Dynamic prediction of mortality in COVID-19 patients in the intensive care unit: A retrospective multi-center cohort study.

Background: The COVID-19 pandemic continues to overwhelm intensive care units (ICUs) worldwide, and improved prediction of mortality among COVID-19 patients could assist decision making in the ICU setting. In this work, we report on the development and validation of a dynamic mortality model specifically for critically ill COVID-19 patients and discuss its potential utility in the ICU. Methods: We collected electronic medical record (EMR) data from 3222 ICU admissions with a COVID-19 infection from 25 different ICUs in the Netherlands. We extracted daily observations of each patient and fitted both a linear (logistic regression) and non-linear (random forest) model to predict mortality within 24 h from the moment of prediction. Isotonic regression was used to re-calibrate the predictions of the fitted models. We evaluated the models in a leave-one-ICU-out (LOIO) cross-validation procedure. Results: The logistic regression and random forest model yielded an area under the receiver operating characteristic curve of 0.87 [0.85; 0.88] and 0.86 [0.84; 0.88], respectively. The recalibrated model predictions showed a calibration intercept of -0.04 [-0.12; 0.04] and slope of 0.90 [0.85; 0.95] for logistic regression model and a calibration intercept of -0.19 [-0.27; -0.10] and slope of 0.89 [0.84; 0.94] for the random forest model. Discussion: We presented a model for dynamic mortality prediction, specifically for critically ill COVID-19 patients, which predicts near-term mortality rather than in-ICU mortality. The potential clinical utility of dynamic mortality models such as benchmarking, improving resource allocation and informing family members, as well as the development of models with more causal structure, should be topics for future research.

Experimental verification of principal losses in a regulatory particulate matter emissions sampling system for aircraft turbine engines.

A sampling system for measuring emissions of nonvolatile particulate matter (nvPM) from aircraft gas turbine engines has been developed to replace the use of smoke number and is used for international regulatory purposes. This sampling system can be up to 35 m in length. The sampling system length in addition to the volatile particle remover (VPR) and other sampling system components lead to substantial particle losses, which are a function of the particle size distribution, ranging from 50 to 90% for particle number concentrations and 10-50% for particle mass concentrations. The particle size distribution is dependent on engine technology, operating point, and fuel composition. Any nvPM emissions measurement bias caused by the sampling system will lead to unrepresentative emissions measurements which limit the method as a universal metric. Hence, a method to estimate size dependent sampling system losses using the system parameters and the measured mass and number concentrations was also developed (SAE 2017; SAE 2019). An assessment of the particle losses in two principal components used in ARP6481 (SAE 2019) was conducted during the VAriable Response In Aircraft nvPM Testing (VARIAnT) 2 campaign. Measurements were made on the 25-meter sample line portion of the system using multiple, well characterized particle sizing instruments to obtain the penetration efficiencies. An agreement of ± 15% was obtained between the measured and the ARP6481 method penetrations for the 25-meter sample line portion of the system. Measurements of VPR penetration efficiency were also made to verify its performance for aviation nvPM number. The research also demonstrated the difficulty of making system loss measurements and substantiates the E-31 decision to predict rather than measure system losses.

Risk of vascular events after nondisabling small and large vessel cerebral ischemia.

BACKGROUND: Small- and large-vessel disease (SVD and LVD, respectively) might have a different pathogenesis and prognosis but the long-term risk of death and recurrent stroke appears to be similar in previous studies. In this study, we investigated the long-term vascular prognosis of patients with LVD and SVD in a large cohort of well-documented patients. METHODS: We included 971 patients with transient ischemic attack (TIA) or nondisabling ischemic stroke of atherosclerotic origin referred to a university hospital in the Netherlands between 1994 and 2005 and followed them for the occurrence of vascular events or death. The primary outcome was a composite of stroke, myocardial infarction and vascular death, whichever happened first. Classification of SVD/LVD was primarily based on brain imaging. We used regression analyses to generate hazard ratios (HRs) with 95% confidence intervals (CIs). Sensitivity analyses were performed in subsets of the population, i.e. patients with subtype classification based on imaging, excluding TIA patients, first-ever stroke patients and LVD patients without a symptomatic carotid stenosis. RESULTS: During a mean follow-up of 6.3 years, new vascular events occurred in 56 of 312 SVD patients (3.3%/year) and in 128 of 659 LVD patients (2.9%/year). These were ischemic strokes in 33 of the 56 events in SVD patients (2.0%/year) and 54 of the 128 events in LVD patients (1.2%/year). The corresponding age- and sex-adjusted HR for all new vascular events for LVD versus SVD was 0.76 (95% CI 0.56-1.05) for the total follow-up period. When this risk was split into early risk (<1 year) and late risk (>1 year), it was not significantly different for the 1-year risk of vascular events (HR 1.04, 95% CI 0.57-1.91); however, after 1 year of follow-up, LVD patients had fewer outcome events compared with SVD patients (HR 0.66, 95% CI 0.46-0.96). For ischemic strokes, the overall HR was 0.60 (95% CI 0.39-0.94). As with the primary outcome, here also the 1-year risk was not significantly different from >1-year risk (HR 1.31, 95% CI 0.62-2.81, and HR 0.36, 95% CI 0.21-0.63, respectively). The sensitivity analyses showed virtually the same results. CONCLUSION: In patients with nondisabling cerebrovascular disease, we found, despite no differences at baseline in terms of vascular risk factors, a better long-term prognosis for patients with LVD for all vascular events, especially for recurrent strokes. Our observations support a different pathogenesis in SVD and LVD patients, and optimal prevention is indicated for patients with what was formerly regarded as 'benign' SVD stroke.

