Study Design

This study had a prospective observational design and was part of a multicenter project that included patients who called the Swedish emergency number (112) and were assessed at the scene by ACs from the beginning of March 2018 through the end of September 2019.

Study Setting

Patients in this study were recruited from one AS in an urban area, the city of Gothenburg, Sweden. The catchment area is approximately 930m ^{Reference Fleischmann, Scherag and Adhikari2} with 690,000 inhabitants. The AS comprises seven ambulance stations and responds to over 80,000 missions annually, of which 59,000 are considered primary missions with initial patient assessments. According to the national requirements, each ambulance unit consists of at least one AC who is a registered nurse (RN) and one emergency medical technician, or alternatively, two RNs. ^{Reference Lindström, Bohm and Kurland31,Reference Suserud32} In Sweden, ACs most commonly assess patients with the aid of a decision support system called the Rapid Triage and Treatment System (RETTS) for adults. This system is based on vital signs (blood pressure, heart rate, respiratory rate, oxygen saturation, degree of consciousness, and body temperature) and Emergency Symptoms and Signs (ESS; assessment of a patient’s condition and medical severity based on anamnesis, symptoms experienced, and signs of illness or injury). A patient’s level of severity is determined as one of four different priority colors used in the AS setting. Red indicates a life-threatening status, orange indicates a potentially life-threatening status, and yellow and green indicate that a patient’s care can wait. Yellow is considered more urgent than green. The priority color is determined based on deviations in either vital signs or ESS. ^{Reference Widgren and Jourak33}

Study Population

In the database of the multicenter project, POC serum lactate was measured in 4,775 patients by the ACs during a 14-month period. During this period, ACs were urged to measure POC serum lactate in all patients who fulfilled the following inclusion criteria: (1) patients aged ≥16 years; and (2) patients assessed by the RETTS as having red, orange, yellow, or green levels if their respiratory rate >22 breaths/minute.

Red, orange, and yellow RETTS were included because they were considered to need more urgent emergency medical care. Respiratory rate >22 breaths/minute was used to identify patients who were considered having a potential need for emergency hospital care. A respiratory rate >22 breaths/minute is one of the criteria in the instrument Quick Sequential Sepsis Related Organ Failure Assessment. ^{Reference Song, Sin, Park, Shim and Lee34}

From the database, 3,603 patients were excluded in the primary analysis for the following reasons: (1) received care at another AS organization; (2) incorrect registration in the database; (3) missing social security number; (4) cardiac arrest/death on arrival; (5) plasma glucose not recorded by the AS; (6) admitted to another hospital where data were not retrievable; and (6) not admitted for in-patient care and remained at the scene.

After exclusion, data from a total of 1,172 patients who were admitted for in-patient care by an ED physician were manually reviewed in the hospital records. Of these patients, 115 were further excluded for the following reasons: patients for whom the same patient assignment had been evaluated twice, and patients with status epilepticus for whom the physician in the ED did not suspect an infection. Patients with status epilepticus for whom an infection was not suspected were excluded because serum lactate levels from these patients could be very high and give misleading results. ^{Reference Magnusson, Herlitz and Höglind35} A total sample of 1,057 patients was included in the primary analysis (Figure 1). Of these, 253 patients were retrospectively classified by the first author as having sepsis based on the Sepsis-3 criteria. ^{Reference Singer, Deutschman and Seymour14}

Figure 1.

Flow Chart of Inclusion and Exclusion.

Abbreviations: AS, ambulance service; POC, point-of-care; ED, emergency department.

In the secondary analysis, patients who remained at the scene and were not admitted to in-patient care were added (n = 907). Of these, 123 patients were excluded for the following reasons: POC plasma glucose missing (n = 13), and patient with status epilepticus in the AS (n = 110) setting. A total sample of 784 was further included. These patients were classified as not having Sepsis-3, based on that certain blood tests required to be classified as having Sepsis-3 were missing. This leads to a high risk of uncertainty whether the patients fulfil the criteria for Sepsis-3 or not.

Data Collection

Biomarkers and Vital Signs—Biomarkers and vital signs were prospectively collected by ACs and registered in the AS electronic medical register (Ambulink; 10Bitdevelopment Co.; Irving, Texas USA). The first measured levels of POC plasma glucose and serum lactate, and the vital signs respiratory rate, oxygen saturation, heart rate, systolic blood pressure, and temperature, measured by the AC on site, were included. Serum lactate was measured with the StatStrip Xpress (SSX) POC device (Nova Biomedical; Waltham, Massachusetts USA). The SSX measures whole blood lactate levels within a range of 0.3 to 20.0mmol/l. The feasibility of the device has been examined in the AS setting, with acceptable concentrations compared to those measured by standard laboratory devices. ^{Reference Léguillier, Jouffroy and Boisson36–Reference Colon-Franco, Lo, Tarima, Gourlay, Drendel and Brook Lerner38} Plasma glucose was measured with the POC Precision Xtra (Abbott Medical; Abbott Park, Illinois USA). Plasma glucose measurements are part of standard practice in the AS setting and were in use before the start of this study. The device measures glucose in the plasma within a range of 1.1 to 28.7mmol/l.

