[00:00:00]

[00:00:07] Erin: Hello and welcome to breakpoints, the Society of Infectious Diseases Pharmacist Podcast. My name is Erin McCreary and I'm the Director of Infectious Diseases Improvement and Clinical Research Innovation at UPMC.

[00:00:18] Angela: And hello and welcome to Communicable the podcast, brought to you by CMI communications estimates, open Access Journal, covering infectious diseases and clinical microbiology.

[00:00:27] Angela: My name is Angela Huttner and I am the editor in chief of CMI Comms and an infectious disease physician at the Geneva University Hospital in Switzerland. Today we are bringing you the second collaboration podcast between communicable and breakpoints. This time we'll tackle the recently published BALANCE trial whose lead investigators were thrilled to have with us today.

It is my pleasure to introduce Nick Daneman, professor of Medicine at the University of Toronto and infectious disease physician at Sunnybrook Health Sciences Center, Toronto. Nick has a long history of research into the use of antibiotics and their delivery, specifically maximizing their benefits while minimizing their harms.

[00:01:07] Nick: Thank you so much for this invitation and this opportunity to talk about BALANCE.

[00:01:10] Erin: And then next we have Rob Fowler, also a professor of medicine at the University of Toronto, a critical care physician for the Department of Critical Care Medicine at Sunnybrook Health Sciences Center.

[00:01:21] Erin: Rob, welcome to Breakpoints and Communicable.

[00:01:23] Rob: Hi

[00:01:23] Rob: both. Thanks so much.

[00:01:25] Angela: So as our listeners know, at Communicable, we start these episodes with a get to Know You question so here's the question. We're coming to end of summer and some of us might have gone to some cool concerts.

[00:01:37] Angela: What is a concert you would want to attend or maybe already did attend? Nick, you first,

[00:01:43] Nick: I guess for me that's, Billie Eilish. My kids really got me hooked on her music over the last few years to the point where she was my number one artist on Spotify 2024 And the missed opportunity for concerts pretty fresh in my mind.

[00:01:55] Nick: So we came back from this glorious family holiday in Japan, over the last couple of weeks. And on that very last full day of the trip, we were walking in Shibuya, Tokyo. And my daughter Chloe looked up at me with these like really big eyes and said, dad,, Billie Eilish is playing here tonight in Tokyo.

[00:02:10] Nick: And there was that moment of thought, but I dismissed it almost instantly because it would be impossible to get those tickets. And if we got them, they would've cost a fortune. But, just thinking back on it, that would be just, an incredible ending to an incredible trip.

[00:02:21] Erin: Wow. You should

[00:02:21] Angela: have gone.

[00:02:23] Angela: Rob, you're up next.

[00:02:25] Rob: Yeah, I like Billie Eilish too. but last year I wasn't able to get to a Bruce Springsteen concert in Toronto, that we had tickets for. but my wife was able to enjoy with a friend of hers, so I'm hoping to catch up with him on tour somewhere in the future.

[00:02:39] Angela: Very cool. Yeah, that Would be on my list, definitely Erin.

[00:02:43] Erin: Okay. This is acute in my brain. and much to anyone who listens to breakpoints or communicable and has heard me talk, I'm actually not going to talk about Taylor Swift. I just saw Gracie Abrams at Red Rocks out in Colorado. But for those of you who may not know, red Rocks is an amphitheater that's truly built into the mountains. And so you have to like, hike up and , there's 70 something rows and each one goes up quite a bit of elevation and it's just carved into the mountains right on the west side of Denver.

[00:03:11] Erin: And it is stunning and it is everything you'd ever. Hope from an acoustic standpoint from music. ' Last week I took my stepdaughter to see Gracie Abrams and it was truly one of those insane life moments where the concert was on a Tuesday night. So we flew out Tuesday morning, landed in Colorado at,, 10:00 AM got ready for the concert, went to the concert, and then flew home Wednesday.

But it was worth every minute of insanity. It was so fantastic. it was awesome.

[00:03:37] Angela: Yeah. Well, my concert experience already happened and I, I really wrecked my brains and I could not think of a future concert that I would care more about than the one I already went to last year. I went to Taylor Swift in Zurich and it was so cool, because I didn't expect ever to be able to go because I knew like it was impossible to get tickets and everything.

[00:04:00] Angela: But two days before the concert in Zurich, we just looked online and there were tickets available for like my daughters and me , and we all just did a road trip and went to this concert like two days later and i've been to some concerts in my time and I'd never really experienced a concert like that where somebody worked so hard to make people happy.

[00:04:21] Erin: So anyone listening to this right now, the moral of the story is pause, buy Yeah.

[00:04:26] Erin: The tickets. Yeah. Till the event you've been thinking about. Just do it. Just do it. You only live once, There's been many a thing I haven't done, But things we have done are very cool clinical trials. So let's pivot now and get back to our topic. So today we are talking about the BALANCE trial. This trial was presented as a late breaker at ID week in 2024, and then published on November 20th, 2024 in the New England Journal of Medicine.

we cover actually Nick's ID week presentation on the Break Points podcast in our December 20th, 2024 episode, which is titled ID Week 2024, late Breakers Recap. So if you want the quick summary, please feel free to listen to that. But for now, be warned, we're going on an epic journey and we're going to deep dive into all things that went into this trial.

[00:05:09] Erin: And so Nick and Rob, we really want to hear about this long arc that brought you to do this trial. It took 10 plus years to complete. And Nick, I think let's start with you because you explained in your ID week presentation that you guys did a lot of retrospective studies and even national surveys across Canada, Australia.

[00:05:28] Erin: so,, you guys are now doing all these very complex adaptive randomized trials, but you started somewhere.

[00:05:33] Erin: So can you give us an idea how you got from there to here?

Yeah. It's pretty hard to believe that it all started as just a hallway conversation between me and Rob debating a patient that he was following on the critical care unit that our stewardship team had recommended upping the duration of treatment instead of descreasing the duration of treatment 14 days.

[00:05:51] Nick: 'cause they had a positive blood culture and that was just the standard of care at the time. Everybody with bacterium got 14 days of treatment. But as we're having the conversation, we felt really uncomfortable with it. there's really no data to support this, it feels like expert opinion. So why don't we try to answer this question.

[00:06:07] Nick: Right from the beginning, though, we know that It means a very large randomized trial, but we don't wanna jump, right to that. So we wanted to lay the preparatory work to see that there was practice variation, equipoise, a need for a trial, and how to best design that trial. And so our research program started, I think, as pretty much every research program shared with a systematic review.

We looked for randomized trials to bring short versus long duration treatment in bacteremia, pretty much nonexistent at the time. so we saw that evidence gap. We did find that it probably was okay to pursue because there were lots of trials of short versus long duration in other syndromes that can be complicated by bloodstream infection.

[00:06:40] Nick: And then the next step was to ask, in the absence of evidence, what are people doing? So, like you said, clinical practice surveys. I find them one of the most fun and informative studies to do. They don't even require any money.

And so we, asked ID doctors, ICU doctors across Canada. Later on in other countries as well. How long are you treating patients with blood stream infection? And we found 14 days was the most common duration for almost all pathogens and syndromes, but there was huge practice variation, lots of people getting seven in 10 days and clear stated, explicit willingness.

[00:07:10] Nick: I would enroll my patients in a trial if you were to do one. That's when we engaged the Canadian Critical Care Trials group and they bought into this research question and 11 people put up their hands saying, our ICUs across the country will join a retrospective observational study that showed the same practice variation and even longer treatment durations of what we were seeing in surveys.

[00:07:29] Nick: It wasn't just like a mode of 14 days. The median treatment duration people were getting in real life was 14 days. the next step from there, I would also highly encourage anyone that's gonna do a large trial to take that step is a feasibility pilot, not a mechanistic pilot to try to show glimmers that a intervention worked in an underpowered way, but to enroll people and show that it's feasible to do that you can get people to recruit into a study. So we started a pilot in ICUs, starting at Sunnybrook, our hospital, and then rolling out to some others and show that it was really quite feasible and that became the vanguard for our main trial.

[00:08:01] Erin: I think it's so interesting that you started in the ICUs and maybe not, maybe people are like, obviously that's the place to start.

[00:08:07] Erin: But in a way it's like those are sicker patients, you know, we see with drug trials, like that's harder patients to enroll. But is it because Rob, you were working in the ICU and those are where patients with bacteremia are initially presenting and it would've been harder to find these patients on the ward.

[00:08:21] Erin: You're obviously not stable in the first 24, 48 hours if you have bacteremia. So kind of describe to us why you went to the ICUs first.

