Pre-hospital antibiotic therapy preceded by blood cultures in a physician-manned mobile emergency care unit.

BACKGROUND: Rapid recognition and antibiotic treatment, preferably preceded by blood cultures (BCs), is a mainstay in sepsis therapy. The objective of this investigation was to determine if pre-hospital BCs were feasible and drawn with an acceptably low level of contamination and to investigate whether pre-hospital antibiotics were administered on correct indications. METHODS: We performed a register-based study in a pre-hospital physician-manned mobile emergency care unit (MECU) operating in a mixed urban/rural area in Denmark. All patients who received pre-hospital antibiotics by the MECU from November 2013 to October 2018 were reviewed. Outcome measures were characterisation of microbial findings and subsequent in-hospital confirmation of the pre-hospital indication for antibiotics. RESULTS: One-hundred-and-nineteen patients received antibiotics pre-hospitally. Six were excluded. One-hundred-and-thirteen patients were included in the study. BCs were drawn in 107 of the 113 patients (94.7% [88.8%-98.0%]). We found a true pathogen of sepsis in 29 (27.1% [19.0%-36.6%]) of these 107 patients. Nine (8.4% [3.9%-15.4%]) patients had contaminated pre-hospital BCs. Forty-nine of all patients (36.3% [27.4%-45.9%]) had causative pathogens in either their BCs or other samples confirming the pre-hospital tentative diagnosis. Eighty-two (72.6% [63.4%-80.5%]) patients received antibiotic therapy in-hospitally, while 27 (23.9% [16.4%-32.8%]) were assigned an in-hospital diagnosis not associated with infection. Four (3.5% [1.0%-8.8%]) patients died in hospital before a diagnosis was established. CONCLUSIONS: Pre-hospital administration of antibiotics preceded by BCs is feasible, although with somewhat high blood culture contamination rates. Antibiotics are administered on reasonable indications.

Patterns of cerebral tissue oxygen tension and cytoplasmic redox state in bacterial meningitis.

BACKGROUND: Compromised cerebral energy metabolism is common in patients with bacterial meningitis. In this study, simultaneous measurements of cerebral oxygen tension and lactate/pyruvate ratio were compared to explore whether disturbed energy metabolism was usually caused by insufficient tissue oxygenation or compromised oxidative metabolism of pyruvate indicating mitochondrial dysfunction. SUBJECT AND METHODS: Ten consecutive patients with severe streptococcus meningitis were included in this prospective cohort study. Intracranial pressure, brain tissue oxygen tension (PbtO2 ), and energy metabolism (intracerebral microdialysis) were continuously monitored in nine patients. A cerebral lactate/pyruvate (LP) ratio <30 was considered indicating normal oxidative metabolism, LP ratio >30 simultaneously with pyruvate below lower normal level (70 µmol/L) was interpreted as biochemical indication of ischemia, and LP ratio >30 simultaneously with a normal or increased level of pyruvate was interpreted as mitochondrial dysfunction. The biochemical variables were compared with PbtO2 simultaneously monitored within the same cerebral region. RESULTS: In two cases, the LP ratio was normal during the whole study period and the simultaneously monitored PbtO2 was 18 ± 6 mm Hg. In six cases, interpreted as mitochondrial dysfunction, the simultaneously monitored PbtO2 was 20 ± 6 mm Hg and without correlation with the LP ratio. In one patient, exhibiting a pattern interpreted as ischemia, PbtO2 decreased below 10 mm Hg and a correlation between LP and PbtO2 was observed. CONCLUSION: This study demonstrated that compromised cerebral energy metabolism, evidenced by increased LP ratio, was common in patients with severe bacterial meningitis while not related to insufficient tissue oxygenation.

Predictors for outcome among cardiac arrest patients: the importance of initial cardiac arrest rhythm versus time to return of spontaneous circulation, a retrospective cohort study.

BACKGROUND: In the past decade, early treatment of cardiac arrest (CA) victims has been improved in several ways, leading to more optimistic over all prognoses. However, the global survival rate after out-of-hospital CA (OHCA) is still not more than 5-10%. With a better knowledge of the predictors for outcome among CA patients, we can improve the management of CA, in order to strengthen the leads in the chain of survival. METHODS: A retrospective cohort study including 172 CA patients admitted to the intensive care unit (ICU) in Odense University Hospital (OUH) in a three-year period was conducted. We determined the 90-day mortality and neurological outcome at discharge for CA patients treated with therapeutic hypothermia (TH), in regard to determine the importance of the predictors for mortality and neurological outcome, with emphasize on combining initial rhythm and time to return of spontaneous circulation (ROSC). RESULTS: The overall mortality was 44% and a favorable neurological outcome was seen among 52%. Strong predictors for survival and favorable neurological outcome were ventricular tachycardia/ventricular fibrillation (VT/VF) as initial rhythm, cardiac etiology and time to ROSC < 20 minutes. Age < 60 years was a predictor for survival only. Patients with the combination of VT/VF and ROSC < 20 minutes had undeniably the best chance of both survival and a favorable neurological outcome. CONCLUSIONS: We found significant predictors for both survival and neurological outcome, in which an initial rhythm of VT/VF and a cardiac etiology were the strongest.

