Model for predicting short-term mortality of severe sepsis

The Utility of C-Reactive Protein, Procalcitonin, and Leukocyte Values in Predicting the Prognosis of Patients with Pneumosepsis and Septic Shock

Background and Objectives: The predictive value of changes in C-reactive protein (CRP), procalcitonin, and leukocyte levels, which are commonly used in the diagnosis of infection in sepsis and septic shock, remains a topic of debate. The aim of this study was to evaluate the effectiveness of changes in CRP, procalcitonin, and leukocyte counts on the prognosis of 230 patients admitted to the intensive care unit (ICU) with the diagnosis of sepsis and pneumonia-related septic shock between 1 April 2022 and 31 December 2023, and to investigate whether any of these markers have a superior predictive value over the others in forecasting prognosis. Materials and Methods: This single-center, retrospective, cross-sectional observational study included patients who developed sepsis and septic shock due to community-acquired pneumonia and were admitted to the ICU. Demographic data, 1-month and 90-day mortality rates, length of stay in the ICU, discharge to the ward or an outside facility, need for dialysis after sepsis, need for invasive or noninvasive mechanical ventilation during the ICU stay and the duration of this support, whether patients admitted with sepsis or septic shock required inotropic agent support during their stay in the ICU and whether they received monotherapy or combination therapy with antibiotics during their admission to the ICU, the Comorbidity Index score (CCIS), CURB-65 score (confusion, uremia, respiratory rate, BP, age ≥ 65), and Acute Physiology and Chronic Health Evaluation II (APACHE-II) score were analyzed. Additionally, CRP, procalcitonin, and leukocyte levels were recorded, and univariate and multivariate logistic regression analyses were performed to evaluate their effects on 1- and 3-month mortality outcomes. In all statistical analyses, a p-value of <0.05 was accepted as a significant level. Results: According to multivariate logistic regression analysis, low BMI, male gender, and high CCIS, CURB-65, and APACHE-II scores were found to be significantly associated with both 1-month and 3-month mortality (p < 0.05). Although there was no significant relationship between the first-day levels of leukocytes, CRP, and PCT and mortality, their levels on the third day were observed to be at their highest in both the 1-month and 3-month mortality cases (p < 0.05). Additionally, a concurrent increase in any two or all three of CRP, PCT, and leukocyte values was found to be higher in patients with 3-month mortality compared with those who survived (p = 0.004). Conclusions: In patients with pneumoseptic or pneumonia-related septic shock, the persistent elevation and concurrent increase in PCT, CRP, and leukocyte values, along with male gender, advanced age, low BMI, and high CCIS, CURB-65, and APACHE-II scores, were found to be significantly associated with 3-month mortality.

Effect of ethanol exposure on innate immune response in sepsis

Alcohol use disorder, reported by 1 in 8 critically ill patients, is a risk factor for death in sepsis patients. Sepsis, the leading cause of death, kills over 270,000 patients in the United States alone and remains without targeted therapy. Immune response in sepsis transitions from an early hyperinflammation to persistent inflammation and immunosuppression and multiple organ dysfunction during late sepsis. Innate immunity is the first line of defense against pathogen invasion. Ethanol exposure is known to impair innate and adaptive immune response and bacterial clearance in sepsis patients. Specifically, ethanol exposure is known to modulate every aspect of innate immune response with and without sepsis. Multiple molecular mechanisms are implicated in causing dysregulated immune response in ethanol exposure with sepsis, but targeted treatments have remained elusive. In this article, we outline the effects of ethanol exposure on various innate immune cell types in general and during sepsis.

Generalization of a Deep Learning Model for Continuous Glucose Monitoring-Based Hypoglycemia Prediction: Algorithm Development and Validation Study

Background

Predicting hypoglycemia while maintaining a low false alarm rate is a challenge for the wide adoption of continuous glucose monitoring (CGM) devices in diabetes management. One small study suggested that a deep learning model based on the long short-term memory (LSTM) network had better performance in hypoglycemia prediction than traditional machine learning algorithms in European patients with type 1 diabetes. However, given that many well-recognized deep learning models perform poorly outside the training setting, it remains unclear whether the LSTM model could be generalized to different populations or patients with other diabetes subtypes.

Objective

The aim of this study was to validate LSTM hypoglycemia prediction models in more diverse populations and across a wide spectrum of patients with different subtypes of diabetes.

Methods

We assembled two large data sets of patients with type 1 and type 2 diabetes. The primary data set including CGM data from 192 Chinese patients with diabetes was used to develop the LSTM, support vector machine (SVM), and random forest (RF) models for hypoglycemia prediction with a prediction horizon of 30 minutes. Hypoglycemia was categorized into mild (glucose=54-70 mg/dL) and severe (glucose<54 mg/dL) levels. The validation data set of 427 patients of European-American ancestry in the United States was used to validate the models and examine their generalizations. The predictive performance of the models was evaluated according to the sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).

Results

For the difficult-to-predict mild hypoglycemia events, the LSTM model consistently achieved AUC values greater than 97% in the primary data set, with a less than 3% AUC reduction in the validation data set, indicating that the model was robust and generalizable across populations. AUC values above 93% were also achieved when the LSTM model was applied to both type 1 and type 2 diabetes in the validation data set, further strengthening the generalizability of the model. Under different satisfactory levels of sensitivity for mild and severe hypoglycemia prediction, the LSTM model achieved higher specificity than the SVM and RF models, thereby reducing false alarms.

Conclusions

Our results demonstrate that the LSTM model is robust for hypoglycemia prediction and is generalizable across populations or diabetes subtypes. Given its additional advantage of false-alarm reduction, the LSTM model is a strong candidate to be widely implemented in future CGM devices for hypoglycemia prediction.

External validation of the ARCH score in patients undergoing aortic arch reconstruction under circulatory arrest

BACKGROUND

Aortic arch surgery with hypothermic circulatory arrest (HCA) carries a higher risk of morbidity and mortality compared to routine cardiac surgical procedures. The newly developed ARCH (arch reconstruction under circulatory arrest with hypothermia) score has not been externally validated. We sought to externally validate this score in our local population.

METHODS

All consecutive open aortic arch surgeries with HCA performed between 2014 and 2023 were included. Univariable and multivariable analyses were performed. Model discrimination was assessed by the C-statistic with 95% confidence intervals as part of the receiver operating characteristic (ROC) curve analysis. Model performance was visualized by a calibration plot and quantified by the Brier score.

RESULTS

A total of 760 patients (38.3% females) were included. The mean age was 61 (±13.6) years, with 56.4% of patients' age >60 years. The procedures were carried out mostly emergently or urgently (59.6%). Total arch replacement was performed in 32.5% of the patients, and aortic root procedures were carried out in 74.6%. In-hospital death occurred in 64 patients (8.4%), and stroke occurred in 5.4%. The C-statistic revealed a low discriminatory ability for predicting in-hospital mortality (area under the ROC curve, 0.62; 95% confidence interval, 0.54-0.69; P = .002); however, model calibration was found to be excellent (Brier score of 0.07).

