 One thing we discussed was the idea of a national configuration so that you could analyze the data from all the hospitals at once. So let's just repeat that because I believe I did cover it last time. As I said, a lot of times if you have 10 labs, a lot of times they have identical configurations because very often you create one for hospital one and then people just copy it multiple times. If you do that, then you have 10 identical copies. In which case it doesn't matter which configuration you use because they're all the same. But of course over time, the different configurations often develop a life of their own. And then, oh that's right, and I did download the new software. We just put the new software in. You know, we're always in the process of making changes and updates. So I'm going to, I downloaded it earlier today. I just didn't have time to install it. So downloads. This is the new software I downloaded this morning. And I'm just wondering why it would, usually doesn't give me that message. Okay. John, I had the same message and also I had problems installing it. So hopefully you have better luck. Okay, well I did just have to, so we have, oh I was saying it's already installed. Okay, well okay good. So it might be related to Distribute Hoonet. We have a secure digital certificate. It's valid for three years. It expired last week and we purchased a new one. So this is the very first time that we are using the new one. So there must be some issue there of configuration. And good, so all right. So I will not show you. I will not do the new software there. And I have one from now Thursday. Well anyway, I have a recent one. Okay, so let me do that. Let me go back to Hoonet. Let me move the go to meeting controls out of the way again so I can actually access my screen. Okay, I'm going to click on cancel. I'm going to go to file. So we looked at this feature called create a lab from a data file. And that's where we started to end up with compatibility issues with the debase and the SQL lite. Let's see whether that's resolved on my computer. So let me choose Ethiopia. And I'm going to choose Ethiopia all laboratories. And I will call this all. And then I just choose examples of, you know, I choose this one from NRL. I choose this one from a different hospital. I choose this one from another hospital. I choose this one from another hospital. So each of these comes from a different facility with different antibiotics and different data fields. I click okay. And then I, it's going to look, now I'm going to click on okay. It's going to look at the contents of each of those four. Okay, so the one, the reason I updated this guy wanted this feature, but apparently I think he did this feature before the last update. So, but I can still describe it. So basically in that feature, and I think it works if I just do debase files. So let me try that again, but using the debase options. So let me just say Ethiopia, Ethiopia, all hospitals, all, and this time I will only choose the older debase files, which is what you have. Let's see, I have to move these controls out of the way, all files. And this is one, this is a debase file, basically anything that doesn't end in SQL Lite. So all of these are different data files from different groups. In fact, even this one is combined from different groups. So I click on okay, and hopefully this will work. Good. Okay, and it says there are too many fields. So there is a maximum of 255. That's just because I chose a bunch of completely diverse ones from UCAST and CLSI. So it's not an error just being soon as warning you, it'll go with the first 255, which for most people is more than plenty. Most people have like 100 or so, even less than that. Okay, so what it's doing is it's analyzing the contents, the data field names, the antibiotic lists, the dropdown lists, and it's creating a configuration that has all of that. So this new configuration can be utilized to analyze any of the facilities. So it's finished. I say yes. And here I have my antibiotics on the right. And these antibiotics are from every single laboratory. Here at the top, you see the disk diffusion. If I go down further, I see the MIC. If I go down further, I see E test. And so it's 207 antibiotic tests. That's why we're so close to, we're just slightly over 255 total fields. We have the antibiotics and then we have the non-antibiotic fields like first name, last name, et cetera. And then if I go to location, this is every location encountered anywhere in any of these databases. Data fields. It has every database, every isolate, every data field found in any of the databases, including these food ones, these animal ones, these locally defined ones that somebody made for their own local purposes. So I'm going to say save. And now when I go back, I have Ethiopia all hospitals. So this would be very valuable for you so that you can feel comfortable what you've been doing previously, I guess, is analyze the hospital one with hospital one configuration, analyze hospital two data with hospital two configuration, analyze hospital three data with hospital three configuration. But using this one, you can analyze all of the data from any of the hospitals using one configuration. That's more convenient. You can keep the break points up to date because it's just the one configuration. And more importantly, you can analyze all 10 hospitals at once without being afraid of losing missing antibiotics. As I explained the other time, if I am analyzing data from hospital two, if I open hospital one and then analyze data from hospital two, they will only find things that they have in common. So if hospital two does some antibiotics that hospital one doesn't test, then I will miss those if I analyze hospital two using hospital one configuration. But if I use Ethiopia all it has all labs, all antibiotics, all locations. So I'll just leave it at that. Are there any questions on how to create a national configuration? Yeah. Maybe I have one question. Yes. Yeah. It's really good to have a national configuration because while you are analyzing hospital one, you are expected to open its own laboratory. So you can analyze the whole data by using this national configuration. My question for you is the national configuration will contain all the possible fields that are found in all hospitals. But sometimes the variable names or the field names might be different, but they have the same information. So in that case, the national configuration will create these variables in a different field, right? So how can you manage these things? I'm just sending a reminder to Rodney just in case he has the invite. So if he's not on his practice, he's just busy. Okay. Good. So that's a very good question. I'm already in. Hi, John. I'm already in. Oh, is that Rodney? Yes, this is Rodney. I'm already in. I'm already in. Oh, good, good, good. No, good. Thank you. Thank you. I did not double check the participant list. Great. Okay, great. It's an excellent question. So it is true that if I have data fields from hospital one, hospital two, hospital three, when you merge them together, who in it will merge columns that have the same name? So hospital one calls it patient ID, hospital two calls a patient ID, hospital four calls a patient ID, and everything will merge perfectly. However, if somebody decides to delete the who net patient ID standard field and put something else called Ethiopian national ID, then who net will merge who net will include the new column in the new file, but it will not merge the column contents because who net doesn't realize it's the same thing. So, you know, you'll have so, so if five hospitals call a patient ID, but one hospital calls it national ID, you'll the new the new configuration will have two different files, but two different fields. Let me go to that. I'm I'm going to Ethiopia all hospitals modify lab data fields. And you can see here at the top all of the normal standard who net fields, which I recommend people do not change. I mean, they can change the length. My first name is 20 characters. I don't care if you change the length, who net will just simply choose the longest length. But you know, all these standard fields, you shouldn't change the name. Here at the bottom, this one says patient state. This one says response to treatment. This one, I don't even know what CHEDO is. It's something that I think this is actually from a Vietnam group I was working with. I forget what that project was. So who net will include everything in the configuration, but it will not merge the data files together. So then if you go to who net data analysis, let's be doing let me do an example. I'm going to click on save. And I'm going to go to data analysis, analysis type, isolate listing and summary. I'm just going to do the list this time. Organisms, let me say all and data files, let me choose all. And you know, I'll just choose some of those exact same files that combined, not the Excel tests, you know, so I'm getting data from a variety of systems. Okay, okay. To get analysis. Now, this is that debase issue. So when the process of my computer, let's see if I can get that to go to, as you can tell, we're having a lot of issues with this debase stuff, which is why it's so important for us to have the SQLite. I'm just going to reinstall the users, the users JS90 and downloads. And Microsoft access, where's the access engine? Access engine. So I'm just going to reinstall that. Usually reinstalling the access engine, 64. No, it's confused because it's 64 and 32 bit. All right, well, that's a pain in the neck. Okay. Okay, so instead, what I'm going to do this time around is because this portion does work with the SQLite. So I'm going to just put SQLite here. So I'm just going to choose a variety of files from different systems. And I'm choosing okay, and I'm choosing, okay, begin analysis. So it's analyzing all the data from these different facilities. But if there are two columns that are similar, who net is not smart enough to know they're similar. So what you see here, a variety of different things from different columns. Here you see market category patient ID, those x long, the diagnosis. So if they have called the same, if they have to call the same content, the same field, the same field name, there's no problem, who then simply merges the contents together. But as you're pointing out, if one hospital calls an identification number, and a different hospital calls at H, C, E, D, O, who net will retain them, but they are in separate columns. Also with the antibiotics, if some of them have imipenem disk and imipenem MIC, if some of them make a mistake and put imipenem five, imipenem 10, who net will reliably capture all of those fields as separate fields, is not going to try to merge them together. So basically, as long as people keep to the standards for the things that you care about nationally, there won't be a problem. It's only things like, you know, these user defined fields, the GHI, CHU column, this one here, those things might be very different between hospitals. And that's okay. It means that I don't care about it at the national level. They care about it locally. They want to know the name of the doctor. So things at the national level that I care about, I want very well standardized. If I don't care about it, you know, I can just ignore it. Those don't have to be standardized. So those comments help? Yeah, thank you. But what is the difference between combining the different hospitals data and having, analyzing each hospital using the national configuration? So, you know, we can convert