鶹ýӰ

Algorithm Helps Manage Fever

— Pediatric cancer patients with non-neutropenic fever could benefit

MedpageToday

This article is a collaboration between MedPage Today and:

TORONTO -- A simple algorithm can help manage pediatric cancer patients with non-neutropenic fever, a researcher said here.

In a retrospective analysis, the five-step guideline accurately classified patients as being at high, medium, and low risk of serious infections, according to Pat Gavigan, MD, of St. Jude Children's Research Hospital in Memphis.

Such an approach could improve recognition and guide initial therapy in cases of non-neutropenic fever, Gavigan said at the annual meeting of the .

Pediatric cancer patients frequently have fever and are at risk of bacterial infection because they are immunosuppressed, with impaired mucosal surfaces from chemotherapy and often central venous catheters, he noted.

Febrile neutropenia is associated with about a 20% risk of serious bacterial infection and there are guidelines for risk stratification and management when such episodes take place, Gavigan said.

But no similar guidelines exist for non-neutropenic fever, in which invasive bloodstream infections have been reported in 6%- 10% of cases, he added, and even in a single institution there are a range of management strategies.

To help fill the gap, he and colleagues tried to devise a tool that would be pragmatic, stratify risk, and include practical management considerations.

The five-step algorithm they came up with begins with the recognition of fever, the suspicion of infection, and collection of blood cultures and serum lactate, as well as a complete blood count.

Patients are at the highest risk if they have signs of sepsis, including such things as septic shock, toxic appearances and rigors, organ dysfunction, or neurological dysfunction. These patients would be treated in the ICU with vancomycin, meropenem, and amikacin or on the ward with vancomycin, cefepime, and amikacin.

If there's no sepsis, Gavigan said, patients could have high-risk features, such as recent chemotherapy or be under treatment for leukemia in induction, re-induction or treatment for relapsed and refractory disease. Treatment would be cefepime, plus vancomycin or meropenem if indicated.

If there are no high-risk features, patients still might be at elevated risk if they have any of several individual characteristics, including age under 3, recent surgery, a CSF shunt, signs and symptoms of a bacterial infection, new fever while on antibiotics, and fever for more than 48 hours. Treatment should be individualized after consulting an infectious disease specialist.

Finally, patients would be considered at low risk for serious infection if they appear well, have a maximum temperature of less than 39°C, have a port with no past bloodstream infection, or have some obvious cause for the fever, such as an upper respiratory infection or drug. No antibiotics are indicated, but parents should be informed, and the patient should be reviewed within 24 hours.

Any patient who doesn't fit those categories but still has fever would be classified as standard risk and should be given ceftriaxone and reviewed within a day, Gavigan said.

To evaluate the algorithm, he and colleagues looked at back at the records of 1,049 patients who had 2,224 episodes of non-neutropenic fever, defined as a temperature of at least 38°C and an absolute neutrophil count of at least 500, from Jan. 1, 2012 to Jan. 16, 2017.

They tested the guideline in the 304 episodes (in 158 patients) where there was a positive blood culture, ICU admission or death, and found that:

  • 88 met criteria for sepsis, including 47 admitted to the ICU
  • 64 had high-risk features
  • 33 had individual features suggesting elevated risk
  • None of the cases would have been classified as low-risk, but 120 had simple uncomplicated courses
  • 120 were classified as standard risk

Gavigan said the guideline appears to be able to identify high-risk patients well but needs to be validated in other settings and in a prospective manner.

"They know that these patients are at higher risk and they're trying to kind of codify the response to that and then evaluate if that algorithm is actually better than just plain old clinical judgment," commented , of Cincinnati Children's Hospital Medical Center, who was not part of the study but who co-moderated the session at which it was presented.

But he cautioned that the study was "purely observational. They're going to have to test it in larger, multi-center kind of way" before it becomes widely accepted.