CONCLUSIONS

The ARCH score for in-hospital mortality showed low discriminatory ability in our local population, although with excellent ability for prediction of mortality.

Establishment and assessment of mortality risk prediction model in patients with sepsis based on early-stage peripheral lymphocyte subsets

This study is aimed to explore the value of lymphocyte subsets in evaluating the severity and prognosis of sepsis. The counts of lymphocytes, CD3+ T cells, CD4+ T cells, CD8+ T cells, CD19+ B cells, and NK cells significantly decreased between day 1 and day 3 in both the survivor and the non-survivor groups. The peripheral lymphocyte subsets (PLS) at day 1 were not significantly different between the survivor and the non-survivor groups. However, at day 3, the counts of lymphocytes, CD3+ T cells, CD4+ T cells, and NK cells were remarkably lower in the non-survivor group. No significant differences in CD8+ T cells, or CD19+ B cells were observed. The PLS index was independently and significantly associated with the 28-day mortality risk in septic patients (OR: 3.08, 95% CI: 1.18-9.67). Based on these clinical parameters and the PLS index, we developed a nomograph for evaluating the individual mortality risk in sepsis. The area under the curve of prediction with the PLS index was significantly higher than that from the model with only clinical parameters (0.912 vs. 0.817). Our study suggests that the decline of PLS occurred in the early stage of sepsis. The new novel PLS index can be an independent predictor of 28-day mortality in septic patients. The prediction model based on clinical parameters and the PLS index has relatively high predicting ability.

Early prediction of mortality at sepsis diagnosis time in critically ill patients by using interpretable machine learning

This study applied machine learning for the early prediction of 30-day mortality at sepsis diagnosis time in critically ill patients. Retrospective study using data collected from the Medical Information Mart for Intensive Care IV database. The data of the patient cohort was divided on the basis of the year of hospitalization, into training (2008-2013), validation (2014-2016), and testing (2017-2019) datasets. 24,377 patients with the sepsis diagnosis time < 24 h after intensive care unit (ICU) admission were included. A gradient boosting tree-based algorithm (XGBoost) was used for training the machine learning model to predict 30-day mortality at sepsis diagnosis time in critically ill patients. Model performance was measured in both discrimination and calibration aspects. The model was interpreted using the SHapley Additive exPlanations (SHAP) module. The 30-day mortality rate of the testing dataset was 17.9%, and 39 features were selected for the machine learning model. Model performance on the testing dataset achieved an area under the receiver operating characteristic curve (AUROC) of 0.853 (95% CI 0.837-0.868) and an area under the precision-recall curves of 0.581 (95% CI 0.541-0.619). The calibration plot for the model revealed a slope of 1.03 (95% CI 0.94-1.12) and intercept of 0.14 (95% CI 0.04-0.25). The SHAP revealed the top three most significant features, namely age, increased red blood cell distribution width, and respiratory rate. Our study demonstrated the feasibility of using the interpretable machine learning model to predict mortality at sepsis diagnosis time.

Prognostic prediction of sepsis patient using transformer with skip connected token for tabular data

Sepsis is known as a common syndrome in intensive care units (ICU), and severe sepsis and septic shock are among the leading causes of death worldwide. The purpose of this study is to develop a deep learning model that supports clinicians in efficiently managing sepsis patients in the ICU by predicting mortality, ICU length of stay (>14 days), and hospital length of stay (>30 days). The proposed model was developed using 591 retrospective data with 16 tabular data related to a sequential organ failure assessment (SOFA) score. To analyze tabular data, we designed the modified architecture of the transformer that has achieved extraordinary success in the field of languages and computer vision tasks in recent years. The main idea of the proposed model is to use a skip-connected token, which combines both local (feature-wise token) and global (classification token) information as the output of a transformer encoder. The proposed model was compared with four machine learning models (ElasticNet, Extreme Gradient Boosting [XGBoost]), and Random Forest) and three deep learning models (Multi-Layer Perceptron [MLP], transformer, and Feature-Tokenizer transformer [FT-Transformer]) and achieved the best performance (mortality, area under the receiver operating characteristic (AUROC) 0.8047; ICU length of stay, AUROC 0.8314; hospital length of stay, AUROC 0.7342). We anticipate that the proposed model architecture will provide a promising approach to predict the various clinical endpoints using tabular data such as electronic health and medical records.

Predictive Modeling Using Artificial Intelligence and Machine Learning Algorithms on Electronic Health Record Data: Advantages and Challenges

The rapid adoption of electronic health record (EHR) systems in US hospitals from 2008 to 2014 produced novel data elements for analysis. Concurrent innovations in computing architecture and machine learning (ML) algorithms have made rapid consumption of health data feasible and a powerful engine for clinical innovation. In critical care research, the net convergence of these trends has resulted in an exponential increase in outcome prediction research. In the following article, we explore the history of outcome prediction in the intensive care unit (ICU), the growing use of EHR data, and the rise of artificial intelligence and ML (AI) in critical care.

An Artificial Intelligence Approach to Guiding the Management of Heart Failure Patients Using Predictive Models: A Systematic Review

Heart failure (HF) is one of the leading causes of mortality and hospitalization worldwide. The accurate prediction of mortality and readmission risk provides crucial information for guiding decision making. Unfortunately, traditional predictive models reached modest accuracy in HF populations. We therefore aimed to present predictive models based on machine learning (ML) techniques in HF patients that were externally validated. We searched four databases and the reference lists of the included papers to identify studies in which HF patient data were used to create a predictive model. Literature screening was conducted in Academic Search Ultimate, ERIC, Health Source Nursing/Academic Edition and MEDLINE. The protocol of the current systematic review was registered in the PROSPERO database with the registration number CRD42022344855. We considered all types of outcomes: mortality, rehospitalization, response to treatment and medication adherence. The area under the receiver operating characteristic curve (AUC) was used as the comparator parameter. The literature search yielded 1649 studies, of which 9 were included in the final analysis. The AUCs for the machine learning models ranged from 0.6494 to 0.913 in independent datasets, whereas the AUCs for statistical predictive scores ranged from 0.622 to 0.806. Our study showed an increasing number of ML predictive models concerning HF populations, although external validation remains infrequent. However, our findings revealed that ML approaches can outperform conventional risk scores and may play important role in HF management.

Epidemiology of sepsis and risk factors for mortality in intensive care unit: a hospital based prospective study in South India

OBJECTIVE

The present study was aimed at elucidating the epidemiology of sepsis, with a special emphasis on identifying the common bacterial aetiology, proportion of infections caused by multi-drug resistant (MDR) bacteria, and risk factors associated with 28-day mortality at a university hospital in South India.

METHODS

A prospective study was undertaken from January 2017 to March 2018. Adult patients with the diagnosis of sepsis requiring intensive care unit (ICU) care were recruited. Baseline clinical, epidemiological, and laboratory data were recorded, and their association with 28-day mortality was assessed using logistic regression models.

