Thank you, Dr. Gutsa. And just now we'll transition to the main part of our grand rounds for today. And it really is an honor for us to have Dr. Christopher Polage here with us today. Dr. Polage is an associate professor of pathology and the medical director of the clinical microbiology lab at Duke University. He started his career out west. He was at the University of New Mexico for medical school before moving on to training at the University of Colorado at the University of Utah. 

He joined the faculty at UC Davis, serving there as the director of their clinical microbiology lab and the medical director of their specimen center as well before moving on to Duke. And really, through his career, he's become an expert in a number of areas, one of them being the diagnosis of Clostridium difficile, and speaking nationally on that topic. 

In his career, he's also won a number of clinical research awards, including the Clinical Research Achievement Award and also served on the editorial board for the Journal of Clinical Microbiology. So really, with that background, we are honored to have Dr. Polage here to speak with us on the complexities of C. diff testing and helping us chart a path forward. And I think this is really pertinent for us now as institution, as we work to try to decrease C. diff rates here. So without further ado, I'll turn it over to Dr. Polage. 

So I typically have a playground voice anyway. But we've got the microphone. So can everybody hear me OK? 

Well, first, I'd like to just thank the Division of Infectious Diseases and the Department of Medicine here at UVA for hosting me and inviting me up to come basically meet with all of the ID faculty and infection prevention and clinical microbiology faculty. It's been a lovely visit so far. And it certainly is an honor for me to be invited here to share with you some of my work and others' work in the area of Clostridium difficile testing. Let's go ahead and get started. 

So as we spoke, I've been at Duke just for about a year now. I direct the clinical microbiology laboratory down there. And I guess that's about it. Let's go. 

So I do have a few disclosures. This area of Clostridium difficile testing has been very controversial for probably quite some time. And so the more I've gotten involved in this, I've also done some consulting. So I actually work with several diagnostic companies. Some of this is related to Clostridium difficile testing. I also work with some probiotic and microbiota development companies as well. 

So just a quick overview, we're obviously going to do some background related to both Clostridium difficile a disease and related to the types of C. difficile tests. And there's several tests, frequently confusing, even for people who work in this area. And then I'm going to tell you about something that I like to refer to as the great Clostridium difficile test debate. We'll talk a bit about the impact of C. difficile testing, because within recent years, it really has become front and center, having both an impact on patient care and diagnosis but also on larger things, like what we report in terms of health care-associated infections and even reimbursement for institutions. 

And then we'll attempt to talk about optimal testing strategies, which has really been the most controversial thing, as to what's the right way to diagnose patients. But we'll take a stab at that. And I'll share my opinions and thoughts with you on that. And then we'll sort of wind up with diagnostic stewardship as one of the paths forward. 

So this really is the obligatory background slide, which probably most of you are aware of. But really, Clostridium difficile was sort of percolating under the surface for much of the '80s and 1990s and then took off in the 2000s with what was recognized as, now in hindsight, as an epidemic of a new, more virulent strain. 

But so basically from about 1999 through about 2011, there was a fairly significant increase in Clostridium difficile infections, both here in the United States, also in Canada and in Europe as well as many other parts of the world. Specifically here in the United States, as of about 2011, we were thought to not quite have about 500,000 cases per year. This was a 200% increase from about a decade earlier. And it was recognized that probably about 1% of adult hospital-admitted patients were affected by Clostridium difficile infection. And this, of course, resulted in excess hospital days and, depending on your estimate anywhere, from $1 to $6 billion in excess health care costs, so a lot of impact this way. 

And then on the right, we've just got to figure from the New England Journal point prevalence study, which is actually slightly outdated. But the new study shows very much the same thing, and that basically is that for several years now, Clostridium difficile has actually assumed the position as the most common single organism causing health care-associated infections in the United States above Staph. aureus and others. So that just kind of says where we've come from. 

So this is just a quick pathogenesis slide to really just remind people that Clostridium difficile infection is mediated by a combination of toxins and inflammation. And so what I've tried to depict here on this slide, these are colonocytes or epithelial cells lining, basically, the mucosa of the colon. And the thought process there being that then, right after those colonocytes or those epithelial cells, we have kind of an inner mucus layer. And then we are supposed to have this very mixed, complex, diverse bacterial population. And this is obviously just a PowerPoint slide, but trying to represent that we have anywhere in the range of 400 or more different species and types of bacteria in a normal healthy microbiota. 

And then what you probably all know is that when you get antibiotic exposure, especially antibiotics that have broad activity and anti-anaerobic activity, what we see is a simplification and basically a killing of many of these bacteria in the gut. And this opens a niche and an opportunity for colonization by, and then overgrowth by, opportunists like Clostridium difficile. 

And so we get colonized by Clostridium difficile. And what I've depicted on the right is this overgrowth. And then the little dots are meant to be little protein toxins of C. difficile, which basically bind to receptors on cells, are taken up, and ultimately result in an SOS-type response in these cells as well as disruption of their cytoskeleton as well as the intracellular junctions. So you get breakdown of the epithelium. And you also trigger inflammation. So it's really these toxins and inflammation that cause the destruction, the inflammation, the damage, and the diarrhea that we see, and in some cases, more fulminant disease. 

This is the kind of an overview of the spectrum of C. difficile disease. And the important part that I'm trying to emphasize here-- and we'll talk more about this in a second. But it's that you should be aware that there's really a sizable fraction of patients that get colonized with Clostridium difficile that remain entirely asymptomatic. And so here I depict them as walking around. But they could easily be lying in a hospital bed and not having diarrhea and not be recognized as being colonized whatsoever. 

So there was actually some very early work that showed that many patients actually have antitoxin antibodies to Clostridium difficile. In the '90s this was recognized. And then there were similar papers right around the turn of the century, in 2000, 2001 that suggested that these antitoxin antibodies might even be protective and that when you basically become colonized but you would have a pre-existing immune response, that this may in fact protect you from symptomatic disease. 

And then what we have is we have then the spectrum of symptomatic disease with kind of a sizable proportion of patients, about half of them, developing a mild to moderate disease but really no inflammation that's notable systemically. And if you were to scope them, they would look relatively normal. Then we have what we classically think of as pseudomembranous colitis. These are basically little volcano plaques that are really represented by protein exudate as well as neutrophils. And then we've got really fulminant, complicated C. difficile. And this is demonstrating toxic megacolon colon here on an image. 

So this is another slide that I put together which was really for a product that I did at my prior institution, where we were looking at transmission from asymptomatic carriers. And the concept here was to try and drive home to people that there's this sizable iceberg reservoir of patients under the surface that go unrecognized. Again, for every single patient who has symptomatic C. difficile disease, that there's easily five patients that are colonized without recognized symptoms or symptomatic disease. And in some units, that number, that ratio could be even greater. It might even be 10 to 1. So again, just trying to illustrate here that for every patient that's symptomatic and recognized, there's quite a few patients that are colonized and unrecognized. 

