Image by Getty Images
Image by Getty Images

Debates in sepsis

Experts disagree about use of antibiotics, IV fluids, and quality measures.

If you want to start a fight at a medical convention, gather an intensivist, an infectious disease specialist, an emergency physician, and a nephrologist and ask them about sepsis.

Although all certainly agree that reducing mortality from sepsis is a critical task facing hospitals and physicians, the experts and medical societies that deal with sepsis have found numerous points of disagreement.

Sources of contention include antibiotics and fluids, with even those who agree on the optimal use of these interventions debating how to put them into wide practice. And although much of this conflict stems from lack of definitive data, an ongoing trial of sepsis treatment has stirred enough opposition that it was covered in the New York Times. Then there's the debate over the diagnostic criteria for sepsis itself.

“With sepsis, we can't even agree on the definition, let alone the presence or absence of it,” said Mitchell Levy, MD, chief of critical care, pulmonary, and sleep medicine and professor of medicine at Brown University in Providence, R.I. “That's unfortunate because what happens is clinicians and the public get a mixed message.”

Perhaps not surprisingly, another expert disagrees with that assessment. “I think it's healthy when scientists with different expertise and different perspectives debate this, and I think rather than looking at the debates as a problem, it's an advantage that people feel strongly. They want to make improvements which often occur when different views are considered,” said Henry Masur, MD, MACP, an infectious disease specialist and chief of critical care medicine for the NIH Clinical Center in Bethesda, Md.

To help hospitalists better understand these arguments among academics and their implications for patient care, ACP Hospitalist recently asked the experts to clarify their concerns, recommendations, and uncertainties about sepsis.


Antibiotics are one source of disagreement and a reason that the Infectious Diseases Society of America (IDSA) declined to endorse the most recent Surviving Sepsis Campaign (SSC) guidelines, which were published in 2016. The updated guidelines recommend that IV antibiotics be initiated within an hour of patients presenting with sepsis or septic shock.

“I myself published data demonstrating that if you come to the emergency room with severe sepsis or with life-threatening organ dysfunction due to your infection, every hour that goes by, your likelihood of developing septic shock goes up by 8%, and your likelihood of dying goes up by 5%,” said Steven Q. Simpson, MD, FACP, a professor of pulmonary and critical care medicine at the University of Kansas Medical Center in Kansas City and a coauthor of the guidelines.

However, the IDSA's response paper (published online by Clinical Infectious Diseases in 2017 and in print May 2, 2018) and other critics of the SSC guidelines worry that the focus on speed will lead to unnecessary prescription of broad-spectrum antibiotics, as hospitals compete to publicly comply with the recommendations.

“If you are being measured by a scorecard that counts only how often patients with syndromes looking like sepsis got antibiotics within an hour, then you want to give everybody who even looks a little bit like sepsis antibiotics within that first hour,” said Dr. Masur. “Many patients, if you had thought about it for two hours or three hours, you would have realized they really don't have a bacterial infection—they might have pancreatitis or influenza and congestive heart failure.”

Patients who unnecessarily receive the drugs risk Clostridium difficile, acute kidney injury, hepatitis, cytopenia, rash, and antibiotic resistance, said a recent JAMA viewpoint critiquing the early use of antibiotics for sepsis. Dr. Masur also worries that the one-hour metric will lead hospitals and clinicians to find ways to game the performance measurement system.

In addition, critics question the evidence gathered by Dr. Simpson and others. The results may be confounded by early antibiotics being “a marker of better care,” said Paul Marik, MBBCh, FACP, chief of pulmonary and critical care medicine at Eastern Virginia Medical School in Norfolk.

Associations between speedier antibiotic delivery and lower sepsis mortality over time may be due to other changes in care, he suggested. “We ventilate patients differently, we give them less blood. Understanding of critical care has evolved, and our treatment has evolved,” Dr. Marik said. Sicker, more complex patients may also get antibiotics later, causing the association between later treatment and higher mortality, the JAMA article noted.

These are limitations of the existing retrospective evidence supporting early antibiotics, acknowledged Dr. Levy, a leader of the SSC. “No one has done a randomized controlled trial looking at whether or not you get antibiotics. It would be hard to enroll people in that trial, needless to say. There are people who would argue that it's not even ethical,” he said.

He argues, however, that the concerns about the negative effects of antibiotics for patients with suspected sepsis are overblown. “Part of that argument is based on the fact that for years in hospitals, once an antibiotic was started, it was continued for days, even though clinicians were no longer convinced there was an infection present,” Dr. Levy said. “I really believe the solution is if the minute you administer an antibiotic, you're already asking yourself, ‘Should I give the next dose? Do I think this patient is infected?’”

