A Mobile Post Anesthesia Care Unit Order Reminder System Improves Timely Order Entry.

Transition to the postanesthesia care unit (PACU) requires timely order placement by anesthesia providers. Computerized ordering enables automated order reminder systems, but their value is not fully understood. We performed a single-center, retrospective cohort study to estimate the association between automated PACU order reminders and primary outcomes (1) on-time order placement and (2) the degree of delay in placement. As a secondary post-hoc analysis, we studied the association between late order placement and PACU outcomes. We included patients with a qualifying postprocedure order from January 1, 2019, to May 31, 2023. We excluded cases transferred directly to the ICU, whose anesthesia provider was involved in the pilot testing of the reminder system, or those with missing covariate data. Order reminder system usage was defined by the primary attending anesthesiologist's receipt of a push notification reminder on the day of surgery. We estimated the association between reminder system usage and timely order placement using a logistic regression. For patients with late orders, we performed a survival analysis of order placement. The significance level was 0.05. Patient (e.g., age, race), procedural (e.g., anesthesia duration), and provider-based (e.g., ordering privileges) variables were used as covariates within the analyses. Reminders were associated with 51% increased odds of order placement prior to PACU admission (Odds Ratio: 1.51; 95% Confidence Interval: 1.43, 1.58; p ≤ 0.001), reducing the incidence of late PACU orders from 17.5% to 12.6% (p ≤ 0.001). In patients with late orders, the reminders were associated with 10% quicker placement (Hazard Ratio: 1.10; 95% CI 1.05, 1.15; p < 0.001). On-time order placement was associated with decreased PACU duration (p < 0.001), decreased odds of peak PACU pain score (p < 0.001), and decreased odds of multiple administration of antiemetics (p = 0.02). An order reminder system was associated with an increase in order placement prior to PACU arrival and a reduction in delay in order placement after arrival.

Machine learning predicts cerebral vasospasm in patients with subarachnoid haemorrhage.

BACKGROUND: Cerebral vasospasm (CV) is a feared complication which occurs after 20-40% of subarachnoid haemorrhage (SAH). It is standard practice to admit patients with SAH to intensive care for an extended period of resource-intensive monitoring. We used machine learning to predict CV requiring verapamil (CVRV) in the largest and only multi-center study to date. METHODS: Patients with SAH admitted to UCLA from 2013 to 2022 and a validation cohort from VUMC from 2018 to 2023 were included. For each patient, 172 unique intensive care unit (ICU) variables were extracted through the primary endpoint, namely first verapamil administration or no verapamil. At each institution, a light gradient boosting machine (LightGBM) was trained using five-fold cross validation to predict the primary endpoint at various hospitalization timepoints. FINDINGS: A total of 1750 patients were included from UCLA, 125 receiving verapamil. LightGBM achieved an area under the ROC (AUC) of 0.88 > 1 week in advance and ruled out 8% of non-verapamil patients with zero false negatives. Our models predicted "no CVRV" vs "CVRV within three days" vs "CVRV after three days" with AUCs = 0.88, 0.83, and 0.88, respectively. From VUMC, 1654 patients were included, 75 receiving verapamil. VUMC predictions averaged within 0.01 AUC points of UCLA predictions. INTERPRETATION: We present an accurate and early predictor of CVRV using machine learning with multi-center validation. This represents a significant step towards optimized clinical management and resource allocation in patients with SAH. FUNDING: Robert E. Freundlich is supported by National Center for Advancing Translational Sciences federal grant UL1TR002243 and National Heart, Lung, and Blood Institute federal grant K23HL148640; these funders did not play any role in this study. The National Institutes of Health supports Vanderbilt University Medical Center which indirectly supported these research efforts. Neither this study nor any other authors personally received financial support for the research presented in this manuscript. No support from pharmaceutical companies was received.

Prognostic clinical decision support for pneumonia in the emergency department: A randomized trial.

BACKGROUND: Hospitalization rates for childhood pneumonia vary widely. Risk-based clinical decision support (CDS) interventions may reduce unwarranted variation. METHODS: We conducted a pragmatic randomized trial in two US pediatric emergency departments (EDs) comparing electronic health record (EHR)-integrated prognostic CDS versus usual care for promoting appropriate ED disposition in children (<18 years) with pneumonia. Encounters were randomized 1:1 to usual care versus custom CDS featuring a validated pneumonia severity score predicting risk for severe in-hospital outcomes. Clinicians retained full decision-making authority. The primary outcome was inappropriate ED disposition, defined as early transition to lower- or higher-level care. Safety and implementation outcomes were also evaluated. RESULTS: The study enrolled 536 encounters (269 usual care and 267 CDS). Baseline characteristics were similar across arms. Inappropriate disposition occurred in 3% of usual care encounters and 2% of CDS encounters (adjusted odds ratio: 0.99, 95% confidence interval: [0.32, 2.95]). Length of stay was also similar and adverse safety outcomes were uncommon in both arms. The tool's custom user interface and content were viewed as strengths by surveyed clinicians (>70% satisfied). Implementation barriers include intrinsic (e.g., reaching the right person at the right time) and extrinsic factors (i.e., global pandemic). CONCLUSIONS: EHR-based prognostic CDS did not improve ED disposition decisions for children with pneumonia. Although the intervention's content was favorably received, low subject accrual and workflow integration problems likely limited effectiveness. Clinical Trials Registration: NCT06033079.