Demographic Information—Information on age, sex, and comorbidities was retrospectively collected from the hospital medical record system (Melior; Oracle Cerner Sweden; Gothenburg, Sweden).

Evaluation of Sepsis-3 Criteria According to the SOFA Score—Data for the evaluation of the Sepsis-3 criteria were retrospectively collected from the AS electronic medical record database (Ambulink) and hospital medical records (Melior). Sepsis-3 was defined within 48 hours from the arrival of the patient by AS in accordance with the Sepsis-3 criteria ^{Reference Singer, Deutschman and Seymour14} (ie, suspected infection in combination with an increased Sequential Organ Failure Assessment [SOFA] score of two points or higher compared with the patient’s baseline status and based on a review of their medical records). The baseline SOFA score was assumed to be zero unless the patient was known to have pre-existing (acute or chronic) organ dysfunction before the onset of infection. ^{Reference Singer, Deutschman and Seymour14} According to the Swedish guidelines, Sepsis-3 should be defined within 36 hours of arrival at the hospital, ³⁹ and according to Seymour, et al, ^{Reference Seymour, Liu and Iwashyna40} a change in the Sepsis-3 criteria can be calculated from 48 hours before to 24 hours after the onset of infection. Since Sepsis-3 was also defined based on vital signs from the AS setting, Sepsis-3 was chosen to be defined within 48 hours of the arrival of the AS. A suspected infection was defined from the physician’s final note in the ED medical record, for example, suspected pneumonia, a urinary tract infection, or erysipelas. The included patients were also verified to have been diagnosed and treated for an infection during the hospital stay by the discharge note in their medical record. To calculate the SOFA score, the Strama Nationell (Softwerk; Växjö, Sweden) app was used. This app converted the oxygen saturation level to the partial pressure of oxygen (PaO2)/fraction of inspired oxygen (FiO2) and the reaction level scale (RLS) score to the Glasgow Coma Scale (GCS) score, ^{Reference Nationell41,Reference Walther, Jonasson and Gill42} since the RLS and not the GCS is most commonly used to measure mental status in the included organization. Patients who did not meet the Sepsis-3 criteria as described were classified as having “other diagnoses.” To ensure the reliability of the classification of the Sepsis-3 criteria, a total of 100 patients were reclassified, and inter-reliability was tested with Cohen’s kappa. ^{Reference McHugh43} The result of the analysis showed an accuracy of 0.857.

Statistical Analysis

For the primary study population, two-group comparisons were analyzed. The Mann-Whitney U test was used for the continuous variables (age, biomarkers, and vital signs) because the continuous variables were non-normally distributed. Data were checked for normality using histograms and the Kolmogorov–Smirnov test. For two-group comparisons of categorical variables, chi-square was used for gender and Fisher’s exact test was used for comorbidity. Results are presented as numbers (percentages) or medians (25th and 75th percentiles).

For both the primary and secondary study populations, the prognostic value of prehospital lactate and glucose levels for Sepsis-3 was evaluated using receiver operating characteristic (ROC) analysis. The values of the areas under the ROC curve (AUC) were evaluated. The optimal cut-off values for plasma glucose and serum lactate were determined using the Youden index. For both the primary and secondary study populations, sensitivity, specificity, and positive and negative predictive values (PPV and NPV) for plasma glucose and serum lactate levels for optimal cut-offs were also calculated. An AUC of 0.5 to <0.7 for plasma glucose and serum lactate levels was considered to indicate poor accuracy, an AUC ≥0.7-0.8 was considered to indicate acceptable accuracy, an AUC ≥0.8-0.9 was considered to indicate excellent accuracy, and an AUC ≥0.9 was considered to indicate outstanding accuracy. ^{Reference Hosmer, Lemeshow, Sturdivant, Hosmer, Lemeshow and Sturdivant44} A P value of <.05 was considered to indicate a significant difference between groups. MedCalc for Windows, version 20.014 (MedCalc Statistical Software; Ostend, Belgium) was used to calculate the sensitivity, specificity, PPV, and NPV, and SPSS version 20.0 (IBM SPSS Statistics for Windows, Version 20.0; Armonk, New York USA) was used for other statistical analyses.

Ethics Approval and Consent to Participate

The study was approved by the regional ethical review board in Gothenburg (No: 2019-000808/1173-18), and the methods procedures followed the Declaration of Helsinki. ⁴⁵ In this study, data were prospectively recorded in the ambulance by the ACs. All patients and/or their relatives were asked whether their identity should remain confidential before being reported in the ambulance patient data records. The patients who asked for their data to remain confidential were not included in the retrospective analysis. However, informed consent was not obtained from the patients for the retrospective analysis of their data. The ethical review board in Gothenburg (No: 2019-000808/1173-18) waived the need for informed consent for the following reasons: (1) lactate levels are already measured in the ED, and thus, the test is only brought forward in the AS phase; (2) the study was a quality evaluation by a health care provider; (3) patients with serious conditions could never be contacted retrospectively, as they either had died or were in poor clinical condition, which would increase the risk of selection bias and the reliability of the results - an approach to patients or their relatives with issues such as these may create anxiety and thus be regarded as unethical; and (4) all patients under the care of the AS have the right to refuse tests based on Swedish legislation.