Enhancing Preoperative Outcome Prediction: A Comparative Retrospective Case-Control Study on Machine Learning versus the International Esodata Study Group Risk Model for Predicting 90-Day Mortality in Oncologic Esophagectomy.

Risk prediction prior to oncologic esophagectomy is crucial for assisting surgeons and patients in their joint informed decision making. Recently, a new risk prediction model for 90-day mortality after esophagectomy using the International Esodata Study Group (IESG) database was proposed, allowing for the preoperative assignment of patients into different risk categories. However, given the non-linear dependencies between patient- and tumor-related risk factors contributing to cumulative surgical risk, machine learning (ML) may evolve as a novel and more integrated approach for mortality prediction. We evaluated the IESG risk model and compared its performance to ML models. Multiple classifiers were trained and validated on 552 patients from two independent centers undergoing oncologic esophagectomies. The discrimination performance of each model was assessed utilizing the area under the receiver operating characteristics curve (AUROC), the area under the precision-recall curve (AUPRC), and the Matthews correlation coefficient (MCC). The 90-day mortality rate was 5.8%. We found that IESG categorization allowed for adequate group-based risk prediction. However, ML models provided better discrimination performance, reaching superior AUROCs (0.64 [0.63-0.65] vs. 0.44 [0.32-0.56]), AUPRCs (0.25 [0.24-0.27] vs. 0.11 [0.05-0.21]), and MCCs (0.27 ([0.25-0.28] vs. 0.15 [0.03-0.27]). Conclusively, ML shows promising potential to identify patients at risk prior to surgery, surpassing conventional statistics. Still, larger datasets are needed to achieve higher discrimination performances for large-scale clinical implementation in the future.

Outcome and management of decompressive hemicraniectomy in malignant hemispheric stroke following cardiothoracic surgery.

Management of malignant hemispheric stroke (MHS) after cardiothoracic surgery (CTS) remains difficult as decision-making needs to consider severe cardiovascular comorbidities and complex coagulation management. The results of previous randomized controlled trials on decompressive surgery for MHS cannot be generally translated to this patient population and the expected outcome might be substantially worse. Here, we analyzed mortality and functional outcome in patients undergoing decompressive hemicraniectomy (DC) for MHS following CTS and assessed the impact of perioperative coagulation management on postoperative hemorrhagic and cardiovascular complications. All patients that underwent DC for MHS resulting as a complication of CTS between June 2012 and November 2021 were included in this observational cohort study. Outcome was determined according to the modified Rankin Scale (mRS) score at 1 and 3-6 months. Clinical and demographic data, anticoagulation management and postoperative hemorrhagic and thromboembolic complications were assessed. In order to evaluate a predictive association between clinical and radiological parameters and the outcome, we used a multivariate logistic regression analysis. Twenty-nine patients undergoing DC for MHS after CTS with a female-to-male ratio of 1:1.9 and a median age of 60 (IQR 49-64) years were identified out of 123 patients undergoing DC for MHS. Twenty-four patients (83%) received pre- or intraoperative substitution. At 30 days, the in-hospital mortality rate and neurological outcome corresponded to 31% and a median mRS of 5 (5-6), which remained stable at 3-6 months [Mortality: 42%, median mRS: 5 (4-6)]. Postoperatively, 15/29 patients (52%) experienced new hemorrhagic lesions and Bayesian logistic regression predicting mortality (mRS = 6) after imputing missing data demonstrated a significantly increased risk for mortality with longer aPPT (OR = 13.94, p = .038) and new or progressive hemorrhagic lesions after DC (OR = 3.03, p = .19). Notably, all but one hemorrhagic lesion occurred before discontinued anticoagulation and/or platelet inhibition was re-initiated. Despite perioperative discontinuation of anticoagulation and/or platelet inhibition, no coagulation-associated cardiovascular complications were noted. In conclusion, Cardiothoracic surgery patients suffering MHS will likely experience severe neurological disability after DC, which should remain a central aspect during counselling and decision-making. The complex coagulation situation after CTS, however, should not per se rule out the option of performing life-saving surgical decompression.

Body computed tomography in sepsis: predictors of CT findings and patient outcomes in a retrospective medical ICU cohort study.

