Sudha R. Raman

Socially vulnerable communities within the contiguous United States (CONUS) face disproportionate heat exposure, yet how these disparities will change under future warming remains unclear. Although socially vulnerable communities already experience higher air temperatures, high‐resolution downscaled climate projections have not been used to assess future exposure disparities under different Global Warming Levels (GWLs). Here, we evaluate past and future daily maximum air temperature exposure across CONUS for high‐ and low‐vulnerability locations using the Social Vulnerability Index. We find the disparity in air temperature exposure for high‐SVI areas increased from ∼1.3°C (∼2.3°F) 1951–1980 to ∼1.6°C (∼2.9°F) 1994–2023. Using an ensemble of downscaled climate models, we assess future impacts at GWLs of 1.5°C, 2°C, and 3°C (2.7°F, 3.6°F, and 5.4°F). We compare the number of days/year that exceed air temperature thresholds recommended for safe fan use (35°C, 39°C, or 95°F, 102.2°F) in high‐SVI and low‐SVI areas. We find that for exposure to a 35°C air temperature threshold, at 1.5°C global warming there is an average annual exposure difference of ∼16 days, at 2°C ∼18 days, and at 3°C ∼22 days. For exposure to a 39°C threshold, at 1.5°C global warming there is an exposure difference of ∼5 days, at 2°C ∼6 days, and at 3°C ∼9 days. These results highlight that future warming will disproportionately increase the number of days exceeding adaptation‐relevant air temperature thresholds in high‐SVI areas. These findings support the need to slow global warming and direct adaptation efforts toward more vulnerable communities.

Introduction: To support a learning health system, we aimed to develop processes and tools to enable collaboration between the Veterans Health Administration (VHA) operations and research sectors in the implementation of promising healthcare innovations. Methods: The collaboration process involved: (1) holding foundational partnership meetings between operations and health research sector leadership (2) reviewing existing research–operations collaborations frameworks, (3) developing a framework with criteria for systematically categorizing innovations into defined pathways that integrate research and operational considerations, and (4) adapting known participatory approaches to collect information for productive collaboration. Clearly delineating the skills, goals, and perspectives of the innovation and research sectors during partnership meetings enabled assessment of the value and trade‐offs of advancing an innovation into practice. The literature review of existing frameworks yielded heterogeneous objectives, domains, and criteria for evaluation. Across framework domains, commonalities (e.g., alignment with institutional goals), differences (e.g., attention to sustainment), and gaps (e.g., health equity focus) were noted. We developed five research–innovation partnership pathways and criteria to vet and categorize potential innovations (i.e., evidence for effectiveness and implementation, risks, equity, feasibility, and sustainability). We developed a menu of participatory processes to elicit feedback on innovations. Results: We applied this process with an innovation to refine and evaluate the process, pathways, and criteria. Surveys of diverse partners (clinicians, administrators, researchers) ranked the innovation using our developed criteria, and placed the innovation on a pathway, which then helped guide next steps for evaluating the innovations. Overall, the process was feasible. We were able to categorize and plan next steps for promising innovations. Conclusion: This theoretically grounded, iterative process may serve as a blueprint to accelerate the implementation of healthcare innovations. Through these intentional and participatory processes to engage operations innovators and health services researchers we can speed the delivery of promising innovations that impact patient care.

