BACKGROUND:The relevance of covert cerebrovascular disease (CCD) in practice is uncertain, partly because estimation of risk in whole clinical populations is difficult. Studies have had success extracting CCD from clinical text using natural language processing (NLP), though they have been limited to specific CCD phenotypes. Here, we used NLP to measure multiple clinically-reported CCD phenotypes in a large clinical cohort and estimated subsequent disease risk in health record data. METHODS:From all people with brain imaging in Scotland (2010-2018), we selected people with no prior hospitalisation for neurological disease (n=367 988). NLP of imaging reports identified: white matter hypoattenuation or hyperintensities (WMH), lacunes, cortical infarcts and cerebral atrophy. Adjusted HRs (aHRs) were estimated between each phenotype and stroke, dementia and Parkinson's disease (conditions previously associated with CCD), epilepsy and colorectal cancer (control conditions). RESULTS:For each phenotype, the aHR of stroke was WMH 1.4 (95% CI 1.3-1.4), lacunes 1.6 (1.5-1.6), cortical infarct 1.8 (1.7-1.9) and cerebral atrophy 1.1 (1.0-1.1). The aHR of dementia was WMH 1.3 (1.3-1.3), lacunes 1.0 (0.9-1.0), cortical infarct 1.1 (1.1-1.2) and cerebral atrophy 1.7 (1.7-1.8). The aHR of Parkinson's disease was WMH 1.1 (1.0-1.2), lacunes 1.1 (0.9-1.2), cortical infarct 0.7 (0.6-0.9) and cerebral atrophy 1.4 (1.3-1.5). The aHRs between CCD phenotypes and epilepsy and colorectal cancer were around the null. CONCLUSION:CCD and atrophy have implications for future disease risk and can be identified at scale using NLP of clinical reports. Prevention of neurological disease in people with CCD should be a priority for healthcare policy makers.

Importance:Patent foramen ovale (PFO) closure decreases recurrent stroke but increases atrial fibrillation (AF). Careful selection of patients in whom PFO is more likely to be the cause of stroke may improve outcomes by avoiding closure in patients unlikely to benefit. Objective:To determine whether the PFO-Associated Stroke Causal Likelihood (PASCAL) classification system identifies who will experience net benefit and net harm from PFO closure. Design, Setting, and Participants:This meta-analysis was a secondary analysis of individual participant-level data from the Systematic, Collaborative, PFO Closure Evaluation (SCOPE) consortium meta-analysis, including all 6 randomized trials of transcatheter PFO closure vs antithrombotic therapy alone. Participants were young and middle-aged adults (mean [SD] age, 45 [10] years) with a PFO and an otherwise cryptogenic stroke. The trials were conducted in hospitals in North America, Europe, Australia, Brazil, and South Korea from 2000 to 2017. The current analysis, involving all trial participants, was performed from January to August 2025. Interventions:Transcatheter PFO closure plus antithrombotic therapy vs antithrombotic therapy alone. Main Outcomes and Measures:The primary efficacy end point was recurrent ischemic stroke. The primary safety end point was first-ever detection of AF beyond the periprocedural period (>45 days after randomization). Results:The 6 trials enrolled 3740 patients (1889 who had PFO closure, 1851 who had medical therapy); 2058 patients (55.0%) were male, and 1682 (45.0%) were female. Among patients in all 6 trials, PASCAL classified PFO relatedness to the index stroke as probable in 1382 patients (37.0%), possible in 1811 (48.4%), and unlikely in 547 (14.6%); among the 2967 patients in the 4 trials with broad entry criteria, PASCAL classified PFO relatedness as probable in 860 patients (29.0%), possible in 1565 (52.7%), and unlikely in 543 (18.3%). The reduction in the absolute rate of recurrent ischemic strokes over 5 years was greater than the increase in first-ever detection of AF in the postperiprocedural period as follows: in the probable group, fewer strokes, -2.5% (95% CI, -4.2% to -1.3%) vs more late AF, 1.3% (95% CI, 0.0% to 2.5%), and in the possible group, fewer strokes -3.4% (95% CI, -5.4% to -1.3%) vs more late AF, 1.1% (95% CI, -0.5% to 2.6%). Reduction in recurrent ischemic strokes was not observed and increase in first-ever detected postperiprocedural AF was magnified in the unlikely group (more strokes, 0.4%; 95% CI, -4.0% to 4.8%, vs more late AF, 4.6%; 95% CI, 0.3% to 8.9%). Conclusion and Relevance:Among young and middle-aged patients with PFO and otherwise cryptogenic stroke, the PASCAL classification algorithm distinguished the 4 of every 5 patients in the probable and possible groups with net benefit and the 1 of every 5 patients in the unlikely group with net harm from closure.

