Skip to content
Munroe & Gershengorn – Peripheral Vasopressor Administration episode artwork

Jan 31, 2024 46 min

Munroe & Gershengorn – Peripheral Vasopressor Administration

from Maryland CC Project

Elizabeth Munroe, MD, MSc is a clinical instructor and post-doctoral research fellow in the dvision of Pulmonary and Critical Care Medicine at the University of Michigan. Hayley Gershengorn, MD, FCCM, […]

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

Munroe & Gershengorn – Peripheral Vasopressor Administration

0:00 46:23
of matches

Transcript Auto-generated, may contain errors

that aren't necessarily prepared for that influx. You also have issues with understaffing, which can exacerbate strain. Then you've got regional and other sort of crises of political and natural crises that refugees and migrants are fleeing that can change the influx and demographics of patients. And then you have things like natural disasters, which unfortunately we are not going to be avoiding, you know, insistently in the future, clearly. So there are many potential contributors to these strains.

And this really can apply not just to ICUs, but any ward. They described it in the analogy of a balloon where you have demands on resources, including the patients who are already in the ICU with varying levels of acuity, the patients who then also need ICU level care, and then the supply being a balloon of things like resources of practitioners that are the nurses, docs, respiratory therapists in the ICU. You have fixed resources like a number of beds, the CRT machines, vent machines, and then you have the actual process efficiency of how those practitioners and resources are able to be organized and mobilized. How efficient those things are organized and mobilized also can reflect sort of the, if you think about the balloon as being more or less stretchy, its ability to...

be compliant, we'll say, to those resource needs. And some ICUs in some hospitals may be more or less compliant to adapt to those needs. These supply domains that we think about can also be thought of in the way that Barbish and Koning had described in 2006 as the elements of surge capacity being staff, stuff, and space, and system, the four S's of surge capacity. And these are really all the same domains, just with different names. And so the question is, what happens when that balloon bursts to adapt and...

be completed in their level of severity of illness, can also influence processes of care in the ICU, like prophylactic care or influencing the rate of adverse events. You also have changes to physician structure and nurse-to-patient ratios and other staffing models that, again, influence those processes of care and outcomes, all potentially contributing to an increased risk of mortality in the setting of strain. So how do you measure something like capacity strain? So classically, it's most commonly measured as census. So you can just have a census that's an unadjusted number of patients in the ICU beds. You can have the proportion of beds that are filled and occupancy, but not all census is created equal. So you can have census that is adjusted for the number of new admissions, the turnover of ICU cases, acuity adjusted, how sick are the patients, what are their actual sickness needs in terms of resources? Like, are they all on ventilators? Are they all on CRRT machines? So census can be adjusted to really capture not only the number of patients, but the acuity and the workload.

There are other ways you can measure a capacity strain. However, you can measure the workload of how many medications need to be administered by respiratory therapists or nurses. Other things like transporting off wards. There are lots of different processes that can increase the workload and resources needed to treat the patients on that unit. So another study from George Nisi, actually the same one that I showed you that graph from previously, he tried to really understand whether capacity strain influences whether patients are triage to the ICU or the floor. And so he focused on patients with sepsis in the emergency room whose disposition in the hospital was going to be discretionary, meaning that they didn't have to go to the ICU. So he looked at patients who weren't on vasopressors, non-invasive or invasive mechanical ventilation, who plausibly could go to the ICU or the ward based on what the team's assessment of their need was. So this way you could actually see whether capacity strain was likely to influence that decision making.

And he found looking at ICU occupancy as a measure of capacity strain, that when the ICU was relatively less full, less occupied, patients with this sort of discretionary level of sepsis were more likely to go to the ICU. And not unexpectedly, when the ICU was pretty full, they were more likely to be triaged to the ward. So then the question is, did that translate into differential outcomes for those patients? Did it make a difference that they were triaged differently? So again, looking at that same group of patients with sepsis that could go either way, when he looked at the patients who were triaged to the ICU, patients who were admitted to the hospital under levels of low ICU occupancy strain, who were more likely to go to the ICU, they ended up with lower mortality, whereas the patients who under high occupancy, high ICU strain, ended up being triaged to the ward had higher mortality.