all the hospitals data, then we can analyze the merge data. And also, another option is to create a new configuration to the national level. Then we can analyze each hospital using this configuration. So what is the difference between the... So our number of related issues that we're discussing here under data files, I can analyze as many files as I want. Yeah. So the files themselves are not being combined. The files are staying the way they were. You know, this file is the way it is now. And after running the analysis, it will still be the way it is. So in this way, I can analyze data from many files. The files will not be combined, but their data will be combined in the statistics and in the analysis. So if I just choose a whole lot of files here, I am not combining the files, but I am combining their data into the analysis. Is no requirement at all to physically combine these files together. If you physically want to combine the files together, you can. I didn't show that today, I've shown it previously. There is an option here called Combine, Export, Earn, Crib Data. So when I do it this way, and I say all files, and I just choose a bunch of the files here, I don't want to choose any big ones. I don't want this to be slow. That's fine. And I just chose some things that are not debased files. That's fine. Okay. And good. That's fine. And I click okay. Good. So all of these files, I'm now going to be combined into Ethiopia over. It's just easier to deal with one file rather than multiple years. So is that distinction clear? Like here, there's no requirement ever to use this feature called Combine. But it can be convenient if you're just getting too many files and you just want to combine them together. You know, for example, what I do when I'm doing a research project is I will have one file per hospital per year. So if I have 10 hospitals, 10 years, I have 100 files. But if I'm doing a publication about those 100 files, I do find it personally convenient to combine all 100 files into one very large file. I don't have to do that, but it just means when I'm doing my analysis, I don't have to select 100 files. I just choose one very large one. After the publication is done, I have two choices. I can delete that file because it really is just temporary. Or I'll zip it up and I'll just keep it as an archival thing and I'll connect it to the publication for later convenience if I ever have to redo anything. So is it now clear the distinction between combining the files physically using this feature to make one large file versus this one data analysis where we're not combining the files, but we are combining their contents into the analysis. Is that distinction clear between those two combines? One is a real combine. We're combining the files. The other one, we're not combining the files, we're just combining the contents in the analysis. That's clear? Yeah. Now this other feature that I mentioned, create a lab from a data file, it is actually not really combining anything. It's not combining any of the data rows. So it's not combining the files, but it is looking at the headings. So if there are 10 different hospitals, it looks at the headings for hospital one, headings for hospital two, headings for each of the hospitals, and it says this hospital has 10 antibiotics. This hospital has 20 antibiotics. It combines the national configuration of one in the same or is there a difference? Well, okay. So, okay, well, so, okay, I'm back at the main screen, the file open. I'm now opening file 01. I'm opening file 01. When I go to data analysis and I choose the data from all of these different facilities, it will only analyze the antibiotics of hospital one. So even though the files are being combined, it is only looking at the data fields and the antibiotics of hospital one. So here, I don't want to, if I'm analyzing data from all Ethiopia, I do not want to open hospital 01 because it will only give me the antibiotics of hospital 01. I would like to use the, when I'm analyzing all of Ethiopia, I want to use Ethiopia all hospitals, but this I had to create and the way to create it is basically we're scanning it. We use this feature called create a lab from a data file to scan the contents of multiple hospitals in order to create this national config. Now that I have analyzed the national, now that I've prepared a national configuration, then I can use this configuration, I open this configuration, then I go to data analysis, and then I choose any of the hospitals, the data from any of the hospitals. Is that still unclear? These are three very different uses. They overlap in the general ideas. So first of all, combine is never needed. There's no reason you ever have to combine data. You combine data only out of condition. You have 12 files, you've just prepared a one file. So the reason for combine is simply to combine things for convenience. In data analysis, you often want to combine data for multiple facilities, but it will only combine the facilities with regards to the open lab. If the open lab is 01, it will only analyze the 01 fields, so the 01 antibiotics. But if I'm analyzing all of Ethiopia, I don't want to do that with hospital 01. I want to do that with Ethiopia all and the way that I create Ethiopia all is I scan the data for multiple facilities. Okay. So this one is about creating a national configuration. It's not combining the data, it's only combining the data definitions. Once I've combined definitions, I now have my new configuration. Now that I have my new configuration, I can utilize that to analyze the data from any of the facilities. Okay. It's more about method than when you're doing the first national configuration. You only have to create the national configuration once, and then from then on, you can do everything with that national configuration. Yeah. Now that I've created the national configuration, I could choose data from hospital 01, or I could just choose the data from hospital 10. So using the national configuration, I can do the national analyses, but I can also do the individual facility analyses. You know, if I'm interested in data from hospital 01, I can open the configuration from hospital 01, or I can just use the national configuration. Either one of those is fine. The advantage of using the all is that I can use hospital 01, or hospital 02, or hospital 03, or all three of them at the same time. Whereas if I choose hospital 01, I really should focus only on hospital 01. That's clear. Thank you. As long as we're discussing this, there's this feature here called copy lab. Let me make a copy, and let me call this Ethiopia. I'll call this all hospitals, important stuff, or minimal, or something like that. Let me call this all minimal. So now I have two identical configurations. One is all hospitals. That's the one I got from scanning all the files, and now I have all hospitals minimal. At the present time, these are identical. There's no difference between them except for the name. Why would I want to do that? I'll show you why. I'm going to go to modify lab. I'm going to go to data fields. I'm going to click again on modified lists. I'm moving down my go to meeting control so I can actually see what I'm doing. And here you see the thing from HO10. I don't care about that. I'm going to remove it. Diagnosis. I'm going to get rid of it because maybe only one hospital did that. Responsive treatment, patient state. I'm just going to get rid of stuff that's not part of my national protocol. Just out of convenience. If I'm doing some tables, I don't want the table to be too long because it has a lot of things I don't care about. For example, I'm going to click OK. I'm going to do the same for the antibiotics. Antibiotics. For example here, like this one, amoxicillin ND25 is an invalid test. Silicide does not have any disdiffusion break points for this. So one of the laboratories is doing it, but it's not a valid test. Therefore, I'm going to delete it. So I'm deleting it because it's invalid or carbenicillin. That's a very old drug. I don't really care about it. And maybe one of the hospitals has data. So this is allowing me to, like peppericillin, is no longer really used. So basically your laboratories maybe test 50 different antibiotics. But maybe there are 30 that you've all agreed on nationally. So the reason I'm doing this minimal is really just out of convenience. I'll just show you an example. I'm doing the minimal now. Data analysis, analysis type. Let me do this summary. Let me do the summary by laboratory. OK. Let me do E. coli. Let me do a different one. Let me do the isolate listing and summary. Let me try the antibiotics and see how that does. Because I didn't pay attention to which files I was selecting. Let me make sure at least there's some real data here. Yeah, let's do a test. Let's make sure some of that. So this one is real data. See, a lot of these files are basically empty. Whenever you see the file size of 16, that's basically just an empty file. OK. Begin analysis. So I've done this analysis. And it is not showing me all of the antibiotics because I have deleted some of the antibiotics. Well, I don't think I saved it. You see here? Here you see the peppericillin. But you see only hospital two tested. Hospital one doesn't test it. So what I'm going to do now is I'm going to go back to my minimal configuration, antibiotics. I'm going to go to that peppericillin. And this time I'm really going to delete it. Last time I said I was, but I didn't actually save it. And I'm just going to delete a bunch of things here. Deleting a bunch of stuff. Delete the MIC stuff, etc. OK. Save. Data analysis. Data analysis. Let me repeat the same analysis. Let me do this by summary. Let me do this by laboratory. Let me do this for E. coli. Let me choose this for two of my test data. This is this one and this one. I click OK. Click on begin analysis. So basically I've just filtered it. Now when I analyze the data, you don't see any peppericillin because I deleted it. Let me go copy graph. Let me do copy table. Let me go to Excel. So now I have it in Excel and I have cleaned this up a bit. This is not polluted with a lot of antibiotics I don't care about. If you have a network of 20 hospitals and one of the hospitals that's taking cycling, it's great for them, but it just kind of messes up my national thing. I ended with a lot of missing columns. So continue. So is that clear? This all hospitals is everything. Every antibiotic, every field, everything, everything. But some of those things I really don't care about and it just messes up my output because if I want to share this at the national level, I don't want to share the national level statistics for peppericillin if only one hospital tests it. So just a way to pre-fill to the data to the things I care about. Instead of calling it minimal, I called it minimal, I can call it, I can call this national protocol. Just so it's only the subset of things that you have all agreed to test and publish and analyze, deleting all the stuff that really is not so important or well standardized or complete at the national level. Is that clear? The advantages of all hospital with everything, everything versus all hospitals where you just kind of prune it to that subset of things that you really do care about? I hope that's clear. If it is clear, then there are no questions. Okay, good. So that's a bit about the national configuration. Now in terms of data cleaning, we talked a bit about this earlier. I'm now going to click on all labs, national protocol, and I'm going to go to my quick analysis and I'm going to choose