Determinants of the calibration of SAPS II and SAPS 3 mortality scores in intensive care: a European multicenter study

The association between the change in severity score from baseline and the outcomes of critically ill patients was enhanced by integration of bioimpedance analysis parameters

The study of the outcomes of critically ill patients has been a hard stuff in the field of intensive care. To explore the relationship between changes of severity scores, bioelectrical impedance analysis (BIA) and outcomes of critically ill patients, we enrolled patients (n = 206) admitted to intensive care unit (ICU) in Jinling Hospital from 2018 to 2021 with records of BIA on the days 1- and 3- ICU. Collected BIA and clinical data including simplified acute physiology score II (SAPS II) and sequential organ failure assessment. According to the baseline and change of severity scores or phase angle (PA) values, the patients were divided into: G-G, baseline good status, 3rd day unchanged; G-B, baseline good status, 3rd day deteriorated; B-G, baseline bad status, 3rd day improved; and B-B, baseline bad status, 3rd day unchanged. According to PA, the mortality of group G-G was 8.6%, and it was greater than 50% in group B-B for severity scores. The new score combining PA and severity scores established. Multivariate logistic regression analysis revealed that PA-SAPS II score was the only independent factor for 90-day mortality (P < 0.05). A linear correlation was found between mortality and PA-SAPS II score (prediction equation: Y ( % ) = 16.97 × X - 9.67 , R2 = 0.96, P < 0.05).

Development and validation of a nomogram for predicting in-hospital mortality of patients with cervical spine fractures without spinal cord injury

BACKGROUND

The incidence of cervical spine fractures is increasing every day, causing a huge burden on society. This study aimed to develop and verify a nomogram to predict the in-hospital mortality of patients with cervical spine fractures without spinal cord injury. This could help clinicians understand the clinical outcome of such patients at an early stage and make appropriate decisions to improve their prognosis.

METHODS

This study included 394 patients with cervical spine fractures from the Medical Information Mart for Intensive Care III database, and 40 clinical indicators of each patient on the first day of admission to the intensive care unit were collected. The independent risk factors were screened using the Least Absolute Shrinkage and Selection Operator regression analysis method, a multi-factor logistic regression model was established, nomograms were developed, and internal validation was performed. A receiver operating characteristic (ROC) curve was drawn, and the area under the ROC curve (AUC), net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were calculated to evaluate the discrimination of the model. Moreover, the consistency between the actual probability and predicted probability was reflected using the calibration curve and Hosmer-Lemeshow (HL) test. A decision curve analysis (DCA) was performed, and the nomogram was compared with the scoring system commonly used in clinical practice to evaluate the clinical net benefit.

RESULTS

The nomogram indicators included the systolic blood pressure, oxygen saturation, respiratory rate, bicarbonate, and simplified acute physiology score (SAPS) II. The results showed that our model had satisfactory predictive ability, with an AUC of 0.907 (95% confidence interval [CI] = 0.853-0.961) and 0.856 (95% CI = 0.746-0.967) in the training set and validation set, respectively. Compared with the SAPS-II system, the NRI values of the training and validation sets of our model were 0.543 (95% CI = 0.147-0.940) and 0.784 (95% CI = 0.282-1.286), respectively. The IDI values of the training and validation sets were 0.064 (95% CI = 0.004-0.123; P = 0.037) and 0.103 (95% CI = 0.002-0.203; P = 0.046), respectively. The calibration plot and HL test results confirmed that our model prediction results showed good agreement with the actual results, where the HL test values of the training and validation sets were P = 0.8 and P = 0.95, respectively. The DCA curve revealed that our model had better clinical net benefit than the SAPS-II system.

CONCLUSION

We explored the in-hospital mortality of patients with cervical spine fractures without spinal cord injury and constructed a nomogram to predict their prognosis. This could help doctors assess the patient's status and implement interventions to improve prognosis accordingly.

Development and validation of a nomogram for predicting in-hospital mortality in patients with nonhip femoral fractures

BACKGROUND

The incidence of nonhip femoral fractures is gradually increasing, but few studies have explored the risk factors for in-hospital death in patients with nonhip femoral fractures in the ICU or developed mortality prediction models. Therefore, we chose to study this specific patient group, hoping to help clinicians improve the prognosis of patients.

METHODS

This is a retrospective study based on the data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Least absolute shrinkage and selection operator (LASSO) regression was used to screen risk factors. The receiver operating characteristic (ROC) curve was drawn, and the areas under the curve (AUC), net reclassification index (NRI) and integrated discrimination improvement (IDI) were calculated to evaluate the discrimination of the model. The consistency between the actual probability and the predicted probability was assessed by the calibration curve and Hosmer-Lemeshow goodness of fit test (HL test). Decision curve analysis (DCA) was performed, and the nomogram was compared with the scoring system commonly used in clinical practice to evaluate the clinical net benefit.

RESULTS

The LASSO regression analysis showed that heart rate, temperature, red blood cell distribution width, blood urea nitrogen, Glasgow Coma Scale (GCS), Simplified Acute Physiology Score II (SAPSII), Charlson comorbidity index and cerebrovascular disease were independent risk factors for in-hospital death in patients with nonhip femoral fractures. The AUC, IDI and NRI of our model in the training set and validation set were better than those of the GCS and SAPSII scoring systems. The calibration curve and HL test results showed that our model prediction results were in good agreement with the actual results (P = 0.833 for the HL test of the training set and P = 0.767 for the HL test of the validation set). DCA showed that our model had a better clinical net benefit than the GCS and SAPSII scoring systems.

CONCLUSION

In this study, the independent risk factors for in-hospital death in patients with nonhip femoral fractures were determined, and a prediction model was constructed. The results of this study may help to improve the clinical prognosis of patients with nonhip femoral fractures.

The impact of the new acute respiratory distress syndrome (ARDS) criteria on Berlin criteria ARDS patients: a multicenter cohort study

OBJECTIVE

The European Society of Intensive Care Medicine (ESICM) recently recommended changes to the criteria of acute respiratory distress syndrome (ARDS), patients with high-flow oxygen were included, however, the effect of these changes remains unclear. Our objectives were to evaluate the performance of these new criteria and to compare the outcomes of patients meeting the new ARDS criteria with those meeting the Berlin ARDS criteria.

