Multiple Electronic Health Record Research Articles

Sexual orientation and gender identity measures and viral suppression for people living with HIV: a protocol for a population-based cohort study

IntroductionThe measure of sexual orientation and gender identity (SOGI) data in electronic health records (EHR) has been critical for addressing health disparities and inequalities, especially for HIV care. Given that...

Linking Patient Encounters across Primary and Ancillary Electronic Health Record Systems: A Comparison of Two Approaches.

To achieve scientific goals, researchers often require integration of data from a primary electronic health record (EHR) system and one or more ancillary EHR systems used during the same patient care encounter. Although studies have demonstrated approaches for linking patient identity records across different EHR systems, little is known about linking patient encounter records across primary and ancillary EHR systems. We compared a patients-first approach versus an encounters-first approach for linking patient encounter records across multiple EHR systems. We conducted a retrospective observational study of 348,904 patients with 533,283 encounters from 2010 to 2020 across our institution's primary EHR system and an ancillary EHR system used in perioperative settings. For the patients-first approach and the encounters-first approach, we measured the number of patient and encounter links created as well as runtime. While the patients-first approach linked 43% of patients and 49% of encounters, the encounters-first approach linked 98% of patients and 100% of encounters. The encounters-first approach was 20 times faster than the patients-first approach for linking patients and 33% slower for linking encounters. Findings suggest that common patient and encounter identifiers shared among EHR systems via automated interfaces may be clinically useful but not "research-ready" and thus require an encounters-first linkage approach to enable secondary use for scientific purposes. Based on our search, this study is among the first to demonstrate approaches for linking patient encounters across multiple EHR systems. Enterprise data warehouse for research efforts elsewhere may benefit from an encounters-first approach.

A Genotype Validated Bimodal Method for the Large-Scale Identification and Phenotyping of Persons with Sickle Cell Disease Using Electronic Health Record Data

Introduction Clinical trials for using hematopoietic stem cell transplant (HSCT) with gene therapies to cure sickle cell disease (SCD) began in 2019. The Food and Drug Administration (FDA) requires that 1) all gene therapy participants receive at least 15 years of ongoing surveillance and 2) their outcomes be compared to a contemporaneous cohort that did not receive gene therapy. Previously we developed an automated contemporaneous cohort of children and adults with SCD from electronic health record (EHR) data (Cronin RM et al. 2023, Blood Adv). Our work was built upon previously explored algorithms utilizing ICD codes and laboratory data to identify persons with sickle cell anemia (SCA), a severe form of SCD (Singh et al. 2018, Blood Adv). Still, our algorithm additionally classifies other predicted genotypes. However, we could not test the SCD phenotypes against the SCD genotypes due to the absence of beta-globin gene sequence data. Here we tested the hypothesis that our Vanderbilt-EHR algorithm for identifying a contemporaneous cohort at Vanderbilt can classify their predicted genotype compared to beta-globin gene sequencing, the gold standard for SCD diagnosis, in the cohort. We subsequently tested the Vanderbilt-EHR algorithm in an established cohort of children from St. Jude Children's Research Hospital that have been deeply phenotyped. Methods We applied the Vanderbilt-EHR algorithm to a cohort of 275 samples from Vanderbilt's biorepository of DNA, BioVU, linked to de-identified medical records in a data warehouse called the Synthetic Derivative. These 275 samples were previously submitted for whole genome sequencing as part of the Genetic Variation of Heart, Lung, and Kidney Disease in Sickle Cell Disease: Pre- and Post-Curative Therapies TOPMed project (HLK-SCD phs002617). The returned genotypes were validated using the hemoglobin variant database HbVar (Giardine BM et al. 2021, Nucleic Acids Res) (Table 1). We applied the hemoglobinopathy genotype classification portion of the algorithm to the Sickle Cell Clinical Research and Intervention Program (SCCRIP) cohort from St. Jude Children's Research Hospital, which includes putative genotype diagnoses derived from a comprehensive evaluation of clinical data for each member. We performed statistical analyses of the identification and classification results of the Vanderbilt cohort and the SCCRIP cohort's classification results. Results Of the 275 genotyped samples, the algorithm correctly predicted 255 SCD cases and 10 non-SCD cases. There were three cases predicted to be false positives; however, a comprehensive analysis showed that all have laboratory values indicative of SCA (high hemoglobin S levels >60%, absent or negligible hemoglobin A levels, elevated reticulocytes, and low mean corpuscular volume) suggesting that there may be a genotyping error. The remaining 7 cases were either indeterminate in the prediction or in the genotype (Table 1). Given the small number of indeterminate results, we performed a sensitivity analysis to identify the range of performance for SCD identification. The method performed very well with over 98% sensitivity and over 96% positive predictive value (PPV) (Table 2). The lower specificity and negative predictive value (NPV) could be largely attributed to the disproportionately high number of SCD cases in the dataset compared to non-SCD cases. The SCD classification algorithm performed similarly in both the VUMC and SCCRIP cohorts (Table 2). The algorithm performs best in identifying SC and SCA types of SCD with nearly 100% accuracy for SC and over 94% accuracy for SCA. Transfusions can confound the diagnosis of SCA vs. S beta thalassemia +, but the latter has a much lower prevalence consistent with our results (Table 2). Discussion and Conclusions Larger SCD cohorts will be identified using these approaches in singular EHR databases and in de-identified data warehouses that EHR companies have developed by aggregating data across multiple EHR systems (e.g., EPIC Cosmos and Cerner Real-World Data). The ongoing need for contemporaneous comparison cohorts of persons with SCD to understand outcomes related to treatment and phenotype can be greatly assisted by accurate automated approaches. The Vanderbilt-EHR algorithm is the first to comprehensively phenotype SCD and other hemoglobinopathies involving variant hemoglobin beta alleles.