my patients and sample statistics, data files, and let me just choose, let me just choose those two hospitals that I just chose. Well, but I promise they both have the same name that doesn't help. So let me choose both of those there and let me just choose some other ones here. I think the Agisar should be here. Let me just choose. I'm just trying to find some real data. Agisar, yes, that's real. And let me choose glass. That's real. And I can also choose the NRL, NRL. So this is real. Okay. Okay. And so we've already done this. I did this last time, but last time I only chose a single laboratory. Now I'm going to click on begin analysis as a reminder. If I click on edit, it's going to do each of these macros in sequence, lab by month, sex by lab, age by lab, location by lab, et cetera. I did this last week, but I did not choose multiple facilities. Okay, great. So here you can see I now have four labs. As I mentioned, two of the two of the files were both from the hospital TST. So this is a combination of those two different rows. We have data from 001, 01 as we've already discussed. This should really be combined. Last time we did review how to clean that, and SKH is simply a different hospital. And then here I can see the distribution. So in terms of data cleaning, I can see their distributions by month. So for TAS, it's from January 95. SKH, the data's missing and your data is some other, always a later time period. That's because it's several years later. So this allows me to see if any data are missing, and then you go back to the hospital and you tell them you're missing January, or you used to have 100 a month and now you have five for last month. So that's one aspect of data cleaning. Here you can see male and female. You can see that hospital 01 and hospital first two have age and gender listed. SKH and TST don't. So the data is missing from there. So you can go to those hospitals and say, I see you do not enter male and female. Is it possible for you to do this in the future? So you're again providing incremental comments so they can improve this over time. So that's male and female. I will click on continue. This is by age group. And again, hospital 001, hospital 01 have ages. The last two columns don't. So it's just it's missing from those files and just a little nudge them. Can you do this next year? This is now about the location. So here's the dental, dialysis, emergency department. It's very specific for that hospital. Gynecology, intensive, LND, maternity, spelled over a bunch of different ways. So that's also another data cleaning issue. Like they have an MSW, that's probably, I don't know what exactly, but it's like surgical ward male. So they're not consistent. So this is part of data cleaning. Can you please be more consistent in the future? Maybe that's the bed, I don't know. And that's from hospital one or hospital two. In fact, that you can tell those codes come from hospital SKH because you see that column is filled in. If I go down further, these locations are from hospital TST. And these are nice and clean. And also hospital 01, we discussed you don't enter that, you entered into department. So it's nothing wrong with that, but it's not the same way the other people have done it. So it just makes it harder to compare the data between facilities. When I click on continue, it now does it by location type. So this is much more standardized. I can see hospital 01 and SKH don't enter that, which is unfortunate. We don't know if it's inpatient or now patient. I think you put that information into department. So it's not that it's not there, but it's there in a way different from the other hospitals. I mean, it's just less, it's more awkward to integrate the data. So we'd like to standardize where people put the information so we can facilitate, you know, how this is, you know, the analysis. Good. And I click on continue. There's now a specimen type. So SKH, most common location, most common specimen is urine. Tests, hospital, also urine, hospital 001, most common is blood, followed by urine. And I'll click on continue. So that's just a part of recognizing different issues that different hospitals may have. Okay, I mentioned about the 0101. I will review what we discussed last week. Last week, how do we fix that? We can go to, there are different ways to do it. One way is I simply open the file. So I'm going to do NRL and there it is. I click on view database and here I see the data. And here you can see some 001s. But if I keep on going down somewhere, you see the 01. So you see those two different things. So as a reminder of what we did last time, how can we fix this? Well, if we want to fix them one at a time, we can click on edit isolate and we make the change on this screen. Or I can simply click on edit table. And I can say that, let me go back to the top of this list. These 01s, this should be, oh, that's right. I don't already fix this. Right now I can't edit this because it's a hidden field. But we can edit these other ones. E coli, E coli, et cetera. The reason we cannot configure lab is if I click on edit isolate, edit isolate. The lab column does not appear on the screen. So because it doesn't appear on this screen, we can't edit it here. Adam fixed that. So there's no reason you couldn't fix it here. So in short, edit isolates one at a time, click on edit isolate. To edit the table, you just click edit the table. This inability to edit 01 was fixed last week and I just have to update it. Okay. And good. So that's when you're doing just minor fixing. Like, let me go to the Bank of Mison. You know, if you find some Bank of Mison resistant, you find that isolate and you click on edit and then you just fix that one isolate. You could either delete it. If you're not sure, you can just go ahead and delete it if you're not sure what the result is. If you retest it, then you can confirm it to the correct value. So let's see what else can I do there. I'm going to go back to view database. I'm going to click on the column. One of the most common errors in the specimen dates. I'm going to click on specimen date once. So here you see these data are from January 1st, 2019. I click on specimen date again and they run on December 30th, 2019. This is nice and clean. These are all data from the year 2019. But a lot of times you find typing mistakes, maybe from December 2018, just before the year started, or people just mistyped. They'll put 1963 because they just put the year of birth rather than the year of the specimen. Or they'll put the year 2030 because they're just typing the wrong date. So you can easily see that. I just sort by anything else. Sort by lab. You see a lot of different institutions here. So that's interesting. These are all different. This is all the same laboratory. This is a laboratory SQLite. No, actually it's your data. This is your NRL. And this is the name of the institution. I didn't even realize that was here. I'm going to show you how we can also utilize that. Let me go to continue. Let me go to exit, data analysis, analysis type. First I'm going to do this. It's a related data cleaning. I'm going to do a summary. Okay. All. Okay. Data files. I think that was NRL. Good. Let me choose that. Okay. Right now it is going to do organism by date. So here I can see the distribution. That's an e-director by money. You see the distribution. I think we even found this outbreak last time. This is organism by month. I'm going to change that and put organism by laboratory. I don't think that's going to be helpful because we only have two labs. Lab zero one and zero zero one, which is not interesting because it's the same lab. So that's organism by lab. But now what I can do is organism by institution. I put this in the wrong place. Organism in the rows, institution in the columns. So here I can see organisms are still in the rows, but each column is the different institution. So let me go on E. Coli. So E. Coli is mostly hospital ABT. Hospital OT. I don't know if that's a real hospital. And then hospital OTH, which I guess is other. So you do need to be consistent. Is it OT? Is it OPH? Or maybe they are different. I don't know. ALH. So we talked about data cleaning, but data cleaning is always related to epidemiology. Do the numbers make sense to you? If you had a ton of things under unknown data, if it simply said unknown, it means that the person simply forgot to type the data. Or if hospital ABT is everybody else's owns or something. I go to the patient's name. I look for the person's name or values. Same thing for organism. If I go to organism, this is a list of every single organism in this database. So there's one value of filters. The second value is I can click on select all because I don't want to see everything. I only want to see the E. Coli. And now I only see the E. Coli. And you notice a few things. You notice I only see the E. Coli. You notice the icon change. This is an active filter. The icon is a little bit different. Now let me just make that a little bit larger so it's easier to see. So I just change the icon, meaning that the filter is on. The other thing that happened, let's have to move my go to meeting control, is that the lines turn blue. The row numbers turn blue. We have row number one, row number four, row number eight, row number nine. So you can see how valuable filters are. If I just wanted to find the CRE, I can just go to the imipenem column and I just look for the small zone diameters. Like six millimeters is complete resistance. So Excel is valuable if you know Excel from doing a lot of these simple manipulations. We cannot edit the data in Excel. Excel will not do that with the Debase or SQLite file, but Excel is very valuable for visualizing your data. So we want to take some of the, and I forgot to show you the search and replace. Well, let me do that. Let me reopen that file, which I just closed. File, open, double check, test. So let you see here that but maybe like the problem is I was going to tell, I'll do this example. You just need to make sure there's enough space. I'm going to replace C-surge with cardiac surgery, which is more obvious. So I'm going to say replace or replace all. So we've replaced all of it. Do you have the option to replace them one at a time, replace all, or you can just do find next. So some of these features that are extremely valuable in Excel, we will start to put them into Hoonat, filter, find, replace, replace once, replace all. So some of your cleaning like the 001 issue, I would suggest to wait until we have that replace feature. Oh, there are the specific examples that you would like to discuss about data cleaning. As I said, for advanced data cleaning, I often do that with Excel, I'm sorry, with Microsoft Access. Like if somebody put in me Penham, I think, well, let me see the one that I know. If somebody put, for example, Cephtrax on 30, that is correct. And somebody else put Cephtrax on 10, that is not correct. How do we fix that? Well, actually, for that one, actually, I can tell you how to do that. We're going to data entry. I can say modified data file structure. Why not? Well, it depends. Do you have Excel 2003 or later than that? Yeah, later. I know that you don't have Excel 2003. So Hoonat uses a very old-fashioned data structure called Debase. If I click, I'm going to save this. I'm going to try clicking on save. File, save. Excel tells you we can no longer save Debase files. To save your changes, click OK and save in the latest format. They're requiring that you update this to Excel. So here you see the list of all of the options. Excel, CSV, Web, Unicode, Excel. Debase