Dr. Masur agrees with Dr. Levy on the importance of de-escalation. “We would like to see more emphasis in the [SSC] bundle that at daily follow-up, the antibiotics would be reassessed to see if it's the right antibiotic, the right dose, and whether the number of antibiotics can be reduced,” he said.

He also sees value in encouraging the speedy delivery of antibiotics. The difference of opinion is on whether the process of physician decision making should fall under this pressure to hurry.

“Physicians need to act as quickly as feasible, but often serial observations or additional laboratory results provide useful information that should inform the decision about whether to start antibiotics, and what antibiotics to start,” Dr. Masur said. “When that doctor puts in a stat antibiotic order, one thing we can measure is how quickly the order gets to the pharmacy, how fast the pharmacy mixes up the drug and gets it back to the patient. What that interval should be has to be a realistic goal: one could argue for 0.5 hour, one hour, two to three hours, depending on how ill the patient is.”

Dr. Levy sees more urgency. “It's hard to believe that anyone believes that if their loved one came to an emergency department and was not quite sure they had an infection, whether any of us would want that emergency room physician to hesitate about giving an antibiotic,” he said.

He also noted that most of the debate only applies to a specific group of patients who make up a sort of middle ground. “People who are in shock and who have a history of pneumonia or it's clear they were infected, that's an automatic. No one would say don't give them antibiotics. People who are sitting up and reading the paper and have no evidence of organ dysfunction and no real evidence of infection, no one is saying they should be given antibiotics,” Dr. Levy said.

Autonomy vs. protocols

If this seems like a lot of drama over whether to give a single dose of antibiotics to a subgroup of patients, that may be because the argument is actually a proxy for a larger debate: physician autonomy versus care protocols. In addition to being guided by the SSC, hospitals' sepsis care has been measured by CMS and publicly reported since 2015.

“I don't think the SSC guideline bundles were initially proposed with the thought that they would become CMS-mandated reporting measures,” said Dr. Simpson.

Yet, in some states, the recommended protocols have been even more firmly enshrined in policy. For example, in 2013, the New York State Department of Health began requiring all hospitals to develop protocols for early identification and treatment of severe sepsis or septic shock and to report their performance. Known as Rory's Regulations, the rules were inspired by the 2012 death of a young patient from septic shock.

“The motivation is laudable,” said Dr. Masur. “I just don't think it should be a process that is used to rank hospital performance, as opposed to a process that hospitals can use to evaluate their performance. There may or may not be good reasons why they have a lower compliance rate than a competitor—having the public or legislature decide that just because your numbers are bad, you're a bad hospital and deserve less reimbursement.”

Another concern, shared by all of the experts, is that medicine changes more quickly than policy, and the regulations could require hospitals to follow care protocols after they are made obsolete by new research.

However, supporters of the regulations believe these risks are outweighed by the potential benefits for patients. Dr. Levy was the lead author on a study of the New York regulations in which patients who were treated under the protocols had a risk-adjusted mortality rate of 24.4%, compared to 28.8% when the regulations began, according to results published by the American Journal of Respiratory and Critical Care Medicine on Sept. 7, 2018.

“Thousands of patients are alive in New York who would not have been alive if they had not done that. With that sort of an outcome, it becomes rather difficult to argue that we shouldn't have done it,” said Dr. Simpson.

But such mandates ignore the heterogeneity of sepsis, argue others. “It really depends on a lot of things, and patients are different,” said Dr. Masur. “Maybe when you actually get to know the patient and look at the record, the patient has always had a blood pressure of 90 and the patient's mentating well, so there's really not as much urgency as if I went in with a blood pressure generally of 140 and suddenly I'm 90 and I'm confused.”

That vision of high-quality individualized care ignores the reality of busy hospitals, contends Dr. Levy. “You get the sense of this physician stooped over the bedside, carefully titrating every individual therapy to the individual needs of every patient,” he said. “The idea that the physician is at the bedside moment by moment is a bit of a fairy tale. You have busy clinicians in a chaotic environment trying to remember to do the right thing.”


Doing the right thing for patients with sepsis may be even more complicated when it comes to fluid resuscitation.

At least the experts generally agree that patients who appear to have severe sepsis or septic shock should receive broad-spectrum antibiotics as soon as possible. When it comes to how much fluid these same patients should get, they are much less certain. “What's the optimal amount of fluid? No one really knows,” said Dr. Levy. “I believe on balance, 30 cc/kg will not be harmful to the vast majority of patients.”

Accordingly, the latest SSC guidelines recommend that patients receive 30 mL of IV crystalloid fluid per kg within the first three hours. After that, they recommend that fluid resuscitation be guided by frequent reassessment of hemodynamic status, and in practice, that means wide variation in care, according to the experts.

