Assessment of Natural Language Processing of Electronic Health Records to Measure Goals-of-Care Discussions as a Clinical Trial Outcome.

Importance: Many clinical trial outcomes are documented in free-text electronic health records (EHRs), making manual data collection costly and infeasible at scale. Natural language processing (NLP) is a promising approach for measuring such outcomes efficiently, but ignoring NLP-related misclassification may lead to underpowered studies. Objective: To evaluate the performance, feasibility, and power implications of using NLP to measure the primary outcome of EHR-documented goals-of-care discussions in a pragmatic randomized clinical trial of a communication intervention. Design, Setting, and Participants: This diagnostic study compared the performance, feasibility, and power implications of measuring EHR-documented goals-of-care discussions using 3 approaches: (1) deep-learning NLP, (2) NLP-screened human abstraction (manual verification of NLP-positive records), and (3) conventional manual abstraction. The study included hospitalized patients aged 55 years or older with serious illness enrolled between April 23, 2020, and March 26, 2021, in a pragmatic randomized clinical trial of a communication intervention in a multihospital US academic health system. Main Outcomes and Measures: Main outcomes were natural language processing performance characteristics, human abstractor-hours, and misclassification-adjusted statistical power of methods of measuring clinician-documented goals-of-care discussions. Performance of NLP was evaluated with receiver operating characteristic (ROC) curves and precision-recall (PR) analyses and examined the effects of misclassification on power using mathematical substitution and Monte Carlo simulation. Results: A total of 2512 trial participants (mean [SD] age, 71.7 [10.8] years; 1456 [58%] female) amassed 44 324 clinical notes during 30-day follow-up. In a validation sample of 159 participants, deep-learning NLP trained on a separate training data set identified patients with documented goals-of-care discussions with moderate accuracy (maximal F1 score, 0.82; area under the ROC curve, 0.924; area under the PR curve, 0.879). Manual abstraction of the outcome from the trial data set would require an estimated 2000 abstractor-hours and would power the trial to detect a risk difference of 5.4% (assuming 33.5% control-arm prevalence, 80% power, and 2-sided α = .05). Measuring the outcome by NLP alone would power the trial to detect a risk difference of 7.6%. Measuring the outcome by NLP-screened human abstraction would require 34.3 abstractor-hours to achieve estimated sensitivity of 92.6% and would power the trial to detect a risk difference of 5.7%. Monte Carlo simulations corroborated misclassification-adjusted power calculations. Conclusions and Relevance: In this diagnostic study, deep-learning NLP and NLP-screened human abstraction had favorable characteristics for measuring an EHR outcome at scale. Adjusted power calculations accurately quantified power loss from NLP-related misclassification, suggesting that incorporation of this approach into the design of studies using NLP would be beneficial.

Crystal Structure of an i-Motif from the HRAS Oncogene Promoter.

An i-motif is a non-canonical DNA structure implicated in gene regulation and linked to cancers. The C-rich strand of the HRAS oncogene, 5'-CGCCCGTGCCCTGCGCCCGCAACCCGA-3' (herein referred to as iHRAS), forms an i-motif in vitro but its exact structure was unknown. HRAS is a member of the RAS proto-oncogene family. About 19 % of US cancer patients carry mutations in RAS genes. We solved the structure of iHRAS at 1.77 Šresolution. The structure reveals that iHRAS folds into a double hairpin. The two double hairpins associate in an antiparallel fashion, forming an i-motif dimer capped by two loops on each end and linked by a connecting region. Six C-C+ base pairs form each i-motif core, and the core regions are extended by a G-G base pair and a cytosine stacking. Extensive canonical and non-canonical base pairing and stacking stabilizes the connecting region and loops. The iHRAS structure is the first atomic resolution structure of an i-motif from a human oncogene. This structure sheds light on i-motifs folding and function in the cell.

Predictors of mortality, limb loss, and discharge disposition at admission among patients with necrotizing skin and soft tissue infections.

BACKGROUND: Necrotizing soft tissue infections (NSTI) represent a heterogeneous group of rapidly progressive skin and soft tissue infections associated with significant morbidity and mortality. Efforts to identify factors associated with death have produced mixed results, and little or no data is available for other adverse outcomes. We sought to determine whether admission variables were associated with mortality, limb loss, and discharge disposition in patients with NSTI. METHODS: We analyzed prospectively collected data of adult patients with surgically confirmed NSTI from an NSTI registry maintained at a quaternary referral center. Factors independently associated with mortality, amputation, and skilled nursing facility discharge were identified using logistic regression. RESULTS: Between 2015 and 2018, 446 patients were identified. The median age was 55 years (interquartile range, 43-62). The majority of patients were male (65%), white (77%), and transferred from another facility (90%). The perineum was most commonly involved (37%), followed by the lower extremity (34%). The median number of operative debridements was 3 (interquartile range, 2-4). Overall mortality was 15%, and 21% of extremity NSTI patients required amputation. Age greater than 60 years; creatinine greater than 2 mg/dL; white blood cell count greater than 30 x 10^ /µl, platelets less than 150 × 10/µL, and clostridial involvement were independently associated with greater odds of death; perineal involvement was associated with lower odds of death. Age greater than 60 years; sex, male; nonwhite race; diabetes; chronic wound as etiology; leg involvement; transfer status; and sodium, less than 130 mEq/L were independently associated with amputation. Age greater than 60 years; sex, female; nonwhite race; perineal involvement; and amputation were associated with skilled care facility discharge. CONCLUSION: Necrotizing soft tissue infections are a heterogeneous group of infections involving significantly different patient populations with different outcomes; efforts to differentiate and predict adverse outcomes in NSTI should include laboratory data, comorbidities, infection site, and/or etiology to improve predictions and better account for this heterogeneity. LEVEL OF EVIDENCE: Prognostic, Level III.