse and either before or after slaughter, you take the sample. So we're not trying to make the animal better, we're about to sell the animal at the market, the food. So for the animal world, if you're taking care of sick animals, it's very important to test animal specific antibiotics. But if you want to compare between animal and human, you know, you want to have human antibiotics. Of course, many antibiotics are used in both sectors. So how do I know what a laboratory is testing? Basically they're two ways, they're two obvious ways. How do I know what they are testing in laboratories one, two, three and four? One way is data driven. That's what I'm doing here is I'm simply looking at their data. I'm looking at their data and I am seeing what they have been testing. That's option number one. Option number two is I just call the laboratory and visit the laboratory and ask them, what are you testing? And then they just give me a list. So I can know by asking the lab. Of course, there's a slight difference that when I ask the lab, they tell me what they plan to test, what their intentions are, what their practices are. But of course, sometimes they run out of disks or sometimes the piece of paper they give me doesn't match reality. I'll say, but you said you're testing this in your protocol. I said, and they said, yeah, but we can't buy it. So they actually didn't change the protocol, but they just started testing something different. So one way to know what they are testing is to ask them. That's quick and easy, but it does need an email. It does need a phone call. So that's one way theoretical. You just ask them and they tell you. The other way that I like here is to simply use the data. And the data will tell you exactly what they're testing as well as how often they test it. Which tells me like if they switched in the middle of the year or if they run out of disks or if they have like some urine drugs and non-urine drugs. Okay. Another question. Early in a surveillance, later in a surveillance program, the focus is epidemiology, treatment guidelines, understanding, advocacy, education, all those things. Early in the surveillance program, the goal is capacity building. Are they testing the right drugs? Are they getting strange results like vancomycin resistant staph aureus? So basically what I've shown you today is I'm extremely happy with the test practices in laboratory four. Laboratory one, I'm pretty happy with that except they could do a better job on these two drugs. You know, we really want them to be above 90%. There's not a magic number. I'd love them to be a hundred percent. I'd love them to be 95%. So you kind of weaken your, you kind of go down in your criteria, but I would say 80% is sort of a good low goal. Laboratory two and laboratory three, I'm not happy that we're missing two key drugs, sephoxidin and Cypro. They do have the oxacillin, as I showed you, that they have oxacillin and I want to ask them, did you actually test oxacillin? If they say, yes, we tested oxacillin, that's a mistake. You should not be testing the oxacillin disc. Laboratory three, I'm not happy with because SXT is 100%, Cypro is 26%. So it's very haphazard. You know, it's as if they're always running out of discs or I've been in some laboratories where they put the antibiotics according to the patient's disease. You know, if the patient is this, this and this, let me put on these two drugs or the patient's in the ICU, let me put on these five drugs. And that may or may not be good for the patient, but it's certainly bad for the statistics. I don't want to make statistics for Cypro for laboratory three, if I only tested 26% of the isolates. So here you can see, I'm very happy with this one. I'm pretty happy with this one. I still want them to do a better job on a few of the drugs. And I'm very happy with hospital four. Hospital three, really, you need to review with them. What are you testing? And you should change your test practices. And laboratory two, it's doing pretty well at these four drugs, but they shouldn't be testing the oxacillin disc. So you can see the value to this to capacity building. Okay, that's the end of what I wanted to show you about reviewing test practices. Let me just show you very quickly. I'm going to change the organism, clear a list, E. coli, okay, begin analysis. And let me do the detailed report first. Detailed report. Well, I know we're not going to go beyond that. I just wanted to give you an idea. So this is the same thing for E. coli and let me change this to all of the antibiotics. So when I look at this for E. coli, I see there are only three drugs I'm happy with. Meropenem, Cipro, and SXT. Those three drugs are tested about 450 times. That's tier number one, always tested. And then there's tier number two. I have one, two, three, four, five, six, seven, eight, nine, 10, where it's a bit over half of the time. You can do the same thing in the table. Look once, look again. These first three antibiotics, Meropenem, SXT, Cipro, 433, you see in total at the top left, they have 482 E. coli. Out of the 482 E. coli, they tested Meropenem 433, 432. So it's not 100%, but it's, I think it's like 90%. Ampicillin, then you have the second tier. Ampicillin 309, 291, and then you have the drop-off. And so you see a lot of drugs that are over 200. And those are the ones I was moderately happy with. And then at the bottom, you see the tetracycline is a drop-off again, that's this last one here in the graph, the last antibiotic, 122. And then if you go down to the bottom, you see things like penicillin, completely inappropriate. Hopefully that was just a simple typing mistake. Immipenem very appropriate, but they do not, they only tested it twice. They don't test imipenem, they test meropenem, which is fine. And I'm glad that everybody has agreed on testing the same drug. So a couple of points, you see that gram negatives have a lot more antibiotics than gram positives do. And there are a lot more choices. You know, for example, here you see the, let me go to the, let me put this alphabetically. Now that I've done this alphabetically. So let me compare these two rows. You see seftraxone, 113 times, 73% resistant. Seftraxone, 249 times, 68% resistant. So the percent resistance is very similar. But, you know, both of them are commonly tested. Seftraxone is about two thirds of the time. Sepotaxone is about one third of the time. They're perfectly fine, but for the national program, it would just be easier if the seftraxone people move over to seftraxone, simply to improve standardization. And then I would not have two rows, I would have one row merged together. It's all I have to say about the E. coli, good. Now I'm going to do this other analysis called resistance profiles. First I'm going to show you resistance profiles using my WHO test data. I go to the WHO test data. I go to data analysis, data analysis, analysis type. And there's this other analysis that we have not been looking at called resistance profiles. There's a listing, similar to a normalized listing. There's a summary, we'll look at both. I'm just showing you one month of data, so let me show you the summary by day. And we'll talk about cluster detection later. I will not click here for right now. And then you see here, I'll come back to this. I'll come back to this. Organisms, something simple, Staph aureus, okay. Data files, let me choose my one month of sample data, okay. And begin analysis. So what you see here is a nice normal listing. There's one row for every Staph aureus. One row for every Staph aureus. There's an over on the right. I don't see all the antibiotics, but I see the most interesting antibiotics for my sample database. These are not Ethiopian data. Then erythroclinid, oxygenta, SXT, Cipro. And then for urine isolates, the nitrofrantone is a useful additional antibiotic. So you see, we don't have a lot of data for the nitrofrantone, but it is interesting for the urine isolates. And I click here once, click here again. There are not a lot of urine isolates. So I see some nitrofrantones, but most of them