120. Clinical Decision Support Systems with Biomerieux

Jame is joined by Chloe from BioMerieux to discuss clinical decision support systems and their application in Microbiology & Infectious Disease! What are they, how do they work, and what are the potential applications in antimicrobial stewardship? Listen on for details! 

Note: This episode was produced in collaboration with BioMérieux. An honorarium received from them will be donated directly to charitable causes.

Message from BioMérieux: Every effort has been made to ensure the accuracy of information shared during our podcast appearances. The views expressed by our representatives are in good faith and based on the information available at the time. We accept no liability for any errors, omissions, or discrepancies. For the most accurate and up-to-date product details, please refer to our official website or contact us directly

Transcript

Jame: 0:18

Hi everyone. Welcome to the Idiots Podcast, that's infectious disease insight of two specialists. I'm Jane, that's Chloe, and we are going to tell you everything you need to know about infectious disease. Chloe, how are you doing?

Chloe: 0:30

I am good. Thank you. How are you?

Jame: 0:32

I'm fine, I'm fine. So this is another, collaboration, with, biome, on the show. Are paying a, uh, stipend for this episode, and that stipend will be donated to charitable causes. So Chloe, what have you come to talk to us about today?

Chloe: 0:49

I've come to talk to you today about our clinical decision support software that we have for antimicrobial stewardship. So it's called a PSS.

Jame: 0:57

Interesting. Okay. And first, a little bit about yourself. How did you get into this, game?

Chloe: 1:03

So I'm a hospital pharmacist by background, with a clinical interest in critical care. And then I then moved into digital innovation where I mainly focused on implementing and optimizing electronic prescribing, and administration records. So that is looking at clinical decision support softwares, how we can use that as well. So when I, when this job came up at Binary, I was obviously very interested in it.

Jame: 1:25

And so your, your interest is not so much in stewardship, but in the clinical decision support software itself.

Chloe: 1:31

And how we can use these tools to, make clinicians lives better and also make patient care, more safe.

Jame: 1:37

Okay, so patient care is a secondary concern. Is it? It's mostly about the doctors.

Chloe: 1:41

I'm a clinician by background, so, you know, thinking about ourselves first.

Jame: 1:45

No, I, I understand and I, I suppose having had experience in the NHS, you will be all too aware of the clunky MS. DOS based systems, which hang around as a legacy of our. It infrastructure in the NHSI mean, I've previously worked with plenty of, so, um, computer systems, which use or programs which used, uh, some were dos based, some were running inside a un shell. There's a lot of stuff like that out there. I, I think the NHS is moving to, to modernize. You've mentioned something here about, in the print notes about the, Lord Darzi and analog to digital. So, can you explain a little bit more about that?

Chloe: 2:28

Yeah, of course., We have a lot of national drivers that are pushing us towards. More, clinical, systems. So we have a strong push, to untap to use data that is in the NHS at the moment. So most of the data that we have in the NHS is generally underused for clinical care, and that's what Lordi has called for us to leverage this technology and then apply it when we are providing care for our patients.

Jame: 2:52

In a research capacity or in an optimization of the health service

Chloe: 2:56

Optimization of the health service, and also for trust to increase their digital maturity. And that's where clinical decision support systems really come in because they are primary marker of if a trust is digitally mature or.

Jame: 3:09

I mean, the, the NHS must be producing terabytes of data every single day. And that data is not really. Going to full use. I know particularly when we talk about releasing that data to. Companies for, for research, there's a lot of reluctance to, to do that. And I think I would be part of that, group. If I had to put myself in any one camp, I don't really want patient data going two companies for commercial exploitation. But the CDSS. Tools that you're talking about today are, are not really that. It's, uh, decision support software that would, that would be applied within the NHS, assuming that, it's able to run on Windows 10 or Windows eight, uh, now that I think about it. So the CDSS that you're going to talk about, that's from a company called, uh, Luer. Is that right?

Chloe: 4:03

Yeah, so, it's a company called Lumad. So, which is actually, we have three pieces of, softwares all that provide data-driven clinical decision tools. So we have a PSS for stewardship. We have zinc for infection prevention and control. And then there's an oncology module, which is, is pretty separate. And it actually started as a PhD project, um, many years ago in Canada. And then it's since grown and commercialized, and it's now a registered medical device, which is really exciting. And it's now making up part of Mary's stewardship. Offering, which is really exciting. So they have the diagnostic tools, which obviously everyone knows about. And then we have the IT solutions, which is more about giving the data that we get from the labs and giving that to the clinicians to use real

Jame: 4:45

And, and, and that's a relatively recent expansion of bar,

Chloe: 4:49

Yeah, yeah, yeah. It's, it's recently, uh, and it's growing pretty rapidly, because we need the whole journey of the patient through the hospital, not just focused on the lab.