This Viewpoint examines adversarial collaboration—bringing investigators with opposing views to design, analyze, and publish studies together, often with a neutral arbiter—as a strategy for credible biomedical science.

ABSTRACT Background To assess acceptability, feasibility, and effectiveness of incorporating individualized risk prediction into clinical assessment, decision making, and communication of risk of type 2 diabetes, with and without preventive interventions, in patients with prediabetes. Methods We integrated a prediction model into the clinical workflow at a U.S. health care organization. We conducted patient and provider focus groups and pre‐ and post‐dissemination surveys among 2500 patients with prediabetes who had primary care visits between May 2018 and December 2019. We compared rates of progression to type 2 diabetes at 3 years between the intervention group and a propensity score‐matched cohort of patients who received usual care. Results Prior to implementing the predictive model, 41.6% of providers and 63.8% of patients felt confident or very confident in their ability to estimate the risk of progression to diabetes for individual patients. After personalized risk information was made available, this increased to 92.8% for providers and 66.9% for patients. People with prediabetes who had a primary care visit where their care team had access to personal risk of developing type 2 diabetes assessed by the EHR‐based prediction model were significantly less likely to progress to diabetes within 3 years, compared to a propensity‐score‐matched cohort who received usual care in the same health system without individualized risk estimates (19.5% vs. 27.6%, p = 0.042). Conclusions Used at the point of care during a primary care visit, the EHR‐based diabetes risk calculator helped providers prioritize patients for diabetes preventive interventions, facilitated communication, and improved health outcomes among patients with prediabetes.

The Epic Sepsis Model version 2 (ESMv2) is a prediction model embedded into the electronic medical record used to warn clinicians which hospitalized patients are at risk for sepsis. We conducted a retrospective cohort study of 31,951 hospitalizations of 25,760 patients to compare analyses conducted at the commonly used patient-level (where a maximum prediction prior to the onset of sepsis is used to measure performance) vs novel prediction-level (where each prediction is used to measure performance). Sepsis, defined by the Sepsis 3 criteria occurred during 1,049 hospitalizations (3.3%). Patient-level analyses suggested excellent discrimination AUC 0.86; [IQR 0.85, 0.87], whereas prediction-level analyses demonstrated lower performance AUC 0.62; [IQR 0.57, 0.65]. Low estimates of the positive predictive value (14.5% at the patient level vs 4% at the prediction level) imply a high number of false alerts. Common evaluation approaches may overstate the performance of dynamic prediction models and mislead clinical decision-making.

BACKGROUND AND OBJECTIVES:Studies of covert cerebrovascular disease (CCD) typically rely on MRI screening, whereas CT is the predominant imaging modality in routine care. We aimed to determine the diagnostic agreement of clinically interpreted CT and MRI for incidentally discovered CCD and to compare the modality-specific prognostic implications of these findings. METHODS:We conducted a retrospective cohort study within Kaiser Permanente Southern California. Adults aged ≥50 years without previous stroke or dementia who underwent both head CT and brain MRI within 30 days (2009-2022) were included. Natural language processing was used to identify covert brain infarction (CBI) and white matter disease (WMD) and to grade WMD severity from reports. The primary outcome was incident ischemic stroke or dementia. Associations between modality-specific CCD findings and outcomes were estimated using Cox proportional hazards models adjusted for age, sex, race/ethnicity, vascular risk factors, and dementia risk factors. RESULTS:Among 18,628 participants (mean age 64.9 years; 59.1% female), CBI prevalence was similar on CT (6.3%) and MRI (6.1%), with modest agreement (κ = 0.27). WMD was more frequently reported on MRI (60.5%) than CT (24.4%), with modest agreement (κ = 0.23). Nearly half of the patients with graded WMD showed discordant severity across modalities, most often (92%) with higher severity on MRI. Over a mean 4.4-year follow-up, incidence rates of stroke or dementia per 1,000 person-years were 12.7 (95% CI 11.5-14.0) for patients negative on both modalities, 22.6 (21.0-24.2) for WMD reported on MRI only, 37.0 (29.8-45.6) for WMD reported on CT only, and 52.2 (48.7-56.0) for WMD reported on both modalities. Compared with patients negative on both modalities, adjusted hazard ratios were 1.23 (95% CI 1.07-1.41) for MRI-only WMD and 1.82 (1.58-2.11) for WMD reported on both modalities. DISCUSSION:MRI detects WMD more frequently and assigns higher severity than CT; however, WMD detected on CT identifies a subgroup at substantially higher risk than those with MRI-only findings. These findings highlight the importance of modality-specific interpretation of CCD in clinical practice and research. Limitations include reliance on radiology reports rather than direct image review and the selected population undergoing both imaging modalities.

