Electronic Health Record-Based Algorithm for Monitoring Respiratory Virus-Like Illness.

Viral respiratory illness surveillance has traditionally focused on single pathogens (e.g., influenza) and required fever to identify influenza-like illness (ILI). We developed an automated system applying both laboratory test and syndrome criteria to electronic health records from 3 practice groups in Massachusetts, USA, to monitor trends in respiratory viral-like illness (RAVIOLI) across multiple pathogens. We identified RAVIOLI syndrome using diagnosis codes associated with respiratory viral testing or positive respiratory viral assays or fever. After retrospectively applying RAVIOLI criteria to electronic health records, we observed annual winter peaks during 2015-2019, predominantly caused by influenza, followed by cyclic peaks corresponding to SARS-CoV-2 surges during 2020-2024, spikes in RSV in mid-2021 and late 2022, and recrudescent influenza in late 2022 and 2023. RAVIOLI rates were higher and fluctuations more pronounced compared with traditional ILI surveillance. RAVIOLI broadens the scope, granularity, sensitivity, and specificity of respiratory viral illness surveillance compared with traditional ILI surveillance.

Electronic Health Record Data for Lyme Disease Surveillance, Massachusetts, USA, 2017-2018.

Lyme disease surveillance based on provider and laboratory reports underestimates incidence. We developed an algorithm for automating surveillance using electronic health record data. We identified potential Lyme disease markers in electronic health record data (laboratory tests, diagnosis codes, prescriptions) from January 2017-December 2018 in 2 large practice groups in Massachusetts, USA. We calculated their sensitivities and positive predictive values (PPV), alone and in combination, relative to medical record review. Sensitivities ranged from 57% (95% CI 47%-69%) for immunoassays to 87% (95% CI 70%-100%) for diagnosis codes. PPVs ranged from 53% (95% CI 43%-61%) for diagnosis codes to 58% (95% CI 50%-66%) for immunoassays. The combination of a diagnosis code and antibiotics within 14 days or a positive Western blot had a sensitivity of 100% (95% CI 86%-100%) and PPV of 82% (95% CI 75%-89%). This algorithm could make Lyme disease surveillance more efficient and consistent.

Validity of inpatient electronic health record-based measures of oxygen-related therapy in the United States: Lessons applicable for studying COVID-19 endpoints.

INTRODUCTION: During the COVID-19 pandemic, inpatient electronic health records (EHRs) have been used to conduct public health surveillance and assess treatments and outcomes. Invasive mechanical ventilation (MV) and supplemental oxygen (O2) use are markers of severe illness in hospitalized COVID-19 patients. In a large US system (n = 142 hospitals), we assessed documentation of MV and O2 use during COVID-19 hospitalization in administrative data versus nursing documentation. METHODS: We identified 319 553 adult hospitalizations with a COVID-19 diagnosis, February 2020-October 2022, and extracted coded, administrative data for MV or O2. Separately, we developed classification rules for MV or O2 supplementation from semi-structured nursing documentation. We assessed MV and O2 supplementation in administrative data versus nursing documentation and calculated ordinal endpoints of decreasing COVID-19 disease severity. Nursing documentation was considered the gold standard in sensitivity and positive predictive value (PPV) analyses. RESULTS: In nursing documentation, the prevalence of MV and O2 supplementation among COVID-19 hospitalizations was 14% and 75%, respectively. The sensitivity of administrative data was 83% for MV and 41% for O2, with both PPVs above 91%. Concordance between sources was 97% for MV (κ = 0.85), and 54% for O2 (κ = 0.21). For ordinal endpoints, administrative data accurately identified intensive care and MV but underestimated hospitalizations with O2 requirements (42% vs. 18%). CONCLUSIONS: In comparison to nursing documentation, administrative data under-ascertained O2 supplementation but accurately estimated severe endpoints such as MV. Nursing documentation improved ascertainment of O2 among COVID-19 hospitalizations and can capture oxygen requirements in adults hospitalized with COVID-19 or other respiratory illnesses.

