Using public clinical trial reports to probe non-experimental causal inference methods.

BACKGROUND: Non-experimental studies (also known as observational studies) are valuable for estimating the effects of various medical interventions, but are notoriously difficult to evaluate because the methods used in non-experimental studies require untestable assumptions. This lack of intrinsic verifiability makes it difficult both to compare different non-experimental study methods and to trust the results of any particular non-experimental study. METHODS: We introduce TrialProbe, a data resource and statistical framework for the evaluation of non-experimental methods. We first collect a dataset of pseudo "ground truths" about the relative effects of drugs by using empirical Bayesian techniques to analyze adverse events recorded in public clinical trial reports. We then develop a framework for evaluating non-experimental methods against that ground truth by measuring concordance between the non-experimental effect estimates and the estimates derived from clinical trials. As a demonstration of our approach, we also perform an example methods evaluation between propensity score matching, inverse propensity score weighting, and an unadjusted approach on a large national insurance claims dataset. RESULTS: From the 33,701 clinical trial records in our version of the ClinicalTrials.gov dataset, we are able to extract 12,967 unique drug/drug adverse event comparisons to form a ground truth set. During our corresponding methods evaluation, we are able to use that reference set to demonstrate that both propensity score matching and inverse propensity score weighting can produce estimates that have high concordance with clinical trial results and substantially outperform an unadjusted baseline. CONCLUSIONS: We find that TrialProbe is an effective approach for probing non-experimental study methods, being able to generate large ground truth sets that are able to distinguish how well non-experimental methods perform in real world observational data.

Principled estimation and evaluation of treatment effect heterogeneity: A case study application to dabigatran for patients with atrial fibrillation.

OBJECTIVE: To apply the latest guidance for estimating and evaluating heterogeneous treatment effects (HTEs) in an end-to-end case study of the Long-term Anticoagulation Therapy (RE-LY) trial, and summarize the main takeaways from applying state-of-the-art metalearners and novel evaluation metrics in-depth to inform their applications to personalized care in biomedical research. METHODS: Based on the characteristics of the RE-LY data, we selected four metalearners (S-learner with Lasso, X-learner with Lasso, R-learner with random survival forest and Lasso, and causal survival forest) to estimate the HTEs of dabigatran. For the outcomes of (1) stroke or systemic embolism and (2) major bleeding, we compared dabigatran 150 mg, dabigatran 110 mg, and warfarin. We assessed the overestimation of treatment heterogeneity by the metalearners via a global null analysis and their discrimination and calibration ability using two novel metrics: rank-weighted average treatment effects (RATE) and estimated calibration error for treatment heterogeneity. Finally, we visualized the relationships between estimated treatment effects and baseline covariates using partial dependence plots. RESULTS: The RATE metric suggested that either the applied metalearners had poor performance of estimating HTEs or there was no treatment heterogeneity for either the stroke/SE or major bleeding outcome of any treatment comparison. Partial dependence plots revealed that several covariates had consistent relationships with the treatment effects estimated by multiple metalearners. The applied metalearners showed differential performance across outcomes and treatment comparisons, and the X- and R-learners yielded smaller calibration errors than the others. CONCLUSIONS: HTE estimation is difficult, and a principled estimation and evaluation process is necessary to provide reliable evidence and prevent false discoveries. We have demonstrated how to choose appropriate metalearners based on specific data properties, applied them using the off-the-shelf implementation tool survlearners, and evaluated their performance using recently defined formal metrics. We suggest that clinical implications should be drawn based on the common trends across the applied metalearners.

Estimating the optimal individualized treatment rule from a cost-effectiveness perspective.

