A Prospective Cluster-Randomized, Controlled Crossover Trial to Validate an Electronic "Resistance Calculator"

2.1 List the clinical hypotheses.

• Primary hypothesis: The "Shamir Resistance Calculator" will shorten the time to initiation of appropriate antimicrobials among hospitalized septic patients.
• Secondary hypotheses: The "Shamir Resistance Calculator" will reduce the rate of new MDRO acquisitions, will reduce the empiric usage of broad-spectrum agents, will reduce the incidences of acute Clostridiodes difficile infections, and will reduce the in-hospital mortality rates among hospitalized septic patients.

2.2 Background \& Rationale, Significance of Selected Topic \& Preliminary Data

The world health organization (WHO) had declared that the issue of resistance to multiple antimicrobials, is one of the current and future biggest challenges and threats to human health.

• Multi-drug resistant organisms (MDRO) and extensively drug resistant organisms (XDRO) are prevalent in hospitals, in acute care settings, and also in the community.2
• The outcomes of infections caused by MDROs and XDROs are worse in comparison to susceptible organisms, mainly due to delay in initiation of appropriate antimicrobial therapy (DAAT) in those infections.3
• DAAT is the strongest independent modifiable predictor for mortality among patients with severe infections.3
• Therefore, it is of paramount importance to avoid and shorten DAAT among hospitalized septic patients.
• However, empirically prescribing unnecessary broad-spectrum agents, to patients with infections resulting from susceptible organisms, increases the burden of resistances and costs to the facility.
• Moreover, narrow-spectrum agents are frequently more potent vs. susceptible organisms (e.g., cefazolin vs. vancomycin for methicillin-susceptible Staphylococcus aureus infection4), and are frequently less toxic (e.g., colistin, vancomycin).4-6
• There are two time points, where the majority of empiric prescription errors and DAAT frequently occurs among hospitalized adults: 1) upon admission (UA) while attending patients at risk of MDRO infection, and 2) during hospitalization, while attending patients at risk of XDRO infection (MDRO are usually covered empirically in nosocomial sepsis).6
• At the first encounter with a septic patient, the primary physician needs a simple and reliable tool to be used bedside, in order to tailor an appropriate empiric antimicrobial regimen.
• Many previous attempts to develop such scores, failed to become a standard of practice, for multiple and various reasons,7-10 lacking the size and the comprehensiveness to address the whole spectrum of clinical issues and decisions.
• We developed an on-line calculator (https://assafharofe.azurewebsites.net), which is free and accessible from all devices, in order to predict the risk of MDRO infection upon admission, and the risk of nosocomial XDRO infection during the hospitalization.11
• The calculator contains only parameters that could be extracted bedside, at the first presentation of a septic patient, including the ability to conduct a complete assessment also among patients with reduced consciousness and/or cognition.
• The study that was executed in order to develop the "Shamir Resistance Calculator", which was recently published in Infection Control and Hospital Epidemiology,11 was a large study that included 4,114 patients: 2,472 patients with sepsis upon admission and 1,642 patients with nosocomial sepsis. This as far as we know, is the largest trial ever published in this field of research.
• Moreover, the calculator performances were very encouraging. The MDRO upon admission score included 10 parameters, and with a cutoff of ≥22 points, it had a ROC AUC of 0.85. The nosocomial XDRO score included 7 parameters, and with a cutoff of ≥36 points, it had a ROC AUC of 0.87.
• However, the main limitation of the on-line calculator, remains the lack of prospective data that trialed the performances of the calculator prospectively, on hard hospitalization's outcomes.
• Therefore, our goal is to trial the calculator prospectively, using the cluster randomized crossover design, which is considered today the gold standard methodology in terms of interventional studies in the field of hospital epidemiology.12

