AT THE BEGINNING of her sepsis update at this year’s Society of Hospital Medicine conference, intensivist Patricia Kritek, MD, EdM, threw up what she called “the obligatory slide” on the high costs of sepsis in the U.S. The disease affects more than 1.5 million American patients a year and costs more than $20 billion.
But Dr. Kritek, a professor at the University of Washington in Seattle and the associate dean for faculty development, called out one bullet point in particular: Sepsis mortality ranges between 20% and 50%.
“That reflects our difficulty in saying what sepsis is,” she pointed out. That difficulty in turn fuels “our desire to come up with better definitions so we can put the right people in the right bucket and call it sepsis.” The problem: Just about every facet of sepsis care is controversial or not yet clear, from how to define and screen for the condition to the value of various treatment bundles.
“We understand sepsis better,” Dr. Kritek pointed out, “but we don’t fully understand it.”
Even the definition of sepsis keeps changing, Dr. Kritek said. The most familiar one—known as Sepsis-1—is nearly 30 years old and defines sepsis as an infection plus at least two of four systematic inflammatory response syndromes (SIRS). But “we were concerned that SIRS produced too many false positives,” she noted.
“We understand sepsis better, but we don’t fully understand it.”
~ Patricia Kritek, MD, EdM
University of Washington
The Sepsis-3 definition published in the Feb. 23, 2016, issue of JAMA defines sepsis as life-threatening organ dysfunction caused by a dysreglated host response to infection. Under Sepsis-1, by contrast, organ dysfunction was part of the definition of severe sepsis.
“So now, we no longer recognize severe sepsis, only sepsis and septic shock—except in the CMS bundle, which we’ll talk about later,” she said. Here’s the new Sepsis-3 definition of septic shock: Sepsis with persistent hypotension requiring vasopressors to maintain a mean arterial pressure (MAP) of more than 65 mm/Hg and a serum lactate of more than 2 mmol/L despite adequate volume resuscitation.
“With a MAP and a lactate,” said Dr. Kritek, “this definition is more specific.”
SOFA and qSOFA
Because organ dysfunction is part of the new sepsis definition, the defining body directed doctors to gauge organ dysfunction according to patients’ SOFA (Sequential Organ Failure Assessment) score. SOFA looks at six different systems, scoring each with points from 0 to 4. The problem, Dr. Kritek said: “This is not nimble. You don’t have all these data”—the patient’s creatinine, for instance, or bilirubin—”when you’re trying to figure out if the patient is septic.”
That’s why the expert body published a paper detailing qSOFA (for quick SOFA score). “What fell out was two of three criteria: tachypneic with a respiratory rate of 22 or higher, altered mental status as measured by a Glasgow Coma Score of 13 or less, or a systolic blood pressure of less than 100,” she explained. “This is not a screening tool but one for prognostication. Two out of three correlate with a bad prognosis, with patients more likely to die or spend a long time in the ICU.”
Many large studies, Dr. Kritek said, have gone on to try to validate qSOFA’s prognostic ability, including two in the Jan. 17, 2017, issue of JAMA. The first looked at qSOFA in the ICU, “and I think the answer was ‘no,’ qSOFA didn’t work particularly well in the ICU,” Dr. Kritek said.
But qSOFA’s predictive value proved better in the study that looked at the ED. Emergency patients with a qSOFA of 2 or more had 24% mortality vs. 3% for those who didn’t have that finding. “This is the message: qSOFA is not effective in the ICU but potentially good outside the ICU to predict who will do poorly,” Dr. Kritek said.
Which scoring system works?
However, knowing whether or not someone will do poorly is not the same as early identification, Dr. Kritek pointed out. When qSOFA came out, many people started talking about using it for screening.
“But some doctors, like me, wanted more data before adopting qSOFA,” she said. “Many physicians worried that waiting until patients are hypotensive and have altered mental status is too late to identify sepsis.” Plus, “many places invested time and energy into screening protocols based on SIRS, and they’ve decided to stick with those.”