Optimal individualized treatment rules (ITRs) provide customized treatment recommendations based on subject characteristics to maximize clinical benefit in accordance with the objectives in precision medicine. As a result, there is growing interest in developing statistical tools for estimating optimal ITRs in evidence-based research. In health economic perspectives, policy makers consider the tradeoff between health gains and incremental costs of interventions to set priorities and allocate resources. However, most work on ITRs has focused on maximizing the effectiveness of treatment without considering costs. In this paper, we jointly consider the impact of effectiveness and cost on treatment decisions and define ITRs under a composite-outcome setting, so that we identify the most cost-effective ITR that accounts for individual-level heterogeneity through direct optimization. In particular, we propose a decision-tree-based statistical learning algorithm that uses a net-monetary-benefit-based reward to provide nonparametric estimations of the optimal ITR. We provide several approaches to estimating the reward underlying the ITR as a function of subject characteristics. We present the strengths and weaknesses of each approach and provide practical guidelines by comparing their performance in simulation studies. We illustrate the top-performing approach from our simulations by evaluating the projected 15-year personalized cost-effectiveness of the intensive blood pressure control of the Systolic Blood Pressure Intervention Trial (SPRINT) study.

Traction Time, Force and Postoperative Nerve Block Significantly Influence the Development and Duration of Neuropathy Following Hip Arthroscopy.

PURPOSE: To (1) evaluate the individual and combined effects of traction time and traction force on postoperative neuropathy following hip arthroscopy, (2) determine if perioperative fascia iliaca block has an effect on the risk of this neuropathy, and (3) identify if the these items had a significant association with the presence, location, and/or duration of postoperative numbness. METHODS: Between February 2015 and December 2016, a consecutive cohort of hip arthroscopy patients was prospectively enrolled. Traction time, force, and postoperative nerve block administration were recorded. The location and duration of numbness were assessed at postoperative clinic visits. Numbness location was classified into regions: 1, groin; 2, lateral thigh; 3, medial thigh; 4, dorsal foot; and 5,preoperative thigh or radiculopathic numbness. RESULTS: A total of 156 primary hip arthroscopy patients were analyzed, 99 (63%) women and 57 (37%) men. Mean traction time was 46.5 ± 20.3 minutes. Seventy-four patients (47%) reported numbness with an average duration of 157.5 ± 116.2 days. Postoperative fascia iliaca nerve block was a significant predictor of medial thigh numbness (odds ratio, 3.36; 95% confidence interval, 1.46-7.76; P = .04). Neither traction time nor force were associated with generalized numbness (P = .85 and P = .40, respectively). However, among those who experienced numbness, traction time and force were greater in patients with combined groin and lateral thigh numbness compared with those with isolated lateral thigh or medial thigh numbness (P = .001 and P = .005, respectively). CONCLUSIONS: Postoperative neuropathy is a well-documented complication following hip arthroscopy. Concomitant pudendal and lateral femoral cutaneous nerve palsy may be related to increased traction force and time, even in the setting of low intraoperative traction time (<1 hour). Isolated medial thigh numbness is significantly associated with postoperative fascia iliaca blockade. LEVEL OF EVIDENCE: IV, case series.

Variation in Adolescent Idiopathic Scoliosis Surgery: Implications for Improving Healthcare Value.

OBJECTIVES: To investigate the variation in care and cost of spinal fusion for adolescent idiopathic scoliosis (AIS), and to identify opportunities for improving healthcare value. STUDY DESIGN: Retrospective cohort study from the Pediatric Health Information Systems database, including children 11-18 years of age with AIS who underwent spinal fusion surgery between 2004 and 2015. Multivariable regression was used to evaluate the relationships between hospital cost, patient outcomes, and resource use. RESULTS: There were 16 992 cases of AIS surgery identified. There was marked variation across hospitals in rates of intensive care unit admission (0.5%-99.2%), blood transfusions (0%-100%), surgical complications (1.8%-32.3%), and total hospital costs ($31 278-$90 379). Hospital cost was 32% higher at hospitals that most frequently admitted patients to the intensive care unit (P = .009), and 8% higher for each additional 25 operative cases per hospital (P = .003). Hospital duration of stay was shorter for patients admitted to hospitals with highest intensive care unit admission rates and higher surgical volumes. There was no association between cost and duration of stay, 30-day readmission, or surgical complications. The largest contribution to hospital charges was supplies (55%). Review of a single hospital's detailed cost accounting system also found supplies to be the greatest single contributor to cost, the majority of which were for spinal implants, accounting for 39% of total hospital costs. CONCLUSIONS: The greatest contribution to AIS surgery cost was supplies, the majority of which is likely attributed to spinal implant costs. Opportunities for improving healthcare value should focus on controlling costs of spinal instrumentation, and improving quality of care with standardized treatment protocols.