And there are other outcomes that have been shown to be affected by high capacity strain. This is another study from another one of my colleagues in our research group, Rachel Cohn, and she focused on 30-day readmission. And she looked at patients who had survived an ICU stay and were transitioned out to the ward and then looked at the factors on that ward where they stayed after their ICU admission to see what potential measures of capacity strain might be associated with their risk of 30-day readmission. And she looked at a number of different patient-specific and ward-specific measures of acuity and potential contributors to strain. And she found some interesting things that a lot of the ward-specific measures of strain were actually highly associated with the risk of 30-day readmission. So things like if the ICU survivor is on a ward where a lot of medications are having to be administered, in other words, there's a high nurse workload, those patients had a higher risk of 30-day readmission. So showing that the workload of a given unit can affect longer-term outcomes for these patients is clearly having some impact on their care received.

So what about during COVID? That was the extreme example of capacity strain. And a number of people, as you might imagine, tried to look at whether capacity strain poorly impacted patient outcomes. This study came out of Australia, and I thought it was very interesting. They created this activity index, which is sort of a composite measure. It's basically census adjusted for severity and workload. So you can see on the numerator here, you have the number of beds that have a high nursing workload, like patients on mechanical ventilation or renal replacement therapy, for example, relative to the total number of staffed ICU beds. And overall, a higher activity index score means the ICU is very busy.

And they looked at the mean activity index score in a given ICU for the first week that a patient was admitted to the ICU and then looked at whether that was associated with mortality. They looked at the first two years of pandemic data, and they adjusted for a number of patient factors, including their chronic comorbidities, their admission diagnosis, severity of illness, demographics, frailty, and a bunch of other hospital factors. And they found that, yes, in fact, a busier ICU had higher mortality. So if a patient was in the very first week of their admission in a very busy ICU, they were more likely to die than the patients who were in their first week of admission in a less busy ICU. So what about in sepsis? Does this also apply? I will say that I'm going to talk more in depth about the processes of care for hospital onset sepsis. There is some data in what we call community onset sepsis, and this is all based on ED care. And this study from Pelton and his group in Intermountain looked at patients with sepsis in the ED, quantified how long it took for them to get antibiotics from when they arrived or to antibiotic time. And then they separated out the...

patients by ED occupancy rates, so basically how busy and how full was the ED when the patient was first seen. And you can see that the time to antibiotics on the x-axis was longer for patients in this fourth quartile of very busy EDs compared to shorter time to treatment for patients who were in the first quartile of less busy EDs. So what was interesting is that they broke down a number of care processes that have to happen before the patient gets antibiotics to see whether each of these sub-processes was affected by that level of occupancy in the ED. And they did find that this first step of ED arrival to being placed in a room was affected by increasing ED occupancy. So this quartile four of highest occupancy, those patients were less likely in those hours to be roomed. Same thing with their first time to evaluation.

The time from their clinical assessment to actual diagnostic data collection also was impeded by high occupancy busy ED times. But the time from diagnostic data collection to antibiotic initiation did not seem to be impeded by ED crowding, which to me tells me that the processes leading up to the decision-making were affected, but the decision-making itself was not particularly subject to a change based on ED crowding. These first three processes are pretty specific to ED care. And so the question really becomes, how does capacity strain affect ED? inpatient antibiotic initiation, which is subject to a really different set of processes. Which processes are affected on the inpatient side? And do we need to really think about that differently for patients with sepsis on the inpatient side from the community onset group in the ED? So that leads us to introducing what hospital onset sepsis is and how it's different from community onset sepsis and how we can try to study it.

So community onset cases really represent the majority of sepsis cases, and these are patients who present with sepsis on admission. Patients who develop sepsis in the hospital at least two days after admission comprise a minority of patients, but they are really not the same. They've been presumed to be the same thing, but I will explain to you how and why they're different. Most research has really focused on the community onset group because it is a larger group, and I will show you why in a second. I think it's also just easier to study.

So this group really is unique from community-onset sepsis. They differ in heart failure, renal disease. They're less likely to receive timely care and timing goals. They also experience higher rates of complication like respiratory failure and shock. They have twice the ICU and hospital length of stay and incur two to three times higher cost of hospitalization overall. And they experience two to three times higher rates of mortality than patients with community-onset sepsis. And in adjusted analyses looking at patients who are admitted without sepsis, if you develop sepsis in the hospital compared to not, you have an increased risk of dying that's threefold.