Implementing a BRAF biomarker electronic clinical quality measure in CancerLinQ.

418 Background: Achieving the benefits of precision oncology is dependent on timely and appropriate biomarker testing. ASCO received funding from the Gordon & Betty Moore Foundation’s Diagnostic Excellence Initiative to develop and test a biomarker electronic Clinical Quality Measure (eCQM) in colorectal cancer for implementation in SmartLinQ, CancerLinQ’s quality platform. Methods: Working collaboratively with a Technical Expert Panel (TEP), ASCO staff successfully developed an eCQM, BRAF Mutational Analysis Results Received for Patients Treated for Metastatic Colorectal Cancer. CancerLinQ tested the implementation of three versions of the measure's technical specifications against data from 61 organizations and multiple electronic health record (EHR) systems. We analyzed the availability of required data elements in native and curated data, and variability and frequency of BRAF mutation testing documentation. The measure was also tested for validity and reliability among volunteer oncology practices. Results: Implementation efforts revealed that BRAF information is infrequently found in structured data fields and often lacks specificity for the specific variant tested. Ultimately, implementation of the measure in CancerLinQ was infeasible due to a lack of BRAF testing data and variant specificity across its active organizations. Reliability testing among oncology practices uncovered a lack of substantial performance gap with an average measure score of 87.4, 95% CI [86.1, 88.8], making the measure unsuitable for Consensus-Based Entity (CBE) endorsement or use in federal reporting programs. Conclusions: While ASCO’s implementation efforts yielded clear benefits, including identification of future CancerLinQ platform enhancements, gaps in the capture of BRAF testing data and inconsistent documentation of BRAF variants across EHR systems prevented successful implementation of the eCQM at this time. These findings underscore the current difficulty in ascertaining the specificity of key molecular variants in structured EHR data alone without human curation or access to unstructured data, as well as an ongoing need for measure stakeholders to advance eCQM standards necessary for supporting future eCQM deployment.

Impact of integrating standardized venous thromboembolism (VTE) prophylaxis orders for patients with multiple myeloma into an electronic health record (EHR).

