Predictors at Admission of Mechanical Ventilation and Death in an Observational Cohort of Adults Hospitalized With Coronavirus Disease 2019.

BACKGROUND: Coronavirus disease (COVID-19) can cause severe illness and death. Predictors of poor outcome collected on hospital admission may inform clinical and public health decisions. METHODS: We conducted a retrospective observational cohort investigation of 297 adults admitted to 8 academic and community hospitals in Georgia, United States, during March 2020. Using standardized medical record abstraction, we collected data on predictors including admission demographics, underlying medical conditions, outpatient antihypertensive medications, recorded symptoms, vital signs, radiographic findings, and laboratory values. We used random forest models to calculate adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for predictors of invasive mechanical ventilation (IMV) and death. RESULTS: Compared with age <45 years, ages 65-74 years and ≥75 years were predictors of IMV (aORs, 3.12 [95% CI, 1.47-6.60] and 2.79 [95% CI, 1.23-6.33], respectively) and the strongest predictors for death (aORs, 12.92 [95% CI, 3.26-51.25] and 18.06 [95% CI, 4.43-73.63], respectively). Comorbidities associated with death (aORs, 2.4-3.8; P < .05) included end-stage renal disease, coronary artery disease, and neurologic disorders, but not pulmonary disease, immunocompromise, or hypertension. Prehospital use vs nonuse of angiotensin receptor blockers (aOR, 2.02 [95% CI, 1.03-3.96]) and dihydropyridine calcium channel blockers (aOR, 1.91 [95% CI, 1.03-3.55]) were associated with death. CONCLUSIONS: After adjustment for patient and clinical characteristics, older age was the strongest predictor of death, exceeding comorbidities, abnormal vital signs, and laboratory test abnormalities. That coronary artery disease, but not chronic lung disease, was associated with death among hospitalized patients warrants further investigation, as do associations between certain antihypertensive medications and death.

Characteristics and Risk Factors of Hospitalized and Nonhospitalized COVID-19 Patients, Atlanta, Georgia, USA, March-April 2020.

We compared the characteristics of hospitalized and nonhospitalized patients who had coronavirus disease in Atlanta, Georgia, USA. We found that risk for hospitalization increased with a patient's age and number of concurrent conditions. We also found a potential association between hospitalization and high hemoglobin A1c levels in persons with diabetes.

ICU and Ventilator Mortality Among Critically Ill Adults With Coronavirus Disease 2019.

OBJECTIVES: To determine mortality rates among adults with critical illness from coronavirus disease 2019. DESIGN: Observational cohort study of patients admitted from March 6, 2020, to April 17, 2020. SETTING: Six coronavirus disease 2019 designated ICUs at three hospitals within an academic health center network in Atlanta, Georgia, United States. PATIENTS: Adults greater than or equal to 18 years old with confirmed severe acute respiratory syndrome-CoV-2 disease who were admitted to an ICU during the study period. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Among 217 critically ill patients, mortality for those who required mechanical ventilation was 35.7% (59/165), with 4.8% of patients (8/165) still on the ventilator at the time of this report. Overall mortality to date in this critically ill cohort is 30.9% (67/217) and 60.4% (131/217) patients have survived to hospital discharge. Mortality was significantly associated with older age, lower body mass index, chronic renal disease, higher Sequential Organ Failure Assessment score, lower PaO2/FIO2 ratio, higher D-dimer, higher C-reactive protein, and receipt of mechanical ventilation, vasopressors, renal replacement therapy, or vasodilator therapy. CONCLUSIONS: Despite multiple reports of mortality rates exceeding 50% among critically ill adults with coronavirus disease 2019, particularly among those requiring mechanical ventilation, our early experience indicates that many patients survive their critical illness.

Timing of Intubation and Mortality Among Critically Ill Coronavirus Disease 2019 Patients: A Single-Center Cohort Study.