METHODS

This was a retrospective cohort. The patients admitted to the intensive care unit (ICU) were diagnosed with ARDS. Patients were classified as meeting Berlin criteria ARDS (n = 4279), high-flow nasal oxygen (HFNO) criteria ARDS (n = 559), or new criteria ARDS (n = 4838).

RESULTS

In comparison with HFNO criteria ARDS and new criteria ARDS, patients with Berlin criteria ARDS demonstrated lower blood oxygen levels assessed by PaO2/FiO2, SpO2/FiO2, and ROX (SpO2/FiO2/respiratory rate) (p < 0.001); and higher severity of illness assessed by the Sequential Organ Failure Assessment (SOFA) score, Acute Physiology And Chronic Health Evaluations (APACHE II), Simplified Acute Physiology Score (SAPS II) (p < 0.001), (p < 0.001), and longer ICU and hospital stays (p < 0.001). In comparison with the HFNO criteria, patients meeting Berlin criteria ARDS had higher hospital mortality (10.6% vs. 16.9%; p = 0.0082), 28-day mortality (10.6% vs. 16.5%; p = 0.0079), and 90-day mortality (10.7% vs. 17.1%; p = 0.0083). ARDS patients with HFNO did not have severe ARDS; Berlin criteria ARDS patients with severe ARDS had the highest mortality rate (approximately 33%). PaO2/FiO2, SpO2/FiO2, and ROX negatively correlated with the SOFA and APACHE II scores. The SOFA and APACHE II scores had high specificity and sensitivity for prognosis in patients with new criteria ARDS.

CONCLUSION

The new criteria of ARDS reduced the severity of illness, length of stay in the ICU, length of hospital stays, and overall mortality. SOFA and APACHE II scores remain important in assessing the prognosis of patients with new criteria ARDS.

TRIAL REGISTRATION

Registration number: ChiCTR2200067084.

Development and Validation of Machine Learning-Based Models to Predict In-Hospital Mortality in Life-Threatening Ventricular Arrhythmias: Retrospective Cohort Study

BACKGROUND

Life-threatening ventricular arrhythmias (LTVAs) are main causes of sudden cardiac arrest and are highly associated with an increased risk of mortality. A prediction model that enables early identification of the high-risk individuals is still lacking.

OBJECTIVE

We aimed to build machine learning (ML)-based models to predict in-hospital mortality in patients with LTVA.

METHODS

A total of 3140 patients with LTVA were randomly divided into training (n=2512, 80%) and internal validation (n=628, 20%) sets. Moreover, data of 2851 patients from another database were collected as the external validation set. The primary output was the probability of in-hospital mortality. The discriminatory ability was evaluated by the area under the receiver operating characteristic curve (AUC). The prediction performances of 5 ML algorithms were compared with 2 conventional scoring systems, namely, the simplified acute physiology score (SAPS-II) and the logistic organ dysfunction system (LODS).

RESULTS

The prediction performance of the 5 ML algorithms significantly outperformed the traditional models in predicting in-hospital mortality. CatBoost showed the highest AUC of 90.5% (95% CI 87.5%-93.5%), followed by LightGBM with an AUC of 90.1% (95% CI 86.8%-93.4%). Conversely, the predictive values of SAPS-II and LODS were unsatisfactory, with AUCs of 78.0% (95% CI 71.7%-84.3%) and 74.9% (95% CI 67.2%-82.6%), respectively. The superiority of ML-based models was also shown in the external validation set.

CONCLUSIONS

ML-based models could improve the predictive values of in-hospital mortality prediction for patients with LTVA compared with traditional scoring systems.

Atrial fibrillation is not an independent determinant of 28-day mortality among critically III sepsis patients

This study was conducted to investigate the relationship between atrial fibrillation and the clinical prognosis of patients with sepsis in intensive care unit. A total of 21,538 sepsis patients were enrolled in the study based on the Medical Information Mart for Intensive Care IV database, of whom 6,759 had AF. Propensity score matching was used to compare the clinical characteristics and outcomes of patients with and without AF. Besides, the inverse probability of treatment weighting, univariate and multivariate Cox regression analyzes were performed. Of the 21,538 patients, 31.4% had AF. The prevalence of AF increased in a step-by-step manner with growing age. Patients with AF were older than those without AF. After PSM, 11,180 patients remained, comprising 5,790 matched pairs in both groups. In IPTW, AF was not associated with 28-day mortality [hazard ratio (HR), 1.07; 95% confidence interval (CI), 0.99-1.15]. In Kaplan-Meier analysis, it was not observed difference of 28-day mortality between patients with and without AF. AF could be associated with increased ICU LOS, hospital LOS and need for mechanical ventilation; however, it does not remain an independent short-term predictor of 28-day mortality among patients with sepsis after PSM with IPTW and multivariate analysis.

Critically ill patient mortality by age: long-term follow-up (CIMbA-LT)

BACKGROUND

The past years have witnessed dramatic changes in the population admitted to the intensive care unit (ICU). Older and sicker patients are now commonly treated in this setting due to the newly available sophisticated life support. However, the short- and long-term benefit of this strategy is scarcely studied.

METHODS

The Critically Ill patients' mortality by age: Long-Term follow-up (CIMbA-LT) was a multicentric, nationwide, retrospective, observational study addressing short- and long-term prognosis of patients admitted to Portuguese multipurpose ICUs, during 4 years, according to their age and disease severity. Patients were followed for two years after ICU admission. The standardized hospital mortality ratio (SMR) was calculated according to the Simplified Acute Physiology Score (SAPS) II and the follow-up risk, for patients discharged alive from the hospital, according to official demographic national data for age and gender. Survival curves were plotted according to age group.