there's no nitrofrantone. So let me just do that again, okay. So here what I see are two columns called profile and resistance profile. So here you see these bacteria are resistant or intermediate to one drug. Resistant to erythromycin alone. Resistant to penicillin alone, there are a lot of those. I'll start at the top. Here you see nothing. So these Staph aureus are sensitive to everything. These are basically the old wild type. Staph aureus sensitive to everything. Staph aureus resistant to one drug, erythro. Staph aureus resistant to one drug, penicillin, there are a lot of those. Followed by a Staph aureus resistant to two drugs. Staph aureus resistant to a different two drugs. Staph aureus resistant to a different two drugs and this one's common. And then I see Staph aureus resistant to three drives and let me just make this column a little bit wider. So here at the bottom you see Staph aureus resistant to everything requested. What did I request? Pen, erythro, tenda, oxagenta, S60, Sipro. It's not all the antibiotics, it's just the antibiotics I want to focus on. The antibiotics that have the most data. So the purpose of this analysis is to show me a list of Staph aureus but to classify them by how resistant they are. The very sensitive bacteria are at the top of the list. The moderately resistant ones are in the middle of the list and the completely resistant ones are at the bottom of the list. And that matches common sense. No, not all Staph aureus are the same. Some are relatively sensitive, some are relatively resistant and that would generally correlate to some degree of course with the genome type. You know, it's resistant to Sipro because it has a resistant gene. It's resistant to oxafoxidin because it has the MEK-A or the MEK-A gene. So what you're seeing here is a phenotypic expression of the bacteria's genotype. Because of the genotype, we see this phenotype. Okay, so this is very useful for tracking purposes. At the bottom we see Staph aureus resistant to everything. And where are they? Well, they're four of them and you look at the patient ID and it's four different people. Two of the people, the last two are in the oncology unit. That is not a surprise. It's not a surprise to have multi-resistant MRSA in the oncology unit. These are sick people, multiple hospitalizations, multiple antibiotic courses, a lot of resistance. Then you have these two odd things here, OP is outpatient. So you also have these two outpatients with a very resistant Staph aureus. And you might think that's strange. Why in the world would an outpatient have a completely resistant Staph aureus? Well, actually, I don't know in this particular example, but it's also not a big surprise. You have to keep in mind that outpatients are not necessarily healthy people. There are a lot of outpatients with cancer. So outpatient, it might be an oncology patient who wasn't inpatient, then went home and then came back a week or two later from the into the outpatient clinic with the multi-resistant Staph. So why does an outpatient have a multi-resistant strain? Specifically for these two people, I don't know, but there are a lot of potential reasons. There might be a nursing home patient. There might be a former or current oncology patient who is in the outpatient setting right now. There might be a family member of an oncology patient. They might come from a different country. They might come from a different patient population. So these final two patients have a multi-resistant MRSA and I know what their risk factor is. They're oncology patients. These two patients, I don't know what their risk factor is, but if I looked at their medical record, if I talked to the patient, I could probably figure out where did you travel? What diseases do you have? Were you hospitalized recently? Do you live or have you visited a nursing home? So it's very valuable to say that not all Staph ores are the same. They are defined by their resistant genes and here we're seeing that expressed as their phenotype. So these two are interesting. These are two, they're not completely resistant. They're resistant to five drugs, sensitive to three drugs, but there are only two of them. One of them was in, and you can see on the left, one was cardiology, one was medicine. Here, this is an interesting example. These are resistant to four drugs. They're resistant to panorhithroclinodipro, but they're sensitive to oxysilin. So these are an SSA, methicillin susceptible Staph ores. And as you can see, two of them, one of them was January 25th, sputum. One was January 28th, sputum. And both of those were in medicine number two. So I do wonder what's happening. Maybe there's a little miniature outbreak in medicine two. And I wonder if they have any connection to this other patient on January 12th with the urine isolated in the intensive care unit. Because you have to keep in mind that these patients move around. So this patient in the ICU maybe went to medicine two. Or maybe this isolate from medicine two is from a patient who was in ICU one last week. So who now will not tell me about the movements? I don't know where these people were. I just know where the sample was collected. So what's interesting here is this is potentially a little miniature outbreak of an MSSA. People pay a lot closer attention to MRSA than they do to MSSA. But that doesn't mean MSSA doesn't cause outbreaks. So this is potentially, I would consider this interesting for a possible outbreak. There are two in medicine two on almost the same day. It's a distinctive phenotype. This database at the top of the screen is I think 86 deforious. Only three of them look like this. And two of them are in the same room on almost the same day. So this is suspicious. Who that will never tell you this is an outbreak? But who that will tell you this looks like something I'm worried about. Maybe it's an outbreak, maybe it isn't. Let me investigate further. Let me look at the zone diameters. Let me look at the risk factors. Let me look at their medical record. Were they hospital, are these hospital infections or community infections? Are they from the same family? Are they from the same ward? What is the historical background like? This is, these are January data. Do we see any in December? Do we see any in December? Were they in medicine too? So I'm hoping you get to see that using these multi-resistance, I study the resistance for two reasons. One is because I'm interested in resistance. I'm interested in treatment. I'm interested in emerging resistance. So one reason I study resistance profiles is because of my public health and clinical interest in resistance. That's one reason. The second reason is I look at multi-resistance as a way to improve outbreak detection. It's a strain identifier. Whether it is very sensitive or very resistant is a different concern, but it's useful to me. Like here's another interesting example. So here you see one is January 17th, one is January 19th, one is the intensive care unit and one is oncology. It's two different people, but you kind of wonder, is the same doctor taking care of both of them or the same nurse or the same medical student or the same respiratory therapist? Was the oncology patient in the ICU or was the ICU patient in oncology? Is the ICU patient an oncology patient? So even in this small sample, real database, I'm seeing all these little things going on here that are suggestive of possible little mini clusters. I'm gonna look at this top section, this top section. This top section is sensitive to everything. And you can see emergency room, emergency room, emergency room. So it's not a surprise, and here you see a lot more, emergency room, emergency room. So I'm gonna click on emergency room. So I'm clicking on alphabetically. So this section here is the emergency room. And as you can see, this is the emergency room and the emergency room does not have multi-resistance. It had three isolates that were completely sensitive, three isolates resistant to one drug erythro, maybe six drugs resistant to penicillin and then a couple of resistant to two resistant