Outcomes in Burn-Injured Patients Who Develop Sepsis.

This study examines health outcomes in burn patients with sepsis. We hypothesized that burn patients with sepsis would have an increased odds risk for in-hospital death and longer intensive care unit (ICU) stays. This was a retrospective cohort of consecutive patients admitted to the burn ICU with total BSA (TBSA) ≥10% and/or inhalation injury between January 2008 and March 2015. Overall 407 burn patients were included; the case-rate for sepsis was 39.1% (n = 159); 20.1% (n = 82) patients were septic and 18.9% (n = 77) patients experienced septic shock. Patients with septic shock had the highest mortality rate (13.31% no sepsis vs 3.7% sepsis vs 49.4% septic shock, P < .01). Median 28-day ICU-free days was higher in patients without sepsis (23 days [Interquartile range (IQR) 14-27] no sepsis vs 0 days [IQR 0-10] sepsis vs 0 days [IQR 0-0] septic shock, P < .01). Sepsis (with or without shock) increased odds of in-hospital death (odds ratio 7.04, 95% confidence interval 1.93-25.7) in reference to the no sepsis group. With each incremental Sequential Organ Failure Assessment (SOFA) score or 10% TBSA increase, the odds risk for in-hospital death increased by 56 and 75%, respectively. Our study characterized outcomes in patients with sepsis after severe burn injury. The odds risk for in-hospital death was greater in patients with sepsis, increasing burn severity according to TBSA and SOFA score.

A Computable Phenotype for Acute Respiratory Distress Syndrome Using Natural Language Processing and Machine Learning.

Acute Respiratory Distress Syndrome (ARDS) is a syndrome of respiratory failure that may be identified using text from radiology reports. The objective of this study was to determine whether natural language processing (NLP) with machine learning performs better than a traditional keyword model for ARDS identification. Linguistic pre-processing of reports was performed and text features were inputs to machine learning classifiers tuned using 10-fold cross-validation on 80% of the sample size and tested in the remaining 20%. A cohort of 533 patients was evaluated, with a data corpus of 9,255 radiology reports. The traditional model had an accuracy of 67.3% (95% CI: 58.3-76.3) with a positive predictive value (PPV) of 41.7% (95% CI: 27.7-55.6). The best NLP model had an accuracy of 83.0% (95% CI: 75.9-90.2) with a PPV of 71.4% (95% CI: 52.1-90.8). A computable phenotype for ARDS with NLP may identify more cases than the traditional model.

Comparison of Automated Methods Versus the American Burn Association Sepsis Definition to Identify Sepsis and Sepsis With Organ Dysfunction/Septic Shock in Burn-Injured Adults.

To develop an algorithm to identify sepsis and sepsis with organ dysfunction/septic shock in burn-injured patients incorporating criteria from the American Burn Association sepsis definition that possesses good test characteristics compared with International Classification of Diseases, Ninth Revision-Clinical Modification (ICD-9) codes and an algorithm previously validated in nonburn-injured septic patients (Martin et al method). This was a retrospective cohort study of consecutive patients admitted to the burn intensive care unit between January 2008 and March 2015. Of the 4761 admitted, 8.6% (n = 407) met inclusion criteria, of which the case rate for sepsis was 34.2% (n = 139; n = 48 sepsis; n = 91 sepsis with organ dysfunction/septic shock). For sepsis identification, the novel algorithm had an accuracy of 86.0% (95% CI: 82.2-89.2%), sensitivity of 66.9% (95% CI: 59.1-74.7%), and specificity of 95.9% (95% CI: 93.5-98.3%). The novel algorithm had better discrimination (0.81, 95% CI: 0.77-0.86) than the ICD-9 method (0.77, 95% CI: 0.73-0.81), although this was not significant (P = .08). For sepsis with organ dysfunction/septic shock, the novel algorithm plus vasopressors (0.67, 95% CI: 0.63-0.72) and the ICD-9 method (0.63, 95% CI: 0.58-0.68) performed equivocal (P = 0.15) but the Martin method (0.76, 95% CI: 0.71-0.81) had superior discrimination than other methods (P < .01). The novel algorithm is an accurate and simple tool to identify sepsis in the burn cohort with good sensitivity and specificity and equivocal discriminative ability to ICD-9 coding. The Martin method had superior discriminative ability for identifying sepsis with organ dysfunction/septic shock in burn-injured patients than either the novel algorithm plus vasopressors or ICD-9 coding.

Patient-Centered Outcomes Research in Practice: The CAPriCORN Infrastructure.

CAPriCORN, the Chicago Area Patient Centered Outcomes Research Network, is one of the eleven PCORI-funded Clinical Data Research Networks. A collaboration of six academic medical centers, a Chicago public hospital, two VA hospitals and a network of federally qualified health centers, CAPriCORN addresses the needs of a diverse community and overlapping populations. To capture complete medical records without compromising patient privacy and confidentiality, the network created policies and mechanisms for patient consultation, central IRB approval, de-identification, de-duplication, and integration of patient data by study cohort, randomization and sampling, re-identification for consent by providers and patients, and communication with patients to elicit patient-reported outcomes through validated instruments. The paper describes these policies and mechanisms and discusses two case studies to prove the feasibility and effectiveness of the network.

The Gene Expression Barcode 3.0: improved data processing and mining tools.

The Gene Expression Barcode project, http://barcode.luhs.org, seeks to determine the genes expressed for every tissue and cell type in humans and mice. Understanding the absolute expression of genes across tissues and cell types has applications in basic cell biology, hypothesis generation for gene function and clinical predictions using gene expression signatures. In its current version, this project uses the abundant publicly available microarray data sets combined with a suite of single-array preprocessing, quality control and analysis methods. In this article, we present the improvements that have been made since the previous version of the Gene Expression Barcode in 2011. These include a variety of new data mining tools and summaries, estimated transcriptomes and curated annotations.

Development of a tool to manage patient health records in support of burn injury research.

Data captured in electronic medical records (EMRs) and paper charts have enormous potential for clinical research and to improve the quality of health care; however, accessing, organizing, and analyzing these data pose significant challenges. To address these challenges, this article reports development of a web-based application that provides for local clinical data capture as well as integration of patient data directly from an institutional EMR. A web-based system was created using an existing institutional application development framework. The application consists of a local clinical data repository, processes that integrate data from an EMR, and programs that enable end-user access, manual data capture, and analysis. Data are maintained in a relational database at the patient level in a time- oriented manner and by clinical data type. The application and data repository have been used to integrate and analyze a broad range of clinical data of 637 patients with burn injury. Research findings have shown that in addition to tracking clinical outcomes, laboratory data provide the ability to risk stratify patient populations to target high-risk individuals for case management and interventions. This effort validates the utility of web-based applications to collect local clinical data and integrate clinical data directly from an institutional EMR. This approach leverages institutionally collected clinical information and provides the flexibility to incorporate disparate data and accommodate system modifications as needed. Although the current efforts have focused on a cohort of patients with burn injury, the approach and system design are extendable to other patient types.