RESULTS

We included 37.118 patients, including 15.8% over 80 years old. The mean SAPS II score was 42.8 ± 19.4. The ICU all-cause mortality was 16.1% and 76% of all patients survive until hospital discharge. The SAPS II score overestimated hospital mortality [SMR at hospital discharge 0.7; 95% confidence interval (CI) 0.63-0.76] but accurately predicted one-year all-cause mortality [1-year SMR 1.01; (95% CI 0.98-1.08)]. Survival curves showed a peak in mortality, during the first 30 days, followed by a much slower survival decline thereafter. Older patients had higher short- and long-term mortality and their hospital SMR was also slightly higher (0.76 vs. 0.69). Patients discharged alive from the hospital had a 1-year relative mortality risk of 6.3; [95% CI 5.8-6.7]. This increased risk was higher for younger patients [21.1; (95% CI 15.1-39.6) vs. 2.4; (95% CI 2.2-2.7) for older patients].

CONCLUSIONS

Critically ill patients' mortality peaked in the first 30 days after ICU admission. Older critically ill patients had higher all-cause mortality, including a higher hospital SMR. A long-term increased relative mortality risk was noted in patients discharged alive from the hospital, but this was more noticeable in younger patients.

A Prediction Nomogram Combining Epworth Sleepiness Scale and Other Clinical Parameters to Predict Obstructive Sleep Apnea in Patients with Hypertension

Background

Obstructive sleep apnea (OSA) is common in patients with hypertension. Nonetheless, OSA is underdiagnosed despite considerable evidence of the association between OSA and adverse health outcomes. This study developed and validated a clinical nomogram to predict OSA in patients with hypertension based on the Epworth Sleepiness Scale (ESS) score and OSA-related parameters.

Methods

A total of 347 hypertensive patients with suspected OSA were retrospectively enrolled and randomly assigned to a training set and a validation set at 70 : 30 (N = 242/N = 105) ratio. OSA was diagnosed through sleep monitoring and was defined as an apnea-hypopnea index ≥5 events/h. Using the least absolute shrinkage and selection operator regression model, we identified potential predictors of OSA and constructed a nomogram model in the training set. The predictive performance of the nomogram was assessed and validated by discrimination and calibration. The nomogram was also compared with ESS scores according to decision curve analysis (DCA), integrated discrimination index (IDI), and net reclassification index (NRI).

Results

ESS scores, body mass index, neck circumference, snoring, and observed apnea predicted OSA are considered. The nomogram showed similar discrimination between the training set (AUC: 0.799, 95% CI: 0.743–0.847) and validation set (AUC: 0.766, 95% CI: 0.673–0.843) and good calibration in the training (P=0.925 > 0.05) and validation (P=0.906 > 0.05) sets. Compared with the predictive value of the ESS, the nomogram was clinically useful and significantly improved reclassification accuracy (NRI: 0.552, 95% CI: 0.282–0.822, P < 0.001; IDI: 0.088, 95% CI: 0.045–0.133, P < 0.001) at a probability threshold of >42%.

Conclusions

We developed a novel OSA prediction nomogram based on ESS scores and OSA-related parameters. This nomogram may help improve clinical decision-making, especially in communities and primary clinics, where polysomnography is unavailable.

SCD14-ST and New Generation Inflammatory Biomarkers in the Prediction of COVID-19 Outcome

Since no definitive cure for COVID-19 is available so far, one of the challenges against the disease is understanding the clinical features and the laboratory inflammatory markers that can differentiate among different severity grades of the disease. The aim of the present study is a comprehensive and longitudinal evaluation of SCD14-ST and other new inflammatory markers, as well as cytokine storm molecules and current inflammatory parameters, in order to define a panel of biomarkers that could be useful for a better prognostic prediction of COVID-19 mortality. SCD14-ST, as well as the inflammatory markers IL-6, IL-10, SuPAR and sRAGE, were measured in plasma-EDTA of ICU COVID-19 positive patients. In this longitudinal study, SCD14-ST resulted significantly higher in patients who eventually died compared to those who were discharged from the ICU. The results suggest that the new infection biomarker SCD14-ST, in addition to new generation inflammatory biomarkers, such as SuPAR, sRAGE and the cytokines IL-6 and IL-10, can be a useful prognostic tool associated with canonical inflammatory parameters, such as CRP, to predict SARS-CoV-2 outcome in ICU patients.

Simplified Acute Physiology Score 3 Performance in Austrian COVID-19 Patients Admitted to Intensive Care Units with and without Diabetes

This study evaluated and compared the performance of simplified acute physiology score 3 (SAPS 3) for predicting in-hospital mortality in COVID-19 patients admitted to intensive care units (ICUs) with and without diabetes in Austria. The Austrian national public health institute (GÖG) data of COVID-19 patients admitted to ICUs (n = 5850) were analyzed. Three versions of SAPS 3 were used: standard equation, Central European equation, and Austrian equation customized for COVID-19 patients. The observed in-hospital mortality was 38.9%, 42.9%, and 37.3% in all, diabetes, and non-diabetes patients, respectively. The overall C-statistics was 0.69 with an insignificant (p = 0.193) difference between diabetes (0.70) and non-diabetes (0.68) patients. The Brier score was > 0.20 for all SAPS 3 equations in all cohorts. Calibration was unsatisfactory for both standard and Central European equations in all cohorts, whereas it was satisfactory for the Austrian equation in diabetes patients only. The SAPS 3 score demonstrated low discrimination and accuracy in Austrian COVID-19 patients, with an insignificant difference between diabetes and non-diabetes. All equations were miscalibrated particularly in non-diabetes patients, while the Austrian equation showed satisfactory calibration in diabetes patients only. Both uncalibrated and calibrated versions of SAPS 3 should be used with caution in COVID-19 patients.

Developing machine learning models for prediction of mortality in the medical intensive care unit

BACKGROUND AND OBJECTIVE

Alert of patient deterioration is essential for prompt medical intervention in the Medical Intensive Care Unit (MICU). Logistic Regression (LR) has been used for the development of most conventional severity-of-illness scoring systems to anticipate the risk of mortality in the MICU. Machine Learning (ML) models such as probabilistic graphical models and Extreme Gradient Boosting (XGB) have demonstrated improved prediction accuracy in patient outcomes compared to LR. The aim was to compare three ML models to the SAPS, SAPS II, SAPS III, SOFA, serial SOFA, LODS, and OASIS for prediction of MICU mortality.

