The impact of uncertainties in air‐sea fluxes and ocean model parameters on the ocean circulation and ocean heat uptake (OHU) is assessed in a novel modeling framework. We use an ocean‐only model forced with the simulated sea surface fields of the CMIP5 climate models. The simulations are performed using control and 1% CO₂ warming scenarios. The ocean‐only ensemble adequately reproduces the mean Atlantic Meridional Overturning Circulation (AMOC) and the zonally integrated OHU. The ensemble spread in AMOC strength, its weakening, and Atlantic OHU due to different air‐sea fluxes is twice as large as the uncertainty range related to vertical and mesocale eddy diffusivities. The sensitivity of OHU to uncertainties in air‐sea fluxes and model parameters differs vastly across basins, with the Southern Ocean exhibiting strong sensitivity to air‐sea fluxes and model parameters. This study clearly demonstrates that model biases in air‐sea fluxes are one of the key sources of uncertainty in climate simulations.

The ideal group size for effective teaching of cardiopulmonary resuscitation is currently under debate. The upper limit is reached when instructors are unable to correct participants' errors during skills practice. This simulation study aimed to define this limit during cardiopulmonary resuscitation teaching. Medical students acting as simulated Basic Life Support course participants were instructed to make three different pre-defined Basic Life Support quality errors (e.g., chest compression too fast) in 7 min. Basic Life Support instructors were randomized to groups of 3-10 participants. Instructors were asked to observe the Basic Life Support skills and to correct performance errors. Primary outcome was the maximum group size at which the percentage of correctly identified participants' errors drops below 80%. Sixty-four instructors participated, eight for each group size. Their average age was 41 ± 9 years and 33% were female, with a median [25th percentile; 75th percentile] teaching experience of 6 [2;11] years. Instructors had taught 3 [1;5] cardiopulmonary resuscitation courses in the year before the study. A logistic binominal regression model showed that the predicted mean percentage of correctly identified participants' errors dropped below 80% for group sizes larger than six. This randomized controlled simulation trial reveals decreased ability of instructors to detect Basic Life Support performance errors with increased group size. The maximum group size enabling Basic Life Support instructors to correct more than 80% of errors is six. We therefore recommend a maximum instructor-to-participant ratio of 1:6 for cardiopulmonary resuscitation courses.

Background: Prediction modelling increasingly becomes an important risk assessment tool in perioperative systems approaches, e.g. in complex patients with open abdomen treatment for peritonitis. In this population, combining predictors from multiple medical domains (i.e. demographical, physiological and surgical variables) outperforms the prediction capabilities of single-domain prediction models. However, the benefit of these prediction models for clinical decision-making remains to be investigated. We therefore examined the clinical utility of mortality prediction models in patients suffering from peritonitis with a decision curve analysis.Methods: In this secondary analysis of a large dataset, a traditional logistic regression approach, three machine learning methods and a stacked ensemble were employed to examine the predictive capability of demographic, physiological and surgical variables in predicting mortality under open abdomen treatment for peritonitis. Calibration was examined with calibration belts and predictive performance was assessed with the area both under the receiver operating characteristic curve (AUROC) and under the precision recall curve (AUPRC) and with the Brier Score. Clinical utility of the prediction models was examined by means of a decision curve analysis (DCA) within a treatment threshold range of interest of 0–30%, where threshold probabilities are traditionally defined as the minimum probability of disease at which further intervention would be warranted.Results: Machine learning methods supported available evidence of a higher prediction performance of a multi- versus single-domain prediction models. Interestingly, their prediction performance was similar to a logistic regression model. The DCA demonstrated that the overall net benefit is largest for a multi-domain prediction model and that this benefit is larger compared to the default “treat all” strategy only for treatment threshold probabilities above about 10%. Importantly, the net benefit for low threshold probabilities is dominated by physiological predictors: surgical and demographics predictors provide only secondary decision-analytic benefit.Conclusions: DCA provides a valuable tool to compare single-domain and multi-domain prediction models and demonstrates overall higher decision-analytic value of the latter. Importantly, DCA provides a means to clinically differentiate the risks associated with each of these domains in more depth than with traditional performance metrics and highlighted the importance of physiological predictors for conservative intervention strategies for low treatment thresholds. Further, machine learning methods did not add significant benefit either in prediction performance or decision-analytic utility compared to logistic regression in these data.

Background: Traumatic brain injury (TBI) remains one of the main causes of mortality and long-term disability worldwide. Maintaining physiology of brain tissue to the greatest extent possible through optimal management of blood pressure, airway, ventilation, and oxygenation, improves patient outcome. We studied the quality of prehospital care in severe TBI patients by analyzing adherence to recommended target ranges for ventilation and blood pressure, prehospital time expenditure, and their effect on mortality, as well as quality of prehospital ventilation assessed by arterial partial pressure of CO₂ (PaCO₂) at hospital admission. Methods: This is a retrospective cohort study of all TBI patients requiring tracheal intubation on scene who were transported to one of two major level 1 trauma centers in Switzerland between January 2014 and December 2019 by Swiss Air Rescue (Rega). We assessed systolic blood pressure (SBP), end-tidal partial pressure of CO₂ (PetCO₂), and PaCO₂ at hospital admission as well as prehospital and on-scene time. Quality markers of prehospital care (PetCO₂, SBP, prehospital times) and prehospital ventilation (PaCO₂) are presented as descriptive analysis. Effect on mortality was calculated by multivariable regression analysis and a logistic general additive model. Results: Of 557 patients after exclusions, 308 were analyzed. Adherence to blood pressure recommendations was 89%. According to PetCO_2, 45% were normoventilated, and 29% had a SBP ≥ 90 mm Hg and were normoventilated. Due to the poor correlation between PaCO₂ and PetCO₂, only 33% were normocapnic at hospital admission. Normocapnia at hospital admission was strongly associated with reduced probability of mortality. Prehospital and on-scene times had no impact on mortality. Conclusions: PaCO₂ at hospital admission is strongly associated with mortality risk, but normocapnia is achieved only in a minority of patients. Therefore, the time required for placement of an arterial cannula and prehospital blood gas analysis may be warranted in severe TBI patients requiring on-scene tracheal intubation.