Association between Preoperative Blood Pressures and Postoperative Adverse Events.

BACKGROUND: The relationship between postoperative adverse events and blood pressures in the preoperative period remains poorly understood. This study tested the hypothesis that day-of-surgery preoperative blood pressures are associated with postoperative adverse events. METHODS: The authors conducted a retrospective, observational study of adult patients having elective procedures requiring an inpatient stay between November 2017 and July 2021 at Vanderbilt University Medical Center to examine the independent associations between preoperative systolic and diastolic blood pressures (SBP, DBP) recorded immediately before anesthesia care and number of postoperative adverse events-myocardial injury, stroke, acute kidney injury, and mortality-while adjusting for potential confounders. The study used multivariable ordinal logistic regression to model the relationship. RESULTS: The analysis included 57,389 cases. The overall incidence of myocardial injury, stroke, acute kidney injury, and mortality within 30 days of surgery was 3.4% (1,967 events), 0.4% (223), 10.2% (5,871), and 2.1% (1,223), respectively. The independent associations between both SBP and DBP measurements and number of postoperative adverse events were found to be U-shaped, with greater risk both above and less than SBP 143 mmHg and DBP 86 mmHg-the troughs of the curves. The associations were strongest at SBP 173 mmHg (adjusted odds ratio, 1.212 vs. 143 mmHg; 95% CI, 1.021 to 1.439; P = 0.028), SBP 93 mmHg (adjusted odds ratio, 1.339 vs. 143 mmHg; 95% CI, 1.211 to 1.479; P < 0.001), DBP 106 mmHg (adjusted odds ratio, 1.294 vs. 86 mmHg; 95% CI, 1.003 to 1.17671; P = 0.048), and DBP 46 mmHg (adjusted odds ratio, 1.399 vs. 86 mmHg; 95% CI, 1.244 to 1.558; P < 0.001). CONCLUSIONS: Preoperative blood pressures both less than and above a specific threshold were independently associated with a higher number of postoperative adverse events, but the data do not support specific strategies for managing patients with low or high blood pressure on the day of surgery.

Effectiveness of Oral Iron Therapy in Anemic Inpatient Pregnant Women: A Single Center Retrospective Cohort Study.

Background and aim Oral iron therapy is effective in treating iron deficiency anemia in outpatient pregnant women but has not been studied in inpatient pregnant women. We aimed to evaluate the effect of oral iron therapy versus no therapy during hospitalization on maternal and neonatal outcomes in women with anemia who are hospitalized for pregnancy-related morbidities (i.e., preterm premature rupture of membranes, preterm labor, pre-eclampsia, abnormal placentation, or fetal monitoring). Methods A retrospective, single-center study was conducted in hospitalized pregnant women (2018 to 2020) with inpatient stays of more than three days. The primary outcome was a change in hemoglobin level from admission to delivery in women treated with oral iron compared with those left untreated. Secondary outcomes included the total amount of iron administered before delivery, the time interval from admission to delivery, and neonatal effects. Results Two hundred sixty-three women were admitted, 79 women had anemia, and 29 (36.7%) received at least one dose of oral iron. Baseline patient characteristics were similar between groups. The median (interquartile range) dose of iron in the oral iron group was 1185.0 (477.0, 1874.0) mg. Neither absolute hemoglobin before delivery (control group: 10.0±1.2 g/dL; iron group: 10.1±1.1 g/dL; p=0.774) nor change in hemoglobin from admission to delivery (control group: -0.1±1.1 g/dL vs. iron group: 0.4±1.1 g/dL; p=0.232) differed between groups. Women in the control group had shorter length of stay (LOS) median (IQR) than women in the iron group (control group: 7.1 (5.0, 13.7) days; iron group: 11.4 (7.4, 25.9) days; p=0.03). There were no differences in maternal mode of delivery, though each group had high rates of cesarean delivery (control group: 53.7%; iron group: 72.4%; p=0.181). There were no differences in estimated blood loss at delivery (control group: 559±401; iron group: 662.1±337.4;p=0.264) in either group. Neonatal birthweight (control group: 1.9±0.7 kg; iron group: 1.9±0.7 kg; p=0.901), birth hemoglobin (control group: 16.3±2.2 g/dL; iron group: 16±2.2 g/dL; p=0.569), neonatal intensive care unit (NICU) admission (control group: 93.3%; iron group: 84.8%;p=0.272 ), or neonatal death (control group: 8.9%; iron group: 3%; p=0.394) were not different between groups. Conclusions Oral iron administered to anemic inpatient pregnant women was not associated with higher hemoglobin concentrations before delivery. Lack of standardized iron regimens and short hospital stays may contribute to the inefficacy of oral iron for this inpatient pregnant population. The small sample size and retrospective nature of this study are limiting factors in drawing conclusive evidence from this study.