METHODS

A Bayesian Network (BN), Naïve Bayes network (NB), and a XGB model were developed. 9893 adult MICU-stays from the MIMIC-III database were studied. The primary outcome was MICU mortality prediction and the secondary outcome was 1-year mortality prediction. Data analyzed consisted on routine physiological measurements collected during 5 hours in the MICU, demographic and diagnoses/procedure features. The performance was evaluated by accuracy statistics, discrimination and calibration measures. Limitations of the study were discussed.

RESULTS

The AUROC for MICU mortality prediction was 0.919 for XGB, 0.905 for BN, and 0.864 for NB, while the conventional systems displayed much lower values with the serial SOFA having the best value (0.814). The Diagnostic Odds Ratio was ≤7.099 for all the conventional systems, reaching values of 30.115 for XGB and 22.648 for BN. The XGB achieved a sensitivity of 0.831 and specificity of 0.86 assuring an acceptable precision (0.528), whose values were much lower for the conventional systems. The Brier score was better for the ML models, except for the NB (0.119), with 0.072 for XGB and 0.081 for BN.

CONCLUSIONS

The XGB and BN substantially outperformed the conventional systems for discrimination, calibration and the accuracy statistics assessed. The NB showed inferior performance to the XGB and BN but improved the discrimination and all accuracy statistics of the conventional systems except for an inferior calibration and 1-year mortality discrimination. The XGB showed the best performance among all models. These ML models have the potential to improve the monitoring of MICU patients, which must be evaluated in future studies.

Effectiveness of LODS, OASIS, and SAPS II to predict in-hospital mortality for intensive care patients with ST elevation myocardial infarction

The relationship between three scoring systems (LODS, OASIS, and SAPS II) and in-hospital mortality of intensive care patients with ST segment elevation myocardial infarction (STEMI) is currently inconclusive. The baseline data, LODS score, OASIS score, SAPS II score, and in-hospital prognosis of intensive care patients with STEMI were retrieved from the Medical Information Mart for Intensive Care IV database. Propensity score matching analysis was performed to reduce bias. Receiver operating characteristic curves (ROC) were drawn for the three scoring systems, and comparisons between the areas under the ROC curves (AUC) were conducted. Decision curve analysis (DCA) was performed to determine the net benefits of the three scoring systems. LODS and SAPS II were independent risk factors for in-hospital mortality. For the study cohort, the AUCs of LODS, OASIS, SAPS II were 0.867, 0.827, and 0.894; after PSM, the AUCs of LODS, OASIS, SAPS II were 0.877, 0.821, and 0.881. A stratified analysis of the patients who underwent percutaneous coronary intervention/coronary artery bypass grafting (PCI/CABG) or not was conducted. In the PCI/CABG group, the AUCs of LODS, OASIS, SAPS II were 0.853, 0.825, and 0.867, while in the non-PCI/CABG group, the AUCs of LODS, OASIS, SAPS II were 0.857, 0.804, and 0.897. The results of the Z test suggest that the predictive value of LODS and SAPS II was not statistically different, but both were higher than OASIS. According to the DCA, the net clinical benefit of LODS was the greatest. LODS and SAPS II have excellent predictive value, and in most cases, both were higher than OASIS. With a more concise composition and greater clinical benefit, LODS may be a better predictor of in-hospital mortality for intensive care patients with STEMI.

Artificial intelligence forecasting mortality at an intensive care unit and comparison to a logistic regression system

Objective

To explore an artificial intelligence approach based on gradient-boosted decision trees for prediction of all-cause mortality at an intensive care unit, comparing its performance to a recent logistic regression system in the literature, and a logistic regression model built on the same platform.

Methods

A gradient-boosted decision trees model and a logistic regression model were trained and tested with the Medical Information Mart for Intensive Care database. The 1-hour resolution physiological measurements of adult patients, collected during 5 hours in the intensive care unit, consisted of eight routine clinical parameters. The study addressed how the models learn to categorize patients to predict intensive care unit mortality or survival within 12 hours. The performance was evaluated with accuracy statistics and the area under the Receiver Operating Characteristic curve.

Results

The gradient-boosted trees yielded an area under the Receiver Operating Characteristic curve of 0.89, compared to 0.806 for the logistic regression. The accuracy was 0.814 for the gradient-boosted trees, compared to 0.782 for the logistic regression. The diagnostic odds ratio was 17.823 for the gradient-boosted trees, compared to 9.254 for the logistic regression. The Cohen’s kappa, F-measure, Matthews correlation coefficient, and markedness were higher for the gradient-boosted trees.

Conclusion

The discriminatory power of the gradient-boosted trees was excellent. The gradient-boosted trees outperformed the logistic regression regarding intensive care unit mortality prediction. The high diagnostic odds ratio and markedness values for the gradient-boosted trees are important in the context of the studied unbalanced dataset.

Preoperative and ICU Scoring Models for Predicting the In-Hospital Mortality of Patients With Ruptured Abdominal Aortic Aneurysms

OBJECTIVES

This study's objective was to compare several preoperative and intensive care unit (ICU) prognostic scoring systems for predicting the in-hospital mortality of ruptured abdominal aortic aneurysms (RAAAs).

DESIGN

Retrospective cohort study.

SETTING

Single tertiary university center.

PARTICIPANTS

The study comprised 157 patients.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

A total of 157 patients (82% male) presented with RAAA at Charité University Hospital from January 2011 to December 2020. The mean age was 74 years (standard deviation ten years). In-hospital mortality was 29% (n = 45), of whom nine patients (6%) died en route to the operating room, 13 (8%) on the operating table, and 23 (15%) in the ICU. A total of 135 patients (86%) were admitted to the ICU. All six models demonstrated good discriminating performance between survivors and nonsurvivors. Overall, the area under the curve (AUC) for RAAA preoperative scores was greater than those for ICU scores. The largest AUC was achieved with the Vascular Study Group of New England (VSGNE) RAAA risk score (AUC = 0.87 for all patients, AUC = 0.84 for patients admitted to the ICU), followed by Hardman Index (AUC = 0.83 for all patients, AUC = 0.81 for patients admitted to the ICU), and Glasgow Aneurysm Score (AUC = 0.74 for all patients, AUC = 0.83 for patients admitted to the ICU). The largest AUC for ICU scores (only patients admitted to the ICU) was achieved with Simplified Acute Physiology Score II (0.75), followed by Sepsis-related Organ Failure Assessment (0.73), and Acute Physiology and Chronic Health Evaluation II (0.71).