Validation of Claims-Based Algorithm for Lyme Disease, Massachusetts, USA.

Compared with notifiable disease surveillance, claims-based algorithms estimate higher Lyme disease incidence, but their accuracy is unknown. We applied a previously developed Lyme disease algorithm (diagnosis code plus antimicrobial drug prescription dispensing within 30 days) to an administrative claims database in Massachusetts, USA, to identify a Lyme disease cohort during July 2000-June 2019. Clinicians reviewed and adjudicated medical charts from a cohort subset by using national surveillance case definitions. We calculated positive predictive values (PPVs). We identified 12,229 Lyme disease episodes in the claims database and reviewed and adjudicated 128 medical charts. The algorithm's PPV for confirmed, probable, or suspected cases was 93.8% (95% CI 88.1%-97.3%); the PPV was 66.4% (95% CI 57.5%-74.5%) for confirmed and probable cases only. In a high incidence setting, a claims-based algorithm identified cases with a high PPV, suggesting it can be used to assess Lyme disease burden and supplement traditional surveillance data.

Pragmatic guidance for embedding pragmatic clinical trials in health plans: Large simple trials aren't so simple.

BACKGROUND: There are unique opportunities related to the design and conduct of pragmatic trials embedded in health insurance plans, which have longitudinal data on member/patient demographics, dates of coverage, and reimbursed medical care, including prescription drug dispensings, vaccine administrations, behavioral healthcare encounters, and some laboratory results. Such trials can be large and efficient, using these data to identify trial-eligible patients and to ascertain outcomes. METHODS: We use our experience primarily with the National Institutes of Health Pragmatic Trials Collaboratory Distributed Research Network, which comprises health plans that participate in the US Food & Drug Administration's Sentinel System, to describe lessons learned from the conduct and planning of embedded pragmatic trials. RESULTS: Information is available for research on more than 75 million people with commercial or Medicare Advantage health plans. We describe three studies that have used or plan to use the Network, as well as a single health plan study, from which we glean our lessons learned. CONCLUSIONS: Studies that are conducted in health plans provide much-needed evidence to drive clinically meaningful changes in care. However, there are many unique aspects of these trials that must be considered in the planning, implementation, and analytic phases. The type of trial best suited for studies embedded in health plans will be those that require large sample sizes, simple interventions that could be disseminated through health plans, and where data available to the health plan can be leveraged. These trials hold potential for substantial long-term impact on our ability to generate evidence to improve care and population health.

Leveraging Electronic Health Record Data for Timely Chronic Disease Surveillance: The Multi-State EHR-Based Network for Disease Surveillance.

CONTEXT: Electronic health record (EHR) data can potentially make chronic disease surveillance more timely, actionable, and sustainable. Although use of EHR data can address numerous limitations of traditional surveillance methods, timely surveillance data with broad population coverage require scalable systems. This report describes implementation, challenges, and lessons learned from the Multi-State EHR-Based Network for Disease Surveillance (MENDS) to help inform how others work with EHR data to develop distributed networks for surveillance. PROGRAM: Funded by the Centers for Disease Control and Prevention (CDC), MENDS is a data modernization demonstration project that aims to develop a timely national chronic disease sentinel surveillance system using EHR data. It facilitates partnerships between data contributors (health information exchanges, other data aggregators) and data users (state and local health departments). MENDS uses query and visualization software to track local emerging trends. The program also uses statistical and geospatial methods to generate prevalence estimates of chronic disease risk measures at the national and local levels. Resulting data products are designed to inform public health practice and improve the health of the population. IMPLEMENTATION: MENDS includes 5 partner sites that leverage EHR data from 91 health system and clinic partners and represents approximately 10 million patients across the United States. Key areas of implementation include governance, partnerships, technical infrastructure and support, chronic disease algorithms and validation, weighting and modeling, and workforce education for public health data users. DISCUSSION: MENDS presents a scalable distributed network model for implementing national chronic disease surveillance that leverages EHR data. Priorities as MENDS matures include producing prevalence estimates at various geographic and subpopulation levels, developing enhanced data sharing and interoperability capacity using international data standards, scaling the network to improve coverage nationally and among underrepresented geographic areas and subpopulations, and expanding surveillance of additional chronic disease measures and social determinants of health.

