Search

BackgroundCOVID-19 has infected millions of people worldwide and is responsible for several hundred thousand fatalities. The COVID-19 pandemic has necessitated thoughtful resource allocation and early identification of high-risk patients. However, effective methods to meet these needs are lacking. ObjectiveThe aims of this study were to analyze the electronic health records (EHRs) of patients who tested positive for COVID-19 and were admitted to hospitals in the Mount Sinai Health System in New York City; to develop machine learning models for making predictions about the hospital course of the patients over clinically meaningful time horizons based on patient characteristics at admission; and to assess the performance of these models at multiple hospitals and time points. MethodsWe used Extreme Gradient Boosting (XGBoost) and baseline comparator models to predict in-hospital mortality and critical events at time windows of 3, 5, 7, and 10 days from admission. Our study population included harmonized EHR data from five hospitals in New York City for 4098 COVID-19–positive patients admitted from March 15 to May 22, 2020. The models were first trained on patients from a single hospital (n=1514) before or on May 1, externally validated on patients from four other hospitals (n=2201) before or on May 1, and prospectively validated on all patients after May 1 (n=383). Finally, we established model interpretability to identify and rank variables that drive model predictions. ResultsUpon cross-validation, the XGBoost classifier outperformed baseline models, with an area under the receiver operating characteristic curve (AUC-ROC) for mortality of 0.89 at 3 days, 0.85 at 5 and 7 days, and 0.84 at 10 days. XGBoost also performed well for critical event prediction, with an AUC-ROC of 0.80 at 3 days, 0.79 at 5 days, 0.80 at 7 days, and 0.81 at 10 days. In external validation, XGBoost achieved an AUC-ROC of 0.88 at 3 days, 0.86 at 5 days, 0.86 at 7 days, and 0.84 at 10 days for mortality prediction. Similarly, the unimputed XGBoost model achieved an AUC-ROC of 0.78 at 3 days, 0.79 at 5 days, 0.80 at 7 days, and 0.81 at 10 days. Trends in performance on prospective validation sets were similar. At 7 days, acute kidney injury on admission, elevated LDH, tachypnea, and hyperglycemia were the strongest drivers of critical event prediction, while higher age, anion gap, and C-reactive protein were the strongest drivers of mortality prediction. ConclusionsWe externally and prospectively trained and validated machine learning models for mortality and critical events for patients with COVID-19 at different time horizons. These models identified at-risk patients and uncovered underlying relationships that predicted outcomes.

Abstract: Importance: SARS-CoV-2 infection can result in ongoing, relapsing, or new symptoms or other health effects after the acute phase of infection; termed post-acute sequelae of SARS-CoV-2 infection (PASC), or long COVID. The characteristics, prevalence, trajectory and mechanisms of PASC are ill-defined. The objectives of the Researching COVID to Enhance Recovery (RECOVER) Multi-site Observational Study of PASC in Adults (RECOVER-Adult) are to: (1) characterize PASC prevalence; (2) characterize the symptoms, organ dysfunction, natural history, and distinct phenotypes of PASC; (3) identify demographic, social and clinical risk factors for PASC onset and recovery; and (4) define the biological mechanisms underlying PASC pathogenesis. Methods: RECOVER-Adult is a combined prospective/retrospective cohort currently planned to enroll 14,880 adults aged ≥18 years. Eligible participants either must meet WHO criteria for suspected, probable, or confirmed infection; or must have evidence of no prior infection. Recruitment occurs at 86 sites in 33 U.S. states, Washington, DC and Puerto Rico, via facility– and community-based outreach. Participants complete quarterly questionnaires about symptoms, social determinants, vaccination status, and interim SARS-CoV-2 infections. In addition, participants contribute biospecimens and undergo physical and laboratory examinations at approximately 0, 90 and 180 days from infection or negative test date, and yearly thereafter. Some participants undergo additional testing based on specific criteria or random sampling. Patient representatives provide input on all study processes. The primary study outcome is onset of PASC, measured by signs and symptoms. A paradigm for identifying PASC cases will be defined and updated using supervised and unsupervised learning approaches with cross– validation. Logistic regression and proportional hazards regression will be conducted to investigate associations between risk factors, onset, and resolution of PASC symptoms. Discussion: RECOVER-Adult is the first national, prospective, longitudinal cohort of PASC among US adults. Results of this study are intended to inform public health, spur clinical trials, and expand treatment options.