CONCLUSIONS

Preoperative and ICU scores can predict the mortality of patients presenting with RAAA. In addition, the discriminatory ability of preoperative scores between survivors and nonsurvivors was larger than that for ICU scores.

Can the cytokine adsorber CytoSorb® help to mitigate cytokine storm and reduce mortality in critically ill patients? A propensity score matching analysis

Background

A cytokine storm is life threatening for critically ill patients and is mainly caused by sepsis or severe trauma. In combination with supportive therapy, the cytokine adsorber Cytosorb^® (CS) is increasingly used for the treatment of cytokine storm. However, it is questionable whether its use is actually beneficial in these patients.

Methods

Patients with an interleukin-6 (IL-6) > 10,000 pg/ml were retrospectively included between October 2014 and May 2020 and were divided into two groups (group 1: CS therapy; group 2: no CS therapy). Inclusion criteria were a regularly measured IL-6 and, for patients allocated to group 1, CS therapy for at least 90 min. A propensity score (PS) matching analysis with significant baseline differences as predictors (Simplified Acute Physiology Score (SAPS) II, extracorporeal membrane oxygenation, renal replacement therapy, IL-6, lactate and norepinephrine demand) was performed to compare both groups (adjustment tolerance: < 0.05; standardization tolerance: < 10%). U-test and Fisher’s-test were used for independent variables and the Wilcoxon test was used for dependent variables.

Results

In total, 143 patients were included in the initial evaluation (group 1: 38; group 2: 105). Nineteen comparable pairings could be formed (mean initial IL-6: 58,385 vs. 59,812 pg/ml; mean SAPS II: 77 vs. 75). There was a significant reduction in IL-6 in patients with (p < 0.001) and without CS treatment (p = 0.005). However, there was no significant difference (p = 0.708) in the median relative reduction in both groups (89% vs. 80%). Furthermore, there was no significant difference in the relative change in C-reactive protein, lactate, or norepinephrine demand in either group and the in-hospital mortality was similar between groups (73.7%).

Conclusion

Our study showed no difference in IL-6 reduction, hemodynamic stabilization, or mortality in patients with Cytosorb^® treatment compared to a matched patient population.

Outcome in patients with open abdomen treatment for peritonitis: a multidomain approach outperforms single domain predictions

Numerous patient-related clinical parameters and treatment-specific variables have been identified as causing or contributing to the severity of peritonitis. We postulated that a combination of clinical and surgical markers and scoring systems would outperform each of these predictors in isolation. To investigate this hypothesis, we developed a multivariable model to examine whether survival outcome can reliably be predicted in peritonitis patients treated with open abdomen. This single-center retrospective analysis used univariable and multivariable logistic regression modeling in combination with repeated random sub-sampling validation to examine the predictive capabilities of domain-specific predictors (i.e., demography, physiology, surgery). We analyzed data of 1,351 consecutive adult patients (55.7% male) who underwent open abdominal surgery in the study period (January 1998 to December 2018). Core variables included demographics, clinical scores, surgical indices and indicators of organ dysfunction, peritonitis index, incision type, fascia closure, wound healing, and fascial dehiscence. Postoperative complications were also added when available. A multidomain peritonitis prediction model (MPPM) was constructed to bridge the mortality predictions from individual domains (demographic, physiological and surgical). The MPPM is based on data of n = 597 patients, features high predictive capabilities (area under the receiver operating curve: 0.87 (0.85 to 0.90, 95% CI)) and is well calibrated. The surgical predictor “skin closure” was found to be the most important predictor of survival in our cohort, closely followed by the two physiological predictors SAPS-II and MPI. Marginal effects plots highlight the effect of individual outcomes on the prediction of survival outcome in patients undergoing staged laparotomies for treatment of peritonitis. Although most single indices exhibited moderate performance, we observed that the predictive performance was markedly increased when an integrative prediction model was applied. Our proposed MPPM integrative prediction model may outperform the predictive power of current models.

Machine learning applied to a Cardiac Surgery Recovery Unit and to a Coronary Care Unit for mortality prediction

Most established severity-of-illness systems used for prediction of intensive care unit (ICU) mortality were developed targeted at the general ICU population, based on logistic regression (LR). To date, no dynamic predictive tool for ICU mortality has been developed targeted at the Cardiac Surgery Recovery Unit (CSRU) and Coronary Care Unit (CCU) using machine learning (ML). CSRU and CCU adult patients from the MIMIC-III critical care database were studied. The ML methods developed extract ICU data during a 5-h window and demographic features to produce mortality predictions and were compared to six established severity-of-illness systems and LR. In a secondary experiment, additional procedure/surgery and ICU features were added to the models. The ML models developed were the Tree Ensemble (TE), Random Forest, XGBoost Tree Ensemble (XGB), Naive Bayes (NB), and Bayesian network. The discrimination, calibration and accuracy statistics were assessed. The AUROC values were superior for the ML models reaching 0.926 and 0.924 for the XGB, and 0.904 and 0.908 for the TE for ICU mortality prediction in the primary and secondary experiments respectively. Among the conventional systems, the serial SOFA obtained the highest AUROC (0.8405). The Brier score was better for the ML models except the NB over the conventional systems. The accuracy statistics less sensitive to unbalanced cohorts were higher for all the ML models. In conclusion, the predictive power of XGB was excellent, substantially outperforming the conventional systems and LR. The ML models developed in this work offer promising results that could benefit CSRU and CCU.

Barriers to Access to Treatment for Hypertensive Patients in Primary Health Care of Less Developed Northwest China: A Predictive Nomogram

Background

This study aims to evaluate the risk factors associated with untreated hypertension and develop and internally validate untreated risk nomograms in patients with hypertension among primary health care of less developed Northwest China.

Methods

A total of 895 eligible patients with hypertension in primary health care of less developed Northwest China were divided into a training set (n = 626) and a validation set (n = 269). Untreated hypertension was defined as not taking antihypertensive medication during the past two weeks. Using least absolute shrinkage and selection operator (LASSO) regression model, we identified the optimized risk factors of nontreatment, followed by establishment of a prediction nomogram. The discriminative ability, calibration, and clinical usefulness were determined using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision analysis. The results were assessed by internal validation in the validation set.