The impact of a medically directed student registered nurse anesthesia staffing model on Postprocedural patient outcomes.

STUDY OBJECTIVE: In 2018, the American Society of Anesthesiologists stated that student registered nurse anesthetists (SRNAs) "are not yet fully qualified anesthesia personnel." It remains unclear, however, whether postprocedural outcomes are affected by SRNAs providing anesthesia care under the medical direction of anesthesiologists, as compared with medically directed anesthesiology fellows or residents, or certified registered nurse anesthetists (CRNAs). We therefore aimed to examine whether medically directed SRNAs serving as in-room anesthesia providers impact surgical outcomes. DESIGN: Retrospective, matched-cohort analysis. SETTING: Adult patients (≥18 years old) undergoing inpatient surgery between 2000 and 2017 at a tertiary academic medical center. PATIENTS: 15,365 patients exclusively cared for by medically directed SRNAs were matched to 15,365 cared for by medically directed CRNAs, anesthesiology residents, and/or fellows. INTERVENTIONS: None. MEASUREMENTS: The primary composite outcome was postoperative occurrence of in-hospital mortality and six categories of major morbidities (infectious, bleeding, serious cardiac, gastrointestinal, respiratory, and urinary complications). In-hospital mortality was analyzed as the secondary outcome. MAIN RESULTS: In all, 30,730 cases were matched using propensity score matching to control for potential confounding. The primary outcome was identified in 2295 (7.5%) cases (7.5% with exclusive medically directed SRNAs vs 7.4% with medically directed CRNAs, residents and/or fellows; relative risk, 1.02; 95% CI, 0.94-1.11). Thus, our effort to determine noninferiority (10% difference in relative risk) with other providers was inconclusive (P = .07). However, the medically directed SRNA group (0.8% [118]) was found to be noninferior (P < .001) to the matched group (1.0% [156]) on in-hospital mortality (relative risk, 0.75; 95% CI, 0.59-0.96). CONCLUSIONS: Among 30,730 patients undergoing inpatient surgery at a single hospital, findings were inconclusive regarding whether exclusive medically directed SRNAs as in-room providers were noninferior to other providers. The use of medically directed SRNAs under this staffing model should be subject to further review. Clinical Trial and Registry URL: Not applicable.

Machine Learning Predicts Cerebral Vasospasm in Subarachnoid Hemorrhage Patients.

Background: Cerebral vasospasm (CV) is a feared complication occurring in 20-40% of patients following subarachnoid hemorrhage (SAH) and is known to contribute to delayed cerebral ischemia. It is standard practice to admit SAH patients to intensive care for an extended period of vigilant, resource-intensive, clinical monitoring. We used machine learning to predict CV requiring verapamil (CVRV) in the largest and only multi-center study to date. Methods: SAH patients admitted to UCLA from 2013-2022 and a validation cohort from VUMC from 2018-2023 were included. For each patient, 172 unique intensive care unit (ICU) variables were extracted through the primary endpoint, namely first verapamil administration or ICU downgrade. At each institution, a light gradient boosting machine (LightGBM) was trained using five- fold cross validation to predict the primary endpoint at various timepoints during hospital admission. Receiver-operator curves (ROC) and precision-recall (PR) curves were generated. Results: A total of 1,750 patients were included from UCLA, 125 receiving verapamil. LightGBM achieved an area under the ROC (AUC) of 0.88 an average of over one week in advance, and successfully ruled out 8% of non-verapamil patients with zero false negatives. Minimum leukocyte count, maximum platelet count, and maximum intracranial pressure were the variables with highest predictive accuracy. Our models predicted "no CVRV" vs "CVRV within three days" vs "CVRV after three days" with AUCs=0.88, 0.83, and 0.88, respectively. For external validation at VUMC, 1,654 patients were included, 75 receiving verapamil. Predictive models at VUMC performed very similarly to those at UCLA, averaging 0.01 AUC points lower. Conclusions: We present an accurate (AUC=0.88) and early (>1 week prior) predictor of CVRV using machine learning over two large cohorts of subarachnoid hemorrhage patients at separate institutions. This represents a significant step towards optimized clinical management and improved resource allocation in the intensive care setting of subarachnoid hemorrhage patients.