[00:08:28] Rob: yeah, that's a great question. I mean, that gets to, I guess a very practical reason that Nick alluded to in that in Canada for, over 30 years we've had this terrific group called the Canadian Critical Care Trials Group, which does a lot of foundational research that oftentimes culminates in a clinical trial.

and so there was a network of, very interested investigators, research coordinators, methodologists and really already mentoring and, helping, lots of groups answer questions like this. it became obvious that we would, , go to a group that I knew pretty well and Nick had engaged with to try to get combination of, mentorship and help.

[00:09:04] Rob: And then also, help with operationalizing, the studies that lead to a trial like this, the sorts of studies that Nick alluded to. , They're, foundational ones for a trial. And oftentimes, if we don't do some of this work upfront, then something pops up when we're trying to do a trial that we say, oh, I wish I had known this earlier.

[00:09:23] Rob: I wish I had discovered this earlier. They can really throw a huge wrench into, a lot of work for a lot of people. And so I think it's super helpful to be able to do this work ahead of time. But it's also, I would say, super responsible to try to do this work ahead of time so that, at the point when you're engaging patients in an experiment, , one, treatment or another, that you've done as much as you can do upfront to make sure that experiment is as, appropriate and efficient and will be as helpful as possible to future, folks.

[00:09:51] Erin: That's fantastic. I love that.

[00:09:53] Angela: so can we, just step back and run down sort of the big overview of the trial for those who have [00:10:00] not actually read it. There may not be that many out there who haven't, but just in case.

[00:10:04] Rob: Yeah, sure. so, the question is shorter versus longer duration antibiotics. That's a pretty ubiquitous one in treating patients with infection.

[00:10:11] Rob: as Nick mentioned, our original question was focused on antibiotic durations and was focused on , a critically ill patient. common syndromes. There are ventilator associated ammonia and bloodstream infection because of catheters and urinary, tract infections that lead to bloodstream infections, et cetera.

[00:10:27] Rob: There had been this really informative trial examining, one, essentially one versus two weeks of treatment for ventilator associated ammonia. Back in about 2003. I remember I had done an observational study, looking at treatments for ventilator associated pneumonia when I was in training and helped with some practice guidelines, but was impressed at how difficult it was, to know if someone had pneumonia or not.

and so Nick and I remember discussing that when we were thinking about syndromes that would be important to study in critically ill patients and eventually in the whole hospital. we performed our initial pilot after the foundational work among hospitalized patients who were admitted to the ICU with a blood culture that was positive for a pathogenic bacteria and randomizing eligible patients to seven or 14 days of appropriate antibiotics, so appropriate to their culture and sensitivity.

Our pilot trial, the outcome was feasibility, so recruitment rate and protocol adherence. And while we measured, in parallel the outcomes that we'd want to use in a larger trial. And that pilot was pretty successful in terms of the feasibility, and it was certainly instrumental in starting, the main trial with the same intervention and comparator, with a primary outcome of 90 D mortality.

[00:11:36] Rob: And we're seeking non-inferiority of seven versus 14 days. And that distinction, as opposed to superiority, because our starting point was that probably, , a week is, , maybe just as good as two weeks. A, good question to study within a non-inferiority margin of about 4% - - and practically this meant we'd need about 3,600 patients to, gain enough confidence that the answer was, true. we subsequently expanded our population to include all hospitalized patients, but not before. We conducted a separate pilot trial among, non ICU patients and found that the trial procedures were still feasible at the hospital level.

interesting to us, and we knew this a little bit, but we didn't know it, in absolute numbers. There were probably three to four times as many patients with bacteremia outside the ICU was in the ICU. And so that was a big practical help with eligibility. once we expanded to the rest of the hospital.

[00:12:28] Nick: By the time we made that hospital wide expansion, our inclusion criteria become very simply, you're hospitalized at the time your blood culture dings positive in the lab. And then, the exclusion criteria, we tried to keep them as tear down as possible, only excluding the most of your immunocompromised, which was neutropenia and transplant, only really exuding one common bacterial pathogen, staph aureus, common contaminants, especially would be relevant for them.

[00:12:51] Nick: And then the syndromes where we know we need longer treatment, which when we set into this, I would've thought would be a longer list. But when it comes down to it, it's endocarditis, bone and joint undrained abscess or un remove prosthetic infection so a pretty, pretty short list.

I did notice that you

[00:13:06] Angela: excluded people who simply had, prosthetic valves, like even without necessarily suspicion of endocarditis.

[00:13:15] Angela: But you said, no, we're not gonna take any chances on that?

[00:13:18] Nick: Our feasibility pilot

[00:13:19] Nick: was, so feasible that we didn't make many adjustments at the time of switching over, from Vanguard to main trial. But that was one tweak to the exclusion criteria. We added, prosthetic valve. 'cause we'd run into the concern around some of the organisms like enterococcus, where people feel, less comfortable. It wasn't really relevant to the gram negatives and so it would've been Right. Easier to allow it in. So it's probably one of the reasons why I didn't even mention it when I was running through our study criteria 'cause I don't think of it as one of the main most important ones.

[00:13:45] Angela: Yeah, yeah.

just thinking you don't wanna tie your hands when you're recruiting, right. And now with these aging populations, there are so many prosthetic valves out there. So it would be a shame, I guess to bar yourselves from those patients if they just have an e coli bacteremia, for example.

[00:14:01] Angela: Yeah. But clearly you guys did fine. You recruited tons of patients, but I guess it did take you a lot of years.

[00:14:07] Nick: Yeah.

[00:14:09] Angela: In a lot of hospitals

[00:14:10] Nick: there was a COVID pandemic in between, that slowed things down quite a bit. And some of those early years the ramp up, that vanguard phase gets included in the total duration.

[00:14:19] Nick: the actual trial really only launches a main trial around 2016 or so.

So let's frame that for the audience. So you guys did end up randomizing 3,631 patients from 74 hospital sites in seven countries over 10 years. That's a lot of work. That's pretty incredible.

[00:14:38] Erin: I wanna just real quick go back to the fact that this was incredibly easy and very broad inclusion criteria, and you said your feasibility was so feasible and you intentionally made that very broad.

From collaborating with folks in Canada and Australia you guys are really good at doing pragmatic low barrier to entry trials to answer these very relevant questions. I would say in the United States, we're not so good at it. our trials tend to exclude a lot of people.

[00:15:04] Erin: , It gets to the point where then you're like. This trial is great, but finding this, , diamond in the rough of this patient, that exactly applies to the trial population, it's challenging. And so the trial's less externally valid, it feels intuitive, but can you just state it for our listeners, why your team and collaborative said, you know what, we're gonna keep this very easy and very broad and kind of take this almost all comer trial outside of staph and candida.

That's a great question. we did start from a point of wanting, answers to be as relevant to as broad a patient population as possible. And I think we're very actively thinking about every exclusion criteria. There are some that are in the BALANCE trial that I think if we were going to do it again, we would say, ah, , we take even more of those exclusions out because we're left thinking that, we don't have as much information as we should to generalize even the results that we do have to some of the other important patient populations.

[00:15:57] Rob: And so I think if anything we'd want to be more pragmatic, have fewer restrictions, in our subsequent work. when I had just started as a clinician at, Nick and my hospital in Toronto, one project that I had done with a resident at the time looked at exclusion criteria, eligibility criteria really of.

[00:16:15] Rob: high impact, trials, in the medical literature at the time. And, it was so interesting to find, I would say, a usual laundry list of exclusions that when we went through and tried to justify in the context of the intervention and the patient population, we were left scratching our head as to why wouldn't this be something to learn about for pregnant women?

[00:16:34] Rob: Or why would it be that, a 14-year-old may not benefit from knowing the results? And that certainly informed a lot of, decisions that I've, been thinking about over the years.

[00:16:42] Erin: I love that, that is very helpful too. So tell us who ended up being in this trial? What were your demographics like in the population enrolled?

So, among those eventually 3,608 patients that you mentioned, there were slightly more males than females, and that's not uncommon for a large, trial that enrolls a large proportion of patients in the ICU. So about 55% were male, median age of 70 years. comorbid conditions that you might expect.

[00:17:05] Rob: So about a third with diabetes, a fifth with cancer, about 10% with some degree of renal insufficiency or on dialysis, heart failure and COPD. one measure we collected the frailty score found that patients were commonly with some degree of vulnerability, sort of in brackets, and they didn't need daily help for things, but, things that might just slow them up a bit.

so a hospital, patient population that we might imagine seeing on our wards in ICU. about three quarters of the patients had developed their bacteremia while in the community before presenting to the hospital. And, of the common syndromes, urinary, sort of syndromes leading to bacteremia most prevalent at about 40%, 20% intraabdominal, 13%, with a lung source and seven catheter related.