Jame: 4:57

uh, People will have encountered clinical decision support software or CDSS as we're going to call it for the rest of the show, so that I. Don't have to say it all the time,, in an infection control, context. I, I certainly. It's encountered infection control software used in the trust that I've worked in where information has been pulled from the EPR, like, you know, this person is colonized with MRSA or they've had an MSA, colonization, result in the past, et cetera, et cetera. And that's been put onto a dashboard for the infection control team to work from a PSS is slightly different though. So this is a applied CDSS program. What does a PSS stand for again?

Chloe: 5:40

So it's antimicrobial prescription surveillance surveillance system.

Jame: 5:45

Antimicrobial prescription surveillance system. Okay. So Big brother for antibiotics. Okay, fine. We're, we're just gonna call it a PSS for the rest of the show again, because I don't want to repeat it, constantly, so that's fine. So tell me a little bit about, uh, A PSS. What, what is it?

Chloe: 6:05

Yeah, so like I said, it's a clinical decision support, system. So we have, there are two different types of decision support system systems you have. One that's a knowledge base. CDSS, and you have one that's a non knowledge base. So ours is a knowledge base, CDSS, which means that we have an algorithm inside it that we have built. So we don't use any, ai. The algorithm is based on knowledge that has been built by pharmacists and infection doctors, and everything is referenced.

Jame: 6:34

So you didn't feel like a black box guessing and making everything up, was the way to go here when making life, life or death decisions. Okay, fine. Not to nail my colors to the master or anything. So do most CDSS is run like that on a knowledge based system. Can you gimme an example of a non knowledge based one?

Chloe: 6:53

Um, I would say the non knowledge based ones are probably more the ones that are used in like radiology imaging that are more ai, based, where someone has fed it numerous pictures and you've developed a knowledge base and then it's going forward from there. Whereas this is one where we have inputted a knowledge base and we're regularly updating it.

Jame: 7:12

Yes. Okay. So it's non machine learning, but it is algorithmic

Chloe: 7:15

Yes, exactly.

Jame: 7:17

These cdss, do you have any evidence that they're better than what came before them? Clinical gestalt.

Chloe: 7:25

I guess what became before that was, was just no decision support. And I'm not saying that A-C-D-S-S is a new technology. You will have encountered it many times and overridden many alerts in your prescribing system, a very basic level, CDs, ss, so they are, they are everywhere. We do have,

Jame: 7:42

Like warning, this patient has a penicillin allergy. Don't give com. That kind of

Chloe: 7:46

Exactly, so that's a very basic CDS. Um, ours is a medical device, so we're actually providing a specific recommendation of what you should do. And there has been meta-analysis done. On CDs, ss, that are used for antimicrobial stewardship and they've had some pretty promising results. So it has been shown to reduce relative mortality rates, and that's due to inappropriate dosing. They reduce cost, they reduce patient's length of stay, and that's due to IV oral switches. And then they're also led to decrease overall antibiotic usage. By about 23%. And the most common interventions that the CDS is making is usually due to an IVR switch, an immediate discontinuation of treatment and spectrum modification.

Jame: 8:36

What do you mean by spectrum modification?

Chloe: 8:37

So advising and switching from a broad spectrum antibiotic to a medium or an narrow.

Jame: 8:43

Okay, fair enough. Yeah, so. Advising, stopping antimicrobials, advising an oral switch, or advising moving to a narrower spectrum, antibiotic. Would it ever advise moving to a broader spectrum? You know, as I sometimes do.

Chloe: 8:58

Yeah. Yeah. It's, um, those were just the most common interventions that the CDSS was used for in this metaanalysis that.

Jame: 9:07

fine, fine. So specifically about a PSS, do you have, as if I didn't already know, a publication of which you wish to hawk to whatever listeners are listening right now?

Chloe: 9:18

Yes, of course we do. So we did a study, at Sherbrooke Hospital in Canada. So it's a 700 bed teaching hospital. Um, and they implemented a PSS and they found that they got a reduced length of stay by two days, and that was mainly due to oral switches and discontinuation of therapy. They had a reduced consumption by 24%. Um, and that was measured days of therapy, um, per 1000 inpatient days. And then they also had a reduced cost, by 28%. And what's really important here is the

Jame: 9:47

Was that reduced cost just of the antibiotic or the entire length of Save the patient.

