Fast-track Blood Test for Suspected Fever by Deficiency of a Kind of White Blood Cells As Main Defense Against Infection

1. Background

1.a. Burden of neutropenic fever Neutropenic fever (NF), or febrile neutropenia, is characterised by high body temperature and low absolute neutrophil count (ANC) following myelosuppressive cancer treatment.[1] It occurs in 5 - 10 percent of patients with early-stage solid tumours, 20 - 25 percent with non-leukaemic haematological cancers, 85 - 95 percent with acute leukaemia,[2] and 13 - 21 percent with metastatic solid tumours.[1] It is more common after the first cycle of chemotherapy. 7.83 per 1,000 cancer patients were hospitalised for NF annually in the United States (US).[3] With earlier cancer recognition, and more prescriptions of chemotherapeutic agents and targeted therapies, the figures are expected to rise.

NF is associated with unplanned chemotherapy interruptions and relative dose intensity (RDI) reductions more than 15 percent, which undermine treatment success rates and overall survival.[2] When complicated by neutropenic sepsis, a dysregulated host response against infection, NF becomes an oncological emergency. The mortality rate is 3 - 18 percent following complications e.g. hypotension, respiratory failure, encephalopathy, cardiac failure and arrhythmia, renal failure, haemorrhage, and admission to intensive care unit (ICU).[4] Risk factors of life-threatening infections are severe neutropenia, protracted neutropenia, and splenectomy.[5] Mortality risk increases with advanced age, comorbidities, clinically documented infection, bacteraemia, leukaemia and lung cancer as underlying malignancies.[4,6]

The cost of managing NF remains substantial for healthcare systems worldwide. The mean direct hospitalisation costs in the US, Germany, and Singapore were US$19,110 (1995-2000), €3,950 (2005-2006), and US$4,913 (2009-2012) respectively.[6-8] Higher costs are associated with inpatient treatment, comorbidities, discharge, deaths, male sex, and infection.[9]

1.b. Diagnostic criteria NF is defined by 1) single oral temperature ≥ 38.3 degree Celsius (101ºF), or ≥ 38.0 degree Celsius (100.4ºF) sustained over 1 hour; and 2) ANC < 1.0 x 109/L ("moderate" neutropenia). Neutropenia becomes "severe", "profound" and "protracted" if ANC < 0.5 x 109/L, < 0.1 x 109/L, and lasts for more than one week, respectively.[10] This definition applies to oncological and haematological participants only.

1. c. Aetiology Neutrophils are recruited early during the acute phase of bacterial and fungal infections.[11] Fever may be the only manifestation of infection during neutropenia because the typical signs of inflammation are obscured. Neutropenia is usually acquired by myelosuppressive cancer treatments, and pre-engraftment phases of haematopoietic stem cell transplantation (HSCT). Bone marrow failure and defective neutrophil maturation are other possible mechanisms.[12] NF can have infectious and non-infectious causes. Fever of unknown origin, chemotherapy-related oral mucositis, tumour-related cytokine release, transfusion-related reaction, drug reaction, graft-versus-host disease, and thromboembolism are common non-infective causes.[13] 30 - 50 percent are infections by clinical presentation or microbiology,[14] whilst only 20 - 30 percent are microbiologically documented infections.[15] Bloodstream infections, bacterial translocation from respiratory tract and perianal region, and central venous catheter are major sources.

1.d. Situation in Hong Kong Local public emergency departments (EDs) have implemented clinical pathways for suspected NF, which expedite medical consultation, septic workup, and broad-spectrum antibiotics such as Meropenem or Piperacillin/Tazobactam. The target time from ED registration to antibiotic administration, or Door-To-Antibiotic (DTA) time, is within 1 hour disregarding ANC, in line with international guidelines.[16-18] DTA times are shortened after implementing clinical pathways in local ED.[19] Local epidemiology, antibiotic sensitivity patterns, and healthcare cost in managing NF are understudied. It has been shown that inadequate antibiotic regimen was more significantly associated with ICU admissions and mortality than longer DTA times. Presence of sepsis or septic shock, prior colonisation with drug resistant strains, and risk stratifying indices, are criteria to judge antibiotic adequacy.[20]

1. e. Antibiotic stewardship for cancer participants Participants with cancer are frequently exposed to antibiotics for treatment and prophylaxis, therefore they are more vulnerable to multi-drug resistance, and are in special need for antibiotic stewardship.[21] Broad-spectrum antibiotics are often started empirically in ED for suspected NF, assuming infection by drug-resistant bacteria. However, clinicians may not proactively de-escalate subsequent antibiotics. Prolonged exposures to parenteral broad-spectrum antibiotic impose risks of nosocomial infection and injection site complications. Gram-negative bacilli are more frequent in neutropenia. Staphylococcus aureus, Acinetobacter and Enterobacter species are more frequent in non-neutropenic bloodstream infections.[22] More drug resistant (MDR) strains, such as Extended-spectrum Beta-lactamase (ESBL)-producing Enterobacteriaceae, MDR Pseudomonas aeruginosa, and MDR Acinetobacter were isolated from neutropenic patients in a survey in China.[23]
2. Research (PICO) question In adult participants with cancer presenting to the ED with suspected NF, can an ED protocol guided by Fast-tRack Absolute Neutrophil Count (FRANC protocol) compared with conventional care guided by clinical suspicion alone improve safe antibiotic stewardship?
3. Objective and purpose