is no longer on this list. So Debase was on this list in Excel 2003. And then Debase is not a Microsoft product. Every year, Excel makes it harder to use Debase files. We are very glad that Excel is still able to read a Debase file. I don't know why they allow you to read it, but they don't allow you to save it. I am glad you are at least allowed to read it. I commonly recommend that people use Excel for visualizing and seeing their Debase files, but Excel no longer supports you to save it back to a Debase. Okay. And you can't save it to the CSV. That's not a problem. So I can save it to Excel. So it's not as if I cannot save the changes. I have saved the changes, but the saved changes are in the Excel file, and Hoonet doesn't understand Excel. So yeah, I could save it to CSV, but then it's not a Hoonet file, and Hoonet would no longer understand it. Okay. Thank you. I never even thought of that of actually using CSV as an option for Hoonet as a data structure. It's not a modern structure, which is why we never supported it, but Debase is for 20 years, and now there are problems every... We had a few problems with Debase in the last few years, and then it started to go in February, and the last two months, we've just been having a lot of issues. I do think it's because that Microsoft did some automatic update. There is even a web page. I don't think I can easily find it, but in January, it's on the Microsoft web page. Microsoft DAO temporarily removed support, and they need something. A lot of you may not know this. I only want things from this year, so I'm going to say anytime I'm going to say from the past year. So you don't have to do the whole thing. Double shooting, Debase, remove, remove. No, I don't find it, but Microsoft in January made a statement, we are no longer going to support the DAO technology, which is what we've been using for 20 years for Debase, which is why we had to rush. We're planning on transforming it at some point, but this year we really had to rush it because Microsoft unexpectedly pulled the carpet from under our legs, and that's why you see these little things coming up, especially the most common features are working, but some of these things like create a lab from a data file is not so common, but we did fix it last week, but I was unable to install it right now, and this issue about modify a data file structure is also not a common feature, so we've been prioritizing, of course, the common features that people are utilizing, but we're getting there, and so we want everything to work exactly as it's supposed to. In fact, these training courses I'm doing have been very valuable to me because when I do my testing, I don't test everything, during these workshops I do a lot more testing, so a lot of times after these workshops I immediately call Adam and say these issues identify during the call, and most of the time you can fix it the same day or the next day. Next question or next issue? Maybe another question? Sure. Yeah, you know, sometimes pathogens may be tested against unrecommended drugs, so my question for you is, is there any mechanism to identify specifically in the unit that has these kind of errors? Yeah, there are different ways to do that. Sure, let me close this. We talked about many ways to identify problems, and then we started about cleaning problems. I just want to summarize them here. Edit isolate and data entry. Edit table and data entry. Modify data file structure, for example, if the problem is the name of the field or antibiotic disc potency. There is, and then there's more advanced things like access, and ask John for assistance. Then you can do this on your own, but for a lot of things you will not need that, but if you do need it the first time we can just do it together. I think those are the main strategies that I was going over. Okay, great. Oh, also, I forgot about this one. Edit in access. Okay, in access, edit individual data points, edit data file structure, or merge columns together, including search and replace. It's very easy. You cannot edit data files with Excel, but you can edit them with access. If you link or import your data to access, you can do a lot of very simple things. Just edit, just like we were doing in Excel. Very simple. Anybody with any knowledge of access can easily do that. Edit data file structure. Most people also know how to do that. Merge columns together is more complicated. You need to know how to do what's called an update query. These are additional strategies. We're trying to get more and more of these features built inside of Hoonat to decrease the need to edit them in some of their software. A lot of it, like search and replace, would already take care of a lot of people's needs. Okay, great. That's all I have to say about that. Now, regarding your new question, you were asking about invalid data. Good. Let me go to something like, let me just go to data entry. Let me go to step orious. So let's say, let's go for something like, okay, so here on the right side of the screen, you see the break points? Case in 15 to 16 is intermediate. I'm hitting enter, enter, and you always see the break points. Okay, but this one, you don't. You see it's Cephapemesidobactam. I don't know who in the world paid. I've never heard of that. It's a brand new drug and there are no break points. So whenever there are no break points, that is a sign that it's not a valid test. It might not be a valid test because there are not break points yet. And this is probably an example of that. Like there might be valid QC ranges. So people might start testing it for QC before CLSI has made break points. So one reason there are no break points is there are no break points yet. Or let me see if I can find Novobiasen. Is that on the screen? I don't see Novobiasen. Okay, I'm just going to keep on hitting enter. So this one, no break points. Cephotaxymclobulinic acid. This is not for antibiotic interpretation. This is for confirming ESPL. So it's not an antibiotic that has break points. Next, next, next. Cepparome. That no longer has break points. There are no CLSI break points. Similar explanation. Ceptoberprosonet is going to tell you. Let me just choose the other example I wanted to do, the Devoucher test hospital. And if I choose, if I do the same thing with the Devoucher test hospital, go to Stephoreus. And Novobiasen. I did notice it on the list. But let me go to the complete list, the only antibiotics. Novobiasen. So let me go. Meslasilin, yes. Minusectin, yes. I'm looking here at the right. I'm looking to see their break points. Minusectin, yes. Nitrofrenterin, yes. Norephloxacin, yes. Novobiasen, no. The word break points for this drug 30 years ago. And then Celesi said, it's not a, it's not a, it's not a reproducible reliable test. So every time you see no break points, that is a clue that it is not a valid test. What happens if I type, for example, 30 millimeters? Hoonet says the interpretation is question mark. So that's one way to know something is not valid is you get, you get the interpretation is question mark. I'm going to go to vancomycin and I'm going to type six millimeters. That also gets a question mark. Vancomycin in a general sense is a valid test, but it is not a valid disc test for staph aureus. You know, it was a valid test in the 1990s, but then they realized it just wasn't reliable. You had a lot of resistance strains that incorrectly were being called susceptible. So in this case, vancomycin, it's not that it's not a valid test in a general sense, but it's not a valid test for, it's not a valid test for staph aureus. You know, similarly, if I put E. coli, you know, and if I go back to all antibiotics and I put vancomycin. So every time you see that question mark, it's a sign that it's not a valid test. That's one way you would know, simply on the data entry screen. Let me go to data analysis and I'm going to do a percent RIS for staph aureus and let me choose the sample database. Okay, okay, but I did not forget to let me choose the database again. Okay, begin and question mark. So here look at this to the vancomycin zero and this is confusing to people. It's zero percent sensitive, which sounds great, but then it's zero point intermediate. It's also zero percent sensitive. It is a hundred percent question mark. And if you look at the break points, there are no break points. So that's a second clue. It's the same thing, but it's on a different screen. So if you see there are no break points, that is a clue that there that this is not a valid test. Good. I'll now show you a third way. I'm going to exit. I'm going to go to file, I'm going to go to quick analysis. I'm going to quick analysis. I'm going to choose the Hoonet standard report, which we saw previously. Hoonet standard report. Let me choose that one month of data. Let me click. Okay. Let me begin the analysis. We saw this earlier about data field, complainers, organisms, antibiotic results. You see section eight, section G is called laboratory configuration. And here it's telling you no break points have been defined for the following antibiotic. The most common reasons for this is there are no approved break points for this antibiotic, meaning it's an invalid drug, or somebody chose the incorrect disc potency. It's supposed to be ampicillin 10 micrograms. If they put ampicillin 20 micrograms, they would also get these question marks. So there's a interpretation equals question mark invalid test. If you enter your national protocol, they should go with the national protocol. If you notice that, you know, the national protocol says 30 and they're doing 10, you know, then that's also a sign you can just check manually in that direction because you'd have to be an expert. I personally top of my head thing, but there are certain patterns. Most of the cephalosporins are 30. So sometimes you'll notice like, like most of the quintalones are five. Most of the macrolides are 15. So sometimes you'll just happen to notice that somebody, somebody just chose a wrong disc potency here. But you can only do this if you know what the correct dispotency in five. But there are experts who really know what they're doing, who have other responsibilities, who might decide to choose something invalid, but they already know they're experts, they know what they're doing. So I don't want to suppress the ability to find these other quote unquote invalid tests, but I do want to give the option, the user to say the short list or the long list. The short list would be the valid ones. And the long list would be every possibility. So this will be so here on the right, because I've been doing this, you see, I have three different sypros. Every time, let me just highlight those three. Let me just put this in alphabetical order. So another sign that there's a problem is whenever you see the same antibiotic more than once. So here I see disc diffusion four different times. Sometimes that's correct. Like ampicillin, there are two different valid ones. But whenever I see the same antibiotic more than once, I want to ask myself, is that a mistake or is that true? Okay, I hope that answers that question. In short, any time there are no breakpoints, that's a sign this is an invalid test. And there are different ways to know that the breakpoint is invalid. Data entry, you'll see a question mark. Data analysis, you'll see a question mark. The UNED data report will tell you there are no breakpoints. Yeah, maybe John. Hello. Yes. You know, once upon a time, we found a problem like they are testing ampicillin for clebsaele anemone. You know, clebsaele anemone