At the same time, “we’re seeing a bit of growth” in people using MEWS (for modified early warning score). Those systems (a similar one in the U.K. is called NEWS for national early warning score) look a lot like qSOFA and SIRS criteria, “but they include the use of supplemental oxygen as one of their parameters.” Further, these tools were designed to suss out who in acute care would need to go to the ICU.
“They work relatively well for that,” she noted, “but they weren’t designed for the ED and they weren’t focused on sepsis.”
Two big studies—one in the April 1, 2017, American Journal of Respiratory and Critical Care Medicine, the other in the June 2018 issue of The BMJ Emergency Medical Journal—compared qSOFA to both SIRS and MEWS/NEWS to detect clinical deterioration or predict in-house mortality and ICU admission.
“They found that SIRS has great sensitivity but horrible specificity,” Dr. Kritek said. “There are a lot of reasons patients can have SIRS that aren’t sepsis.” qSOFA, on the other hand, was much more specific, “but its sensitivity is much lower, and that’s concerning if we think we’re going to miss some people with sepsis.” MEWS/NEWS fell somewhere in between.
“No perfect answer”
The Surviving Sepsis Campaign’s recommends that hospitals and health systems have a performance improvement program for sepsis, one that includes screening for high-risk, acutely ill patients.
“But they don’t tell you how to do that, and that’s intentional,” she pointed out. “There is no perfect answer for screening.” What is true is that you need a low-enough bar to get all the patients with sepsis, and SIRS plus infection does that.
“But then we also need a way to say, ‘These are the people to be worried about,’ and maybe that’s MEWS or a combination of SIRS and qSOFA. I honestly think we haven’t found the right tool yet.” But “if your hospital is already invested in a system that people understand and use, it’s probably good to stick with it and tweak it over time as opposed to re-educating people to use a totally different tool.”
Principles of early treatment
For many years, sepsis treatment was driven by a 2001 study on early goal-directed therapy (EGDT), “a small study that had an amazing impact on our care,” Dr. Kritek pointed out. However, three big trials—the ProCESS, ARISE and ProMISE studies—all found EGDT to be no better than standard care, as set out in a meta-analysis in the June 8, 2017, New England Journal of Medicine. “We have edged away from it, which I think is appropriate.”
At the same time, patients in those three trials all got early antibiotics and fluid resuscitation before they were enrolled. “So our usual care has changed a lot,” she noted, “and I don’t think we want to lose these really important parts of intervening early that have become part of usual care.”
That evolution, Dr. Kritek added, is reflected in the Surviving Sepsis guidelines, and the campaign’s recommendations on antibiotics are “probably its least controversial section.” Among its many recommendations on antibiotics, Dr. Kritek highlighted two strong ones backed by moderate evidence.
The first: Administer IV antimicrobials as soon as possible after admission and within one hour for both sepsis and septic shock. “Certainly, in septic shock, the quicker the better,” she said. “But in sepsis, we need to be on the stewardship side of antimicrobials, and I don’t necessarily think you have to give IV antibiotics all the time—although a first dose of IV antibiotics isn’t a bad idea, followed by PO.”
The second recommendation is to use broad-spectrum therapy with one or more antimicrobials to cover all possible pathogens. “Give broad-spectrum antibiotics right up front,” she said, “then be equally rigorous in paring those back, once we know either what the bug or the syndrome is. We can’t keep giving antibiotics just because we’re a little worried.”
Fluid of choice?
Dr. Kritek then turned to the topic of early aggressive resuscitation. Polling the audience on their resuscitation fluid of choice, about half indicated normal saline.
But an equal number raised their hands for Lactated Ringer’s. “I wish I could take a picture of this, a bunch of medicine docs raising their hands for Lactated Ringer’s,” she said. Very few indicated they used PlasmaLyte or albumin, and no one chose hetastarch.
The Surviving Sepsis guidelines include a strong recommendation (with low quality of evidence) for giving patients with sepsis-induced hypoperfusion at least 30 mL/kg of IV crystalloid within the first three hours.
“There’s a lot of debate about whether every single patient needs 30 mL/kg of fluid, and I don’t think we have evidence that that’s the right amount,” Dr. Kritek said. “But I do think the vast majority of patients can tolerate that, even those with end-stage renal disease, although probably not people with cardiogenic shock.”