Evaluating Confounding Control in Estimations of Influenza Antiviral Effectiveness in Electronic Health Plan Data.

Observational studies of oseltamivir use and influenza complications could suffer from residual confounding. Using negative control risk periods and a negative control outcome, we examined confounding control in a health-insurance-claims-based study of oseltamivir and influenza complications (pneumonia, all-cause hospitalization, and dispensing of an antibiotic). Within the Food and Drug Administration's Sentinel System, we identified individuals aged ≥18 years who initiated oseltamivir use on the influenza diagnosis date versus those who did not, during 3 influenza seasons (2014-2017). We evaluated primary outcomes within the following 1-30 days (the primary risk period) and 61-90 days (the negative control period) and nonvertebral fractures (the negative control outcome) within days 1-30. We estimated propensity-score-matched risk ratios (RRs) per season. During the 2014-2015 influenza season, oseltamivir use was associated with a reduction in the risk of pneumonia (RR = 0.72, 95% confidence interval (CI): 0.70, 0.75) and all-cause hospitalization (RR = 0.54, 95% CI: 0.53, 0.55) in days 1-30. During days 61-90, estimates were near-null for pneumonia (RR = 1.04, 95% CI: 0.95, 1.15) and hospitalization (RR = 0.94, 95% CI: 0.91, 0.98) but slightly increased for antibiotic dispensing (RR = 1.14, 95% CI: 1.08, 1.21). The RR for fractures was near-null (RR = 1.09, 95% CI: 0.99, 1.20). Estimates for the 2016-2017 influenza season were comparable, while the 2015-2016 season had conflicting results. Our study suggests minimal residual confounding for specific outcomes, but results differed by season.

Validation of diagnosis codes to identify hospitalized COVID-19 patients in health care claims data.

PURPOSE: Health plan claims may provide complete longitudinal data for timely, real-world population-level COVID-19 assessment. However, these data often lack laboratory results, the standard for COVID-19 diagnosis. METHODS: We assessed the validity of ICD-10-CM diagnosis codes for identifying patients hospitalized with COVID-19 in U.S. claims databases, compared to linked laboratory results, among six Food and Drug Administration Sentinel System data partners (two large national insurers, four integrated delivery systems) from February 20-October 17, 2020. We identified patients hospitalized with COVID-19 according to five ICD-10-CM diagnosis code-based algorithms, which included combinations of codes U07.1, B97.29, general coronavirus codes, and diagnosis codes for severe symptoms. We calculated the positive predictive value (PPV) and sensitivity of each algorithm relative to laboratory test results. We stratified results by data source type and across three time periods: February 20-March 31 (Time A), April 1-30 (Time B), May 1-October 17 (Time C). RESULTS: The five algorithms identified between 34 806 and 47 293 patients across the study periods; 23% with known laboratory results contributed to PPV calculations. PPVs were high and similar across algorithms. PPV of U07.1 alone was stable around 93% for integrated delivery systems, but declined over time from 93% to 70% among national insurers. Overall PPV of U07.1 across all data partners was 94.1% (95% CI, 92.3%-95.5%) in Time A and 81.2% (95% CI, 80.1%-82.2%) in Time C. Sensitivity was consistent across algorithms and over time, at 94.9% (95% CI, 94.2%-95.5%). CONCLUSION: Our results support the use of code U07.1 to identify hospitalized COVID-19 patients in U.S. claims data.

Association of COVID-19 vs Influenza With Risk of Arterial and Venous Thrombotic Events Among Hospitalized Patients.