RESULTS

400 subjects with a qSOFA score ≥2 at the time of ICU admission were included in the study. The mean age was 55.7 ± 16.6 years, and 69% were males. The mean SOFA score at the time of admission was 9.9 ± 2.7. Bacterial aetiology of sepsis was established in 53.5% of cases and 24% were caused by MDR pathogens. Carbapenem resistance was observed in 37% of the Gram-negative isolates. Escherichia coli (34.1%) was the leading pathogen. Overall, the 28-day mortality in ICU was 40%. 38% died within 48 h of ICU admission. Hypertension and SOFA > 9, male gender, and baseline-creatinine values >2.4 mg/dl were risk factors for mortality.

CONCLUSIONS

Male gender, hypertension, SOFA > 9, and increased creatinine were identified as the predictors for mortality. Infectious aetiology remained undetected in nearly half of the cases using routine microbiology culture methods. Mortality within the first 48 h of admission to ICU is high and prompts the need for increasing awareness about early sepsis diagnosis in community health care settings.

Establishment and validation of the predictive model for the in-hospital death in patients with sepsis

BACKGROUND

Identifying sepsis patients with risk of in-hospital death early can improve the prognosis of patients. This study aimed to develop a model to predict in-hospital death of sepsis patients based on the Medical Information Mart for Intensive Care-Ⅲ (MIMIC-Ⅲ) database, and use clinical data to externally validate the model.

METHODS

A total of 1,839 sepsis patients were used for model development, and 125 clinical cases were used for external validation. The discriminatory ability of the model was determined by calculating the area under the curve (AUC) with 95% confidence intervals (CI).

RESULTS

The AUC of the random forest model and logistic regression model was 0.754 (95%CI, 0.732-0.776) and 0.703 (95%CI, 0.680-0.727), respectively, and the random forest model had higher AUC (Z = 3.070, P = .002). External validation showed that the AUC of the random forest model was 0.539 (95%CI, 0.440-0.628). Further validation was carried out according to gender and SOFA score. The AUC of the model in males and females was 0.648 and 0.412, respectively. In addition, the AUC of the model in the population with SOFA scores of 3-8, 9-12, and 13-15 were 0.705, 0.495, and 0.769, respectively.

CONCLUSIONS

The random forest model had a better predictive ability and a good applicability to external populations with SOFA score of 13-15.

Validation of the Predisposition Infection Response Organ (PIRO) dysfunction score for the prognostic stratification of patients with sepsis in the Emergency Department

OBJECTIVE

There are many different methods for computing the Predisposition Infection Response Organ (PIRO) dysfunction score. We compared three PIRO methods (PIRO1 (Howell), PIRO2 (Rubulotta) and PIRO3 (Rathour)) for the stratification of mortality and high level of care admission in septic patients arriving at the Emergency Department (ED) of an Italian Hospital.

DESIGN, SETTING AND PARTICIPANTS

We prospectively collected clinical data of 470 patients admitted due to infection in the ED to compute PIRO according to three different methods. We tested PIRO variables for the prediction of mortality in the univariate analysis. Calculation and comparison were made of the area under the receiver operating curve (AUC) for the three PIRO methods, SOFA and qSOFA.

RESULTS

Most of the variables included in PIRO were related to mortality in the univariate analysis. Increased PIRO scores were related to higher mortality. In relation to mortality, PIRO 1 performed better than PIRO2 at 30 d ((AUC 0.77 (0.716-0.824) vs. AUC 0.699 (0.64-0.758) (p=0.03) and similarly at 60 d (AUC 0.767 (0.715-0.819) vs AUC 0.709 (0.656-0.763)(p=0.55)); PIRO1 performed similarly to PIRO3 (AUC 0.765 (0.71-0.82) at 30 d, AUC 0.754 (0.701-0.806) at 60 d, p=ns). Both PIRO1 and PIRO3 were as good as SOFA referred to mortality (AUC 0.758 (0.699, 0.816) at 30 d vs. AUC 0.738 (0.681, 0.795) at 60 d; p=ns). For high level of care admission, PIRO proved inferior to SOFA.

CONCLUSIONS

We support the use of PIRO1, which combines ease of use and the best performance referred to mortality over the short term. PIRO2 proved to be less accurate and more complex to use, suffering from missing microbiological data in the ED setting.

Ceftolozane/tazobactam versus meropenem in patients with ventilated hospital-acquired bacterial pneumonia: subset analysis of the ASPECT-NP randomized, controlled phase 3 trial

Background

Ceftolozane/tazobactam is approved for treatment of hospital-acquired/ventilator-associated bacterial pneumonia (HABP/VABP) at double the dose approved for other infection sites. Among nosocomial pneumonia subtypes, ventilated HABP (vHABP) is associated with the lowest survival. In the ASPECT-NP randomized, controlled trial, participants with vHABP treated with ceftolozane/tazobactam had lower 28-day all-cause mortality (ACM) than those receiving meropenem. We conducted a series of post hoc analyses to explore the clinical significance of this finding.

Methods

ASPECT-NP was a multinational, phase 3, noninferiority trial comparing ceftolozane/tazobactam with meropenem for treating vHABP and VABP; study design, efficacy, and safety results have been reported previously. The primary endpoint was 28-day ACM. The key secondary endpoint was clinical response at test-of-cure. Participants with vHABP were a prospectively defined subgroup, but subgroup analyses were not powered for noninferiority testing. We compared baseline and treatment factors, efficacy, and safety between ceftolozane/tazobactam and meropenem in participants with vHABP. We also conducted a retrospective multivariable logistic regression analysis in this subgroup to determine the impact of treatment arm on mortality when adjusted for significant prognostic factors.

Results

Overall, 99 participants in the ceftolozane/tazobactam and 108 in the meropenem arm had vHABP. 28-day ACM was 24.2% and 37.0%, respectively, in the intention-to-treat population (95% confidence interval [CI] for difference: 0.2, 24.8) and 18.2% and 36.6%, respectively, in the microbiologic intention-to-treat population (95% CI 2.5, 32.5). Clinical cure rates in the intention-to-treat population were 50.5% and 44.4%, respectively (95% CI − 7.4, 19.3). Baseline clinical, baseline microbiologic, and treatment factors were comparable between treatment arms. Multivariable regression identified concomitant vasopressor use and baseline bacteremia as significantly impacting ACM in ASPECT-NP; adjusting for these two factors, the odds of dying by day 28 were 2.3-fold greater when participants received meropenem instead of ceftolozane/tazobactam.

Conclusions

There were no underlying differences between treatment arms expected to have biased the observed survival advantage with ceftolozane/tazobactam in the vHABP subgroup. After adjusting for clinically relevant factors found to impact ACM significantly in this trial, the mortality risk in participants with vHABP was over twice as high when treated with meropenem compared with ceftolozane/tazobactam.

Trial registration

clinicaltrials.gov, NCT02070757. Registered 25 February, 2014, clinicaltrials.gov/ct2/show/NCT02070757.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-021-03694-3.