Objectives/Goals: Trainees in clinical and translational science (CTS) take courses in biostatistics, epidemiology, and other quantitative areas. To be most successful, trainees require competency in algebra. We developed a quantitative assessment and study guide to assess trainee’s quantitative skills and provide review material to address weaknesses. Methods/Study Population: The Tufts CTS Graduate Program is the training core of the Tufts CTSI and its associated pre- and post-doctoral T32 awards. Approximately 10 trainees with a range of backgrounds (e.g., physicians, medical students, master’s-level researchers, and basic science PhDs) and varying math education experiences matriculate each year. We wanted to address the resulting range of quantitative skills to help students succeed in our program. In Spring 2023, we met with faculty teaching quantitative courses to identify core algebra concepts needed to succeed in their classes. A graduate student in computational mathematics with extensive tutoring experience then drafted assessment questions, a comprehensive study guide, and brief cheat sheet. The material was reviewed and revised with input from quantitative faculty. Results/Anticipated Results: We developed a 20-item quantitative assessment covering properties of operators; identity elements and inverses; simplification of arithmetic and algebraic expressions; solving algebraic equations; functions; equations of a line; and exponents/logarithms. A cheat sheet provided trainees with a brief refresher for these topics. A study guide provided more detailed instruction, example exercises and solutions, and referenced publicly available, online resources (e.g., Khan Academy). During the introductory summer course for the Tufts CTS Program, trainees were allowed to use the cheat sheet and were given 1 hour to complete the assessment. Trainees who got questions incorrect were directed to relevant sections in the study guide. We anticipate collecting formal feedback to evaluate the material. Discussion/Significance of Impact: Trainees must have adequate foundational algebra skills to succeed in CTS graduate programs and as future researchers. Developing a quantitative assessment allowed us to identify areas of weakness resulting from educational disparities or reflecting other aspects of their backgrounds and to provide material to reinforce their preparation.

A patent foramen ovale (PFO), an opening between the right and left atria during normal fetal development that fails to close after birth, is present in approximately 25% of all adults. Paradoxical embolism, a venous thromboembolism that travels to the systemic circulation typically through a PFO, accounts for about 5% of all strokes and 10% of strokes in younger patients. Approximately 50% of patients 60 years or younger with an embolic stroke of undetermined source (cryptogenic stroke) have a PFO, compared with 25% of the general population. The Risk of Paradoxical Embolism (RoPE) score incorporates clinical characteristics (age, history of stroke or transient ischemic attack, diabetes, hypertension, smoking, cortical infarct on imaging) to predict the likelihood that embolic stroke of undetermined source was caused by a PFO. Among patients in the lowest RoPE score category (score <3), PFO prevalence was similar to that in the general population (23%), while PFO prevalence was 77% in patients with a RoPE score of 9 or 10. The PFO-Associated Stroke Causal Likelihood (PASCAL) classification system combines the RoPE score and anatomical criteria from echocardiography (large shunt, atrial septal aneurysm) to classify PFO as the “probable,” “possible,” or “unlikely” cause of otherwise cryptogenic stroke. PFO closure reduces recurrent ischemic stroke in patients 60 years or younger with cryptogenic stroke. In a pooled analysis of 6 trials (3740 patients), the annualized incidence of stroke over a median follow-up of 57 months was 0.47% (95% CI, 0.35%-0.65%) with PFO closure vs 1.09% (95% CI, 0.88%-1.36%) with medical therapy (adjusted hazard ratio, 0.41 [95% CI, 0.28-0.60]). However, the benefits and harms of closure were highly heterogeneous across the trial populations. In patients categorized as PASCAL “probable” (ie, younger patients without vascular risk factors and high-risk PFO anatomical features), there was a 90% decreased relative rate of recurrent ischemic stroke after PFO closure at 2 years (hazard ratio, 0.10 [95% CI, 0.03-0.35]; absolute risk reduction, 2.1% [95% CI, 0.9%-3.4%]). PASCAL “unlikely” patients (eg, older patients with vascular risk factors and no high-risk PFO anatomical features) did not have a lower recurrent stroke rate with PFO closure but had higher risk of procedure- and device-related adverse events, such as atrial fibrillation. Patent foramen ovale is present in approximately 25% of all adults and is a common cause of stroke in young and middle-aged patients. The PASCAL classification system can help guide patient selection for PFO closure. Percutaneous PFO closure substantially reduces the risk of stroke recurrence in well-selected patients younger than 60 years after cryptogenic stroke.