[00:17:48] Rob: And accordingly to, the syndromes, e coli was the most responsible causative organism in just over 40% of infections with other gram negatives and positives, as you might expect in any bacteremia observational study, with the exception of, staph aureus, which was an exclusion criteria.

And 55% of your patients were in the ICU.

21% were intubated and 30% were on vasopressors, I think at the time of enrollment. I don't know about you Angela, but for me, for my institutions, this is what was really meaningful for us because I think at this point people are comfortable treating stable patients with a little less if possible.

[00:18:25] Erin: I think in general, the world understands we wanna limit antibiotic exposures whenever we can. But admittedly, even those of us who have adopted some shorter course paradigms for any disease syndrome, often our guidance is okay, but they have to be totally stable and so this really expands to quite a sick population and then quite a patient population of advanced age. I think your median age was 70 years old, so that's what really stood out to us from this trial and the patients who enrolled.

can you guys actually tell us on what day of the bacteremia did patients get randomized? Because I, didn't actually see that in the main text.

[00:18:59] Nick: Yeah. we encourage people to approach as early as possible and randomize as early as possible, but they were allowed to randomize anywhere up to the seventh day about equity.

[00:19:08] Nick: Oh, wow. Treatment.

[00:19:09] Angela: Okay. And what was your median.

it was gonna be around four to five days

[00:19:13] Erin: Okay. So kind of after that acute illness. Yeah.

Yeah. Because I, I could imagine that it would be hard to get buy-in from your physicians,, to say, well, you guys have to randomize this patient who's in the ICU on day one or day two.

don't know how he's gonna do. but randomize 'em now to seven days or 14 days, make that decision now. Whereas clinically, we, don't really make that decision right off. Right. We're thinking about the choice of antibiotic. We wanna give the way we wanna administer it. We're not necessarily thinking about duration at that point because the patient's in the acute phase.

I just wanted to encourage community approaches and early randomization. 'cause another decision, for design could have been, and there were a couple team members that had recommended this would be to randomize only on day seven. And then you're either stopping today or you're giving another week of treatment, you're gonna get less protocol.

[00:19:59] Nick: Non-adherence. Then [00:20:00] I'm sure we'll come back to talk about non-adherence, but you're changing the question and I think that really does impact generalizability. 'cause you're really focusing only on the patients that are really well by day seven. We wanted this to be a treatment strategy trial from the beginning.

Would I be uncomfortable treating people with buzz infection with seven as opposed to traditional 14 days of treatment? I

mean, yeah, in a way I think you did kind of a massive sensitivity analysis. You basically said, okay, we are gonna be blind to how this patient's actually doing. At the time when we need to make a decision as to stopping or continuing, we're going to, preempt that and we're going to already determine the duration no matter,, what's gonna happen with this patient in the next 72 hours.

[00:20:40] Angela: The acute, really early onset where the bacterial loads are still really high, and so you did it upstream and it still showed non-inferiority of seven days. So it's like really rock solid. That's how I take it.

and you make a good point there. I guess we didn't say explicitly there was prolonged allocation concealment, so that if you randomize on day one or two, or three or four, et cetera.

[00:21:02] Nick: The team wouldn't get to know what group they were in until they seven.

[00:21:06] Angela: That's fantastic. 'cause you don't want the team changing their behavior when they know, you don't want them getting a full body CT scan or something the night before stopping the antibiotic.

[00:21:16] Angela: Yeah. Make sure there are no abscesses.

Angela, what you just said is actually so important. It sounds like Nick and Rob, you appreciate this amongst your colleagues and peers is that this is all about changing behavior and, science's data is important, but it really comes down to behavioral change and management and there's a lot of mind over matter here. So I think if you had approached people on day seven and said, you need to stop today, you would've had a lot less spot in. People are not mentally prepared for that, especially back then when they were enrolling before it became kind of something we accept now, especially in the gram negative space, you're giving them a couple of days to mentally brace for the fact that They could stop at day seven and they've accepted that they're ready for that change. And,, Angela, like you just said, they're not like, well. I might stop, but let me do all this extra imaging and just be , super sure. Right? We wanna first do no harm, but often, extending antibiotics doesn't necessarily not do harm.

[00:22:00] Erin: And so I love that, that it was the intention of seven days is acceptable and then that leads to more overall buy-in.

[00:22:07] Rob: As you said this notion of changing behavior is so challenging.

[00:22:11] Rob: And even in the context of the trial, I think we could appreciate. qualitatively that the first enrollment, for any particular clinician, probably was associated with a bit of nail biting, on the treating physician's behalf when it is potentially changing, a treatment strategy that we've become much more accustomed to longer, durations.

[00:22:31] Rob: And, I think as, people went through the trial, and we see this at individual hospitals, got a little bit more comfort, at randomizing early. So, after we have, an organism identified and, it became much easier as, people got more comfortable with the concept. And I think that sort of mirrors probably outside the trial as well, but it takes some time for people to warm up, as you say, maybe randomizing a few days ahead of the stopping point was helpful in that way.

[00:22:57] Rob: Yeah.

[00:22:57] Angela: Interesting. Yeah, I think we both have so many questions and I'm gonna try to keep them ordered. try to follow a, certain line. so we are still officially at screening and recruitment, in terms of, actual inclusion. If we take the consort flowchart, you did screen over 36,000 patients, , congratulations just on that.

[00:23:18] Angela: That is a huge feat. That's really a lot of work. And, bravo to your teams. You screen 36,000, for eligibility and. We know that ultimately 3,600 were included, but 13,000 would've made it in, 10,000 of those 13,000 didn't get in because they either declined to participate or their physician declined participation.

[00:23:41] Angela: Speaking of, , pre-standing biases, so in the end, only 10% of all patients screened made it into the study. How much do you think the non-inclusion of those specific 10,000 that would've gotten in , based on their medical profile, how much could that have changed the results?

[00:23:59] Angela: how representative are those 10,000 that didn't get in because they declined or because their physicians declined? Yeah.

[00:24:05] Rob: It's such a important observation. I would say trials often have this very humbling flow diagram where, the top number to those that are enrolled and followed up and depending on the trial, but it's frequently in this sort of 10% range.

even in this case, despite trying to make the eligibility criteria pretty broad, and minimize exclusions, of those 13,000 eligible patients, really, about half the clinician of the patient or the SDM declined to participate. I don't know that we can say this to be true yet, but.

[00:24:37] Rob: Suspicion is that number diminished over time as people became, more comfortable with the trial. Some patients had no one to provide consent, but, that's a minority and very few eventually, once they got into the trial, withdrew. And so that's good. But the, generalizability question amongst a lot of patients that are not in the trial to begin with would, probably, as you've highlighted, track along lines of, you know, the very sickest of patients.

[00:25:04] Rob: Sometimes people would have, some concerns about, patients that,, were very likely to pass away in their minds, et cetera. Or on the flip side, a patient that's super well and they think, oh yes, they were uroseptic and got bacteremic, but I think I'm gonna send them home tomorrow.

[00:25:20] Rob: Or I think I'm gonna send them home in a day. Some of those patients, are not gonna get in and for sure that's, important. Having said that, looking at the eventual patients that did get in, Erin, you highlighted, there were a lot of sick patients in the trial. And so even though we had a big drop off from eligible to those included, I think the spectrum of, demographics and characteristics of the patients eventually enrolled help us feel a little bit less bad about, that typical drop off in the consort diagram.

[00:25:48] Rob: So I think we still got a broad spectrum and a lot of patients that were really sick, but you're right. that typical, everyone you screened, everyone you included is often a pretty, pretty humbling figure.

[00:25:57] Erin: Yeah, but you know what too, I was thinking about this. 'cause now it's like, yeah, seven days, right?

[00:26:01] Erin: We give seven days for pneumonia, blah, blah, blah. But. 10 years ago, you're asking people to cut the treatment duration in half. . That's a big ask, I think. And so it's, amazing that so many people did to flip it.

[00:26:12] Nick: Even with those refusals, when you look at the patients in BALANCE, they look almost exactly like patients in a bloodstream infection registry or observational cohort, right?

[00:26:20] Nick: You have median age 70, slight male predominance, almost the identical frequencies of comorbidities that I would see when I do like a population-wide study here in Ontario. And then if you look at that top pathogen list, it's almost in the exact order. From the surveillance, networks, the only one missing is staph aureus up at number two after e coli 'cause we excluded it.