Chloe: 9:52

Of the antibiotics.

Jame: 9:54

Okay, fine.

Chloe: 9:54

But the savings that were, found here, outweighed the cost of the intervention. So they outweighed what they'd spent on the A PSS license, the salary of one full-time pharmacist, working there, and one hour of a consultant per day. So that's really exciting. And then all of these, findings have actually been sustained three years post intervention. So like

Jame: 10:18

is a PSS still working, is still in that hospital.

Chloe: 10:21

Yes.

Jame: 10:22

Yes. Okay, fine. Yeah, so it wasn't just a,

Chloe: 10:25

Just a

Jame: 10:25

Hawthorne effect or a first mover, and then it, then it revert to the mean. It's a sustained change in antimicrobial prescribing practice, which is the holy grail of guideline developers and antimicrobial stewards, worldwide.

Chloe: 10:38

exactly.

Jame: 10:39

Well, we'd better tell the listener a little bit about what a PSS actually looks like. We haven't talked really about the software itself, so does this talk to existing. Patient management systems like Epic and and, the big hitters like that, or does it integrate with everything or?

Chloe: 10:56

Yeah, exactly. So we're pulling in, data from any electronic system that we need in your hospital. So the electronic healthcare record, your prescribing, an administration systems, laboratory information system. So we integrate with

Jame: 11:10

Does it only work with ER limbs? Like if, if, if I'm operating on a BD Phoenix, will it refuse to talk until you, until you throw it out of the window?

Chloe: 11:19

No, we're not that mean yet. So it will integrate with anything, yeah, anything that's digital and in a standard format we can tap into. And we're also moving bidirectionally, so we're not just taking the information from them. If we make a change, we make a, intervention note, we're gonna then give that, um, data back to the electronic healthcare record so that there is, we are maintaining the one source of truth for.

Jame: 11:43

Yes. So you what you pull the information in and then you output a, note, an advice note or something onto the EPR, except I suppose if it's still paper notes, in which case you'll have to schlep over to the ward and hand write it.

Chloe: 11:58

If it's paper notes, this probably isn't the,

Jame: 12:01

so what inputs does A PSS take in the The microbiological results as well, presumably? Allergy status, CRP, white count.

Chloe: 12:11

Yeah, any clinical parameter. So all your blood tests, anything that's been sent, we're pulling in. We're pulling in all your micro results, your radiology results. We can have the reports in there as well. Allergy status, all the drugs that you are on, your weight, your height, everything. We then pull that into a timeline visualization so you have a holistic view of a patient just in one glance. And I can see everything just without having to click into lots of different systems.

Jame: 12:38

Uh, and you've shown me to this, this sort of looks like a line graph over time, so times on the x axis, the Y axis has, multiple different, AEs. And you can sort of see the CRP, the white count, what the date that x-rays or cts or imaging is, is done. Is there a little sort of dot there? Do I, I seem to

Chloe: 12:57

yeah, that goes, if I hover over any point on the graph, it can tell me exactly when it was taken and gimme a bit more

Jame: 13:03

Yeah. And what it is,

Chloe: 13:04

what it is.

Jame: 13:05

And sort of color coded too.

Chloe: 13:07

Everything's color coded. Um, it looks, it looks very attractive. Um, if I hover over a micro result, it'll tell me the sensitivities of what was tested as well.

Jame: 13:17

Yeah, maybe we could include a screenshot of suitably anonymized data to put in the, um, in the prep notes or in the, in the show notes for the, uh, for the podcast episode. Um,

Chloe: 13:29

definitely.

Jame: 13:30

And that will get around the fact that podcasting is not a visual medium. But yeah, but the, yeah, the pulling in of data is, is sort of quite impressive. And when you, you showed it to me, I. One last month or the month before, I found myself thinking this would actually be quite good for consult reviews and not just stewardship, stuff. Like it's obvious application is in stewardship. You know, this person on Cipro, why? Maybe I could get'em off it on something else, you know? Or maybe not. But I also thought the data that was coming into a PSS was most, but not all of the information that I would need to do a pretty decent ID consult review, to, so your staph aureus battery means your gram negative battery means that kind of thing. But yeah. And so then once the information comes in, what can you do?