Evaluation of Factors Driving Cost Variation for Distal Radius Fracture Open Reduction Internal Fixation.

PURPOSE: Distal radius fracture open reduction and internal fixation (ORIF) represents a considerable cost burden to the health care system. We aimed to elucidate demographic-, injury-, and treatment-specific factors influencing surgical encounter costs for distal radius ORIF. METHODS: We retrospectively reviewed adult patients treated with isolated distal radius ORIF between November 2014 and October 2016 at a single tertiary academic medical center. Using our institution's information technology value tools-which allow for comprehensive payment and cost data collection and analysis on an item-level basis-we determined relative costs (RC) for each factor potentially influencing total direct costs (TDC) for distal radius ORIF using univariate and multivariable gamma regression analyses. RESULTS: Of the included 108 patients, implants and facility utilization costs were responsible for 48.3% and 37.9% of TDC, respectively. Factors associated with increased TDC include plate manufacturer (RC 1.52 for the most vs least expensive manufacturer), number of screws (RC 1.03 per screw) and distal radius plates used (RC 1.67 per additional plate), surgery setting (RC 1.32 for main hospital vs ambulatory surgery center), treating service (RC 1.40 for trauma vs hand surgeons), and surgical time (RC 1.04 for every 10 min of additional surgical time). Open fracture was associated with increased costs (RC 1.55 vs closed fracture), whereas other estimates of fracture severity were nonsignificant. In the multivariable model controlling for injury-specific factors, variables including implant manufacturer, and number of distal radius plates and screws used, remained as significant drivers of TDC. CONCLUSIONS: Substantial variations in surgical direct costs for distal radius ORIF exist, and implant choice is the predominant driver. Cost reductions may be expected through judicious use of additional plates and screws, if hospital systems use bargaining power to reduce implant costs, and by efficiently completing surgeries. CLINICAL RELEVANCE: This study identifies modifiable factors that may lead to cost reduction for distal radius ORIF.

Cost Implications of Varying the Surgical Technique, Surgical Setting, and Anesthesia Type for Carpal Tunnel Release Surgery.

PURPOSE: Carpal tunnel release (CTR) is a common surgical procedure, representing a financial burden to the health care system. The purpose of this study was to test whether the choice of CTR technique (open carpal tunnel release [OCTR] vs endoscopic carpal tunnel release [ECTR]), surgical setting (operating room vs procedure room [PR]), and anesthetic type (local, monitored anesthesia care [MAC], Bier block, general) affected costs or payments. METHODS: Consecutive adult patients undergoing isolated unilateral CTR between July 2014, and October 2017, at a single academic medical center were identified. Patients undergoing ECTR converted to OCTR, revision surgery, or additional procedures were excluded. Using our institution's information technology value tools, we calculated total direct costs (TDCs), total combined payment (TCP), hospital payment, surgeon payment, and anesthesia payment for each surgical encounter. Cost data were normalized using each participant's surgical encounter cost divided by the average cost in the data set and compared across 8 groups (defined by surgery type, operation location, and anesthesia type). RESULTS: Of 479 included patients, the mean age was 55.3 ± 16.1 years, and 68% were female. Payer mix included commercial (45%), Medicare (37%), Medicaid (13%), workers' compensation (2%), self-pay (1%), and other (3%) insurance types. The TDC and TCP both differed significantly between each CTR group, and OCTR in the PR under local anesthesia was the lowest. The OCTR/local/operating room, OCTR/MAC/operating room, and ECTR/operating room, were associated with 6.3-fold, 11.0-fold, and 12.4-16.6-fold greater TDC than OCTR/local/PR, respectively. CONCLUSIONS: Performing OCTR under local anesthetic in the PR setting significantly minimizes direct surgical encounter costs relative to other surgical methods (ECTR), anesthetic methods (Bier block, MAC, general), and surgical settings (operating room). CLINICAL RELEVANCE: This study identifies modifiable factors that may lead to cost reductions for CTR surgery.

Operative Time and Flap Failure in Unilateral and Bilateral Free Flap Breast Reconstruction.

