Systematic Review and Comparison of Publicly Available ICU Data Sets-A Decision Guide for Clinicians and Data Scientists.

OBJECTIVE: As data science and artificial intelligence continue to rapidly gain traction, the publication of freely available ICU datasets has become invaluable to propel data-driven clinical research. In this guide for clinicians and researchers, we aim to: 1) systematically search and identify all publicly available adult clinical ICU datasets, 2) compare their characteristics, data quality, and richness and critically appraise their strengths and weaknesses, and 3) provide researchers with suggestions, which datasets are appropriate for answering their clinical question. DATA SOURCES: A systematic search was performed in Pubmed, ArXiv, MedRxiv, and BioRxiv. STUDY SELECTION: We selected all studies that reported on publicly available adult patient-level intensive care datasets. DATA EXTRACTION: A total of four publicly available, adult, critical care, patient-level databases were included (Amsterdam University Medical Center data base [AmsterdamUMCdb], eICU Collaborative Research Database eICU CRD], High time-resolution intensive care unit dataset [HiRID], and Medical Information Mart for Intensive Care-IV). Databases were compared using a priori defined categories, including demographics, patient characteristics, and data richness. The study protocol and search strategy were prospectively registered. DATA SYNTHESIS: Four ICU databases fulfilled all criteria for inclusion and were queried using SQL (PostgreSQL version 12; PostgreSQL Global Development Group) and analyzed using R (R Foundation for Statistical Computing, Vienna, Austria). The number of unique patient admissions varied between 23,106 (AmsterdamUMCdb) and 200,859 (eICU-CRD). Frequency of laboratory values and vital signs was highest in HiRID, for example, 5.2 (±3.4) lactate values per day and 29.7 (±10.2) systolic blood pressure values per hour. Treatment intensity varied with vasopressor and ventilatory support in 69.0% and 83.0% of patients in AmsterdamUMCdb versus 12.0% and 21.0% in eICU-CRD, respectively. ICU mortality ranged from 5.5% in eICU-CRD to 9.9% in AmsterdamUMCdb. CONCLUSIONS: We identified four publicly available adult clinical ICU datasets. Sample size, severity of illness, treatment intensity, and frequency of reported parameters differ markedly between the databases. This should guide clinicians and researchers which databases to best answer their clinical questions.

Intraovarian regulation of gonadotropin-dependent folliculogenesis depends on notch receptor signaling pathways not involving Delta-like ligand 4 (Dll4).

BACKGROUND: In-situ hybridisation studies demonstrate that Notch receptors and ligands are expressed in granulosa cells (GCs) and in the theca layer vasculature of growing follicles. Notch signaling involves cell-to-cell interaction mediated by transmembrane receptors and ligands. This signaling pathway may represent a novel intraovarian regulator of gonadotropin-dependent follicular development to the preovulatory stage. We hypothesized that blocking Notch pathways would disrupt follicular maturation in the mouse ovary. METHODS: Hypophysectomized CD21 female mice were administered pregnant mare serum gonadotropin (PMSG) for 3 days to stimulate follicular development. In one experiment, a pan-notch inhibitor, compound E, was initiated 2 days prior to and throughout stimulation (n = 10), while in a second experiment, a humanized phage Dll4 blocking antibody, YW152F, was used (n = 5). After sacrifice, ovarian histology, serum estradiol levels and uterine weights were compared to controls. The ovarian morphology was evaluated with hematoxylin/eosin staining and immunohistochemistry was performed for Notch1, Notch2, Notch3, Notch4, Jagged1, Dll4, platelet endothelial cell adhesion molecule (PECAM) and alpha-smooth muscle actin (α-SMA) detection. RESULTS: We localized specific Notch ligands and receptors in the following structures: Dll4 is specific to theca layer endothelial cells (ECs); Notch1/Notch4 and Jagged1 are expressed in theca layer ECs and vascular smooth muscle cells (VSMCs), whereas Notch3 is restricted to VSMCs; Notch2 is expressed mostly on GCs of small follicles. Administration of a pan-Notch inhibitor, compound E, inhibits follicular development to the preovulatory stage (8.5 preovulatory follicles in treatment vs. 3.4 preovulatory follicles in control, p < 0.01; average number per ovary) with significant secondary effects on ovarian and uterine weight and estradiol secretion in a setting of uninhibited vascular proliferation, but disorganized appearance of ECs and VSMCs. Inhibition of endothelial Notch1 function through the inactivation of its ligand Dll4 with the blocking antibody YW152F induces mild disorganisation of follicular vasculature, but has no significant effect on gonadotropin-dependent folliculogenesis. CONCLUSIONS: Our experiments suggest that the complete blockage of the Notch signaling pathway with compound E impairs folliculogenesis and induces disruption of gonadotropin stimulated angiogenesis. It seems the mechanism involves Notch1 and Notch3, specifically, causing the improper assembly of ECs and VSMCs in the theca layer, although the potential role of non-angiogenic Notch signaling, such as Jagged2 to Notch2 in GCs, remains to be elucidated.