Improving Neuromuscular Monitoring Through Education-Based Interventions and Studying Its Association With Adverse Postoperative Outcomes: A Retrospective Observational Study.

BACKGROUND: We assessed the association between education-based interventions, the frequency of train-of-four (TOF) monitoring, and postoperative outcomes. METHODS: We studied adults undergoing noncardiac surgery from February 1, 2020 through October 31, 2021. Our education-based interventions consisted of 3 phases. An interrupted time-series analysis, adjusting for patient- and procedure-related characteristics and secular trends over time, was used to assess the associations between education-based interventions and the frequency of TOF monitoring, postoperative pulmonary complications (PPCs), 90-day mortality, and sugammadex dosage. For each outcome and intervention phase, we tested whether the intervention at that phase was associated with an immediate change in the outcome or its trend (weekly rate of change) over time. In a sensitivity analysis, the association between education-based interventions and postoperative outcomes was adjusted for TOF monitoring. RESULTS: Of 19,422 cases, 11,636 (59.9%) had documented TOF monitoring. Monitoring frequency increased from 44.2% in the first week of preintervention stage to 83.4% in the final week of the postintervention phase. During the preintervention phase, the odds of TOF monitoring trended upward by 0.5% per week (odds ratio [OR], 1.005; 95% confidence interval [CI], 1.002-1.007). Phase 1 saw an immediate 54% increase (OR, 1.54; 95% CI, 1.33-1.79) in the odds, and the trend OR increased by 3% (OR, 1.03; 95% CI, 1.01-1.05) to 1.035, or 3.5% per week (joint Wald test, P < .001). Phase 2 was associated with a further immediate 29% increase (OR, 1.29; 95% CI, 1.02-1.64) but no significant association with trend (OR, 0.96; 95% CI, 0.93-1.01) of TOF monitoring (joint test, P = .04). Phase 3 and postintervention phase were not significantly associated with the frequency of TOF monitoring (joint test, P = .16 and P = .61). The study phases were not significantly associated with PPCs or sugammadex administration. The trend OR for 90-day mortality was larger by 24% (OR, 1.24; 95% CI, 1.06-1.45; joint test, P = .03) in phase 2 versus phase 1, from a weekly decrease of 8% to a weekly increase of 14%. However, this trend reversed again at the transition from phase 3 to the postintervention phase (OR, 0.82; 95% CI, 0.68-0.99; joint test, P = .05), from a 14% weekly increase to a 6.2% weekly decrease in the odds of 90-day mortality. In sensitivity analyses, adjusting for TOF monitoring, we found similar associations between study initiatives and postoperative outcomes. TOF monitoring was associated with lower odds of PPCs (OR, 0.69; 95% CI, 0.55-0.86) and 90-day mortality (OR, 0.79; 95% CI, 0.63-0.98), but not sugammadex dosing (mean difference, -0.02; 95% CI, -0.04 to 0.01). CONCLUSIONS: Our education-based interventions were associated with both TOF utilization and 90-day mortality but were not associated with either the odds of PPCs or sugammadex dosing. TOF monitoring was associated with reduced odds of PPCs and 90-day mortality.

Oxygen-Saturation Targets and Cognitive and Functional Outcomes in Mechanically Ventilated Adults.

Rationale: Among mechanically ventilated critically ill adults, the PILOT (Pragmatic Investigation of Optimal Oxygen Targets) trial demonstrated no difference in ventilator-free days among lower, intermediate, and higher oxygen-saturation targets. The effects on long-term cognition and related outcomes are unknown.Objectives: To compare the effects of lower (90% [range, 88-92%]), intermediate (94% [range, 92-96%]), and higher (98% [range, 96-100%]) oxygen-saturation targets on long-term outcomes.Methods: Twelve months after enrollment in the PILOT trial, blinded neuropsychological raters conducted assessments of cognition, disability, employment status, and quality of life. The primary outcome was global cognition as measured using the Telephone Montreal Cognitive Assessment. In a subset of patients, an expanded neuropsychological battery measured executive function, attention, immediate and delayed memory, verbal fluency, and abstraction.Measurements and Main Results: A total of 501 patients completed follow-up, including 142 in the lower, 186 in the intermediate, and 173 in the higher oxygen target groups. Median (interquartile range) peripheral oxygen saturation values in the lower, intermediate, and higher target groups were 94% (91-96%), 95% (93-97%), and 97% (95-99%), respectively. Telephone Montreal Cognitive Assessment score did not differ between lower and intermediate (adjusted odds ratio [OR], 1.36 [95% confidence interval (CI), 0.92-2.00]), intermediate and higher (adjusted OR, 0.90 [95% CI, 0.62-1.29]), or higher and lower (adjusted OR, 1.22 [95% CI, 0.83-1.79]) target groups. There was also no difference in individual cognitive domains, disability, employment, or quality of life.Conclusions: Among mechanically ventilated critically ill adults who completed follow-up at 12 months, oxygen-saturation targets were not associated with cognition or related outcomes.