OBJECTIVES: Increasing time to mechanical ventilation and high-flow nasal cannula use may be associated with mortality in coronavirus disease 2019. We examined the impact of time to intubation and use of high-flow nasal cannula on clinical outcomes in patients with coronavirus disease 2019. DESIGN: Retrospective cohort study. SETTING: Six coronavirus disease 2019-specific ICUs across four university-affiliated hospitals in Atlanta, Georgia. PATIENTS: Adults with laboratory-confirmed severe acute respiratory syndrome coronavirus 2 infection who received high-flow nasal cannula or mechanical ventilation. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Among 231 patients admitted to the ICU, 109 (47.2%) were treated with high-flow nasal cannula and 97 (42.0%) were intubated without preceding high-flow nasal cannula use. Of those managed with high-flow nasal cannula, 78 (71.6%) ultimately received mechanical ventilation. In total, 175 patients received mechanical ventilation; 44.6% were female, 66.3% were Black, and the median age was 66 years (interquartile range, 56-75 yr). Seventy-six patients (43.4%) were intubated within 8 hours of ICU admission, 57 (32.6%) between 8 and 24 hours of admission, and 42 (24.0%) greater than or equal to 24 hours after admission. Patients intubated within 8 hours were more likely to have diabetes, chronic comorbidities, and higher admission Sequential Organ Failure Assessment scores. Mortality did not differ by time to intubation (≤ 8 hr: 38.2%; 8-24 hr: 31.6%; ≥ 24 hr: 38.1%; p = 0.7), and there was no association between time to intubation and mortality in adjusted analysis. Similarly, there was no difference in initial static compliance, duration of mechanical ventilation, or ICU length of stay by timing of intubation. High-flow nasal cannula use prior to intubation was not associated with mortality. CONCLUSIONS: In this cohort of critically ill patients with coronavirus disease 2019, neither time from ICU admission to intubation nor high-flow nasal cannula use were associated with increased mortality. This study provides evidence that coronavirus disease 2019 respiratory failure can be managed similarly to hypoxic respiratory failure of other etiologies.

Characteristics Associated with Hospitalization Among Patients with COVID-19 - Metropolitan Atlanta, Georgia, March-April 2020.

The first reported U.S. case of coronavirus disease 2019 (COVID-19) was detected in January 2020 (1). As of June 15, 2020, approximately 2 million cases and 115,000 COVID-19-associated deaths have been reported in the United States.* Reports of U.S. patients hospitalized with SARS-CoV-2 infection (the virus that causes COVID-19) describe high proportions of older, male, and black persons (2-4). Similarly, when comparing hospitalized patients with catchment area populations or nonhospitalized COVID-19 patients, high proportions have underlying conditions, including diabetes mellitus, hypertension, obesity, cardiovascular disease, chronic kidney disease, or chronic respiratory disease (3,4). For this report, data were abstracted from the medical records of 220 hospitalized and 311 nonhospitalized patients aged ≥18 years with laboratory-confirmed COVID-19 from six acute care hospitals and associated outpatient clinics in metropolitan Atlanta, Georgia. Multivariable analyses were performed to identify patient characteristics associated with hospitalization. The following characteristics were independently associated with hospitalization: age ≥65 years (adjusted odds ratio [aOR] = 3.4), black race (aOR = 3.2), having diabetes mellitus (aOR = 3.1), lack of insurance (aOR = 2.8), male sex (aOR = 2.4), smoking (aOR = 2.3), and obesity (aOR = 1.9). Infection with SARS-CoV-2 can lead to severe outcomes, including death, and measures to protect persons from infection, such as staying at home, social distancing (5), and awareness and management of underlying conditions should be emphasized for those at highest risk for hospitalization with COVID-19. Measures that prevent the spread of infection to others, such as wearing cloth face coverings (6), should be used whenever possible to protect groups at high risk. Potential barriers to the ability to adhere to these measures need to be addressed.

Characteristics and Clinical Outcomes of Adult Patients Hospitalized with COVID-19 - Georgia, March 2020.

SARS-CoV-2, the novel coronavirus that causes coronavirus disease 2019 (COVID-19), was first detected in the United States during January 2020 (1). Since then, >980,000 cases have been reported in the United States, including >55,000 associated deaths as of April 28, 2020 (2). Detailed data on demographic characteristics, underlying medical conditions, and clinical outcomes for persons hospitalized with COVID-19 are needed to inform prevention strategies and community-specific intervention messages. For this report, CDC, the Georgia Department of Public Health, and eight Georgia hospitals (seven in metropolitan Atlanta and one in southern Georgia) summarized medical record-abstracted data for hospitalized adult patients with laboratory-confirmed* COVID-19 who were admitted during March 2020. Among 305 hospitalized patients with COVID-19, 61.6% were aged <65 years, 50.5% were female, and 83.2% with known race/ethnicity were non-Hispanic black (black). Over a quarter of patients (26.2%) did not have conditions thought to put them at higher risk for severe disease, including being aged ≥65 years. The proportion of hospitalized patients who were black was higher than expected based on overall hospital admissions. In an adjusted time-to-event analysis, black patients were not more likely than were nonblack patients to receive invasive mechanical ventilation† (IMV) or to die during hospitalization (hazard ratio [HR] = 0.63; 95% confidence interval [CI] = 0.35-1.13). Given the overrepresentation of black patients within this hospitalized cohort, it is important for public health officials to ensure that prevention activities prioritize communities and racial/ethnic groups most affected by COVID-19. Clinicians and public officials should be aware that all adults, regardless of underlying conditions or age, are at risk for serious illness from COVID-19.