is intrinsically resistant to ampicillin. So ampicillin, it has a breakpoint if you see it in the UNED. So that kind of issues may come to us. So how can we hand this kind of? Sure. Sure. What you were just describing is not in is there's no problem and I will explain why. But before doing that, you see how I did this. I'm going to open the EPHI laboratory. I'm going to click this is the one that you sent to me. So I'm going to click a modify lab antibiotics. I'm just going to take a quick look here. Oh, I definitely clicked on cancel. I actually removed it before I meant to modify lab. So you asked me about invalid antibiotics. I already have a clue off the top. And I'm looking. So here you see, is it for mice in twice? The 15 is valid, but the 30 is not valid. But like, I know which one is valid. You may not know, but the fact that there's there twice is a flag. And so the 30 is invalid. I could simply remove it. But I don't know if there any data there is that if that column if so, let me even open up this file. This is this is related to data cleaning. Let me open up that file. Let me click on the database. Oh, I must have chosen. Sorry, I chose an empty. I chose the wrong file data entry, which is the real. This is the real data. No, I just open the database. And here, if I go over to the right, you're going to see that this drug appears twice. So here you see a zithromycin 15 and a zithromycin 30. This is also a very good aspect of data cleaning. Does this column have data? Let me click on this once. This column has data. Let me click on this other one. This column has four results. So what I would like to do to clean this up is edit the table. Six, 25, 24, 23. Does that make sense? I'm just moving it over. Okay. They have 15, they have 30. My guess is they do not have a disk of 30 micrograms. I don't even think they sell them. So so sometimes people test the wrong drug. I can't do anything about that. They tested the wrong drug. Sometimes they tested the right drug, but they incorrectly configured it in Hunet. My guess is the laboratory tested the 15, but somebody accidentally chose the 30. They entered four results. So this is a nice example of data cleaning. I just simply, I just simply moved it over to the right column. Does that make sense? Yeah, yes. So now that I've done that, I don't actually have to delete this column. This column is empty. So there's no need to delete it. It's just going to be ignored. But if you wanted to clean things up, yes, you could go ahead and delete the column as well to really clean it up. So now if I click on continue, safe changes, yes, and exit, and I go to file, and I go to modify laboratory antibiotics. So the call, I still have it on the list. I have azithromycin and azithromycin 15 and 30. I can remove the 30 with no bad consequences. The column is empty. I did not actually just delete the column, but who that's not going to look for it when it only looks for things on this list. So let me go down further. The cephabemesidobactym, I don't know if you actually have that disc. It's a brand new antibiotic. These are actually wonderful new drugs, but they're still in development. There are no breakpoints. You see here also look here, it says ND blank. Blank means there are no, there's no disc potency, meaning who that doesn't know what disc potency they are working on. So that's another clue. If you see there's no disc potency and you're doing this, there's obviously some mistake. Every disc has a potency. Let me go down further to see if I notice other things of the similar nature. You're just a comment. Nobody tests carbenicilin anymore. It's a very old drug. Cepheroxymaxitil, people no longer test. They just test cepheroxyne. I think dorypennum, I think it's been removed from the market. I'm not sure of that. I think it was and then it wasn't. I'm not sure. Here we have two genomycines, but that's correct. Genomycin 10 is the normal one for most bacteria. Genomycin high is the one for enterococcus. Linazolid, merpenum, peprosilin by itself. Nobody tests that anymore. It's just too old. Venkomycin. Here it's twice, but that's okay. This is the Venkomycin disc. You can tell because it has the letter D and it also says disc at the bottom. This is Venkomycin MIC. That's fine. I mean, these are just two different tests. Good. That's another way. You just simply look at the list. You need a bit more knowledge, but you do this enough. You sort of look for things that just don't make sense. The same drug is there multiple times. The disc potency is a missing. It's some drug you've never ever heard of that you know nobody is really testing. That was my further comment because I did want to do this. I wanted some example, which was not only my data. I did want to use some of yours. This was a good example for that. Regarding this issue about CLSI, and we just have 13 more minutes, so I'm just watching the time. Oh, no, we don't. We have 40 minutes. That's right. We go until 9 30. Okay, good. Let me go to the internet. It's just a good opportunity to talk about the CLSI document as a resource. Where's my Google Chrome? It's always moving because I've got too many icons. CLSI free. I'm just doing a Google search, a web search for CLSI free. And I bring up free resources from CLSI. They do not make everything available for free, but they do make some of the key documents for free. M100 is routine bacteriology. M60 is yeast, the fungi. M23 is not for you. That's for people making breakpoints. You're not making breakpoints and you're not making QC ranges. M23S is the supplement. Vet 08 is a free document that is routine veterinary bacteria. There's also the documents that are not free. M45 is fastidious bacteria or rare bacteria like bioterrorism. M61 is for mold. Human for mold. Vet 06 is for fastidious veterinary. Vet 03, Vet 04 are for aquatic animals such as fish or shrimp or shellfish or lobsters. So we're going to go to the M100. That's the basic human one. Operating corner. I click on click here to use guest access. And you see I have these three free human documents. Had I clicked on the veterinary, it would have shown me the veterinary documents. M23 is for making breakpoints, making QC ranges. That's only for the experts. M60 is for the yeast and M100 is for the human routine bacteria. Let me open that up. So you do not need a login, no money, no password. This document is free. On the left it says TOC, table of contents. And let's jump down immediately to table 2A. Well, okay. At the top, very valuable overview of changes from last year. So if people do not have the money to buy this document every year, they can look at, and even if you do have the money, you still want to just know what the changes are. So if I click on overview of changes, this just summarizes they replace the word infection control with the word infection prevention. They replace coagulants, negative staff with the word other stuff with coccyx. So that's nomenclature changes. And then if I go down further, eventually they change table one, table one is what do they recommend for testing? Table two are the breakpoints. So table two, clarify added, polystyne, it's either added breakpoint, deleted breakpoint, comment about the breakpoint, clarification about the breakpoint. So I'm going back to the TOC, the table of contents. So the overview of changes is very valuable. And then table 1A is what they recommend for testing. That's another way to know what they should be testing. Table 1A in a general sense is what CLSA recommends to the world. Nobody does exactly this. This is just a general guide. They say like group A is always test always report. Group B is always test selectively report. Like in the United States, we have antimicrobial stewardship. If you have a simple outpatient urine infection sensitive to 15 drugs, don't tell doctor all 15. Just tell the doctor the first few, the cheap oral drugs, first line drugs we want the doctor to choose from. On the other hand, if it's an ICU patient, if it's multi-resistant, if I'm doing this for epidemiology, I want to see everything. So group A is a general recommendation, always test always report. Group B, always test selectively report, including some extremely new drugs like meropenum vapor bactam. That's valuable for the CRE. So these are new combinations. Group C is selectively test. Don't bother testing these always. Just test them if you feel it's appropriate. Second line, if you have personal interest. And group you are urine drugs. So here at the top, you see table, let me do that again. Table one A, this is enterobacterialis, pseudomonas, staph, and enterococcus, continued here with the non-permentors. Acinetobacter, burglderis, genotrophomonas, other non-enterobacterialis followed by table one B. Well, table one B, I have to go back to my table of contents. Table one B is what CLSI recommends for testing for the fastidious organisms, homophilus, Nigeria, strep pneumonia, or the other strap. That's group A. Group A is always test always report. Group B always test selectively report. Group C selectively test. And group you urine, but these are not urine pathogens. I'll go back to table of contents. I go to table one C. Table one C is what they recommend for testing for the anaerobes. And a lot of you do not test anaerobes. See these are for gram-negative, gram-positive. So these are general recommendations for what labs can be testing. And I recommend you do not use this list as is. This list is not appropriate for anybody. The concept is relevant. You choose the drugs that you feel are relevant. And a lot of people, especially if you're testing a limited number of agents, you always test and always report. You do not have to do selective test, selective reporting. We do that just to kind of steer doctors to the cheaper drugs. Let me go back to the table of contents. Let me now go to table 2A. I'm giving you a roundabout answer to your question, but this is a different education. This is also, these are also valuable details. So table 2A, I'm going to, you mentioned about plebsiola and ampicillin. So this table is for enterbacteriallis, which is the same as enterbacteriaceae, which is family genus, you know, a kingdom family order, one is family, one is order. So this table is for enterbacteriallis, such as e. coli and plebsiola. So here you see the breakpoints for ampicillin. So there are no exceptions. All enterbacteriallis have the same breakpoints, including clepsiola and ammonia. So it is valid to test clepsiola and ammonia for ampicillin. The expectation is the result is probably going to be resistant, but it might not be. That's why we have the breakpoints. As long as I'm on the screen, you can see results of ampicillin can be used to predict results for amoxicillin. You do not see amoxicillin anywhere on this list with the breakpoint. There is amoxicillin clavulinic acid, but that's different. That's amoxicillin plus something else. So amoxicillin by silicide is not a valid disk diffusion test. Instead, test the ampicillin and whatever you get for the ampicillin, just tell the doctor, if the doctor asks them what the amoxicillin result is, just tell them the ampicillin result because it's a proxy. It's a substitute. So in answer to your question, if you test clepsiola and ammonia, and the zone diameter is 30, then the report is sensitive. It's unusual, but it is sensitive. There is another table here. Let me go to another table here called intrinsic resistance. It's one of the appendices. So here you see appendix B called intrinsic resistance. This is natural resistance. So what they define, let's