Despite all of this high risk and documented delays in care and different outcomes, this group really remains understudied and observational research and has largely been excluded from clinical trials and sepsis because trials are just a little bit easier to enroll in the ED. A lot of my work has focused on this population as a particularly vulnerable group where some of the decision-making and processes of care may be particularly at risk under a capacity strain. So there are a few reasons why the inpatient sepsis patient is different from the emergency room. So the emergency room is really a place where patients who come in with sepsis often will have a rapid decline and a rapid escalation of care. They often come in sort of fully formed in the sense that they already have a presentation that has developed enough to prompt their presentation to the ED. They will have a bowl of data all collected at once. And this is in a place that's really designed for rapid triage of undifferentiated patients. Whereas on the inpatient side,

Some patients with hospital onset sepsis will have a rapid decline and rapid escalation of care, but many will have more of an incremental clinical decline on top of something else that already brought them into the hospital. Their presentation is really evolving in front of your eyes in real time. And so there is sort of a dwindling period where things are potentially unclear with more diagnostic uncertainty as to what exactly is happening. And data is not collected at the same rate on the inpatient side as it is in the ED. You're getting routine labs in the morning unless some clinical change has prompted you to get new labs. And this is a space that's really designed for ongoing care of patients with established problems and established diagnoses.

So on the inpatient side, we're a little bit more subject to cognitive errors like anchoring bias where we lock into the initial diagnosis and we're not intentionally but cognitively reluctant to adjust our approach and assessment and our thinking in the face of new data as it starts to come through. We also are subject to search satisfying where you stop a search because you've found an initial result and you don't keep looking. There's also something called vertical line failure where you really fail to think outside the box, outside of the bounds of what you already know about the patient. And in hospital onset sepsis, I think traditionally, as I said, it's been approached as the same thing as community onset because we've had this one size fits all approach to care, whereby all of the patients are nails and we have a hammer. But there's really this growing recognition of heterogeneity in sepsis, especially in the distinction between community onset and hospital onset populations, as well as potential phenotypes within those groups. But it's prompted calls for more precision approaches to care. And the idea is that different patients and different circumstances might need different tools for treatment.

So my research has really been framed by that idea, really trying to ask, can outcomes in this high-risk group be improved by better targeting our interventions to the right patients at the right time, which really requires that we identify the patients who are high-risk and those care contexts that are most vulnerable to care delays and care gaps and poor outcomes. So I told you that one of the reasons this group is understudied is because it's kind of hard to study. There's really no gold standard to define sepsis onset or time zero of clinical sepsis. And that's challenging in any sepsis case, but particularly on the inpatient side. Most of the community onset studies use somewhat of a process measure to decide when sort of the clock starts ticking for these patients. So often, like that study I showed you with antibiotic delays in the ED, they use the time that they arrive, their time of triage, as a standardized time that you can apply to all patients as when their disease started.

But in hospital onset sepsis, on the inpatient side, there's no standard approach to that because there's no equivalent entry time into observation. And so you really then have to think about clinically, when did the disease start? And we just don't have a clear answer for that yet. So I have, in an attempt to try to sort out that problem, implemented clinical case criteria that are evidence-based to identify cases. And then I'll show you how I've looked at that to look at care timing. So I'll walk you through the criteria for identifying patients who meet SPSA-3 criteria for research. So first, when you're looking back at admission, you find a time period during the admission when the patient has orders that look like the clinicians were worried that they had infection. So SPSA-3 defines this as a blood culture being drawn and new antibiotics being started.

And then within a certain window around that, you look to see if that patient also had organ dysfunction. And then they've met all the criteria of having an infection and organ dysfunction that's plausibly attributable to that infection. And this graphic is adapted from the CDC criteria to identify patients for epidemiologic surveillance. And you can see how you might apply this to an example patient where you've met the first criteria. This patient on day six has a blood culture drawn. You look for a window period around that. You see that they're started on antimicrobials within that window period. They've met criteria for presumed infection. You also see that they developed organ dysfunction in that window. They've now met all three criteria. And the first of those criteria to occur is what is considered as

time zero. So historically, that's been applied to a day of sepsis onset for epidemiology. And what I've done is use that as a time of sepsis onset. Now, I realize that that is sepsis onset. The first of these criteria to occur is somewhat of a restruct, but we can think of it like an EHR signature of sepsis onset. That is, like with community onset, a standard approach that you're applying to all patients that allows you to examine relationships relative to that time point. And it's a window really into what the clinicians have seen, the earliest time point that they have data that would suggest the patient is septic.