469 Background: VTE is a serious complication in multiple myeloma patients, who have a 9-fold increased risk compared to the general population. Prophylaxis for VTE is recommended by international guidelines in multiple myeloma patients based on VTE risk assessment, but published studies have demonstrated prescribing may be suboptimal. Our aim was to evaluate the impact of integrating standardized VTE prophylaxis orders for multiple myeloma patients on active treatment into our EHR, which services over 2300 providers within 500 community cancer treatment locations across the United States. Methods: National Comprehensive Cancer Network Guideline-recommended VTE prophylaxis agents were integrated as optional orders into multiple myeloma EHR regimen templates from June to July 2022. Optional informational orders were also included: a “VTE Prophylaxis Reminder” order served as a trigger for the care team if prophylaxis was not prescribed and a “VTE Prophylaxis Exception” order served as documentation if the patient did not qualify for anticoagulation. An educational guide about VTE risk stratification and prophylaxis in multiple myeloma was developed and embedded in regimen templates. To evaluate the impact of this intervention, a structured EHR data export was used to identify multiple myeloma patients with regimen templates ordered during the prespecified time frames. A retrospective chart review was conducted to identify documentation of anticoagulation or reason for exception on or prior to cycle 1 day 1 of the multiple myeloma regimen. Data were compared using the Fisher’s exact test. We also evaluated ordering of VTE prophylaxis and use of informational orders from regimens during the post-implementation phase. Results: A total of 297 patients were identified, with 128 during the pre-implementation phase (May 2022) and 169 during the post-implementation phase (September 2022). Documentation of anticoagulation or reason for exception on or prior to cycle 1 day 1 of the multiple myeloma regimen increased from 70% (89/128) during the pre-implementation phase to 81% (137/169) during the post-implementation phase (two-sided p=0.0276). Aspirin was the most common agent prescribed in 73% and 64% of patients, respectively. During the post-implementation phase, VTE prophylaxis was prescribed directly from the regimen in 16% of patients. The “VTE Prophylaxis Reminder” order was utilized for 89% of patients and the “VTE Prophylaxis Exception” order was utilized for 1 patient. Conclusions: Integration of standardized orders into EHR regimen templates may facilitate prescribing and serve as reminders for the care team. We saw an 11% increase in documentation of VTE prophylaxis, which may potentially reduce risk of VTE in this patient population. Efforts are ongoing to highlight the need for anticoagulation and importance of documentation.

SAP BusinessObjects in Medical Informatics.

Electronic health record (EHR) generates a large amount of data filled with opportunities to enhance documentation compliance, quality improvement, and other metrics. Various software tools exist, but many clinicians are unaware of them. Our institution switched from a hybrid of paper and multiple small EHRs to one all-inclusive EHR system. We faced significant challenges beyond the typical new software deployment phase that affected our departmental regulatory compliance, quality measures, and research initiatives. We aimed to navigate these issues through the use of medical informatics. We used a multidimensional database software analysis tool called SAP BusinessObjects® (SAP SE. Released 2020. SAP BusinessObjects, Version 14.2.8.3671. Waldorf, Germany) to design automated queries for the patient database to generate various reports for our department. As a result, We improved our anesthesia documentation non-compliance from 13-17% of all cases to 4% within months. We have also used this tool to automatically generate various reports such as preoperative beta-blocker administrations, caseloads, case complications, procedure logs, and medication records. Even today many departments rely on manual checks for even the most basic documentation and quality metric compliance, which can be time consuming and costly. Using medical informatics tools is a highly efficient alternative. Fortunately, many software tools exist within most modern EHR packages, and most people can learn to use these tools productively.

Assessing document section heterogeneity across multiple electronic health record systems for computational phenotyping: A case study of heart-failure phenotyping algorithm.

The incorporation of information from clinical narratives is critical for computational phenotyping. The accurate interpretation of clinical terms highly depends on their associated context, especially the corresponding clinical section information. However, the heterogeneity across different Electronic Health Record (EHR) systems poses challenges in utilizing the section information. Leveraging the eMERGE heart failure (HF) phenotyping algorithm, we assessed the heterogeneity quantitatively through the performance comparison of machine learning (ML) classifiers which map clinical sections containing HF-relevant terms across different EHR systems to standard sections in Health Level 7 (HL7) Clinical Document Architecture (CDA). We experimented with both random forest models with sentence-embedding features and bidirectional encoder representations from transformers models. We trained MLs using an automated labeled corpus from an EHR system that adopted HL7 CDA standard. We assessed the performance using a blind test set (n = 300) from the same EHR system and a gold standard (n = 900) manually annotated from three other EHR systems. The F-measure of those ML models varied widely (0.00-0.91%), indicating MLs with one tuning parameter set were insufficient to capture sections across different EHR systems. The error analysis indicates that the section does not always comply with the corresponding standardized sections, leading to low performance. We presented the potential use of ML techniques to map the sections containing HF-relevant terms in multiple EHR systems to standard sections. However, the findings suggested that the quality and heterogeneity of section structure across different EHRs affect applications due to the poor adoption of documentation standards.