Results

Five independent risk factors were derived from LASSO regression model and entered into the nomogram: age, herdsman, family income per member, altitude of habitation, and comorbidity. The nomogram displayed a robust discrimination with an AUC of 0.859 (95% confidence interval: 0.812–0.906) and good calibration. The nomogram was clinically useful when the intervention was decided at the untreated possibility threshold of 7% to 91% in the decision curve analysis. Results were confirmed by internal validation.

Conclusions

Our nomogram showed favorable predictive accuracy for untreated hypertension in primary health care of less developed Northwest China and might help primary health care assess the risk of nontreatment in patients with hypertension.

Performance of intensive care unit severity scoring systems across different ethnicities in the USA: a retrospective observational study

BACKGROUND

Despite wide use of severity scoring systems for case-mix determination and benchmarking in the intensive care unit (ICU), the possibility of scoring bias across ethnicities has not been examined. Guidelines on the use of illness severity scores to inform triage decisions for allocation of scarce resources, such as mechanical ventilation, during the current COVID-19 pandemic warrant examination for possible bias in these models. We investigated the performance of the severity scoring systems Acute Physiology and Chronic Health Evaluation IVa (APACHE IVa), Oxford Acute Severity of Illness Score (OASIS), and Sequential Organ Failure Assessment (SOFA) across four ethnicities in two large ICU databases to identify possible ethnicity-based bias.

METHODS

Data from the electronic ICU Collaborative Research Database (eICU-CRD) and the Medical Information Mart for Intensive Care III (MIMIC-III) database, built from patient episodes in the USA from 2014-15 and 2001-12, respectively, were analysed for score performance in Asian, Black, Hispanic, and White people after appropriate exclusions. Hospital mortality was the outcome of interest. Discrimination and calibration were determined for all three scoring systems in all four groups, using area under receiver operating characteristic (AUROC) curve for different ethnicities to assess discrimination, and standardised mortality ratio (SMR) or proxy measures to assess calibration.

FINDINGS

We analysed 166 751 participants (122 919 eICU-CRD and 43 832 MIMIC-III). Although measurements of discrimination were significantly different among the groups (AUROC ranging from 0·86 to 0·89 [p=0·016] with APACHE IVa and from 0·75 to 0·77 [p=0·85] with OASIS), they did not display any discernible systematic patterns of bias. However, measurements of calibration indicated persistent, and in some cases statistically significant, patterns of difference between Hispanic people (SMR 0·73 with APACHE IVa and 0·64 with OASIS) and Black people (0·67 and 0·68) versus Asian people (0·77 and 0·95) and White people (0·76 and 0·81). Although calibrations were imperfect for all groups, the scores consistently showed a pattern of overpredicting mortality for Black people and Hispanic people. Similar results were seen using SOFA scores across the two databases.

INTERPRETATION

The systematic differences in calibration across ethnicities suggest that illness severity scores reflect statistical bias in their predictions of mortality.

FUNDING

There was no specific funding for this study.

Association Between Preadmission Metformin Use and Outcomes in Intensive Care Unit Patients With Sepsis and Type 2 Diabetes: A Cohort Study

Background: Sepsis is a deadly disease worldwide. Effective treatment strategy of sepsis remains limited. There still was a controversial about association between preadmission metformin use and mortality in sepsis patients with diabetes. We aimed to assess sepsis-related mortality in patients with type 2 diabetes (T2DM) who were preadmission metformin and non-metformin users. Methods: The patients with sepsis and T2DM were included from Medical Information Mart for Intensive Care -III database. Outcome was 30-day mortality. We used multivariable Cox regression analyses to calculate adjusted hazard ratio (HR) with 95% CI. Results: We included 2,383 sepsis patients with T2DM (476 and 1,907 patients were preadmission metformin and non-metformin uses) between 2001 and 2012. The overall 30-day mortality was 20.1% (480/2,383); it was 21.9% (418/1,907), and 13.0% (62/476) for non-metformin and metformin users, respectively. After adjusted for potential confounders, we found that preadmission metformin use was associated with 39% lower of 30-day mortality (HR = 0.61, 95% CI: 0.46-0.81, p = 0.007). In sensitivity analyses, subgroups analyses, and propensity score matching, the results remain stable. Conclusions: Preadmission metformin use may be associated with reduced risk-adjusted mortality in patients with sepsis and T2DM. It is worthy to further investigate this association.

Prevalence and predictive nomogram of depression among hypertensive patients in primary care: A cross-sectional study in less developed Northwest China

Hypertensive patients commonly co-exist persistent depressive symptoms. However, these issues are not always identified, especially in primary health care, which may worsen the prognosis of hypertension. Therefore, the aim of this study was to determine the prevalence and risk factor of depression, and to develop risk nomogram of depression in hypertensive patients from primary health care Northwest China.We used a stratified multistage random sampling method to obtain 1856 hypertensives subjects aged ≥18 years in Xinjiang between April and October 2019. The subjects were randomly divided into a training set (n = 1299) and a validation set (n = 557). Depression was evaluated by Hospital Anxiety and Depression Scale (HADS), with a cut-off score ≥8. Using the least absolute shrinkage and selection operator (LASSO) regression model, we identified optimized risk factors of depression in the training set, followed by the establishment of prediction nomogram. The discriminative ability, calibration, and clinical usefulness of nomogram were assessed. The results were verified by internal validation in validation set.13.7% hypertensive subjects displayed depression. Seven independent risk factors of depression were identified and entered into the nomogram including age, region, ethnicity, marital status, physical activity, sleep quality, and control of hypertension. The nomogram displayed robust discrimination with an AUC of 0.760 [95% confidence interval (CI): 0.724-0.797)] and 0.761 (95%CI: 0.702-0.819), and good calibration in training set and validation set, respectively. The decision curve analysis and clinical impact curve demonstrated clinical usefulness of predictive nomogram.There is a considerable prevalence of depression in patients with hypertension from primary care of Xinjiang, Northwest China. Our nomogram may help primary care providers assess the risk of depression in patients with hypertension.