All right, so now I will give you sort of an overview of what we've done to look at how capacity strain and other sort of variations of it may relate to antimicrobial timing for hospital onset sepsis. So I'll give you actually first a little bit of context. A lot of hospitals in trying to deal with the patients who have sepsis in the hospital have tried to turn to automated alerts to try to detect sepsis in the hospital and improve the rapidity of recognition and treatment. And my research is actually motivated by one such failed attempt. And this is the value of negative studies. So I, a good number of years ago now, was involved in implementing and analyzing this predictive early warning system alert. It was based on a machine learning algorithm to predict the development of sepsis and septic shock in the hospital. And it had decent...

positive likelihood ratio and positive predictive value. And it had a median of about five hours of alerting prior to sepsis onset. And we paired this alert with a message that was sent to the clinical team to let them know their patient was high risk. And ultimately, we really found no change in meaningful processes of care and no change in the outcomes for the patients. And we had also paired this with a survey study to understand how clinicians used and perceived the tool. And basically, they hated it. They said it didn't change any of their decision making. It very rarely flagged patients who they weren't already worried about. It wasn't very good at finding the patients that they needed help with.

So that really brought me back to this question of how do we identify the patients who are high risk and the times that would be particularly vulnerable to lapses in care where clinicians might need help with some intervention. So in order to look at the patients who develop sepsis in the hospital and understand the care that they've received and then be able to try to identify those high risk patients, we really had to work with granular data to understand the processes in real time. So along with one of my mentors, Gary Weissman, we've put together this really granular data set that we call CERCI. And it contains basically every single inpatient admission to five different hospitals in our health system. And we have encounter-level data for these patients, including demographics, admission and discharge times, mortality, ICD codes. But we also have hour-level data that really gets at the nitty-gritty of the care that was received. So we have data for timestamps of orders and care received for all of their workup and their treatments. We also have their location at any given time, basically every single hour of their admission.

So this allows us to then try to find cases clinically that meet these criteria and then look at their care received. So from that data set, I have pulled out a cohort of patients who meet either the sepsis 3 or adult sepsis event criteria for hospital onset sepsis. So their onset has to happen at least 48 hours after admission, onset of the infection and the organ dysfunction. And we've identified over 5,000 patients who meet these criteria. And this is the cohort that I will show you a few studies from.

So with that hourly data, then we can look at the entire period from admission leading up to sepsis onset, all of the details of the patient's presentation, their workup, their clinical manifestations, but also the system and process factors that we're leading up to and at the time of sepsis onset. And then we can look at their timing of treatment initiation relative to what we've labeled as their EHR signature of onset, and then see what their inpatient complications and outcomes are subsequent to that. So I'll give you a snapshot of the timing estimates and some of the outcomes for this cohort. So first, for the Haas cohort that met adult sepsis event criteria, these patients developed sepsis on a median of a week into their hospital stay. They were administered antimicrobials about four hours after onset, and there was a pretty high mortality rate of 19%, and it got even higher among the patients whose sepsis was complicated by subsequent respiratory failure. Almost 40% of those patients died. The sepsis 3 cohort had a very similar timing of sepsis in their hospital stay, almost about a week in.

And here we actually broke down the time from onset to the order being placed and then from the order to administration of the medication for a median time to treatment of about 10 hours, which is longer but not unexpected because sepsis 3 criteria identifies a slightly less sick population of patients with sepsis, as we can see with a lower mortality rate of 15%. And again, a very high mortality rate even among this group if they developed respiratory failure. So then with this cohort, we went to ask three questions. First, are antibiotics for hospital onset sepsis delayed under times of high levels of capacity strain? And are antibiotic delays more likely at certain times of day that have variation in resource availability? And then ultimately, do those delays harm patients? So these first two questions, looking at antimicrobial timing under capacity strain and the time of day, the rationale for this is that things like clinical decision fatigue or competing patient care obligations, staffing ratios, and other hospital processes will all vary by level of resource strain and time of day. And you may have varying levels of risk of cognitive error in that context. And all of that may impact the timeliness of treatment.