The next point I want to make is that, even though we often have a limited differential and we think about C. difficile being the primary cause of diarrhea in hospitalized patients, the reality is that actually, it only causes a fraction of diarrhea in hospitalized patients. And so this is a table that's adapted from a review article I published a few years ago. But basically trying to drive home to people that the majority of diarrhea in hospitals is either due to medications or that it's due to underlying illness, and that only a subset of that is actually due to Clostridium difficile infection, even though often, when a patient's develops diarrhea, Clostridium difficile is the first thing that comes to our mind and often the only thing on our differential. 

Now, this is a slide that's kind of a modified Venn diagram but one that I put together to try and summarize what I think is going on in terms of the epidemiology of both diarrhea and Clostridium difficile infection and colonization in the hospital. And the reality is that when we're trying to accurately diagnose patients in the hospital, this challenge is complicated by the fact that there's colonized patients, immunity, and non-C. difficile diarrhea. 

So if you think of this overall rectangle here as being all hospitalized patients, we talked earlier that it's been recognized that a subset of patients, typically about 50%, are potentially immune and have these antitoxin antibodies. So we've got a sizable portion of our population that potentially has pre-existing immunity and is protected. And then you have the other half that is potentially susceptible without the pre-existing immunity. 

You layer on top of this the fact that anywhere from 10% to 20% of patients in the hospital are colonized with C. difficile-- we actually did screening in isolation activities in UC Davis where I came from. And we saw rates of colonization in some of ICUs as high as 25% to 30%. So this number really is real, that there's a sizable fraction of patients that are colonized with C. difficile. 

Here, I've depicted that it overlaps immune patients and susceptible patients. And then the reality is just like we talked about. There's diarrhea due to all causes. And if we just imagine that this overlaps each of these, my suspicion is that what we have is that when you've got a susceptible patient, gets colonized with C. difficile and develops diarrhea, that that's true C. difficile infection caused by Clostridium difficile. 

You've got a lot of patients who don't have C. difficile and have diarrhea due to other causes. But then there's potentially this other group here that's possibly immune, colonized by C. difficile, and yet has diarrhea due to another cause and is potentially really a carrier but not having C. difficile infection. 

So now, because I'm a diagnostic test guy, we're going to take a step and a journey through diagnostic tests and give you a quick overview of those. So there's really two main categories of diagnostic tests. There's tests that focus on the toxins, so really detecting free fecal toxin protein in the stool sample directly, which is important, because again, this is what mediates actual disease. And the examples of this are the toxin immunoassay, which many people have developed a dislike for over recent years. And this is really the thing that's generated most of our argument over recent years. And then the historic reference assay, which was literally using cells in tissue culture to detect toxins, because human cells and other cells are very sensitive to C. difficile toxins, as you might imagine. 

And then we've got other tests that focus on just detecting the organism, either the bacteria or the organism is the way I think about it. And here, we're either basically doing culture, or we're doing PCR or some flavor of nucleic acid amplification detection. And the important part here is that a positive result is really just detecting the bacteria or the organism, but doesn't necessarily say anything about whether toxin is being produced and is actually present in the fecal sample. 

Now I just want to give you an overview of some of these examples of tests. We're not going to spend much time on it. But the point is, there's multiple manufacturers. Some of these are enzyme immunoassays. And then there's the traditional cell cytotoxicity, which again, is the reference test. 

Here's some examples of these, plate-based ELISA. There's some cartridge-based assays. And here's an example of the traditional cytotoxicity assay, a negative sample that then when you get exposed to toxin, you get the same kind of cell cytotoxicity rounding up and dying of cells that we see in the gut. 

If we transition to these organism tests, we've got several examples here. There's some of them that detect antigens that are conserved and shared by all organisms. In Clostridium difficile, this is glutamate dehydrogenase. And we've got several that detect nucleic acid, DNA. And most of the time, this is toxin genes. So it indicates that it's a potentially pathogenic organism, but not that it's producing toxins. 

Or you can do culture. And here's an example of culture where you've got an anaerobic plate. You've grown the colonies. And then what you can do, actually, is you have to marry that with that cell cytotoxicity assay to demonstrate that this organism that you've grown in culture can in fact produce toxins. And that combination is called toxigenic culture. 

Comparison of these tests, the main thing, I don't want to beat this too much. But again, just emphasizing that we've got toxin tests and organism tests. And there's several different ways to do it. And then trying to focus on the fact that we've got analytical sensitivity and specificity in the lab. But really, the thing we've been arguing about is what's the clinical sensitivity, and what's the clinical specificity for actual disease in our hospitalized patients. 

And the one I really want to zero in on our zoom in on is really, this is the thing we've been arguing about for several decades now, is basically, when you've got these toxin enzyme immunoassays, what's the sensitivity for clinical disease. And some people have been concerned that this was really low. And other people thought it wasn't so bad. But that's really been the controversy. 

So I like to think of this as the great Clostridium difficile test debate. And I'd like to start with this quote, which is really from one of the godfathers in the Clostridium difficile area. He's an infectious disease doctor, currently works in Chicago. But this is Dale Gerding. And this is a lovely quote from 1993. 

"The major controversy in the diagnosis of symptomatic gastrointestinal infection due to Clostridium difficile is whether laboratory evidence of the organism in culture is sufficient or if evidence of one of the toxins in stool should be required." So it's this whole deal of, should we focus on the toxin, or is detecting the organism alone enough. 

And so what we have over here on the right, this is some data from some of my own work. And basically, on the y-axis, we have the bacterial burden, the organism burden, the concentration of organism in the fecal sample. And all four of these columns are the same 100 patients. But we have two columns that are toxin tests, one that's the toxin immunoassay, one that's that cell cytotoxicity assay. 

And then we have the two organism or bacterial tests. We've got the toxigenic culture and PCR. And what I really want to highlight for you is that the toxin tests, those patients typically have a higher concentration of bacteria, a higher burden of disease. So they have a higher concentration on the y-axis. And that's where most of the toxin-positive patients are. 

And most of the patients that we're picking up with either toxigenic culture or with PCR basically have lower concentrations of organism and are typically, but not always, negative for toxin. 

So now I want to talk about where we've come over the last 15 years or so. And what I want to basically make the argument is that we had this epidemic in the early 2000s. There was a true epidemic from 1999 through about 2008. This was this novel, hypervirulent strain. It was basically the ribotype 027 BI/NAP1 strain. This was a bona fide epidemic with more virulent disease and more harm. And this scared a lot of people, both clinically and in laboratories. 

And this was during a period when most labs use the same tests. They used all-toxin tests, typically. And therefore, we diagnosed C. difficile disease based on toxin positivity. And if you were negative for toxins by these tests, you were basically not detected and not reported. And we knew that there were occasional missed cases of C. difficile infection. But this was what we'd done, and we'd done this for several decades. 