A New Method for Optimizing Sepsis Therapy by Nivolumab and Meropenem Combination: Importance of Early Intervention and CTL Reinvigoration Rate as a Response Marker

Background: Recently, there has been a growing interest in applying immune checkpoint blockers (ICBs), so far used to treat cancer, to patients with bacterial sepsis. We aimed to develop a method for predicting the personal benefit of potential treatments for sepsis, and to apply it to therapy by meropenem, an antibiotic drug, and nivolumab, a programmed cell death-1 (PD-1) pathway inhibitor.

Methods: We defined an optimization problem as a concise framework of treatment aims and formulated a fitness function for grading sepsis treatments according to their success in accomplishing the pre-defined aims. We developed a mathematical model for the interactions between the pathogen, the cellular immune system and the drugs, whose simulations under diverse combined meropenem and nivolumab schedules, and calculation of the fitness function for each schedule served to plot the fitness landscapes for each set of treatments and personal patient parameters.

Results: Results show that treatment by meropenem and nivolumab has maximum benefit if the interval between the onset of the two drugs does not exceed a dose-dependent threshold, beyond which the benefit drops sharply. However, a second nivolumab application, within 7–10 days after the first, can extinguish a pathogen which the first nivolumab application failed to remove. The utility of increasing nivolumab total dose above 6 mg/kg is contingent on the patient's personal immune attributes, notably, the reinvigoration rate of exhausted CTLs and the overall suppression rates of functional CTLs. A baseline pathogen load, higher than 5,000 CFU/μL, precludes successful nivolumab and meropenem combination therapy, whereas when the initial load is lower than 3,000 CFU/μL, meropenem monotherapy suffices for removing the pathogen.

Discussion: Our study shows that early administration of nivolumab, 6 mg/kg, in combination with antibiotics, can alleviate bacterial sepsis in cases where antibiotics alone are insufficient and the initial pathogen load is not too high. The study pinpoints the role of precision medicine in sepsis, suggesting that personalized therapy by ICBs can improve pathogen elimination and dampen immunosuppression. Our results highlight the importance in using reliable markers for classifying patients according to their predicted response and provides a valuable tool in personalizing the drug regimens for patients with sepsis.

Using machine learning methods to predict in-hospital mortality of sepsis patients in the ICU

Background

Early and accurate identification of sepsis patients with high risk of in-hospital death can help physicians in intensive care units (ICUs) make optimal clinical decisions. This study aimed to develop machine learning-based tools to predict the risk of hospital death of patients with sepsis in ICUs.

Methods

The source database used for model development and validation is the medical information mart for intensive care (MIMIC) III. We identified adult sepsis patients using the new sepsis definition Sepsis-3. A total of 86 predictor variables consisting of demographics, laboratory tests and comorbidities were used. We employed the least absolute shrinkage and selection operator (LASSO), random forest (RF), gradient boosting machine (GBM) and the traditional logistic regression (LR) method to develop prediction models. In addition, the prediction performance of the four developed models was evaluated and compared with that of an existent scoring tool – simplified acute physiology score (SAPS) II – using five different performance measures: the area under the receiver operating characteristic curve (AUROC), Brier score, sensitivity, specificity and calibration plot.

Results

The records of 16,688 sepsis patients in MIMIC III were used for model training and test. Amongst them, 2949 (17.7%) patients had in-hospital death. The average AUROCs of the LASSO, RF, GBM, LR and SAPS II models were 0.829, 0.829, 0.845, 0.833 and 0.77, respectively. The Brier scores of the LASSO, RF, GBM, LR and SAPS II models were 0.108, 0.109, 0.104, 0.107 and 0.146, respectively. The calibration plots showed that the GBM, LASSO and LR models had good calibration; the RF model underestimated high-risk patients; and SAPS II had the poorest calibration.

Conclusion

The machine learning-based models developed in this study had good prediction performance. Amongst them, the GBM model showed the best performance in predicting the risk of in-hospital death. It has the potential to assist physicians in the ICU to perform appropriate clinical interventions for critically ill sepsis patients and thus may help improve the prognoses of sepsis patients in the ICU.

Validation of the Predisposition Infection Response Organ (PIRO) dysfunction score for the prognostic stratification of patients with sepsis in the Emergency Department

OBJECTIVE

There are many different methods for computing the Predisposition Infection Response Organ (PIRO) dysfunction score. We compared three PIRO methods (PIRO1 (Howell), PIRO2 (Rubulotta) and PIRO3 (Rathour)) for the stratification of mortality and high level of care admission in septic patients arriving at the Emergency Department (ED) of an Italian Hospital.

DESIGN, SETTING AND PARTICIPANTS

We prospectively collected clinical data of 470 patients admitted due to infection in the ED to compute PIRO according to three different methods. We tested PIRO variables for the prediction of mortality in the univariate analysis. Calculation and comparison were made of the area under the receiver operating curve (AUC) for the three PIRO methods, SOFA and qSOFA.

RESULTS

Most of the variables included in PIRO were related to mortality in the univariate analysis. Increased PIRO scores were related to higher mortality. In relation to mortality, PIRO 1 performed better than PIRO2 at 30 d ((AUC 0.77 (0.716-0.824) vs. AUC 0.699 (0.64-0.758) (p=0.03) and similarly at 60 d (AUC 0.767 (0.715-0.819) vs AUC 0.709 (0.656-0.763)(p=0.55)); PIRO1 performed similarly to PIRO3 (AUC 0.765 (0.71-0.82) at 30 d, AUC 0.754 (0.701-0.806) at 60 d, p=ns). Both PIRO1 and PIRO3 were as good as SOFA referred to mortality (AUC 0.758 (0.699, 0.816) at 30 d vs. AUC 0.738 (0.681, 0.795) at 60 d; p=ns). For high level of care admission, PIRO proved inferior to SOFA.

CONCLUSIONS

We support the use of PIRO1, which combines ease of use and the best performance referred to mortality over the short term. PIRO2 proved to be less accurate and more complex to use, suffering from missing microbiological data in the ED setting.

Establishment and evaluation of a multicenter collaborative prediction model construction framework supporting model generalization and continuous improvement: A pilot study

BACKGROUND AND OBJECTIVE

In recent years, an increasing number of clinical prediction models have been developed to serve clinical care. Establishing a data-driven prediction model based on large-scale electronic health record (EHR) data can provide a more empirical basis for clinical decision making. However, research on model generalization and continuous improvement is insufficiently focused, which also hinders the application and evaluation of prediction models in real clinical environments. Therefore, this study proposes a multicenter collaborative prediction model construction framework to build a prediction model with greater generalizability and continuous improvement capabilities while preserving patient data security and privacy.

MATERIALS AND METHODS

Based on a multicenter collaborative research network, such as the Observational Health Data Sciences and Informatics (OHDSI), a multicenter collaborative prediction model construction framework is proposed. Based on the idea of multi-source transfer learning, in each source hospital, a base classifier was trained according to the model research setting. Then, in the target hospital with missing calibration data, a prediction model was established through weighted integration of base classifiers from source hospitals based on the smoothness assumption. Moreover, a passive-aggressive online learning algorithm was used for continuous improvement of the prediction model, which can help to maintain a high predictive performance to provide reliable clinical decision-making abilities. To evaluate the proposed prediction model construction framework, a prototype system for colorectal cancer prognosis prediction was developed. To evaluate the performance of models, 70,906 patients were screened, including 70,090 from 5 US hospital-specific datasets and 816 from a Chinese hospital-specific dataset. The area under the receiver operating characteristic curve (AUC) and the estimated calibration index (ECI) were used to evaluate the discrimination and calibration of models.