Importance:A patent foramen ovale (PFO), an opening between the right and left atria during normal fetal development that fails to close after birth, is present in approximately 25% of all adults. Paradoxical embolism, a venous thromboembolism that travels to the systemic circulation typically through a PFO, accounts for about 5% of all strokes and 10% of strokes in younger patients. Observations:Approximately 50% of patients 60 years or younger with an embolic stroke of undetermined source (cryptogenic stroke) have a PFO, compared with 25% of the general population. The Risk of Paradoxical Embolism (RoPE) score incorporates clinical characteristics (age, history of stroke or transient ischemic attack, diabetes, hypertension, smoking, cortical infarct on imaging) to predict the likelihood that embolic stroke of undetermined source was caused by a PFO. Among patients in the lowest RoPE score category (score <3), PFO prevalence was similar to that in the general population (23%), while PFO prevalence was 77% in patients with a RoPE score of 9 or 10. The PFO-Associated Stroke Causal Likelihood (PASCAL) classification system combines the RoPE score and anatomical criteria from echocardiography (large shunt, atrial septal aneurysm) to classify PFO as the "probable," "possible," or "unlikely" cause of otherwise cryptogenic stroke. PFO closure reduces recurrent ischemic stroke in patients 60 years or younger with cryptogenic stroke. In a pooled analysis of 6 trials (3740 patients), the annualized incidence of stroke over a median follow-up of 57 months was 0.47% (95% CI, 0.35%-0.65%) with PFO closure vs 1.09% (95% CI, 0.88%-1.36%) with medical therapy (adjusted hazard ratio, 0.41 [95% CI, 0.28-0.60]). However, the benefits and harms of closure were highly heterogeneous across the trial populations. In patients categorized as PASCAL "probable" (ie, younger patients without vascular risk factors and high-risk PFO anatomical features), there was a 90% decreased relative rate of recurrent ischemic stroke after PFO closure at 2 years (hazard ratio, 0.10 [95% CI, 0.03-0.35]; absolute risk reduction, 2.1% [95% CI, 0.9%-3.4%]). PASCAL "unlikely" patients (eg, older patients with vascular risk factors and no high-risk PFO anatomical features) did not have a lower recurrent stroke rate with PFO closure but had higher risk of procedure- and device-related adverse events, such as atrial fibrillation. Conclusions and Relevance:Patent foramen ovale is present in approximately 25% of all adults and is a common cause of stroke in young and middle-aged patients. The PASCAL classification system can help guide patient selection for PFO closure. Percutaneous PFO closure substantially reduces the risk of stroke recurrence in well-selected patients younger than 60 years after cryptogenic stroke.

Background: External validations are essential to assess the performance of a clinical prediction model (CPM) before deployment. Apart from model misspecification, also differences in patient population, the standard of care, predictor definitions, and other factors influence a model's discriminative ability, as commonly quantified by the AUC (or c-statistic). We aimed to quantify the variation in AUCs across sets of external validation studies and propose ways to adjust expectations of a model's performance in a new setting. Methods: The Tufts-PACE CPM Registry holds a collection of CPMs for prognosis in cardiovascular disease. We analyzed the AUC estimates of 469 CPMs with at least one external validation. Combined, these CPMs had a total of 1603 external validations reported in the literature. For each CPM and its associated set of validation studies, we performed a random-effects meta-analysis to estimate the between-study standard deviation tau among the AUCs. Since the majority of these meta-analyses have only a handful of validations, this leads to very poor estimates of tau. So, instead of focusing on a single CPM, we estimated a log-normal distribution of tau across all 469 CPMs. We then used this distribution as an empirical prior. We used cross-validation to compare this empirical Bayesian approach with frequentist fixed and random-effects meta-analyses. Results: The 469 CPMs included in our study had a median of 2 external validations with an IQR of [1-3]. The estimated distribution of tau had a mean of 0.055 and a standard deviation of 0.015. If tau = 0.05, then the 95% prediction interval for the AUC in a new setting has a width of at least +/- 0.1, no matter how many validations have been done. When there are fewer than 5 validations, which is typically the case, the usual frequentist methods grossly underestimate the uncertainty about the AUC in a new setting. Accounting for tau in a Bayesian approach achieved near nominal coverage. Conclusion: Due to large heterogeneity among the validated AUC values of a CPM, there is great irreducible uncertainty in predicting the AUC in a new setting. This uncertainty is underestimated by existing methods. The proposed empirical Bayes approach addresses this problem which merits wide application in judging the validity of prediction models.