Background: For helicopter emergency service systems (HEMS), the prehospital time consists of response time, on-scene time and transport time. Little is known about the factors that influence on-scene time or about differences between adult and paediatric missions in a physician-staffed HEMS. Methods: We analysed the HEMS electronic database of Swiss Air-Rescue from 01-01-2011 to 31-12-2021 (N = 110,331). We included primary missions and excluded missions with National Advisory Committee for Aeronautics score (NACA) score 0 or 7, resulting in 68,333 missions for analysis. The primary endpoint ‘on-scene time’ was defined as first physical contact with the patient until take-off to the hospital. A multivariable linear regression model was computed to examine the association of diagnosis, type and number of interventions and monitoring, and patient's characteristics with the primary endpoint. Results: The prehospital time and on-scene time of the missions studied were, respectively, 50.6 [IQR: 41.0–62.0] minutes and 21.0 [IQR: 15.0–28.6] minutes. Helicopter hoist operations, resuscitation, airway management, critical interventions, remote location, night-time, and paediatric patients were associated with longer on-scene times. Conclusions: Compared to adult patients, the adjusted on-scene time for paediatric patients was longer. Besides the strong impact of a helicopter hoist operation on on-scene time, the dominant factors contributing to on-scene time are the type and number of interventions and monitoring: improving individual interventions or performing them in parallel may offer great potential for reducing on-scene time. However, multiple clinical interventions and monitoring interact and are not single interventions. Compared to the impact of interventions, non-modifiable factors, such as NACA score, type of diagnosis and age, make only a minor contribution to overall on-scene time.

Patients undergoing cardiac surgery are at increased cardiovascular risk, which includes altered lipid status. However, data on the effect of cardiac surgery and cardiopulmonary bypass (CPB) on plasma levels of key lipids are scarce. We investigated potential effects of CPB on plasma lipid levels and associations with early postoperative clinical outcomes. This is a prospective bio-bank study of patients undergoing elective cardiac surgery at our center January to December 2019. The follow-up period was 1 year after surgery. Blood sampling was performed before induction of general anesthesia, upon weaning from cardiopulmonary bypass (CPB), and on the first day after surgery. Clinical end points included the incidence of postoperative stroke, myocardial infarction, and death of any cause at 30 days after surgery as well as 1-year all-cause mortality. A total of 192 cardiac surgery patients (75% male, median age 67.0 years (interquartile range 60.0–73.0), median BMI 26.1 kg/m² (23.7–30.4)) were included. A significant intraoperative decrease in plasma levels compared with preoperative levels (all p < 0.0001) was observed for total cholesterol (TC) (Cliff’s delta d: 0.75 (0.68–0.82; 95% CI)), LDL-Cholesterol (LDL-C) (d: 0.66 (0.57–0.73)) and HDL-Cholesterol (HDL-C) (d: 0.72 (0.64–0.79)). At 24h after surgery, the plasma levels of LDL-C (d: 0.73 (0.650.79)) and TC (d: 0.77 (0.69–0.82)) continued to decrease compared to preoperative levels, while the plasma levels of HDL-C (d: 0.46 (0.36–0.55)) and TG (d: 0.40 (0.29–0.50)) rebounded, but all remained below the preoperative levels (p < 0.001). Mortality at 30 days was 1.0% (N = 2/192), and 1-year mortality was 3.8% (N = 7/186). Postoperative myocardial infarction occurred in 3.1% of patients (N = 6/192) and postoperative stroke in 5.8% (N = 11/190). Adjusting for age, sex, BMI, and statin therapy, we noted a protective effect of postoperative occurrence of stroke for pre-to-post-operative changes in TC (adjusted odds ratio (OR) 0.29 (0.07–0.90), p = 0.047), in LDL-C (aOR 0.19 (0.03–0.88), p = 0.045), and in HDL-C (aOR 0.01 (0.00–0.78), p = 0.039). No associations were observed between lipid levels and 1-year mortality. In conclusion, cardiac surgery induces a significant sudden drop in levels of key plasma lipids. This effect was pronounced during the operation, and levels remained significantly lowered at 24 h after surgery. The intraoperative drops in LDL-C, TC, and HDL-C were associated with a protective effect against occurrence of postoperative stroke in adjusted models. We demonstrate that the changes in key plasma lipid levels during surgery are strongly correlated, which makes attributing the impact of each lipid to the clinical end points, such as postoperative stroke, a challenging task. Large-scale analyses should investigate additional clinical outcome measures.