So this first question, looking at capacity strain, we wanted to know if timing of antimicrobial initiation was associated with ward capacity strain. And we measured that by unit census at the hour of sepsis onset. And we were looking at ward patients here. So the unit census, we measured every single hour on the ward. And then we looked at the hour in which the patient met sepsis onset criteria and just looked to see if that level of census was, you know, and this was all standardized relative to that ward's experience over multiple years. So how far out of the norm is it for that ward that is associated with how quickly they got antibiotics.

And so we measure that outcome in the number of hours from sepsis onset to receipt of the antimicrobials. This was all adjusted for hospital factors, seasonal factors, and patient severity wellness factors. And we did find that indeed with higher levels of capacity strain, patients were slower to receive antimicrobial initiation. So on the x-axis here on this graph, you have standardized ward census. So how far off of the mean ward census, which is zero in the middle, how far above or how below, how far below is the census at the time of sepsis onset for that given ward. And you can see that as census increases from left to right, time to antimicrobial initiation on the y-axis also goes up.

On the extremes of low census, antimicrobials were initiated for these patients at a median of 3.6 hours, and at the extreme highs of census, it was as long as 6.8 hours. Ultimately, we quantified that for every three additional patients on a ward, the time to starting antibiotics increased by one hour. We did a little bit more flexible modeling in a secondary analysis, and we found that this relationship was really most marked at the extreme high of census, which is not entirely surprising, but I'll just say that the extreme lows of census weren't necessarily associated with very rapid antibiotics. It was more so that highs of census were associated with delays.

So this next question, looking at the time of day. So again, we looked at exposures at the time of sepsis onset. So we looked at two things, actually. One was the hour of day in general. And then the second was the hour of day at the time that a patient became septic. And I'll explain why.

we looked at two different outcomes. The first outcome was the probability of starting antibiotics at that given hour for any patient with hospital onset sepsis, accounting for how long they've been septic and all the other things. At any given hour of day, what is the likelihood that a clinician will start antimicrobials for hospital onset sepsis? And then the second question was for a given patient who gets sick at that time, what is their time to treatment? Which are two slightly different questions.

And so for this first question, we did see that, yes, the likelihood of treatment initiation for hospital onset sepsis varied by time of day. And we saw a couple interesting things. So here on the x-axis is time of day starting at 7 a.m., the start of day shift, going through 6 a.m., the end of night shift. So you have day shift on the left, night shift on the right. And you can see two interesting things. We have a nadir, basically a lower likelihood of starting antimicrobials at 7 a.m. and 7 p.m., which is change. And night shift were all associated with increased time to antimicrobial initiation.

And really, to me, this identifies some potential context that we were looking for, which care contexts and times are particularly vulnerable for treatment delays. More work really needs to be done to understand what the mechanisms are for these delays so that we can actually target the care improvements. Why is it that at night, patients who get sick at night are less likely to get antibiotics quickly? That is a question that still remains to be determined. We also saw that those delays were associated with increased odds of respiratory failure and death. And again, we saw some interesting things that the median time to requiring intubation was almost 30 hours. And it suggests that there is a good window of opportunity after the patient really is first starting to get sick to try to improve our care delivery and prevent those poor outcomes.

So then the million dollar question, what can we actually do about this? Well, a group out of Colorado actually asked a bunch of academic centers what their experience is with strain. And they found three main themes about this challenge. The first is that it's complicated and it's really hard to predict, which makes it very difficult to mobilize the resources that are needed in time to actually respond. The second important theme was that what people really found work best to mitigate resource strain is resources like people and time.