[00:26:40] Nick: So I feel pretty good about the people that got in, even though we'll never know for certain about those folks that didn't, decide to take part. Participating in SNAP, I'm also reassured by the fact that another very large blood infection trial sees almost the exact same 30% consent rate among eligible patients.

[00:26:57] Nick: You're just not gonna get everybody agreeing to the trial when they're super sick.

[00:27:00] Nick: regardless of what the questions are.

[00:27:02] Erin: So about 10% was excluded for staph aureus, which would put that you had, about 40, 45% e coli, 15% kleb. So urinary source is what we expect.

[00:27:11] Erin: Abdominal, staph would've been your number three here. and then enterococcus as your fourth 7% had enterococcus. But just to round out our conversation about the consort, something that really stood out to me, probably because I do a lot of quality work was that, 10,000 plus patients were excluded because of a contaminant, which when you do the math, 29% of patients had contamination, which is why they were excluded.

[00:27:32] Erin: That is a crazy amount of contamination. Right? just like as a side note, there's all these pearls we can get from these trials that teach us about practice as a whole. Other hypothesis generating things and like, wow, we have a long way to go in blood culture contamination, But to move on BALANCE was essentially randomized one-to-one. you guys stratified randomization by ICU, admission and hospital site.

[00:27:52] Erin: You talked about how it was a prolonged allocation concealment to day seven, open label, no placebo. And we ended up with these 3,600 patients. So let's talk about the pathogens because gram-negative, , we feel good about, but you guys did, you had 7% enterococcus. And then when you add up all the species of streptococcus, you're at about 10% having streptococcus too.

[00:28:12] Erin: So there's a lot. Of gram positives, did you have a pretrial probability that you would be so successful in enrolling the gram positive patients?

Not sure about a pretrial probability, but we felt optimistic and I think it was one of the value, add of all that preparatory work to be able to see that people had equipoise, and practice variation in the treatment of those organisms.

[00:28:33] Nick: So it helped in terms of us designing the trial. It also helped with site startup visits as we were able to present that data and say, look, you may always treat this for 14 days, but look at your colleagues across ICU and ID and internal medicine. They're giving seven, 10, or 14 days, hugely variable treatment.

[00:28:50] Nick: And even the ones that are giving long treatment are saying explicitly, I think I'm giving too long. I need some evidence to shorten my treatment. So, at least the preparatory work seemed to show that we. Should be able to enroll these patients.

[00:29:01] Rob: this is a bit of a tangent now, to the question, but Nick's comment, around, the importance of doing surveys and observational work.

[00:29:07] Rob: There's a number of times when we presented the idea to different groups and were met with, , a very reasonable, comment that, oh, we never do that. or we would never really treat that long anymore. in Canada at least, when we had those conversations, we asked people, what do you think you do?

[00:29:23] Rob: And state, perceived practice and then you go and measure and you realize, oh my goodness, that's not all we really do. so it was really informative to do some of that comparative work about what we think we do and then what we actually do. For lots of,, reasons, the amount of antibiotics administered don't end up.

[00:29:37] Rob: Mirroring what we might hope or what we might think we do. So it was super important.

[00:29:41] Angela: your trial, because it spanned so many years, from the beginning to the end phases, did you notice a difference in, culture really in the very beginning?

[00:29:52] Angela: Was it harder to get people to accept seven days, versus the end? Because by the end of your trial, two other [00:30:00] randomized trials were already published when you started those two trials were just taking shape. One was of course Dafna Yahav with Mical Paul in Israel and in Italy.

[00:30:10] Angela: And the second one was our Swiss trial, which I wrote the protocol not knowing you guys were already, on your way. 10 years later we had Daphna's results. We had the Swiss results. did it almost flip towards the end where people like, oh, I don't really wanna give 14 days. why would you have us do that now?

[00:30:29] Nick: I would say there was a huge Gulf ship, promoted by your amazing study and Dafna's amazing study.

[00:30:33] Nick: Where the thought seven days then was what many people would see becoming the standard of care. So, we could still have people be willing to enroll into our trial and have equipoise because we'd be able to replicate the results. but with the added sample size, increase the precision of the cumulative findings.

so hopefully there weren't too many people not enrolled at the participating sites 'cause of the existing data. I think that those studies really helped, get people wrap their heads around the idea of, the value of shortening treatment and the safety of it.

[00:31:02] Angela: It was interesting when we started, what we had to do, Nick, was we actually looked at the pediatric literature to kind of comfort people, to reassure our funder, was National Science Foundation.

[00:31:13] Angela: and the doctors when we first started recruiting because. pediatricians are very good at giving short durations. At that time, you could only find it in pediatric literature that people were giving even under seven days. And that actually really helped our argument.

[00:31:27] Angela: But when we started, there was nothing , all the retrospective data we could find were much more for longer durations. So it was very hard to reassure people, until we found the pediatric literature actually. because we did have that extra arm, we had this crazy CRP arm where we actually stopped treatment at five days if the CRP had fallen, by 75%.

[00:31:50] Angela: So, that scared some of our doctors, I think in the beginning. We had to reassure people a lot. but then Dafna's study was published and it, really did help us, I would say.

[00:32:00] Nick: What would be super fascinating would be a longitudinal, observational study looking back over the last maybe 20 years, to antibody treatment, durations for bloodstream infection.

We have great population based data in our province, Ontario with 14 million people, but we lack inpatient, medication use. the detail that we love out, but if we could somehow look, it'd be really neat to see how things shifted, from by the time BALANCE start to the time BALANCE finished with, your, study in between and Dafna's study in between.

[00:32:27] Nick: And then to see, uh. What more work needs to be done. Yeah,

[00:32:30] Angela: yeah. And like you were saying, other populations, right? Like we felt sort of obliged to exclude transplant patients, certainly patients with neutropenia. those patients historically get, like, three weeks of, in fact, I think it's still in guidelines to give, renal transplant patients up to three weeks of antibiotics for bacteremia.

[00:32:51] Angela: David Vine is working on that. He's trying to get off the ground. A duration trial just for that population. there's so much more now we can do, I think BALANCE really opens it up, right? Like you've done it for, the general patient profile.

[00:33:06] Angela: and on the basis of that, we can now go into different patient populations and be more bold,.

[00:33:11] Erin: Yeah. I always say the best details are always in the supplement. So when you dig into the supplement and to kind of round out our conversation about the patients in this trial, to Angela's point and how this has really just broadened what we think about for shorter courses, you guys also included polymicrobial, bacteremias, which I would say have traditionally been excluded.

[00:33:28] Erin: We wanna be pure, we wanna understand the intervention to the pathogen, but six to 16%, depending on what group you look at, were polymicrobial infections. And then you also included non fermentors. So about 5% of the patients had pseudomonas. Can you talk to us about the decisions to include those patients?

[00:33:43] Erin: Because honestly, thank you. We're so glad you did.

Rob and I are pretty pragmatic people, so I think anytime there was a decision to be made, we always lean towards the pragmatic direction. So we wanted to include as many patients as possible to help answer questions that people, didn't really have answers to.

[00:33:59] Nick: there seem to be collective equipoise out there for all these questions across clinicians, any individual clinician. May disagree on specific points. I'm not gonna randomize poly microbial patients, or I'm not gonna randomize pseudomonas patients. But there are other clinicians out there that would, and there were so many sites enrolling and there was randomization stratified within sites.

[00:34:15] Nick: we were able to address it. I also love seeing that stat, like you said, that six to 16% of poly microbial, it's kind of another little piece of evidence that we have a generalizable, population.

[00:34:25] Angela: I'm so happy you guys included pseudomonas because we didn't, and I always felt bad about that. We had a lot of debate about it.

[00:34:32] Angela: It is a bad organism if you are. Very immunosuppressed. It's an opportunist. Thank you so much for including it because now, you, broke the taboo, I think, which we weren't able to do.

[00:34:45] Rob: of the things that having such a large number of patients is you say a good chunk of patients that have pseudomonas and enterococcus and things that you might be a little bit more interested in getting into the details about how those subgroups performed, with the intervention.

[00:35:00] Rob: And that's something that we can tease out a little bit and hopefully in the near future be able, to give a little bit more information on that.

[00:35:07] Nick: We're very diligent in not looking at them during the initial.

[00:35:09] Nick: study manuscript production Adhered really closely to the statistical analysis plan so that we wouldn't be, multiple hypothesis testing, but there's, definite plans to pursue some of these subgroups.

[00:35:21] Angela: Yes, we need to know this. Very excited to hear how it went by pathogen.

