Antibiotic overconsumption has been considered as one of the major contributive factors of the emergence of multidrug resistant bacteria, a serious threat particularly in intensive care units. Antibiotic stewardship programs are set up to meet this problem. Shortening the duration of antimicrobial therapy seems to be one of the strongest tools of these programs. Nevertheless, the decision to stop antibiotics in a critical care patients remains often challenging in real-life practice.

Procalcitonin (PCT), an inflammatory biomarker, has a promising profile and scores better than traditionally biomarkers as c-reactive protein (crp) and leucocytosis. Although two big multicenter randomised controlled trials showed a positive impact of PCT use in Intensive Care Unit (ICU), as it led to reduction of antibiotic exposure, the efficiency of this biomarker is still a point of debate. Notably the cost of PCT determination is a counterargument for its routinely use as it is a quite expensive test and its cost-benefit ratio has not been well studied.

The objective of this study is to test a PCT-algorithm for stopping antibiotics in a real life setting by assessing its impact on antibiotic consumption. The investigators hypothesize that it will shorten antimicrobial courses and will decrease overconsumption, with a possible positive impact on the increase of antimicrobial resistance and with no apparent adverse outcome.

This is a single center, before-after, matched cohort study in a ICU of a non-university hospital. It puts the focus on neurocritical care patients as they are prone to infectious complications and these cases are often less straightforward in terms of diagnosis and management. The investigators hypothesise that such an algorithm, in addition to clinical judgement of the ICU physician, will reduce the duration of antibiotic treatment and the overall antibiotic use, could reduce selective pressure, without increase of mortality or recurrent infections.

All patients admitted to ICU with a primary non-infectious neurologic pathology will be screened for eligibility. Eligible patients can be enrolled if they meet all the inclusion criteria, there are no exclusion criteria, a matched control is available and an informed consent has provided by the patient or his or her legal representative. Patients included in the intervention group will be 1:1 matched with a patient out of the historical control group. Therefore, three fixed and two variable criteria for matching will be used. The fixed ones are: 1. pathology (traumatic brain injury; intracerebral/subarachnoid hemorrhage due to cerebral aneurysm or arteriovenous malformation; ischemic stroke; hemorrhagic stroke or other intracranial hemorrhage from; and the whether or not need for intervention (none, endovascular or neurosurgical intervention) . 2. need for mechanical ventilation within the first 12 hours after admission and 3. Glasgow coma scale (GCS) < 8 versus ≥ 8 at ICU admission or before starting sedation or anesthesia.

The two variable criteria are age (with a range of plus or minus 10 years) and the Acute Physiology and Chronic Health Evaluation (APACHE)-II score (with a range of plus or minus 15 points).

Matching will be manually, blinded for the outcome. A control group has been composed by analysing retrospectively patients admitted to our ICU from January 2016 on.

In the intervention group, doctors are additionally provided with PCT levels on regular base and with a non-binding advice on continuation or discontinuation of antibiotic therapy by electronic patient data management system feedback and by the involved microbiologist. The first PCT value will be used just as base-line, without any therapeutic consequence. Measurements will be repeated every 3-4 days, the last one is scheduled at day 27.

Patients in both groups are followed until hospital discharge, allowing assesment of hospital length of stay, hospital mortality and prevalence of multidrug resistance during hospitalisation.

Following data were collected at initiation of the trial : age, sex, BMI, pre-existing comorbidities including presence of known multidrug resistant bacteria before ICU admission, previous location before admission, reason for admission, GCS at admission to ICU, intervention, reason for antibiotic start, day of ICU admission at starting point of the study, APACHE II, the presence and type of organ dysfunction using the sequential organ-failure assessment (SOFA) score and use of mechanical ventilation.

Subsequently, following parameters will be recorded during ICU stay: SOFA score and type of organ or system failure at every PCT measurement, daily need for mechanical ventilation, source of infection when known and results of microbiological cultures. Additionally, the inflammatory biomarkers crp and leucocytosis as well as maximum temperature will be daily registered.

At the end of follow-up, every suspected infectious episode will be sorted by the investigator into four groups: 1. microbiologically documented infection (presence of a clinical and/or radiological infectious focus and pathogen identification); 2. clinically documented infection (presence of a clinical and/or radiological infectious focus, without causative pathogen identification); 3. absence of infection (absence of a clinical or radiological infectious focus) 4. possible infection (all other situations).

Data management will be performed by the investigators or research nurses. Patient's anonymity will be maintained by identifying enrolled patients of both groups with a trial identification number. The list containing the subjects name and allocation numbers are kept in strict confidence. Electronical clinical research files (CRF) or other subject related data will be protected by a password and will be put on a secured server within the hospital. Similarly, written papers (including CRF on paper) containing privacy sensitive information will be kept behind locked doors.

For the power analysis, the ICU databank has been queried, analysing the antibiotic duration of the first, uninterrupted antibiotic course of 74 patients who stayed at least 10 days in ICU in 2016 and got empirical and/or directed antibiotic therapy. A lognormal distribution with a peak value at 11.78 days has been found. After log-transformation, a mean of 2.466 with standard deviation of 0.433 was found.

The investigators aim for 20% antibiotic reduction, supported by the results of the PRORATA and the SAPS trial. This goal is ambitious, but is not unrealistic, taking into account that the study population group is traditionally characterized by high figures of antibiotic consumption. Based on these assumptions, a number of 60 patients in both groups is required for obtaining a power of 80%. This number has been increased with 10%, anticipating for patients who will die due to non-infectious reasons (pe. expected poor neurologic outcome with therapy restrictions or withdrawal) between day 7 and the end of the intervention period (day 28). In conclusion, a total of 66 patients will be included in the intervention group and will be matched with 66 patients out of the historical control group.

Inclusion Criteria:

1. patients admitted to the ICU with a primary non-infectious neurological pathology:

   • Traumatic Brain Injury
   • Intracerebral Bleeding (pe. subarachnoid bleeding) due to aneurysm or arteriovenous malformation
   • Ischemic Stroke Stroke
   • Hemorrhagic stroke or other intracranial haemorrhage
   • Other non-infectious neurologic condition (as hydrocephalus, status epilepticus, postoperative complication after elective neurosurgery, ...)

   AND

2. requiring antibiotics within the first week (day 0 - day 6) after ICU-admission for a suspected bacterial infection

Exclusion Criteria:

• severe immunodeficiency and/or neutropenia: defined as (1) solid-organ transplant recipients with immunosuppressive therapy (monotherapy with corticosteroids is allowed), (2) recent chemotherapy in last 6 months, (3) hematologic malignancy with active therapy in last 2 years, (4) bone marrow transplant, (5) HIV patient with clinical complications (Pneumocystis jirovecii, Kaposi's sarcoma, lymphoma, tuberculosis, toxoplasmosis, …) or CD4 count < 200/mm3, while neutropenia has been defined as white cell count < 1000/ml.
• microbiologically proven infection with Pseudomonas, Acinetobacter baumannii, Lysteria or atypical pathogen as Chlamydia, Legionella or Mycoplasma; or Staphylococcal aureus bacteremia
• microbiologically proven meningitis or ventriculitis
• compartmentalised infection: pe. abscess, empyema
• microbiologically proven (co-)infection making a prolonged antibiotic course necessary, such as endocarditis, prosthetic joint infection or septic arthritis, osteomyelitis, chronic prostatitis, ...
• already > 24h on antibiotics before ICU admission
• (expected) ICU length of stay < 7 days
• no match available in the historical 'Standard of Care' group
• no Informed Consent obtained