BACKGROUND: There is an increasing trend toward bilateral breast reconstruction. Using the National Surgical Quality Improvement Program (NSQIP) database, we sought to understand the association between unilateral and bilateral free flap breast reconstruction and operative time and flap failure. METHODS: We selected a cohort of patients undergoing free flap breast reconstruction using the 2005 to 2010 NSQIP database. Cases were divided into unilateral and bilateral reconstruction. Subgroup analyses were performed dividing cases into delayed and immediate reconstruction. The effect of patient characteristics including age, body mass index (BMI), history of diabetes, and the American Society of Anesthesiologists' classification on operative time and flap failure was examined using univariable and multivariable regression models. Rates and odds ratios (OR) were reported using the multivariable gamma and logistic regression models, respectively. RESULTS: There were 691 free flap breast reconstructions performed in the cohort and 29.1% were bilateral cases. There was a 78-minute increase in the median operative time when comparing unilateral and bilateral reconstruction (p = 0.005). Patients undergoing bilateral reconstructions were generally younger and had fewer comorbidities compared with unilateral reconstructions. There was no significant association between bilateral reconstruction and flap failure. Immediate bilateral reconstructions had a significant increase in median operative time compared with immediate unilateral reconstructions (563 versus 480 minutes, p = 0.002) but no significant increase in operative time was noted when comparing delayed unilateral and delayed bilateral reconstructions. Prolonged operative time was associated with flap failure after adjusting for age and BMI (OR 1.17, p < 0.001). CONCLUSIONS: Bilateral free flap breast reconstruction can be performed safely despite an increase in operative time when compared with unilateral reconstruction.

Impact of removed tumor volume and location on patient outcome in glioblastoma.

Glioblastoma is an aggressive primary brain tumor with devastatingly poor prognosis. Multiple studies have shown the benefit of wider extent of resection (EOR) on patient overall survival (OS) and worsened survival with larger preoperative tumor volumes. However, the concomitant impact of postoperative tumor volume and eloquent location on OS has yet to be fully evaluated. We performed a retrospective chart review of adult patients treated for glioblastoma from January 2006 through December 2011. Adherence to standardized postoperative chemoradiation protocols was used as an inclusion criterion. Detailed volumetric and location analysis was performed on immediate preoperative and immediate postoperative magnetic resonance imaging. Cox proportional hazard modeling approach was employed to explore the modifying effects of EOR and eloquent location after adjusting for various confounders and associated characteristics, such as preoperative tumor volume and demographics. Of the 471 screened patients, 141 were excluded because they did not meet all inclusion criteria. The mean (±SD) age of the remaining 330 patients (60.6% male) was 58.9 ± 12.9 years; the mean preoperative and postoperative Karnofsky performance scores (KPSs) were 76.2 ± 10.3 and 80.0 ± 16.6, respectively. Preoperative tumor volume averaged 33.2 ± 29.0 ml, postoperative residual was 4.0 ± 8.1 ml, and average EOR was 88.6 ± 17.6%. The observed average follow-up was 17.6 ± 15.7 months, and mean OS was 16.7 ± 14.4 months. Survival analysis showed significantly shorter survival for patients with lesions in periventricular (16.8 ± 1.7 vs. 21.5 ± 1.4 mo, p = 0.03), deep nuclei/basal ganglia (11.6 ± 1.7 vs. 20.6 ± 1.2, p = 0.002), and multifocal (12.0 ± 1.4 vs. 21.3 ± 1.3 months, p = 0.0001) locations, but no significant influence on survival was seen for eloquent cortex sites (p = 0.14, range 0.07-0.9 for all individual locations). OS significantly improved with EOR in univariate analysis, averaging 22.3, 19.7, and 13.2 months for >90, 80-90, and 70-80% resection, respectively. Survival was 22.8, 19.0, and 12.7 months for 0, 0-5, and 5-10 ml postoperative residual, respectively. A hazard model showed that larger preoperative tumor volume [hazard ratio (HR) 1.05, 95% CI 1.02-1.07], greater age (HR 1.02, 95% CI 1.01-1.03), multifocality (HR 1.44, 95% CI 1.01-2.04), and deep nuclei/basal ganglia (HR 2.05, CI 1.27-3.3) were the most predictive of poor survival after adjusting for KPS and tumor location. There was a negligible but significant interaction between EOR and preoperative tumor volume (HR 0.9995, 95% CI 0.9993-0.9998), but EOR alone did not correlate with OS after adjusting for other factors. The interaction between EOR and preoperative tumor volume represented tumor volume removed during surgery. In conclusion, EOR alone was not an important predictor of outcome during GBM treatment once preoperative tumor volume, age, and deep nuclei/basal ganglia location were factored. Instead, the interaction between EOR and preoperative volume, representing reduced disease burden, was an important predictor of reducing OS. Removal of tumor from eloquent cortex did not impact postoperative KPS. These results suggest aggressive surgical treatment to reduce postoperative residual while maintaining postoperative KPS may aid patient survival outcomes for a given tumor size and location.