to three. So the emergency room, they see mostly sensitive strains. I'm now gonna go down to the ICU and the ICU has some of these multi-resistance strains. Let me go to oncology. Oncology, we have a lot more oncology. They have a lot of these multi-resistance strains. Of course, ICU patients and oncology patients also have sensitive strains. An oncology patient might have a sensitive staph aureus sepsis from the community, it'd be very sensitive. So I'm not saying that all oncology patients have resistant bacteria. They have a high risk for having resistant bacteria, but they also get sensitive bacteria. Any questions on what I've discussed so far? It's how to use multi-resistance profiles. If I'm looking for an outbreak of staph aureus, there's too many staph aureus. It's just better if I separate them, separate the sensitive and the resistant ones and then I can study them separately. Any questions or comments? I hope you can all hear me, and I'm also assuming that. Can you just try to select by date? So we see which week is the most concerning in this report. Sure. And before I do that, I'm looking at, I already showed you the emergency room. Now I'm showing you outpatient. You see, again, the outpatient is very sensitive, by and large. Very sensitive except for these last three. So these last three are kind of wonder, and these are the ones I mentioned, resistant to everything. These are not normal outpatients. These are outpatients with a history either. Okay, so we've already discussed about that. So one way to know the dates is here, you see the dates right here. But what you would like to see, I'm gonna click on continue, and I wanna see a graph by date. And that's what we are doing now. We're not looking at the individual listing. We're now looking at the distribution by date. So here you see the graph. So this is the graph for staff or is sensitive to everything. Graph resistant to one, graph resistant to one. So this is the most common staff today is not sensitive to everything. Sensitive to everything was common 100 years ago, but today most staff or is our penicillin resistant. So penicillin resistant staff, there's one, one, zero, zero, one, two. There's a little blip in the middle of the month. Is that an outbreak? I don't know, but then it kind of disappears again. January 26th, it goes up. So there's nothing obvious here, the saying there's an outbreak. This is interesting. We have three people at the end of the month on the same day. What row is that this row here? We have, you look in the table, January 8th has one isolate, and January 25th has three isolates. I don't know if that's an outbreak. It would be nice to know, is this the same room? So there are a number of things you would want to consider. Are these outpatient infections or inpatient infections? So it's not an obvious outbreak, because the numbers do vary a lot. So is that an outbreak? It is suspicious. It's three on the same day with the same resistance pattern. This doesn't, what's interesting here, penicillin cipro, what I find interesting here is for the first three weeks of the month, there were none. January 26th, one, January 29th, 30th. So again, we see all these little evidences of like little micro outbreaks. Are they real outbreaks? I don't know, but there's a group at NYU Medical Center that does genome sequencing. So whenever they get a blip to say this looks like a potential thing I'm worried about, let me send it for sequencing, then they can start follow up on this. I want to know here, what room are they in? Are they in the same room? The first things I want to know, are they in the same room? Second, are they hospital infections or community infections? If they're community infections, are they from the same family? If they're hospital infections, did they overlap? Same surgeon, same room. Maybe they're not in the same room now, but they were in the same room three weeks ago. And what they find is sometimes it does confirm and sometimes it doesn't. This is interesting. We see penicillin erythromycin. We see five on the same day. Of course, keep in mind sometimes this happens because things are quiet on the weekend and then Monday, you know, things open up and things get busy again. So I'm seeing all these little evidences. Keep in mind, I only have one month of data here. What you would really like to do is a year of data to see what is the long-term trend. How variable is the long-term trend? I'm going to go to the very last row here. Here's just one islet, one islet, one islet. Well, two islets. The last one is resistant to everything and there are four islets that we already discussed. Two of them are ICU, or I'm sorry, two of them are oncology, two of them are outpatient. One was on January 6th, two were on January 13th, one was on January 28th. Is there an outbreak here? There's not an obvious outbreak. You know, it just, it does vary. Sometimes you have one, sometimes you have zero, sometimes you have two. And of course, what you would really like is a year of data so you can get a better sense of what the variability is like. So that was the answer to your question about the dates and that's why we have a listing with all of the details and a summary. The summary is by date, but you can also do the summary by laboratory. So you can see what is the most common resistance phenotype for laboratory one, for laboratory two, for laboratory three, so we will come to that. Here you can see I did not do all the antibiotics. I did not do vancomycin, I did not do amiccin, I did not do toramycin. I just chose these seven drugs. You get to choose which antibiotics HUNET will use. I'm gonna click on analysis type. Here at the bottom, HUNETS is automatic. It says here edit profiles. Here you see for step four is it does these seven drugs. Penn, erythro, clenta, oxa, genta, SXT, cipro as the primary focus. It also looks at nitroferantoin. I showed that to you earlier. I go to the right. So these seven drugs, the first seven drugs, penicillin to cipro, are almost always tested. Nitroferantoin is there, it's for display purposes. It's not usually tested. So HUNET distinguishes, and then the statistics, I'm going to the statistics, the aggregate statistics. I don't include nitroferantoin here. There's just not enough data. It's usually missing. You see here, not tested. There's one example here somewhere. Let me make this a little bit wider. So here you see there's a hyphen. The hyphen simply means the drug wasn't tested. Missing data. Okay, so that's why here under edit profiles, it says nitroferantoin is a secondary supplemental drug that's interesting for the listing, but it's not interesting for the summary because it's usually missing. So we get to choose which antibiotics to use for which organisms. So that's what it chose for staff. That's what it chose for enterococcus. What it chose for gram negative, like E. coli, prosidomonas, et cetera. I'm now going to leave my WHO test database and I'm going to continue with the Ethiopian data. Clicking on exit. Before I continue with the Ethiopian data, any questions? Okay, seems that. Click on, I click on file, open laboratory. I go to the Ethiopia all laboratories I clicked on. I'm not going to go to open laboratory now. I'm going to click on modify laboratory. Antibiotics profiles. So right now, Hoonet is doing something very stupid. It's taking the first eight antibiotics in alphabetical order. Staff, is it through my SIN and my disk? Is it through my SIN and my seat? So Fox, Flora, Cipro, this makes no scientific sense. Hoonet is just sticking in the first antibiotics in alphabetical order. I'm going to click on edit. So this, I told you, you get to choose which antibiotics you want. And I don't want this list. I don't want this list. This is a nonsense list that was done by Hoonet automatic in alphabetical order. The antibiotics that I want are these six drugs. Suffoxitin, Cipro, let me put. I'm going to put penicillin and suffoxitin together because they're similar drugs. In other words, they're in the same class. They're beta-lactinants, okay. So