RESULTS

Regarding the colorectal cancer prognosis prediction in our prototype system, compared with the reference models, our model achieved a better performance in model calibration (ECI = 9.294 [9.146, 9.441]) and a similar ability in model discrimination (AUC = 0.783 [0.780, 0.786]). Furthermore, the online learning process provided in this study can continuously improve the performance of the prediction model when patient data with specified labels arrive (the AUC value increased from 0.709 to 0.715 and the ECI value decreased from 13.013 to 9.634 after 650 patient instances with specified labels from the Chinese hospital arrived), enabling the prediction model to maintain a good predictive performance during clinical application.

CONCLUSIONS

This study proposes and evaluates a multicenter collaborative prediction model construction framework that can support the construction of prediction models with better generalizability and continuous improvement capabilities without the need to aggregate multicenter patient-level data.

A prospective, multicenter case control study of risk factors for acquisition and mortality in Enterobacter species bacteremia

BACKGROUND

Enterobacter is among the main etiologies of hospital-acquired infections. This study aims to identify the risk factors of acquisition and attributable mortality of Enterobacter bacteremia.

METHODS

Observational, case-control study for risk factors and prospective cohort for outcomes of consecutive cases with Enterobacter bacteremia. This study was conducted in five hospitals in Spain over a three-year period. Matched controls were patients with negative blood cultures and same sex, age, and hospitalization area.

RESULTS

The study included 285 cases and 570 controls. E. cloacae was isolated in 198(68.8%) cases and E. aerogenes in 89(31.2%). Invasive procedures (hemodialysis, nasogastric tube, mechanical ventilation, surgical drainage tube) and previous antibiotics or corticosteroids were independently associated with Enterobacter bacteremia. Its attributable mortality was 7.8%(CI95%2.7-13.4%), being dissimilar according to a McCabe index: non-fatal=3.2%, ultimately fatal=12.9% and rapidly fatal=0.12%. Enterobacter bacteremia remained an independent risk factor for mortality among cases with severe sepsis or septic shock (OR 5.75 [CI95%2.57-12.87], p<0.001), with an attributable mortality of 40.3%(CI95%25.7-53.3). Empiric therapy or antibiotic resistances were not related to the outcome among patients with bacteremia.

CONCLUSIONS

Invasive procedures, previous antibiotics and corticosteroids predispose to acquire Enterobacter bacteremia. This entity increases mortality among fragile patients and those with severe infections. Antibiotic resistances did not affect the outcome.

Frequency and mortality of septic shock in Europe and North America: a systematic review and meta-analysis

Background

Septic shock is the most severe form of sepsis, in which profound underlying abnormalities in circulatory and cellular/metabolic parameters lead to substantially increased mortality. A clear understanding and up-to-date assessment of the burden and epidemiology of septic shock are needed to help guide resource allocation and thus ultimately improve patient care. The aim of this systematic review and meta-analysis was therefore to provide a recent evaluation of the frequency of septic shock in intensive care units (ICUs) and associated ICU and hospital mortality.

Methods

We searched MEDLINE, Embase, and the Cochrane Library from 1 January 2005 to 20 February 2018 for observational studies that reported on the frequency and mortality of septic shock. Four reviewers independently selected studies and extracted data. Disagreements were resolved via consensus. Random effects meta-analyses were performed to estimate pooled frequency of septic shock diagnosed at admission and during the ICU stay and to estimate septic shock mortality in the ICU, hospital, and at 28 or 30 days.

Results

The literature search identified 6291 records of which 71 articles met the inclusion criteria. The frequency of septic shock was estimated at 10.4% (95% CI 5.9 to 16.1%) in studies reporting values for patients diagnosed at ICU admission and at 8.3% (95% CI 6.1 to 10.7%) in studies reporting values for patients diagnosed at any time during the ICU stay. ICU mortality was 37.3% (95% CI 31.5 to 43.5%), hospital mortality 39.0% (95% CI 34.4 to 43.9%), and 28-/30-day mortality 36.7% (95% CI 32.8 to 40.8%). Significant between-study heterogeneity was observed.

Conclusions

Our literature review reaffirms the continued common occurrence of septic shock and estimates a high mortality of around 38%. The high level of heterogeneity observed in this review may be driven by variability in defining and applying the diagnostic criteria, as well as differences in treatment and care across settings and countries.

Electronic supplementary material

The online version of this article (10.1186/s13054-019-2478-6) contains supplementary material, which is available to authorized users.

Influence of empirical double-active combination antimicrobial therapy compared with active monotherapy on mortality in patients with septic shock: a propensity score-adjusted and matched analysis

Objectives

To evaluate the influence on mortality of empirical double-active combination antimicrobial therapy (DACT) compared with active monotherapy (AM) in septic shock patients.

Methods

A retrospective study was performed of monomicrobial septic shock patients admitted to a university centre during 2010-15. A propensity score (PS) was calculated using a logistic regression model taking the assigned therapy as the dependent variable, and used as a covariate in multivariate analysis predicting 7, 15 and 30 day mortality and for matching patients who received DACT or AM. Multivariate models comprising the assigned therapy group and the PS were built for specific patient subgroups.

Results

Five-hundred and seventy-six patients with monomicrobial septic shock who received active empirical antimicrobial therapy were included. Of these, 340 received AM and 236 DACT. No difference in 7, 15 and 30 day all-cause mortality was found between groups either in the PS-adjusted multivariate logistic regression analysis or in the PS-matched cohorts. However, in patients with neutropenia, DACT was independently associated with a better outcome at 15 (OR 0.29, 95% CI 0.09-0.92) and 30 (OR 0.25, 95% CI 0.08-0.79) days, while in patients with Pseudomonas aeruginosa infection DACT was associated with lower 7 (OR 0.12, 95% CI 0.02-0.7) and 30 day (OR 0.26, 95% CI 0.08-0.92) mortality.

Conclusions

All-cause mortality at 7, 15 and 30 days was similar in patients with monomicrobial septic shock receiving empirical double-active combination therapy and active monotherapy. However, a beneficial influence of empirical double-active combination on mortality in patients with neutropenia and those with P. aeruginosa infection is worthy of further study.

Risk factors for long-term mortality in patients admitted with severe infection

Background

Severe infection is a main cause of mortality. We aim to describe risk factors for long-term mortality among inpatients with severe infection.

Methods

Prospective cohort study in a 600-bed university hospital in Portugal including all patients with severe infection admitted into intensive care, medical, surgical, hematology and nephrology wards over one-year period. The outcome of interest was 5-year mortality following infection. Variables of patient background and infectious episode were studied in association with the main outcome through multiple logistic regression. There were 1013 patients included in the study. Hospital and 5-year mortality rates were 14 and 37%, respectively.