Chloe: 14:18

So then we have this visualization, but that's, that's sort of the side, side part of the product. So we take in all the prescriptions, so all the antimicrobial prescriptions, and then they run through our algorithm, and then we produce an alert if there's a non-optimal prescription. So this could be something like, if this patient's eligible for an oral switch, and for that, we'd be looking at, at things like, how long have they been on the antibiotic, but what is their white cell count? What is their temperature? Where is the infection? Are they on vasopressors? So it's, it's more intelligent and alert. It's not just have they been on antibiotics for 48 hours.

Jame: 14:53

So more of an evidence-based, reason to, yeah. I suppose the question I'd ask at this point is, is the algorithm, hyper cautious, or. Brave. Is it a brave little boy?

Chloe: 15:06

The algorithm is whatever you want it to be in your hospital,

Jame: 15:10

does that mean that I could modify it to oral switch everything with a pulse and some things without

Chloe: 15:14

Yeah, if you, if that's what you said you wanted, we have our base level of knowledge base and alert, but we would then work with the stewardship team in the hospital to say, actually this is the criteria we want for oral switch, and

Jame: 15:27

Mm. Yeah.

Chloe: 15:29

And, and that's on your, your watch.

Jame: 15:32

So, and remind me, have you rolled this out in the UK yet?

Chloe: 15:36

Not yet. We launched just a few weeks ago at, um, BSAC conference. So we are pretty widespread in Canada. We're in France and in Belgium and trying to get into other European, hospitals. But we are just launching now in the uk so looking for a collaboration site.

Jame: 15:53

Yes. Okay. And so when you are just taking, digging into oral switching a little bit, for a moment, you know that there's recently been a UK HSA sort of oral switching guideline document, which was produced by Delphi Consensus as, as far as I can tell, and includes some stuff that is evidence-based and some stuff which very much is not. Would your oral switching algorithm be set sort of according to that? Would you tailor it for the UK is what I'm asking.

Chloe: 16:21

Yeah, so our knowledge base works in levels. We have a, a global database, then we have a uk and then we go down to trust and then put a hospital. So it's tailored for the uk. But then we would then review that because each hospital has their own local guidelines as well.

Jame: 16:39

And our own foibles.

Chloe: 16:40

exactly, exactly. So we know that the NHS is not completely standardized.

Jame: 16:44

Oh, you're aware of that. Are you okay, good. And so then in terms of outputs that you can then push back into the patient CPR.

Chloe: 16:54

Yeah, so I've been flagged my patient. I need to go review. I would then review it and then I can see if I agree with the alert. If I agree with the alert, I would press accept and I can also, refer it to a more senior person as well. So put it into a separate work workflow. If I press accept it, then pre-fills like an intervention that has a summary of the alert. I can auto fill different clinical parameters and then I press. Okay. But this is my recommendation. And then that is what would go back into the EPR to document what my advice is, what, or something that I've,

Jame: 17:30

Okay, fine. And then when you say, escalate to senior, does that kick it up to someone like me, like an ID consultant, a microbiologist, that kind of thing?

Chloe: 17:39

Exactly, and you would just have one workflow that you could look through and there'd only be like five patients in there instead of looking at loads. And it would only be people that already a stewardship, pharmacist or clinician has looked at. So you would only be seeing a few patients. So it's all about prioritizing and identifying people who we actually need to see, and getting rid of, the ones that we're just not gonna make a difference on for antibiotic.

Jame: 18:03

So, yeah, and, but it's also another, possible sort of source of patients for your complex patient MDT, you might want to refer some of the more difficult people to oat. And you know, this would be a way of highlighting those people that could benefit from, senior ID input. So, yeah. Okay, good. You've answered this next question, I think, which is where is it being used? You've mentioned Canada and some parts of Europe, anywhere else. Have into the states yet, have you moved to Japan? Have you.

Chloe: 18:32

not yet. Not yet. They'll be pending. But the, Europe and the UK were identified as the first places we wanted to launch. Mainly'cause like the maturity of the market, and like how there already is a lot of electronic systems in our hospitals and obviously we need electronic data for this kind of system to work.

Jame: 18:50

Yeah. Although as previously stated, electronic, abilities vary by, by trust, and sometimes from hospitals within the same trust. Yeah. This, next and, semi-final question is, is somewhat leading. Do you think it would work in the NHS person who is trying to sell everybody a clinical decision software tool?