[00:35:26] Erin: It might be time to move on to the main results, which we've alluded to for sure and people probably know 'cause they've read the trial. But drum roll please. Nick, do you wanna do the honors and, share with our listeners what your primary outcome ended up being?

[00:35:42] Nick: Yes. Our primary outcome was 90 day all cause mortality. We feel like that's pretty objective as it gets and patient important in the seven day treatment arm. 14.5% of people had died by day 90. In the 14 day treatment arm, 16.1% of patients had died by day 90. So the risk difference is minus actually favoring seven days of treatment.

[00:36:04] Nick: The 95% confidence interval spans from minus 4.0 to plus only 0.8. So that upper limit plus 0.8 is really far from that plus 4% non-inferiority margin. So we can say really clearly at the end of this, at seven days, treatment was non-inferior to 14 days treatment for the outcome of 90 day mortality.

[00:36:22] Nick: That was in our intention to treat analysis. we also ran for protocol analysis where we limited it to the people that got within plus or minus two days of their assigned duration. Same result, and a modified intention to treat analysis where we removed anybody that died in the first seven days of treatment, which is before your treatment arms diverge.

[00:36:40] Nick: And again, found, the exact same thing. And our results were also consistent across. Secondary outcomes of interest and also pre-specified patient pathogen, syndrome subgroups. So a pretty consistent finding to say that seven days of antibiotic treatment is non-inferior to 14 days of antibiotic treatment for patients with bloodstream infection,

[00:36:57] Angela: I think in infectious disease clinical trials, it's rare to have such tight confidence intervals and it's beautiful. . We don't often get that level of certainty, So thank you.

[00:37:08] Nick: took a long time to get there. We're starting to see that more and more though. I mean, look at the bling three trial. Incredible. Like enrolled 7,000 patients.

[00:37:14] Nick: Such a tough, intervention to study it's amazing because some people call that a negative trial. 'cause I think the P value is zero six or something. But you have. 7,000 patients, you're not gonna get more than that. And you have very tight precision around the estimate.

[00:37:27] Nick: So we know the answer of how much it works and it works. modest amount.

[00:37:31] Angela: I think we have to thank Canada, Australia, these are countries that are ready to fund, to the degree that's needed. Good clinical trials.

[00:37:42] Angela: Where we actually get to answer questions and not lead to more doubt, because oh, , the results sort of neither here nor there because the power is missing. it's nice to see this trend even in our field, even in infectious diseases.

[00:37:55] Erin: Well, it's to quote our good friend Vance Fowler, who wrote the editorial for this trial, who said, you guys did such an amazing job, such an inclusive patient population, such precise results, non-inferior across all three analyses, the only thing this trial is missing is the United States, which is true, right?

[00:38:10] Erin: We couldn't enroll patients in the United States 'cause of some barriers here. But all in all. just so amazing.

Did anything surprise you about these results the data comes in and you get to open it with your investigators. Is that what you were expecting or were you surprised?

[00:38:24] Rob: I think we started with this notion that, seven was,, probably as good as 14 to sort of phrase non-inferiority in common terms.

[00:38:31] Rob: But I don't know. I have to admit that when we were unblinded to the findings, the notion that, 90 day mortality veers towards an improvement for the seven day treatment group, I mean, I, I don't wanna overemphasize that and it was fairly consistent across the pre-specified subgroups.

[00:38:45] Rob: I think that was, if not surprising, it was intriguing. and a bit reassuring, that at least the directionality was in a way that we had wondered about and hoped. But, you don't know until, you do the trial.

[00:38:57] Angela: May I interrupt you there? Rob, why, ? Would there be this directionality? Was it like c diff for everyone who got 14 days and they dehydrated what, were the, costs of having 14 days of antibiotics that could explain the difference in the end?

[00:39:12] Rob: Yeah, I it's a great obvious question. I guess be clear, we showed non-inferiority. We didn't show that, one treatment strategy is better than the other, and so you'll get a point estimate that's not gonna be bang on zero. But, it's gonna be a little bit on one side or the other.

[00:39:25] Rob: And so just to say, they're non-inferior, . We upfront wanted to look at these reasons why it could be better or it could be worse. And from the data, I don't know, Nick, I don't think we have anything that we can point to that says, this is the intermediary or the mechanistic, result that, helps us understand, that there is really any difference.

[00:39:45] Rob: it was a bit surprising to show, a shorter hospital length of stay with seven day treatment. And, that as an outcome as opposed to mortality, we could say, well, you're thinking about seven or 14 days of antibiotics, and some people are on IV and, maybe there's some practical reasons why people [00:40:00] might be able to get outta hospital earlier and come up with, a rationale as to why that might be the case.

[00:40:05] Rob: But in terms of like mortality, non-inferior. Encourage that it's going in the direction of our original hypothesis, but, hard to come up with a reason why it might be better. Nick, what do you think?

Yeah, for me. Pleasant surprise that the point estimate was so far favoring seven days of treatment.

[00:40:19] Nick: I don't know, for Angela question, I guess dysbiosis and antibiotic resistance and the sequalae of that, we weren't able to show that in this study. We captured antibiotic resistance only through routinely collected clinical specimens. So any other blood cultures or urine cultures or, other specimens the patient had collected followup which most of which would've been in the index hospitalization.

[00:40:40] Nick: So we don't have super rich data capture on that. So if that's the mechanism I wouldn't necessarily be able to establish it.

[00:40:47] Erin: to me as a pharmacist and a steward, and Angela, you mentioned we like to talk about c diff, and of course that is a concern. And of course antibiotic exposure is our most major modifiable risk factor for C difficile infection.

[00:40:57] Erin: But now that we have more large trials in the ID space, we continually see a c diff rate less than 2%, and even in arms. You would hope to see the 14 day group would have, you know, a.

8% c diff rate and a 1% in the seven day. And we could really pin to that. And repeatedly we've seen in very large trials, that's simply not true. C diff is actually quite rare still, thank goodness. and despite more antibiotic exposure. So that's interesting 'cause that's clearly a steward thing we hang on.

[00:41:21] Erin: Right? It's like you wanna decrease your likelihood of c diff. But let's talk a little bit about your subgroups. That's the punchline. That's the three groups in the main study. So I think now with the trials of Angela's fantastic trial, your trial, the other trials, I think for gram-negative bacteremia caused by a urinary or intraabdominal source of source control. Seven days the end.

[00:41:40] Erin: Like this is the nail in the coffin for that syndrome. And hopefully people are on board with that. And I don't think there's a lot of debate in the community anymore. ICU patients though. So when you talk about the patients that were enrolled in the ICU, that confidence interval is much wider. And then when we get to the Apache scores greater than or equal to 25, or patients who had vasopressor use , then you start to cross that line and you get to perhaps within that subgroup where it's not non-inferior.

Now we can't make a call in the subgroup, but what are you guys hearing amongst your community? What's the debate here? Are people like, yep, I still feel very good about seven days. Or are they saying, eh, those bigger lines make me nervous?

[00:42:15] Nick: Yeah.

[00:42:16] Nick: There's a great, systematic review in JAMA Internal Medicine, five or six years ago, I would say, where they looked at all RCTs that made a subgroup claim in their abstract. And,, most were not pre-specified, most induced physical tests. Hypothesis, which would be a test of interaction and zero were corroborated in future follow-up trials focused on that subgroup.

[00:42:35] Nick: So like most subgroup, findings tend to be, spurious in the end. The other thing I, would say, when people read the forest plot of sub, figure three maybe. the eyes get drawn to the ends of the confidence intervals and whether or not they're crossing the line.

[00:42:49] Nick: So the crossing line of no difference or crossing the line of non-inferiority or try to convince people to direct their eyes more to the point estimates, because competence intervals, we draw them as flat straight lines, but the probability isn't like a flat distribution. Probability is like kinda have a bell curve , most of the probability is around the point estimate and only the little bits of probability left at the tips.

[00:43:08] Nick: And then we design the trial, the sample size calculation based on the overall study. So subgroups are by definition. underpowered for that noninferiority margin. So in BALANCE, when I look, I don't, worry too much about some of the confidence interval being wider across the line. I, feel like my general message when I read down it is that we have a pretty robust finding across the different, subgroups.

[00:43:29] Nick: We also did do statistical tests of interaction. None of the subgroups show a significant difference across the different, categories of the subgroup. So I feel like for the most part, I wouldn't worry too much about those subgroups, but I know I'm not gonna convince of the critics out there.

[00:43:43] Nick: And then they have particular worries about, one group or another.