Then what I'd like to suggest is that starting around 2009 roughly through now we had a second, test-related epidemic. And what we did is we basically switched test methods en masse. We had the first FDA-cleared PCR test. And labs switched fairly dramatically en masse to PCR or other flavors of nucleic acid amplification tests. 

And now, all of a sudden, we had a definition change. We weren't just reporting toxin-positive patients as having C. difficile infection. Anybody that had the organism present, by PCR, we called C. difficile infection. And the rates went through the roof a second time. 

So here's some data. This is data from Baylor when they switched from their toxin detection method to a PCR method. And they saw 104% increase overnight when they turned that assay on. And they actually went back and became concerned that a lot of these patients didn't have real clinically significant disease. 

What we also saw is at a population level, or at a state level, we also saw a sizable jump in reported C. difficile cases with the implementation of these more sensitive tests focusing on the organism rather than the toxin. This was a collaboration between the CDC and several state public health laboratories. And they basically showed that as labs introduced PCR tests, we saw, not surprisingly, an increase in reported C. difficile cases. 

So what I'd like to suggest happened is that we really had a shift in clinical priorities, ordering practices, and test methods over time. And all of this has contributed to where we find ourselves now. So historically, in the past, we really prized specificity over sensitivity. We preferred that. We focused on toxin testing. 

And I'd also like to suggest that because C. difficile was not felt to be a significant common issue, we had a high bar for testing and diagnosis. We actually had a higher threshold for testing. We tested less often. People had to have more significant diarrhea, typically, before we tested. We didn't just jump to test patients for C. difficile. Usually, we watched. And they had to have fairly significant diarrhea before we did. And we didn't use culture, because it was slow and non-specific. And the belief was that most toxin-negative patients didn't have CDI. 

More recently, we basically have favored sensitivity over specificity. We've had a much lower bar for testing and diagnosis. And as we've already talked about, we've shifted towards organism detection. And we switched our belief to saying that all toxin-negative, PCR-positive patients also had CDI. And that's really been the rub. 

So I want to now walk you through some cases and some studies that have really tried to explore this issue of do these discordant patients that are negative by toxin but positive by culture or PCR, do they have C. diff infection and need treatment, or do they not have C. diff infection and not need treatment. That's really what we've been arguing about. 

And this is a lovely study done while I was back in high school, again by Gerding. And it was in 1986. We'd never be able to reproduce this study today. One of the things they did, it was at a VA hospital in Chicago. And one of the things they did, they actually included case patients and diarrhea-free controls. Now, I'm all for the controls. We could question whether it was a good idea to do diarrhea-free controls. But it was really a lovely case and control design. 

And the really important thing is that a sizable proportion of the patients got colonoscopy in an attempt to have some kind of reference standard for clinical disease. And they actually had really stringent inclusion criteria. We would almost never wait for six stools over 36 hours now. We have a lower threshold for diarrhea now. They also excluded patients with no other causes. And almost 2/3 of their patients got lower GI endoscopy. 

And what's interesting here, and the part I like to highlight is, if you look at these here, this is culture-positive, toxin-positive, culture-positive, toxin-negative, or negative by culture, toxin-positive, et cetera, et cetera, four categories like that. We basically see that the vast majority of patients in their cohort were positive for the organism, positive for the toxin, and that the rate of people who had an abnormal colonoscopy was 51% in that population. 

Now, with this very stringent criteria, excluding other causes, lots of diarrhea in these patients, they saw a few patients who were negative for toxins but positive by endoscopy. But I'd like to emphasize that the rate here was five times higher than that rate. So they saw a clinical difference between these two groups of patients. And here's their interpretation from that paper. 

They basically said that toxin sensitivity depends on your reference for disease. They said if you basically take anybody who has diarrhea and then has the organism and you say this is C. difficile-associated diarrhea, your sensitivity to the toxin assay is about 70%, suggesting you might be missing some. If you use colonoscopically-proven pseudomembranous colitis as your reference for disease, toxin assay actually does very well. And others have even shown that the sensitivity might be as high as 95%. But it really depends on how you define disease. 

The more important thing, they suggested, was they thought that majority of the cases diagnosed only by stool culture were very similar to their controls, many of whom also had asymptomatic colonization. And they actually dared to suggest that they thought that only one in five of these patients had real C. difficile disease, even back in 1986. And they really thought that most of the patients with real disease were toxin-positive. So this was early evidence supporting a different clinically between these. 

Now here's a four-hospital study from the UK published in 2013. And I really don't want to spend much time on this slide. But the main emphasis here is that they observed a mortality difference between the group that was toxin-positive and organism-positive versus a group that was toxin-negative and organism-positive versus the group that was negative in both categories. 

And you can see here that the all-cause mortality was not quite 17% in the toxin-positive group, about 9%, maybe 10% in the discordant group, and similar, not statistically significantly different from here, in the group that was negative by all causes. And it had a similar effect when you looked at the rate of death per patient days. 

So now, enter me. You can see that I'm up there in this little circle up here. I was also concerned about whether or not all of these toxic-negative, PCR-positive patients really had disease. And we thought that we needed to pause before we implemented a PCR test. And we actually needed to evaluate clinical outcomes. 

So I put together a bunch of eager beaver undergraduate students and some research assistants and things like this, including actually, a postdoc here at UVA right now who was one of the people that was instrumental in that study. We decided we were going to embark on a two-year journey. We ultimately studied almost 1,400 patients or so. And we actually only used toxin enzyme immunoassay clinically. This is what we'd done for years. 

Before we did this study, we did a five-year look-back at 6,000 patients, because we thought, well, if we're missing bunch of C. difficile patients, there ought to be a bunch of evidence of harm. So we went back and looked for patients that were getting harm with complicated C. difficile disease but negative toxin assays. And we couldn't find any. 

So we decided we're going to continue doing the toxin assay that we've been doing for literally 10 years at this point in time. And then we were going to do PCR and culture in the background, and that we would only report the toxin assay. And here's the breakdown of our patients. We had about 9% that were positive by both, about 11% that were negative for toxins but positive by PCR, and then about 80% that were negative by both assays. 

We looked at how many diarrheal stools they had for 24 hours or 48 hours. And we basically said there were about 15% who didn't meet a low bar threshold for diarrhea but basically about 84% had some diarrhea by some criteria, depending on whether you have a low bar for threshold for diarrhea or a historic higher bar. 

Then what we did, here's our data. This is a Kaplan-Meier curve. But we're really looking at how long and what proportion of patients continued to have diarrhea. So this is on the day of testing. You can see that virtually all patients, most patients had diarrhea, just like we were talking about, met a criteria for diarrhea on the day of testing. But then with subsequent days, the proportion of patients that still have diarrhea goes down, depending on how far out you are. 