Prognostic value of the Rose questionnaire: a validation with future coronary events in the SMART study.

AIM: The Rose questionnaire was developed in epidemiological studies to obtain a reproducible diagnosis of angina pectoris. We studied the prognostic value of this questionnaire with respect to the occurrence of future coronary events. METHODS AND RESULTS: We studied 7916 consecutive patients (mean age 56 years; 67% men) with clinically manifest vascular disease or cardiovascular risk factors, enrolled in the Second Manifestations of ARTerial disease (SMART) study from 1996 to 2009. At inclusion, all patients completed the Rose questionnaire. We investigated the prognostic value of four definitions of angina pectoris that were based on the following elements of the questionnaire (1) the full questionnaire; (2) three key questions concerning chest pain; (3) one question about discomfort or pain in the chest; (4) two questions about complaints when slowing down or stopping activities (the definition that is used in the SMART study). All patients were followed for new coronary events and interventions for an average of 4.6 years. Analyses were with multivariable Cox regression models. Discriminatory ability of the four definitions as assessed with areas under the receiver-operator characteristics curves was similar (range 0.708-0.726) for coronary events in isolation as well as in combination with coronary interventions. The models were assessed for their ability to improve risk stratification compared with each other; differences between definitions are small. CONCLUSION: Our data implicate that the use of a subset of questions of the Rose questionnaire performs equally well compared with the full Rose questionnaire regarding the prediction of coronary events.

Differential propensity for major hemorrhagic events in patients with different types of arterial disease.

AIMS: Atherosclerosis is the most frequent cause of coronary artery disease (CAD), cerebrovascular disease (CVD), and peripheral arterial obstructive disease (PAD). We previously found that patients with CVD or PAD had a two-fold higher risk of major hemorrhagic complications than patients with CAD. We investigated whether this difference was attributable to baseline risk factors or genetic variants involved in hemostasis. METHODS AND RESULTS: We included 2622 consecutive patients from a single university hospital who presented with non-disabling CAD, CVD, or PAD. All patients were followed for the occurrence of major hemorrhagic complications for a mean of 6.6 years. Major hemorrhagic events included intracranial hemorrhagic events, fatal hemorrhagic events, and any hemorrhagic complications requiring hospitalization, irrespective of interventions. Major hemorrhagic complications occurred in 122 patients (annual event rate of 0.77%). Patients with CVD or PAD had more hemorrhagic complications than patients with CAD (hazard ratio [HR] 2.05, 95% confidence interval [CI] 1.39-3.01). Hypertension, diabetes, renal failure and use of oral anticoagulants or antiplatelet therapy did not explain the difference (HR adjusted for all characteristics 1.74; 95% CI 1.14-2.61). Additional adjustment for genetic variants did not further change the HR. CONCLUSION: Patients with CVD or PAD are at higher risk for major hemorrhagic events than patients with CAD. This difference could not be explained by known risk factors, use of antithrombotic agents, or genetic variants involved in hemostasis. Further research to find the reason for this difference and possible differences in pathogenesis is warranted.

Prognostic modelling in ischaemic stroke study, additional value of genetic characteristics. Rationale and design.

BACKGROUND AND AIM: The prediction of prognosis after cerebral infarction might be improved by genetic information. The aim of the Prognostic Modelling in Ischaemic Stroke study is to develop 2 different prognostic models on the basis of traditional vascular risk factors and genetic information in patients who have suffered from cerebral ischaemia of arterial origin, 1 concerning new ischaemic and the other new haemorrhagic events. METHODS: Polymorphisms and haplotypes describing the haemostatic system and those that influence antithrombotic drug activity will be identified in a cohort of 1,200 patients with cerebral ischaemia of arterial origin who will be followed up for a mean of 6.5 years. In total, 312 ischaemic and 78 haemorrhagic events are anticipated. With a prevalence of a genetic characteristic of 10% a relative risk of 1.4 (95% confidence interval = 1.1-1.8) for ischaemic events and of 1.8 (95% confidence interval = 1.0-3.2) for haemorrhagic events can be estimated with sufficient precision. To determine the additional prognostic value of genetic characteristics the area under the ROC curves of 2 separate models will be compared: 1 based on non-genetic risk factors only, the other also including genetic data.

Characterization of an imatinib-sensitive subset of high-grade human glioma cultures.

High-grade gliomas, including glioblastomas, are malignant brain tumors for which improved treatment is urgently needed. Genetic studies have demonstrated the existence of biologically distinct subsets. Preliminary studies have indicated that platelet-derived growth factor (PDGF) receptor signaling contributes to the growth of some of these tumors. In this study, human high-grade glioma primary cultures were analysed for sensitivity to treatment with the PDGF receptor inhibitor imatinib/Glivec/Gleevec/STI571. Six out of 15 cultures displayed more than 40% growth inhibition after imatinib treatment, whereas seven cultures showed less than 20% growth inhibition. In the sensitive cultures, apoptosis contributed to growth inhibition. Platelet-derived growth factor receptor status correlated with imatinib sensitivity. Supervised analyses of gene expression profiles and real-time PCR analyses identified expression of the chemokine CXCL12/SDF-1 (stromal cell-derived factor 1) as a predictor of imatinib sensitivity. Exogenous addition of CXCL12 to imatinib-insensitive cultures conferred some imatinib sensitivity. Finally, coregulation of CXCL12 and PDGF alpha-receptor was observed in glioblastoma biopsies. We have thus defined the characteristics of a novel imatinib-sensitive subset of glioma cultures, and provided evidence for a functional relationship between imatinib sensitivity and chemokine signaling. These findings will assist in the design and evaluation of clinical trials exploring therapeutic effects of imatinib on malignant brain tumors.