Background: With CMS core measures requiring timely use of antibiotics in patients with fever and suspected sepsis, many patients receive antibiotics up front that may ultimately end up having another non-bacterial etiology as the cause of their fever. On the one hand overuse of antibiotics can increase bacterial resistance, healthcare costs, and potential side effects. On the other hand, withholding antibiotics from patients with bacterial infections can increase morbidity and mortality. The authors of this trial wanted to determine whether a procalcitonin-guided algorithm could be used to reduce antibiotic regimens in the ED.
What They Did:
- Higher diagnostic accuracy and cost effectiveness using procalcitonin in the Treatment of Emergency Medicine Patients with fever (HiTEMP)
- Multicenter, non-inferiority, randomized clinical trial performed in two Dutch EDs
- Adult patients with fever (≥38.2 C or 100.8 F) in triage randomized to standard diagnostic workup (control group) and procalcitonin-guided therapy (PCT group)
- PCT group = standard diagnostic workup + PCT
- PCT <0.5ug/L = bacterial infection unlikely
- PCT ≥0.5ug/L = bacterial infection likely
- Efficacy: Number of prescribed antibiotic regimens
- Safety: Composite outcome of 30d mortality, ICU admission, return ED visit within 2 weeks (used as surrogate for treatment failure)
- Accuracy: Sensitivity, specificity and AUC of PCT for bacterial infections
- Cost savings (including medical and non-medical costs)
- Protocol non-adherence
- Efficacy of PCT-guided therapy for subgroup of patients with respiratory sources of infection
- Adult patients ≥18 years with fever (≥38.2 C or 100.8 F) in triage
- Immunocompromised conditions (neutropenia with ANC <0.5 x 109cells/L, current chemotherapy, or post solid organ transplantation)
- Moribund patients
- Patients <72 hours after surgery or requiring primary surgical intervention
- 551 patients included
- Number of Prescribed Antibiotic Regimens:
- PCT: 73%
- Standard: 77%
- P = 0.28
- 30d Mortality, ICU Admission, and Return ED Visit Within 2 Weeks:
- PCT: 11%
- Standard: 16%
- P = 0.16
- AUC for confirmed Bacterial Infections for PCT was 0.681 (95% CI 0.633 – 0.730)
- Sensitivity: 0.52; 95% CI 0.45 – 0.60
- Specificity: 0.73; 95% CI 0.68 – 0.78
- AUC for confirmed Bacterial Infections for CRP was 0.619 (95% CI 0.569 – 0.669)
- All physicians attended an educational session on PCT-guided therapy before the initiation of the study and also carried a pocket card that had the PCT-guided algorithm
- Suspected bacterial infections were corroborated by two independent physicians
- Small loss to follow up at 30 days – 25 patients (i.e. 5% loss to follow up)
- Baseline characteristics were balanced between groups
- There was no blinding of group allocation for patients or physicians
- Antibiotic regimens were defined as initiation of antibiotics and lasting for several days, however, there was no corroboration that the antibiotics selected were ideal for the given infection
- 51% of all patients received intravenous antibiotics due to the presence of SIRS as part of ED protocol before any labs resulted.
- Authors missed 550 eligible patients during their enrollment period
- Protocol non-adherence in the PCT-guided therapy arm was 44%, which was most commonly giving patients antibiotics when PCT was <0.5ug/L (i.e. suspected non-bacterial etiology)
- Study not powered to show a difference in costs between groups
- Temperature was used as the sole inclusion criteria, which is not a perfect marker of infectious disease. Also patients with normothermia can have sepsis (and may represent a group with the greatest clinical equipoise) and were not included in this study creating the strong likelihood of selection bias.
- Patients with altered mental status and the sickest patients were excluded in this study due to the fact that consent could not be obtained
- The location of infection was most commonly respiratory (39%) followed by urinary tract (20%)
- With a protocol non-adherence rate of 44% in patients with PCT <0.5ug/L it is likely that physicians value their clinical judgment more than a PCT level. The results of this trial and the ProACT trial do confirm one thing…changing the antibiotic prescription behavior of physicians can be a difficult task.
- In the subgroup analysis of patients with respiratory sources of infection, there was no difference in percentage of antibiotic prescriptions based on which arm patients were randomized to (i.e. 75% vs 72%; p = 0.76). Unfortunately, the sample size was relatively small making absolute conclusions difficult to make in this patient population.
- One important discussion point is although a single PCT level may not have good accuracy for bacterial infection, subsequent levels may help with early discontinuation of antibiotics in admitted patients.
- A PCT cut-off value of 0.5ug/L may simply be too high to find low risk patients. Most studies have used lower cut-off levels, such as 0.25ug/L, which may increase accuracy.
- Patients with skin and soft tissue infection and urinary tract infections, there is little diagnostic uncertainty as most of these infections are due to bacterial etiologies, therefore in patients with obvious cellulitis and urinary tract infections, a PCT level may add little in these patients.
- Previous investigations have suggested that PCT’s most valuable role might be in conjunction with an antibiotic stewardship effort. Broad use of the test across all infections, given the heterogeneity of such an application is unlikely to show significant benefit.
- One area where PCT-guided therapy could be of benefit is in patients where there is diagnostic uncertainty. Rick Pescatore, DO has discussed this before on REBEL EM in elderly patients with questionable UTIs.
Author Conclusion:“PCT-guided therapy was non-inferior in terms of safety but did not reduce prescription of antibiotic regimens in an ED population with fever. In this heterogenous population, the accuracy of PCT in diagnosing bacterial infections was poor.”
Clinical Take Home Point: Procalcitonin is still a test looking for a disease, at least in the emergency department. Use of a PCT-guided antibiotic therapy for broad application should not be recommended or used at this time as it has a poor accuracy for bacterial infections in a wider population of patients with fever and suspected sepsis. One area in which PCT-guided therapy may have some benefit is in patients with diagnostic uncertainty—further studies with more narrowly-defined research settings are needed.
- Van der Does Y et al. Procalcitonin-Guided Antibiotic Therapy in Patients with Fever in a General Emergency Department Population: A Multicentre Non-Inferiority Randomized Clinical Trial (HiTEMP Study). Clin Microbiol Infect 2018. PMID: 29870855
- Schuetz P et al. Procalcitonin in Patients with Fever: One Approach Does not Fit All. Clin Microbiol Infect 2018. PMID: 30076970
Post Peer Reviewed By: Rick Pescatore, DO (Twitter: @Rick_Pescatore)