Results

Two different models were developed (with and without acute-illness severity scores) and factors independently associated with 5-year mortality were [adjusted odds ratio (95% confidence interval)]: age = 1.03 per year (1.02-1.04), cancer = 4.36 (1.65–11.53), no comorbidities = 0.4 (0.26–0.62), Karnovsky Index < 70 = 2.25 (1.48–3.40), SAPS (Simplified Acute Physiology Score) II = 1.05 per point (1.03–1.07), positive blood cultures = 1.57 (1.01–2.44) and infection by an ESKAPE pathogen (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeroginosa and Enterobacter species) = 1.61 (1.00– 2.60); and in the second model [without SAPS II and SOFA (Sequential Organ Failure Assessment) scores]: age = 1.04 per year (1.03–1.05), cancer = 5.93 (2.26–15.51), chronic haematologic disease = 2.37 (1.14–4.93), no comorbidities = 0.45 (0.29–0.69), Karnovsky Index< 70 = 2.32 (1.54– 3.50), septic shock [reference is infection without SIRS (Systemic Inflammatory Response Syndrome)] = 3.77 (1.80–7.89) and infection by an ESKAPE pathogen = 1.61 (1.00–2.60). Both models presented a good discrimination power with an AU-ROC curve (95% CI) of 0.81 (0.77–0.84) for model 1 and 0.80 (0.76–0.83) for model 2. If only patients that survived hospital admission are included in the model, variables retained are: age = 1.03 per year (1.02–1.05), cancer = 4.69 (1.71–12.83), chronic respiratory disease = 2.27 (1.09–4.69), diabetes mellitus = 1.65 (1.06–2.56), Karnovsky Index < 70 = 2.50 (1.63–3.83) and positive blood cultures = 1.66 (1.04–2.64) with an AU-ROC curve of 0.77 (0.73–0.81).

Conclusions

Age, previous comorbidities, and functional status and infection by an ESKAPE pathogen were consistently associated with long-term prognosis. This information may help in the discussion of individual prognosis and clinical decision-making.

Risk of Recurrence After Surviving Severe Sepsis: A Matched Cohort Study

OBJECTIVES

To examine the risk of recurrence in adults who survived first-episode severe sepsis for at least 3 months.

DESIGN

A matched cohort study.

SETTING

Inpatient claims data from Taiwan's National Health Insurance Research Database.

SUBJECTS

We analyzed 10,818 adults who survived first-episode severe sepsis without recurrence for at least 3 months in 2000 (SS group; mean age, 62.7 yr; men, 54.7%) and a group of age/sex-matched (1:1) population controls who had no prior history of severe sepsis. All subjects were followed from the study entry to the occurrence of end-point, death, or until December 31, 2008, whichever date came first.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Primary end-point was severe sepsis that occurred after January 1, 2001 (the study entry). Relative risk of the end-point was assessed using competing risk regression model. During the follow-up period, severe sepsis and death occurred in 35.0% and 26.5% of SS group and in 4.3% and 18.6% of controls, respectively, representing a covariate-adjusted sub-hazard ratio of 8.89 (95% CI, 8.04-9.83) for the risk of recurrence. In stratified analysis by patient characteristics, the sub-hazard ratios ranged from 7.74 in rural area residents to 23.17 in young adults. In subgroup analysis by first-episode infection sites in SS group, the sub-hazard ratios ranged from 4.82 in intra-abdominal infection to 9.99 in urinary tract infection.

CONCLUSIONS

Risk of recurrence after surviving severe sepsis is substantial regardless of patient characteristics or infection sites. Further research is necessary to find underlying mechanisms for the high risk of recurrence in these patients.

Predictors of mortality of severe sepsis among adult patients in the medical Intensive Care Unit

Background:

Sepsis is an important cause of mortality in the Intensive Care Units (ICUs) worldwide. Information regarding early predictive factors for mortality and morbidity is limited.

Aims and Objectives:

The primary objective of the study was to estimate the mortality of severe sepsis among adult patients admitted into the medical ICU. The secondary objective was to identify the predictors associated with mortality.

Materials and Methods:

Adult patients admitted with severe sepsis in the medical ICU were studied. The primary outcome was the mortality among the study population. Baseline demographic, clinical, and laboratory data were recorded upon inclusion into the study. Risk factors associated with mortality were studied by univariate analysis. The variables having statistical significance were further included in multivariate analysis to identify the independent predictors of mortality.

Results:

Out of eighty patients, 54 (67.5%) died. Univariate analysis showed that age >60 years, tachycardia, hypotension, elevated C-reactive protein (CRP) and lactate, thrombocytopenia, need of mechanical ventilation, and high Acute Physiology and Chronic Health Evaluation (APACHE) II and Sequential Organ Failure Assessment scores were variables associated with high mortality. The independent predictors of mortality identified by multivariate regression analysis were platelet count below 1 lakhs, serum levels of CRP >100, APACHE II score >25 on the day of admission to the ICU with severe sepsis, and the need for invasive mechanical ventilation.

Conclusions:

Low platelet count, elevated serum levels of CRP, APACHE score >25, and the need for invasive mechanical ventilation were found to be independent predictors of mortality of severe sepsis among adult patients with severe sepsis in the medical ICU.

Long-term outcome of critically ill adult patients with acute epiglottitis

Background

Acute epiglottitis is a potentially life threatening disease, with a growing incidence in the adult population. Its long-term outcome after Intensive Care Unit (ICU) hospitalization has rarely been studied.

Methodology and Principal Findings

Thirty-four adult patients admitted for acute epiglottitis were included in this retrospective multicentric study. The mean age was 44±12 years (sex ratio: 5.8). Sixteen patients (47%) had a history of smoking while 8 (24%) had no previous medical history. The average time of disease progression before ICU was 2.6±3.6 days. The main reasons for hospitalization were continuous monitoring (17 cases, 50%) and acute respiratory distress (10 cases, 29%). Microbiological documentation could be made in 9 cases (26%), with Streptococcus spp. present in 7 cases (21%). Organ failure at ICU admission occurred in 8 cases (24%). Thirteen patients (38%) required respiratory assistance during ICU stay; 9 (26%) required surgery. Two patients (6%) died following hypoxemic cardiac arrest. Five patients (15%) had sequelae at 1 year. Patients requiring respiratory assistance had a longer duration of symptoms and more frequent anti inflammatory use before ICU admission and sequelae at 1 year (p<0.05 versus non-ventilated patients). After logistic regression analysis, only exposure to anti-inflammatory drugs before admission was independently associated with airway intervention (OR, 4.96; 95% CI, 1.06-23.16).

Conclusions and Significance

The profile of the cases consisted of young smoking men with little comorbidity. Streptococcus spp. infection represented the main etiology. Outcome was favorable if early respiratory tract protection could be performed in good conditions. Morbidity and sequelae were greater in patients requiring airway intervention.

PREDICT: a diagnostic accuracy study of a tool for predicting mortality within one year: who should have an advance healthcare directive?

BACKGROUND

CARING is a screening tool developed to identify patients who have a high likelihood of death in 1 year.