But what's interesting is that, fairly rapidly, although it's similar up at the beginning, fairly rapidly, the toxin-positive, PCR-positive group starts to separate. And they have a higher proportion of patients that continue to be symptomatic after about the third day after testing. And although there's a little bit of separation here early on, the negative-positive discordant group, before long, is more or less entirely on top of the group that clearly doesn't have C. difficile by either of the tests that we looked at. 

And then this is really just looking at what's the odds in an individual patient of having diarrhea on subsequent days and really just putting numbers to this to say that there's a higher rate of diarrhea in the toxin-positive group on subsequent days. And mind you, this is despite treatment in those patients. 

We also looked at more serious outcomes that everybody is scared of. We looked at historic complications of Clostridium difficile. And we looked at death. And we actually did blinded chart reviews with adjudicators who actually had no knowledge of the PCR test results. And we basically had them adjudicate our complications, being ICU transfer, megacolon, or colectomy to say, yes or no, was this due to or possibly due to or related to C. difficile. 

We looked at all of our deaths. And then we basically combined these outcomes. And similar to what we observed previously in our retrospective study, we really saw that the vast majority of either of these was in the toxin-positive group. There was one patient who died in the toxin-negative group. And basically, their death might have been contributed to by C. difficile but not immediately attributed to or due to C. difficile. So we have one questionable patient. But the vast majority of them were in this toxin-positive group. So that was reassuring. 

I guess the other thing I should back up and point out is that the other thing that's interesting is that in both this and in the previous slide with the Kaplan-Meier, we actually saw surprisingly little difference between the patients who were completely negative by all assays and the negative-positive patients. So our interpretation of this was that toxin-negative, PCR-positive patients had a shorter duration of symptoms, few or no complications that we would attribute to C. difficile infection, and similar outcomes to the patients that were negative by all test methods. And in fact, this was despite the majority of them not getting treated. 

We had a subset of patients that got partial treatment but really only a few that got full treatment. So we thought that this was pretty strongly suggestive that maybe these people either don't need treatment or that they possibly don't have C. difficile infection at all, and they're just colonized with another cause of diarrhea. 

Now, here's a little bit more data supporting the potential clinical value and safety of toxin tests. And the reality is that there's sort of a slowly accumulating literature group suggesting that there are less symptoms and better outcomes in toxin-negative patients. It's not perfect, because there's some that showed different findings. But a sizable proportion do. 

Interestingly, in clinical trials of C. diff drugs, what's emerged is a story that the drugs actually have better efficacy in patients that are toxin-positive and lower efficacy or no evidence of effect, depending on your statistical power, in toxin-negative patients, suggesting that there may not be treatment-responsive disease in those patients, which is kind of interesting. 

And then the thing that to me has been fascinating-- and UVA has done some work in this area and others have-- is that the concentration, the burden of organism is a proxy for whether or not toxins are present. And people have started to say, well, if you've got more organism present, if you've got a higher burden, that's consistent with more complications, more severe disease. And really, what they're saying is that if toxins are present and you have a lower PCR crossing threshold or a higher burden, you have more severe disease and worse outcomes, which is basically the same as saying that if toxins are present, you have more severe disease and worse outcomes. 

And then what we also see, because I wanted to go back and look at the safety of this, is that when you really try and cull the literature and see what's the rate of true complicated toxin-negative C. difficile infection, it does occur. You can find cases. There's no doubt about it. But it's actually relatively uncommon. It's not as common as we might think it is. 

The highest rate here was from a recent study in Spain. We could quibble about the exact details. But they're the highest rate. Several centers had rates of zero. Pittsburgh, actually from their very lovely, long study, basically surveying all of their data, by my count, just had less than 1%. Similar data in Switzerland and in Cleveland. So relatively uncommon but does occur. 

Now I just want to acknowledge there are some conflicting studies. There's some case series that absolutely document that complicated C. diff can occasionally occur in toxin-negative patients. Some of these are very old studies. Some of these are more recent. 

Here's the classic Pittsburgh study that was one of the ones pointing to the 027 epidemic. And then there's a few studies that actually don't show a statistical difference in clinical signs or symptoms. And in a second, we'll talk about why might it be that some studies see a clinical difference and other studies don't. How could that possibly be the case? 

So let's talk about some causes of toxin-negative, PCR-positive results. So one thing I would suggest is that this is just colonization of a patient who really doesn't have clinically significant disease or symptoms. So this is basically inappropriate testing. And you guys have been addressing this here with diagnostic stewardship and the BPA that you've probably confronted and you may or may not have loved. 

Another option is that you've got somebody who is colonized, and they really do have diarrhea. But their diarrhea is due to something else. So they have an alternative cause. Could that be laxatives? Could that be other medications? Could they have received prior antibiotic treatment? 

So it turns out that prior antibiotic treatment, either for C. difficile or not for C. difficile, like metronidazole for some other kind of indication, can result in a false negative assay. So that also occurs. And then there's been some question over time about whether or not you could actually have the toxin break down before you analyzed it in the lab. And that would be a cause of false negative results. 

So here's my stab at trying to explain why is there so much confusion in this area. Why have we been fighting about this for so long? Why do we not agree on this? One of the issues is that there's really no reference test, no true reference standard for clinical disease. And we often compare our tests with one reference standard or the other. But none of these really is an adequate reference or gold standard for clinical disease, and that's because the signs and symptoms of CDI are non-specific. And the reality is that, as we talked earlier, diarrhea and, colonization, both vastly more common than actual C. difficile infection. So it's really tricky to say what's true C. difficile disease. 

And then we've had this anecdotal negative experience which, quite frankly, has scared people. And this has caused us to really not feel comfortable trusting toxin assays. We've seen occasional patients where, when they test negative by toxins once, and then we test them again later, and they would suddenly become positive. And we'd go, oh, that stupid toxin assay, I'm not trusting that assay again. I'm going to test everybody three times. This is going to be the last time I'm going to be affected by that. 

And this fueled this desire for absolute sensitivity. And we stopped worrying so much about clinical specificity. We didn't want to get burned. 

And then we had a lot of laboratory studies without clinical correlation that promoted the belief that all toxin-negative patients really had CDI, regardless of what was going on clinically. And these all, I think, exaggerated the clinical insensitivity of toxin tests. 

This is my attempt at trying to say why do some of these studies not see clinical differences between these patients. I think there's a variety of methodologic and statistical reasons for this. Some of it is just inadequate power. They didn't have very many complicated cases. In some cases, their study design was honestly biased. They had anchoring bias in some of these cases. Or they misattributed their symptoms to C. difficile, because they knew of the PCR results. 

Sometimes they didn't try to look for other causes of clinical symptoms. And very few studies actually had control groups. So there's lots of issues. 

And then I guess probably the biggest one is this challenge that there's shared risk factors between infection, colonization, and non-C. difficile diarrhea. So it's really hard to tease these out. 