BACKGROUND:Glucagon-like peptide-1 receptor agonists (GLP-1RAs) and sodium-glucose cotransporter-2 inhibitors (SGLT2is) have favorable cardiovascular outcomes compared with dipeptidyl peptidase-4 inhibitors (DPP4is) and sulfonylureas in adults with type 2 diabetes and high cardiovascular risk. How these benefits vary across lower levels of cardiovascular risk is unknown. METHODS:We used nationwide claims data to emulate a comparative effectiveness trial and examine the heterogeneity of treatment effects of GLP-1RAs, SGLT2is, DPP4is, and sulfonylureas on major adverse cardiovascular events (MACEs) among adults with type 2 diabetes and moderate cardiovascular risk (annualized MACE risk 1%-5%, estimated using the annualized claims-based MACE estimator). RESULTS:Among 386 276 included adults with type 2 diabetes, 25.2% had baseline ACME-predicted MACE risk >1% to ≤2% (lower-risk patients) and 13.3% had ACME-predicted risk >4% to ≤5% (higher-risk patients). By year 3 of treatment, higher-risk patients derived greater absolute benefit than lower-risk patients when treated with GLP-1RAs versus sulfonylureas (absolute reduction in the estimated rate of MACE of 3.1% in higher-risk patients and 1.6% in lower-risk patients), SGLT2is versus sulfonylureas (absolute reduction, 3.9% in higher-risk patients and 1.3% in lower-risk patients), and GLP-1RAs versus DPP4is (absolute reduction, 1.6% in higher-risk patients and 0.5% in lower-risk patients). The relative benefits for MACE were also greater in higher-risk than lower-risk patients with SGLT2is versus DPP4is (hazard ratio [HR], 0.78 [95% CI, 0.70-0.87] in higher-risk patients; HR, 0.99 [95% CI, 0.88-1.12] in lower-risk patients). Conversely, the relative benefits of DPP4is and GLP-1RAs versus sulfonylureas were greater in lower-risk patients: HR 0.76 (95% CI, 0.71-0.81) in lower-risk and HR 0.91 (95% CI, 0.97-0.96) in higher-risk patients for DPP4is versus sulfonylureas; HR 0.67 (95% CI, 0.58-0.78) in lower-risk and HR 0.80 (95% CI, 0.70-0.93) in higher-risk patients for GLP-1RAs versus sulfonylurea. Benefits of SGLT2is and GLP-1RAs were comparable across all risk levels. CONCLUSIONS:Cardiovascular benefits of SGLT2is and GLP-1RAs exist across all levels of moderate cardiovascular risk, reinforcing the importance of choosing glucose-lowering therapies that can prevent MACE in all people with type 2 diabetes.

Importance:The Predictive Approaches to Treatment Effect Heterogeneity (PATH) Statement of 2020 proposed predictive modeling for identifying heterogeneity in treatment effects (HTE) in randomized clinical trials (RCTs). It described 2 approaches: risk modeling, which develops a multivariable model predicting individual baseline risk of study outcomes and then examines treatment effects across strata of predicted risk, and effect modeling, which develops a model that directly predicts individual treatment effects using a variety of regression and machine learning methods. Objective:To identify, describe, and evaluate findings from reports that cited the PATH Statement and presented predictive modeling of HTE in RCTs. Evidence Review:Reports were identified using PubMed, Google Scholar, Web of Science, and SCOPUS through July 5, 2024. Using double review with adjudication, reports were assessed for consistency with PATH Statement recommendations, credibility of HTE findings (applying criteria adapted from the Instrument to Assess Credibility of Effect Modification Analyses), and clinical importance of credible findings. Findings:A total of 65 reports (presenting 31 risk models and 41 effect models) analyzing 162 RCTs were identified, with credible, clinically important HTE in 24 reports (37%). Contrary to PATH Statement recommendations, only 25 of 48 studies with positive overall findings included a risk model. Most effect models were exploratory, including multiple predictors with little prior evidence for HTE. Claims of HTE were noted in 23 risk modeling and 31 effect modeling reports but were more likely to meet credibility criteria with risk modeling (20 of 23 reports [87%]) than effect modeling (10 of 31 reports [32%]). For effect modeling, validation of HTE findings in external datasets was critical in establishing credibility. Credible HTE from either approach was usually judged clinically important (24 of 30 reports [80%]). In the 19 reports from RCTs suggesting overall treatment benefits, modeling identified subgroups of 5% to 67% of patients predicted to experience no benefit or net treatment harm. In the 5 reports that found no overall benefit, subgroups of 25% to 60% of patients were nevertheless predicted to benefit. Conclusions and Relevance:This scoping review of 65 reports of multivariable predictive modeling of HTE in RCTs identified credible, clinically important HTE in 37%. Risk modeling was more likely than effect modeling to find credible HTE, but external validation of HTE findings served to increase the credibility of findings from exploratory effect models.