Temporal Events Detector for Pregnancy Care (TED-PC): A rule-based algorithm to infer gestational age and delivery date from electronic health records of pregnant women with and without COVID-19.

Identifying the time of SARS-CoV-2 viral infection relative to specific gestational weeks is critical for delineating the role of viral infection timing in adverse pregnancy outcomes. However, this task is difficult when it comes to Electronic Health Records (EHR). In combating the COVID-19 pandemic for maternal health, we sought to develop and validate a clinical information extraction algorithm to detect the time of clinical events relative to gestational weeks. We used EHR from the National COVID Cohort Collaborative (N3C), in which the EHR are normalized by the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). We performed EHR phenotyping, resulting in 270,897 pregnant women (June 1st, 2018 to May 31st, 2021). We developed a rule-based algorithm and performed a multi-level evaluation to test content validity and clinical validity, and extreme length of gestation (<150 or >300). The algorithm identified 296,194 pregnancies (16,659 COVID-19, 174,744 without COVID-19) in 270,897 pregnant women. For inferring gestational age, 95% cases (n = 40) have moderate-high accuracy (Cohen's Kappa = 0.62); 100% cases (n = 40) have moderate-high granularity of temporal information (Cohen's Kappa = 1). For inferring delivery dates, the accuracy is 100% (Cohen's Kappa = 1). The accuracy of gestational age detection for the extreme length of gestation is 93.3% (Cohen's Kappa = 1). Mothers with COVID-19 showed higher prevalence in obesity or overweight (35.1% vs. 29.5%), diabetes (17.8% vs. 17.0%), chronic obstructive pulmonary disease (0.2% vs. 0.1%), respiratory distress syndrome or acute respiratory failure (1.8% vs. 0.2%). We explored the characteristics of pregnant women by different gestational weeks of SARS-CoV-2 infection with our algorithm. TED-PC is the first to infer the exact gestational week linked with every clinical event from EHR and detect the timing of SARS-CoV-2 infection in pregnant women. The algorithm shows excellent clinical validity in inferring gestational age and delivery dates, which supports multiple EHR cohorts on N3C studying the impact of COVID-19 on pregnancy.

CEDAR: communication efficient distributed analysis for regressions.

Electronic health records (EHRs) offer great promises for advancing precision medicine and, at the same time, present significant analytical challenges. Particularly, it is often the case that patient-level data in EHRs cannot be shared across institutions (data sources) due to government regulations and/or institutional policies. As a result, there are growing interests about distributed learning over multiple EHRs databases without sharing patient-level data. To tackle such challenges, we propose a novel communication efficient method that aggregates the optimal estimates of external sites, by turning the problem into a missing data problem. In addition, we propose incorporating posterior samples of remote sites, which can provide partial information on the missing quantities and improve efficiency of parameter estimates while having the differential privacy property and thus reducing the risk of information leaking. The proposed approach, without sharing the raw patient level data, allows for proper statistical inference. We provide theoretical investigation for the asymptotic properties of the proposed method for statistical inference as well as differential privacy, and evaluate its performance in simulations and real data analyses in comparison with several recently developedmethods.

Linkage of multiple electronic health record datasets using a "spine linkage" approach compared to all "pairwise linkages".