I'm going to put the penicillin first, often also put the older drugs at the beginning. Penicillin, suffoxitin, and then the sequence, I don't care because they're all different classes. Cipro next, clindamycin next, erythromycin next, and the rest of the list, I'm just doing alphabetically. Trimethoprimsulpha, supplemental drugs. I'll put in the nitroferantoin just, no, no. I'm going to put it down here in the lower right. Oops, I have to remove it first from the top, right? Now I'll put it at the bottom, right? So I can show you the difference between the profile antibiotics that are used for classification purposes and supplemental drugs that are simply showed on the listing but are not included in the statistics. Save changes, okay, okay, save. What I just did is I chose the resistance profiles, the resistance profile antibiotics. I could have done the same thing right here. This is what I showed you for my hospital. I could have shown you the same thing here. I did not do that because I can't save it permanently. When you go to Modify Laboratory and you go to antibiotics and you define the profiles, it saves it forever until you change it again. So that's why I did Modify Lab because I wanted to make this a permanent change. If I make the change inside of data analysis, that's good if you're just playing around in a couple of demonstrations, but it doesn't save it when you leave Hoonet, it goes back to its original settings. So to define the profile drugs, you go to Modify Lab, antibiotics, profiles, put in the antibiotics and then you save it. Antibiotics, data analysis, analysis type, resistance profiles, listing and summary, if I click on edit profiles, you see it's those six drugs that I put in. The six profile drugs as primary and nitrofrancine is secondary. I did not do these other ones. So these other ones all have stupid lists. I think one of them, I think I did spend some attention on, like I may have done some of them, but most of this list is a nonsense garbage list. Well, it doesn't matter. Okay, so the staff worries is a meaningful list generated from the data. So that's why I'm gonna stay with the staff worries. I click on okay, I click on okay, choose organism, staff worries, okay, data files. Let me choose that. Before I show you the resistance profiles, there's another analysis that we've already covered called isolate listing and summary. I wanna show you how similar these two analyses are. Isolate listing and summary, okay, staff worries. This is simply a list, laboratory one, laboratory two, location, specimen number, and then you see all of the antibiotics, isithromycin, suffoxidin, doxycycline. You see a long, long list of antibiotics. That's isolate listing. It does not have the resistance profiles. This is the isolate listing summary. And it's organism by date. And it looks funny because you have these isolates here from January 10th, I'm sorry, January 2010. Somebody made a type of mistake. Here you see data in November 2010, obviously. Correct, okay. So that is the analysis called isolate listing and summary. Now we're doing resistance profile and it's almost the same thing. These two analyses are identical, one row for each isolate, except for a few small changes. One change is that we now have the column called resistance profiles. Here we have sensitive to everything. Resistance to one drug, resistance to one drug, resistance to two drugs, resistance to three drugs. And then at the bottom, resistance to everything. So here you see these four isolates, resistance to everything. We're gonna talk to you about that later. So how is this different from an isolate listing? Well, first of all, these columns are there. These columns are there, but we're not there previously, NDR multi-drug resistance. So here at the bottom, let me just go to the bottom, multi-drug resistant, possible drug, XDR is extensively drug resistant, PDR, pan drug resistant, resistance to everything. But we'll come back to that. So one difference is that it's focusing on a small subset of drugs giving me the resistance profile. The other difference is in the columns, it's not showing me all the antibiotics, it's only showing me those seven drugs. The six drugs that are primary, plus it's showing me the nitroferantoin. So you see how these two analyses are almost the same. One is isolate listing without profiles, the other is isolate listing with profiles. So I'll go through this column called resistance profiles more slowly now. Here you can see sensitive to everything. This is the old wild type from 100 years ago, sensitive to these one, two, three, four, five, six drugs. I'm not including nitroferantoin because it's usually not tested. Sensitive to everything. And then we get to the annoying part. The annoying part is the hyphen. The hyphen means not tested. So if I look at the data to the right, you'll see here, the penicillin is missing. Here you see the cipro and the clindor are missing. Here you see the penicillin, the erythromycin are missing. So here you see the annoying part where they're only different, not tested, not tested, not tested. Here nothing is tested. Here they're not tested because the different laboratory. So here you see on the left, if you look at the column on the left, these are laboratory two and laboratory three because laboratories two and laboratories three are not doing the same degree of testing that hospital one and hospital four is. So a lot of these hyphens are simply because the different test practices. This one's different simply because it was not tested. They didn't test any of the antibiotics. Oh, actually, there should be six for that. Okay, well, yeah, that is six, that is six, sorry. So not tested against anything. So then you see here resistant to, these are the ones I like is there are no hyphens. Hyphen simply means I don't know if it's sensitive. I don't know it's resistant. It's simply there's missing data. So here we have resistant to SXT by itself from laboratory number one, resistant to erythromycin, resistant to erythro, resistant to clindor. These isolates here are rather common, resistant to penicillin alone. And then here you see resistance to one, but hyphen, one of the drugs is missing. A lot of it is not interesting because some of the drugs are missing. So I'm most interested here in laboratories one and four. I'm especially interested in laboratory four because laboratory four did very good testing. So this is interesting. You see resistant to these two drugs, laboratory one, three times, laboratory four ones. Here you see again, laboratory one, three times, laboratory four ones. Right at the very bottom, very bottom, very bottom. You see here resistant to everything, to all six drugs. Two of them came from laboratory one. Two of them came from laboratory four. One reason is that they have, one possible reason is they have the most resistance. Another reason is they simply have the most testing. I don't know. So like here, if you look at this whole block, this complete block is laboratory number two. It's the left-hand column, laboratory two. So this block is resistant to everything that was tested. But two of the drugs are missing. And I know one of them, it's the missing suffoxidin. I do know they have the oxysilin, but that's in a different column. So basically if I had to, I could rescue them. I could merge the oxysilin and suffoxidin, but there's not an easy way to do that. So here from laboratory two, we see resistance to everything that was tested. And then if I go over to the right-hand side, you see that simply re-emphasized, right-hand side. Look at the six millimeters, six, six, six. So these are very resistant. You know, there could be some 18s and 14s, but it's mostly very resistant. But the Cipro is missing and the suffoxidin is missing. So I don't know if they're sensitive or resistant. Okay, I'm gonna click on continue. This is now a summary. Let me go down to the bottom. Let me go down to the bottom. Oh, I meant to do it on the graph. Let me go to the bottom of the graph. So here we see resistance to everything. The