AIM

This study sought to validate a modified CARING tool (termed PREDICT) using a population of patients presenting to the Emergency Department.

SETTING/PARTICIPANTS

In total, 1000 patients aged over 55 years who were admitted to hospital via the Emergency Department between January and June 2009 were eligible for inclusion in this study.

DESIGN

Data on the six prognostic indicators comprising PREDICT were obtained retrospectively from patient records. One-year mortality data were obtained from the State Death Registry. Weights were applied to each PREDICT criterion, and its final score ranged from 0 to 44. Receiver operator characteristic analyses and diagnostic accuracy statistics were used to assess the accuracy of PREDICT in identifying 1-year mortality.

RESULTS

The sample comprised 976 patients with a median (interquartile range) age of 71 years (62-81 years) and a 1-year mortality of 23.4%. In total, 50% had ≥1 PREDICT criteria with a 1-year mortality of 40.4%. Receiver operator characteristic analysis gave an area under the curve of 0.86 (95% confidence interval: 0.83-0.89). Using a cut-off of 13 points, PREDICT had a 95.3% (95% confidence interval: 93.6-96.6) specificity and 53.9% (95% confidence interval: 47.5-60.3) sensitivity for predicting 1-year mortality. PREDICT was simpler than the CARING criteria and identified 158 patients per 1000 admitted who could benefit from advance care planning.

CONCLUSION

PREDICT was successfully applied to the Australian healthcare system with findings similar to the original CARING study conducted in the United States. This tool could improve end-of-life care by identifying who should have advance care planning or an advance healthcare directive.

External validation of new risk prediction models is infrequent and reveals worse prognostic discrimination

OBJECTIVES

To evaluate how often newly developed risk prediction models undergo external validation and how well they perform in such validations.

STUDY DESIGN AND SETTING

We reviewed derivation studies of newly proposed risk models and their subsequent external validations. Study characteristics, outcome(s), and models' discriminatory performance [area under the curve, (AUC)] in derivation and validation studies were extracted. We estimated the probability of having a validation, change in discriminatory performance with more stringent external validation by overlapping or different authors compared to the derivation estimates.

RESULTS

We evaluated 127 new prediction models. Of those, for 32 models (25%), at least an external validation study was identified; in 22 models (17%), the validation had been done by entirely different authors. The probability of having an external validation by different authors within 5 years was 16%. AUC estimates significantly decreased during external validation vs. the derivation study [median AUC change: -0.05 (P < 0.001) overall; -0.04 (P = 0.009) for validation by overlapping authors; -0.05 (P < 0.001) for validation by different authors]. On external validation, AUC decreased by at least 0.03 in 19 models and never increased by at least 0.03 (P < 0.001).

CONCLUSION

External independent validation of predictive models in different studies is uncommon. Predictive performance may worsen substantially on external validation.

Development of a daily mortality probability prediction model from Intensive Care Unit patients using a discrete-time event history analysis

As studies have pointed out, severity scores are imperfect at predicting individual clinical chance of survival. The clinical condition and pathophysiological status of these patients in the Intensive Care Unit might differ from or be more complicated than most predictive models account for. In addition, as the pathophysiological status changes over time, the likelihood of survival day by day will vary. Actually, it would decrease over time and a single prediction value cannot address this truth. Clearly, alternative models and refinements are warranted. In this study, we used discrete-time-event models with the changes of clinical variables, including blood cell counts, to predict daily probability of mortality in individual patients from day 3 to day 28 post Intensive Care Unit admission. Both models we built exhibited good discrimination in the training (overall area under ROC curve: 0.80 and 0.79, respectively) and validation cohorts (overall area under ROC curve: 0.78 and 0.76, respectively) to predict daily ICU mortality. The paper describes the methodology, the development process and the content of the models, and discusses the possibility of them to serve as the foundation of a new bedside advisory or alarm system.

Predisposition, insult/infection, response and organ dysfunction (PIRO): a pilot clinical staging system for hospital mortality in patients with infection

PURPOSE

To develop a clinical staging system based on the PIRO concept (Predisposition, Infection, RESPONSE and Organ dysfunction) for hospitalized patients with infection.

METHODS

One year prospective cohort study of all hospitalized patients with infection (n = 1035), admitted into a large tertiary care, university hospital. Variables associated with hospital mortality were selected using logistic regressions. Based on the regression coefficients, a score for each PIRO component was developed and a classification tree was used to stratify patients into four stages of increased risk of hospital mortality. The final clinical staging system was then validated using an independent cohort (n = 186).

RESULTS

Factors significantly associated with hospital mortality were • for Predisposition: age, sex, previous antibiotic therapy, chronic hepatic disease, chronic hematologic disease, cancer, atherosclerosis and a Karnofsky index<70; • for Insult/Infection: type of infection • for

RESPONSE

abnormal temperature, tachypnea, hyperglycemia and severity of infection and • for Organ dysfunction: hypotension and SOFA score≥1. The area under the ROC curve (CI95%) for the combined PIRO model as a predictor for mortality was 0.85 (0.82-0.88). Based on the scores for each of the PIRO components and on the cut-offs estimated from the classification tree, patients were stratified into four stages of increased mortality rates: stage I: ≤5%, stage II: 6-20%, stage III: 21-50% and stage IV: >50%. Finally, this new clinical staging system was studied in a validation cohort, which provided similar results (0%, 9%, 31% and 67%, in each stage, respectively).

CONCLUSIONS

Based on the PIRO concept, a new clinical staging system was developed for hospitalized patients with infection, allowing stratification into four stages of increased mortality, using the different scores obtained in Predisposition, RESPONSE, Infection and Organ dysfunction. The proposed system will likely help to define inclusion criteria in clinical trials as well as tailoring individual management plans for patients with infection.

Late recognition and illness severity are determinants of early death in severe septic patients

OBJECTIVE

To identify the independent variables associated with death within 4 days after the first sepsis-induced organ dysfunction.

METHODS

In this prospective observational study, severe sepsis and septic shock patients were classified into 3 groups: Group 1, survivors; Group 2, late non-survivors; and Group 3, early non-survivors. Early death was defined as death occurring within 4 days after the first sepsis-induced organ dysfunction. Demographic, clinical and laboratory data were collected and submitted to univariate and multinomial analyses.

RESULTS

The study included 414 patients: 218 (52.7%) in Group 1, 165 (39.8%) in Group 2, and 31 (7.5%) in Group 3. A multinomial logistic regression analysis showed that age, Acute Physiology and Chronic Health Evaluation II score, Sepsis-related Organ Failure Assessment score after the first 24 hours, nosocomial infection, hepatic dysfunction, and the time elapsed between the onset of organ dysfunction and the sepsis diagnosis were associated with early mortality. In contrast, Black race and a source of infection other than the urinary tract were associated with late death. Among the non-survivors, early death was associated with Acute Physiology and Chronic Health Evaluation II score, chronic renal failure, hepatic dysfunction Sepsis-related Organ Failure Assessment score after 24 hours, and the duration of organ dysfunction.