Performance of intensive care unit severity scoring systems across different ethnicities

BACKGROUND

Despite wide utilisation of severity scoring systems for case-mix determination and benchmarking in the intensive care unit, the possibility of scoring bias across ethnicities has not been examined. Recent guidelines on the use of illness severity scores to inform triage decisions for allocation of scarce resources such as mechanical ventilation during the current COVID-19 pandemic warrant examination for possible bias in these models. We investigated the performance of three severity scoring systems (APACHE IVa, OASIS, SOFA) across ethnic groups in two large ICU databases in order to identify possible ethnicity-based bias.

METHOD

Data from the eICU Collaborative Research Database and the Medical Information Mart for Intensive Care were analysed for score performance in Asians, African Americans, Hispanics and Whites after appropriate exclusions. Discrimination and calibration were determined for all three scoring systems in all four groups.

FINDINGS

While measurements of discrimination -area under the receiver operating characteristic curve (AUROC) -were significantly different among the groups, they did not display any discernible systematic patterns of bias. In contrast, measurements of calibration -standardised mortality ratio (SMR) -indicated persistent, and in some cases significant, patterns of difference between Hispanics and African Americans versus Asians and Whites. The differences between African Americans and Whites were consistently statistically significant. While calibrations were imperfect for all groups, the scores consistently demonstrated a pattern of over-predicting mortality for African Americans and Hispanics.

INTERPRETATION

The systematic differences in calibration across ethnic groups suggest that illness severity scores reflect bias in their predictions of mortality.

FUNDING

LAC is funded by the National Institute of Health through NIBIB R01 EB017205. There was no specific funding for this study.

Score performance of SAPS 2 and SAPS 3 in combination with biomarkers IL-6, PCT or CRP

OBJECTIVE

We aimed to evaluate the effects of combining the Simplified-Acute-Physiology-Score (SAPS) 2 or the SAPS 3 with Interleukin-6 (IL-6) or Procalcitonin (PCT) or C-Reactive Protein (CRP) concentrations for predicting in-hospital mortality.

MATERIAL AND METHODS

This retrospective study was conducted in an interdisciplinary 22-bed intensive care unit (ICU) at a German university hospital. Within an 18-month period, SAPS 2 and SAPS 3 were calculated for 514 critically ill patients that were admitted to the internal medicine department. To evaluate discrimination performance, the area under the receiver operating characteristic curves (AUROCs) and the 95% confidence intervals (95% CIs) were calculated for each score, exclusively or in combination with IL-6 or PCT or CRP. DeLong test was used to compare different AUROCs.

RESULTS

The SAPS 2 exhibited a better discrimination performance than SAPS 3 with AUROCs of 0.81 (95% CI, 0.76-0.86) and 0.72 (95% CI, 0.66-0.78), respectively. Overall, combination of the SAPS 2 with IL-6 showed the best discrimination performance (AUROC 0.82; 95% CI, 0.77-0.87), albeit not significantly different from SAPS2. IL-6 performed better than PCT and CRP with AUROCs of 0.75 (95% CI, 0.69-0.81), 0.72 (95% CI, 0.66-0.77) and 0.65 (95% CI, 0.59-0.72), respectively. Performance of the SAPS 3 improved significantly when combined with IL-6 (AUROC 0.76; 95% CI, 0.69-0.81) or PCT (AUROC 0.73; 95% CI, 0.67-0.78).

CONCLUSIONS

Our analysis provided evidence that the risk stratification performance of the SAPS 3 and, to a lesser degree, also of the SAPS 2 can increase when combined with IL-6. A more accurate detection of aberrant or dysregulated systemic immunological responses (by IL-6) may explain the higher performance achieved by SAPS 3 + IL-6 vs. SAPS 3. Thus, implementation of IL-6 in critical care scores can improve prediction outcomes, especially in patients experiencing acute inflammatory conditions; however, statistical results may vary across hospital types and/or patient populations with different case mix.

Risk factors and outcome of patients with periprosthetic joint infection admitted to intensive care unit

INTRODUCTION

Prosthetic joint infection (PJI) is a severe complication after total joint replacement surgery. The current study analyzes the outcome and mortality of patients admitted to an intensive care unit following PJI.

METHODS

Retrospective analysis of all patients treated between 2012 and 2016 due to PJI in the surgical intensive care unit of a university hospital.

RESULTS

A total of 124 patients were included. The mean age was 75 ± 11 years. Of those 124 patients, 85 patients (68.5%) suffered a prosthetic infection of the hip, 33 patients (27.3%) of the knee, six patients (4.8%) of hip and knee. 52 patients were male (40.9%). The assessed mean Simplified Acute Physiology Score II (SAPSII) was 29.6 ± 5.9. The mortality rate was 21% (26/124). Of surviving patients, 53.1% were discharged home, 25.5% were transferred to a nursing home, and 21.4% were transferred to a geriatric rehabilitation center. Comparing survivors to non-survivors, the non-survivor group showed a higher incidence of renal replacement therapy (46.1 vs 3.0%; p < 0.01), higher SAPSII on admission (35.7 vs. 29.0; p = 0.01) and higher Charlson Comorbidity Indices (CCI) (5.5 vs. 2.82; p < 0.01). The multivariate regression identified CCI (odds ratio 1.49; p < 0.01) and renal replacement therapy (odds ratio 12.4; p < 0.01) as independent risk factors for increased mortality.

CONCLUSIONS

Admission to an intensive care unit was associated with a mortality rate of 21%. Factors associated with poor outcomes included renal replacement therapy, higher admission SAPII scores, and higher admission Charlson comorbidity index. These factors could be used for individual risk assessment on admission to the ICU.

Mortality prediction models in the adult critically ill: A scoping review

BACKGROUND

Mortality prediction models are applied in the intensive care unit (ICU) to stratify patients into different risk categories and to facilitate benchmarking. To ensure that the correct prediction models are applied for these purposes, the best performing models must be identified. As a first step, we aimed to establish a systematic review of mortality prediction models in critically ill patients.

METHODS

Mortality prediction models were searched in four databases using the following criteria: developed for use in adult ICU patients in high-income countries, with mortality as primary or secondary outcome. Characteristics and performance measures of the models were summarized. Performance was presented in terms of discrimination, calibration and overall performance measures presented in the original publication.