Chloe: 19:11

Uh, yeah, I think it would work in the NHS. Obviously, I mean, I've worked in the NHS, for quite a lot of years, so I understand like fully the pressures that are faced. So stewardship teams are really varied in the NHS. We know that on average a hospital of about 700 beds has one. A clinician or pharmacist who is carrying out stewardship activities for the whole hospital. And so that's, it's, it's an unmanageable workload. And often they're using really manual processes to try and find their patients. So they're like going to ward rounds and just asking who's on antibiotics and things like that. And they're spending a lot of time finding patients. So a system like this is really needed. Also we've had the NHS, England Digital Vision for antimicrobial stewardship report that has been released. And so our system really does fit that. So they've basically said that we need to be able to identify and prioritize patients. We need to have timely and relevant information given to our clinicians, and it needs to be, at the right time in the right place. We need to have decision support tools that are giving alerts that are targeted. Know, no alert fatigue. So Ined, 91% of our alerts are accepted by clinicians.

Jame: 20:27

Oh, that's, that's a very interesting, let's just focus on that for a sec.'cause I am, I am reminded, uh, earlier on we mentioned a allergy status flagging up. If you try and, prescribe, say, ceftriaxone for a penicillin allergy and one of my pet hates and one thing. Where I think actually CDSS has gone really wrong is the enforcement of dogmas for which there's no evidence. So we've listeners will be sick to the back teeth of me talking about the cross reactivity of penicillins and kelo sporin and the the lack of any evidence for anything beyond the first generation Keli SSPORs causing a cross reaction with penicillin and the historical context. In fact, we're about to cover it in our upcoming. Episode, but when I try and prescribe, ceftriaxone or cef Taine or even as Triam, in some systems for people with a penicillin algae, it will dutifully pop up and reminds me that this patient is allergic to beta lactams. And that's completely infuriating. And I can. Help educate, the juniors and the the SHOs and the nurses and the pharmacists and the hospital at home team, and the Ambulator care team. And the pharmacists as much as I like, which I did in my own trust, but. The people are gonna have a lot of difficulty actually putting appropriate allergy based, prescribing into practice. If every time they try and do it, the EPR pops up little popup that which is in red, saying You're about to kill the patient. Are you sure you want to do this? And they have to tick a box saying, yes, I am sure I'm going to, take my registration in my own hands, you know, ide, et cetera, et cetera. So I think that's an example of CDSS really enshrining something that we don't really want to enshrine, and once that stuff goes in, it is really difficult to pull it out.

Chloe: 22:21

Yeah. And that's an example of a, yeah, it's an example of a, a really basic level CDSS then that's the kind of CDSS that, uh, EPR takes, whereas you need one that's agile and that's gonna fit your current practice in the hospital.

Jame: 22:36

Because I suppose your target audience is not the general, doctor population. It's specifically antimicrobial stewardship teams that will be infection doctors and it will be, antimicrobial pharmacists and nurses and stewardship officers and, and sort of people like that. So it's a smaller audience, let's say.

Chloe: 22:55

Exactly. Exactly. And they, they have a different requirement. And we can put different sort of intelligent alerts into the system instead of just every basic alert that is under the sun that you could produce about medicine.

Jame: 23:08

Yes, absolutely. Yeah.

Chloe: 23:11

Um, something else that the NHS has said that we need that's actually really important is, um, data visualizations that are live in the systems that clinicians can use. So, at the moment. Maybe if you wanted to run a report, you'd have to contact an analyst who'd then run a SQL report for you. And it's, it's a really lengthy process. Whereas actually for something like antibiotic prescribing, you really need live data and you need to be able to visualize things and surveil prescriptions and resistance rates and things like that. So that's something that's really important that CDSS can help you with, as well and producing live antibiograms. People can actually use and interpret, every day.

Jame: 23:52

Yeah. Even though they're from the Phoenix.

Chloe: 23:56

Even. Even if they're from the Phoenix.

Jame: 23:58

Well, you, you heard it here first. They're all friends over there at, uh, biome and bd.

Chloe: 24:04

Exactly.

Jame: 24:05

Chloe, thanks very much for coming on the show. Uh, is there anything else you want to say about a PSS?

Chloe: 24:10

No, that's great. Thank you. Thank you for having me.

Jame: 24:13

that's fine. No discount code for the first 10 people who sign up.

Chloe: 24:16

Well, you can, you can email me and we can try.

Jame: 24:19

Okay, well, we'll see how it goes. Uh, Chloe, thanks very much for coming on the show.