Tissue-based genomic classifiers (GCs) have been developed to improve prostate cancer (PCa) risk assessment and treatment recommendations. To summarize the impact of the Decipher, Oncotype DX Genomic Prostate Score (GPS), and Prolaris GCs on risk stratification and patient-clinician decisions on treatment choice among patients with localized PCa considering first-line treatment. MEDLINE, EMBASE, and Web of Science published from January 2010 to August 2024. Two investigators independently identified studies on risk classification and treatment choice after GC testing for patients with localized PCa considering first-line treatment. Relevant data extracted by 1 researcher and overread by a second. Risk of bias (ROB) was assessed in duplicate. Ten studies reported risk reclassification after GC testing. In low ROB observational studies, very low- or low-risk patients with PCa were more likely to have their risk levels classified as the same or lower (GPS, 100% to 88.1%; Decipher, 87.2% to 82.9%; Prolaris, 76.9%). However, 1 randomized trial found that GC testing with GPS reclassified 34.5% of very low-risk and 29.4% of low-risk patients to a higher risk category. Twelve observational studies indicated that treatment decisions after GC testing either remained unchanged or slightly favored active surveillance. In contrast, analyses from a single randomized trial found fewer choices for active surveillance after GPS testing. Heterogeneity in screening patterns, risk-determination cutoffs, pathology, and clinical practices. Studies on treatment choice were moderate to high ROB. Although GC tests do not consistently influence risk classification or treatment decisions, the differences observed between observational and randomized studies highlight a need for well-designed trials to explore the role of GC tests in patients with newly diagnosed PCa considering first-line treatment. U.S. Department of Veterans Affairs. (PROSPERO: CRD42022347950).

Multiple imputation (MI) models can be improved with auxiliary covariates (ACs), but their performance in high-dimensional data remains unclear. We aimed to develop and compare high-dimensional MI (HDMI) methods using structured and natural language processing (NLP)–derived AC in studies with partially observed confounders. We conducted a plasmode simulation with acute kidney injury as outcome and simulated 100 cohorts with a null treatment effect, incorporating creatinine labs, atrial fibrillation (AFib), and other investigator-derived confounders in the outcome generation. Missingness was imposed on creatinine based on creatinine itself and AFib. Different HDMI candidate ACs were created using structured and NLP-derived features, and we mimicked scenarios where AFib was unobserved by omitting it from all analyses. Using the least absolute shrinkage and selection operator, we selected HDMI covariates for MI and propensity score models. The treatment effect was estimated after propensity score matching in MI datasets, and HDMI methods were compared to baseline imputation and complete case analysis. High-dimensional MI using claims data showed the lowest bias (0.072). Combining claims and sentence embeddings led to an improvement in the efficiency with a root mean square error (RMSE) of 0.173 and 94% coverage. Natural language processing–derived AC alone did not outperform baseline MI. High-dimensional MI approaches may decrease bias in studies where confounder missingness depends on unobserved factors.

Missing data in confounding variables present a frequent challenge in generating evidence using real-world data, including electronic health records (EHR). Our objective was to apply a recently published toolkit for characterizing missing data patterns and based on the toolkit results about likely missingness mechanisms, illustrate the decision-making process for analyses in an empirical case example. We utilized the Structural Missing Data Investigations (SMDI) toolkit to characterize missing data patterns in the context of a pharmacoepidemiology study comparing cardiovascular outcomes of initiating sodium-glucose-cotransporter-2 inhibitors (SGLT2i) and dipeptidyl peptidase‐4 inhibitors (DPP‐4i) among older adults. The study used a linked EHR-Medicare claims dataset from Duke Health patients (2015–2017), focusing on partially observed confounders from EHR data (HbA1c lab and body mass index [BMI] values). Our analysis incorporated SMDI's descriptive functions and diagnostic tests to explore missingness patterns and determine missingness mitigation approaches. We used findings from these investigations to inform estimation of adjusted hazard ratios comparing the two classes of medications. High levels of missingness were noted for important confounding variables including HbA1c (63.6%) and BMI (16.5%). Diagnostic tests resulted in output that described: 1) the distributions of patient characteristics, exposure, and outcome between patients with or without an observed value of the partially observed covariate, 2) the ability to predict missingness based on observed covariates, and 3) estimate if the missingness of a partially observed covariate is differential with respect to the outcome. There was evidence that missingness could be sufficiently described using observed data, which allowed multiple imputation by chained equations using random forests to address missing confounder data in estimating treatment effects. Multiple imputation resulted in improved alignment of effect estimates with previous studies. We were able to demonstrate the practical application of the SMDI toolkit in a real-world setting. Application of the SMDI toolkit and the resulting insights of potential missingness patterns can inform the choice of appropriate analytic methods and increase transparency of research methods in handling missing data. This type of approach can inform analytic decision making and may increase our ability to generate evidence from real-world data.