But often the solutions that hospital systems deploy end up being the less expensive solutions that end up not being as helpful and potentially have unanticipated negative consequences, like detracting from actual delivery of care. And then lastly, unfortunately, and it's not news, I think, to most people is that there's just going to be conflict that arises from that tension between the lack of sufficient resources, the resources that you need, that ineffective tactics that are being deployed and your inability to do your job because of all of those things ends up having a very negative impact on the workforce. They asked these centers what kinds of interventions had been deployed in their centers. And these had some with more, some with less, most with less success. I won't go through all of these in detail, but I wanted to highlight a few that I think had some interesting sort of implications. One is what's called level loading. So we're essentially trying to balance the capacity strain by admitting across teams in a balanced way to try to balance out the census so that you don't have surges in census on one team and then an ebb of census on another, which I know is a...

can be logistically challenging for teams ending up admitting basically every day, but is, I think, an interesting and novel workflow that has some potential. There's a lot of talk of surge coverage, backup calls, bringing in moonlighters, EPPs, things like that. The trouble is that with that lack of predictability for surges and capacity strain, bringing in, understanding those staffing concerns and then being able to respond on time, all of that usually lags and it's very difficult to time it well. And then there's also a lot of financial challenges to being overstaffed and understaffed. Things like a complex discharge team sound really helpful to help with discharges from the floor that end up taking a lot of time, but it also comes at a decent high cost to the health system to employ people who focus on that. Unit-based teams are always a great idea because there's a lot of time lost and some good literature to show how teams who have patients on quote-unquote ectopic geographic locations, those patients receive less good care and it's very inefficient, but it's really difficult to maintain geographic-based teams when you are at bed capacity strain. It's really hard to move patients around to where they need to go.

There's something called a discharge lounge where patients who are ready to go and just waiting for paperwork or meds or things like that can go hang out. That sounded like a very intriguing idea that I'd never heard of before, but it turns out all of these academic centers had a very negative experience with it. Essentially, the nurses and the docs had very little buy-in because they were all concerned about the level of acuity of the patient. They didn't want them sitting in a waiting room. They wanted to keep an eye on them until they really left.

Also discharge huddles. I think a lot of hospitals have really implemented multidisciplinary discharge huddles. Also discharge goals like discharge two before two or discharge before noon, things like that. It turns out that that was pretty universally disliked because it ended up detracting from actual delivery of care because people were checking their phones, their emails had to meet for a multidisciplinary meeting that they didn't find to be particularly efficient or effective. But multidisciplinary communication platforms, some sort of a text platform where you can get the nurse, the doc, the pharmacist, the social worker all-in-one text thread was actually very helpful and efficiently enacting discharge plans.

And then lastly, there's something called an ED-based triageist, which we have recently enacted at Penn. And this person is sort of ours, at least as a hospitalist, who's embedded in the ED and helps with bed flow, things like that, but also, unfortunately, is an expensive intervention. And what is interesting is that basically none of these centers felt like they had adequately tackled the problem and that they had really good solutions. So unfortunately, as I said, I don't really have a silver bullet. But I think that we do need to start thinking about capacity strain as on a continuum of strain. There are usual variations to strain. There are also seasonal variations and then in the more extreme crises and catastrophic.

levels of capacity strain. And this, you know, we don't have to go through this in detail, but this chart is a really good summary of the different S's of resources, systems-based stuff and stuff that would need to be sort of triaged differently under these different levels. And if we start to think about it on this continuum, then I think we can start to learn a little bit more from those extreme cases like during the pandemic that really unmask our weaknesses. And I thought this was a really nice chart. You're not gonna be able to read most of it, but certainly a reference to look up. It's a group out of Harvard that developed this tool for hospital leaders and clinical leaders

It's called the Hospital Surgeon Preparedness Response Index, which is similar to something I'll present to you in just a second, but not exactly the same. But it essentially is a tool to link action items to strain triggers so that you, for each of these domains of staff, stuff, space, and supplies, have actual triggers to say, okay, we've exceeded our capacity by these metrics, and these are the action items of next steps. And the idea is that you can anticipate these system needs and figure out how you're going to be able to flex and expand your balloon of capacity. And hospitals, I think, over the last 20 years, certainly since 9-11 and through a number of disasters since, have increased their preparedness. So this comes from one of your own. David developed this hospital medical surge preparedness index. Sounds very similar to the last one, but this is an index of the score that essentially can quantify how prepared a hospital is for surge capacity needs. And it quantifies things in those four S's to give a total score. And you can see, as they studied in 2021, over time in this graph on the left.

Hospitals of different sizes of beds, these two lines represent hospitals with different numbers of beds, over time from 2005 to 2014 have increased in their preparedness score. They are improving. But in this map of the U.S., you can see that that increase was not universal. Some states having a darker color here shows that they've had a greater increase in their preparedness score and others less so. So there still remains a lot of variation across regions and hospitals.