Public Health Informatics and the Perioperative Physician: Looking to the Future.

The role of informatics in public health has increased over the past few decades, and the coronavirus disease 2019 (COVID-19) pandemic has underscored the critical importance of aggregated, multicenter, high-quality, near-real-time data to inform decision-making by physicians, hospital systems, and governments. Given the impact of the pandemic on perioperative and critical care services (eg, elective procedure delays; information sharing related to interventions in critically ill patients; regional bed-management under crisis conditions), anesthesiologists must recognize and advocate for improved informatic frameworks in their local environments. Most anesthesiologists receive little formal training in public health informatics (PHI) during clinical residency or through continuing medical education. The COVID-19 pandemic demonstrated that this knowledge gap represents a missed opportunity for our specialty to participate in informatics-related, public health-oriented clinical care and policy decision-making. This article briefly outlines the background of PHI, its relevance to perioperative care, and conceives intersections with PHI that could evolve over the next quarter century.

Analysis of the factors contributing to residual weakness after sugammadex administration in pediatric patients under 2 years of age.

BACKGROUND: Sugammadex reverses the neuromuscular blockade induced by rocuronium and vecuronium and is approved by the U.S. Food and Drug Administration for use in patients aged over 2 years. There is, however, a paucity of data regarding its dosing profile in infants and children younger than 2 years. AIMS: The aim of this study was to assess the risk of recurarization, or re-paralysis, in children under 2 years of age to increase awareness on the importance of appropriate neuromuscular blocked monitoring and reversal. METHODS: All patients aged ≤24 months who underwent an operative procedure at a tertiary medical center between January 1, 2018, and December 31, 2021, and received both rocuronium for neuromuscular blockade and sugammadex for neuromuscular blockade reversal, were identified in the electronic medical record. Patients were excluded from analysis if they (1) received vecuronium, cisatracurium, atracurium, or succinylcholine for neuromuscular blockade, (2) received neostigmine for reversal, or (3) underwent more than one operation within 24 h. We performed a survival analysis of sugammadex redose using a Cox proportional hazards model. RESULTS: We reviewed 2923 records. Sugammadex was redosed in 123 (4.2%) cases. The median [IQR] time to redose was 7 [4-17] min, and the median [IQR] amount of redose administered was 2.74 [1.96-3.99] mg/kg. Increasing patient age (p < .01) and weight (p < .01) were associated with reduced hazard rate of sugammadex redose. For a patient of median weight, increasing age from 3 to 13 months was associated with a 53% risk reduction (HR: 0.47; 95% CI: 0.24-0.91). For a patient of median age, increasing weight from 4.7 to 9.2 kg was associated with 41% risk reduction (HR: 0.59; 95% CI: 0.32-1.07). We failed to detect any other associations. CONCLUSIONS: In this single-center, retrospective cohort study of pediatric surgery patients, there was an association between the hazard of sugammadex redose with both increased age and weight.

External validation of a predictive model for reintubation after cardiac surgery: A retrospective, observational study.

STUDY OBJECTIVE: Explore validation of a model to predict patients' risk of failing extubation, to help providers make informed, data-driven decisions regarding the optimal timing of extubation. DESIGN: We performed temporal, geographic, and domain validations of a model for the risk of reintubation after cardiac surgery by assessing its performance on data sets from three academic medical centers, with temporal validation using data from the institution where the model was developed. SETTING: Three academic medical centers in the United States. PATIENTS: Adult patients arriving in the cardiac intensive care unit with an endotracheal tube in place after cardiac surgery. INTERVENTIONS: Receiver operating characteristic (ROC) curves and concordance statistics were used as measures of discriminative ability, and calibration curves and Brier scores were used to assess the model's predictive ability. MEASUREMENTS: Temporal validation was performed in 1642 patients with a reintubation rate of 4.8%, with the model demonstrating strong discrimination (optimism-corrected c-statistic 0.77) and low predictive error (Brier score 0.044) but poor model precision and recall (Optimal F1 score 0.29). Combined domain and geographic validation were performed in 2041 patients with a reintubation rate of 1.5%. The model displayed solid discriminative ability (optimism-corrected c-statistic = 0.73) and low predictive error (Brier score = 0.0149) but low precision and recall (Optimal F1 score = 0.13). Geographic validation was performed in 2489 patients with a reintubation rate of 1.6%, with the model displaying good discrimination (optimism-corrected c-statistic = 0.71) and predictive error (Brier score = 0.0152) but poor precision and recall (Optimal F1 score = 0.13). MAIN RESULTS: The reintubation model displayed strong discriminative ability and low predictive error within each validation cohort. CONCLUSIONS: Future work is needed to explore how to optimize models before local implementation.