Importance: The incidence of arterial thromboembolism and venous thromboembolism in persons with COVID-19 remains unclear. Objective: To measure the 90-day risk of arterial thromboembolism and venous thromboembolism in patients hospitalized with COVID-19 before or during COVID-19 vaccine availability vs patients hospitalized with influenza. Design, Setting, and Participants: Retrospective cohort study of 41 443 patients hospitalized with COVID-19 before vaccine availability (April-November 2020), 44 194 patients hospitalized with COVID-19 during vaccine availability (December 2020-May 2021), and 8269 patients hospitalized with influenza (October 2018-April 2019) in the US Food and Drug Administration Sentinel System (data from 2 national health insurers and 4 regional integrated health systems). Exposures: COVID-19 or influenza (identified by hospital diagnosis or nucleic acid test). Main Outcomes and Measures: Hospital diagnosis of arterial thromboembolism (acute myocardial infarction or ischemic stroke) and venous thromboembolism (deep vein thrombosis or pulmonary embolism) within 90 days. Outcomes were ascertained through July 2019 for patients with influenza and through August 2021 for patients with COVID-19. Propensity scores with fine stratification were developed to account for differences between the influenza and COVID-19 cohorts. Weighted Cox regression was used to estimate the adjusted hazard ratios (HRs) for outcomes during each COVID-19 vaccine availability period vs the influenza period. Results: A total of 85 637 patients with COVID-19 (mean age, 72 [SD, 13.0] years; 50.5% were male) and 8269 with influenza (mean age, 72 [SD, 13.3] years; 45.0% were male) were included. The 90-day absolute risk of arterial thromboembolism was 14.4% (95% CI, 13.6%-15.2%) in patients with influenza vs 15.8% (95% CI, 15.5%-16.2%) in patients with COVID-19 before vaccine availability (risk difference, 1.4% [95% CI, 1.0%-2.3%]) and 16.3% (95% CI, 16.0%-16.6%) in patients with COVID-19 during vaccine availability (risk difference, 1.9% [95% CI, 1.1%-2.7%]). Compared with patients with influenza, the risk of arterial thromboembolism was not significantly higher among patients with COVID-19 before vaccine availability (adjusted HR, 1.04 [95% CI, 0.97-1.11]) or during vaccine availability (adjusted HR, 1.07 [95% CI, 1.00-1.14]). The 90-day absolute risk of venous thromboembolism was 5.3% (95% CI, 4.9%-5.8%) in patients with influenza vs 9.5% (95% CI, 9.2%-9.7%) in patients with COVID-19 before vaccine availability (risk difference, 4.1% [95% CI, 3.6%-4.7%]) and 10.9% (95% CI, 10.6%-11.1%) in patients with COVID-19 during vaccine availability (risk difference, 5.5% [95% CI, 5.0%-6.1%]). Compared with patients with influenza, the risk of venous thromboembolism was significantly higher among patients with COVID-19 before vaccine availability (adjusted HR, 1.60 [95% CI, 1.43-1.79]) and during vaccine availability (adjusted HR, 1.89 [95% CI, 1.68-2.12]). Conclusions and Relevance: Based on data from a US public health surveillance system, hospitalization with COVID-19 before and during vaccine availability, vs hospitalization with influenza in 2018-2019, was significantly associated with a higher risk of venous thromboembolism within 90 days, but there was no significant difference in the risk of arterial thromboembolism within 90 days.

Can Machine Learning Help Identify Patients at Risk for Recurrent Sexually Transmitted Infections?

BACKGROUND: A substantial fraction of sexually transmitted infections (STIs) occur in patients who have previously been treated for an STI. We assessed whether routine electronic health record (EHR) data can predict which patients presenting with an incident STI are at greatest risk for additional STIs in the next 1 to 2 years. METHODS: We used structured EHR data on patients 15 years or older who acquired an incident STI diagnosis in 2008 to 2015 in eastern Massachusetts. We applied machine learning algorithms to model risk of acquiring ≥1 or ≥2 additional STIs diagnoses within 365 or 730 days after the initial diagnosis using more than 180 different EHR variables. We performed sensitivity analysis incorporating state health department surveillance data to assess whether improving the accuracy of identifying STI cases improved algorithm performance. RESULTS: We identified 8723 incident episodes of laboratory-confirmed gonorrhea, chlamydia, or syphilis. Bayesian Additive Regression Trees, the best-performing algorithm of any single method, had a cross-validated area under the receiver operating curve of 0.75. Receiver operating curves for this algorithm showed a poor balance between sensitivity and positive predictive value (PPV). A predictive probability threshold with a sensitivity of 91.5% had a corresponding PPV of 3.9%. A higher threshold with a PPV of 29.5% had a sensitivity of 11.7%. Attempting to improve the classification of patients with and without repeat STIs diagnoses by incorporating health department surveillance data had minimal impact on cross-validated area under the receiver operating curve. CONCLUSIONS: Machine algorithms using structured EHR data did not differentiate well between patients with and without repeat STIs diagnosis. Alternative strategies, able to account for sociobehavioral characteristics, could be explored.

