Introduction: The adverse effects of delayed admission to the intensive care unit (ICU) have been recognized in previous studies. However, the definitions of delayed admission vary across studies. This study proposed a model to define `delayed admission’, and explored the effect of ICU waiting time on patients’ outcome. Methods: This retrospective cohort study included nontraumatic adult patients on mechanical ventilation in the emergency department (ED), from July 2009 to June 2010. The primary outcomes measures were 21-ventilator-day mortality and prolonged hospital stays (over 30 days). Models of Cox regression and logistic regression were used for multivariate analysis. The non-delayed ICU waiting was defined as a period in which the time effect on mortality was not statistically significant in a Cox regression model. To identify a suitable cutoff point between `delayed’ and `non-delayed’ subsets from the overall data were made based on ICU waiting time and the hazard ratio of ICU waiting hour in each subset was iteratively calculated. The cutoff time was then used to evaluate the impact of delayed ICU admission on mortality and prolonged length of hospital stay. Results: The final analysis included 1,242 patients. The time effect on mortality emerged after 4 hours, thus we deduced ICU waiting time in the ED of >4 hours as delayed. By logistic regression analysis, delayed ICU admission affected the outcomes of 21-ventilator-day mortality and prolonged hospital stay, with an odds ratio of 1.41 (95% confidence interval, 1.05 to 1.89) and 1.56 (95% confidence interval, 1.07 to 2.27) respectively. Conclusions: For patients on mechanical ventilation in the ED, delayed ICU admission is associated with higher probability of mortality and additional resource expenditure. A benchmark waiting time of no more than 4 hours for ICU admission is recommended.

In emergency departments, the most common cause of death associated with suspected infected patients is sepsis. In this study, deep learning algorithms were used to predict the mortality of suspected infected patients in a hospital emergency department. During January 2007 and December 2013, 42,220 patients considered in this study were admitted to the emergency department due to suspected infection. In the present study, a deep learning structure for mortality prediction of septic patients was developed and compared with several machine learning methods as well as two sepsis screening tools: the systemic inflammatory response syndrome (SIRS) and quick sepsis-related organ failure assessment (qSOFA). The mortality predictions were explored for septic patients who died within 72 h and 28 days. Results demonstrated that the accuracy rate of deep learning methods, especially Convolutional Neural Network plus SoftMax (87.01% in 72 h and 81.59% in 28 d), exceeds that of the other machine learning methods, SIRS, and qSOFA. We expect that deep learning can effectively assist medical staff in early identification of critical patients.

Clinical differentiation of influenza from dengue and other febrile illnesses (OFI) is difficult, and available rapid diagnostic tests have limited sensitivity. We conducted a retrospective study to compare clinical and laboratory findings between (i) influenza and dengue and (ii) influenza and OFI. Of 849 enrolled patients, the mean time between illness onset and hospital presentation was 1.7, 3.7, and 3 days for influenza, dengue, and OFI, respectively. Among pediatric patients (≤18 years) (445 influenza, 24 dengue, and 130 OFI), we identified absence of rashes, no leukopenia, and no marked thrombocytopenia (platelet counts 18 years) (81 influenza, 124 dengue, and 45 OFI), no leukopenia and no marked thrombocytopenia distinguished influenza from dengue, while rhinorrhea and malaise differentiated influenza from OFI. A diagnostic algorithm developed to distinguish influenza from dengue using rash, leukopenia, and marked thrombocytopenia showed >90% sensitivity to identify influenza in pediatric patients. This study identified simple clinical and laboratory parameters that can assist clinicians to distinguish influenza from dengue and OFI. These findings may help clinicians diagnose influenza and facilitate appropriate management of affected patients, particularly in resource-poor settings.

The links of air pollutants to health hazards have been revealed in literature and inflammation responses might play key roles in the processes of diseases. WBC count is one of the indexes of inflammation, however the l iterature reveals inconsistent opinions on the relationship between WBC counts and exposure to air pollutants. The goal of this population-based observational study was to examine the associations between multiple air pollutants and WBC counts. This study recruited community subjects from Kaohsiung city. WBC count, demographic and health hazard habit data were collected. Meanwhile, air pollutants data (SO₂, NO₂, CO, PM₁₀, and O₃) were also obtained. Both datasets were merged for statistical analysis. Single- and multiple-pollutants models were adopted for the analysis. A total of 10,140 adults (43.2% males; age range, 33~86 years old) were recruited. Effects of short-term ambient concentrations (within one week) of CO could increase counts of WBC, neutrophils, monocytes, and lymphocytes. However, SO₂ could decrease counts of WBC, neutrophils, and monocytes. Gender, BMI, and smoking could also contribute to WBC count increases, though their effects are minor when compared to CO. Air pollutants, particularly SO₂, NO₂ and CO, may thus be related to alterations of WBC counts, and this would imply air pollution has an impact on human systematic inflammation.

Little is known about which methods are best for detecting adverse events in the emergency department (ED). This study compared the ability of trigger tool and reporting methods to capture adverse events in the ED and investigated the characteristics of the adverse events identified by each. This 1-year prospective observational cohort study evaluated a monitoring system that combined 2 reporting methods and 5 trigger tool methods to capture adverse events in the ED of an academic medical center. Measurement outcomes included the number, type, and physical impact of the captured adverse events. Among 69,327 adult nontrauma ED visits, 285 adverse events were identified. Of these adverse events, 77.2% were identified using reporting methods, 26% using trigger tool methods, and 3.2% using both methods. Most patients (81.7%) incurred temporary, minor physical impacts. Of the adverse events that occurred, 86.7% were related to clinical performance. Compared with reporting methods, trigger tool methods had a lower positive predictive rate to identify adverse events (odds ratio [OR], 0.1; 95% confidence interval [CI], 0.09–0.16), a greater proportion of adverse events occurring during the preinterventation and postintervention phases (OR, 17.0; 95% CI, 8.48–34.16), and more cases of severe physical impact or death (OR, 5.4; 95% CI, 2.62–11.10). The reporting methods more effectively captured greater numbers of adverse events, whereas the adverse events captured by the trigger tool methods were more likely to be severe physical impacts. The combined use of the different methods had synergistic benefits for monitoring adverse events in the ED. This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.