BACKGROUND: Sepsis is a life-threatening condition that requires immediate focus identification and control. However, international sepsis guidelines do not provide information on imaging choice. PURPOSE: To identify predictors of CT findings and patient outcomes in a population of septic patients from a medical ICU. MATERIAL AND METHODS: A full-text search in the radiological information system (RIS) retrieved 227 body CT examinations conducted to identify infectious sources in 2018. CT reports were categorized according to identified foci and their diagnostic certainty. Diagnostic accuracy of CT was compared to microbiological results. Clinical and laboratory information was gathered. Statistical analysis was performed using nonparametric tests and logistic regression analysis. RESULTS: CT revealed more positive infectious foci 52.4% (n = 191/227) than microbiological tests 39.3% (n = 79/201). There were no significant differences between focus-positive CT scans with regard to positive microbiological testing (p = 0.32). Sequential organ failure assessment (SOFA) scores were slightly but nonsignificantly higher in patients with a focus-positive CT, odds ratio (OR) = 0.999 (95% CI 0.997-1.001) with p = 0.52. Among C-reactive protein (CRP), procalcitonin (PCT), and leukocytes, in focus-positive versus focus-negative CT scans, CRP showed a minor but statistically significant elevation in the group with focus-positive CT scans (OR = 1.004, 95% CI = 1.000-1.007, p = 0.04). No significant association was found for PCT (OR = 1.007, 95% CI = 0.991-1.023; p = 0.40) or leukocytes (OR = 1.003, 95% CI = 0.970-1.038; p = 0.85). In 33.5% (n = 76/227) of cases, the CT findings had at least one therapeutic consequence. In 81.6% (n = 62/76), the CT findings resulted in one consequence, in 14.5% (n = 11/76) in two consequences, and in 3.9% (n = 3/76) in three consequences. There was no significant association between focus-positive CT scans and mortality (p = 0.81). CONCLUSION: In this population of septic patients in medical intensive care, microbiological analysis complemented CT findings. Both clinical and laboratory parameters were not predictive of CT findings. While therapeutic consequences of CT findings in this study population underline the role of CT for decision making in septic patients, CT findings do not predict patient outcomes in this retrospective analysis.

Cerebrovascular Events in Suspected Sepsis: Retrospective Prevalence Study in Critically Ill Patients Undergoing Full-Body Computed Tomography.

Purpose: This study aimed at retrospectively evaluating full-body computed tomography (CT) examinations for the prevalence of cerebrovascular events in patients with suspected sepsis treated in the intensive care unit (ICU). Methods: All full-body CT examinations, i.e., both cranial CT (cCT) and body CT including chest, abdomen and pelvis, for focus search in septic patients over a 12-months period were identified from three ICUs, using full-text search. From this retrospective cohort, we fully analyzed 278 cCT examinations for the occurrence of acute cerebral findings. All acute cerebrovascular events were independently reviewed by two blinded readers. Clinical and laboratory findings were extracted. The data were statistically analyzed using contingency tests. Results: In our population of patients with suspected sepsis, 10.8% (n = 30/278) were identified to have major cerebral events, including 7.2% (n = 20/278) major cerebrovascular events and 4.3% (n = 12/278) generalized parenchymal damage. 13.4% (n = 22/163) of patients with a severe coma as compared with non-severe coma, 4.4% (n = 3/68), showed a major cerebral event (p = 0.04). Patients referred from the cardiology/nephrology ICU ward showed major cerebral events in 16.3% (n = 22/135), as compared with 4.9% (n = 3/61) in patients from pulmonary ICU and 6.1% (n = 5/82) major cerebral events with surgical referral (p = 0.02). Conclusion: Our study provides further evidence that septic patients may suffer from cerebral events with relevance to their prognosis. Severe coma and the referring ward were associated with acute cerebral conditions. Full-body CT has the advantage of both detecting of septic foci and possibly identifying ischemic or hemorrhagic stroke in this vulnerable patient population.

Does adherence to a quality indicator regarding early weaning from invasive ventilation improve economic outcome? A single-centre retrospective study.

OBJECTIVES: To measure and assess the economic impact of adherence to a single quality indicator (QI) regarding weaning from invasive ventilation. DESIGN: Retrospective observational single-centre study, based on electronic medical and administrative records. SETTING: Intensive care unit (ICU) of a German university hospital, reference centre for acute respiratory distress syndrome. PARTICIPANTS: Records of 3063 consecutive mechanically ventilated patients admitted to the ICU between 2012 and 2017 were extracted, of whom 583 were eligible adults for further analysis. Patients' weaning protocols were evaluated for daily adherence to quality standards until ICU discharge. Patients with <65% compliance were assigned to the low adherence group (LAG), patients with ≥65% to the high adherence group (HAG). PRIMARY AND SECONDARY OUTCOME MEASURES: Economic healthcare costs, clinical outcomes and patients' characteristics. RESULTS: The LAG consisted of 378 patients with a median negative economic results of -€3969, HAG of 205 (-€1030), respectively (p<0.001). Median duration of ventilation was 476 (248; 769) hours in the LAG and 389 (247; 608) hours in the HAG (p<0.001). Length of stay (LOS) in the LAG on ICU was 21 (12; 35) days and 16 (11; 25) days in the HAG (p<0.001). LOS in the hospital was 36 (22; 61) days in the LAG, and within the HAG, respectively, 26 (18; 48) days (p=0.001). CONCLUSIONS: High adherence to this single QI is associated with better clinical outcome and improved economic returns. Therefore, the results support the adherence to QI. However, the examined QI does not influence economic outcome as the decisive factor.