So one of the issues is that there are lots of different contributors and potential solutions to strain, and people don't necessarily agree on those. So this study was actually done by one of my mentors, Mita Carlin, a while ago now, a good 10 years ago. But she did a six-month cohort study where she had qualitative interviews with ICU charge nurses and physicians in ICUs to ask them about their perceived level of strain on a given day in the ICU. And you can see this chart that I have in the middle that their score on a given day didn't correlate all that well. There was really only moderate correlation in their perceived level of strain.

And what's interesting is that they also differed in what they perceived as contributing to strain. So the charge nurses felt like the high strain days were when ICU census was strained and there was high ICU acuity of patients and when the ward census was high. Whereas physicians really, it was just about bed availability, just about the census in the ICU. That's when they felt strain. And they also differed in what they perceived as the potential resource needs in response to that strain. So the nurses really saw a need for nurse-based

care and personnel, they needed more ICU nurses. Whereas physicians on high days of strain felt like they needed more beds, more residents and more advanced practice providers, which, you know, it's not entirely surprising that these different groups had different perceptions of what causes strain and what they need because they have different roles and in those roles, potentially different priorities. But I think it speaks to the subjectivity of strain that different team members will feel strain differently and really the need for a multidisciplinary approach to handle it across those different mechanisms of strain and to be able to address it for all the team members. So if you think about it, something that might mitigate strain for physicians like opening up new ICU beds may actually worsen strain for nurses if the nurse staffing for those ICU beds isn't there and it's inadequate. And so, you know, I think we think often about strain as a bed question and there is this ever increasing number of ICU beds across the United States. But I think this is a good example to show that ICU beds don't necessarily address all the different sources of strains for the different providers within the team.

So how do we put this all together? I think the way that I think about it is the idea is that you would ideally want to identify and target the specific mechanisms of care gaps and learn from what is being unmasked during pandemics and other extreme examples of strain. So where does the care fall through? Where do the gaps develop under those high levels of strain? It'll show us the weaknesses in our processes. And the idea is to really identify those times, the units, the places, the time of day, the context in which delays and lapses and errors in care are more likely and use those also to identify potential disparities in care.

does strain exacerbate our cognitive biases and other existing disparities that are pre-hospital in having differential outcomes for different groups of patients. And this really is going to need a more granular analysis of the mechanisms to understand if it's not just a question of giving us more beds, what is the actual gap and need that needs to be filled? And ideally, we would align disease-specific initiatives like a sepsis-QI group's initiatives with those system initiatives for surge preparedness. I'll give you just two examples to think through. I don't have specific solutions, but ideas. So if you're thinking about surge preparedness, you want to identify the patients who are actually harmed by the delays, target those specific mechanisms, learn from the gaps that are unmasked under capacity strain, and potentially use capacity strain as a trigger to enact those action items. So for example, if you've identified that your hospital onset patients with septic shock are particularly high risk of poor outcomes, or the patients in the ED who have sepsis and a discretionary disposition are subject to differential triage under high capacity strain,

Or patients with high diagnostic uncertainty who you're worried are more likely to have errors because of cognitive bias under times of strain. Those are the patients who you may want to target more specifically or surge preparedness as opposed to this blanket hammer and nail approach. Your care specific mechanisms say that you identify that. antibiotic delays are happening on a unit, particularly at times when medication burden is very high. Perhaps you have a resource nurse for the hospital who's deployed to specific units when a certain threshold of medication administration capacity strain has been reached so that they can go offload medication administration for the floor nurses and help reduce delays in antimicrobial administration, for example. Say that you discover that very laborious discharges are actually one of the mechanisms of antimicrobial or other delays. Perhaps you employ a social worker and pharmacy remote support to be activated at certain times of high strain.