RiskScape: A Data Visualization and Aggregation Platform for Public Health Surveillance Using Routine Electronic Health Record Data.

Automated analysis of electronic health record (EHR) data is a complementary tool for public health surveillance. Analyzing and presenting these data, however, demands new methods of data communication optimized to the detail, flexibility, and timeliness of EHR data.RiskScape is an open-source, interactive, Web-based, user-friendly data aggregation and visualization platform for public health surveillance using EHR data. RiskScape displays near-real-time surveillance data and enables clinical practices and health departments to review, analyze, map, and trend aggregate data on chronic conditions and infectious diseases. Data presentations include heat maps of prevalence by zip code, time series with statistics for trends, and care cascades for conditions such as HIV and HCV. The platform's flexibility enables it to be modified to incorporate new conditions quickly-such as COVID-19.The Massachusetts Department of Public Health (MDPH) uses RiskScape to monitor conditions of interest using data that are updated monthly from clinical practice groups that cover approximately 20% of the state population. RiskScape serves an essential role in demonstrating need and burden for MDPH's applications for funding, particularly through the identification of inequitably burdened populations.

A COVID-19-ready public health surveillance system: The Food and Drug Administration's Sentinel System.

The US Food and Drug Administration's Sentinel System was established in 2009 to use routinely collected electronic health data for improving the national capability to assess post-market medical product safety. Over more than a decade, Sentinel has become an integral part of FDA's surveillance capabilities and has been used to conduct analyses that have contributed to regulatory decisions. FDA's role in the COVID-19 pandemic response has necessitated an expansion and enhancement of Sentinel. Here we describe how the Sentinel System has supported FDA's response to the COVID-19 pandemic. We highlight new capabilities developed, key data generated to date, and lessons learned, particularly with respect to working with inpatient electronic health record data. Early in the pandemic, Sentinel developed a multi-pronged approach to support FDA's anticipated data and analytic needs. It incorporated new data sources, created a rapidly refreshed database, developed protocols to assess the natural history of COVID-19, validated a diagnosis-code based algorithm for identifying patients with COVID-19 in administrative claims data, and coordinated with other national and international initiatives. Sentinel is poised to answer important questions about the natural history of COVID-19 and is positioned to use this information to study the use, safety, and potentially the effectiveness of medical products used for COVID-19 prevention and treatment.

Identifying prescribing cascades in Alzheimer's disease and related dementias: The calcium channel blocker-diuretic prescribing cascade.

PURPOSE: Prescribing cascades occur when a physician prescribes a new drug to address the side-effect of another drug. Persons with Alzheimer's disease and related dementias (ADRD) are at increased risk for prescribing cascades. Our objective was to develop an approach to estimating the proportion of calcium channel blocker-diuretic (CCB-diuretic) prescribing cascades among persons with ADRD in two U.S. health plans. METHODS: We identified patients aged ≥50 on January 1, 2017, dispensed a drug to treat ADRD in the 365-days prior to/on cohort entry date. Patients had medical/pharmacy coverage for 1 year before and through cohort entry. We excluded individuals with an institutional stay encounter in the 45 days prior to cohort entry and censored patients based on: disenrollment from coverage, death, or end of data. We identified incident and prevalent CCB use in the 183-days following cohort entry, and identified subsequent incident diuretic use among incident and prevalent CCB-users within 365-days from cohort entry. RESULTS: There were 121 538 eligible patients. Approximately 62% were female, with a mean age of 79.5 (SD ±8.6). Overall 2.1% of the cohort experienced a prevalent CCB-diuretic prescribing cascade with 1586 incident diuretic-users among 36 462 prevalent CCB-users (4.3%, 95% CI 4.1-4.6%]); and there were161 incident diuretic-users among 3304 incident CCB-users (4.9%, 95% CI 4.2-5.7%) (incident CCB-diuretic cascade). CONCLUSIONS: We describe an approach to identify prescribing cascades in persons with ADRD, which can be used to assess the proportion of prescribing cascades in large cohorts. We determined the proportion of CCB-diuretic prescribing cascades was low.

