Procalcitonin-guided care in sepsis shows no difference in early antibiotic initiation but lower 28-day mortality

BACKGROUND: Sepsis is a common and serious condition, defined as a dysregulated host response to infection, that leads to life-threatening organ dysfunction. In emergency department settings, accurate diagnosis can be challenging, as many non-infectious conditions have similar presenting features and there is no gold standard diagnostic test, which can lead to misdirected use of antibiotics. Procalcitonin is a well characterised biomarker that responds rapidly and with high specificity to the presence of bacterial infection. We aimed to investigate whether supplementing current practice with rapid procalcitonin testing would improve recognition of sepsis and allow reduced antibiotic prescribing with at least no change in overall mortality. METHODS: A parallel, two-arm, open-label, individually randomised controlled trial was done in 20 hospital emergency departments within 17 National Health Service (NHS) Trusts or Health Boards across England and Wales. Patients aged 16 years and older with suspected sepsis were randomly assigned to either usual care or procalcitonin-guided care in a 1:1 ratio via a centrally controlled web-based randomisation programme. Participants, research staff, those assessing outcomes, and statisticians analysing the data were not masked to group assignment. Participants in the usual care group were assessed according to standard clinical management based on National Early Warning Score 2 (NEWS2). In the procalcitonin-guided care group, rapid procalcitonin testing was used in combination with NEWS2 assessment by use of a guidance-only algorithm for clinicians. This algorithm for clinical management was used for both usual care and procalcitonin-guided care groups and clinicians were free to use, ignore, or deviate from the algorithm. The co-primary endpoints were intravenous antibiotic initiation at 3 h (superiority) and 28-day mortality (non-inferiority) from triage assessment, assessed in all randomly assigned consenting participants with data for both co-primary outcomes available. Non-inferiority was concluded for 28-day mortality if the upper bound of the 90% CI was below a 2·5% margin on the risk difference scale. All patients who were randomly assigned to one of the two groups and who consented to data collection at baseline were included in adverse event analysis for the period they were included in the study. The study was registered with ISRCTN (ISRCTN54006056) and is complete. FINDINGS: Between Nov 20, 2020, and Nov 1, 2023, a total of 7667 patients were recruited and randomly assigned to the usual care group (n=3836) or the procalcitonin-guided care group (n=3831). 5453 patients (2748 female, 2703 male, 1 non-binary, 1 data for gender missing) were included in the primary analysis population (2715 usual care, 2738 procalcitonin-guided care). The last 28-day follow-up was on Nov 29, 2023. There was no difference in intravenous antibiotic initiation at 3 h from triage assessment between the two groups: 48·4% (1325/2738 participants) in the procalcitonin-guided care group versus 48·2% (1308/2715 participants) in the usual care group (adjusted risk difference -0·08 percentage points, 95% CI -2·58 to 2·42; p=0·95). Mortality at 28 days was lower in the procalcitonin-guided care group: 13·6% (372/2738 participants) in the procalcitonin-guided care group versus 16·6% (450/2715 participants) in the usual care group (adjusted risk difference -3·12 percentage points, 90% CI -4·68 to -1·57; p=0·0009). The upper bound of the 90% CI was below the non-inferiority margin of 2·5 percentage points and the point of no effect, implying both non-inferiority and superiority at the one-sided 5% level. In the primary analysis population, the procalcitonin test result was considered in clinical decision making in 64·7% (1771/2738) of participants in the procalcitonin-guided care group. Improved mortality was not explained by findings of planned subgroup, sensitivity, or secondary analyses. A total of 180 adverse events were recorded, 53·3% (96 events in 57 [1·9%] of 2968 participants) in the usual care group and 46·7% (84 events in 66 [2·2%] of 3042 participants) in the procalcitonin-guided care group. One (<1%) of 2042 participants in the procalcitonin-guided care group reported a serious adverse event probably or definitely attributable to the procalcitonin test. INTERPRETATION: Making a procalcitonin-guided algorithm available to clinicians in emergency departments did not change intravenous antibiotic initiation at 3 h in patients managed as suspected sepsis, but a decrease in 28-day mortality was seen and further research is needed to understand this finding. FUNDING: National Institute for Health and Care Research.