data are by month. So one of these was November 2018. I don't know which hospital. One of them was January 2019. One were two isolates in May of 2019. That's interesting. There were two in the same month. I wonder if they were in the same hospital. I wonder if they were in the same room. To simplify my discussion here, I'm going to go to laboratory number four. I like laboratory number four. They test everything all the time. So here you see sensitive to everything for isolates. Resistant to erythromycin, one isolate. Resistant to clindot, one isolate. Resistant to penicillin, that's relatively common. When I say they tested everything all the time, it's not every time, you see it once in a while. Infrequently they don't test some of the drugs. And that could simply be because they ran out of discs. Or you know, sometimes there's diffusion. The drug falls funny. And if it falls funny on the plate, you can't read it. So they test everything almost all of the time. Here, this is not interesting because there are a couple of drugs that are missing. If I go over to the right, here the suffoxidant's missing, the syphrose's missing, the clindus missing, the SXT's missing, those various isolates. So those three isolates are not interesting. Resistant to two. I wonder what run they're in? Well, it simply says, oh, let me see, Joyce. That's interesting. So remember here, okay, it's both of them are PWB and PWA. I don't know if PW is, but basically it seems to me to be, maybe, you know, it might be pediatric woman A, you know, like, I don't know what it is, but PWA and PWB seem to be related from the codes. They have the same resistance pattern. What's interesting is that they're different dates. One was in April, I have an American computer. So this is April 15th, 2019. This is November 8th. So they are several months apart. So there's not an obvious connection. But sometimes there's an indirect thing like if you have a doctor or a nurse who's there for several months, it's possible that the doctor infected some person in April and infected a different person in November. So who know, we'll never tell you that, but it just gives you these ideas. What about this one here? Let me go down further to the more resistant ones. Of course, the resistant ones are the more interesting ones. But keep in mind, you know, sensitive bacteria also cause outbreaks. So the last two isolates are resistant to everything. And look at the patient ID. These are two different people on the same day. One was a urine, one was a pus, one was an inpatient, one was an outpatient, medical ward B, I don't know what that is. And oh, that's probably the other one. The other one is probably pediatric ward A, pediatric ward B, that's a guess. This is probably medicine ward B and E and D, you know, ears, nose, and throat. So what's very interesting is I have here one year of data and I have one year of data and we have this twice on the same day in the same hospital. One was medicine ward B, one was E and T. There's not an obvious connection from who not. But if you looked at the patient's chart, it would be interesting to know, is there some connection, maybe the same family, maybe the same doctor, the same student? How can I be sure? Well, look at the medical record to see if you can find a risk factor or do molecular typing to see how similar they are. Look at the zone diameters. The zone diameters are almost identical, six, six, six, it's perfect, six, six, six, six, completely resistant, except for the suffoxidin. It's off by a couple of millimeters. Is that the same, is that different? If you retested the isolate, you could see how reproducible it was. For the first one, nitroferantoin is 28, so it actually is sensitive to nitroferantoin and as a reminder, this is indeed a urine isolate. The second one is a pus, so of course for a pus, they didn't test it. So when I say this is resistant to everything, I don't mean it's resistant to everything in the world, I mean it's resistant to the six drugs that I requested, the six drugs that you see at the top. In fact, it is sensitive to nitroferantoin, but I didn't include it because it's usually missing. And I'm sure it's resistant, I'm sure it's sensitive to vancomycin and endoplamycin linusolid. So in this analysis, when I say resistant to everything, I mean it's resistant to everything at the listed here, the six drugs that are focused on. Great, so that's how you can use this to find potential outbreaks. I'm gonna click on continue. And this is the summary and this is just a summary from hospital number four. So here I have one year, well actually I have two years of data. It seems to start, let me go with something sensitive. Let me go with the penicillin. So here the data go from November of 2018 to August of 2019 for this profile. You know, it's different months for different profiles. But basically we have here approximately years worth of data. So in one year of data, we only, hospital four only had this multi-resistant profile once. And it was twice on the same day. This is very suspicious. The baseline is zero. There were no other isolates like this. And then all of a sudden we have two on the same day. One was outpatient in ENT, one was in medicine Ward B. So who knows, I don't know what the risk factor is, but it is very suspicious. I'm just looking to see if there are any obvious outbreaks here. What was this one here? Yeah, not an obvious thing, but you know, it's resistant to penicillin, clindamycin, and erythromycin. This is a year of data. One of them was in October 2018. One was in December. I wonder if it's in the same room. The listing would have told me, if I stayed on the listing, I would know if it's the same room or not. I'd know the specimen type, I'd know the zone diameters. In a future version of Hoonhead, if you did double click on this, you'll be able to see more details. We'll do that in the future. Is it would be, I would like to know are these in the same room or not? And in the future, you'll be able to do that. So is that an outbreak? I don't know, but it is suspicious. It's a very unusual profile. By unusual, there's nothing strange about it. Penicillin, clindamycin, it's not a common, it's not a rare phenotype in the world. But in this particular laboratory, in this particular year, it only happened twice. And it was about a five to seven weeks apart. Maybe there's a connection, maybe not. No outbreak, the numbers are too small. Not an obvious outbreak, not an obvious outbreak here. This one's interesting. Here we have some November, December, January, February, March, April, nothing. And then there's this three or four, I think that's three. And then we have three in the same month. Are they in the same room? I don't know. Also, you see things like time of year, sometimes in January, especially like salmonella. You see more salmonella in the United States in July and August. It's when people are outside having picnic and chicken and mayonnaise that gets contaminated. But I do find it interesting that there were none in February, March, and April. And then there were three in May. It maybe is worth an additional look. Let me go up, see if I see anything else. This is interesting, nothing for several months. And then two in June. So I'm not saying, who knows, we'll never say this is an outbreak. Who knows simply showing you data that you may or may not want to investigate. Let me click on continue. I'm gonna go to my analysis detection routine. Gonna change the threshold to something kind of high. The list is unchanged. First to show me the list of the staph aureus from hospital number four. And I don't know why it's taking so long. I know why, it's because all of you are watching me. That's why it's taking so long. It's nervous, there it is. Okay, there's the list. I already finished. Oh, so here you see on the left, everything you see here is laboratory number four. Okay. Where's that one example? This is the example I showed you. There's nothing very unusual about this resistance pattern. Penn, pleniplythera. What's unusual is that it simply only happened