Reduction in wound healing complications and infection rate by lumbar CSF drainage after decompressive hemicraniectomy.

OBJECTIVE: Wound healing disorders and surgical site infections are the most frequently encountered complications after decompressive hemicraniectomy (DHC). Subgaleal CSF accumulation causes additional tension of the scalp flap and increases the risk of wound dehiscence, CSF fistula, and infection. Lumbar CSF drainage might relieve subgaleal CSF accumulation and is often used when a CSF fistula through the surgical wound appears. The aim of this study was to investigate if early prophylactic lumbar drainage might reduce the rate of postoperative wound revisions and infections after DHC. METHODS: The authors retrospectively analyzed 104 consecutive patients who underwent DHC from January 2019 to May 2021. Before January 2020, patients did not receive lumbar drainage, whereas after January 2020, patients received lumbar drainage within 3 days after DHC for a median total of 4 (IQR 2-5) days if the first postoperative CT scan confirmed open basal cisterns. The primary endpoint was the rate of severe wound healing complications requiring surgical revision. Secondary endpoints were the rate of subgaleal CSF accumulations and hygromas as well as the rate of purulent wound infections and subdural empyema. RESULTS: A total of 31 patients died during the acute phase; 34 patients with and 39 patients without lumbar drainage were included for the analysis of endpoints. The predominant underlying pathology was malignant hemispheric stroke (58.8% vs 66.7%) followed by traumatic brain injury (20.6% vs 23.1%). The rate of surgical wound revisions was significantly lower in the lumbar drainage group (5 [14.7%] vs 14 [35.9%], p = 0.04). A stepwise linear regression analysis was used to identify potential covariates associated with wound healing disorder and reduced them to lumbar drainage and BMI. One patient was subject to paradoxical herniation. However, the patient's symptoms rapidly resolved after lumbar drainage was discontinued, and he survived with only moderate deficits related to the primary disease. There was no significant difference in the rate of radiological herniation signs. The median lengths of stay in the ICU were similar, with 12 (IQR 9-23) days in the drainage group compared with 13 (IQR 11-23) days in the control group (p = 0.21). CONCLUSIONS: In patients after DHC and open basal cisterns on postoperative CT, lumbar drainage appears to be safe and reduces the rate of surgical wound revisions and intracranial infection after DHC while the risk for provoking paradoxical herniation is low early after surgery.

Fluconazole Increases Osmotic Water Transport in Renal Collecting Duct through Effects on Aquaporin-2 Trafficking.

BACKGROUND: Arginine-vasopressin (AVP) binding to vasopressin V2 receptors promotes redistribution of the water channel aquaporin-2 (AQP2) from intracellular vesicles into the plasma membrane of renal collecting duct principal cells. This pathway fine-tunes renal water reabsorption and urinary concentration, and its perturbation is associated with diabetes insipidus. Previously, we identified the antimycotic drug fluconazole as a potential modulator of AQP2 localization. METHODS: We assessed the influence of fluconazole on AQP2 localization in vitro and in vivo as well as the drug's effects on AQP2 phosphorylation and RhoA (a small GTPase, which under resting conditions, maintains F-actin to block AQP2-bearing vesicles from reaching the plasma membrane). We also tested fluconazole's effects on water flow across epithelia of isolated mouse collecting ducts and on urine output in mice treated with tolvaptan, a VR2 blocker that causes a nephrogenic diabetes insipidus-like excessive loss of hypotonic urine. RESULTS: Fluconazole increased plasma membrane localization of AQP2 in principal cells independent of AVP. It also led to an increased AQP2 abundance associated with alterations in phosphorylation status and ubiquitination as well as inhibition of RhoA. In isolated mouse collecting ducts, fluconazole increased transepithelial water reabsorption. In mice, fluconazole increased collecting duct AQP2 plasma membrane localization and reduced urinary output. Fluconazole also reduced urinary output in tolvaptan-treated mice. CONCLUSIONS: Fluconazole promotes collecting duct AQP2 plasma membrane localization in the absence of AVP. Therefore, it might have utility in treating forms of diabetes insipidus (e.g., X-linked nephrogenic diabetes insipidus) in which the kidney responds inappropriately to AVP.