Validation of an electronic algorithm for Hodgkin and non-Hodgkin lymphoma in ICD-10-CM.

PURPOSE: Lymphoma is a health outcome of interest for drug safety studies. Studies using administrative claims data require the accurate identification of lymphoma cases. We developed and validated an International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM)-based algorithm to identify lymphoma in healthcare claims data. METHODS: We developed a three-component algorithm to identify patients aged ≥15 years who were newly diagnosed with Hodgkin (HL) or non-Hodgkin (NHL) lymphoma from January 2016 through July 2018 among members of four Data Partners within the FDA's Sentinel System. The algorithm identified potential cases as patients with ≥2 ICD-10-CM lymphoma diagnosis codes on different dates within 183 days; ≥1 procedure code for a diagnostic procedure (e.g., biopsy, flow cytometry) and ≥1 procedure code for a relevant imaging study within 90 days of the first lymphoma diagnosis code. Cases identified by the algorithm were adjudicated via chart review and a positive predictive value (PPV) was calculated. RESULTS: We identified 8723 potential lymphoma cases via the algorithm and randomly sampled 213 for validation. We retrieved 138 charts (65%) and adjudicated 134 (63%). The overall PPV was 77% (95% confidence interval: 69%-84%). Most cases also had subtype information available, with 88% of cases identified as NHL and 11% as HL. CONCLUSIONS: Seventy-seven percent of lymphoma cases identified by an algorithm based on ICD-10-CM diagnosis and procedure codes and applied to claims data were true cases. This novel algorithm represents an efficient, cost-effective way to target an important health outcome of interest for large-scale drug safety and public health surveillance studies.

Leveraging the Capabilities of the FDA's Sentinel System To Improve Kidney Care.

The Sentinel System is a national electronic postmarketing resource established by the US Food and Drug Administration to support assessment of the safety and effectiveness of marketed medical products. It has built a large, multi-institutional, distributed data network that contains comprehensive electronic health data, covering about 700 million person-years of longitudinal observation time nationwide. With its sophisticated infrastructure and a large selection of flexible analytic tools, the Sentinel System permits rapid and secure analyses, while preserving patient privacy and health-system autonomy. The Sentinel System also offers enhanced capabilities, including accessing full-text medical records, supporting randomized clinical trials embedded in healthcare delivery systems, and facilitating effective collection of patient-reported data using mobile devices, among many other research programs. The nephrology research community can use the infrastructure, tools, and data that this national resource offers for evidence generation. This review summarizes the Sentinel System and its ability to rapidly generate high-quality, real-world evidence; discusses the program's experience in, and potential for, addressing gaps in kidney care; and outlines avenues for conducting research, leveraging this national resource in collaboration with Sentinel investigators.

More Screening or More Disease? Gonorrhea Testing and Positivity Patterns Among Men in 3 Large Clinical Practices in Massachusetts, 2010-2017.