Background: Some vaccines have a small risk of triggering Guillain–Barré syndrome (GBS), an autoimmune disorder where nerve damage leads to paralysis. There is a CDC precaution for patients whose GBS was associated with an influenza or tetanus toxoid‐containing vaccine (GBS occurring within 42 days following vaccination). Methods: We described vaccine patterns before and after a GBS diagnosis with a matched cohort design in a 20% random sample of fee‐for‐service Medicare enrollees. We defined the index date as an ICD‐9‐CM or ICD‐10‐CM GBS diagnosis code in the primary position of an inpatient claim. We matched each GBS patient to five non‐GBS comparators on sex, exact age, racial and ethnic category, state of residence and the month of preventive health visits during baseline; used weighting to balance covariates; and measured frequency of vaccines received per 100 people during year before and after the index date using the weighted mean cumulative count (wMCC). Results: We identified 1567 patients with a GBS diagnosis with at least 1 year of prior continuous enrollment in Medicare A and B that matched to five comparators each. The wMCCs in the 1 year before the index date were similar for both groups, with a wMCC of 74 vaccines/100 people in the GBS group (95% CI 71, 77). Within 1 year after the index date, patients with GBS had received 26 vaccines/100 people (95% CI 23, 28), which was 41 fewer vaccines than matched non‐GBS comparators (95% CI −44, −38). Among GBS patients, 11% were diagnosed with GBS within 42 days after a vaccine. Conclusions: GBS diagnosis has a strong impact on reducing subsequent vaccination even though there is no warning or precaution about future vaccines for most patients diagnosed with GBS. These data suggest discordance between clinical practice and current vaccine recommendations.

Background: Despite great promise, trials that ascertain patient clinical data from electronic health records (EHR), referred to here as “EHR-sourced” trials, are limited by uncertainty about how existing trial sites and infrastructure can be best used to operationalize study goals. Evidence is needed to support the practical use of EHRs in contemporary clinical trial settings. Main text: We describe a demonstration project that used EHR data to complement data collected for a contemporary multi-center pharmaceutical industry outcomes trial, and how a central coordinating center supported participating sites through the technical, governance, and operational aspects of this type of activity. We discuss operational considerations related to site selection, data extraction, site performance, and data transfer and quality review, and we outline challenges and lessons learned. We surveyed potential sites and used their responses to assess feasibility, determine the potential capabilities of sites and choose an appropriate data extraction strategy. We designed a flexible, multimodal approach for data extraction, enabling each site to either leverage an existing data source, create a new research datamart, or send all data to the central coordinating center to produce the requisite data elements. We evaluated site performance, as reflected by the speed of contracting and IRB approval, total patients enrolled, enrollment yield, data quality, and compared performance by data collection strategy. Conclusion: While broadening the type of sites able to participate in EHR-sourced trials may lead to greater generalizability and improved enrollment, sites with fewer technical resources may require additional support to participate. Central coordinating center support is essential to facilitate the execution of operational processes. Future work should focus on sharing lessons learned and creating reusable tools to facilitate participation of heterogeneous trial sites.