Development and Validation of an Automated Tool to Retrieve and Curate Faculty Publications of Academic Departments.

Introduction Academic departments need to monitor their faculty's academic productivity for various purposes, such as reporting to the medical school dean, assessing the allocation of non-clinical research time, evaluating for rank promotion, and reporting to the Accreditation Council for Graduate Medical Education (ACGME). Our objective was to develop and validate a simple method that automatically generates query strings to identify and process distinct department faculty publications listed in PubMed and Scopus. Methods We created a macro-enabled Excel workbook (Microsoft, Redmond, WA) to automate the retrieval of faculty publications from the PubMed and Scopus bibliometric databases (available at https://bit.ly/get-pubs). Where the returned reference includes the digital object identifier (doi), a link is provided in the workbook. Duplicate publications are removed automatically, and false attributions are managed. Results At the University of Miami, between 2020 and 2021, there were 143 anesthesiology faculty-authored publications with a PubMed identifier (PMID), 95.8% identified by the query and 4.2% missed. At Vanderbilt University Medical Center, between 2019 and 2021, there were 760 anesthesiology faculty-authored publications with a PMID, 94.3% identified by the query and 5.7% missed. Recall, precision, and the F1 score were all above 93% at both medical centers. Conclusions We developed a highly accurate, simple, transportable, scalable method to identify publications in PubMed and Scopus authored by anesthesiology faculty. Manual checking and faculty feedback are required because not all names can be disambiguated, and some references are missed. This process can greatly reduce the burden of curating a list of faculty publications. The methodology applies to other academic departments that track faculty publications.

Postdischarge Survival After Sepsis: A Cohort Study.

BACKGROUND: After hospital discharge, patients who had sepsis have increased mortality. We sought to estimate factors associated with postdischarge mortality and how they vary with time after discharge. METHODS: This was a retrospective study of hospital survivors of sepsis using time-varying Cox proportional hazard models, which produce a baseline hazard ratio (HR) and a second number (δHR) that reflects the amount by which the baseline HR changes with time. RESULTS: Of the 32,244 patients who survived sepsis at hospital discharge, 13,565 patients (42%) died (mean ± standard deviation: 1.41 ± 1.87 years) after discharge from the index hospitalization, while 18,679 patients were still alive at follow-up (4.98 ± 2.86 years). The mortality rate decreased with time after discharge: approximately 8.7% of patients died during the first month after discharge, 1.1% of patients died during the 12th month after discharge, and 0.3%% died during the 60th month; after Kaplan-Meier analysis, survival was 91% (95% confidence interval [CI], 91%-92%) at 1 month, 76% (95% CI, 76%-77%) at 1 year, 57% (95% CI, 56%-58%) at 5 years, and 48% (95% CI, 47%-48%) at 10 years after discharge. Organ dysfunction at discharge was associated with worse survival. In particular, elevated urea nitrogen at discharge (HR, 1.10 per 10 mg/dL, 95% CI, 1.08-1.12, P < .001) was associated with increased mortality, but the HR decreased with time from discharge (δHR, 0.98 per 10 mg/dL per year, 95% CI, 0.98-0.99, P < .001). Higher hemoglobin levels were associated with lower mortality (HR, 0.92 per g/dL, 95% CI, 0.91-0.93, P < .001), but this association increased with increasing time after discharge (δHR, 1.02 per g/dL per year, 95% CI, 1.01-1.02, P < .001). Older age was associated with an increased risk of mortality (HR, 1.29 per decade of age, 95% CI, 1.27-1.31, P < .001) that grew with increasing time after discharge (δHR, 1.01 per year of follow-up per decade of age, 95% CI, 1.00-1.02, P < .001). Compared to private insurances Medicaid as primary insurance was associated with an increased risk of mortality (HR, 1.17, 95% CI, 1.10-1.25, P < .001) that did not change with time after discharge. In contrast, Medicare status was initially associated with a similar risk of mortality as private insurance at discharge (HR, 1), but was associated with greater risk as time after discharge increased (δHR, 1.04 per year of follow-up, 95% CI, 1.03-1.05, P < .001). CONCLUSIONS: Acute physiologic derangements and organ dysfunction were associated with postdischarge mortality with the associations decreasing over time.