BACKGROUND: Gonorrhea diagnosis rates in the United States increased by 75% during 2009-2017, predominantly in men. It is unclear whether the increase among men is being driven by more screening, an increase in the prevalence of disease, or both. We sought to evaluate changes in gonorrhea testing patterns and positivity among men in Massachusetts. METHODS: The analysis included men (aged ≥15 years) who received care during 2010-2017 in 3 clinical practice groups. We calculated annual percentages of men with ≥1 gonorrhea test and men with ≥1 positive result, among men tested. Log-binomial regression models were used to examine trends in these outcomes. We adjusted for clinical and demographic characteristics that may influence the predilection to test and probability of gonorrhea disease. RESULTS: On average, 306 348 men had clinical encounters each year. There was a significant increase in men with ≥1 gonorrhea test from 2010 (3.1%) to 2017 (6.4%; adjusted annual risk ratio, 1.12; 95% confidence interval, 1.12-1.13). There was a significant, albeit lesser, increase in the percentage of tested men with ≥1 positive result (1.0% in 2010 to 1.5% in 2017; adjusted annual risk ratio, 1.07; 95% confidence interval, 1.04-1.09). CONCLUSIONS: We estimated significant increases in the annual percentages of men with ≥1 gonorrhea test and men with ≥1 positive gonorrhea test result between 2010 and 2017. These results suggest that observed increases in gonorrhea rates could be explained by both increases in screening and the prevalence of gonorrhea.

Underuse of oral anticoagulants in privately insured patients with atrial fibrillation: A population being targeted by the IMplementation of a randomized controlled trial to imProve treatment with oral AntiCoagulanTs in patients with Atrial Fibrillation (IMPACT-AFib).

BACKGROUND: Many studies showing underuse of oral anticoagulants (OACs) in patients with atrial fibrillation (AF) predated the advent of the non-vitamin K antagonist OACs. We retrospectively examined use of OACs in a large commercially insured population. METHODS: Administrative claims data from 4 research partners participating in FDA-Catalyst, a program of the Sentinel Initiative, were queried in September 2017. Patients were included if they were ≥30 years old with ≥365 days of medical/pharmacy coverage, and had ≥2 diagnosis codes for AF, a CHA2DS2-VASc score ≥2, absence of contraindications to OAC use, and no evidence of OAC use in the 365 days before the index AF diagnosis. The main outcome measures of the current analysis were rates of OAC use in the prior 12 months of cohort identification and factors associated with non-use. RESULTS: A total of 197,806 AF patients met the eligibility criteria prior to assessment of OAC treatment. Of these, 179,580 (91%) patients were ≥65 years old and 73,286 (37%) patients were ≥80 years old. Half of the patients (98,903) were randomized to the early intervention arm in the IMPACT-AFib trial and constitute the cohort for this analysis. Of these, 32,295 (33%) had no evidence of OAC use in the prior 12 months. Compared with patients with evidence of OAC use in the prior 12 months, patients without OAC use were more likely to be ≥80 years old, women, and have a history of anemia (51% vs 47%) and less likely to have diabetes (41% vs 44%), history of stroke or TIA (15% vs 19%), and history of heart failure (39% vs 48%). CONCLUSIONS: Despite a high risk of stroke, one-third of privately insured patients with AF and no obvious contraindications to an OAC were not treated with an OAC. There is an unmet need for evidence-based interventions that could lead to greater use of OACs in patients with AF at risk for stroke.

Using electronic health records to identify candidates for human immunodeficiency virus pre-exposure prophylaxis: An application of super learning to risk prediction when the outcome is rare.

Human immunodeficiency virus (HIV) pre-exposure prophylaxis (PrEP) protects high risk patients from becoming infected with HIV. Clinicians need help to identify candidates for PrEP based on information routinely collected in electronic health records (EHRs). The greatest statistical challenge in developing a risk prediction model is that acquisition is extremely rare. METHODS: Data consisted of 180 covariates (demographic, diagnoses, treatments, prescriptions) extracted from records on 399 385 patient (150 cases) seen at Atrius Health (2007-2015), a clinical network in Massachusetts. Super learner is an ensemble machine learning algorithm that uses k-fold cross validation to evaluate and combine predictions from a collection of algorithms. We trained 42 variants of sophisticated algorithms, using different sampling schemes that more evenly balanced the ratio of cases to controls. We compared super learner's cross validated area under the receiver operating curve (cv-AUC) with that of each individual algorithm. RESULTS: The least absolute shrinkage and selection operator (lasso) using a 1:20 class ratio outperformed the super learner (cv-AUC = 0.86 vs 0.84). A traditional logistic regression model restricted to 23 clinician-selected main terms was slightly inferior (cv-AUC = 0.81). CONCLUSION: Machine learning was successful at developing a model to predict 1-year risk of acquiring HIV based on a physician-curated set of predictors extracted from EHRs.