Generalizable calibrated machine learning models for real-time atrial fibrillation risk prediction in ICU patients.

BACKGROUND: Atrial Fibrillation (AF) is the most common arrhythmia in the intensive care unit (ICU) and is associated with increased morbidity and mortality. Identification of patients at risk for AF is not routinely performed as AF prediction models are almost solely developed for the general population or for particular ICU populations. However, early AF risk identification could help to take targeted preemptive actions and possibly reduce morbidity and mortality. Predictive models need to be validated across hospitals with different standards of care and convey their predictions in a clinically useful manner. Therefore, we designed AF risk models for ICU patients using uncertainty quantification to provide a risk score and evaluated them on multiple ICU datasets. METHODS: Three CatBoost models, utilizing feature windows comprising data 1.5-13.5, 6-18, or 12-24 hours before AF occurrence, were built using 2-repeat-10-fold cross-validation on AmsterdamUMCdb, the first freely available European ICU database. Furthermore, AF Patients were matched with no-AF patients for training. Transferability was validated using a direct and a recalibration evaluation on two independent external datasets, MIMIC-IV and GUH. The calibration of the predicted probability, used as an AF risk score, was measured using the Expected Calibration Error (ECE) and the presented Expected Signed Calibration Error (ESCE). Additionally, all models were evaluated across time during the ICU stay. RESULTS: The model performance reached Areas Under the Curve (AUCs) of 0.81 at internal validation. Direct external validation showed partial generalizability with AUCs reaching 0.77. However, recalibration resulted in performances matching or exceeding that of the internal validation. All models furthermore showed calibration capabilities demonstrating adequate risk prediction competence. CONCLUSION: Ultimately, recalibrating models reduces the challenge of generalization to unseen datasets. Moreover, utilizing the patient-matching methodology together with the assessment of uncertainty calibration can serve as a step toward the development of clinical AF prediction models.

Evaluating equitable care in the ICU: Creating a causal inference framework to assess the impact of life-sustaining interventions across racial and ethnic groups.

Background: Variability in the provision of intensive care unit (ICU)-interventions may lead to disparities between socially defined racial-ethnic groups. Research Question: We used causal inference to examine the use of invasive mechanical ventilation (IMV), renal replacement therapy (RRT), and vasopressor agents (VP) to identify disparities in outcomes across race-ethnicity in patients with sepsis. Study Design and Methods: Single-center, academic referral hospital in Boston, Massachusetts, USA. Retrospective analysis of treatment effect with a targeted trial design categorized by treatment assignment within the first 24 hours in the MIMIC-IV dataset (2008- 2019) using targeted maximum likelihood estimation. Of 76,943 ICU stays in MIMIC-IV, 32,971 adult stays fulfilling sepsis-3 criteria were included. The primary outcome was in-hospital mortality. Secondary outcomes were hospital-free days, and occurrence of nosocomial infection stratified by predicted mortality probability ranges and self-reported race-ethnicity. Average treatment effects by treatment type and race-ethnicity, Racial-ethnic group (REG) or White group (WG), were estimated. Results: Of 19,419 admissions that met inclusion criteria, median age was 68 years, 57.4% were women, 82% were White, and mortality was 18.2%. There was no difference in mortality benefit associated with the administration of IMV, RRT, or VP between the REG and the WG. There was also no difference in hospital-free days or nosocomial infections. These findings are unchanged with different eligibility periods. Interpretation: There were no differences in the treatment outcomes from three life-sustaining interventions in the ICU according to race-ethnicity. While there was no discernable harm from the treatments across mortality risk, there was also no measurable benefit. These findings highlight the need for research to understand better the risk-benefit of life-sustaining interventions in the ICU.