ObjectivesTo compare two approaches for linking multiple datasets: using all "pairwise linkages", linking each dataset to every other dataset; versus linking each dataset to a designated "spine dataset", by: considering the differences between these approaches, and illustrating using real-world data on patients undergoing emergency bowel cancer surgery.&#x0D; ApproachWe linked an administrative hospital dataset (Hospital Episode Statistics; HES) capturing patients admitted to hospitals in England, and two clinical datasets comprising patients undergoing emergency bowel surgery (National Emergency Laparotomy Audit; NELA) and patients diagnosed with bowel cancer (National Bowel Cancer Audit; NBOCA), with study period from 31 October 2013 to 30 April 2018. We compared pairwise linkage to spine linkage, designating HES as the spine dataset, by considering the number of eligible patients linked by each approach, characteristics of linked patients, levels of missing data, and whether analysis results were sensitive to the approach used.&#x0D; ResultsThe spine linkage approach resulted in an analysis cohort of 15,826 patients, equating to 98.3% of the 16,100 patients identified with the pairwise linkage approach. Of 274 additional patients captured in the pairwise approach, approximately two-thirds were only in the emergency surgery dataset (NELA) and one-third were only in the bowel cancer dataset (NBOCA). There were no systematic differences in patient characteristics between these analysis cohorts. Associations of patient and tumour characteristics with mortality, complications, and length of stay were not sensitive to the linkage approach. When eligibility criteria were applied before linkage, spine linkage included 14,509 patients (90.0% compared to pairwise linkage).&#x0D; ConclusionSpine linkage can be an efficient alternative to pairwise linkage, if case ascertainment in the spine dataset and data quality of linkage variables are high. These aspects should be systematically evaluated in the nominated spine dataset before spine linkage. Results are sensitive to order of linkage steps.

Defining Opioid-related Problems Using a Health Care Safety Net Institution's Inpatient Electronic Health Records: Limitations of Diagnosis-based Definitions.

Measuring clinically relevant opioid-related problems in health care systems is challenging due to the lack of standard definitions and coding practices. Well-defined, opioid-related health problems (ORHPs) would improve prevalence estimates and evaluation of clinical interventions, crisis response, and prevention activities. We sought to estimate prevalence of opioid use disorder (OUD), opioid misuse, and opioid poisoning among inpatients at a large, safety net, health care institution. Our study included events documented in the electronic health records (EHRs) among hospitalized patients at Denver Health Medical Center during January 1, 2017 to December 31, 2018. Multiple EHR markers (ie, opioid-related diagnostic codes, clinical assessment, laboratory results, and free-text documentation) were used to develop diagnosis-based and extended definitions for OUD, opioid misuse, and opioid poisoning. We used these definitions to estimate number of hospitalized patients with these conditions. During a 2-year study period, 715 unique patients were identified solely using opioid-related diagnostic codes; OUD codes accounted for the largest proportion (499/715, 69.8%). Extended definitions identified an additional 973 unique patients (~136% increase), which includes 155/973 (15.9%) who were identified by a clinical assessment marker, 1/973 (0.1%) by a laboratory test marker, and 817/973 (84.0%) by a clinical documentation marker. Solely using diagnostic codes to estimate prevalence of clinically relevant ORHPs missed most patients with ORHPs. More inclusive estimates were generated using additional EHR markers. Improved methods to estimate ORHPs among a health care system's patients would more fully estimate organizational and economic burden to more efficiently allocate resources and ensure capacity to provide clinical services.

Intimate Partner Violence and the Pediatric Electronic Health Record: A Qualitative Study.

To explore expert perspectives on risks associated with the pediatric electronic health record (EHR) for intimate partner violence (IPV) survivors and their children and to identify strategies that may mitigate these risks. We conducted semistructured interviews with multidisciplinary pediatric IPV experts (nursing, physicians, social workers, hospital security, IPV advocates) recruited via snowball sampling. We coded interview transcripts using thematic analysis, then consolidated codes into themes. Twenty-eight participants completed interviews. Participants identified the primary source of risk as an abuser's potential access to a child's EHR by legal and illegal means. They noted that abuser's access to multiple pediatric EHR components (eg, online health portals, clinical notes, contact information) may result in escalated violence, stalking, and manipulation of IPV survivors. Suggested risk mitigation strategies included limited and coded documentation, limiting EHR access, and discussing documentation with the IPV survivor. Challenges to using these strategies included healthcare providers' usual practice of detailed documentation and that information documented may confer both risk and benefit concurrently. Reported potential benefits of the pediatric EHR for IPV survivors included ensuring continuity of care, decreasing need to repeatedly talk about trauma histories, and communication of safety plans. Our findings suggest the pediatric EHR may confer both risks and benefits for IPV survivors and their children. Further work is needed to develop best practices to address IPV risks related to the pediatric EHR, to ensure consistent use of these practices, and to include these practices as standard functionalities of the pediatric EHR.