Meta-analysis of the effect of different exercise modalities in the prevention and treatment of perinatal depression.

BACKGROUND: The incidence of perinatal depression is increasing and has become a global public health problem to be addressed. OBJECTIVE: To explore the prevention and treatment effects of different exercise methods on perinatal depression. METHODS: A meta-analysis was conducted by searching databases for published "exercise interventions for perinatal depression "related randomized controlled trials, up to July 20, 2022. RESULTS: 48 randomized controlled trials were included, with a total of 5282 pregnant women. (1) Exercise prevention of prenatal depression has a low effective stress intervention effect, ranking from high to low as yoga, aerobic+resistance. (2) Exercise therapy for prenatal depression has a significant intervention effect, followed by gymnastics, pelvic floor muscle training, aerobic exercise, aerobic+resistance, and yoga. (3) Exercise prevention of postpartum depression has a low effective intervention effect, followed by yoga, aerobic exercise, aerobic+resistance, and gymnastics. (4) Exercise has a moderate equivalent stress intervention effect on treating postpartum depression, followed by aerobic exercise, water exercise, yoga, fertility dance, and stroller walking. LIMITATIONS: Due to the small number of included literature on single exercise modalities, and maternity is a special population, most of the trial procedures included in the text were not blinded, which has a certain risk of bias and affects the accuracy of the Meta-analysis results. CONCLUSIONS: The therapeutic effect of exercise in the prevention and treatment of perinatal depression is superior to the preventive effect, and the effect of prenatal prevention and treatment is better than that of postpartum, with a moderate effect.

Tough polyvinyl alcohol-gelatin biological macromolecules ionic hydrogel temperature, humidity, stress and strain, sensors.

High strength, high toughness and high sensitivity were some of the most popular characteristics of flexible sensors. However, the mechanical properties and reproducibility of current single biomacromolecule gelatin hydrogel sensors are lower, and few hydrogel sensors have been able to provide excellent mechanical properties and flexibility at the same time so far. To address this challenge, a simple method to prepare tough polyvinyl alcohol (PVA) and gelatin hydrogel was proposed in this study. The PVA-gelatin-Fe3+ biological macromolecules hydrogel was prepared by a freeze-casting-assisted solution substitution method, which exhibited high strength (2.5 MPa), toughness (7.22 MJ m-3), and excellent temperature, humidity, stress, strain, and human motion sensing properties. This combination of mechanical properties and flexibility makes PVA-gelatin biological macromolecules hydrogel a promising material for flexible sensing. In addition, an ionic immersion strategy could also impart multiple functions to the hydrogel and be applied to various hydrogel sensor materials. Thus, this work provided an all-around solution for the preparation of advanced and robust sensors with good application prospects.

Clinical utility gains from incorporating comorbidity and geographic location information into risk estimation equations for atherosclerotic cardiovascular disease.