Utilized or Underutilized: A Qualitative Analysis of Building Coherence During Early Implementation of a Tele-Intensive Care Unit.

Background: Generating, reading, or interpreting data is a component of Telemedicine-Intensive Care Unit (Tele-ICU) utilization that has not been explored in the literature. Introduction: Using the idea of "coherence," a construct of Normalization Process Theory, we describe how intensive care unit (ICU) and Tele-ICU staff made sense of their shared work and how they made use of Tele-ICU together. Materials and Methods: We interviewed ICU and Tele-ICU staff involved in the implementation of Tele-ICU during site visits to a Tele-ICU hub and 3 ICUs, at preimplementation (43 interviews with 65 participants) and 6 months postimplementation (44 interviews with 67 participants). Data were analyzed using deductive coding techniques and lexical searches. Results: In the early implementation of Tele-ICU, ICU and Tele-ICU staff lacked consensus about how to share information and consequently how to make use of innovations in data tracking and interpretation offered by the Tele-ICU (e.g., acuity systems). Attempts to collaborate and create opportunities for utilization were supported by quality improvement (QI) initiatives. Discussion: Characterizing Tele-ICU utilization as an element of a QI process limited how ICU staff understood Tele-ICU as an innovation. It also did not promote an understanding of how the Tele-ICU used data and may therefore attenuate the larger promise of Tele-ICU as a potential tool for leveraging big data in critical care. Conclusions: Shared data practices lay the foundation for Tele-ICU program utilization but raise new questions about how the promise of big data can be operationalized for bedside ICU staff.

Global trends in the awareness of sepsis: insights from search engine data between 2012 and 2017.

BACKGROUND: Sepsis is an established global health priority with high mortality that can be curtailed through early recognition and intervention; as such, efforts to raise awareness are potentially impactful and increasingly common. We sought to characterize trends in the awareness of sepsis by examining temporal, geographic, and other changes in search engine utilization for sepsis information-seeking online. METHODS: Using time series analyses and mixed descriptive methods, we retrospectively analyzed publicly available global usage data reported by Google Trends (Google, Palo Alto, CA, USA) concerning web searches for the topic of sepsis between 24 June 2012 and 24 June 2017. Google Trends reports aggregated and de-identified usage data for its search products, including interest over time, interest by region, and details concerning the popularity of related queries where applicable. Outlying epochs of search activity were identified using autoregressive integrated moving average modeling with transfer functions. We then identified awareness campaigns and news media coverage that correlated with epochs of significantly heightened search activity. RESULTS: A second-order autoregressive model with transfer functions was specified following preliminary outlier analysis. Nineteen significant outlying epochs above the modeled baseline were identified in the final analysis that correlated with 14 awareness and news media events. Our model demonstrated that the baseline level of search activity increased in a nonlinear fashion. A recurrent cyclic increase in search volume beginning in 2012 was observed that correlates with World Sepsis Day. Numerous other awareness and media events were correlated with outlying epochs. The average worldwide search volume for sepsis was less than that of influenza, myocardial infarction, and stroke. CONCLUSIONS: Analyzing aggregate search engine utilization data has promise as a mechanism to measure the impact of awareness efforts. Heightened information-seeking about sepsis occurs in close proximity to awareness events and relevant news media coverage. Future work should focus on validating this approach in other contexts and comparing its results to traditional methods of awareness campaign evaluation.

Influence of non-invasive blood pressure measurement intervals on the occurrence of intra-operative hypotension.