Leveraging electronic health records for data science: common pitfalls and how to avoid them.

Analysis of electronic health records (EHRs) is an increasingly common approach for studying real-world patient data. Use of routinely collected data offers several advantages compared with other study designs, including reduced administrative costs, the ability to update analysis as practice patterns evolve, and larger sample sizes. Methodologically, EHR analysis is subject to distinct challenges because data are not collected for research purposes. In this Viewpoint, we elaborate on the importance of in-depth knowledge of clinical workflows and describe six potential pitfalls to be avoided when working with EHR data, drawing on examples from the literature and our experience. We propose solutions for prevention or mitigation of factors associated with each of these six pitfalls-sample selection bias, imprecise variable definitions, limitations to deployment, variable measurement frequency, subjective treatment allocation, and model overfitting. Ultimately, we hope that this Viewpoint will guide researchers to further improve the methodological robustness of EHR analysis.

Understanding critically ill sepsis patients with normal serum lactate levels: results from U.S. and European ICU cohorts.

While serum lactate level is a predictor of poor clinical outcomes among critically ill patients with sepsis, many have normal serum lactate. A better understanding of this discordance may help differentiate sepsis phenotypes and offer clues to sepsis pathophysiology. Three intensive care unit datasets were utilized. Adult sepsis patients in the highest quartile of illness severity scores were identified. Logistic regression, random forests, and partial least square models were built for each data set. Features differentiating patients with normal/high serum lactate on day 1 were reported. To exclude that differences between the groups were due to potential confounding by pre-resuscitation hyperlactatemia, the analyses were repeated for day 2. Of 4861 patients included, 47% had normal lactate levels. Patients with normal serum lactate levels had lower 28-day mortality rates than those with high lactate levels (17% versus 40%) despite comparable physiologic phenotypes. While performance varied between datasets, logistic regression consistently performed best (area under the receiver operator curve 87-99%). The variables most strongly associated with normal serum lactate were serum bicarbonate, chloride, and pulmonary disease, while serum sodium, AST and liver disease were associated with high serum lactate. Future studies should confirm these findings and establish the underlying pathophysiological mechanisms, thus disentangling association and causation.

Extracorporeal membrane oxygenation use has increased by 433% in adults in the United States from 2006 to 2011.

Recent studies have shown the benefits of extracorporeal membrane oxygenation (ECMO) in supporting adults with severe respiratory or cardiac failure refractory to conventional treatments. The purpose of this investigation was to analyze the usage of ECMO in adults to identify recent trends within the United States. The usage of ECMO, the survival rates, and the hospitalization costs from 2006 to 2011 were analyzed using the Nationwide Inpatient Sample from the Healthcare Cost and Utilization Project. The rate of ECMO cases per million adult discharges increased 433% from 11.4 (95% confidence interval, 6.1-16.8) in 2006 to 60.9 (95% confidence interval, 28.1-93.7) in 2011 (p for trend = 0.001). There was a trend toward improved survival rates, but this was not statistically significant (p for trend = 0.14). The costs per day have not changed significantly (p for trend = 0.07) nor have the total costs per patient (p for trend = 0.87). In conclusion, there was a huge increase in the usage of ECMO in adults from 2006 to 2011 with a trend toward improved survival rates and no increase in hospitalization costs.

National trends in the utilization of short-term mechanical circulatory support: incidence, outcomes, and cost analysis.