[00:43:47] Erin: Well, I'm glad you said it, you heard it here. At least we can point back to this now. 'cause the same applies then for the other subgroups. I think of interest are the gram positives and the polymicrobials, right? So where you run into that same situation, those confidence intervals are wider.

[00:43:59] Erin: But you're right, when you just look down, those black dots are all beautifully to the left, seven days non-inferior. to me, I take these data and think, wow, we just have tremendous opportunity for uncomplicated, strep enterococcus. my wheels are already turning in my head for how I can help patients across my massive system.

[00:44:15] Erin: think it's really important. And something you guys wrote, and Nick, something you said in your ID week presentation, which I loved, is that really in totality what we can do with all these data and patients within subgroups is. Approach it with the mentality that seven days is going to be okay. And then let the patient guide it , and some patients may need a little bit more and that's probably okay, but have that mentality rather than saying everyone needs 14 and having to justify less go in that says standard care is seven.

[00:44:44] Erin: Some patients, Angela, to your point, may end up at five. They're totally fine. Some patients may need a little bit longer, but let the patient declare that and have the mentality that seven days is where we should start. And I love that framing. Well, we've talked about everything we love. Angela, do you wanna guide us through some things that we can critique?

[00:44:59] Angela: yeah. Can you guys just tell us what did the reviewers not like? What did they give you a hard time about when you

[00:45:05] Erin: Oh, I love that. Yeah.

[00:45:07] Angela: what were the sticking points with your peer reviewers?

[00:45:10] Rob: That's a good question. I don't know if I can, replicate, in my mind what they put on the page, but some of the themes that you've brought up, I think are ones that did come up in the review process and so.

[00:45:20] Rob: , For any trial, for lots of reasons we'd like to, blind, a number of participants to the intervention and patients and family and clinicians and evaluators, et cetera. we found even in the observational study work that it would be practically impossible to blind for instance, have placebos for all of the anti-microbials that are used in the treatment.

[00:45:42] Rob: And so even though that was one we thought about, we, entertained it, it just, wasn't gonna be possible for this kind of a pragmatic trial.

[00:45:49] Rob: we talked a fair bit about exclusions of patients that I think probably did come up in the review process and, we have to decide if these are generalizable to patient populations that weren't represented as much.

[00:46:00] Rob: And a few populations have come up as we've talked like. organ transplant patients that we excluded, neutropenic patients that we excluded. , I don't know what Nick might imagine in redoing the eligibility criteria, but I think now we might think differently about that and say, , we should consider including these patients.

[00:46:18] Rob: so that came up in the process and while we can't generalize the results to Staph Aureus, another great trial ongoing, looking at that organism, we certainly talked about it, upfront. And so other things that I think we might have considered differently and that came up in the review process to some degree, non-adherence, that's a, super important one.

[00:46:38] Rob: We knew there was always gonna be non-adherence. And importantly as you both said, we wanted this to be a trial that, clinicians could modify treatments, should a patient remain, seriously ill or have an unresolved infection with clinical signs and symptoms.

[00:46:53] Rob: on the flip side, I suppose get better very quickly. but that upfront strategy of seven to 14 days, did lead to, median actual treatment. Durations very close to this, but the idea of retaining clinician judgment for individualizing therapy to some degree for very specific patients,, hopefully some objective.

[00:47:12] Rob: Signs, symptoms, et cetera. but I think the generalizability question we've addressed lack of blinding in terms of an internal validity question. I think we just found we couldn't do. And, the non-adherence issue, I think to me is probably about the most important one to try to work through and to make sure we can incorporate this into, our clinical decision making at the bedside.

[00:47:33] Erin: Yes, Rob, thank you. I do think the non- adherence point is very important and we've alluded to it several times now. For our listeners, I'm gonna read specifically what was in the trial. So patients were treated for longer than the assigned duration of treatment in 23.1% of the patients in the seven day group and 10.7% of patients in the 14 day group, which, people are gonna clinging to that 23% in the seven day and say, oh, they got eight, they got 10, but 10, 11% in the 14 day group is also kind of a lot, and that a lot of antibiotics we're giving for bacteremia.

[00:48:05] Erin: So I think both of those are important and again. Adherence was defined in the study as seven or 14 days plus or minus two. So if you got five to nine days, you were considered adherent in the seven day group. And if you got 12 to 16 days, you were considered adherent in the 14 day group.

[00:48:22] Erin: That's also very practical. And we see that even in observational bacteremia studies, which I like the group out of Hopkins does this very nicely. They say, we intend seven days and healthcare is imperfect. operationally you might get one dose at 1:00 AM that was supposed to be given the day prior.

[00:48:37] Erin: And if you're counting calendar days, , if you're on Q eight hourly medicine, what's a day? Is it one dose on any given day? Do you have to get 50% of the total daily dose? And that's, I think, something we struggle with. But I think the cleanest is to do a calendar day, at which point, you're gonna have this plus or minus.

[00:48:52] Erin: And we've talked about how you guys stated it really beautifully is that it's the strategy to use a shortened duration that's non-inferior to committing all patients to 14 days, which I think is so important.

[00:49:05] Erin: And then you did that very robust per protocol analysis that was limited just to people who were adherent and that still showed no difference. So I'm not as stressed about the non-adherence, I know people are. Nick, what are your thoughts on that?

[00:49:17] Nick: of all the limitations in the study, it is the non-adherence one that gave me the most angst, throughout BALANCE.

For the listeners that had to think through this. in terms of a superiority versus a non-inferiority trial, non-adherence is a way bigger threat in a non-inferiority trial like this. Because in a superiority trial, if you have a lot of non-adherence, it's gonna dilute out your signal and you won't be able to find that superiority.

[00:49:38] Nick: Result. But in the non-inferiority trial, when it dilutes out, the difference between the groups pushes you towards finding falsely a non-inferiority result. So we didn't wanna be able to find non-inferiority when non-inferiority didn't actually exist. I'm personally very reassured by Rob's philosophical approach.

[00:49:55] Nick: Like we've said, that we think about it as a treatment strategy study. The non-adherence is not [00:50:00] as big of an issue. But I imagine that some of your listeners won't, be reassured on that point. So I just wanna provide some in press reassurance or under review reassurance. So Sean Ong is a brilliant infectious disease, clinician from Singapore, doing a joint PhD at the University of Toronto with us and University of Melbourne, Steve Tong and Josh and others, and focusing on, innovative, work in clinical trials.

[00:50:23] Nick: And so one of his thesis projects was using BALANCE, so he had to wait for BALANCE to get done to do this one. But look at. Non-adherence in great depth. So the first thing we wanted to do is look at what are the predictors of non-adherence so that people could learn like what types of patients in BALANCE had treatment prolongation, so people that had treatment prolonged beyond their duration seven hours.

[00:50:43] Nick: You pointed out even beyond 14 people, over 70 years of age, people in ICU at baseline, people with a high organ failure score, people with persistent fever, people with persistent bacteremia. So a lot of things that kind of, maybe make sense in terms of things that came out as predictors of shortening of treatment was just vascular catheter source of infection and a highly resistant microorganism.

[00:51:03] Nick: Actually,

Nick, did he also look at the day that the patient was randomized? were those who were randomized earlier on in their bacteremia? Were they less likely to be adherent

[00:51:13] Nick: That's a great question. I don't think we did look at that, but something to.

[00:51:16] Nick: look at, cause there should be that trade off of higher potential protocol adherence with earlier randomization. But what I wanted to point out was that Sean took like this plan way further than we had anticipated and actually uncovered in the literature some really novel techniques to account for the impact of non-adherence on the outcome of a trial.

[00:51:34] Nick: So there's some, ways of doing this. Inverse probability of treatment weighting and instrumental variable analysis, which people have probably seen used in observational studies to account for, selection bias. They can actually be used in a trial. No one really does this. I think some systematic could be found that only like 2% of trials have done this.

[00:51:50] Nick: But you can use these methods to account for that post randomization confounding that people that don't adhere are different than people that do adhere. And when Sean applied all of these methods, the result of seven versus 14 days treatment looks exactly the same. So that was super reassuring, me that we can believe the result at seven days is not inferior to 14 days.

that even though there was non adherence as we expected, anticipated, it doesn't change that finding really. So that's, that will hopefully be in that infectious disease journal, for people to look at sometime soon. that might make people feel more comfortable about, shortening.

That's really great that you guys actually turned this into a subject of study.

[00:52:25] Angela: I wanna learn more about these methods because it's obviously a problem that comes up again and again. oh, my patient's different. I agreed to randomize, but no, , no, it's not gonna work for my patient. It's that tension, right? . And as we get more, emboldened to question dogma, to question guidelines non-evidence based. Recommendations, let's say, we're going to need, Sean's methods because, there always will be this problem of non-adherence.