Nurse-Directed Blood Glucose Management in a Medical Intensive Care Unit.

BACKGROUND: Insulin-delivery algorithms for achieving glycemic control in the intensive care unit require frequent checks of blood glucose level and thus increase nursing workload. Hypoglycemia is a serious complication associated with intensive insulin therapy. OBJECTIVES: To evaluate a nurse-directed protocol for blood glucose management that allows individualized insulin delivery within a predefined blood glucose corridor, intended to avoid hypoglycemia while maintaining adequate control of blood glucose level without increasing nursing workload. METHODS: A nurse-directed protocol for blood glucose management was developed by an interprofessional team, and the protocol's performance was investigated in 175 patients compared with 384 historical controls. RESULTS: With the nurse-directed protocol, hypoglycemia incidents declined significantly (31% vs 12%, P < .001), and minimum blood glucose levels increased significantly (80 mg/dL vs 93 mg/dL, P < .001). Mean and maximum blood glucose levels, the proportion of glucose readings within the target range (31% vs 26%, P = .06), and the number of blood glucose checks (59 vs 58, P = .85) remained unchanged with use of the protocol. CONCLUSION: Implementation of the nurse-directed protocol for blood glucose management did not increase nursing workload but reduced hypoglycemia incidents significantly while maintaining adequate glycemic control.

Predictive ability of the stability and workload index for transfer score to predict unplanned readmissions after ICU discharge.

OBJECTIVE: Unplanned readmission of hospitalized patients to an ICU is associated with an increased mortality and hospital length of stay. The ability to identify patients at risk, who would benefit from prolonged ICU treatment, is limited. The aim of this study is to validate a previously published numerical index named the Stability and Workload Index for Transfer in a heterogeneous group of ICU patients. DESIGN: In this retrospective data analysis, the Stability and Workload Index for Transfer score was calculated for all patients, and the ability of the score to predict readmission was compared with the original publication. SETTING: Four ICUs, one intermediate care unit, and one postanesthesia care unit of the department of anesthesia and intensive care of a university hospital. PATIENTS: All consecutive patients treated in one of the units. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Unplanned ICU readmissions or unexpected death within 7 days of ICU discharge. The data of 7,175 patients were included in the analysis. Five hundred ninety-six patients were readmitted or died within 7 days of discharge. The patients who are readmitted to the ICU are significantly older and have significantly higher scores that define the severity of disease at the time of admission and discharge of their first ICU stay. The source of admission for the initial ICU stay did not differ (p = 0.055), and the last Glasgow Coma Scale and the last PaO2/FIO2 ratio before discharge from the ICU were higher in patients who did not need a readmission to the ICU. The performance of the Stability and Workload Index for Transfer score is poor with an area under the receiver operator curve of 0.581 (95% CI, 0.556-0.605; p < 0.001). CONCLUSIONS: Based on the data from our patients, the proposed Stability and Workload Index for Transfer score by Gajic et al is not ideal in aiding the clinician in the decision, if a patient can be discharged safely from the ICU and further research is necessary to define the patients at risk for readmission.

Catecholamine dosing and survival in adult intensive care unit patients.

BACKGROUND: Volume management and vasopressor support remain the gold standard of critical care for patients with shock. However, prolonged therapy with catecholamines in high doses is associated with a negative patient outcome. The aim of the present study was to analyze the administered levels of catecholamines over time with respect to survival, and to identify a cut-off to allow a prediction of survival. METHODS: Consecutively, 9,108 adult patients during 22 months were evaluated. This group included 1,543 patients treated with epinephrine and/or norepinephrine with any dose at any time. Time and dosages of the applied drugs, the sequential organ failure assessment and acute and chronic health evaluation II scores on admission and daily, the length of intensive care unit stay, and the outcomes were recorded. RESULTS: The non-survivors received higher doses of norepinephrine and epinephrine than the survivors (p < 0.001). The receiver operator characteristic curve for the area under the curve with non-survival as the classifier revealed a cut-off level of 294.33 µg/kg for norepinephrine with a sensitivity of 74.73 % and a specificity of 70.48 % and a cut-off for epinephrine of 70.36 µg/kg with a sensitivity of 83.87 % and a specificity of 72.79 %. Dose-dependent time curves using these cut-off values were calculated. CONCLUSIONS: Survival of patients with prolonged therapy with norepinephrine and epinephrine above the evaluated thresholds is poor, whereas short-term application of high-dose catecholamines is not associated with poor outcome. Therefore, it remains for the individual clinician, patients, and their surrogates to decide whether the use of high doses of vasopressors is appropriate in view of the low probability of survival.