The guidelines also—and this is a strong recommendation with a moderate level of evidence—recommend crystalloid, “but they don’t say what crystalloid to use.” She actively advises against hetastarch because of its increased risk of renal injury. And while albumin wouldn’t be a wrong choice, “it wouldn’t be a cost-effective one either.”
That leaves normal saline, Lactated Ringer’s and PlasmaLyte. “If I’d asked five years ago, I wouldn’t have seen half of you raise your hand for Lactated Ringer’s, and that shows a big evolution in the last few years,” she said. In large part, that change is due to the 2018 SMART trial, which randomized ICU patients to either saline or balanced crystalloids (Lactated Ringer’s or PlasmaLyte, although the vast majority of those in that arm received Lactated Ringer’s. “This is effectively a comparison of saline to LR.”)
Among several outcomes including in-hospital death, researchers found balanced crystalloids bested saline in terms of fewer major adverse kidney events at 30 days. “It was small but statistically significant,” Dr. Kritek said. “That’s going to point me toward using LR—but not so much that I wouldn’t use a bag of saline if it was right in front of me.”
Two large randomized trials are now looking at just this issue, she added. “Maybe metabolic acidosis or the hyperchloremia of lots of saline causes harm,” she said, noting that in a subgroup analysis of the SMART trial, Lactated Ringer’s “had a stronger signal” for patients who received large volume resuscitation and those with sepsis. “I think there’s something there in term of hyperchloremia not being good, but we need more to be convinced.”
Then there are the CMS SEP-1 bundles, issued in 2015. (There’s one for severe sepsis, “which we call ‘sepsis’ now,” Dr. Kritek pointed out, as well as one for septic shock.)
Each of the two bundles lays out what clinicians should be doing within three and six hours. Included in the three-hour sepsis bundle, for instance, is measuring lactate, getting blood cultures before administering antibiotics, administering antibiotics and then redoing a lactate if the initial one is elevated more than 2. For shock, the three-hour bundle includes those same three actions as well as resuscitation with 30mL/kg of crystalloids.
Within six hours for shock patients, you’re supposed to start vasopressors as well as reassess patients’ volume status—”and there is no gold standard to do that, so I think that’s really hard to mandate,” she noted. And while “I’m sure many of you are very focused on these bundles in your institution, there’s mixed evidence about their value.”
That said, Dr. Kritek noted that New York state has mandated the use of bundles for early sepsis identification and treatment in hospitals since 2013. A 2017 analysis found that New York hospitals have increased their bundle compliance (from 41.5% to 55.2%) and decreased sepsis mortality from 30.2% to 25.4%.
“This says,” she summed up, “that bundles help.”
However, a study published by Critical Care Medicine in October 2018 looked at compliance with the CMS’ SEP-1 bundles across seven hospitals; out of 851 cases, only 33% completed the bundle. While hospitals had higher sepsis mortality when they didn’t complete the bundles (18.4% vs. 11.0%), “that difference went away when the authors adjusted for severity of illness, and that’s important,” Dr. Kritek said. “The only single statistically significant factor in terms of the difference between the two groups was delay in antibiotics.”
How preventable is sepsis mortality?
Researchers have also explored this question: Even when clinicians do everything right in terms of bundle compliance, is death from sepsis preventable? Writing in JAMA Network Open in February this year, researchers looked at sepsis deaths in six hospitals and found that well under 10% were either definitely or moderately likely to be preventable.
“So people are dying of sepsis, but maybe they’re not dying because we’re treating their sepsis poorly,” Dr. Kritek pointed out. “There’s a lot of debate going on now about how we implement bundles. We’re getting a signal from New York that we’re paying attention to, but we don’t know whether all aspects of the bundle are the solution.”
In 2018, the Surviving Sepsis Campaign announced that it “was going to double down on bundles, and have you do it all in one hour,” Dr. Kritek said. “And there was incredible backlash, with the Society of Critical Care Medicine publicly saying, ‘Ignore this guideline.’ They were worried about giving too many people antibiotics and large fluid volumes.”