I think we as reviewers, as readers, we need to simply understand that you are never gonna get perfect internal validity. not if you want it to be externally valid.

can I ask two, questions that we had No good place to put them. we do like a little rapid fire wrap up of the supplement mostly.

[00:53:10] Erin: So in table SS six, ' you said, patients that received any NAL treatment was about 60%. Okay. So 60% of patients got some kind of oral treatment.

[00:53:20] Erin: Can you guys clarify, was that all step down, everything at the end or were there, dare I ask, were there any patients in the study who got exclusively oral or oral upfront or things like that?

[00:53:29] Angela: Hey Erin, Aren't we not supposed to say Step down?

Early

[00:53:33] Angela: oral switch. Early. Oral

[00:53:34] Erin: switch. Yes. EOS.

[00:53:36] Angela: Yeah. Step down would imply that it's not as powerful if it's oral.

[00:53:39] Erin: okay.

[00:53:39] Erin: So you can't, you're say

[00:53:40] Angela: step down. Yeah.

[00:53:41] Erin: Okay.

stepped

[00:53:42] Erin: up, stepped up to

[00:53:42] Angela: oral. It's better step sideways. Switch

[00:53:46] Nick: control. Find step down change to switch.

[00:53:50] Erin: Okay, but that was all therapy. Continuation was 60% on orals. That's also, again, that's not really what you sought to study, but that's a lot of patients. That's a lot of people were treating with oral therapy and feel comfortable treating with oral therapy, which I think is an important take home from that as well.

[00:54:03] Erin: 'cause that was all clinician guided. You weren't telling them to do that or not do that. So I think that's very important. And then, this happens all the time clinically where we have a patient, you know, e coli, bacteremia. They perhaps are on inactive therapy as the get go, but then by day two they're better, they're clinically better.

[00:54:23] Erin: And we don't get their susceptibilities till day two, day three, sometimes day four, just depending on your lab or whatnot. And you get those susceptibilities back and you're like, oh shoot, they've been on Pip-tazo and it's an e coli with an interesting phenotype and it's pip-tazo resistant, but ceftriaxone susceptible.

[00:54:37] Erin: And then we have this existential crisis of do I need to now give them seven days and not count the first four days, even though the patient's better? I think in your trial day one counted as the first day of active therapy. Right? So those patients that were on inactive therapy. Okay, so Nick is giving me a thumbs up for people listening who cannot see him right now, but

[00:54:58] Nick: answering the second question first, it was based on the number of cumulative calendar days in which you got at least one dose of a drug active against your, index blood culture pathogen that got you enrolled in the study.

[00:55:10] Nick: Time couldn't go back to before that book culture is collected. So even if you were on something a couple days before the blood culture is collected, blood culture collection date at most could be the first day. And each day thereafter you had something active we couldn't take into account pharmacokinetics and pharmacodynamics and dose and frequency.

[00:55:25] Nick: And we left that all up to the treating team. For your first question around the switch, it was I think 99.3% I wanna say had at least some intravenous treatment. So these weren't people typically being started on oral therapy 'cause it was such a sick population with bloodstream infection. Almost everyone got intravenous And then like you said, 60% got some oral along the way and there'll be more in the 14 day arm that got oral than the 7 day arm because they're on treatment longer and had more opportunity to switch over later in the course.

[00:55:52] Nick: And we do have lots of plans to do some secondary analysis to dig into that in more detail and look at the timing of oral switch. compare outcomes with people getting beta lactam and non-beta lactam. And a few other things, , in the oral group.

[00:56:04] Erin: Oh, that's fantastic because you have such a rich data set

I think in that same supplement is about 9% of people that, had inappropriate initial, treatment and we had to react to that and so that's just real life and Yeah. Started the clock from active therapy, as you said.

I think that's what we do too. We do seven days from, active therapy.

[00:56:22] Erin: But it does beg the question, right. If they're better, it's like, could those days have counted?

[00:56:27] Angela: Rob and Nick, if you could do this over again, your 10 year study, what would you do differently if anything?

[00:56:33] Rob: one thing I wouldn't do differently is I wouldn't pick another co-principal investigator. ',

[00:56:38] Angela: That is so sweet. Nick's turning red. Everybody,

Nick would say the same thing, like, never do one of these trials on your own .

And then if you do you have to have someone to do it with. Hundreds people would've joined onto this study.

[00:56:50] Nick: so many countries without the, the Rob Fowler effect never would've happened.

[00:56:53] Angela: I mean, you guys were tested 10 years of this trial and all the fundraising you had to do and all the logistic nightmares, I can just imagine.

[00:57:02] Nick: That's the one thing we both agree on, that we would change would be that non-inferiority margin, we really wanted it to be much smaller than previous studies and serious bacteria infections so that we could really know that, it was safe to do this, but.

[00:57:13] Nick: If we could have just gone with 5% instead of 4%, we would've finish this years ago.

[00:57:17] Angela: I have to confess, I think in the entire publication, that is the number that jumped out at me, honestly. Like that 4% mortality, margin, not many trials in infectious disease had that much courage.

[00:57:32] Angela: 4% is tiny For our typical ID trials, you guys were so intellectually honest. it was so, courageous because every percentage point means, God, another year of my life on this trial. when we were trying to see how, small we could, realistically get that non-inferiority. When you ask people and it's arbitrary, right? There's no one number, something that seems really high and something that seems that's impossible, but when you're asking people, would you accept, you know, up to 4% difference.

[00:58:03] Rob: People think about that and say, oh, wait a minute. No, I don't, want one treatment to be 4%. Okay, what about 3%? and we get down to asking groups of people. And at 1% people say, okay, fine. 1%. and you're like,

[00:58:15] Angela: oh my God, I have to randomize the whole North American continent.

[00:58:19] Rob: You need everybody. but so focusing as Nick said on the, point estimate eventually, with some, degree of confidence around that. If the point estimate was, , 3.9% absolute difference favoring, higher mortality in a seven day arm, that might not have been very comfortable.

[00:58:34] Rob: And I think we'd be having a very different conversation looking at the point estimates though, , and maybe,, 5%, non-inferior margin maybe that's a better trade off than what we initially, chose. but as you say, live and learn a little bit and maybe we do it differently.

Well, guys, congratulations. , It's seriously amazing work. It's been amazing discussion and the time has come for the break points. Faithful for the I feel nerdy segment of the podcast. I feel nerdy is meant to be a safe place and a closing segment for panelists to nerd out over their favorite ID topics, quirks, and fun facts.

[00:59:04] Erin: And this Angela is making fun of me. 'cause So communicable does the, like get to know you. It's just supposed to be non-science related whatsoever. And I put this as the get to know you question and Angela's like, Erin, you're such a nerd. Like, no, the get to know you is supposed to be about concerts.

Question is, what's next? What's the next question you want to answer? And it doesn't have to be something you're necessarily already working on, but like, if money just rained from the sky and you could do any trial in the world that you wanted to do, what trial would you design?

[00:59:32] Nick: I guess I'm pretty fortunate that we have somehow gotten funding for that next, extended trial that we do wanna do.

[00:59:38] Nick: And the idea there is, we have a traditional single question trial, a platform trial asking multiple questions at once. We're drilling down just to the gram negative, blood stream infections compared to BALANCE.

[00:59:49] Nick: So it's a little bit more homogeneous, but it's still a mix of pathogens and, sources of infection. And there's five questions. The first one would be deescalation to the blood culture sensitivity result [01:00:00] versus no deescalation. next one in, people ready to switch to oral therapy, beta lactam versus non-beta lactam oral therapy.

[01:00:07] Nick: And then the small subset with a low risk, now that we call 'em low risk CE organisms, randomizing them to carbapenem versus, ceftriaxone treatment. And then. There's a diagnostic question, which is, do we need to repeat the follow up blood culture? And so randomized people to follow up blood culture versus no follow up blood culture.

And the one syndrome specific question that we started with, which is the trickiest one to do is, should we replace or retain the central eye in the patient with less infection in central?

[01:00:33] Angela: Wow. I love it. Okay, so I am surrounded by nerds because Aaron's question, was meant to be , what trial do you wanna do next?

[01:00:42] Angela: And then Nick's and Rob's answer is, oh, well we're already doing it. ,

[01:00:46] Erin: So his answer was not one but five trials, which is,

[01:00:49] Angela: which is amazing. Rob, Anything you wanna add to your dream trial that you're already running?