The American Society of Anesthesiologists Standards for Basic Monitoring recommends blood pressure (BP) measurement every 5 min. Research has shown distractions or technical factors can cause prolonged measurement intervals exceeding 5 min. We investigated the relationship between prolonged non-invasive BP (NIBP) measurement interval and the incidence of hypotension, detected post-interval. Our secondary outcome was to determine independent predictors of these prolonged NIBP measurement intervals. Retrospective data were analyzed from 139,509 general anesthesia cases from our institution's Anesthesia Information Management System (AIMS). Absolute hypotension (AH) was defined a priori as a systolic BP < 80 mmHg and relative hypotension (RH) was defined as a 40% decrease in systolic BP from the preoperative baseline. Odds ratios (OR) with 95% confidence intervals and Pearson's Chi square Test reported the association of prolonged NIBP measurement intervals on hypotension detected post-NIBP measurement interval. Logistic regression models were developed to determine independent predictors of NIBP measurement intervals. The analysis revealed that NIBP measurement intervals greater than 6 and 10 min are associated with an approximately four times higher incidence of a patient transitioning into hypotension (AH/RH > 6 min OR 4.0 / 3.6; AH/RH > 10 min OR 4.3 / 3.9; p < 0.001). A key finding was that the "> 10-minute AH model" indicated that age 41-80, increased co-morbidity profile, obesity and turning (repositioning) of the operative room table were significant predictors of prolonged NIBP measurement intervals (p < 0.001). While we do not suggest NIBP measurement intervals cause hypotension, intervals greater than 6 and 10 min are associated with a fourfold increase in the propensity of an undetected transition into both RH or AH. These data support current monitoring guidelines.

Temporal trends and current practice patterns for intraoperative ventilation at U.S. academic medical centers: a retrospective study.

BACKGROUND: Lung protective ventilation strategies utilizing lower tidal volumes per predicted body weight (PBW) and positive end-expiratory pressure (PEEP) have been suggested to be beneficial in a variety of surgical populations. Recent clinical studies have used control groups ventilated with high tidal volumes without PEEP based on the assumption that this reflects current clinical practice. We hypothesized that ventilation strategies have changed over time, that most anesthetics in U.S. academic medical centers are currently performed with lower tidal volumes, and that most receive PEEP. METHODS: Intraoperative data were pooled for adults undergoing general anesthesia with tracheal intubation. Median tidal volumes per kilogram of PBW were categorized as > 10, 8-10 and < 8 mL per kg of PBW. The percentages of cases in 2013 that were performed with median tidal volumes < 8 mL per kg of PBW and PEEP were determined. As a secondary analysis, a proportional odds model using institution, year, height, weight and gender determined the relative associations of these factors using categorical and interquartile odds ratios. RESULTS: 295,540 cases were analyzed from 5 institutions over a period of 10 years. In 2013, 59.3% of cases used median tidal volumes < 8 mL per kg of PBW, 83.3% used PEEP, and 51.0% used both. Of those cases with PEEP, 60.9% used a median pressure of ≥ 5 cmH2O. Predictors of lower categories of tidal volumes included height (odds ratio (OR) 10.83, 95% confidence interval [10.50, 11.16]), institution (lowest OR 0.98 [0.96, 1.00], highest OR 9.63 [9.41, 9.86]), year (lowest OR 1.32 [1.21, 1.44], highest OR 6.31 [5.84, 6.82]), male gender (OR 1.10 [1.07, 1.12]), and weight (OR 0.30 [0.29, 0.31]). CONCLUSION: Most general anesthetics with tracheal intubation at the institutions surveyed are currently performed with a median tidal volume < 8 mL per kg of PBW, most are managed with PEEP of ≥ 5 cmH2O and approximately half utilize both. Given the diversity of the institutions included, this is likely reflective of practice in U.S. academic medical centers. The utilization of higher tidal volumes without PEEP in control groups for clinical research studies should be reconsidered.

Design and implementation of a hospital wide waveform capture system.

The use of telemetry and invasive monitoring is exceptionally common in modern healthcare. To date the vast majority of this information is not stored for more than a brief duration on the local monitor. This prohibits extensive investigation into waveform data. We describe a system to collect such data in a quaternary care facility. Using standardized "packet sniffing" technology along with routine manual documentation, we reverse engineered the Unity network protocol used to transmit waveform data across the University of Michigan mission critical monitor network. Data was subsequently captured using a proprietary piece of software writing waveform data to local disks. Nightly, this data is post-processed using data from the admit-discharge-transfer system into individual patient waveforms for the day regardless of location. Over a 10 month period, over 2,785 individual patients had a total of 65,112 waveforms captured 15,978 from the operating rooms and 49,134 from the ICUs. The average OR case collected over 11 MB of data. The average single day data collection consisted of 8.6 GB of data. Entire hospital waveform data collection is possible using internally developed software enabling research on waveform data with minimal technical burden. Further research is required to determine the long-term storage and processing of such data.

Predicting risk of postoperative lung injury in high-risk surgical patients: a multicenter cohort study.