Background and objectivesSuccessful clinical integration of genomic sequencing (GS) requires evidence of its utility. While GS potentially has benefits (utilities) or harms (disutilities) across multiple domains of life for both patients and their families, there is as yet no empirically informed conceptual model of these effects. Our objective was to develop an empirically informed conceptual model of perceived utility of GS that captures utilities and disutilities for patients and their families across diverse backgrounds.MethodsWe took a patient-centered approach, in which we began with a review of existing literature followed by collection of primary interview data. We conducted semi-structured interviews to explore types of utility in a clinically and sociopolitically diverse sample of 60 adults from seven Clinical Sequencing Evidence-Generating Research (CSER) consortium projects. Interviewees had either personally received, or were parents of a child who had received, GS results. Qualitative data were analyzed using thematic analysis. Findings from interviews were integrated with existing literature on clinical and personal utility to form the basis of an initial conceptual model that was refined based on expert review and feedback.ResultsFive key utility types that have been previously identified in qualitative literature held up as primary domains of utility and disutility in our diverse sample. Interview data were used to specify and organize subdomains of an initial conceptual model. After expert refinement, the five primary domains included in the final model are clinical, emotional, behavioral, cognitive, and social, and several subdomains are specified within each.ConclusionWe present an empirically informed conceptual model of perceived utility of GS. This model can be used to guide development of instruments for patient-centered outcome measurement that capture the range of relevant utilities and disutilities and inform clinical implementation of GS.

PurposeThere is a critical need for genomic medicine research that reflects and benefits socioeconomically and ancestrally diverse populations. However, disparities in research populations persist, highlighting that traditional study designs and materials may be insufficient or inaccessible to all groups. New approaches can be gained through collaborations with patient/community stakeholders. Although some benefits of stakeholder engagement are recognized, routine incorporation into the design and implementation of genomics research has yet to be realized.MethodsThe National Institutes of Health-funded Clinical Sequencing Evidence-Generating Research (CSER) consortium required stakeholder engagement as a dedicated project component. Each CSER project planned and carried out stakeholder engagement activities with differing goals and expected outcomes. Examples were curated from each project to highlight engagement strategies and outcomes throughout the research lifecycle from development through dissemination.ResultsProjects tailored strategies to individual study needs, logistical constraints, and other challenges. Lessons learned include starting early with engagement efforts across project stakeholder groups and planned flexibility to enable adaptations throughout the project lifecycle.ConclusionEach CSER project used more than 1 approach to engage with relevant stakeholders, resulting in numerous adaptations and tremendous value added throughout the full research lifecycle. Incorporation of community stakeholder insight improves the outcomes and relevance of genomic medicine research.

BackgroundGlobally, several diabetes prevention interventions have been shown to be cost-effective, yet they have had limited adaptation, implementation, and evaluation in the Caribbean and among Caribbean-descent individuals, where the burden of type 2 diabetes is high. We report on the protocol for the Lifestyle Intervention with Metformin Escalation (LIME) study - an evidence-based diabetes prevention intervention to reduce the incidence of diabetes among Caribbean-descent individuals with prediabetes.MethodsLIME is a hybrid type-I effectiveness-implementation quasi-experimental study taking place in 4 clinical sites in Barbados, Trinidad, the U.S. Virgin Islands, and Puerto Rico. LIME targets individuals who self-identify as Caribbean or Caribbean-descent and have high-risk prediabetes with a hemoglobin A1c (HbA1c) between 6 and 6.4%. Eligible participants in the intervention arm are enrolled in a six-week lifestyle modification workshop. Six months later, individuals who have not lost at least 5% of their bodyweight or continue to have an HbA1c of 6% or higher are prescribed metformin medication. In total, participants are followed for one year. The primary effectiveness outcome is proportion of individuals who lower their HbA1c below 6%.DiscussionLIME is a unique diabetes prevention intervention for Caribbean and Caribbean-descent individuals. LIME utilizes a tailored lifestyle change curriculum, incorporates appropriate metformin prescribing when lifestyle change alone is insufficient, targets the highest-risk individuals with prediabetes, and is based in a clinical setting to ensure sustainability.

IntroductionImplementation of genome-scale sequencing in clinical care has significant challenges: the technology is highly dimensional with many kinds of potential results, results interpretation and delivery require expertise and coordination across multiple medical specialties, clinical utility may be uncertain, and there may be broader familial or societal implications beyond the individual participant. Transdisciplinary consortia and collaborative team science are well poised to address these challenges. However, understanding the complex web of organizational, institutional, physical, environmental, technologic, and other political and societal factors that influence the effectiveness of consortia is understudied. We describe our experience working in the Clinical Sequencing Evidence-Generating Research (CSER) consortium, a multi-institutional translational genomics consortium.MethodsA key aspect of the CSER consortium was the juxtaposition of site-specific measures with the need to identify consensus measures related to clinical utility and to create a core set of harmonized measures. During this harmonization process, we sought to minimize participant burden, accommodate project-specific choices, and use validated measures that allow data sharing.ResultsIdentifying platforms to ensure swift communication between teams and management of materials and data were essential to our harmonization efforts. Funding agencies can help consortia by clarifying key study design elements across projects during the proposal preparation phase and by providing a framework for data sharing data across participating projects.ConclusionsIn summary, time and resources must be devoted to developing and implementing collaborative practices as preparatory work at the beginning of project timelines to improve the effectiveness of research consortia.