Now, I just want to talk about how C. diff diagnostic testing turns out to impact many aspects of care. Some of this is because of the last few years. So it obviously impacts things at the level of the patient, diagnosis, treatment. I've tried to make an argument that it impacts outcomes. But the reality is it's super important for infection prevention, because we use it for isolation. We worry about nosocomial transmission and outbreak detection. 

One interesting paradox is that as we switched to PCR tests, we've actually missed some norovirus outbreaks, because we were falsely thinking that patients actually had C. difficile infection, and we weren't looking for norovirus. So this actually goes both ways. You can worry about missing a C. difficile outbreak, but you can also miss other causes of illness, because you're convinced that the patient has a C. difficile infection. 

And then we've really started to see problems with public reporting and, of course, value-based care. It also affects antimicrobial stewardship. There's actually some evidence that inappropriate treatment can increase the rates of other multidrug-resistant organisms. And most recently, this affects perceptions of hospital quality and reimbursement, because the feds have basically put a really big stick behind our rates of C. difficile infection. 

So impact of C. difficile testing a little more, it kind of goes without saying that if you're not testing patients, you can have under-diagnosis. This is one here. This is the early paper, 1986, saying you can have false negative test results that can lead to missing cases. Here's one from Europe saying, if you don't test patients and you miss ones who actually have disease, you can have underdiagnosis. 

So yes, underdiagnosis occurs, but there's also overdiagnosis. This is what I'm emphasizing. And this is a really lovely case report that came out with my paper that I think is really illustrative of this. So I just want to walk through it. 

The basic idea, this is a 70-year-old female with metastatic renal cell carcinoma, multiple episodes of PCR-positive diarrhea over a two-year period. We often think about patients having recurrent C. difficile disease, multiple episodes over a substantial period of time. This person had multiple episodes of diarrhea, always PCR-positive, for two years. This sounds problematic. 

They presented again, similar to previous times, 12 to 15 non-bloody water or bowel movements per day, abdominal cramps, afebrile, slightly elevated white blood cell count, not dramatic but a little bit. They were started immediately on PO vancomycin and metronidazole for recurrent C. difficile infection. We know what this is, no problem. This patient has had this multiple times before. 

Day six, an astute clinician recognized, hey, the diarrhea doesn't seem to be getting any better. It doesn't appear to be responding to this double treatment of PO vancomycin and metronidazole. It's relatively unchanged. So they actually took a step back and re-evaluated the diagnosis. They've got a pharmacist involved. And they started looking at the patient's medication list. So we're pretty astute and thoughtful. 

Chart review showed that prior treatment courses also had minimal effect on diarrhea. Prior tests for other pathogens had often been negative. Abdominal CT scan and showed a mild colitis but really wasn't that suspicious of anything significant going on. And they actually had a colonoscopy recently before this. It was non-diagnostic and negative for pseudomembranes. 

They then reviewed their medication history. And they recognized that the onset of the diarrhea started very close in proximity right after a monoclonal antibody for this renal cell carcinoma was actually started clinically. So this axitinib was prescribed for renal cell carcinoma. And diarrhea is a common adverse effect of this. 

So they had a conversation with the oncologist. And they actually were willing to stop this monoclonal antibody. And they also stopped the antibiotics for C. difficile. And they decided, let's watch this patient. And the patient showed dramatic improvement within the next three days. And their diagnosis was that this was medication-induced diarrhea with persistent C. difficile colonization that had been misdiagnosed and inappropriately treated for two years. 

A few more consequences of overdiagnosis, antibiotics are not benign. As we start to study the microbiota, we recognize this. You can also get MDROs. And there's even a little signal out there that sepsis rates may be increased in patients who have received recent antibiotics. So if you get antibiotics for, say, community-acquired pneumonia, your rate of sepsis in the follow-up period may actually be increased. And this may be due to some disruption of the gut microbiota axis. 

Misdiagnosis, we talked about this, resulting in delays of recognition of true cause of illness or outbreaks. We just gave an example. 

And then this has really been the big stick that's caused a lot of people to wake up. And I'm not happy about that. None of us are happy about this. But a few years ago, Centers for Medicare and Medicaid Services added hospital onset Clostridium difficile infection as one of the metrics that they use to set reimbursement penalties for hospitals. And hospitals starting to get penalized for these excess cases. 

Now, the CDC actually has had an adjustment for which type of test method you use. But this is data from a couple of years ago from University of Iowa, basically suggesting that the CDC's adjustment for enzyme immunoassay versus PCR testing was inadequate to correct for that difference. And they basically suggested that their rate of C. difficile events that were reported to NHSN and their resulting normalized SIR calculation were different depending on the test method and not corrected for by the risk adjustment calculation, hence their title, "Failure of Risk Adjustment by Test Method for C. difficile Laboratory-Identified Event Reporting." And the problem here is this might not be that big a deal if it wasn't tied to value-based performance and pay for performance reimbursement that caused institutions to get millions of dollars in penalties. 

This is just a survey of a few hospitals I worked with in California demonstrating a similar difference between these HO-CDI events depending on what test method you use, PCR versus toxin EIA. And now what I want to do, this is kind of my disclaimer slide, because I'm not trying to encourage anybody to switch test methods simply to try and decrease their NHSN rates. I don't want to advocate for that. It's really, for me, about what's the most accurate way of diagnosing patients clinically. 

The CDC actually attempted to address this issue. They redid their risk adjustment. And they re-baselined their calculation in 2015. So it appears to be much better at this point in time, since that Iowa publication. So that is actually an improvement. And the reality is that the effective test type varies between facilities. And like I say, I really don't think that trying to reduce your C. difficile rates by switching your test method is an acceptable idea. And it's probably unethical. 

So what are the best diagnostic strategies? We really have two approaches, one of them recommended by the Europeans, and one of them more recommended recently by IDSA. We'll talk about these in a second. They have slightly different approaches. IDSA actually gave two options, one of them was to basically say you could use nucleic acid amplification testing if you have diagnostic stewardship criteria to minimize inappropriate testing. 

And this is what's done currently here at UVA. So you have diagnostic stewardship to try and minimize inappropriate testing. And then you use the more sensitive tests. The other thing they suggested was that if you can't minimize inappropriate testing, you could use a multi-test algorithm similar to what the Europeans recommend. 

So I'm going to start with the European model. This is what's known as the 2-step with different reporting or the multi-test with toxin. So this is out of those European guidelines. They say, start out at the top with a highly sensitive test. And this could be like our enzyme immunoassay for glutamate dehydrogenase or just detecting the DNA for the organism. 

So let's say we start with PCR or NAAT testing up here. If it's positive, you then go to a more specific test, like a good, high-performing toxin EIA. There's some toxin EIAs that are less good than others. And you use that to help you discriminate, depending on what's the likelihood of C. difficile. 