twice in this particular facility. And one was in the PICU, the pediatric ICU in pus, no, pulmonary, pulmonary. And the other was MOPD, a medical outpatient department, I guess. So there's not an obvious connection. But, and what about the ages? I don't know if I have the age. Let me take a closer look. Okay, one of them is a baby, age one in the PICU, the pediatric ICU. One of them is a male age 30. So, is this an outbreak? I mean, there's nothing, the only thing that has me thinking is that it's the same resistance pattern, but one's a baby in the inpatient, I don't see an obvious connection, unless if one was the father of the baby. So, continue. Okay, great. So now there, so of all of these, there is one that potentially maybe looks like a possible outbreak. Let me go to that here. And here it is. This is the one I highlighted for you. It comes up in red. So it says, you know, out of all of the data, this stands out as statistically unusual. There were two in the same month. In fact, it was two on the same day, May 22nd until May 22nd. The P value is not that small because the data volume is small. So we normally would have expected point one isolate, but we didn't see point one. We saw two people. So you can see what happens as soon as looking for possible outbreaks. But what Huned is saying, this is statistically unusual. It might be an outbreak. It might be a contamination problem. It might be because you simply started testing. Like if you find an outbreak of anaerobes, it's probably not a true outbreak. It's probably you never tested them in the past and now you're testing them. So Huned is simply saying, this is statistically unusual. It might be an outbreak, but it might have some other stupid explanation like ran out of disks. And then you started testing again and then you have an quote unquote outbreak of oxysilin results. And it questions on this. So I'm showing you how you can link the phenotype to outbreak detection because for this current analysis, it's not treating staph aureus as one organism. It's treating it as different quote unquote organisms defined by the resistance pattern. These resistance patterns sensitive to everything, sensitive to a few things, sensitive to a lot of things. There's nothing really strange about the graphs. But then the last one, it was two on the same day and that's unusual. Okay, I'm now gonna do this with all of the hospitals. So this is laboratory number four. I go to isolates, I click on clear all criteria, clear all criteria, okay, begin analysis. So this is now all the data, laboratories one to four. It's gonna find statistical clusters. I guarantee you that a lot of the statistical clusters will be uninteresting simply because of missing data. I don't know why it's slow, but it was slow before and now this is for laboratories, okay. This analysis I'm doing by date. So I can see how many for each profile in January, February, March, April. I don't have to do it by date. I can do it by animal. Do the humans, animals and food to the chickens and turkeys and cattle have the same resistance patterns? And what you'll often find is that the most common salmonella in cattle is not the most common salmonella in chickens and it may or may not be the most common in humans. Or for example, if you have an outbreak of salmonella derby resistant to these five drugs, you wanna see which animals have that. If you never see it in cattle or swine, but you do see it in chickens, it makes you think that the cause of the outbreak is the chickens. To be sure you do your molecular typing and you do see that a lot of molecular typing, the first time people do genome sequencing and they get a sequencer and they start learning bioinformatics is because of foodborne outbreaks, partially because of the public health consequence on humans eating this, but also very much because of the economic impact. You can no longer export, you can no longer sell, you'll have to close down the factory or more importantly, you don't know what the cause of the problem is. Some people are getting sick but you don't know what's causing it. So you need to figure out which food product and which factory and once you find it, maybe it's all been destroyed, maybe it's old, maybe the food has all been used, but you need to figure out what the cause of the outbreak was. I'm still not sure why this has taken so long. I'm sure it's fine, but I'll discuss with Adam, maybe there's a reason. I'll just continue talking. Okay, so we can do the same analysis by animal, by species as I just discussed. You can also do it by laboratory. Are the most common resistance patterns in hospital one, the same as the most common resistance patterns in hospital two. For example, we can say in Ethiopia, these are the five most common kinds of staph aureus, but for each facility, there might be different, there might be in a different sequence, there might be very rare phenotype that only shows up in hospital five. If it only shows up in hospital five, maybe it started there, but it might move someplace else. I work with a network of hospitals in one of the US States and sometimes you find a resistance pattern that starts in one facility and then it slowly moves to the surrounding facilities and the nursing homes. I'm gonna click on stop analysis, click on yes. The two things I can do to make this go faster, I'm not gonna do 999, I'm just gonna do 999, that's faster. It's the thing called Monte Carlo simulations and well, I'm gonna use this feature. If any antibiotics are missing, I'm gonna exclude it from the analysis. Oops, I'm not that one. I'll mit the ice then. I'm just gonna get rid of the missings and I'm just gonna take advantage to do this by laboratory. So I'm not gonna do this resistance pattern by date. I'm gonna do this resistance pattern by laboratory and that's a lot faster, okay. So how many isolates do I have? 69 isolates that we're tested against everything. Most of them are gonna be hospital one and hospital four. These hospitals two and three we're not as complete as testing, okay. What I wanna show you here is not this list. We've already discussed the list. I'm showing you the summary. So here what you see is good. Let me look at the most resistant one. So here you see resistant to everything from hospital two and hospital four. There are not a lot of hospital twos in the hospital threes simply because they don't do as much testing. It's just a lot of missing data. So here you can see that hospital one, I'm in the graph for hospital one, hospital one mostly has penicillin resistance by itself which is common. Followed by PET that's penerythro trimethoprimsulpha. So they're relatively susceptible. Blank here on the left, blank means susceptible to everything but they do have these couple of isolates at the right that are resistant to all six drugs. That's the situation in hospital one. Hospital four, similar. They have a lot of these resistant to penicillins. You see small differences. I'm gonna click on hospital one once and again. So you can see if I look at the top four phenotypes, whoops, the top five phenotypes. The top five phenotypes are pen sensitive to everything resistant to three, resistant to two, resistant to two. Here you see Fox. So that is your MRSA. Every time you see Fox, there's a Fox. This is a very sensitive Fox. This is a very sensitive MRSA. This MRSA is resistant only to penicillin and sephoxidin. That's what people will often refer to as community type MRSA. This is also MRSA, but this MRSA is resistant to everything. But keep on going. This is also MRSA, resistant to three drugs, resistant to MRSA, resistant to four, resistant to three, resistant to three. So everything with a Fox in it is an MRSA. So let me go back to the top five. So here are the top five. So I'm looking at the top five right now for hospital one. But look at this one here. This pen erythro-SXT was four times. It was isolate number three. It was profile number three in hospital one, but hospital four didn't have any of them. So this is something very specific for hospital one that hospital four does