BackgroundStakeholders (ie, patients, policymakers, clinicians, advocacy groups, health system leaders, payers, and others) offer critical input at various stages in the research continuum, and their contributions are increasingly recognized as an important component of effective translational research. Natural experiments, in particular, may benefit from stakeholder feedback in addressing real-world issues and providing insight into future policy decisions, though best practices for the engagement of stakeholders in observational studies are limited in the literature.MethodsThe Natural Experiments for Translation in Diabetes 2.0 (NEXT-D2) network utilizes rigorous methods to evaluate natural experiments in health policy and program delivery with a focus on diabetes-related outcomes. Each of the 8 partnering institutions incorporates stakeholder engagement throughout multiple study phases to enhance the patient-centeredness of results. NEXT-D2 dedicates a committee to Engagement for resource sharing, enhancing engagement approaches, and advancing network-wide engagement activities. Key stakeholder engagement activities include Study Meetings, Proposal Development, Trainings & Educational Opportunities, Data Analysis, and Results Dissemination. Network-wide patient-centered resources and multimedia have also been developed through the broad expertise of each site's stakeholder group.ConclusionsThis collaboration has created a continuous feedback loop wherein site-level engagement approaches are informed via the network and network-level engagement efforts are shaped by individual sites. Emerging best practices include: incorporating stakeholders in multiple ways throughout the research, building on previous relationships with stakeholders, enhancing capacity through stakeholder and investigator training, involving stakeholders in refining outcome choices and understanding the meaning of variables, and recognizing the power of stakeholders in maximizing dissemination.

Conceptual frameworks are useful in research because they can highlight priority research domains, inform decisions about interventions, identify outcomes and factors to measure, and display how factors might relate to each other to generate and test hypotheses. Discovery, translational, and implementation research are all critical to the overall mission of genomic medicine and prevention, but they have yet to be organized into a unified conceptual framework. To fill this gap, our diverse team collaborated to develop the Genomic Medicine Integrative Research (GMIR) Framework, a simple but comprehensive tool to aid the genomics community in developing research questions, strategies, and measures and in integrating genomic medicine and prevention into clinical practice. Here we present the GMIR Framework and its development, along with examples of its use for research development, demonstrating how we applied it to select and harmonize measures for use across diverse genomic medicine implementation projects. Researchers can utilize the GMIR Framework for their own research, collaborative investigations, and clinical implementation efforts; clinicians can use it to establish and evaluate programs; and all stakeholders can use it to help allocate resources and make sure that the full complexity of etiology is included in research and program design, development, and evaluation.

The Clinical Sequencing Evidence-Generating Research (CSER) consortium, now in its second funding cycle, is investigating the effectiveness of integrating genomic (exome or genome) sequencing into the clinical care of diverse and medically underserved individuals in a variety of healthcare settings and disease states. The consortium comprises a coordinating center, six funded extramural clinical projects, and an ongoing National Human Genome Research Institute (NHGRI) intramural project. Collectively, these projects aim to enroll and sequence over 6,100 participants in four years. At least 60% of participants will be of non-European ancestry or from underserved settings, with the goal of diversifying the populations that are providing an evidence base for genomic medicine. Five of the six clinical projects are enrolling pediatric patients with various phenotypes. One of these five projects is also enrolling couples whose fetus has a structural anomaly, and the sixth project is enrolling adults at risk for hereditary cancer. The ongoing NHGRI intramural project has enrolled primarily healthy adults. Goals of the consortium include assessing the clinical utility of genomic sequencing, exploring medical follow up and cascade testing of relatives, and evaluating patient-provider-laboratory level interactions that influence the use of this technology. The findings from the CSER consortium will offer patients, healthcare systems, and policymakers a clearer understanding of the opportunities and challenges of providing genomic medicine in diverse populations and settings, and contribute evidence toward developing best practices for the delivery of clinically useful and cost-effective genomic sequencing in diverse healthcare settings.

There is a need for diabetes prevention efforts targeting vulnerable populations. Our community--academic partnership, the East Harlem Partnership for Diabetes Prevention, conducted a randomized controlled trial to study the impact of peer led diabetes prevention workshops on weight and diabetes risk among an economically and racially diverse population in East Harlem, New York. We recruited overweight/obese adults from more than 50 community sites and conducted oral glucose tolerance testing and completed other clinical assessments and a health and lifestyle survey. We randomized prediabetic participants to intervention or delayed intervention groups. Intervention participants attended eight 90-minute peer-led workshop sessions at community sites. Participants in both groups returned for follow-up assessments 6 months after randomization. The main outcomes were the proportion of participants who achieved 5% weight loss, percentage weight loss, and change in the probability of developing diabetes over the next 7.5 years according to the San Antonio Diabetes Prediction Model. We enrolled 402 participants who were mainly female (85%), Latino (73%) or Black (23%), foreign born (64%), and non-English speaking (58%). At 6 months, the intervention group lost a greater percentage of their baseline weight, had significantly lower rise in HbA1c (glycated hemoglobin), decreased risk of diabetes, larger decreases in fat and fiber intake, improved confidence in nutrition label reading, and decrease in sedentary behavior as compared with the control group. Thus, in partnership with community stakeholders, we created an effective low-resource program that was less intensive than previously studied programs by incorporating strategies to engage and affect our priority population.