Discrete-time survival analysis in the critically ill: a deep learning approach using heterogeneous data.

Prediction of survival for patients in intensive care units (ICUs) has been subject to intense research. However, no models exist that embrace the multiverse of data in ICUs. It is an open question whether deep learning methods using automated data integration with minimal pre-processing of mixed data domains such as free text, medical history and high-frequency data can provide discrete-time survival estimates for individual ICU patients. We trained a deep learning model on data from patients admitted to ten ICUs in the Capital Region of Denmark and the Region of Southern Denmark between 2011 and 2018. Inspired by natural language processing we mapped the electronic patient record data to an embedded representation and fed the data to a recurrent neural network with a multi-label output layer representing the chance of survival at different follow-up times. We evaluated the performance using the time-dependent concordance index. In addition, we quantified and visualized the drivers of survival predictions using the SHAP methodology. We included 37,355 admissions of 29,417 patients in our study. Our deep learning models outperformed traditional Cox proportional-hazard models with concordance index in the ranges 0.72-0.73, 0.71-0.72, 0.71, and 0.69-0.70, for models applied at baseline 0, 24, 48, and 72 h, respectively. Deep learning models based on a combination of entity embeddings and survival modelling is a feasible approach to obtain individualized survival estimates in data-rich settings such as the ICU. The interpretable nature of the models enables us to understand the impact of the different data domains.

Serum levels of neurofilament light chain, neuron-specific enolase and S100 calcium-binding protein B during acute bacterial meningitis: a prospective cohort study.

PURPOSE: Acute bacterial meningitis (ABM) is a severe disease with an overall poor outcome. Neurofilament (NFL) has shown to be a promising biomarker of neuroaxonal injury in various neurological disorders but has not been investigated in ABM. The aims of this study were (i) to obtain a temporal profile of NFL, neuron-specific enolase (NSE) and S100B in serum during ABM, and (ii) to evaluate their use as biomarkers of severity (Glasgow coma score) and prognosis (Glasgow Outcome Score, GOS and death) in severe ABM. METHODS: Fifteen adults with severe community-acquired ABM who were admitted to the intensive care unit (ICU) and fulfilled the inclusion criteria were included. Lumbar puncture and blood tests were performed on admission, and blood tests were performed three times daily during the ICU stay. GOS was obtained day 30. RESULTS: Serum NFL was significantly elevated in ABM patients compared to healthy controls, both at admission and throughout the observation period (p < .01). NFL increased significantly from day 1 up to day 3-6 (p < .0001), peaking day 6. NSE increased significantly from admission up to day 3 (p < .01). At day 5-6, the serum values were not significantly different from values at admission. The highest median serum value of S100B was observed at admission (0.10 µg/L, IQR 0.06-0.14), significantly decreasing day 4-6 (p < .05). None of the investigated biomarkers revealed significant correlation with severity and prognosis. CONCLUSION: This study represents a first clinical observation of the temporal profile of NFL in serum, in severe ABM. No correlation with severity or prognosis.

Dynamic and explainable machine learning prediction of mortality in patients in the intensive care unit: a retrospective study of high-frequency data in electronic patient records.