Practical challenges in the conduct of pragmatic trials embedded in health plans: Lessons of IMPACT-AFib, an FDA-Catalyst trial.

IMPACT-AFib was an 80,000-patient randomized clinical trial implemented by five US insurance companies (health plans) aimed at increasing the use of oral anticoagulants by individuals with atrial fibrillation who were at high risk of stroke and not on treatment. The underlying thesis was that patients could be change agents to initiate prescribing discussions with their providers. We tested the effect of mailing information to both patients and their providers. We used administrative medical claims and pharmacy dispensing data to identify eligible patients, to randomize them to an early or delayed intervention, and to assess clinical outcomes. The core data were analysis-ready datasets each site had created and curated for the FDA's Sentinel System, supplemented by updated "fresh" pharmacy and enrollment data to ensure eligibility at the time of intervention. Following mutually agreed upon procedures, sites linked to additional internal source data to implement the intervention-educational information mailed to patients and their providers in the early intervention arm, and to providers of patients in the delayed intervention arm approximately 12 months later. The primary analysis compares the early intervention arm to the delayed intervention arm, prior to the delayed intervention being conducted (i.e. compares intervention to non-intervention). The endpoints of interest were evidence of initiation of anticoagulation (primary) as well as clinical endpoints, including stroke and hospitalization for bleeding. Major challenges, some unanticipated, identified during the planning phase include convening multi-stakeholder investigator teams and advisors, addressing ethical concerns about not intervening in a usual care comparison group, and identifying and avoiding interference with sites' routine programs that were similar to the intervention. Needs and challenges during the implementation phase included the fact that even limited site-specific programming greatly increased time and effort, the need to refresh research data extracts immediately before outreach to patients and providers, potential difficulty identifying low-cost medications such as warfarin that may not be reimbursed by health plans and so not discoverable in dispensing data, the need to develop workarounds when "providers" in claims data were facilities, difficulty addressing clustering of patients by provider because providers can have multiple identifiers within and between health plans, and the need to anticipate loss to follow up because of health plan disenrollment or change in benefits. As pragmatic trials begin to shape evidence generation within clinical practice, investigators should anticipate issues inherent to claims data and working with multiple large sites. In IMPACT-AFib, we found that investing in collaboration and communication among all parties throughout all phases of the study helped ensure common understanding, early identification of challenges, and streamlined actual implementation.

Using HIV Surveillance Laboratory Data to Identify Out-of-Care Patients.

HIV-associated laboratory tests reported to public health surveillance have been used as a proxy measure of care engagement of HIV+ individuals. As part of a Health Resources and Services Administration (HRSA) Special Projects of National Significance (SPNS) Initiative, the Massachusetts Department of Public Health (MDPH) worked with three pilot clinical facilities to identify HIV+ patients whose last HIV laboratory test occurred at the participating facility but who then appeared to be out of care, defined as an absence of HIV laboratory test results reported to MDPH for at least 6 months. The clinical facilities then reviewed medical records to determine whether these patients were actually not in care, or if there was another reason that they did not have a laboratory test performed, and provided feedback to MDPH on each of the presumed out-of-care patients. In the first year of the pilot project, 37% of patients who appeared to be out of care based on laboratory data were confirmed to be out of care after review of clinical health records. Of those patients who were confirmed to be out of care, 55% had a subsequent laboratory test within 3 months, and 72% had a laboratory test within 6 months, indicating that they had re-engaged with a care provider. MDPH found that it was essential to have clinical staff confirm the care status of patients who were presumed to be out of care based on surveillance data.