BACKGROUND: Acute respiratory distress syndrome (ARDS) remains a serious postoperative complication. Although ARDS prevention is a priority, the inability to identify patients at risk for ARDS remains a barrier to progress. The authors tested and refined the previously reported surgical lung injury prediction (SLIP) model in a multicenter cohort of at-risk surgical patients. METHODS: This is a secondary analysis of a multicenter, prospective cohort investigation evaluating high-risk patients undergoing surgery. Preoperative ARDS risk factors and risk modifiers were evaluated for inclusion in a parsimonious risk-prediction model. Multiple imputation and domain analysis were used to facilitate development of a refined model, designated SLIP-2. Area under the receiver operating characteristic curve and the Hosmer-Lemeshow goodness-of-fit test were used to assess model performance. RESULTS: Among 1,562 at-risk patients, ARDS developed in 117 (7.5%). Nine independent predictors of ARDS were identified: sepsis, high-risk aortic vascular surgery, high-risk cardiac surgery, emergency surgery, cirrhosis, admission location other than home, increased respiratory rate (20 to 29 and ≥30 breaths/min), FIO2 greater than 35%, and SpO2 less than 95%. The original SLIP score performed poorly in this heterogeneous cohort with baseline risk factors for ARDS (area under the receiver operating characteristic curve [95% CI], 0.56 [0.50 to 0.62]). In contrast, SLIP-2 score performed well (area under the receiver operating characteristic curve [95% CI], 0.84 [0.81 to 0.88]). Internal validation indicated similar discrimination, with an area under the receiver operating characteristic curve of 0.84. CONCLUSIONS: In this multicenter cohort of patients at risk for ARDS, the SLIP-2 score outperformed the original SLIP score. If validated in an independent sample, this tool may help identify surgical patients at high risk for ARDS.

Replacement of anesthesia machines improves intraoperative ventilation parameters associated with the development of acute respiratory distress syndrome.

BACKGROUND: The impact of anesthetic equipment on clinical practice parameters associated with development of acute respiratory distress syndrome (ARDS) has not been extensively studied. We hypothesized a change in anesthesia machines would be associated with parameters associated with lower rates of ARDS. METHODS: We performed a retrospective cohort study on a subset of data used to evaluate intraoperative ventilation. Patients included adults receiving a non-cardiac, non-thoracic, non-transplant, non-trauma, general anesthetic between 2/1/05, and 3/31/09 at the University of Michigan. Existing anesthesia machines (Narkomed IIb, Drager) were exchanged for new equipment (Aisys, General Electric). The initial subset compared the characteristics of patients anesthetized between 12/1/06 and 1/31/07 (pre) with those between 4/1/07 and 5/30/07 (post). An extended subset examined cases two years pre and post exchange. Using the standard predicted body weight (PBW), we calculated and compared the tidal volume (total Vt and mL/kg PBW) as well as positive end-expiratory pressure (PEEP), peak inspiratory pressure (PIP), Delta P (PIP-PEEP), and FiO2. RESULTS: A total of 1,414 patients were included in the 2-month pre group and 1,635 patients included in the post group. Comparison of ventilation characteristics found statistically significant differences in median (pre v post): PIP (26 ± 6 v 21 ± 6 cmH2O, p < .001), Delta P (24 ± 6 v 19 ± 6 cmH2O, p < .001), Vt (588 ± 139 v 562 ± 121 ml, p < 0.001; 9.3 ± 2.2 v 9.0 ± 1.9 ml/kg predicted body weight, p < .001), FiO2 (0.57 ± 0.17 v 0.52 ± 0.18, p < .001). Groups did not differ in age, ASA category, PBW, or BMI. The two year subgroup had similar parameters. Risk adjustment resulted in minimal differences in the analysis. New anesthesia machines were associated with a non-statistically significant reduction in postoperative ARDS. CONCLUSIONS: In this study, a change in ventilator management was associated with an anesthesia machine exchange. The smaller Vt and lower PIP noted in the post group may imply a lower risk of volutrauma and barotrauma, which may be significant in at-risk populations. However, there was not a statistically significant reduction in the incidence of post-operative ARDS.

Preoperative and intraoperative predictors of postoperative acute respiratory distress syndrome in a general surgical population.