Development of a machine learning model to predict intraoperative transfusion and guide type and screen ordering.

STUDY OBJECTIVE: To develop an algorithm to predict intraoperative Red Blood Cell (RBC) transfusion from preoperative variables contained in the electronic medical record of our institution, with the goal of guiding type and screen ordering. DESIGN: Machine Learning model development on retrospective single-center hospital data. SETTING: Preoperative period and operating room. PATIENTS: The study included patients ≥18 years old who underwent surgery during 2019-2022 and excluded those who refused transfusion, underwent emergency surgery, or surgery for organ donation after cardiac or brain death. INTERVENTION: Prediction of intraoperative transfusion vs. no intraoperative transfusion. MEASUREMENTS: The outcome variable was intraoperative transfusion of RBCs. Predictive variables were surgery, surgeon, anesthesiologist, age, sex, body mass index, race or ethnicity, preoperative hemoglobin (g/dL), partial thromboplastin time (s), platelet count x 109 per liter, and prothrombin time. We compared the performances of seven machine learning algorithms. After training and optimization on the 2019-2021 dataset, model thresholds were set to the current institutional performance level of sensitivity (93%). To qualify for comparison, models had to maintain clinically relevant sensitivity (>90%) when predicting on 2022 data; overall accuracy was the comparative metric. MAIN RESULTS: Out of 100,813 cases that met study criteria from 2019 to 2021, intraoperative transfusion occurred in 5488 (5.4%) of cases. The LightGBM model was the highest performing algorithm in external temporal validity experiments, with overall accuracy of (76.1%) [95% confidence interval (CI), 75.6-76.5], while maintaining clinically relevant sensitivity of (91.2%) [95% CI, 89.8-92.5]. If type and screens were ordered based upon the LightGBM model, the predicted type and screen to transfusion ratio would improve from 8.4 to 5.1. CONCLUSIONS: Machine learning approaches are feasible in predicting intraoperative transfusion from preoperative variables and may improve preoperative type and screen ordering practices when incorporated into the electronic health record.

PeRiOperative Glucose PRAgMatic (PROGRAM) trial protocol and statistical analysis plan for comparing automated intraoperative reminders to standardise insulin administration in surgical patients at high risk of hyperglycaemia.

INTRODUCTION: Studies finding perioperative hyperglycaemia is associated with adverse patient outcomes in surgical procedures spurred the development of blood glucose guidelines at many institutions. In this trial, we will assess the implementation of a clinical decision support tool that is integrated into the intraoperative portion of our electronic health record and provides real-time best practice recommendations for intraoperative insulin dosing in surgical patients at high risk for hyperglycaemia. METHODS AND DESIGN: We will assess this intervention using a sequential and repeated cross-over design at the institutional level with periods of time for wash-out, control and study intervention. The unit of analysis will be the surgical case. The primary outcome will be the frequency of hyperglycaemia (>180 mg/dL (10 mmol/L)) at first postoperative anaesthesia care unit measurement. There are several prespecified secondary analyses focused on perioperative glycaemic control. DISCUSSION: This protocol and statistical analysis plan describes the methodology, primary and secondary analyses. The PeRiOperative Glucose PRAgMatic (PROGRAM) trial was approved by the Vanderbilt University Institutional Review Board (IRB), Vanderbilt University Medical Center, Nashville, Tennessee, USA (IRB, 220991). The study results will be disseminated via publication in a peer-reviewed journal and presented at national scientific conferences. The results of PROGRAM trial will inform best practice for perioperative standardised insulin administration in surgical patients at high risk of hyperglycaemia. TRIAL REGISTRATION NUMBER: NCT05426096.

Factors Associated With Initiation of Mechanical Ventilation in Patients With Sepsis: Retrospective Observational Study.