So if both of these are positive, they say, look, C. difficile is likely. Go ahead and treat the patient. If the result is negative, they don't say you fully excluded C. difficile infection. But what I like about this is they basically say C. difficile is less likely. Carriage is quite possible. And they do a really radical thing. They actually suggest, why don't you evaluate the patient. Why don't you look at them and see whether or not they have severe C. difficile infection? 

So you start with their eye sensitivity test. You rule out C. diff. You then have a high-specificity of the test, kind of like HIV testing. And then the radical one at the end, they say, put your thinking cap on. Go back to your patients. If they have severe disease, you're really concerned they've got severe C. difficile infection, go ahead and treat them. And if they don't, maybe just wait a little while. Maybe stop your laxatives. Use some watchful waiting. See whether or not they appear to really have C. difficile disease or if it gets better. So evaluate them before treating. I like this approach. 

Now, the other thing here, the option is what the IDSA is recommending. And they say, well, perhaps NAAT that is an acceptable alternative by itself if you can actually decrease some of this unnecessary testing and inappropriate testing. So here's some data from Stanford. And many others have observed the same thing, basically suggesting that at baseline, if we're left to our own design, we have a tendency to test lots of patients who maybe are unlikely to have Clostridium difficile disease. 

So this with data from Stanford showing that 17% of patients tested didn't have diarrhea at all. 50% of them received laxatives or enemas. And that seems a little bit curious. And then there was really just 33% that hadn't been getting laxatives and enemas and also had clinically significant diarrhea, so suggesting that there was a huge burden of inappropriate testing. 

And what they did at Stanford that was really cute is they basically said, let's attempt to decrease this one and that one there. We're going to actually put a report in place. We're going to force our nurses-- ask our nurses gently. We're going to work with them in a multidisciplinary way to start documenting all the patients' bowel movements and output. And then we're going to run a report in the laboratory and say, has this patient had diarrhea. Are they having diarrhea? And have they received laxatives? 

And if they don't have diarrhea or have received laxatives, we're going to cancel testing. And the doc has to call us on the phone and give us a good reason as to why they want testing to proceed. So kind of an aggressive intervention, but it really was trying to get rid of these here. And they saw a sizable decrease both in test perform-- not surprising, because they were blocking them-- and also in their rate of C. difficile cases. 

And they're now following up these interventions and looking at outcomes. And things seem to be working OK. 

So now, this one here, I want to basically give thanks to actually Drs. Madden and Sifri here. This is borrowed from a review paper that they published in Diagnosis just this year looking at diagnostic stewardship interventions. So this concept that, well, maybe you can just still use PCR testing if you can just decrease inappropriate testing has caused a lot of people to say, can I have my cake and eat it too. Can I have my sensitive test? Can I decrease my inappropriate testing with diagnostic stewardship? And a lot of people have been investigating this with exciting results. 

What's interesting here from this table from Madden et al. in this year is that everybody's kind of all over the map. Everybody's trying different approaches. And you can see, there's lots of institutions doing this, all within the last couple of years in the wake of value-based purchasing penalties for C. difficile infection. But the interventions are really kind of all over the map. There's no standardized intervention. 

Some of them are using this physician order entry changes to discourage repeats and testing without indication. Here's, again, physician order entry criteria requiring approval and monitoring and feedback. This, I actually suggest, is one of the strongest ones here. 

I know these people at UC Irvine. They actually did several things. They said you had to have diarrhea, you couldn't have laxatives, you couldn't be doing it in kids. And if there was another cause of diarrhea, you also couldn't be doing it. And they actually were monitoring. And if they found that people were inappropriately ordering, they actually first gave the doc a call on the phone. And they said, come on, my friend, are you sure you really want to test for C. difficile. And if that didn't discourage anybody, they actually then wrote them a letter that was signed by the chief medical officer, basically saying, cease or desist. You're doing patient harm and institutional harm. 

So I view that as a very strong intervention. And they saw a 56% decrease in effect. 

Now, other institutions have had less strong interventions and have not seen a significant decrease in test ordering or CDI rates. Let's see. Oh, here's UVA right up at the top. You guys have experienced a remarkable effect here with a relatively gentle intervention. But that's that. 

So now here's my question. So just asking the question of does the IDSA recommendation that if you've got better testing, you can still keep your NAAT and avoid overdiagnosis, does that stand up to scrutiny? This has really not been studied in a careful, systematic way yet. But I went back to my data from my JAMA paper. And I basically, because we had the data for whether or not they had laxatives, we had the data for what their bowel output was and whether or not they had diarrhea, we said, well, let's see if we artificially remove those patients who didn't have diarrhea, artificially removed the patients who had gotten laxatives, does the evidence of overdiagnosis go away. 

Here's the original one with all comers. Let them test anybody they want. Here's the one if we limit to patients with a three loose stool per 24-hour diarrhea criteria. You can see it maybe gets a little bit better. But the overdiagnosis is really, for the most part, still there. This looks very similar to that. 

If you block out all the laxative ones, the shape of it starts to change a little bit. It maybe gets better again. But the point is, there's still a sizable difference between these, suggesting that there's still a difference in clinical outcomes and recovery rate between these two groups. So there's probably still overdiagnosis going there. 

So I think reducing inappropriate testing is clearly a good idea. I love it. I support it. I'm actually working on it as well. I'm just suggesting it may not solve all of our ills. 

Other emerging options. This is actually data from 2012 where we basically showed this relationship, again, between toxin-positivity and organism burden versus toxin-negativity with lower organism burden. And we derived one potential cutoff based on an ROC curve that we thought would separate these. But what you can see is that these types of approaches with modifying your cutoff for positivity, they, of course, are not perfect. And they still have some overlap. 

Now, the folks at Stanford, Niaz Banaei et al., they've actually taken this one step further. Rather than just recognizing this opportunity, they actually said, let's go ahead and derive our cutoff. And then let's use this PCR cycle threshold to actually predict toxin-positivity. And if it's basically low like this, they call these CT toxins-negative. And they report those as a negative result. If it's up here, they report those as CT toxins-positive. And they report that as a positive result for toxins, even though they never did a toxin test. So this is kind of an alternate strategy without having to do the second test. 

The other emerging options here, there's several groups, several companies working on ultra-sensitive toxin testing. I should go ahead and make a disclaimer here that this figure specifically is for a company Singulex that I happen to consult for, but I'm not specifically promoting this test. This is from a publication that I had no relationship with. 

But the idea here is that, could we do a better using a toxin test if we had one that was more sensitive. And the idea with this group was that they were trying to peg it to that traditional cell cytotoxicity assay. 

So the first thing here, this is basically the concentrations of toxins on the x-axis. And then here's the cycle threshold with PCR, again, like that. And what you can see again, in the same way, with PCR, lower crossing thresholds equals more organism burden, higher organism burden. Higher crossing threshold is actually lower organism burden. That's because it takes more PCR amplification to detect a positive result. So this is actually less organism. This is more organism. 