see, I was hoping they would say some specific words but they don't. But let's look at clepsiola and ammonia. So clepsiola and ammonia, let me get the column heading. So clepsiola and ammonia is intrinsically resistant to ampicillin and intrinsically resistant to tyrosilin. Whereas, I'm not going to expect to hear, what is interesting, I did not know this. Well, clepsiola erogenes used to be enter-bacter erogenes. So clepsiola erogenes is intrinsically resistant to augmentin, but it's not intrinsic, but clepsiola and ammonia is not. So this is a very valuable table. What CELOSI does not say is they do not say automatically change the interpretation. They're using this as a guide to quality control. So what happens inside of Hoonat? Let me go to Hoonat. Well, I think I had Hoonat open already. So, okay, let me just say, save and let me go to entry and do data file and replace. Let me go to E. coli and let me put ampicillin 30. No comment, completely normal. Let me click clepsiola and let me delete that. Let me click clepsiola and ammonia. And let me type ampicillin 30 now. And you get what's called a low priority alert. Susceptible isolates are rare. Check for a possible laboratory error. Maybe it is not ampicillin sensitive. Maybe somebody mismeasured it or maybe they measured the wrong disk or maybe they put the wrong disk. Or maybe it is sensitive, but maybe it's not clepsiola and ammonia. Maybe the identification was wrong. So CELOSI does not tell you to change the interpretation. They tell you, please double check your work and double check your answer to make sure there's not a mistake. What you can do here is you can manually change it to resistant. If you're not safe, if you don't feel comfortable, I would not feel comfortable telling the doctor that the clepsiola and ammonia was ampicillin sensitive, unless I really checked it and double checked it and double checked it because the clepsiola and ammonia is not 100% resistant, but it is about 95, 98% resistant. So if I have clepsiola and ammonia, ampicillin sensitive, I would retest it. If I do not have the ability to retest it or in the short time, I might immediately tell the doctor resistant because it's probably resistant. So the reason HUNET does not change the interpretation is that CELOSI officially does not tell us to change the interpretation. So that's why there's not a mistake in HUNET where you're simply following CELOSI recommendations and CELOSI says, yes, this is a potential quality control issue, but CELOSI does not say to change the interpretation. Sometimes they do. There are other examples where they say to change the interpretation. For example, I'm now going to put cephalococcus aureus and I'm now going to say, I'm going to say 30 for everything, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30. Just as a reminder, the vancomycin has a question mark because it's not a valid test, just reminding you of what we talked about a little while ago. Now I'm going to change the suffoxicin disc to six millimeters. So right now you see S everywhere. As soon as I change the letter 30 millimeters to six millimeters, several of the drugs have changed to R. The oxysilin is now R and the penicillin is now R. Well, not several. They're only, so basically the beta lactants have changed to resistant. So CELOSI has a recommendation. If this is AC, if this is an MRSA, just go ahead and change these other things to resistant, ignoring the zone diameter. Just ignore the zone diameter. CELOSI does not say to do that for Cephalococcus aureus and ampicillin. They do say to do that for confirmed MRSA. So does that answer your question? Exactly. Can't you? Let me show you something using the ICID alerts. Let me go to data analysis, analysis type, and ICID alerts. Yes, is there a question? Okay, so ICID alerts. I'm going to, so Hunet has a number of nice ICID alerts. I'm going to click on options. Hunet has about 190 ICID alerts categorized into high, medium, and low priority, categorized into important species, important resistance, and quality control, plus a few other things. In answer to your question, I don't want to, I want to see all of the alerts. I, right now, I do not want to see, I, right now, I only want to see the quality control alerts. So, for the microbiologist, epidemiologist, infectious control staff, maybe pharmacy, they would like to see the important species and the important resistance. Very valuable. MRSA, VRE, imipenem, vibrocholera. So for those people, they want to see important species and important resistance. But into your enter on quality, which is a value to the national data managers and to the microbiologist, right now, I'm only interested in quality control alerts. I'll put all three high, medium, and low priority alerts. And let me select okay. Let me select organisms and say all organisms. I can say data files and let me choose the Ethiopia data. No, I'm sorry, that's the wrong one. This one here. Okay, and begin analysis. Hunet is showing me, this is not showing me all of the isolates. And if you look at the top of the list, this is wonderful. You only have 58 isolates out of the entire database. Only 58 isolates had a quality control concern. And here they are. And the red does not mean it's an error. The error means that the red means it might be an error. It might be true, but it might be an error. And let me go over to the right and you'll see a comment of why that is there. Let me just find, let me make this column a bit wider. And let me make this column a bit wider. Okay, so here you see enterococcus fecalis. So enterococcus fecalis is a, where is it, ampicillin? Fecium is usually ampicillin resistant, but enterococcus fecalis is usually ampicillin susceptible. So it's just basically susceptible isolates are rare. Or was another example, like pseudomonas sensitive. So here we have pseudomonas. Let me go back to some of the reds. So these pseudomonas are pipicillin sensitive, but it's usually resistant. Some of these rules I want to adjust. Some of the rules I disagree with, but so all of these are basically, okay, here's another one. Staphylococcus testipitis diffusion. Staphylococcus diffusion is not recommended for oxacillin. It should be the suffoxitin disc. So there are different kinds of quality control results here. Or for here, you see discordant results. Let me just, I'm going to highlight the row beneath it so I can see the red. So let me, where's my red? I'm looking for the immunoglobulic sites. So here you see that the amication is resistant. Amication resistant, but gentamicin sensitive. Gentamicin resistant, I'm sorry, let me rephrase that. Amication resistant, gentamicin sensitive is not common. Amication is much newer than gentamicin. It's much more expensive than gentamicin. Around the world, it is very rare. It does exist. So it's not as if it's impossible, but it's relatively rare. So UNED is, again, that's a kind of alert. It's called discordant results. It's just the immunoglycoside results. Usually agree with each other, but if they don't agree with each other, they don't agree with each other in a good way, in an expected way. Amication sensitive, gentamicin resistant is common. Amication resistant, gentamicin sensitive is not common. So somewhere on here, I don't know about in your database, but in my database, I do have that example about Klebsiella anacetylins sensitive. I don't know if you have any examples like that. So I'm looking, no, it just goes in alphabetical order. So here we have Klebsiella ozanii, which is augmentin resistant, but peppericillin sensitive. See, that doesn't really make sense. If it resists into amoxicillin, clavulinae, I guess it should also be resistant to peppericillin. So that gets an alert like that. The red shows me precisely exactly these issues of just unexpected results that may or may not be mistakes. Click on continue, and it will give me a summary. So the most common, so here we have a lot of low priority alerts. These are really just basic education that microbiologists should know on their own. They really should not need to find these things. But it is good for educating new staff, or if you're the national data manager, you're busy people, you know, it's nice to have a nice simple way. And here it's also separated by lab. So you can see that lab 01 has these quality issues, laboratory 001, which is the same lab as these issues. So you can see which quality control issues to each of the different laboratories have, invalid tests, rare resistance, discordant results. So then you can focus in and say that some of these rare results might be true. And firstly, I'm not an expert. I know that SNE Director Balmani is usually resistant to quintalones. And here there was one isolate where it was quintalone sensitive. That doesn't mean it's a mistake. It's just pointing out for you, it's rare. And it's important. I mean, it's nice that it's sensitive. But the doctor has a sick patient with them with SNE Director. I don't want the doctor to give the Cypher Phloxacin, unless if I double check my work, you know, we have what are called minor errors and major errors and very major errors. An example of a major error is that the patient has a sensitive strain. Let's take the example of Cypher Phloxacin. A major error is the patient's strain is sensitive, but the laboratory said the strain was resistant. This is bad because the doctor could have used quintalones, but they didn't use the quintalone. They didn't use the Cypher because you told the doctor it was resistant. That's a major error because there was a lost opportunity. You could have used Cypher, but you didn't. That's a major error. There's something even worse than a major error, which is a very major error. The very major error is the patient has a resistant strain, but the laboratory said it was sensitive. The doctor says, oh, the bacteria is sensitive to Cypro. Let me give the patient Cypro. This is very dangerous because you've given the patient an antibiotic that is not going to work. Whenever you see, like, Clepsilandemonia, if you tell the doctor, Clepsilandemonia, Empaciland sensitive, that could be dangerous because the bacteria, it might really be resistant and the lab made a mistake. So that's a very major error because the patient may get a wrong drug. The major error, the patient's not going to get a wrong drug, but you could have used a cheaper drug that could have worked, but, you know, you didn't because you thought it was resistant. Other questions. I have one more other comment about the Clepsilandemonia Empaciland sensitive. It is possible. It is, this is a chromosomal, it's a beta-lactamase chromosomal class A enzyme that causes the Empaciland resistance, but that enzyme can be lost. The gene can be lost, or the inducer can be lost. So there are Clepsilandemonia that, indeed, are Empaciland sensitive. It is relatively rare, but even in that case, I still wouldn't want to use it because maybe the laboratory did make a mistake, or because we only tested the bacteria overnight, you know, in an 18-hour incubation, you know, the bacteria appears to be Empaciland sensitive and that may be true, but if the patient's going to be treated with Empaciland for three or four or five days, the bacteria might become resistant a few days later, especially if the gene is present but not being produced because it's just an inactive gene, but after two or three days by selective pressure, the gene may turn itself on again. So these are some therapy issues where intrinsic resistance is very valuable. If it's intrinsically resistant, but the labs are sensitive, you