2.3 Study Design

• The study design is an interventional prospective cluster randomized crossover trial.12
• The crossover design is considered today the gold standard methodology to execute interventional trials in the field of hospital epidemiology, infection control, and antimicrobial stewardship.12 This design enables to control for multiple confounders associated with the Hawthorne effect. The intervention is implemented in same types of units. The intervention is implemented in one unit, and afterwards in the second unit, separated with washout periods prior, between, and at the end of the intervention in the second unit (i.e., to control for carryover effect of the intervention). Eventually, the independent performances of the intervention could be measured, since every unit is compared to the other unit but also the intervention period is compared to the control period in the same unit (see figure in section 2.6).
• The study will be conducted at two large Israeli acute-care hospitals: Shamir (Assaf Harofeh) Medical Center (895 beds) and Mayanei Hayeshua Medical Center (350 beds).
• Each intervention period will last 6 months and each washout period (prior to the first intervention initiation, between phases, and following the last intervention period) will last one month (see figure in section 2.6).
• The study will include all adult (>18 years) septic patients who are hospitalized and requires empiric antibiotic treatment.
• Patients with known pathogen/s for which antibiotic treatment is initiated, will not be included.
• No written informed consent will be obtained from participants, as this intervention imposes a minimal neglected potential risk to patients' safety.
• According to Israeli regulations, these types of interventions, in the field of Infection Control / Antimicrobial Stewardship, does not necessitates obtaining written informed consent from participants. However, the protocol must be submitted, reviewed and authorized by the local IRB at both facilities, which will approve the study prior to its initiation.
• Of note, every prescriber has the option to over-rule or ignore the calculator's results (see figure below).
• The calculator is not aiming to replace good medical clinical judgement or practices. It is only an adjunct tool to assist physicians.
• The calculator also serves to guide prescribers of when to involve Infectious Diseases (ID) specialist, i.e., further contributing to patient's safety.
• The calculator has low false negative rates (14%) and excellent negative predictive value rates (98%), i.e., implying that prescribers could trustily rely on a negative result of the calculator, with high level of confidence.

2.4 Diversity \& Inclusion

Diversity:

All patients above 18 years old will be included in the study. No diversity issues are relevant.

Inclusion criteria:

• Adult patients (age>18 years)
• Patients who are hospitalized at Shamir MC or at Mayanei Hayeshua MC during the study period.
• Patients admitted to the ER and later-on discharged home, or discharged from the ER to a different facility, will be excluded as it will not enable us to follow all the designated outcomes.
• Patients for which antibiotic treatment was initiated, empirically, due to acute sepsis.
• Patients for which antibiotics were initiated as part of a main-consolidative regimen, or as part of a pre-emptive regimen, will not be included in the study.
• Sepsis will be defined in accordance to established criteria1.

2.7 Study Duration

The study will last 15 months (09/01/2021 - 11/30/2022).

• Study interventions are un-blinded.
• We will be able to monitor whether prescribers had used the calculator during or not during the designated intervention in their hospital, since every usage of the calculator is monitored and captured in an electronical database (i.e., this is an on-line free calculator and all results are captured and stored in secured settings).

Variables/Time Points of Interest

• Time to initiation of appropriate antimicrobials (i.e., according to the in-vitro susceptibility results of the offending isolate/s)
• New acquisitions of MDRO by patients (monitored regardless of study's purposes for Infection Control purposes).
• The amount of empiric usage of broad-spectrum agents (e.g., carbapenems, vancomycin)
• Acute C. difficile infections
• In-hospital mortality

2.8 Statistical Analysis and Sample Size Justification

Statistical Methods The performances of the calculator with regards to patients' outcomes as listed above, will be analyzed by Poisson regression models.

Power/Sample Size:

Based on our retrospective trial and preliminary data, of the patients with sepsis upon admission, 60.6% of patients with MDRO infection will suffer from DAAT for >48 hours, compared to 9.2% of patients with non-MDRO infection. This implies that at least 56 patients will need to be enrolled to each cluster (i.e., overall, 224 patients), in order to prove superiority to the intervention in 5% significance (α) and 80% power (β). Based on our preliminary retrospective trial at Shamir MC, we will expect at least ~1,000 eligible patients (i.e., patients presenting with acute sepsis for which empirical regimens are initiated) at Shamir MC during the 6-month intervention, and ~500 patients at Mayanei Hayeshua MC. Therefore, we expect the study to be way over-powered to prove the superiority of the intervention, and we plan to enroll a minimum of 120 patients for each cluster (i.e., a minimum of 480 patients overall).

2.9 Specific Drug Supply Requirements

The on-line calculator will be installed on every computer on which the electronical medical record of the facility is installed, which is used by physicians (i.e., both desktops and laptops).

2.10 Adverse Experience Reporting

AE section will be added by local PV per global templates. The investigator will act accordingly.