CONCLUSION

Factors related to patients' intrinsic characteristics and disease severity as well as the promptness of sepsis recognition are associated with early death among severe septic patients.

Diagnostic performance of a multiple real-time PCR assay in patients with suspected sepsis hospitalized in an internal medicine ward

Early identification of causative pathogen in sepsis patients is pivotal to improve clinical outcome. SeptiFast (SF), a commercially available system for molecular diagnosis of sepsis based on PCR, has been mostly used in patients hospitalized in hematology and intensive care units. We evaluated the diagnostic accuracy and clinical usefulness of SF, compared to blood culture (BC), in 391 patients with suspected sepsis, hospitalized in a department of internal medicine. A causative pathogen was identified in 85 patients (22%). Sixty pathogens were detected by SF and 57 by BC. No significant differences were found between the two methods in the rates of pathogen detection (P = 0.74), even after excluding 9 pathogens which were isolated by BC and were not included in the SF master list (P = 0.096). The combination of SF and BC significantly improved the diagnostic yield in comparison to BC alone (P < 0.001). Compared to BC, SF showed a significantly lower contamination rate (0 versus 19 cases; P < 0.001) with a higher specificity for pathogen identification (1.00, 95% confidence interval [CI] of 0.99 to 1.00, versus 0.94, 95% CI of 0.90 to 0.96; P = 0.005) and a higher positive predictive value (1.00, 95% CI of 1.00 to 0.92%, versus 0.75, 95% CI of 0.63 to 0.83; P = 0.005). In the subgroup of patients (n = 191) who had been receiving antibiotic treatment for ≥24 h, SF identified more pathogens (16 versus 6; P = 0.049) compared to BC. These results suggest that, in patients with suspected sepsis, hospitalized in an internal medicine ward, SF could be a highly valuable adjunct to conventional BC, particularly in patients under antibiotic treatment.

Therapeutic impact and diagnostic performance of multiplex PCR in patients with malignancies and suspected sepsis

OBJECTIVES

New molecular methods allow rapid pathogen detection in patients with sepsis, but their impact on treatment decisions remains to be established. We evaluated the therapeutic usefulness of multiplex PCR testing in patients with cancer and sepsis.

METHODS

110 patients with cancer and sepsis were included prospectively and underwent LightCycler® SeptiFast (LC-SF) multiplex PCR testing in addition to standard tests. Two independent panels of experts assessed the diagnosis in each patient based on medical record data; only one panel had the LC-SF results. The final diagnosis established by a third panel was the reference standard.

RESULTS

The final diagnosis was documented sepsis in 50 patients (55 microorganisms), undocumented sepsis in 54, and non-infectious disease in 6. LC-SF detected 17/32 pathogens recovered from blood cultures (BC) and 11/23 pathogens not recovered from BC; 12 microorganisms were detected neither by BC nor by LC-SF. LC-SF produced false-positive results in 10 cases. The LC-SF results would have significantly improved treatment in 11 (10%) patients and prompted immediate antimicrobial therapy not given initially in 3 patients.

CONCLUSIONS

In cancer patients with suspected sepsis, LC-SF detected 11/55 (20%) true pathogens not recovered from BCs and would have improved the initial management in 11/110 (10%) patients.

Skin and soft tissue infections in hospitalized and critically ill patients: a nationwide population-based study

Background

The proportional distributions of various skin and soft tissue infections (SSTIs) with/without intensive care are unclear. Among SSTI patients, the prevalence and significance of complicating factors, such as comorbidities and infections other than skin/soft tissue (non-SST infections), remain poorly understood. We conducted this population-based study to characterize hospitalized SSTI patients with/without intensive care and to identify factors associated with patient outcome.

Methods

We analyzed first-episode SSTIs between January 1, 2005 and December 31, 2007 from the hospitalized claims data of a nationally representative sample of 1,000,000 people, about 5% of the population, enrolled in the Taiwan National Health Insurance program. We classified 18 groups of SSTIs into three major categories: 1) superficial; 2) deeper or healthcare-associated; and 3) gangrenous or necrotizing infections. Multivariate logistic regression models were applied to identify factors associated with intensive care unit (ICU) admission and hospital mortality.

Results

Of 146,686 patients ever hospitalized during the 3-year study period, we identified 11,390 (7.7%) patients having 12,030 SSTIs. Among these SSTI patients, 1,033 (9.1%) had ICU admission and 306 (2.7%) died at hospital discharge. The most common categories of SSTIs in ICU and non-ICU patients were "deeper or healthcare-associated" (62%) and "superficial" (60%) infections, respectively. Of all SSTI patients, 45.3% had comorbidities and 31.3% had non-SST infections. In the multivariate analyses adjusting for demographics and hospital levels, the presence of several comorbid conditions was associated with ICU admission or hospital mortality, but the results were inconsistent across most common SSTIs. In the same analyses, the presence of non-SST infections was consistently associated with increased risk of ICU admission (adjusted odds ratios [OR] 3.34, 95% confidence interval [CI] 2.91-3.83) and hospital mortality (adjusted OR 5.93, 95% CI 4.57-7.71).

Conclusions

The proportional distributions of various SSTIs differed between ICU and non-ICU patients. Nearly one-third of hospitalized SSTI patients had non-SST infections, and the presence of which predicted ICU admission and hospital mortality.

The influence of infection sites on development and mortality of ARDS

OBJECTIVE

Infection is the most frequent cause of acute respiratory distress syndrome (ARDS). However, little is known about the influence of infection sites on ARDS. This study aimed to assess the associations of infection sites with ARDS development and mortality in critically ill infected patients.

DESIGN

Prospective observational study.

SETTING

Adult intensive care units (ICUs) of an academic medical center.

PATIENTS

Study population included 1,973 consecutive patients admitted to ICUs with bacteremia, pneumonia or sepsis. During follow-up, 549 patients developed ARDS and 212 of them died within 60 days.

MAIN RESULTS

The distribution of infection sites in ARDS patients was: lung (77.2%), abdomen (19.3%), skin/soft tissues (6.0%), urinary tract (4.7%), unknown (2.6%), and multiple sites (17.7%). On multivariate analysis, lung was the only infection site associated with increased ARDS risk [adjusted odds ratio (OR) 3.49]. Urinary tract (adjusted OR 0.43), skin/soft tissue (adjusted OR 0.64), and unknown-site infections (adjusted OR 0.38) were associated with decreased risk. No association was found between individual infection site and ARDS mortality. However, unknown-site [adjusted hazard ratio (HR) 3.08] and multiple-site infections (adjusted HR 1.63) were associated with increased ARDS mortality. When grouping patients into pulmonary, nonpulmonary, and combined infections, nonpulmonary infection was associated with decreased ARDS risk (adjusted OR 0.28) and combined infections was associated with increased ARDS mortality (adjusted HR 1.69), compared with pulmonary infection.

CONCLUSIONS

In critically ill infected patients, pulmonary infection is associated with higher risk of ARDS development than are infections at other sites. Pulmonary versus nonpulmonary infection significantly affects ARDS development but not mortality.