As a result of that pushback, the campaign changed the wording about its one-hour bundle on its Web site to say that it “should be viewed as a quality improvement opportunity moving toward an ideal state.” As Dr. Kritek pointed out, “There’s a lot of controversy about pushing too hard.”
Steroids and vitamin C
AMONG A HOST of topics covered during her sepsis update at this year’s Society of Hospital Medicine conference, intensivist Patricia Kritek, MD, EdM, who’s with Seattle’s University of Washington, spoke about steroid use in refractory sepsis.
Two trials on that topic were published in the March 1, 2018, New England Journal of Medicine. The first, from Australian researchers, “was larger, had more surgical patients, and randomized patients with septic shock to either hydrocortisone or placebo,” Dr. Kritek explained. The second trial, which was done in France, had more medical patients and treated them with both hydrocortisone and fludrocortisone vs. placebo.
The Australian trial found no difference in mortality, although patients randomized to hydrocortisone got off vasopressors earlier. But the French trial “did show a difference in mortality. Patients also got off pressors earlier, and there was less organ dysfunction overall,” she pointed out. A meta-analysis that included data from both trials was published in the Dec. 21, 2018, JAMA Internal Medicine.
That analysis found that steroids conferred a mortality benefit at 28 days but not at 90, “so that doesn’t really answer our question about whether steroids work to decrease mortality,” Dr. Kritek noted. Here’s her own practice with patients who have refractory shock: “I put them on steroids, hydrocortisone 30 mgs q6, because I think getting off pressors earlier is a meaningful outcome. Maybe there’s a mortality benefit.”
And while she personally doesn’t use fludrocortisone, “if you want to be evidence-based, the trial with fludrocortisone showed a mortality benefit,” she said. “If you want to be physiologically based, you may feel that fludrocortisone is insignificant in the setting of that much hydrocortisone. It could go either way.”
Dr. Kritek also mentioned a single-center study that appeared in the June 2017 Chest in which everyone who came into the ICU with sepsis received hydrocortisone, thiamine and vitamin C. According to pre- and post-implementation data, “they had remarkable results, with patients who had a predicted mortality of 40% having actual mortality of 9%.” Those results made headlines, and “I’ve had patients’ families ask for this therapy.”
However, “I do not think we have adequate data to change our standard of care.” At the same time, she noted, “it’s low cost and probably low risk, so I understand why some people are moving toward it.” Three large randomized trials are now testing the therapy.
WHEN IT COMES to some Surviving Sepsis recommendations, some healthy skepticism may be in order.
At a sepsis update at this year’s Society of Hospital Medicine annual meeting, intensivist Patricia Kritek, MD, EdM, who’s with the University of Washington in Seattle, discussed two Surviving Sepsis recommendations on resuscitation that are both weak recommendations with low-quality evidence.
The first recommends using dynamic over static variables to predict fluid responsiveness. Dynamic variables include pulse pressure variability, IVC diameter change and straight leg raise.
“We know dynamic measures better predict how people are going to do in terms of augmenting cardiac output,” she pointed out. “But that doesn’t tell you anything about patient-centered outcomes.” Nor do we know, she added, which dynamic measure to use.
“I would ask you to be skeptical about these things,” she said, pointing out that pulse pressure variability should be done on mechanically ventilated patients who have tidal volumes above 8 ml/kg or higher and normal sinus rhythm or sinus tachycardia. “If you put 10 intensivists in a room, you’ll get 11 opinions on what dynamic measure to use, so we don’t know for sure what to do with this.”
Another recommendation—to guide resuscitation to normalize lactate as a marker of tissue hypoperfusion—is also controversial.
“We are really into lactate, in part because of the CMS sepsis bundles,” said Dr. Kritek. However, a JAMA study published earlier this year compared using capillary refill time—”which they standardized”—to lactate clearance to guide resuscitation.
The results? The outcomes were the same, but an analysis of a less sick subgroup—those with a SOFA score under 10—found they had lower mortality when resuscitation was guided by capillary refill time.
“That makes we worry, and this is conjecture now, that if we are too slavishly resuscitating to lactate clearance, we may be causing harm,” she said. “We need to understand this better, and there is more to come.”