Background: Acute care inpatient admissions outside of psychiatric facilities have been increasingly identified as a critical touchpoint for opioid use disorder (OUD) treatment. We sought to describe non-opioid overdose hospitalizations with documented OUD and examine receipt of post-discharge outpatient buprenorphine. Methods: We examined acute care hospitalizations with an OUD diagnosis in any position within US commercially-insured adults age 18–64 years (IBM MarketScan claims, 2013–2017), excluding opioid overdose diagnoses. We included individuals with ≥ 6 months of continuous enrollment prior to the index hospitalization and ≥ 10 days following discharge. We described demographic and hospitalization characteristics, including outpatient buprenorphine receipt within 10 days of discharge. Results: Most (87%) hospitalizations with documented OUD did not include opioid overdose. Of 56,717 hospitalizations (49,959 individuals), 56.8% had a primary diagnosis other than OUD, 37.0% had documentation of an alcohol-related diagnosis code, and 5.8% end in a self-directed discharge. Where opioid use disorder was not the primary diagnosis, 36.5% were due to other substance use disorders, and 23.1% were due to psychiatric disorders. Of all non-overdose hospitalizations who had prescription medication insurance coverage and who were discharged to an outpatient setting (n = 49, 237), 8.8% filled an outpatient buprenorphine prescription within 10 days of discharge. Conclusions: Non-overdose OUD hospitalizations often occur with substance use disorders and psychiatric disorders, and very few are followed by timely outpatient buprenorphine. Addressing the OUD treatment gap during hospitalization may include implementing medication for OUD for inpatients with a broad range of diagnoses.

Background: As opioid‐related hospitalizations rise, hospitals must be prepared to evaluate and treat patients with opioid use disorder (OUD). We implemented a hospitalist‐led program, Project Caring for patients with Opioid Misuse through Evidence‐based Treatment (COMET) to address gaps in care for hospitalized patients with OUD.Objective: Implement evidence‐based treatment for inpatients with OUD and refer to postdischarge care.Design, Setting, and Participants: Project COMET launched in July 2019 at Duke University Hospital (DUH), an academic medical center in Durham, NC.Intervention, Main Outcomes, and Measures: We engaged key stakeholders, performed a needs assessment, and secured health system funding. We developed protocols to standardize OUD treatment and employed a social worker to facilitate postdischarge care. Electronic health records were utilized for data analysis.Results: COMET evaluated 512 patients for OUD during their index hospitalization from July 1, 2019 through June 30, 2021. Seventy‐one percent of patients received medication for OUD (MOUD) during admission. Of those who received buprenorphine during admission, 64% received a discharge prescription. Of those who received methadone during admission, 83% of eligible patients were connected to a methadone clinic. Among all patients at DUH with OUD, MOUD use during hospitalization and at discharge increased in the post‐COMET period compared to the pre‐COMET period (p < .001 for both).Conclusion: Our program is one of the first to demonstrate successful implementation of a hospitalist‐led, comprehensive approach to caring for hospitalized patients with OUD and can serve as an example to other institutions seeking to implement life‐saving, evidence‐based treatment in this population.

Objective: To empirically explore how pragmatic clinical trials (PCTs) that used real-world data (RWD) assessed study-specific fitness-for-use. Methods: We conducted interviews and surveys with PCT teams who used electronic health record (EHR) data to ascertain endpoints. The survey cataloged key concerns about RWD, activities used to assess data fitness-for-use, and related barriers encountered by study teams. Patterns and commonalities across trials were used to develop recommendations for study-specific fitness-for-use assessments. Results: Of 15 invited trial teams, 7 interviews were conducted. Of 31 invited trials, 15 responded to the survey. Most respondents had prior experience using RWD (93%). Major concerns about EHR data were data reliability, missingness or incompleteness of EHR elements, variation in data quality across study sites, and presence of implausible or incorrect values. Although many PCTs conducted fitness-for-use activities (eg, data quality assessments, 11/14, 79%), less than a quarter did so before choosing a data source. Fitness-for-use activities, findings, and resulting study design changes were not often publically documented. Overall costs and personnel costs were barriers to fitness-for-use assessments. Discussion: These results support three recommendations for PCTs that use EHR data for endpoint ascertainment. Trials should detail the rationale and plan for study-specific fitness-for-use activities, conduct study-specific fitness-for-use assessments early in the prestudy phase to inform study design changes before the trial begins, and share results of fitness-for-use assessments and description of relevant challenges and facilitators. Conclusion: These recommendations can help researchers and end-users of real-world evidence improve characterization of RWD reliability and relevance in the PCT-specific context.