still probably really want to avoid it and not give it to the doctor, not tell the doctor to give it to the patient. I do want to make a change in Hounet to even help this further to protect patient care. Hounet has this area, well, if I go to data analysis, Hounet has these things called expert rules, options, use expert interpretations. For example, if it is MRSA, but Penicillin sensitive, it isn't. Change the Penicillin result to resistant. So Hounet already does change the interpretations for a few things. If I click on file, modify laboratory, antibiotics, breakpoints, expert interpretation rules. So all of you are familiar with this screen. I hope you're familiar with this screen about where you do the breakpoints, but we never really discussed this option here called the expert rules. Hounet only has a small number of rules. In fact, these two rules are no longer recommended. These were CLSI rule that they no longer recommend. If you have an MRSA, it is resistant to the other beta-lactams. If you have a Mophilus influenza or inducible clindamycin, if you have a strain with inducible clindamycin resistance positive, then you need to change the clindamycin to resistant, even though it appears to be sensitive. So these are Hounet rules that change the interpretation. These are official CLSI recommendations. What I would like to do here is to also put in the intrinsic rules. I showed you that from the CLSI website. What I would like to do in Hounet is Hounet does use these rules for the alerts. I already showed you that it does use these rules for the alerts, but it does not change the interpretation. And I do want to offer that to the user as an option for our data clinical reporting purposes. If it's intrinsically supposed to be resistant, I want to tell the doctor resistant just to protect against very major errors. Don't tell the doctor sensitive when it's really resistant. So these intrinsic resistance tables will help us to do that. So in a future version of Hounet, not August, because we've got a lot to do, but in the fall, September, October, you will see another option here for intrinsic resistance rules. And then you will see another option in data entry for clinical reporting and in data analysis for turning the intrinsic resistance rules on and off. So this would allow to do what your expectation was. You thought that Hounet would change it to resistant. Hounet doesn't because that is not a CLSI recommendation. However, a lot of people do it anyway because they're trying to protect patient safety. There are many labs that do change it to resistant precisely for the reasons that I mentioned. It's not a CLSI recommendation, but a lot of people do it anyway. Hounet doesn't, but it could. And we'll put that in in the next few months. Next question. As long as we're on the screen, I do want to take advantage to re-emphasize the value of paying attention to these rows and columns. So summary, a lot of things, you are national responsibilities. A lot of these I do by lab so that you can do the benchmarking. I select summary. I can do organism by lab. I can do anything by lab, or I can do it by organism and lab by specimen date, by month, or by year. So I do encourage you to be creative about how you define your rows and your columns. Instead of thinking about I see what Hounet can and cannot do, think of a table that somebody has asked you for. Think of an analysis that would be interesting for you. By being creative with these rows and the columns, you can usually do all the combinations that you might think of, at least the normal combinations. We're going to make some changes here. Right now, Hounet does this by specimen type. That's great, but the problem with specimen types is they get too detailed. Like blood and urine is great and sputum and stool are great, but Hounet also has bile, pleural fluid, joint fluid. There's just too many. So sometime in the next few months, we are going to have a new option. In addition to specimen type, we will have something called specimen category. So we'll have respiratory, genital, urinary. Because I did show you, and I did show you, let me go back to the quick analysis and let me show you the sample statistics and let me go back to that real data file, which is right here. Let me go to begin analysis. This is one. This is lab by month. Sex by lab, age group by lab, location by lab, location type by lab, and here is the specimen type by lab, which I went over as a useful thing. But here you see cervix, ears, eyes, joint fluid, nail, pericardial fluid. The numbers get very small. Prostatic fluid, if I click a number of isolates here at the top, I have the valuable ones at the top. Steer, I don't know if you're so valuable, but ear for me is kind of respiratory, you know, like ear infections, vagina fluid. And then at the bottom, there was one swab, one pericardial fluid, one bronchiolabial lavage. So this is going to be, well, we will leave this, but in addition to this, we will offer category, blood, urinary, respiratory, genital, which is at the highest level, you're just interested in those broad categories. Next, 9.15. So 15 more minutes. Anybody there? Can someone please confirm the sound that's working? Yeah, Gabriez, does this answer some of the issues that you're expecting to review today? Yes, Fern, these are the, especially the quality control, the cleaning and the specimen cleaning, the one he talked about, the antibiotic resistance, especially the adult is very important to inform clinicians at facility level. And in fact, also for our epidemiological reports. So this is what we really need and achieved. I have an idea that would be very valuable for this discussion. I'm going to do an RIS summary analysis by institution. And I'm going to click on okay, because this is just the one, I only have your one lab. For you, you actually have the separate labs. But I do have multiple institutions. I'm going to go to E. Coli. Well, let's start with the simpler one. Let's just go to Staph aureus. Okay. There's a grand positive that fewer antibiotics in the grand negatives. Okay, okay. And first, let me do this overall organism by percent susceptible. Here you see that there were 91 Staphylococcus aureus. Okay. So a lot of people are interested in all these early columns. But from the perspective of quality of testing, not quality of data, but quality of testing, I want to remember there were 91 isolates of Staph aureus. But let me go to the right. And now I see my denominators. They tested azithromycin 36 times, suffoxidin 71 times, chloramphenyl 4 times, Cipro 64, clindus 75. I'm really not that happy with this because it's all over the map. Actually, let me do this as a graph. Let me go back and redo this. But instead of the summary, let me show you the graph. And so let me go to number tested. So here I can see you have a lot of results for penicillin, oxicillin, suffoxidin, clindamycin, Cipro, and SXT. It's a lot, but it's not even consistent. You know, you test more clinda than you do Cipro. So what I would like to see going forward eventually is people test like they pick six drugs and they test those six drugs all of the time. Okay. Good. So that's one comment. I'm not, I'm happy that you have six drugs with a lot of data. But even though six drugs with a lot of data, you have about 70 suffoxidins, but you only have like 40, you only have like 55 penicillins. It's better to pick a certain core minimal list and focus on that. Genomycin is valid. And there's a lot of that. Tobromycin is valid. So the other thing I'm looking at is it valid or not? And here we have vancomycin. The vancomycin disc test is not a valid test. If this is vancomycin MIC, and I can tell from the graph, let me just look at the, I mean, not the graph, the table. So vancomycin disc, so that is not a valid test. So they should not be testing the vancomycin disc, at least if you're in the United States. Because we just immediately go to MIC, we just do MIC testing. Of course, if MIC testing, this is another reason why people do invalid tests is they'd like to do things the recommended way that CLSI recommends, but they don't have the resources to do E-test or MIC on everything. So there are a number of labs that do vancomycin disc testing. They know it's an invalid test. And as long as it's sensitive, they don't, they assume it's sensitive. If it's resistant, then that's when they would want to follow up and do an MIC test. So sometimes people do invalid tests because they know what they are supposed to do, but they don't have the resources to do what they're supposed to do. E-test is an expensive thing. I will not do E-test routinely. It's just too expensive. If I have an MIC machine, we do MIC testing routinely, so it's not an issue. But if we find a vancomycin resistant strain, we retest it first on the machine as an MIC, and then we also do it as an E-test as a special MIC. So that's another reason why sometimes do invalid tests. They know what they're supposed to do, but they just can't. John, can I have a question? John? Yes? Can I have a question? Yes, of course, yes. Sorry for the interruption. Is there any, I mean, mechanism that we can know any emerging pathogen that has a public health emergency trait during the cleaning of data? Is there any mechanism that we can know that any emerging species is coming up that will have a public health emergency importance? Okay, so let's see. Well, let me finish the screen, and then I'll come back to that question. So I also mentioned earlier the idea of about Ethiopia all laboratories. I also mentioned the concept of doing a minimal list. And as you can see here, they test a little bit of Tobra, a little bit of deptomycin, a little bit of nitrofrancine for the urines, a little bit of chloramphenicol. And as you can see, they do take a lot of space. And if I'm having a minimal list, I would just, I would not, you know, there's certain things, there's just not a lot of data, like deptomycin, my guess is sometimes it's all in one laboratory, they only do it for confirmation. So I would have Ethiopia, all hospitals, all antibiotics, what could do everything. But if I'm doing my annual report, my annual report, I don't put everything in my annual report, I just put the most important drugs. So there'd be one use of a minimal national configuration, I would exclude the deptomycin. Chloramphenicol, it might be valuable for other organisms, I don't know if I delete it, because other organisms it might be common. But the deptomycin, I would just go ahead and delete it from the minimal list because there's just not enough data. Okay, so that's a comment about that. Okay, I'm coming back, anything else on this screen? Okay, first of all, on this screen, you see the nice graph of what they're testing. So basically, there's six drugs that they usually test, the genomycin, the azithromycin, the erythromycin, the tetracycline, are tested about half the time. And the question is, did they change practices like sometimes at the beginning of the project, they do things in a historical way, later in the project, they standardized it, they tried to go to a more official standard list, or sometimes they have first line testing, second line testing. So if it's resistant, the problem if they're doing second line testing is the second line drugs are going to have an inbuilt bias. If you only test the amiccation or the imipanum on multi-resistant strains, you will get reliable results that are not representative. You know, if I only tested imipanum five times, I'm happy to see my five results, but those five results do