OBJECTIVE: There are over 363 customized risk models of the American College of Cardiology and the American Heart Association (ACC/AHA) pooled cohort equations (PCE) in the literature, but their gains in clinical utility are rarely evaluated. We build new risk models for patients with specific comorbidities and geographic locations and evaluate whether performance improvements translate to gains in clinical utility. MATERIALS AND METHODS: We retrain a baseline PCE using the ACC/AHA PCE variables and revise it to incorporate subject-level information of geographic location and 2 comorbidity conditions. We apply fixed effects, random effects, and extreme gradient boosting (XGB) models to handle the correlation and heterogeneity induced by locations. Models are trained using 2 464 522 claims records from Optum©'s Clinformatics® Data Mart and validated in the hold-out set (N = 1 056 224). We evaluate models' performance overall and across subgroups defined by the presence or absence of chronic kidney disease (CKD) or rheumatoid arthritis (RA) and geographic locations. We evaluate models' expected utility using net benefit and models' statistical properties using several discrimination and calibration metrics. RESULTS: The revised fixed effects and XGB models yielded improved discrimination, compared to baseline PCE, overall and in all comorbidity subgroups. XGB improved calibration for the subgroups with CKD or RA. However, the gains in net benefit are negligible, especially under low exchange rates. CONCLUSIONS: Common approaches to revising risk calculators incorporating extra information or applying flexible models may enhance statistical performance; however, such improvement does not necessarily translate to higher clinical utility. Thus, we recommend future works to quantify the consequences of using risk calculators to guide clinical decisions.

The shaky foundations of large language models and foundation models for electronic health records.

The success of foundation models such as ChatGPT and AlphaFold has spurred significant interest in building similar models for electronic medical records (EMRs) to improve patient care and hospital operations. However, recent hype has obscured critical gaps in our understanding of these models' capabilities. In this narrative review, we examine 84 foundation models trained on non-imaging EMR data (i.e., clinical text and/or structured data) and create a taxonomy delineating their architectures, training data, and potential use cases. We find that most models are trained on small, narrowly-scoped clinical datasets (e.g., MIMIC-III) or broad, public biomedical corpora (e.g., PubMed) and are evaluated on tasks that do not provide meaningful insights on their usefulness to health systems. Considering these findings, we propose an improved evaluation framework for measuring the benefits of clinical foundation models that is more closely grounded to metrics that matter in healthcare.

A comparison of approaches to improve worst-case predictive model performance over patient subpopulations.

Predictive models for clinical outcomes that are accurate on average in a patient population may underperform drastically for some subpopulations, potentially introducing or reinforcing inequities in care access and quality. Model training approaches that aim to maximize worst-case model performance across subpopulations, such as distributionally robust optimization (DRO), attempt to address this problem without introducing additional harms. We conduct a large-scale empirical study of DRO and several variations of standard learning procedures to identify approaches for model development and selection that consistently improve disaggregated and worst-case performance over subpopulations compared to standard approaches for learning predictive models from electronic health records data. In the course of our evaluation, we introduce an extension to DRO approaches that allows for specification of the metric used to assess worst-case performance. We conduct the analysis for models that predict in-hospital mortality, prolonged length of stay, and 30-day readmission for inpatient admissions, and predict in-hospital mortality using intensive care data. We find that, with relatively few exceptions, no approach performs better, for each patient subpopulation examined, than standard learning procedures using the entire training dataset. These results imply that when it is of interest to improve model performance for patient subpopulations beyond what can be achieved with standard practices, it may be necessary to do so via data collection techniques that increase the effective sample size or reduce the level of noise in the prediction problem.

An efficient approach for optimizing the cost-effective individualized treatment rule using conditional random forest.

Evidence from observational studies has become increasingly important for supporting healthcare policy making via cost-effectiveness analyses. Similar as in comparative effectiveness studies, health economic evaluations that consider subject-level heterogeneity produce individualized treatment rules that are often more cost-effective than one-size-fits-all treatment. Thus, it is of great interest to develop statistical tools for learning such a cost-effective individualized treatment rule under the causal inference framework that allows proper handling of potential confounding and can be applied to both trials and observational studies. In this paper, we use the concept of net-monetary-benefit to assess the trade-off between health benefits and related costs. We estimate cost-effective individualized treatment rule as a function of patients' characteristics that, when implemented, optimizes the allocation of limited healthcare resources by maximizing health gains while minimizing treatment-related costs. We employ the conditional random forest approach and identify the optimal cost-effective individualized treatment rule using net-monetary-benefit-based classification algorithms, where two partitioned estimators are proposed for the subject-specific weights to effectively incorporate information from censored individuals. We conduct simulation studies to evaluate the performance of our proposals. We apply our top-performing algorithm to the NIH-funded Systolic Blood Pressure Intervention Trial to illustrate the cost-effectiveness gains of assigning customized intensive blood pressure therapy.