“Basic bedside approaches vary, including how and when to deploy fluids and vasopressors after antibiotic therapy, and these variations are not paired with clear evidence of benefit or harm despite widespread use,” said Nathan Ivan Shapiro, MD, MPH, a professor of emergency medicine at Beth Israel Deaconess Medical Center and Harvard Medical School in Boston.

Dr. Shapiro is one of the leaders of an ongoing study of fluid resuscitation in sepsis. The CLOVERS (Crystalloid Liberal or Vasopressors Early Resuscitation in Sepsis) trial is enrolling patients with sepsis-induced hypotension. After receiving between one and three liters of fluid, patients are randomized to either the liberal fluids group, in which they receive additional fluid prior to initiation of vasopressors, or an early vasopressors group, in which the initiation of vasopressors is prioritized before receiving additional fluid and norepinephrine titrated to a mean arterial pressure between 65 mm Hg and 75 mm Hg.

This study protocol has come under fire, with Public Citizen, a nonprofit consumer advocacy organization, calling it dangerous and asking the National Institutes of Health (which funds the study) to investigate it.

“It deviates from normal care in terms of the balancing and titration of both fluids and vasopressors based upon severity of illness. It is possible—we believe likely—that both arms could be exposing subjects to increased risk of harm, including death and organ failure,” said Michael A. Carome, MD, a nephrologist and director of the health research group for Public Citizen.

Dr. Simpson disagrees. “Anyone who works in an ICU knows that the upper and lower arm of what they're doing in CLOVERS are well within the standard practice,” he said. “It's relatively high volume, but it's not beyond the realm of reason. There are other people worried that the one liter isn't enough, but there are plenty of people who practice that way as well.”

The physicians do at least agree on why the arms are so different. “They sought to design a trial that was most likely to find a difference between the groups that they were studying. The two groups are substantially different in their management,” said Dr. Carome.

But while he sees that as an unacceptable risk to patient safety, Dr. Simpson considers it a typical component of medical research. “It's unfortunate that in every clinical trial, one group's going to do better than the other. We wouldn't want to do that to people. We would want to give them the best. But when we don't know what the best is, we have to do something to find out.”

Dr. Carome also expressed concern that the results won't provide that answer, because both care pathways represent extremes, rather than the likely best practice. Dr. Shapiro disagrees. “If the liberal arm is superior, then clinicians currently practicing a restrictive approach should re-evaluate their practice,” he said. “If the restrictive arm is superior, then clinicians who favor a liberal fluid approach should re-evaluate their practice.”

Moving forward

As they wait for more answers, the experts, even those who want standardized care, are taking a variety of approaches. “In my hospital, instead of 30 mL/kg, we try to personalize the fluid bolus using a noninvasive monitor of cardiac output, because we know the reason we're giving fluids is to augment cardiac output,” said Dr. Simpson. “So if we reach the point where fluids are no longer increasing the cardiac output, we know that we've given enough.”

Dr. Marik favors consideration of even more factors. “One needs some flexibility based on the patient's age, hemodynamics, underlying comorbidities,” he said. He cited a study, published by Nature Medicine on Oct. 22, 2018, that used artificial intelligence to guide use of IV fluids and vasopressors for patients with sepsis. “They've tried to individualize management for each patient,” he said. “It kind of goes against Surviving Sepsis, which is every patient gets treated the same.”

Individualized care, or precision medicine for sepsis, would be ideal, but the science isn't there yet, according to Dr. Levy. “The ability to really direct therapy according to individual physiologic responses is still more theory than reality,” he said. “We may give talks on precision medicine, but the ability to identify markers and identify individual sets of patients that respond differently” is still in the works.

A biomarker that identifies sepsis “is sort of like a Holy Grail,” said Dr. Simpson. “One day we will understand the pathophysiology well enough, we will have a biomarker you can get from the blood, maybe you can get it from the sweat, maybe somebody can spit in a tube for you. I give it three or four years.”

Once such a test is available, “there will be a lot less argument,” he said.

Dr. Masur is not so sure that a “magic bullet” will come along to resolve the debates and uncertainties. “Saying that a specific cytokine or cytokine panel is diagnostic for sepsis, and therefore I must give an antibiotic, or I must modulate that cytokine profile is nice to hope for, but it's clearly highly complex.”

Yet he too is optimistic about the future of sepsis care, for a perhaps surprising reason—the collaboration he sees among the experts.

“One of the real advances over the last 15 years is that the professional societies like SCCM and IDSA and federal agencies such as NIH, CDC, and FDA are working together. Now we share information, we share perspectives, and I think we have a much healthier situation than 15 years ago, when everybody was working in isolation. I think we're making a lot more advances,” Dr. Masur said.