not represent all E. coli. So yes, there was a comment. I guess I have a question for you, is why do we only have genomycin half the time? Do you test it half the time, or did you not test it in the first half of the year, but you did test it in the second half of the year? So why do we only have half of the I-staph aureus have those four drugs, genomycin, azithromycin, erythroin, tetracycline? Maybe it will be lack of antibiotics. Sometimes there's a shortage of supplies. I think that's what I suspect, but I am sure this is exactly better. What I know is every time there is a shortage of antibiotics. So they used what is available in the lab. And it's nice to know what the reason is, because the reason you just said does not introduce a bias. They wanted to test it, but they couldn't. So the data are not 100% complete, but they are representative. So the reason we only have genomycin half the time is because they ran out of discs, then the percent resistance is still reliable, because there's no bias because of missing discs. On the other hand, if they only test genomycin on multi-resistant ICU patients, then we have a bias. So the reason is valuable with regard to understanding. Does that make sense? If the reason you didn't test it as you ran out of discs, there's not a bias there. It's just randomly distributed. But if the reason you tested it was because it was an ICU patient or a resistant isolate from the first line testing, that introduces a bias. Yes. So you always want to understand why is this so rare? Or sometimes it's a typing mistake. Like you see E. coli with penicillin. It's just usually a typing mistake. That's also data cleaning. I did this overall for the country. Now what I will do is I will go back to analysis type and I will do this by institution. I'll say, okay, begin analysis. Oh, and I didn't mean to do that. You see, it just gets too busy. So let me just redo that as the summary, just so it's easier to talk about. There's nothing wrong with what I did, but this is a lot simpler. Because I'm not interested in the RAO statistics. I'm really just interested in the denominators. So here you see, oh, I didn't, did I do it? Where's my institution? I thought I clicked on it, but I didn't. Institution, okay, begin analysis. So now you see one row for each institution. And let me put this in. I'm going to copy this over to Excel. Do I still have Excel open? Yes, I do. You see the percent sensitive? Don't care. I don't care about that right now. I only care about my denominators. And I can delete this. I can delete the staph aureus because I know it's all staph aureus. So here, what time is it? I'm just checking the time five minutes. Okay, let me do this because I think it's important. Max tested. So you see that hospital ABT had 40 isolates of staph aureus, but they only tested is ithromysin 15 times. They only tested spasitin 30 times. They only tested chloramphenicol ones. There's a variable in Excel called max. Equals max of this. Okay, I'm going to remove the word number, placed by blank, upset, replace all. And I'm just making the columns narrower. These are all my denominators, because it's a little bit too narrow. Let me move that. Let me just put like five characters. Okay. Max. Do you see how the maximum, so that hospital 12 has 12 staph aureus, and they tested clindamycin 12 times. They tested oxacillin 12 times. They always tested certain drugs. ABT has 40 staph aureus, but the most they had was 34 chloramphenicols. So this max is valuable. So they don't always test. So equals this divided by this. And let me paste that. I have to copy. Let me paste that. And let me just put this here as a percentage with one decimal point, zero decimal points. So here 100% tested, but this one, they only tested 85% of the isolates, because you don't always test. Sometimes you test it because you run out of all the discs. Sometimes if the patient at staph aureus five times already, you don't have to test it again. Or if it's an E coli in the urine, I'm sorry, if it's an E coli in the urine with a low colony count, sometimes you do not do susceptibility testing. Or if it's a wound, people don't always susceptibility test wounds. So there are good reasons why you don't test. Like here they had four staph aureus, but the only susceptibility tested three of them. So this number tells me that most of them, they have it, they tested 100% testing. This one only does 85% testing. And you might want to ask them, well, why did you not test it? And usually they have a good reason. Well, the reason might be we ran out of discs, all discs. So that's about completeness of testing. And let me, let me repeat, let me copy this over. Okay. And I'm going to do a new formula here. Equals one divided by 12. I'm doing this in a special way using dollar signs. I want to call them C to stay the same. I'm going to copy this all the way across. I need to see my headings. Okay. Copy, paste. And let me just reformat that as a nice percentage with zero decimal points. And I'm going to highlight. So, okay, just as an example, I need to see my denominators. Let me just focus on, I'm going to just show you the first four drugs. Let me hide the other drugs. I'm focusing on the first four drugs. Okay. Oops, I did too many. Okay. So here we have, let me go to a nice example. Okay. Well, good enough. Okay. So they have 12 staph aureus. They tested a zithromycin once. One out of 12 is 8%. 11 out of 12 is 92%. Chloramphenicol was not tested. So that's 0%. So do you understand this? We're basically is how completely was the antibiotic tested. So this allows me to see which drugs have good data and which drugs are missing data. There's actually a nice thing called conditional formatting. Highlight cells greater than 90%. And highlight that in green. So I am happy that every, so look how green column foxes. So I am happy that all the hospitals are doing a pretty good job at that, except SMH is only half the time. Zithromycin, ARH is testing it always, OHC. So we were seeing here 100% tested whereas this one only tested 50%. I went through quickly with mechanics, but I'm trying to, I'm hoping you see the value of this. This allows you with a short number of steps that we could repeat more slowly on the next call, allows you to get a better sense of the different test practices in the different hospitals. And you could even minimize this further. Let me, let me, I want to wrap this up obviously because of the time. But let me do, I only did this, did you show you the four columns? Let me, oh, this is easy. All I have to do is conditional formatting highlights cells greater than 90%. That's a 90% green. So I just made some small mistake. And conditional formatting highlight greater than 90%. And let's highlight it in green because green is good. Okay. So here I'm going to say I don't like, is it, what basically chloramphenicol almost nobody tests that. Arithromycin, it's a great drug. I'm disappointed that it's only 50%, 60%. I'm hiding that. Lenezolid, genomycin, nitrofrantin, penicillin. I'm trying to things that are consistently tested across all hospitals. And this is about the, so I'm very happy with sephoxetine and oxacillin, but even there are some of the hospitals that's only 60%. So this is allowing me to come up with a minimal set of the antibiotics that are most tested around the network. What I would like to do is two things. I would like to get the hospital to comply with this list. And this list should be bigger. Arithromycin should be here. And this is the idea of creating a list of minimal testing that you can then score them on. I don't want to score them on deptomycin. If they're not testing deptomycin, if they are testing deptomycin, I'm happy. But they're not testing deptomycin. There's no expectation that they should be. So I'd like to say that for these, for example, six drugs, we're going to give them a score. This hospital does great. This hospital is perfect. Well, of course, there's only one isolate, that's a different issue. But this hospital always tests all of these drugs. This hospital tests all of the drugs except for oxacillin, which is in fact correct. You're not supposed to be testing oxacillin except as a proxy for sephoxetine. So I'll just leave it at that because we are over, but there are there any questions about what I just presented? I'm happy to repeat exactly what I just did on the next call, because I do think it's very valuable now that we're moving to discussing not only national planning, but facility feedback. We want to provide facilities feedback on probable errors, but we also want to provide the facility feedback on which antibiotics they shouldn't be testing. If this availability is an issue, this is evidence for that. And then you can use that to try to come up with a national strategy, you know, for coming up with a disk, mass purchases, plan ahead of time, et cetera, et cetera. Thank you. Thank you very much, John. Yeah, we're a little past our time. I don't know if they have, if the Ethiopia team has any questions. I don't, we can cover on the next call, but I don't remember what Gebrae's question was. I did not get to his final question. I just want to make sure we record it. Yeah, yeah, yeah. My question was during data cleaning, is there any mechanism or any ways that we can detect emerging pathogens that have a public health relevance, public health emergency relevance? I will answer that quickly because it's so close to relate to what we just did. I'm thinking I should alerts, I should alerts options. You saw how I did this. I only asked for the quality control before. Yeah. Now I don't want the quality control. I want, at the national level, maybe only the high priority. Yes. High and medium priority. That's up to you. I would do high priority. I only want the high priority. Yes. So now I'm going to do this and I'm going to say, okay, and I'm going to do begin analysis. Okay. And that was only staff aureus. Let me change this to all organisms. That's why the list is so short. So everything here has a high priority alert. And the high priority alerts are mostly the imipenem resistance or the urtipenem resistance. So all of these, by the HUNET definition, are high priority alerts. In fact, every one of them. The last three we did discuss in our previous call are vancomycin intermediate or vancomycin resistant staff. So in this database, this is pretty clear, the CRE, which is probably true, at least many of them or most of them. I hope it's true. Well, from a data quality perspective, I hope this is true. From the patient perspective, I hope it's false because it's a lot of resistance. And the vancomycin is probably just a laboratory mistake. So there is an overlap between high priority and quality control issues. You always have to wonder is it one or is it the other? So this is the answer to your question. Yes. Just go ahead and just, you know, just use this feature called ICID alerts. Either ask for important species and or important resistance and or quality control. This is a summary. 22 of them was carbopenem resistance, two or and then two of them relate and three of them relate to the vancomycin. The HUNET list is a list I give to you. I go to modify lab. Here's the list of alerts. All of the lists, all the alerts we talked about are on the list here, like the Cepcella sensitive to ampicillin, but you can make your own alerts if you would like. That's usually not needed at the beginning, but the more confident you feel, the better you can make your own alerts and then we can cover how to do that. But I'll just leave that for another time.