BACKGROUND: Patients with sepsis are at risk for mechanical ventilation. This study aimed to identify risk factors for initiation of mechanical ventilation in patients with sepsis and assess whether these factors varied with time. METHODS: Data from the electronic health record were used to model risk factors for initiation of mechanical ventilation after the onset of sepsis. A time-varying Cox model was used to study factors that varied with time. RESULTS: Of 35 020 patients who met sepsis criteria, 28 747 were eligible for inclusion. Mechanical ventilation was initiated within 30 days after sepsis onset in 3891 patients (13.5%). Factors that were independently associated with increased likelihood of receipt of mechanical ventilation were race (White: adjusted hazard ratio [HR], 1.59; 95% CI, 1.39-1.83; other/unknown: adjusted HR, 1.97; 95% CI, 1.54-2.52), systemic inflammatory response syndrome (adjusted HR [per point], 1.23; 95% CI, 1.17-1.28), Sequential Organ Failure Assessment score (adjusted HR [per point], 1.28; 95% CI, 1.26-1.31), and congestive heart failure (adjusted HR, 1.30; 95% CI, 1.17-1.45). Hazard ratios decreased with time for Sequential Organ Failure Assessment score and congestive heart failure and varied with time for 4 comorbidities and 3 culture results. CONCLUSIONS: The risk for mechanical ventilation associated with different factors varied with time after sepsis onset, increasing for some factors and decreasing for others. Through a better understanding of risk factors for initiation of mechanical ventilation in patients with sepsis, targeted interventions may be tailored to high-risk patients.

The IMpact of PerioperAtive KeTamine on Enhanced Recovery after Abdominal Surgery (IMPAKT ERAS): protocol for a pragmatic, randomized, double-blinded, placebo-controlled trial.

BACKGROUND: Multimodal analgesic strategies that reduce perioperative opioid consumption are well-supported in Enhanced Recovery After Surgery (ERAS) literature. However, the optimal analgesic regimen has not been established, as the contributions of each individual agent to the overall analgesic efficacy with opioid reduction remains unknown. Perioperative ketamine infusions can decrease opioid consumption and opioid-related side effects. However, as opioid requirements are drastically minimized within ERAS models, the differential effects of ketamine within an ERAS pathway remain unknown. We aim to pragmatically investigate through a learning healthcare system infrastructure how the addition of a perioperative ketamine infusion to mature ERAS pathways affects functional recovery. METHODS: The IMPAKT ERAS trial (IMpact of PerioperAtive KeTamine on Enhanced Recovery after Abdominal Surgery) is a single center, pragmatic, randomized, blinded, placebo-controlled trial. 1544 patients undergoing major abdominal surgery will be randomly allocated to receive intraoperative and postoperative (up to 48 h) ketamine versus placebo infusions as part of a perioperative multimodal analgesic regimen. The primary outcome is length of stay, defined as surgical start time until hospital discharge. Secondary outcomes will include a variety of in-hospital clinical end points derived from the electronic health record. DISCUSSION: We aimed to launch a large-scale, pragmatic trial that would easily integrate into routine clinical workflow. Implementation of a modified consent process was critical to preserving our pragmatic design, permitting an efficient, low-cost model without reliance on external study personnel. Therefore, we partnered with leaders of our Investigational Review Board to develop a novel, modified consent process and shortened written consent form that would meet all standard elements of informed consent, yet also allow clinical providers the ability to recruit and enroll patients during their clinical workflow. Our trial design has created a platform for subsequent pragmatic studies at our institution. TRIAL REGISTRATION: NCT04625283, Pre-results.

Effects of Postoperative Blood Pressure Management on Delirium Among Patients in the Intensive Care Unit After Cardiac Surgery: An Observational Cohort Study.

OBJECTIVES: This study aimed to determine whether blood pressure control in the early postoperative period was associated with postoperative delirium in the cardiovascular intensive care unit (ICU). DESIGN: Observational cohort study. SETTING: Single large academic institution with a high volume of cardiac surgery. PARTICIPANTS: Cardiac surgery patients admitted to the cardiovascular ICU after surgery. INTERVENTIONS: Observational study. MEASUREMENTS AND MAIN RESULTS: A total of 517 cardiac surgery patients had mean arterial pressure (MAP) data recorded minute-by-minute for 12 postoperative hours. The time spent in each of the 7 prespecified blood pressure bands was calculated, and the development of delirium was recorded in the ICU. A multivariate Cox regression model was developed using the least absolute shrinkage and selection operator approach to identify associations between time spent in each MAP range band and delirium. Compared with the reference band of 60-to-69 mmHg, longer durations spent in 3 bands were independently associated with a lower risk of delirium: 50-to-59 mmHg band (adjusted hazard ratio [HR] 0.907 [per 10 minutes]; 95% CI 0.861-0.955); 70-to-79 mmHg band (adjusted HR 0.923 [per 10 minutes]; 95% CI 0.902-0.944); 90-to-99 mmHg band (adjusted HR 0.898 [per 10 minutes]; 95% CI 0.853-0.945). CONCLUSIONS: The MAP range bands above and below the authors' reference band of 60-to- 69 mmHg were associated with decreased risk of ICU delirium development; however, this was difficult to reconcile with a plausible biologic mechanism. Therefore, the authors did not find a correlation between early postoperative MAP control and increased risk of the development of ICU delirium after cardiac surgery.