Background:Covert cerebrovascular disease (CCD), comprising covert brain infarction (CBI) and white matter disease (WMD), is common in older adults and linked to increased risk of stroke and dementia. While most CCD research relies on MRI, CT remains the predominant imaging modality in clinical care. The influence of imaging modality on detection and prognosis of incidentally discovered CCD remains unclear. Methods:We identified 18,626 patients aged ≥50 years from Kaiser Permanente Southern California who underwent both CT and MRI brain scans within 30 days between 2009-2022. Patients with known prior stroke or dementia were excluded. Natural language processing algorithms were applied to radiology reports to identify CBI and WMD status and WMD severity (none, mild, moderate, severe). We assessed prevalence, cross-modality agreement (Cohen's kappa), and reclassification patterns. Prognostic associations with incident stroke or dementia were estimated using Cox Proportional Hazards regression adjusted for vascular and cognitive risk factors. Findings:CBI prevalence was similar for CT (6.3%) and MRI (6.1%), but agreement was modest (κ=0.27). WMD was reported far more often on MRI (60.5%) than CT (24.4%). Among 15,551 patients with classifiable severity on both modalities, 47.9% (n=7,441) had discordant grades, with 92.3% upgraded on MRI. The incidence rates of stroke or dementia per 1,000 person-years were 12.7 (95% CI 11.5 - 14.0) for patients without WMD on either modality (36.3% of the cohort), 22.6 (21.0 - 24.2) for WMD detected on MRI only (39.2% of the cohort), and 52.2 (48.69 to 55.95) for WMD detected on both CT and MRI (21.2% of the cohort). In adjusted Cox models, WMD detected on MRI only was associated with a 23% higher hazard of stroke or dementia (HR=1.23, 95% CI 1.07-1.41) compared with no WMD on either modality, while WMD detected on both CT and MRI was associated with an 82% higher hazard (1.82, 1.58-2.11). Interpretation:MRI detects substantially more WMD than CT; however, WMD visible on CT has stronger prognostic significance, despite CT's low sensitivity. These findings emphasize modality-based diagnostic and prognostic differences and support the need for modality-specific approaches when translating CCD research into clinical risk assessment and patient counselling.

Objectives/Goals: Trainees in clinical and translational science (CTS) must learn to effectively communicate their research ideas and findings to a range of audiences. As part of our science communication curriculum, we developed ORAL and WRITTEN science communication rubrics for our trainees to use across their courses and research activities. Methods/Study Population: The Tufts CTS Graduate Program is the training core of the Tufts CTSI and its associated pre- and post-doctoral T32 awards. Approximately 10 trainees with a range of backgrounds (e.g., physicians, medical students, master’s-level researchers, and basic science PhDs) matriculate each year. Faculty members and staff with expertise in science communication and pedagogy formed a committee to develop the rubrics. Because oral and written communication require different skills, we developed separate rubrics for each. We reviewed our current science communication curriculum, reviewed existing communication rubrics, and identified common mistakes students make. Following pilot testing by students and faculty pilot for one semester, we modified the rubrics based on informal feedback. Results/Anticipated Results: Both rubrics include a section to identify the target audience and specific items organized by theme. Oral rubric themes include presentation content, slides, verbal communication, nonverbal communication, and following instructions. Written rubric themes include overall, manuscript/proposal sections, and following instructions. The rubrics serve as feedback tools for faculty and students to evaluate work others produce and as self-evaluation tools. Feedback elements include a 4-point rating for each rubric item, open text feedback for each theme, and an open text holistic assessment. We now use the rubrics in our study design course, which features student presentations of planned research, and in our writing course. We anticipate collecting formal student feedback to further evaluate the rubrics. Discussion/Significance of Impact: Our rubrics can supplement existing science communication training and can be integrated into all CTS coursework and research activities. For future clinical and translational scientists to have the greatest impact, they must learn to effectively communicate findings to multiple audiences, ranging from experts in their field to the general public.