not have yet. So if hospital four gets one, they should know, yes, this does exist in Ethiopia, but it doesn't exist in hospital four. So if you see it in hospital four the first time, you may want to put that patient under isolation because it probably was brought in from hospital one or from the general community, because obviously a lot of the circulation is happening out in the community. So those are the top five for hospital one. If I look at the top five, well, let me just put this over to Excel. It'll lead you to discuss there. And let me just change this to text. Let me replace that. And I'm just gonna leave the zero zero one, which is not a real hospital. Okay, good. And the sort sequence, let me just re-sort it. Oh, no, nevermind. I didn't mean to do that yet. Okay. Okay, so you see the top five for hospital zero one. I'm now gonna do the top five for zero four. And let me copy those over. And let me put those into Excel and paste. And let me just, let me just move some stuff around. Let me paste this here. Let me move this over here. And let me delete these. And let me delete these. And let me just delete these. Okay. And I'm now gonna sort this alphabetically. And I'm gonna sort this alphabetically. So if I look at the top, so, top five. So this is in the top five for both. Both. Both. Both. Both. And I do need to make this column a little bit wider. Let me make this a little bit wider. Okay. Let's see, what else? Whoa. So of the top five, they only share two in common. So let me, so I hope you see that. So this is basically hospital zero one. And this is hospital zero four. So even the top phenotypes are different. So hospital four has a lot of these pencips, but that did not make the top five for hospital one. Hospital four has a lot of these, but didn't make it to hospital one. The only ones they have in common are sensitive to everything, which is in the community. Sensitive to everything except for penicillin. So those are, those they share in common, but the other top five are different. So this allows you to see some of the different epidemiology that you might see in different parts of Ethiopia in a university hospital versus a community hospital. All of this can be backed up by whole genome sequencing, but we don't have the time and money and expertise to sequence everything. So the phenotype is a great way to get started. And then if you wanted it, and even here, they do have the, both of them have penicillin resistance alone, but it's the same penicillin resistant. And that's where you could choose a couple of these and a couple of these, this one had 18, this one had 18. It's interesting it had the same number, but you don't have to sequence all 36. When you could sequence maybe four of these and four of these, and to see how similar those are. So these resistant profiles are also a great aid to genome sequencing, because we don't have the time and money to sequence everything. It just helps you to find a few representative isolates, you know, worth testing further. So I've talked a lot, there's still 15 minutes, are there questions on this or other things you'd like to discuss? Is this clear? Is the value of resistance profiles clear? It's basically a way for you to say, instead of saying I have an outbreak, a possible outbreak of staph aureus, it's an outbreak of which staph aureus? Resistance to these three, sensitive to those five. So it just allows you to look at the molecular epidemiology at a lower detail, which is very valuable for outbreak detection, correlating between human animal and food and different kinds of animals. Okay, questions and comments. Yeah, unfortunately, Gabrie doesn't appear to be on to give us any insights into their analysis for their annual AMR report. That's too bad. I'm just going to go back and find the participant list. Go to meeting no, no, sorry, go on. Well, anyway, okay. I mean, I can think of other things, or we could end. First step to sneeze three times. Bless you. Okay, I'm ready. Other applications. Okay, as I pointed out earlier, this analysis is worthless if you have not chosen a meaningful set of drugs. Because if you choose a meaningless set of drugs, you just end up with a lot of missing data and missing data and missing data. Also, if the laboratories are testing different drugs, I can do a nice analysis on hospital four by itself. I can do a nice analysis on hospital two. So let me get rid of, I'm going to go back, where's my honet? Okay, I'm going to go back to analysis type and I'm going to go to my antibiotics and I'm going to go to Steph or his edit and I'm going to get rid of the penicillin and the sephoxytin and the Cipro. So now I only have three drugs. I'm going to repeat this analysis and there's something gained and something lost. There's an improvement and there's a worsening. So now you see here, almost everything is tested. I have sensitive to everything, sensitive to everything, sensitive to everything. The problem though is I've lost three drugs. So now it's much nicer. Now I have all four hospitals. So now I can see this is what hospital one looks like. Hospital two, hospital three, hospital four. So now I've gotten all of my hospitals in but I only have three drugs. So I've lost information on the microbiology side but I've improved the data capture because I've deleted antibiotics that have missing data. So in the future, what you would like to do is to try to get hospitals two and three and everybody to sort of agree on a core set just to make it easier to do this kind of comparison. So now that we have a nice clean data set because I got rid of the problematic drugs, what do we have in hospital one? Hospital one, mostly sensitive. And here you see on the left sensitive to everything. When I say sensitive to everything, I don't mean everything because we'd already know that they're resistant to penicillin but penicillin is not one of the drugs I requested. So this blank means sensitive to everything among these three. So in fact, most of them are resistant to penicillin. So mostly penicillin sensitive. Let's see, well, this is interesting. So if I look only at SXT, that's mostly a problem in hospital two. This is not a problem in hospital two. It doesn't exist in hospital three. This is mostly a problem in hospital two. So you see it's quite different. So that's the different profiles by lab. Here I'll now choose the lab. So here hospital one is mostly blank, probably all of them. So all of them blank is the most common thing which is penicillin resistant by itself. But if I look at the other antibiotics, it does vary between the facilities or between the animals, between the human animal food. I'm now going to do this by date, specimen date by month, okay. And I'm interested in the outbreak detection features. So here, let me look how many possible clusters. There's just one here. Yeah, this one, okay. March of 2019, resistant to erythromycin alone. That's very strange. And I want to investigate that a bit further. Is it the same hospital? So let me just look at closer, look at the list. So if I take a closer look at the list, okay, I'm looking, here it is. You see this whole block here, resistant only to erythromycin, it's a very big block. Well, I'm going to show you something slightly different. Let me go to isolates and I'm going to go to erythromycin and I'm going to say erythromycin non-susceptible, okay. So now I'm doing this and the only thing you see here, everything here is resistant. Oh, I meant to say sensitive to everything else. Anyway, I'll just focus on the first part. Now let me just turn this up further and let me say SXT, let's say, termetoprim sulfa susceptible and let's say, clindamycin susceptible. I don't have to do this. I'm just showing you how to select based on antibiotics just so I can clean up the display, clindamycin susceptible. Good, okay, begin. That's fine. Let me just go to the sets again. Good, I'm not going to look for outbreaks now because I've removed everything again. So everything you see here is the same, resistant to erythromycin alone. How many are there? I