[01:00:55] Rob: I mean, there's, there's so many that are, important to do , and trying to get to those questions as efficiently as we can. the idea, looking back at answering one main question over about a decade, it's great in a sense, but at the same time you think, oh my gosh, really?

[01:01:10] Rob: Could we have done something more efficient in this notion of trying to appropriately blend in other questions to a defined patient population as Nick described in BALANCE plus? I think that's, clearly, where most people are thinking in trials if we went back in time, Nick, like, 15 plus years, I think we might, Tell, our past selves, hey, you should consider, , the approach that, people are taking now and, BALANCE plus is taking. So I guess I would wanna highlight all of Nick said, and, and the importance of trying to get through these questions as efficiently as we can.

[01:01:41] Nick: do you wanna tell about the sixth domain.

[01:01:43] Nick: It doesn't exist yet, but that we're thinking about brought today's conversation.

[01:01:47] Rob: Yeah, that's a good point. Case doesn't come up. So I suppose we, we've kind of. Implied a couple times during the conversation that, shorter is seemingly just as good as longer, in this case, seven, 14 days.

[01:01:58] Rob: But the notion that, we ideally would want to, individualize our, treatment it's tough to know when to stop antibiotics and that's come up a few times and, figuring out the best ways to do that. What patient characteristics, what diagnostic characteristics can we use?

and I think a sixth domain that we're working towards is to do something that would be seven days versus a shorter duration, for patients that, would be, you know, clinically reasonable based upon, we'll say likely some degree of improvement and, lack of markers that would, push us to prolong Nick mentioned one of the. terrific collaborator, Sean Ong. he's also looked at the, BALANCE data set to say what happens to patients physiologically and biochemically over the course of their treatment. in an observational over time way, it does seem that by day, I'll say by day four, but day three to five, most people have a resolution in a lot of those things that we would, think are signs and symptoms of infection, at least active infection.

and so looking at a, patient population that by, , somewhere day three to five, maybe day four has improved, could we shorten antibiotics, in a group? And so that's probably , a next question that we hope to incorporate into the BALANCE plus gram negative, bacteremia trial.

[01:03:12] Angela: I think erin, you and I have talked about this on previous episodes. Maybe I'm convinced that with gram negative organisms, I'm not gonna speak for gram positives just yet, but for Gram negatives, I'm convinced that anything you do, after day three is probably just detail.

[01:03:28] Nick: Hmm. You know,

[01:03:29] Angela: I think once you get that, bacterial load down in the beginning, I really think a lot of the stuff we do afterwards, it's just window dressing, you know, oral switch.

[01:03:38] Angela: And with gram negatives, especially e coli, in our trial of 500, we did not have one metastatic complication of any gram negative. not even one. We had some local superlative, you know, renal abscess next to a pyelonephritis, but we did not have one metastatic complication of any gram negative.

[01:03:58] Angela: you know, it just. Doesn't come back to haunt you like the horror show. That is staph aureus, ? And so I am thrilled to hear that you guys are gonna do this. We need to establish the lower limit.

I am thrilled you're asking both questions. 'cause Nick, you mentioned, which is a burning question in the community of do I need a fluoroquinolone or trimethoprim-sulfamethoxazole or can I use a beta-lactam? And then, you know, US CAST just did a really nice exploration into the beta-lactam PKPD and from A-P-K-P-D standpoint, the beta-lactams.

[01:04:26] Erin: Shouldn't work for gram-negative bacteremia, but we use them clinically all the time. Angela and I've talked about this, it's like, is the question fluoroquinolone or beta-lactam or is the question four versus seven days, right? if you get three days of ceftriaxone, does the last four days even matter?

[01:04:41] Erin: And so you guys are asking both, which is really cool. And I can't wait to see that come out. Hopefully not in 2035, hopefully before, but we'll see.

[01:04:51] Nick: Platform trials are more efficient 'cause you can ask more questions at once. But then it also cause problems, 'cause questions can, interact and cause each other some difficulties.

[01:04:59] Nick: Like why would you randomize patient X to be lactam versus non lactam? For that last part of treatment, if you would also be willing to randomize them to

[01:05:06] Erin: three versus seven days. Right, exactly. Or four versus seven days. So you're right, that it is a problem. But it's still pretty awesome.

I will go back and add a little caveat. I really, believe strongly that individualizing is the, key word. It's not just, oh, let's get to the shortest. Possible duration. Let's get to the lower limit. I still think we need to very carefully look at the pathogen and the host.

It's not for everybody to do three days. I don't wanna be cavalier., But I do think if you, use some good clinical sense, right, biomarkers such as temperature, what are the kinetics of the fever dropping, simple things like that. I'm not saying everyone needs a CRP, , but, looking at your patient, listening to your patient, obviously knowing the pathogen, I think, the individualization is key, but you can bring durations down in some populations. some infection types, of course.

[01:05:56] Erin: Well, and, and not to, like, we're gonna have a whole nother hour here on your, on your future trial.

[01:06:00] Erin: We're gonna have to have you guys back in five years when you finish this one. But the follow up blood cultures, thank you for doing that as well. That's very important because I think some of that nuance gets missed. The 2019 landmark gram negative trial, which we've discussed that did require follow up blood cultures.

[01:06:13] Erin: So I have some providers who won't do seven days without follow up blood cultures. They're like, I'm in if I follow the trial protocol, which was follow up. And so, I think that's a really important thing, to assess too. In my residency I was taught you never need follow ups for gram-negatives and you always need them for gram-positives.

[01:06:28] Erin: And the right answer, as with all things is somewhere in the middle. Right. Some gram positives. You probably don't need to document clearance. And some gram negatives, you probably should, especially if the source is prostate or something like that, that can, you know, who knows? But

[01:06:39] Nick: you have to know that's the one enrolling like wild fire compared to the other domains.

[01:06:42] Nick: So follow up culture already. Yeah, it's first interim analysis passed the 1000 barriers. So we might have. Answers for you soon.

[01:06:49] Erin: Fantastic. Alright, well guys, this has been incredible. Incredible. Yes. Thank you so much for the discussion.

[01:06:56] Angela: So. to our two wonderful guests. As we wrap up, do you two have any last messages for our listeners?

Yes, we got the great fun of being here for this conversation, but, we'd like to thank so many people, so should mention Asgar Rishu who is our BALANCE project manager, who coordinated this entire international undertaking, our brilliant biostatistician, Ruxandra Pinto, our amazing BALANCE steering committee, our wonderful DMC, more than a hundred individual site investigators who were such an honor to meet and work with.

[01:07:27] Nick: And then this incredible group of research coordinators across these 70 plus sites doing all the hard work of screening, enrolling patients and, entering data. and then last but not least, 3,600 patients and their families who generously gave participating in BALANCE and, helped push forward the science and antibiotic treatment treat should.

[01:07:46] Angela: Beautiful. Thank you so much to our guests, Nick Daneman and Rob Fowler in Toronto, Canada.

[01:07:52] Angela: And thank you to Erin McCreary in Pittsburgh, USA for co-hosting with me, Angela Huttner in Geneva, Switzerland. Thank you for listening to communicable the CMI Comms Podcast and breakpoint, the Society of Infectious Disease Pharmacists Podcast.

[01:08:09] Erin: This episode was peer reviewed by Arjana Zerja from Mother Teresa University Hospital Center in Albania, and edited by Katie Host Oy. Communicable theme. Music was composed by Joseph McDade and the breakpoint theme music was recorded by SIDP member Dr. Steve Smoke. You can subscribe to Communicable and Breakpoints on Apple Podcasts, Spotify, or wherever you get your podcasts. Thank you for listening and helping CMI, comms and ESCMID move the conversation in infectious diseases and clinical biology further along and helping SIDP achieve our vision of safe and effective antimicrobials for now and the future.

[01:08:47] Angela: you guys are awesome. Thank you. I'm so sorry this took so long. It was too much fun.

[01:08:52] Nick: No, that was so much fun. I, I think you cured me cough unless I was having so much fun. I was coughing and not even noticing that I was coughing.

[01:08:58] Angela: No, you didn't cough. I actually don't think you coughed even once, Nick, I think we cured you.

[01:09:02] Nick: Yeah, that's it. I think we have,

[01:09:03] Angela: I think we have to do this more often. Yeah.

[01:09:05] Angela: for, sitting through this. Erin, I don't think we've ever recorded for the almo.

[01:09:08] Angela: Like don't worry, we're gonna cut this, but we have never gone to almost two hours actual recording time. It's a good thing we did reserve the two hours. Yes, we have. Angela, you and I talk a lot. I know, actually, yeah. Sorry you guys didn't know what you were getting into.