Predict Sepsis; the Predictive Value of Bedside Measures in the Ambulance

Overview

Sepsis is a condition with a high mortality. Septic patients are frequently difficult to identify because of their non-specific presentations. There is also a low sensitivity of clinical judgment among health care personnel, and of existing screening tools, which are in turn typically based on vital parameters. Despite prior research, no unique sepsis biomarker has been identified so far. There is a need for new strategies to identify sepsis which do not rely on vital parameters and traditional laboratory blood tests alone. The hypothesis of the investigators is that a combination of clinical variables measurable in the ambulance can be used to predict sepsis. The aim of the current study is to determine the predictive value of keywords related to symptom presentation, vital parameters and point-of-care (POC) blood tests, alone and in combination, with respect to the outcome sepsis. The study is performed in the Stockholm ambulance setting from April 2017. A total of 956 adult non-trauma patients will be included.

Introduction Sepsis affects at least 19 000 people in Sweden each year. Despite being a time critical condition with a high mortality, sepsis is often not identified in a timely fashion. The current study aims to identify predictors of sepsis in the ambulance setting. Early treatment is a cornerstone in sepsis. Time to treatment in the emergency department (ED) has been shown to be shortened by identification of the septic patient in the ambulance. Identification of the septic patient in the ambulance may also move the treatment to the prehospital setting. At present, identification is dependent on the suspicion of sepsis and by health care professionals applying clinical judgment. Traditionally, the definition of sepsis has been based on the presence of infection and a minimum of two SIRS criteria (Systemic Inflammatory Response Syndrome), which have, however, been shown to be both un-sensitive and non-specific. The investigators have in previous studies proved proof of principle that a screening tool increases the identification of the septic patient and is superior to clinical judgment among both ambulance personnel and ED doctors. Moreover, these studies also illustrate that current screening tools do not identify all septic patients. Information from the patients' medical history has not previously been incorporated in a screening tool. The investigators have previously identified keywords related to the septic patients' medical history as presented to the emergency medical services (EMS). Eight keywords showed a prevalence exceeding 20% for all septic patients, and examples were gastrointestinal symptoms and a history of abnormal temperature. However, the specificity of these keywords with respect to sepsis remains to be determined. Also, point of care blood tests may add value in sepsis identification in the ambulance when combined with other measures. Glucose is currently measured by the Swedish EMS. Lactate testing is standard in Swedish EDs, but yet not in all ambulance services, and is used as a marker of hypoperfusion and response to treatment in sepsis. Soluble urokinase-type plasminogen activator (suPAR) has been shown to be a prognostic marker in sepsis. Heparin-binding protein (HBP) is a promising biomarker and as an early indicator of circulatory failure in sepsis. To date, no studies have analyzed the added value of suPAR or HBP to history and vital parameters. The hypothesis is that the identification of sepsis within ambulance care can be increased by implementing a screening tool based on a combination of vital parameters, keywords related to patients´ medical history and point of care tests. Such a screening tool is expected to enable timely treatment, and thus improve the prognosis for the septic patient. The current study aims to determine the predictive value of parameters that could be included in a sepsis screening tool tailored to the ambulance setting. Objectives: To determine the predictive value of vital parameters, keywords related to patients´ medical history and point-of-care blood tests, alone and in combination, with respect to identification of sepsis in the ambulance. Methods: This is a prospective study in the ambulance setting. Adult (≥18 years) non-trauma patients with a new onset of infection (prehospital infection group) or considered to belong to the prehospital control group (the next consecutive patient cared for in the same ambulance, immediately following the patient suspected to have a new infection but not him/ herself considered suffering from an infection) will be invited to participate. A Case Report Form (CRF, the