And you can see similarly that the majority of the toxin-positive patients have a higher burden of organism or a lower PCR cycle threshold. And the ones that have a lower concentration of toxins have less organism burden and a higher PCR crossing threshold. 

And so here's where this company set their cutoff. And then what I want to illustrate is that if you look at a conventional toxin enzyme immunoassay like the one we used in our study, they typically have a threshold limit of detection that's about here. So you could say they're missing these patients plus those patients. But the question is, do these patients have clinically significant disease. Do these people need treatment? What's going on with them? 

And then here's the overlay of the conventional cytotoxin assay. And the data's a little bit different. But the point is that this company and their new ultra-sensitive assay, they're trying to be inside this bar as to what their limit of detection is. But it remains to be seen if detecting these additional patients is a good thing or if it will cause overdiagnosis similar to the PCR story. 

Now I just want to give a shout out to some people here at UVA. In preparation for my talk, I was impressed to see that Greg Madden and Dr. Sifri actually have published quite a few very lovely works in this area, a couple of nice reviews. They're also looking at PCR crossing threshold. 

And this is really cute, I thought. This is a figure from one of their papers, basically suggesting that when you start to look at the PCR crossing thresholds of what would be deemed inappropriate test orders versus appropriate test orders, that there's also a little evidence that the PCR cycle threshold, meaning the burden of organisms, differs between those two patient populations. So additional evidence that those tests were probably inappropriate and not really indicated clinically and that it was less likely that that was a clinical disease. And I understand that they're during follow-up work to look at outcomes in those patients. So you guys are doing a lot of great stuff here. 

My windup to this is that more work is needed. This is from the IDSA. The reality is that the testing criteria that we use are actually based on very low quality evidence, the reality there, weak recommendation. Similarly, the recommendation for test methods is still based on very low quality evidence, including some of my studies, et cetera. There's no RCTs in this area, et cetera. So this is all very low quality evidence. 

And the reality is that diagnostic stewardship interventions were really just in the early days. And there's no evidence here either. And there might even be a little bit of signal that some of these, if not well thought out, could in fact cause patient harm. So we do need to look at the safety of these in the same way as we look at the safety elsewhere. 

And I think I'm out of time. But that's it. Here's my takeaway messages. Recent studies show the perils of indiscriminate use of clinically non-specific tests. We have a couple options for optimal testing strategies. Diagnostic stewardship can absolutely improve utilization diagnosis and outcomes. But more work is needed. And that's it. 

[APPLAUSE] 

Questions for Dr. Polage? 

[INAUDIBLE] 

I think there is. This still remains an outstanding question. Probably the thing that we use the most for that is that there definitely is higher transmission and certainly a higher rate of recurrence in those. And there was a fidaxomicin trial, suggesting that there might be a difference depending on the way you treated them in your risk of subsequent recurrence. 

So there probably is a role for that. It's still a little bit debatable. And not everybody pays attention to it. Colleen? 

So we discussed this already. But I just wanted to make one point of clarification [INAUDIBLE], the PCR NAAT test is not the same as the [INAUDIBLE] or just routine microculture, where all you're identifying is the [INAUDIBLE] coordinates of it. The NAAT test does indentify [INAUDIBLE], it's the additional tests that tell you whether you can detect toxin in that [INAUDIBLE]. So that's a clarification that's important to me, because I don't want them to just group one of two ways-- [INAUDIBLE] 

Sure. 

And so do you think that at least [INAUDIBLE] whether or not toxin production is [INAUDIBLE] 

Right. That's a very important point. The point there being that the PCR tests do specifically detect the gene that codes for toxin production. So it does demonstrate that you've got an organism that has the gene and is theoretically capable of producing toxin. What it doesn't do is say whether or not toxin is being actively produced at the moment or whether there's clinical disease at the moment. Sir? 

[INAUDIBLE] 

Yeah. 

[INAUDIBLE] and then the question is, you obviously did a study where [INAUDIBLE] 

That's the thing. Again, I did that review paper, in part because I wanted to educate myself. And I also wanted to put information out there into the literature, because the funny story that that came from my experience was the clinicians at UC Davis. 

We just had this incredibly limited differential, which it was like if we had diarrhea in the hospital, we thought, oh, it's got to be C. diff, test for C. diff. If the C. diff test came back negative, we still thought it must be C. diff. Just test them again. If the test came back negative again, we'd say, oh, that stupid toxin text. I hate this thing. I'm going to treat him anyway, empirically, for C. diff. 

It just never enters our mind that there might be some other cause of diarrhea. So I think having a broader differential is absolutely important. I think one of the challenges is that the majority of diarrhea in the hospital, or at least a sizable proportion, is related to medications and underlying illness. And I like to also joke that we don't have tests for those things. 

So the only way we can really diagnose those is if we're comfortable ruling out infection. And if our knee-jerk response is, the second we find a pathogen, we say that must be it, we're not comfortable ruling out infection or at least asking ourselves, could it be something else. And if we don't go back to the medications and say, could it be this, or could it be either underlying illness or other things like that, their 2B's, their lactulose, all of these things. 

Sir? 

I'm just wondering does the Clostridium bacteria only have one kind of toxin, either A or B, or could it be [INAUDIBLE] other kinds of toxins which [INAUDIBLE] 

Sure. The question is, does Clostridium difficile only have these toxins, toxin A and B, or are there other toxins that the toxin EIA is missing. And this is an astute question. 

Depending on the strain, for the most part, toxins A and B really are the predominant ones produced. And most strains only have those. But in recent years, this 027 strain, which became the epidemic strain, it did have a third toxin, which was this binary toxin. 

To be honest, it's been a little bit debated as to what the clinical significance was of that. There actually has been one report suggesting that binary toxin strains might be possible of causing disease without the A and B, suggesting that that alone would be sufficient. But I think that's really questionable. So I think the role of the binary toxin relative to these others in terms of those things is still very much up for debate. Yes, sir? 

You mentioned [INAUDIBLE] 

Yeah. 

[INAUDIBLE] positive right there is a lot there. 

Mm-hmm. 

Are clinical labs usually [INAUDIBLE] they got a result that says positive. But they hacked into the system and don't report that [INAUDIBLE] research protocol. Are they following these patients that are [INAUDIBLE] negative but [INAUDIBLE] 

Yeah. So I would say several things. The first question, are most clinical labs comfortable doing these kinds of things. I would say the blanket statement is no. It really depends on the comfort level of your individual laboratory and their directors or leadership. Some laboratories are more comfortable making these kinds of decisions. And others are honestly very uncomfortable, and possibly fearful, of making these decisions. 