3.a. Objectives

In adult participants presenting to the ED with suspected NF:

• To investigate the effectiveness of the FRANC protocol to improve antibiotic stewardship by restricting meropenem use except for confirmed neutropenic or clinically unstable participants; and
• To investigate the safety of the FRANC protocol. 3.b. Primary hypothesis In participants presenting to ED with suspected NF, the FRANC protocol significantly reduces unnecessary use of Meropenem compared with normal Standard of Care (SoC). 3.c. Secondary hypothesis In participants presenting to ED with suspected NF, there is no significant difference between those receiving the FRANC protocol and those receiving SoC for serious adverse events including death.

1. Methods 4.a. Participants 4.a.1. Target population This trial targets to adult participants attending ED because of fever, who are at risk of neutropenia related to cancer therapies and underlying conditions. 4.a.2. Study location The trial will be conducted in the ED of Queen Mary Hospital with an average annual attendance of 125,000. The hospital is a tertiary referral centre for HSCT and oncological services which received more than 500 adults with suspected NF via the ED in 2019. 4.a.3. Recruitment and screening Alert cards are routinely issued to participants at risk of neutropenia during follow-ups in Clinical Oncology and Haematology centres. They are reminded to visit EDs as soon as possible when fever occurs. A neutropenic risk alert is set to pop up in the Clinical Management System (CMS) that is shared among all public medical facilities. It is valid for 6 weeks from the last chemotherapy against solid tumours, or lifelong for haematological malignancies, and HSCT.

When febrile participants with valid alerts attend, the triage nurse will declare "1 - Critical" or "2 - Emergency" categories according to clinical state. Consultations by emergency physician will start within 15 minutes. Participants who receive cancer treatment in other medical facilities will also be included if they meet the same criteria.

Participants will then be screened for eligibility by trained research staff for inclusion and exclusion criteria in the ED. Potential participants will be invited to provide informed consent in written, signed, and dated forms. They can decline at any time.

4.a.4. Sample size For antibiotic stewardship A sample size of 344 participants (172 per group) will achieve 80 percent power to detect a superiority difference between two group rates of meropenem prescription. We set the superiority margin at 10 percent and assume that the control group has Meropenem usage rate of 91.8 percent (based on preliminary data) and 71.8 percent for intervention group. The calculation is based on a one-sided Z test at the significance level of 0.05.

4.a.5. Randomisation \& allocation concealment We aim to achieve balanced treatment assignments in 1:1 ratio with simple randomisation by a computer-generated code list. The code is not broken until the last participant is enrolled and has completed 180 days of follow-up. It will be implemented using independent electronic case data files to ensure allocation concealment.

4.b. Data processing and analysis 4.b.1. Data processing All data will be entered electronically by research staff to password-protected, secured, web-based system with tailor-made recording forms using tablet computers. 5 percent of all data will be checked for accuracy. All hard copies will be restricted and locked in unit cabinets. All prevailing regulations by The University of Hong Kong and the Hospital Authority will be strictly followed. An audit trail will be made to include the number of participants screened, approached, recruited, and excluded (with reasons). Participant and clinical staff responses, and data completeness will be evaluated.

4.b.2. Data analysis 4.b.2.a. Primary analysis Baseline characteristics between two groups will be assessed for potential imbalances, which will be adjusted for when comparing outcomes. Intention-to-treat (ITT) analysis will be used by imputing all non-responses at follow-up with baseline values, which yields more conservative estimates of effect sizes. The missing data for the primary outcome is expected to be < 1 percent because most data is retrievable from electronic participant records with electronic time stamps. Statistical analysis will be done using Statistical Package of Social Sciences (SPSS) version 26 (IBM SPSS Statistics, New York, US) with biostatistician support.

Main effect: Intervention vs. Control on proportion of participants receiving Meropenem in 7 days using Chi-squared test.

4.b.2.b. Secondary analyses

• All secondary outcomes at different time points using generalised mixed effect model for multiple intra-participant and inter-participant observations. Main and interaction effects will be assessed.
• Subgroup analysis to primary outcome based on age, gender, type of cancer/ conditions, disease status, therapy received, time interval from last chemotherapy or HSCT to ED registration, at 30, 60, 90 days (4) Sensitivity of ITT and per-protocol analyses for pre-defined co-variates, missing data (including participants lost to follow-up), will be done. Imputation method will depend on actual pattern of missing data.
• If the attrition rate is more than 5 percent, risks of attribution bias and influence to statistical power will be assessed.
• Complication rates over time will be compared with Poisson mixed effects models. Uses of inotrope, mechanical ventilation, renal replacement therapy, and ICU are analysed against mortality using appropriate survival analyses with time to event data.