same for both prehospital groups), including eight short yes/no-questions representing keywords related to the patients´ medical history will be used for all included patients. In addition, the CRF will prompt ambulance personnel to document the six vital parameters (respiratory rate, oxygen saturation, heart rate, systolic blood pressure, level of consciousness, temperature) in accordance to current guidelines. Level of consciousness is described according to GCS (Glasgow Coma Scale) or AVPU (Alert, Voice responsive, Pain responsive or Unresponsive)-since both scales are currently used in the ambulance. Blood tests will be taken in the ambulance (the same tests for both prehospital groups). A total of maximum 24 mL venous blood will be drawn from each included patient. P-(plasma) glucose (\<1 ml, will be analyzed in the ambulance) An additional 5 tubes of venous blood will be taken: 1. P-lactate (Sodium fluoride tube, 4 ml), 2. P-SuPAR (EDTA; Ethylenediamine tetraacetic acid tube, 4 ml), 3. P-HBP (EDTA tube, 4 ml), 4. EDTA tube 6 ml: whole blood for cytokines (caspase-1 generated interleukins: IL-1β, IL-18, IL-37), glucose metabolites and downstream regulators of immunometabolic processes (pyruvate, lactate, HIF; Hypoxia inducible factor, mTOR; mammalian target of rapamycin etc) and 5. PAXgene blood RNA (ribonucleic acid) tube 2,5 ml: whole blood for priming step of the inflammasome mRNA levels of pro-IL and NLRP3 (NOD-like receptor family, pyrin domain containing 3), etc. Inflammatory markers in tube 4 and 5 are included in a sub study performed in collaboration with Örebro University. Tube 1-5 (P-lactate, P-SuPAR, P-HBP and inflammatory markers) will, be centrifuged and frozen at the receiving hospital (weekdays 15:00 pm-8:00 am and weekends) or the Karolinska University Hospital study laboratory in Solna (weekdays daytime) depending on time of day and day of the week. The tests will be stored at -70 °C at Karolinska University Hospital study laboratory until analysis. The ED nurse receiving the ambulance will be collect the blood tests taken in the ambulance. CRFs and informed consent forms will be left in a, for the study specific purpose labeled mailbox, and is present in the ED. Inclusion in the study and the drawing of blood tests will be documented in the patient´s ambulance record. Except for P-Glucose, no laboratory results will be registered in the patients´ medical records. Positive outcome ("cases") are defined as presence of sepsis within 36 hours from ED arrival, according to predefined criteria, based on an analysis of medical records from the receiving hospital. The definition of sepsis is in accordance with Sepsis-3. This outcome measure will be compared with that of the definition of severe sepsis according to the traditional criteria, i.e. based on SIRS criteria and according to the investigators´ previously developed severe sepsis criteria, adapted to emergency care. There are three possible categories; 1. Sepsis, 2. Infection but not sepsis, and 3. No infection. Preliminary calculation of sample size and statistics: According to sample size calculations; 210 patients with the outcome sepsis is sufficient for analysis of the predictive value of 21 variables (eight keywords related to patients´ medical history, six vital parameters, glucose, lactate, SuPAR and HBP in addition to 3 demographic variables: age, gender, Charlson comorbidity score) in the final multivariate regression model, as ten events are needed for each predictor variable in logistic regression. The exact size of the required study sample is not possible to determine before the pilot study has been conducted, since the variation is not known. Models for multivariate regression analysis are based on, and require, variation of the analyzed variable within the population. If there is no such variation, only Fischer´s test can be performed. Again, this will be determined after the pilot study. Pilot study: A pilot study of 100 patients will be conducted to test the CRF, analyze the logistical aspects of the study, estimate the required duration of the study period, and to estimate variation of variables within the population. The pilot study is planned to be initiated in April 2017, and to involve the northern ambulance stations of the ambulance entrepreneur Samariten Ambulans AB in Stockholm. After the pilot: Following the pilot, the study is planned to be performed county-wide, ie in Stockholm Count Council, involving all three ambulance entrepreneurs (AISAB; Ambulansen i Storstockholm AB, Falck ambulans AB, Samariten Ambulans AB) and all seven emergency hospitals (Södersjukhuset, Karolinska Huddinge, Karolinska Solna, St Göran, Danderyd, Norrtälje, Södertälje) of Stockholm.