have to move this thing, it's a top 26 isolates and look at the rooms. A lot of them were laboratories. A lot of them were laboratory one, three of them were laboratory two, several in laboratory three. So whatever the problem is, it's mostly a problem in hospital one and I'm going to understand also with this by date, specimen date. Did I say it before it was March of 2019? Let me just take a look at the graph. I just, yeah, March of 2019. So you see there are a lot of these in March of 2019. So if I look at March of 2019, let me sort this. March of 2019 is this block. And this is a mixture of mostly hospital zero one and hospital zero three. And there was one isolate from zero four. So if for this we're all the same hospital, I'd be very worried. So there's something strange going on. This is, so the graph is very obvious, something clear is going on. I don't know what that is. Are they entering more data? Is there a little outbreak? I don't know. It's unfortunate that I only have three drugs. If I had seven drugs, I could do a better job of separating them out. So that's the advantage of a small number of drugs is I get more data, I get more isolates, but I'm losing antibiotics. The advantage of more antibiotics is I get better microbiology, better refinement, but then you start losing hospitals because they don't test those drugs. So this is an interesting example that I wonder what's happening there. Sometimes what happens is somebody's doing a study. You know, you're doing a study this month, we're all gonna do MRSA. So that month you have a lot of MRSA. That's not an outbreak, that's simply just a study that people are collecting more than they usually do. Let me just take a quick look. I can show you how to do a date selection. Let me go to isolates specimen date. Let me put March 1st, 2019 until March 31st, 2019. And I say, okay. Let's, and okay. Oh, I got it backwards, you know, good. And I say, okay, okay, begin. So it's mostly pus, all of it's pus, except for one pulmonary and two ears. I'm sorry to interrupt, what was the comment? Now I was just saying, we're getting close to the end. So if you leave just a minute or two for one question in the chat box, that would be good. Oh, sure. Is there a question in the chat box? Yes, it takes us back to how to transform the Hoonet D-Base files into SQ Lite. Sure, very easy to answer. By the way, here I have these definitions, MDR, XDR, PDR. We use specific definitions. I'm trying to go to PubMed. So PubMed, and if I look for selling MDR, I don't know if that'll do it. Yeah, I'm one of the co-authors on this 2012 paper from the ECDC, the European Center for Disease Control and Prevention. It involves, you know, Srinivani is from the US CDC. It involves Janet Hinler, she's from CLSI. So it's a wide set of input, Todd Weber, CDC at the ECDC. So there are certain definitions here. So when Hoonet says MDR, XDR, PDR, it's using those definitions. Let me click on Continue and let me click on Exit and let's click on here. So my Hoonet is from beginning of September. We have a newer version. We have finally made some wonderful progress that I've told you many times about the D-Base compatibility issues. We have completely solved that 100% for Hoonet data analysis. So you will no longer have the compatibility issue for Hoonet data analysis. This week, we are almost finished the work for data entry and for backlink. So the main, there are different reasons to switch from D-Base to SQLite. One of the reasons to SQLite is faster, more secure, more features, et cetera, because it's a more modern system. That's a good reason to switch to SQLite. Another reason to switch to SQLite is because of these bad, awful, evil, Windows compatibility issues. But I'm happy to tell you that we have completely solved those compatibility issues for Hoonet data analysis. And this week, we will finish the work for backlink and Hoonet data entry. So if you want to switch to SQLite, great, because there are a lot of advantages. But we don't have to worry about the compatibility issues any longer. The way we did that is we just rewrote some things at Microsoft, because the Microsoft engine, the Microsoft library was completely unreliable. So we made our own version of the Microsoft engine. That's why we were able to solve this issue. So if you want to switch to SQLite, great, but you don't have to. Whereas before you had to if you had the compatibility issue. Okay, great. So how do I update? I go to data entry. I go to update data files to SQLite. And then I choose data files and this is already a SQLite file. I need to find something which is not SQLite or something which is not, it's a test. There's a Dbf file or this is a real data file from some group. I don't remember which group, okay, good. So this is a Dbf file. And now if I click on begin conversion, the file was, the file will not be deleted. The file is called test January, August, you know, I'm sorry, January 8 to 20 Dbf. The new file that is going to be created is called test January. And if you want, you can change it to something else. But who now just giving you a suggestion? No, so let me just choose that again and let me choose all files. And let me just choose that file again. Where's my test? There it is. Okay, okay. So by default, the new file will simply change the Dbf to a SQLite. What should I do with the old file? Who now will not delete it? You have two choices. You can move it to a backup folder to get it out of the way or you can just leave it where it is. So then you have two copies in the same place. I'm going to move it. So I'm not going to delete it, but I'm just going to move it to a backup folder. So I click on begin conversion. Conversion has been completed. And now I have two files. Let me go to C drive, Hoonat data. One of them is called, this is the file I just created test January with today's date and time, September 17th, 929 a.m. here in Boston. That's the new file that I just created. The old file has been moved. Forget this one called archive, it's called backup. We go to backup. Here is the original file from August of 2020. So it didn't delete it, just moved it out of the way. That way you're not confused by having multiple copies of the same thing in the same folder. Okay, if that answered the question, great. If it didn't answer the question, then please ask further. Does that answer the question or do you have another comment? Just in summary, if you want to upgrade your database, to SQLite, you go to data entry, update data file to SQLite. You can choose as many files as you want. It will upgrade them one after the next, after the next. And then with the original file, you can leave it where it is, or you can move it to the backup folder. Yeah, Roger replied. I think he understands that he replied to a message in the box. Well, we reached the top of our to our session. So as usually, as usual, thank you. Thank you so much, John, for your time. This was really interesting. I learned a lot today. I think we, I'm gonna be sending last week, about two weeks ago, link and the notes together with this session because we were a little delayed because of additional workload. So I'm gonna be sending all of those links together this week. So with that, yeah, I wanna thank you everyone. Well, actually, just one comment. We're showing you how to do data, data, data, data, interpret the data, but I keep on emphasizing the importance of utilizing these data to improve practices. Get people to standardize. So what I do recommend is you talk with each other. What antibiotics do we recommend? Which antibiotics for a staph aureus? Which antibiotics for the introductory ACA? Because I don't wanna come back a year from now and the test practices have not changed. So please I hope you take some of these lessons to heart. It's not only about data analysis for theoretical purposes. It's data analysis to help you make a better surveillance program. And with regard to resistance profiles and clinical reporting and epidemiology to have a minimum course set would be a wonderful step ahead. Thank you.