Search

Abstract Accurate sepsis diagnosis is paramount for treatment decisions, especially at the emergency department (ED). To improve diagnosis, clinical decision support (CDS) tools are being developed with machine learning (ML) algorithms, using a wide range of variable groups. ML models can find patterns in Electronic Health Record (EHR) data that are unseen by the human eye. A prerequisite for a good model is the use of high-quality labels. Sepsis gold-standard labels are hard to define due to a lack of reliable diagnostic tools for sepsis at the ED. Therefore, standard clinical tools, such as clinical prediction scores (e.g. modified early warning score and quick sequential organ failure assessment), and claims-based methods (e.g. ICD-10) are used to generate suboptimal labels. As a consequence, models trained with these “silver” labels result in ill-trained models. In this study, we trained ML models for sepsis diagnosis at the ED with labels of 375 ED visits assigned by an endpoint adjudication committee (EAC) that consisted of 18 independent experts. Our objective was to evaluate which routinely measured variables show diagnostic value for sepsis. We performed univariate testing and trained multiple ML models with 95 routinely measured variables of three variable groups; demographic and vital, laboratory and advanced haematological variables. Apart from known diagnostic variables, we identified added diagnostic value for less conventional variables such as eosinophil count and platelet distribution width. In this explorative study, we show that the use of an EAC together with ML can identify new targets for future sepsis diagnosis research.

Objectives To evaluate the prognostic value of the coefficient of variance of axial light loss of monocytes (cv-ALL of monocytes) for adverse clinical outcomes in patients suspected of infection in the emergency department (ED). Methods We performed an observational, retrospective monocenter study including all medical patients ≥18 years admitted to the ED between September 2016 and June 2019 with suspected infection. Adverse clinical outcomes included 30-day mortality and ICU/MCU admission Results A total of 3526 of patients were included. The OR for cv-ALL of monocytes alone was 2.21 (1.98–2.47) for 30-day mortality and 2.07 (1.86–2.29) for ICU/MCU admission Conclusions Cv-ALL of monocytes is a readily available biomarker that is useful as prognostic marker to predict 30-day mortality. Furthermore, it can be used to improve routine prediction of adverse clinical outcomes at the ED. Clinical trial registration Registered in the Dutch Trial Register (NTR) und number 6916.

Early recognition of sepsis is essential for improving outcomes and preventing complications such as organ failure, depression, and neurocognitive impairment. The emergency department (ED) plays a key role in the early identification of sepsis, but clinicians lack diagnostic tools. Potentially, biomarkers could be helpful in assisting clinicians in the ED, but no marker has yet been successfully implemented in daily practice with good clinical performance. Pancreatic stone protein (PSP) is a promising biomarker in the context of sepsis, but little is known about the diagnostic performance of PSP in the ED. We prospectively investigated the diagnostic value of PSP in such a population for patients suspected of infection. PSP was compared with currently used biomarkers, including white blood cell count (WBC) and C-reactive protein (CRP). Of the 156 patients included in this study, 74 (47.4%) were diagnosed with uncomplicated infection and 26 (16.7%) patients with sepsis, while 56 (35.9%) eventually had no infection. PSP was significantly higher for sepsis patients compared to patients with no sepsis. In multivariate regression, PSP was a significant predictor for sepsis, with an area under the curve (AUC) of 0.69. Positive and negative predictive values for this model were 100% and 84.4%, respectively. Altogether, these findings show that PSP, measured at the ED of a tertiary hospital, is associated with sepsis but lacks the diagnostic performance to be used as single marker.

Objectives The COVID-19 pandemic pressurised healthcare with increased shortage of care. This resulted in an increase of awareness for code status documentation (ie, whether limitations to specific life-sustaining treatments are in place), both in the medical field and in public media. However, it is unknown whether the increased awareness changed the prevalence and content of code status documentation for COVID-19 patients. We aim to describe differences in code status documentation between infectious patients before the pandemic and COVID-19 patients.Setting University Medical Centre of Utrecht, a tertiary care teaching academic hospital in the Netherlands.Participants A total of 1715 patients were included, 129 in the COVID-19 cohort (a cohort of COVID-19 patients, admitted from March 2020 to June 2020) and 1586 in the pre-COVID-19 cohort (a cohort of patients with (suspected) infections admitted between September 2016 to September 2018).Primary and secondary outcome measures We described frequency of code status documentation, frequency of discussion of this code status with patient and/or family, and content of code status.Results Frequencies of code status documentation (69.8% vs 72.7%, respectively) and discussion (75.6% vs 73.3%, respectively) were similar in both cohorts. More patients in the COVID-19 cohort than in the before COVID-19 cohort had any treatment limitation as opposed to full code (40% vs 25%). Within the treatment limitations, ‘no intensive care admission’ (81% vs 51%) and ‘no intubation’ (69% vs 40%) were more frequently documented in the COVID-19 cohort. A smaller difference was seen in ‘other limitation’ (17% vs 9%), while ‘no resuscitation’ (96% vs 92%) was comparable between both periods.Conclusion We observed no difference in the frequency of code status documentation or discussion in COVID-19 patients opposed to a pre-COVID-19 cohort. However, treatment limitations were more prevalent in patients with COVID-19, especially ‘no intubation’ and ‘no intensive care admission’.