The proliferation of algorithm-assisted decision making has prompted calls for careful assessment of algorithm fairness. One popular fairness metric, equal opportunity, demands parity in true positive rates (TPRs) across different population subgroups. However, we highlight a critical but overlooked weakness in this measure: at a given decision threshold, TPRs vary when the underlying risk distribution varies across subgroups, even if the model equally captures the underlying risks. Failure to account for variations in risk distributions may lead to misleading conclusions on performance disparity. To address this issue, we introduce a novel metric called adjusted TPR (aTPR), which modifies subgroup-specific TPRs to reflect performance relative to the risk distribution in a common reference subgroup. Evaluating fairness using aTPRs promotes equal treatment for equal risk by reflecting whether individuals with similar underlying risks have similar opportunities of being identified as high risk by the model, regardless of subgroup membership. We demonstrate our method through numerical experiments that explore a range of differential calibration relationships and in a real-world data set that predicts 6-month mortality risk in an in-patient sample in order to increase timely referrals for palliative care consultations.

BACKGROUND:Insertable cardiac monitoring (ICM) detects atrial fibrillation (AF) in substantial proportions of cryptogenic stroke, noncryptogenic ischemic stroke without known AF, and nonstroke patients who are at risk of underlying AF. Given differences in patient characteristics across studies, there may be differences in AF detection rates on ICM across these subgroups that have not been identified. We investigate whether AF detection rates on ICM are higher in cryptogenic stroke or transient ischemic attack (C-IS/TIA) patients compared with individuals with noncryptogenic stroke or without stroke, when accounting for differences in study populations. METHODS:This is an individual-participant data meta-analysis of prospective studies and randomized controlled trials of ICM in C-IS/TIA, noncryptogenic ischemic stroke, and nonstroke patients. Multilevel multivariable logistic regression models were used to test whether C-IS/TIA is associated with increased AF detection relative to other categories. We performed multiple imputation to derive values for variables with <20% missing data and used Rubin's rules to estimate adjusted odds ratios by combining 100 postimputation data sets. The primary outcome was detection of AF. The attributable risk was derived by application of Bayes' Theorem. RESULTS:Two randomized controlled trials and 12 prospective studies were included with a total of 1562 C-IS/TIA patients and 474 non-C-IS/TIA patients. In adjusted multilevel logistic regression analyses, AF detection was higher in C-IS/TIA patients (adjusted odds ratio, 1.90 [95% CI, 1.18-3.06]; P=0.009), indicating that 47% of AF detected in C-IS/TIA is pathogenic. Limiting the comparator group to ischemic stroke or history of stroke yielded similar results (adjusted odds ratio, 2.83 [95% CI, 1.47-5.44]; P=0.002). Days to AF detection were significantly shorter in C-IS/TIA patients (median 65 versus 169; P<0.001). CONCLUSIONS:In this individual-participant data meta-analysis of patients undergoing ICM, AF detection was higher in C-IS/TIA patients, with shorter time to AF detection compared with noncryptogenic/nonstroke individuals. These findings suggest that some of the AF detected in patients with C-IS/TIA may be pathogenic.

BACKGROUND:Risk-based analyses are increasingly popular for understanding heterogeneous treatment effects (HTEs) in clinical trials. For time-to-event analyses, the assumption that high-risk patients benefit most on the clinically important absolute scale when hazard ratios (HRs) are constant across risk strata might not hold. Absolute treatment effects can be measured as either the risk difference (RD) at a given time point or the difference in the restricted mean survival time (ΔRMST), which aligns more closely with utilitarian medical decision-making frameworks. We examined risk-based HTE analyses strata in time-to-event analyses to identify the patterns of absolute HTE across risk strata and whether the ΔRMST may lead to better treatment decisions than the RD. METHODS:Using artificial and empirical time-to-event data, we compared the RD-the difference between Kaplan-Meier estimates at a certain time point-and the ΔRMST-the area between the Kaplan-Meier curves-across risk strata and show how these metrics can prioritize different subgroups for treatment. We explored scenarios involving constant HRs while varying both the overall event rates and the discrimination of the risk models. RESULTS:When event rates and discrimination were low, the RD and the ΔRMST increased monotonically, with high-risk patients benefitting more than low-risk patients. As the event rate increased and/or discrimination increased, a 'sweet spot' pattern emerged: intermediate-risk patients benefit more than low-risk and high-risk patients. When the RD was used, the 'sweet spot' pattern emerged, even in circumstances in which the ΔRMST increased across the risk groups, thus understating the benefit for higher-risk patients and potentially leading to treatment mistargeting. CONCLUSION:The pattern of HTE characterized by the RD may diverge substantially from the ΔRMST, potentially leading to treatment mistargeting. Therefore, we recommend the ΔRMST for assessing the absolute HTE in time-to-event data.