Linkage of multiple electronic health record datasets using a 'spine linkage' approach compared with all 'pairwise linkages'.

Methods for linking records between two datasets are well established. However, guidance is needed for linking more than two datasets. Using all 'pairwise linkages'-linking each dataset to every other dataset-is the most inclusive, but resource-intensive, approach. The 'spine' approach links each dataset to a designated 'spine dataset', reducing the number of linkages, but potentially reducing linkage quality. We compared the pairwise and spine linkage approaches using real-world data on patients undergoing emergency bowel cancer surgery between 31October2013 and 30April2018. We linked an administrative hospital dataset (Hospital Episode Statistics; HES) capturing patients admitted to hospitals in England, and two clinical datasets comprising patients diagnosed with bowel cancer and patients undergoing emergency bowel surgery. The spine linkage approach, with HES as the spine dataset, created an analysis cohort of 15 826 patients, equating to 98.3% of the 16 100 patients identified using the pairwise linkage approach. There were no systematic differences in patient characteristics between these analysis cohorts. Associations of patient and tumour characteristics with mortality, complications and length of stay were not sensitive to the linkage approach. When eligibility criteria were applied before linkage, spine linkage included 14 509 patients (90.0% compared with pairwise linkage). Spine linkage can be used as an efficient alternative to pairwise linkage if case ascertainment in the spine dataset and data quality of linkage variables are high. These aspects should be systematically evaluated in the nominated spine dataset before spine linkage is used to create the analysis cohort.

COVID-19 trajectories among 57 million adults in England: a cohort study using electronic health records

BackgroundUpdatable estimates of COVID-19 onset, progression, and trajectories underpin pandemic mitigation efforts. To identify and characterise disease trajectories, we aimed to define and validate ten COVID-19 phenotypes from nationwide linked electronic health records (EHR) using an extensible framework.MethodsIn this cohort study, we used eight linked National Health Service (NHS) datasets for people in England alive on Jan 23, 2020. Data on COVID-19 testing, vaccination, primary and secondary care records, and death registrations were collected until Nov 30, 2021. We defined ten COVID-19 phenotypes reflecting clinically relevant stages of disease severity and encompassing five categories: positive SARS-CoV-2 test, primary care diagnosis, hospital admission, ventilation modality (four phenotypes), and death (three phenotypes). We constructed patient trajectories illustrating transition frequency and duration between phenotypes. Analyses were stratified by pandemic waves and vaccination status.FindingsAmong 57 032 174 individuals included in the cohort, 13 990 423 COVID-19 events were identified in 7 244 925 individuals, equating to an infection rate of 12·7% during the study period. Of 7 244 925 individuals, 460 737 (6·4%) were admitted to hospital and 158 020 (2·2%) died. Of 460 737 individuals who were admitted to hospital, 48 847 (10·6%) were admitted to the intensive care unit (ICU), 69 090 (15·0%) received non-invasive ventilation, and 25 928 (5·6%) received invasive ventilation. Among 384 135 patients who were admitted to hospital but did not require ventilation, mortality was higher in wave 1 (23 485 [30·4%] of 77 202 patients) than wave 2 (44 220 [23·1%] of 191 528 patients), but remained unchanged for patients admitted to the ICU. Mortality was highest among patients who received ventilatory support outside of the ICU in wave 1 (2569 [50·7%] of 5063 patients). 15 486 (9·8%) of 158 020 COVID-19-related deaths occurred within 28 days of the first COVID-19 event without a COVID-19 diagnoses on the death certificate. 10 884 (6·9%) of 158 020 deaths were identified exclusively from mortality data with no previous COVID-19 phenotype recorded. We observed longer patient trajectories in wave 2 than wave 1.InterpretationOur analyses illustrate the wide spectrum of disease trajectories as shown by differences in incidence, survival, and clinical pathways. We have provided a modular analytical framework that can be used to monitor the impact of the pandemic and generate evidence of clinical and policy relevance using multiple EHR sources.FundingBritish Heart Foundation Data Science Centre, led by Health Data Research UK.

Integrated Health Record Viewers and Reduction in Duplicate Medical Imaging: Retrospective Observational Analysis.