Secondly, relative to the other pieces, this was all done at their institution under the umbrella of performance improvement. And they have legitimate reasons for that. But they are doing additional research analysis. But they basically validated the cut off first. Under performance improvement, they implemented it for patient care. And now they're just watching to basically make sure it's safe. Ma'am? 

Two questions. One is, what do think that the role of either a stool for white cells [INAUDIBLE] is in terms of determining who actually may have an inflammatory colitis that could be due to C. difficile if there's nothing else there? And then the second one is, what do you think is the importance of terminology, watery stool versus-- well, there's diarrhea, which could be, for some patients, six or more stools per day. But water stool versus unformed stool or mucoid stool-- the Bristol Stool Scale-- 

Right. So these are both great questions. The first one has to do with whether or not there's a role for measuring inflammation directly in feces or in the gut. And I actually do think that there's a role for this. But we still have to kind of figure out the best way to do it. 

I really do think the third leg of the stool is going to be to look at post-response. The question is just what's the best way. One of the challenges is that in the hospital, a large proportion of patients-- and we saw this in my studies. They do have some level of inflammation in their feces or in their gut. And so if you use a sensitive, qualitative assay, you get yourself into some of the same trouble, where it becomes wildly indiscriminate. 

What we did see in my work-- and I think that this is a potential pathway. Is that if you look at what the level is of inflammation-- and there's possibly other markers that might be more interesting. We could talk about what those would be. You can see differences. 

And so when we look at quantitation, how much inflammation was present, we could see that a lot of patients, even patients who didn't have C. diff at all-- a lot of patients would have a little bit of inflammation, because they're just sick. They're in the hospital. They got stuff going on. 

But it was really only a subset of patients, and almost entirely toxin-positive-- not exclusively, but almost entirely toxin-positives that had really elevated levels of inflammation. There's other things, inflammatory bowel disease, it's non-specific. So there's other things that can cause it, other causes of colitis. So all of these things are non-specific. 

The second question, relative to what do we think about scoring or paying attention to the consistency of stool relative to the number of bowel movements produced. I want to make sure I say the traditional saying from my GI med school days, that it's both volume and frequency. But at any rate, the Bristol Stool Chart, I think, a lot of people are moving towards. But my impression is that the Bristol Stool Chart was originally developed primarily to focus on constipation and doesn't do a very good job of categorizing diarrhea. 

There's a little bit of distance in the literature as to which of those we classify as diarrhea. Do you accept five, six, and seven? Do you accept only six and seven? Does it have to be watery? 

Laboratories will often use a criteria of just, does it conform to the shape of the container. But this is problematic too. It would seem really black and white. But nurses will sometimes smear poop on the sidewall. They got a tongue depressor. They scoop some poop. They smear it on the sidewall. And the tech in the lab looks at it and says, well, it looks to me like it's adhering to the sidewall and conforms to the shape of the container. But it could easily be smearing formed stool on the inside of the container. So that's an imperfect-- 

When you study poop, you get kind of into these things. Yeah. 

Sir? 

So if the folks who are PCR-positive and toxin-negative [INAUDIBLE] PCR-negative, toxin-negative, why aren't we just doing toxin [INAUDIBLE]? Why would we have a high-sensitivity [INAUDIBLE] 

So I would just say, my easy response is that despite my confidence in the data, this is an extremely controversial area. And I would acknowledge that there are legitimate, bona fide cases that have been reported and observed, mostly at an anecdotal level, of patients who had clear complicated C. diff infection, but they were negative for toxin tests. 

What frustrates me sometimes is that people often are not thorough enough to go back and say, did they get antibiotics prior to that. What was going on with this patient that might explain that? 

And honestly, these cases are rare. They are very rare. So I think what happens then is that those cases, true complicated patients who have fulminate C. diff disease that test negative for C. difficile toxins-- we could talk about why that occurs. But those are rare as hen's teeth. 

And then the challenge is that all the more common patients, those clinical symptoms are very non-specific. And it's hard to tell which patient has diarrhea due to another cause versus diarrhea due to C. difficile, because they don't have obvious colitis. Their colon is not expanded. It's not thickened. There's no thumbprinting, all of those kinds of things. 

So it's sort of like dealer's choice. Do you want to believe this is C. difficile, or do you want to say it might not be C. difficile and be something else? So we've spent years fighting about this. More safety data is coming. 

[INAUDIBLE] delayed who fall and break their hip [INAUDIBLE]. 

Yeah. Yeah, I'll admit, I'm actually doing a five hospital project right now that I'm finishing up in California where we're looking at this issue. So I think more data is coming. And one of the things that we have been really remiss at not doing is that we have not really evaluated, are there ever negative consequences of overtreating and overdiagnosing. We tend to undervalue, underappreciate the impact of overdiagnosing or overtreating. And we only focus on the missed case. 

Ma'am? 

[INAUDIBLE] I just have a questin of what is your ideal isolation [INAUDIBLE] for PCR-positive but toxin-negative [INAUDIBLE] 

Yes. 

[INAUDIBLE] let's say they're even PCR-positive, [INAUDIBLE] but even one person [INAUDIBLE] 

Yep. 

And you know, [INAUDIBLE] most of the infection [INAUDIBLE] 

Mm-hmm. 

[INAUDIBLE] 

Yeah, no. This is obviously a very interesting area. And there's a lot of movement and discussion in this area. There's really two things going on. One of them is that you're absolutely right. Your patient population has got a much higher rate of diarrhea due to medication and treatment. But at the same time, they also have increased risk for C. difficile colonization, multiple health care exposures, lots of antibiotic exposure, even just the dysbiosis due to their cytotoxic chemotherapy, their monoclonal antibiotic therapy, their radiation, all of the above. So it's kind of a perfect storm, double whammy type of situation. 

And interestingly, because their microbiota is disrupted, their C. difficile will overgrow. So it will overgrowth. And they may even be toxin-positive. So it's not just as simple as, oh, the PCR test is no good, and toxin would solve all our ills. It's not that simple. It's just that this is a complicated patient population with a high rate of treatment-related diarrhea and a high rate of colonization. 

Back to your question of should we isolate the patients who we don't think have C. diff disease or that are toxic-negative. I think we probably should. I've done some work, and some others have done some work suggesting that the toxin-negative but PCR-positive patients probably contribute less to transmission than the toxin-positive, PCR-positive ones. But they don't contribute not at all to transmission. That's maybe not good grammar. 

But the point is that they can cause some transmission, maybe not as much as the other ones. But we want to be safe. And in a group like yours, it probably is an issue. And with as much colonization as you have, I would absolutely do that. 

And of course, the other thing that I don't really want to talk about but the people are obviously interested in is whether or not there might be a role for prophylaxis, which is a good way to spend a ton of money and mess up people's microbiota, but might have a short-term benefit. 

Good. Well, we'll have to hold questions there. Thanks so much for your time. 

Sure. 

[APPLAUSE]