Perhaps you have ED triage criteria that really target those patients that you've identified are more likely to have poor triage and poor outcomes during high capacity strain. And then if we think about early warning systems, which I told you are frequently failed, those also need to learn from these capacity strain gaps. The idea is flag the high risk patients and potentially use capacity strain to guide your thresholds. So for patients with hospital onset sepsis who are particularly subject to poor outcomes under strain. Potentially patients with a high-risk phenotype that's discovered or presentations that are difficult diagnostically, they're hypothermic or they're euthermic, they don't have the classic 102 fever. Patients with cancer who are particularly high-risk and have multiple reasons to have SIRS physiology. And then think about the context in which you want your early morning system intervention to be deployed. Is it more important to have a low threshold to alert at night because we know that care gaps happen at night? Perhaps during shift change, your system may deploy some arm to actually enact care during shift change when people are busy.

Using severity adjusted census, for example, to help understand and trigger and adjust those thresholds, perhaps you want to have a different threshold during viral season. The idea really is, I'll go back one, to learn from these gaps and try to be a little bit more specific in the interventions to target those mechanisms. And I will thank a number of my collaborators and mentors at the Paris Center and at Penn broadly, some of whom I presented their work, Dr. Weissman, Kohn, and Anisi, who may still be on the line. I've learned immensely from them, and they've really been sort of publishing seminal things in the field. So go look at a lot of their work. And I'm curious to see if we have any time, what interventions you guys looked familiar to you, what you've tried, what has worked, what has not, and your general thoughts on this very big challenge that I don't necessarily have a solution to.

Jen, this work was really enlightening. And I think there are definitely moments of it where it's sort of scary. Like, how are we going to survive and how is healthcare going to survive? And how do we fix these problems that I feel like are sometimes so much bigger than us or individual hospitals? But also reassuring that at two different hospitals, two different big systems, we're experiencing, I think, a lot of the same pinch points and crises and issues that you are. I've certainly opened it up for anyone else who wants to chime in with ideas or suggestions or things that they found have worked and not worked. I think we do also have some early warning sign kind of system set up. I think that most people probably completely ignore it. A score is generated. The patients are like color-coded based on if they're like red, yellow, or green.

and the most crisis level. When you ask people how the score is generated, it seems like it's not even clear what variables go into it and the variables kind of change at someone's discretion. And I think I had heard at some point that our rapid response team would like call the house staff and say, hey, your patient on the floor is getting this early warning system score or whatever. And they were like, thanks, but no thanks. You know, it sort of offended them in some ways and they felt like I'm standing in the patient's room and they look just fine. So I think we're good. You know, there's a very big discrepancy.

So I think we've seen a lot of those same things. I'm curious if you can comment some on when you looked at your data and you're really focusing on your ward patients and this in-hospital sepsis, did you divide up or look differently at the wards, meaning like a cancer center ward versus like a telemetry ward? And also like, were there differences in the medicine wards versus the surgical wards? Because I always worry with sepsis.

Beyond Hope Project with Jason Tharp Beyond Hope Project Welcome to 'Beyond Hope Project,' a beacon of inspiration in the podcast world. Join Jason Tharp as we embark on a transformative journey, exploring stories that awaken the will to start anew and find the strength to rise from setbacks. 'Beyond Hope Project' is about lighting the way to your starting line, not dictating your path. We delve into narratives that foster optimism and belief in a future where dreams become achievements. Each episode is a tapestry of positivity, enthusiasm, and comforting support, resonating with the essence of hope. This podcast goes beyond storytelling; it's a movement of positive change rooted in authenticity and transparency. We provide real tools for real journeys, infusing more hope and joy into the world. 'Beyond Hope Project' stands as a testament to the transformative power of hope and the limitless potential of the human spirit. Join us in transcending boundaries and breaking barriers. Here, we prove dai Veteran Chat Project Kyle Turner Our goal is to reach as many Veterans as we can. We noticed a lot of the podcast personalities tend to be unrelatable to the every day joe. Not taking anything away from anyone just hoping to give more people a place to feel connected to something. KZ and Kyle are Army vets who deployed together for OIF 07-09 and OND. Now close to 15 years later we are finding a way to open up about our issues, and find a way to build a community. Paramedic Philanthropy Project Melissa Ade Podcast by Melissa Ade Bigfoot Information Project Podcast Bigfoot Information Project Occasional conversations with leading researchers and investigators regarding the animal commonly referred to as bigfoot.

Frequently asked questions

How long is this episode of Maryland CC Project?

This episode is 46 minutes long.

When was this Maryland CC Project episode published?

This episode was published on January 31, 2024.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this Maryland CC Project episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.