BackgroundHealth information exchange and multiplatform health record viewers support more informed medical decisions, improve quality of care, and reduce the risk of adverse outcomes due to fragmentation and discontinuity in care during transition of care. An example of a multiplatform health record viewer is the VA/DoD Joint Longitudinal Viewer (JLV), which supports the Department of Veterans Affairs (VA) and Department of Defense (DoD) health care providers with read-only access to patient medical records integrated from multiple sources. JLV is intended to support more informed medical decisions such as reducing duplicate medical imaging when previous image study results may meet current clinical needs.ObjectiveWe estimated the impact of provider usage of JLV on duplicate imaging for service members transitioning from the DoD to the VA health care system.MethodsWe conducted a retrospective cross-sectional study in fiscal year 2018 to examine the relationship between providers’ use of JLV and the likelihood of ordering duplicate images. Our sample included recently separated service members who had a VA primary care visit in fiscal year 2018 within 90 days of a DoD imaging study. Patients who received at least one imaging study at VA within 90 days of a DoD imaging study of the same imaging mode and on the same body part are considered to have received potentially duplicate imaging studies. We use a logistic regression model with “JLV provider” (providers with 1 or more JLV audits in the prior 6 months) as the independent variable to estimate the relationship between JLV use and ordering of duplicate images. Control variables included provider image ordering rates in the prior 6 months, provider type, patient demographics (age, race, gender), and clinical characteristics (Elixhauser comorbidity score).ResultsProviders known to utilize JLV in the prior 6 months order fewer duplicate images relative to providers not utilizing JLV for similar visits over time (odds ratio 0.44, 95% CI 0.24-0.78; P=.005). This effect is robust across multiple specifications of linear and logistic regression models. The provider’s practice pattern of ordering image studies and the patient’s health status are powerful confounders.ConclusionsThis study provides evidence that adoption of a longitudinal viewer of health records from multiple electronic health record systems is associated with a reduced likelihood of ordering duplicate images. Investments in health information exchange systems may be effective ways to improve the quality of care and reduce adverse outcomes for patients experiencing fragmentation and discontinuity of care.

Early Identification of Maternal Cardiovascular Risk Through Sourcing and Preparing Electronic Health Record Data: Machine Learning Study.

BackgroundHealth care data are fragmenting as patients seek care from diverse sources. Consequently, patient care is negatively impacted by disparate health records. Machine learning (ML) offers a disruptive force in its ability to inform and improve patient care and outcomes. However, the differences that exist in each individual’s health records, combined with the lack of health data standards, in addition to systemic issues that render the data unreliable and that fail to create a single view of each patient, create challenges for ML. Although these problems exist throughout health care, they are especially prevalent within maternal health and exacerbate the maternal morbidity and mortality crisis in the United States.ObjectiveThis study aims to demonstrate that patient records extracted from the electronic health records (EHRs) of a large tertiary health care system can be made actionable for the goal of effectively using ML to identify maternal cardiovascular risk before evidence of diagnosis or intervention within the patient’s record. Maternal patient records were extracted from the EHRs of a large tertiary health care system and made into patient-specific, complete data sets through a systematic method.MethodsWe outline the effort that was required to define the specifications of the computational systems, the data set, and access to relevant systems, while ensuring that data security, privacy laws, and policies were met. Data acquisition included the concatenation, anonymization, and normalization of health data across multiple EHRs in preparation for their use by a proprietary risk stratification algorithm designed to establish patient-specific baselines to identify and establish cardiovascular risk based on deviations from the patient’s baselines to inform early interventions.ResultsPatient records can be made actionable for the goal of effectively using ML, specifically to identify cardiovascular risk in pregnant patients.ConclusionsUpon acquiring data, including their concatenation, anonymization, and normalization across multiple EHRs, the use of an ML-based tool can provide early identification of cardiovascular risk in pregnant patients.

Use of electronic health record systems in accountable care organizations.

To assess the ability of accountable care organizations (ACOs) to use electronic health record (EHR) data for quality. Cross-sectional study of ACOs participating in the Medicare Shared Savings Program(MSSP). A national survey of MSSP ACOs included questions on the number of EHR systems used across all providers in the ACO and barriers to reporting EHR-based quality measures. Just 9% of ACOs use a single EHR system, whereas 77% use 6 or more EHR systems. The more EHR systems an ACO uses, the less likely it is to report having the infrastructure to aggregate EHR data and the more concerned it is about the short-term viability and accuracy of EHR-based quality measures. ACOs have diverse structures that often result in the usage of multiple EHR systems. This has the potential to cause serious delays when CMS begins requiring ACOs to report their quality measures through their EHRs in2022.