Considerations in the reliability and fairness audits of predictive models for advance care planning.

Multiple reporting guidelines for artificial intelligence (AI) models in healthcare recommend that models be audited for reliability and fairness. However, there is a gap of operational guidance for performing reliability and fairness audits in practice. Following guideline recommendations, we conducted a reliability audit of two models based on model performance and calibration as well as a fairness audit based on summary statistics, subgroup performance and subgroup calibration. We assessed the Epic End-of-Life (EOL) Index model and an internally developed Stanford Hospital Medicine (HM) Advance Care Planning (ACP) model in 3 practice settings: Primary Care, Inpatient Oncology and Hospital Medicine, using clinicians' answers to the surprise question ("Would you be surprised if [patient X] passed away in [Y years]?") as a surrogate outcome. For performance, the models had positive predictive value (PPV) at or above 0.76 in all settings. In Hospital Medicine and Inpatient Oncology, the Stanford HM ACP model had higher sensitivity (0.69, 0.89 respectively) than the EOL model (0.20, 0.27), and better calibration (O/E 1.5, 1.7) than the EOL model (O/E 2.5, 3.0). The Epic EOL model flagged fewer patients (11%, 21% respectively) than the Stanford HM ACP model (38%, 75%). There were no differences in performance and calibration by sex. Both models had lower sensitivity in Hispanic/Latino male patients with Race listed as "Other." 10 clinicians were surveyed after a presentation summarizing the audit. 10/10 reported that summary statistics, overall performance, and subgroup performance would affect their decision to use the model to guide care; 9/10 said the same for overall and subgroup calibration. The most commonly identified barriers for routinely conducting such reliability and fairness audits were poor demographic data quality and lack of data access. This audit required 115 person-hours across 8-10 months. Our recommendations for performing reliability and fairness audits include verifying data validity, analyzing model performance on intersectional subgroups, and collecting clinician-patient linkages as necessary for label generation by clinicians. Those responsible for AI models should require such audits before model deployment and mediate between model auditors and impacted stakeholders.

In vivo and in vitro anti-inflammatory effects of a novel derivative of icariin.

Icariin is the major active constituent of Epimedii Herba. Our recent study showed that icariin displayed anti-inflammatory potential. One novel derivate of icariin is 3,5,7-Trihydroxy-4'-methoxy-8-(3-hydroxy-3-methylbutyl)-flavone (ICT). Little is known about ICT's pharmacological activities. In our study, the anti-inflammatory properties of ICT were evaluated. Murine RAW264.7 cells and C57BL/6J mice stimulated by lipopolysaccharide (LPS) was used as in vitro and in vivo inflammatory model, respectively. Our data showed that ICT (1-100 µg/mL) significantly inhibited LPS-induced tumor necrosis factor-α (TNF-α), nitric oxide (NO), prostaglandin E2 (PGE2) production in vitro. These effects did not depend on cytotoxicity. The in vivo assay displayed that pretreatment of C57BL/6J mice with ICT (25-100 mg/kg, by gavage) for 3 days decreased LPS-induced serum levels of TNF-α, PGE2, and neutrophils CD11b expression dose-dependently. Furthermore, our data suggested that ICT reduced NO and PGE2 levels by inhibiting inducible NO synthase and cyclooxygenase-2 protein expression. To our knowledge, it is the first time that the anti-inflammatory effects of ICT have been evaluated.

Does Non-Neurologic Multiorgan Dysfunction After Out-of-Hospital Cardiac Arrest among Children Admitted in Coma Predict Outcome 1 Year Later?

In this article, we investigated whether non-neurologic multiorgan dysfunction syndrome (MODS) following out-of-hospital cardiac arrest (OHCA) predicts poor 12-month survival. We conducted a secondary data analysis of therapeutic hypothermia after pediatric cardiac arrest out-of-hospital randomized trial involving children who remained unconscious and intubated after OHCA ( n = 237). Associations between MODS and 12-month outcomes were assessed using multivariable logistic regression. Non-neurologic MODS was present in 95% of patients and sensitive (97%; 95% confidence interval [CI]: 93-99%) for 12-month survival but had poor specificity (10%; 95% CI: 4-21%). Development of non-neurologic MODS is not helpful to predict long-term neurologic outcome or survival after OHCA.