BACKGROUND: The number of alternatives to intra-aortic balloon counterpulsation in the treatment of anticipated and established acute circulatory failure is growing. Despite the clinical importance and significant cost of short-term mechanical circulatory support (MCS) devices, the state of their present use has not been analyzed on a national scale. OBJECTIVES: The purpose of this study was to characterize the demographics, treatment practices, survival rates, and cost of short-term MCS. METHODS: In this serial cross-sectional study, we analyzed all adult patients receiving short-term MCS in the United States from 2004 to 2011 by using the Nationwide Inpatient Sample from the Healthcare Cost and Utilization Project. RESULTS: From 2007 to 2011, use of percutaneous devices for short-term MCS increased by 1,511% compared with a 101% increase in nonpercutaneous devices. Mortality rates declined over this period (p for trend = 0.027) from 41.1% in 2004 to 2007 to 33.4% in 2008 to 2011. A similar trend was observed for the subset of patients with cardiogenic shock, decreasing from 51.6% to 43.1% (p for trend = 0.012). Hospital costs also declined over this period (p for trend = 0.011). Multivariable analysis revealed balloon pumps (odds ratio [OR]: 2.00; 95% confidence interval [CI]: 1.58 to 2.52), coagulopathy (OR: 2.35; 95% CI: 1.88 to 2.94), and cardiopulmonary resuscitation (OR: 3.50; 95% CI: 2.20 to 5.57) before short-term MCS were among the most significant predictors of mortality. CONCLUSIONS: Use of short-term MCS in the United States has increased rapidly, whereas rates of in-hospital mortality have decreased. These changes have taken place in the context of declining hospital costs associated with short-term MCS.

Endometrial adenocarcinoma presenting as a hematotrachelos.

BACKGROUND: Hematotrachelos, distension of the uterine cervix with accumulated blood, is an extremely rare condition resulting from a congenital anomaly or an acquired condition. We present a case in which an acquired hematotrachelos was the presenting sign of endometrial cancer. CASE: An asymptomatic 66-year-old woman was found to have a bulging cervix during a well-woman visit. Further workup revealed a hematotrachelos and an underlying endometrial adenocarcinoma. She was treated with surgery and adjuvant radiotherapy. CONCLUSION: A hematotrachelos, although rare, can prevent vaginal bleeding, which is often the earliest symptom of a uterine malignancy. This case report illustrates the potential importance of the pelvic examination as part of the well-woman physical examination, because it led to the discovery of early-stage endometrial cancer.

Vascular endothelial growth factor receptor 2 (VEGFR-2) functions to promote uterine decidual angiogenesis during early pregnancy in the mouse.

Implantation of an embryo induces rapid proliferation and differentiation of uterine stromal cells, forming a new structure, the decidua. One salient feature of decidua formation is a marked increase in maternal angiogenesis. Vascular endothelial growth factor (VEGF)-dependent pathways are active in the ovary, uterus, and embryo, and inactivation of VEGF function in any of these structures might prevent normal pregnancy development. We hypothesized that decidual angiogenesis is regulated by VEGF acting through specific VEGF receptors (VEGFRs). To test this hypothesis, we developed a murine pregnancy model in which systemic administration of a receptor-blocking antibody would act specifically on uterine angiogenesis and not on ovarian or embryonic angiogenesis. In our model, ovarian function was replaced with exogenous progesterone, and blocking antibodies were administered prior to embryonic expression of VEGFRs. After administration of a single dose of the anti-VEGFR-2 antibody during the peri-implantation period, no embryos were detected on embryonic d 10.5. The pregnancy was disrupted because of a significant reduction in decidual angiogenesis, which under physiological conditions peaks on embryonic d 5.5 and 6.5. Inactivation of VEGFR-3 reduced angiogenesis in the primary decidual zone, whereas administration of VEGFR-1 blocking antibodies had no effect. Pregnancy was not disrupted after administration of anti-VEGFR-3 or anti-VEGFR-1 antibodies. Thus, the VEGF/VEGFR-2 pathway plays a key role in the maintenance of early pregnancy through its regulation of peri-implantation angiogenesis in the uterine decidua. This newly formed decidual vasculature serves as the first exchange apparatus for the developing embryo until the placenta becomes functionally active.