RESULTS

In total, 43 mortality prediction models were included in the final analysis. In all, 15 models were only internally validated (35%), 13 externally (30%) and 10 (23%) were both internally and externally validated by the original researchers. Discrimination was assessed in 42 models (98%). Commonly used calibration measures were the Hosmer-Lemeshow test (60%) and the calibration plot (28%). Calibration was not assessed in 11 models (26%). Overall performance was assessed in the Brier score (19%) and the Nagelkerke's R² (4.7%).

CONCLUSIONS

Mortality prediction models have varying methodology, and validation and performance of individual models differ. External validation by the original researchers is often lacking and head-to-head comparisons are urgently needed to identify the best performing mortality prediction models for guiding clinical care and research in different settings and populations.

Development and Validation of a Nomogram to Predict Type 2 Diabetes Mellitus in Overweight and Obese Adults: A Prospective Cohort Study from 82938 Adults in China

BACKGROUND

The twin epidemic of overweight/obesity and type 2 diabetes mellitus (T2DM) is a major public health problem globally, especially in China. Overweight/obese adults commonly coexist with T2DM, which is closely related to adverse health outcomes. Therefore, this study aimed to develop risk nomogram of T2DM in Chinese adults with overweight/obesity.

METHODS

We used prospective cohort study data for 82938 individuals aged ≥20 years free of T2DM collected between 2010 and 2016 and divided them into a training (n = 58056) and a validation set (n = 24882). Using the least absolute shrinkage and selection operator (LASSO) regression model in training set, we identified optimized risk factors of T2DM, followed by the establishment of T2DM prediction nomogram. The discriminative ability, calibration, and clinical usefulness of nomogram were assessed. The results were assessed by internal validation in validation set.

RESULTS

Six independent risk factors of T2DM were identified and entered into the nomogram including age, body mass index, fasting plasma glucose, total cholesterol, triglycerides, and family history. The nomogram incorporating these six risk factors showed good discrimination regarding the training set, with a Harrell's concordance index (C-index) of 0.859 [95% confidence interval (CI): 0.850-0.868] and an area under the receiver operating characteristic curve of 0.862 (95% CI: 0.853-0.871). The calibration curves indicated well agreement between the probability as predicted by the nomogram and the actual probability. Decision curve analysis demonstrated that the prediction nomogram was clinically useful. The consistent of findings was confirmed using the validation set.

CONCLUSIONS

The nomogram showed accurate prediction for T2DM among Chinese population with overweight and obese and might aid in assessment risk of T2DM.

The predictive power of SAPS-3 and SOFA scores and their relations with patient outcomes in the Surgical Intensive Care Unit

Objectives

Individual risk of surgical patients is more often underestimated and there is not an absolute criterion demonstrating which patient deserves intensive care. Since a nominative assessment of these patients to quantify the intensity of critical illness is not appropriate, prognostic scores are used to assess the mortality rate and prognosis for critical patients including surgical ones. This study aimed to test the calibration power of SAPS-3 score and SOFA score of surgical patients undergoing gastrointestinal surgery, and identify any relation with patient outcomes in the department of surgical ICU.

Material and Methods

This retrospective observational study was conducted during the period between August 2017 and December 2017. It was performed at a Gastroenterological Surgical ICU, a tertiary care hospital in Ankara, Türkiye. To calculate SAPS-3 and SOFA score, physiological data and laboratory analysis on the day of ICU admission were used. Records were reviewed from hospitalization to medical discharge or hospital mortality. Statistical analysis included Mann Whitney U-test and ROC-curves to predict 30-day mortality.

Results

A total of 233 patients admitted to the Gastroenterological Surgical ICU were included into the study and the main reason for ICU admission was surgical problems. Mortality rate was 2.6 % (6 patients). Average SAPS -3 score was 32.5 and SOFA score was 30.1. A significant correlation was observed with the SAPS-3 score, but not with the SOFA score statistically in mortality as a dependent factor. The discriminative power, assessed using the AUC and the probability of death estimation, was satisfactory with the SAPS-3 scores (AUC 0.754) while it was lower with the SOFA score (AUC 0.631).

Conclusion

We found that SAPS-3 score was significantly correlated not only with mortality rate, but also with LOS in the ICU. Nonetheless, SOFA score was not related to mortality, but related to LOS in the ICU. Prognostic score systems are used to estimate mortality but they may be used to identify LOS in the ICU and postoperative complications. It can be concluded that SAPS-3 and SOFA scores may be used to prognosticate postoperative ICU requirement.

Development and internal validation of the Simplified Mortality Score for the Intensive Care Unit (SMS-ICU)

BACKGROUND

Intensive care unit (ICU) mortality prediction scores deteriorate over time, and their complexity decreases clinical applicability and commonly causes problems with missing data. We aimed to develop and internally validate a new and simple score that predicts 90-day mortality in adults upon acute admission to the ICU: the Simplified Mortality Score for the Intensive Care Unit (SMS-ICU).

METHODS

We used data from an international cohort of 2139 patients acutely admitted to the ICU and 1947 ICU patients with severe sepsis/septic shock from 2009 to 2016. We performed multiple imputations for missing data and used binary logistic regression analysis with variable selection by backward elimination, followed by conversion to a simple point-based score. We assessed the apparent performance and validated the score internally using bootstrapping to present optimism-corrected performance estimates.

RESULTS

The SMS-ICU comprises seven variables available in 99.5% of the patients: two numeric variables: age and lowest systolic blood pressure, and five dichotomous variables: haematologic malignancy/metastatic cancer, acute surgical admission and use of vasopressors/inotropes, respiratory support and renal replacement therapy. Discrimination (area under the receiver operating characteristic curve) was 0.72 (95% CI: 0.71-0.74), overall performance (Nagelkerke's R² ) was 0.19 and calibration (intercept and slope) was 0.00 and 0.99, respectively. Optimism-corrected performance was similar to apparent performance.

CONCLUSIONS

The SMS-ICU predicted 90-day mortality with reasonable and stable performance. If performance remains adequate after external validation, the SMS-ICU could prove a valuable tool for ICU clinicians and researchers because of its simplicity and expected very low number of missing values.