BACKGROUND: Acute respiratory distress syndrome (ARDS) is a devastating condition with an estimated mortality exceeding 30%. There are data suggesting risk factors for ARDS development in high-risk populations, but few data are available in lower incidence populations. Using risk-matched analysis and a combination of clinical and research data sets, we determined the incidence and risk factors for the development of ARDS in this general surgical population. METHODS: We conducted a review of common adult surgical procedures completed between June 1, 2004 and May 31, 2009 using an anesthesia information system. This data set was merged with an ARDS registry and an institutional death registry. Preoperative variables were subjected to multivariate analysis. Matching and multivariate regression was used to determine intraoperative factors associated with ARDS development. RESULTS: In total, 50,367 separate patient admissions were identified, and 93 (0.2%) of these patients developed ARDS. Preoperative risk factors for ARDS development included American Society of Anesthesiologist status 3-5 (odds ratio [OR] 18.96), emergent surgery (OR 9.34), renal failure (OR 2.19), chronic obstructive pulmonary disease (OR 2.16), number of anesthetics during the admission (OR 1.37), and male sex (OR 1.65). After matching, intraoperative risk factors included drive pressure (OR 1.17), fraction inspired oxygen (OR 1.02), crystalloid administration in liters (1.43), and erythrocyte transfusion (OR 5.36). CONCLUSIONS: ARDS is a rare condition postoperatively in the general surgical population and is exceptionally uncommon in low American Society of Anesthesiologists status patients undergoing scheduled surgery. Analysis after matching suggests that ARDS development is associated with median drive pressure, fraction inspired oxygen, crystalloid volume, and transfusion.

Automated alerting and recommendations for the management of patients with preexisting hypoxia and potential acute lung injury: a pilot study.

BACKGROUND: Acute lung injury (ALI) is associated with high mortality. Low tidal volume (Vt) ventilation has been shown to reduce mortality in ALI patients in the intensive care unit. Anesthesiologists do not routinely provide lung-protective ventilation strategies to patients with ALI in the operating room. The authors hypothesized that an alert, recommending lung-protective ventilation regarding patients with potential ALI, would result in lower Vt administration. METHODS: The authors conducted a randomized controlled trial on anesthesia providers caring for patients with potential ALI. Patients with an average or last collected ratio of partial pressure of arterial oxygen to inspired fraction of oxygen less than 300 were randomized to providers being sent an alert with a recommended Vt of 6 cc/kg predicted body weight or conventional care. Primary outcomes were Vt/kg predicted body weight administered to patients. Secondary outcomes included ventilator parameters, length of postoperative ventilation, and death. RESULTS: The primary outcome was a clinically significant reduction in mean Vt from 508-458 cc (P = 0.033), with a reduction in Vt when measured in cc/kg predicted body weight from 8 to 7.2 cc/kg predicted body weight (P = 0.040). There were no statistically significant changes in other outcomes or adverse events associated with either arm. CONCLUSIONS: Automated alerts generated for patients at risk of having ALI resulted in a statistically significant reduction in Vt administered when compared with a control group. Further research is required to determine whether a reduction in Vt results in decreased mortality and/or postoperative duration of mechanical ventilation.

Application of Machine Learning Models to Biomedical and Information System Signals From Critically Ill Adults.

BACKGROUND: Machine learning (ML)-derived notifications for impending episodes of hemodynamic instability and respiratory failure events are interesting because they can alert physicians in time to intervene before these complications occur. RESEARCH QUESTION: Do ML alerts, telemedicine system (TS)-generated alerts, or biomedical monitors (BMs) have superior performance for predicting episodes of intubation or administration of vasopressors? STUDY DESIGN AND METHODS: An ML algorithm was trained to predict intubation and vasopressor initiation events among critically ill adults. Its performance was compared with BM alarms and TS alerts. RESULTS: ML notifications were substantially more accurate and precise, with 50-fold lower alarm burden than TS alerts for predicting vasopressor initiation and intubation events. ML notifications of internal validation cohorts demonstrated similar performance for independent academic medical center external validation and COVID-19 cohorts. Characteristics were also measured for a control group of recent patients that validated event detection methods and compared TS alert and BM alarm performance. The TS test characteristics were substantially better, with 10-fold less alarm burden than BM alarms. The accuracy of ML alerts (0.87-0.94) was in the range of other clinically actionable tests; the accuracy of TS (0.28-0.53) and BM (0.019-0.028) alerts were not. Overall test performance (F scores) for ML notifications were more than fivefold higher than for TS alerts, which were higher than those of BM alarms. INTERPRETATION: ML-derived notifications for clinically actioned hemodynamic instability and respiratory failure events represent an advance because the magnitude of the differences of accuracy, precision, misclassification rate, and pre-event lead time is large enough to allow more proactive care and has markedly lower frequency and interruption of bedside physician work flows.