Patient Selection for Intensive Blood Pressure Management Based on Benefit and Adverse Events.

BACKGROUND: Intensive systolic blood pressure (SBP) treatment prevents cardiovascular disease (CVD) events in patients with high CVD risk on average, though benefits likely vary among patients. OBJECTIVES: The aim of this study was to predict the magnitude of benefit (reduced CVD and all-cause mortality risk) along with adverse event (AE) risk from intensive versus standard SBP treatment. METHODS: This was a secondary analysis of SPRINT (Systolic Blood Pressure Intervention Trial). Separate benefit outcomes were the first occurrence of: 1) a CVD composite of acute myocardial infarction or other acute coronary syndrome, stroke, heart failure, or CVD death; and 2) all-cause mortality. Treatment-related AEs of interest included hypotension, syncope, bradycardia, electrolyte abnormalities, injurious falls, and acute kidney injury. Modified elastic net Cox regression was used to predict absolute risk for each outcome and absolute risk differences on the basis of 36 baseline variables available at the point of care with intensive versus standard treatment. RESULTS: Among 8,828 SPRINT participants (mean age 67.9 years, 35% women), 600 CVD composite events, 363 all-cause deaths, and 481 treatment-related AEs occurred over a median follow-up period of 3.26 years. Individual participant risks were predicted for the CVD composite (C index = 0.71), all-cause mortality (C index = 0.75), and treatment-related AEs (C index = 0.69). Higher baseline CVD risk was associated with greater benefit (i.e., larger absolute CVD risk reduction). Predicted CVD benefit and predicted increased treatment-related AE risk were correlated (Spearman correlation coefficient = -0.72), and 95% of participants who fell into the highest tertile of predicted benefit also had high or moderate predicted increases in treatment-related AE risk. Few were predicted as high benefit with low AE risk (1.8%) or low benefit with high AE risk (1.5%). Similar results were obtained for all-cause mortality. CONCLUSIONS: SPRINT participants with higher baseline predicted CVD risk gained greater absolute benefit from intensive treatment. Participants with high predicted benefit were also most likely to experience treatment-related AEs, but AEs were generally mild and transient. Patients should be prioritized for intensive SBP treatment on the basis of higher predicted benefit. (Systolic Blood Pressure Intervention Trial [SPRINT]; NCT01206062).

Tonifying kidney therapy for stable chronic obstructive pulmonary disease: a systematic review.

OBJECTIVE: To evaluate the efficacy and safety of tonifying kidney therapy (Bushen, TK) for stable chronic obstructive pulmonary disease (COPD). METHODS: Randomized controlled trials (RCTs) of TK use for treatment of stable COPD were searched in four databases including PubMed, the Cochrane Library, Chinese Biomedical Literature Database, and China National Knowledge Infrastructure Database from inception to December 2017. Two reviewers independently screened the literature, extracted the data, and assessed the risk of bias in the included studies. RevMan 5.3 software was used for the Meta-analysis. RESULTS: Eight RCTs involving 809 patients with stable COPD were included. Compared with the conventional Western Medicine (CWM) group, the TK group (TK combined with CWM) showed significant improvements in the effectiveness rates (RR = 1.37, 95% CI 1.22 to 1.53, P < 0.000 01) and 6-min walk distance in meters (MD 11.92, 95% CI 3.52 to 20.32, P = 0.005), this study also showed that the TK group can decrease The Traditional Chinese Medicine Syndrome Score (MD -8.01, 95% CI －12.89 to －3.13, P = 0.001). The lung function [forced expiratory volume in one second% (FEV1%), FEV1/forced vital capacity] showed no difference between the TK and control groups. CONCLUSION: For patients with stable COPD, TK can improve the clinical effectiveness and exercise capacity but fail to improve the patient's symptoms. Because of the low methodological quality of the included trials, additional high-quality and large-scale RCTs are required.