BACKGROUND:Transcatheter aortic valve intervention (TAVI) has revolutionized the care of older adults with aortic stenosis. OBJECTIVES:The objectives of the study were to examine associations between chronic conditions and outcomes after TAVI and to describe palliative care utilization rates. METHODS:This cohort study used the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy registry to identify patients who underwent TAVI and were eligible for linkage with Centers for Medicare & Medicaid Services claims data. The exposure was multiple chronic conditions (MCCs) in the year before TAVI. Associations between chronic conditions and outcomes were assessed using multivariable logistic regression. RESULTS:A total of 188,629 TAVI procedures were linked to Centers for Medicare & Medicaid Services claims. The median (IQR) age was 82.0 (76.0-87.0) years; 86,841 (46%) were female. Chronic conditions were associated with worse 1-year mortality (high MCC [≥6 conditions] vs low MCC [<4 chronic conditions], adjusted HR: 2.33 [95% CI: 2.22-2.44]). Chronic conditions were associated with lower Kansas City Cardiomyopathy Questionnaire at baseline (high MCC median score 37.5 [21.4-56.8] vs low MCC median score 55.7 [37.5-75.0], P < 0.001); however, the average improvement in Kansas City Cardiomyopathy Questionnaire after TAVI was large and appeared independent of chronic disease burden (median score change high MCC 28.7 [9.9-48.4] vs low MCC 24.5 [8.3-42.2], standardized difference +13.8%). Palliative care encounters were rare (8,946, 4.7%) and varied significantly across centers (range 0% to 25% of cases). CONCLUSIONS:Chronic conditions are associated with worse survival after TAVI. However, most patients with high MCC are alive 1 year after treatment, and quality of life improvements appear independent of chronic disease burden. These data help clarify expected health gains for patients with chronic conditions and symptomatic aortic stenosis.

BACKGROUND:The impact of covert cerebrovascular disease on falls in the general population is not well-known. Here, we determine the time to a first fall following incidentally detected covert cerebrovascular disease during a clinical neuroimaging episode. METHODS:This longitudinal cohort study assessed computed tomography (CT) and magnetic resonance imaging from 2009 to 2019 of patients aged >50 years registered with Kaiser Permanente Southern California which is a healthcare organization combining health plan coverage with coordinated medical services, excluding those with before stroke/dementia. We extracted evidence of incidental covert brain infarcts (CBI) and white matter hyperintensities/hypoattenuation (WMH) from imaging reports using natural language processing. We examined associations of CBI and WMH with falls requiring medical attention, using Cox proportional hazards regression models with adjustment for 12 variables including age, sex, ethnicity multimorbidity, polypharmacy, and incontinence. RESULTS:We assessed 241 050 patients, mean age 64.9 (SD, 10.42) years, 61.3% female, detecting covert cerebrovascular disease in 31.1% over a mean follow-up duration of 3.04 years. A recorded fall occurred in 21.2% (51 239/241 050) during follow-up. On CT, single fall incidence rate/1000 person-years (p-y) was highest in individuals with both CBI and WMH on CT (129.3 falls/1000 p-y [95% CI, 123.4-135.5]), followed by WMH (109.9 falls/1000 p-y [108.0-111.9]). On magnetic resonance imaging, the incidence rate was the highest with both CBI and WMH (76.3 falls/1000 p-y [95% CI, 69.7-83.2]), followed by CBI (71.4 falls/1000 p-y [95% CI, 65.9-77.2]). The adjusted hazard ratio for single index fall in individuals with CBI on CT was 1.13 (95% CI, 1.09-1.17); versus magnetic resonance imaging 1.17 (95% CI, 1.08-1.27). On CT, the risk for single index fall incrementally increased for mild (1.37 [95% CI, 1.32-1.43]), moderate (1.57 [95% CI, 1.48-1.67]), or severe WMH (1.57 [95% CI, 1.45-1.70]). On magnetic resonance imaging, index fall risk similarly increased with increasing WMH severity: mild (1.11 [95% CI, 1.07-1.17]), moderate (1.21 [95% CI, 1.13-1.28]), and severe WMH (1.34 [95% CI, 1.22-1.46]). CONCLUSIONS:In a large population with neuroimaging, CBI and WMH are independently associated with greater risks of an index fall. Increasing severities of WMH are associated incrementally with fall risk across imaging modalities.