Optimizing Identification of People Living with HIV from Electronic Medical Records: Computable Phenotype Development and Validation.

Electronic health record (EHR)-based computable phenotype algorithms allow researchers to efficiently identify a large virtual cohort of Human Immunodeficiency Virus (HIV) patients. Built upon existing algorithms, we refined, improved, and validated an HIV phenotype algorithm using data from the OneFlorida Data Trust, a repository of linked claims data and EHRs from its clinical partners, which provide care to over 15 million patients across all 67 counties in Florida. Our computable phenotype examined information from multiple EHR domains, including clinical encounters with diagnoses, prescription medications, and laboratory tests. To identify an HIV case, the algorithm requires the patient to have at least one diagnostic code for HIV and meet one of the following criteria: have 1+ positive HIV laboratory, have been prescribed with HIV medications, or have 3+ visits with HIV diagnostic codes. The computable phenotype was validated against a subset of clinical notes. Among the 15+ million patients from OneFlorida, we identified 61,313 patients with confirmed HIV diagnosis. Among them, 8.05% met all four inclusion criteria, 69.7% met the 3+ HIV encounters criteria in addition to having HIV diagnostic code, and 8.1% met all criteria except for having positive laboratories. Our algorithm achieved higher sensitivity (98.9%) and comparable specificity (97.6%) relative to existing algorithms (77-83% sensitivity, 86-100% specificity). The mean age of the sample was 42.7 years, 58% male, and about half were Black African American. Patients' average follow-up period (the time between the first and last encounter in the EHRs) was approximately 4.6 years. The median number of all encounters and HIV-related encounters were 79 and 21, respectively. By leveraging EHR data from multiple clinical partners and domains, with a considerably diverse population, our algorithm allows more flexible criteria for identifying patients with incomplete laboratory test results and medication prescribing history compared with prior studies.

Improving Student Confidence With Electronic Health Record Order Entry.

Despite near-universal utilization of electronic health records (EHRs) by physicians in practice, medical students in most ambulatory settings gain limited experience with placing EHR orders. In this study, an individual preceptor site investigated the usefulness of a targeted curriculum in improving students' EHR confidence and clinical reasoning skills. Family medicine clerkship students assigned to one community health center were invited to participate in this prospective, survey-based study. In their first week, students observed a preceptor performing EHR tasks. For the remainder of the 4-week clerkship, students utilized decision support tools, assigned a working diagnosis, entered unsigned orders in the EHR, proposed an assessment, and discussed a plan with a preceptor. Students completed weekly questionnaires to self-report confidence across several EHR domains while preceptors synchronously evaluated students' accuracy with entering orders correctly. From February 2017 to March 2020, all 49 eligible students completed the study. One hundred percent of students reported that placing EHR orders was beneficial to their medical education. The difference over time in learner confidence with placing EHR orders was statistically significant across every domain (eg, writing prescriptions, ordering labs and imaging). Preceptors' evaluations of students' accuracy with placing orders also showed significant improvement between each week. Clerkship-wide EHR training may be limited by multiple sites with multiple EHR products. This pilot study suggests that committed faculty at an individual preceptor site can offer a targeted curriculum to help students develop EHR confidence. We propose other preceptors similarly offer students this opportunity to maximize clerkship education.

PCN239 Development of the NOBLE (NOvartis Braf+ meLanoma patients obsErvational) Dataset: a Novel Approach to Creating a Longitudinal Melanoma Data Source by Harmonizing Multiple Electronic Health Record Databases

Although clinical trials have shown that novel immuno-oncology (IO) and targeted therapies (TT) improve clinical outcomes, monitoring longitudinal treatment patterns and outcomes among patients in real-world settings is crucial. Here, we describe the development of NOBLE (Novartis Braf+ meLanoma patients ObsErvational) dataset, to inform critical questions about risk factors, treatment patterns, real-world outcomes, and complement clinical trial data.