BACKGROUND: Many mortality prediction models have been developed for patients in intensive care units (ICUs); most are based on data available at ICU admission. We investigated whether machine learning methods using analyses of time-series data improved mortality prognostication for patients in the ICU by providing real-time predictions of 90-day mortality. In addition, we examined to what extent such a dynamic model could be made interpretable by quantifying and visualising the features that drive the predictions at different timepoints. METHODS: Based on the Simplified Acute Physiology Score (SAPS) III variables, we trained a machine learning model on longitudinal data from patients admitted to four ICUs in the Capital Region, Denmark, between 2011 and 2016. We included all patients older than 16 years of age, with an ICU stay lasting more than 1 h, and who had a Danish civil registration number to enable 90-day follow-up. We leveraged static data and physiological time-series data from electronic health records and the Danish National Patient Registry. A recurrent neural network was trained with a temporal resolution of 1 h. The model was internally validated using the holdout method with 20% of the training dataset and externally validated using previously unseen data from a fifth hospital in Denmark. Its performance was assessed with the Matthews correlation coefficient (MCC) and area under the receiver operating characteristic curve (AUROC) as metrics, using bootstrapping with 1000 samples with replacement to construct 95% CIs. A Shapley additive explanations algorithm was applied to the prediction model to obtain explanations of the features that drive patient-specific predictions, and the contributions of each of the 44 features in the model were analysed and compared with the variables in the original SAPS III model. FINDINGS: From a dataset containing 15 615 ICU admissions of 12 616 patients, we included 14 190 admissions of 11 492 patients in our analysis. Overall, 90-day mortality was 33·1% (3802 patients). The deep learning model showed a predictive performance on the holdout testing dataset that improved over the timecourse of an ICU stay: MCC 0·29 (95% CI 0·25-0·33) and AUROC 0·73 (0·71-0·74) at admission, 0·43 (0·40-0·47) and 0·82 (0·80-0·84) after 24 h, 0·50 (0·46-0·53) and 0·85 (0·84-0·87) after 72 h, and 0·57 (0·54-0·60) and 0·88 (0·87-0·89) at the time of discharge. The model exhibited good calibration properties. These results were validated in an external validation cohort of 5827 patients with 6748 admissions: MCC 0·29 (95% CI 0·27-0·32) and AUROC 0·75 (0·73-0·76) at admission, 0·41 (0·39-0·44) and 0·80 (0·79-0·81) after 24 h, 0·46 (0·43-0·48) and 0·82 (0·81-0·83) after 72 h, and 0·47 (0·44-0·49) and 0·83 (0·82-0·84) at the time of discharge. INTERPRETATION: The prediction of 90-day mortality improved with 1-h sampling intervals during the ICU stay. The dynamic risk prediction can also be explained for an individual patient, visualising the features contributing to the prediction at any point in time. This explanation allows the clinician to determine whether there are elements in the current patient state and care that are potentially actionable, thus making the model suitable for further validation as a clinical tool. FUNDING: Novo Nordisk Foundation and the Innovation Fund Denmark.

Diagnostic Performance of T2Candida Among ICU Patients With Risk Factors for Invasive Candidiasis.

BACKGROUND: Invasive candidiasis (IC) comprises candidemia and deep-seated candidiasis. Blood culture (BC) is the gold standard test, but sensitivity is low. T2Candida is a new diagnostic test. We investigated the performance of T2Candida, BC, and Candida mannan antigen (MAg) for detection of IC in a high-risk intensive care unit (ICU) population. METHODS: One-hundred twenty-six ICU patients at high risk of IC with sepsis despite 3 days of broad-spectrum antibiotics were included. Paired BC, T2Candida, and MAg were obtained twice weekly (334 sets). Patients were classified into proven, likely, possible, or unlikely IC based on patient record review. RESULTS: At enrollment, 92 (77%) patients were receiving antifungal therapy (mainly fluconazole 66%). Fifteen (11.9%) patients were positive by BC (n = 4), T2Candida (n = 11), or MAg (n = 10). The T2Candida species distribution at inclusion (Candida albicans/Candida tropicalis: 8/11 [72.3%] and Candida glabrata/Candida krusei: 3/11 [27.3%]) was supported by the identification of BC or colonizing isolates in 10/11 cases. Patients were classified with proven (11), likely (6), possible (11), and unlikely (98) IC. Defining IC as proven/proven&likely/proven&likely&possible, respectively, the sensitivity was as follows: T2Candida (55%/59%/39%), BC (45%/29%/ 8%), and MAg (36%/41%/32%). The negative predictive value was similar across the tests for proven vs others and proven/likely vs others (94%-96% and 90%-95%, respectively). For test combinations including T2Candida, the sensitivity increased to 64%-65%, without hampering the positive predictive value. CONCLUSIONS: In conclusion, although the diagnostic performance was modest for all the tests, the combination of T2Candida and BC seemed to have the best diagnostic performance, and thus implementation of T2Candida may improve the diagnosis of IC.

Antibiotic therapy as personalized medicine - general considerations and complicating factors.

The discovery of antibiotic drugs is considered one of the previous century's most important medical discoveries (Medicine's 10 greatest discoveries. New Haven, CT: Yale University Press, 1998: 263). Appropriate use of antibiotics saves millions of lives each year and prevents infectious complications for numerous people. Still, infections kill unacceptable many people around the world, even in developed countries with easy access to most antibiotic drugs. Optimal use of antibiotics is dependent on the identification of primary and secondary focus, and knowledge on which pathogens to expect in a specific infectious syndrome and information on general patterns of regional antibiotic resistance. Furthermore, sampling for microbiological analysis, knowledge of patient immune status and organ functions, travel history, pharmacokinetics and -dynamics of the different antibiotics and possible biofilm formation are among several factors involved in antibiotic therapy of infectious diseases. The present review aims at describing important considerations when using antibacterial antibiotics and to describe how this is becoming substantially more personalized. The parameters relevant in considering the optimal use of antibiotics to treat infections are shown in Fig. 1 - leading to the most relevant antibiotic therapy for that specific patient. To illustrate this subject, the present review's focus will be on challenges with optimal dosing of antibiotics and risks of underdosing. Especially, in cases highly challenging for achieving the aimed antibiotic effect against bacterial infections - this includes augmented renal clearance (ARC) in sepsis, dosing challenges of antibiotics in pregnancy and against biofilm infections.

[Disease burden and definition of sepsis in adults].

Sepsis is the terminal event for most infectious diseases and is now defined as life-threatening organ dysfunction caused by a dysregulated host response to infection. Sepsis is frequent, deathly and associated with high burden for patients, relatives and societies. Sepsis is underreported in diagnostic coding, making estimation difficult of the true burden of the disease in Denmark. It is likely, however, that sepsis contributes to 15% of all deaths in Denmark.

[Augmented renal clearance of antibiotics may lead to inadequate dosings for patients with sepsis]. / Forøget renal udskillelse af antibiotika kan resultere i underdosering ved sepsis.

Early, relevant antibiotic treatment is of pivotal significance for survival for patients with sepsis. Recent reports reveal, that critically ill patients may need substantially higher dosing of antibiotics than the usual standards. The background is the newly described augmented renal clearance (ARC), which can be observed in critically ill patients with sepsis. ARC results in significantly reduced antibiotic concentrations in patients with sepsis and can potentially have fatal consequences. The purpose of this paper is to make medical doctors aware of the phenomenon and to compensate for the increased antibiotic clearance.

The pharmacokinetic profile of plasma-derived mannan-binding lectin in healthy adult volunteers and patients with Staphylococcus aureus septicaemia.

Mannan-binding lectin (MBL) is a member of the innate immune system, and MBL-deficiency affects 10-15% of Caucasians. With development of a plasma-derived MBL, substitution has become a therapeutic option in diseases associated with MBL insufficiency. The pharmacokinetics of injected MBL is weakly described, particularly in patients with infectious diseases. The pharmacokinetic profile of MBL following administration of 0.08 mg/kg to 20 healthy MBL-deficient volunteers and 0.2 mg/kg to 2 patients with Staphylococcus aureus septicaemia was established. In the volunteers, the maximal concentration was 2849 microg/l; the mean half-life (T(1/2)) was 69.6 h (14.6-114.9 h). The normalized clearance was 9x10(-6) l/minxkg, and the mean residence time was 82 h. In the patients the serum-MBL versus time curves were similar to those in the volunteers, and T(1/2) values were 36 and 40 h. In conclusion, MBL is distributed into a median volume of 3.4 l similar to the plasma volume, and the elimination in septicaemic patients was within the range of the controls. Due to the large individual variation in T(1/2), we recommend that MBL therapy, with respect to dose and infusion intervals, is based on the chosen therapeutic target (> or =1000 microg/l) and MBL serum determinations following the first infusion.

[Antibiotic strategies in the treatment of infection in critically ill patients]. / Antibiotikastrategi ved behandling af systemiske infektioner hos kritisk syge patienter.

Inappropriate antimicrobial therapy of pneumonia, severe sepsis and bacteraemia is associated with high mortality. Adjustment of insufficient initial antibiotics after susceptibility testing does not reduce mortality. Guidelines on critically ill patients should focus on antibiograms for each intensive care unit to ensure full coverage of initial therapy with a broad-spectrum antibiotic with high tissue penetration, minimal organ toxicity and low risk of resistance development. Early reassessment and withdrawal of antibiotics, if not indicated, are important elements in reducing antibiotic resistance. The initial broad-spectrum should be narrowed according to bacterial susceptibility data. This evidence-based intervention ensures appropriate antimicrobial treatment to critically ill patients to improve outcome and reduction of the use of broad-spectrum antibiotics.