Electronic Health Record Phenotyping Research Articles

1085-P: Disparities in SGLT2i Prescription for Primary Prevention of ASCVD in Patients with T2D

Introduction and Objective: Identify disparities in SGLT2i prescription (SGLT2i-Rx) for primary prevention of Atherosclerotic Cardiovascular Disease (ASCVD) in adults with T2D. Methods: Retrospective analysis of a large hospital system Electronic Health Record (EHR) database. Results: From June 2016-August 2023, 791,873 ambulatory encounters were included. Dialysis and pregnancy were excluded. EHR phenotyping [≥2 appropriate ICD10 codes on subsequent visits AND diabetes defining labs or T2D medications] identified 38,435 with T2D and without ASCVD (age 61.5±9 yrs, 52% women, 48% Non Hispanic White (NHW), 37% private insurance (PI), 32% Medicare and 71% within less deprived ADI Quintiles (1-3). Cardiometabolic comorbidities included obesity 57%, HTN 78%, dyslipidemia (DL) 54%, CKD 13%, heart failure (HF) 3%, and smoking 7%. SGLT2i-Rx was very low (n=6275, 16.3%) and more common in younger age (61 ± 12 vs 62 ± 13 yrs), men (57%), NHW (49%), less deprived ADI (70%), PI (41%), obesity (57%), higher A1C (7.7% vs 7.0%), HF (4.0 vs 2.8%,) DL (62 vs 51%), HTN (82 vs 77%) and CKD (17 vs 12%) vs. no SGLT2i-Rx. Logistic regression analysis showed men were significantly more likely to have SGLT2i-Rx after adjustment for age, race/ethnicity, socioeconomics, and comorbidities. Conclusion: In a large primary prevention cohort of T2D with high cardiometabolic burden, SGLT2i-Rx was low overall, and disproportionately lower in women, indicating possible provider and patient gender biases. Disclosure A.R. Sadhu: Advisory Panel; Abbott Diagnostics. E. Avenatti: None. R. Gullapelli: None. K. Nasir: Research Support; Novartis Pharmaceuticals Corporation, ESPERION Therapeutics, Inc. Advisory Panel; Amgen Inc, Merck Sharp & Dohme Corp, ER Squib & Sons. Research Support; National Institutes of Health, PCORI. S.R. Kashyap: Research Support; Fractyl Health, Inc. Advisory Panel; GI Dynamics. Research Support; Janssen Pharmaceuticals, Inc, Novartis Pharmaceuticals Corporation.

1086-P: Disparities in GLP-1RA Prescription for Primary Prevention of ASCVD in Patients with T2D

Introduction and Objective: Identify disparities in GLP1RA prescription (GLP1RA-Rx) for primary prevention of Atherosclerotic Cardiovascular disease (ASCVD) in adults with T2D. Methods: Retrospective analysis of a large health system Electronic Health Record (EHR) database. Results: From June 2016-August 2023, 791,873 ambulatory encounters were included. Dialysis and pregnancy were excluded. EHR phenotyping [≥2 appropriate ICD10 codes on subsequent visits AND either abnormal diabetes defining labs or T2D medications] identified 38,435 with T2D and without ASCVD (age 61.5±9 yrs, 52% women, 48% Non Hispanic White (NHW), 37% private insurance (PI), 32% Medicare and 71% within less deprived ADI Quintiles (1-3). Cardiometabolic comorbidities included obesity 57%, HTN 78%, dyslipidemia (DL) 54%, CKD 13%, heart failure (HF) 3%, and smoking 7%. GLP1RA-Rx was low (n=11352, 29.5%) and more often in younger age (57±11 vs 62±13 yrs), women (56%), NHW (49%), PI (49%), less deprived ADI (72%), obesity (70%), DL (62% vs 51%) and CKD (15% vs 13%) vs. no GLP1RA-Rx. There was no difference in HTN, smoking or mean A1C levels. Logistic regression analysis showed women were more likely to have GLP1RA-Rx after adjustment for age, race/ethnicity, socioeconomics, and comorbidities. Conclusion: In a large primary prevention cohort with T2D and high cardiometabolic risk, GLP1RA-Rx was low and disproportionately higher in women regardless of age, socioeconomic or ASCVD risk factors. Disclosure A.R. Sadhu: Advisory Panel; Abbott Diagnostics. E. Avenatti: None. R. Gullapelli: None. K. Nasir: Research Support; Novartis Pharmaceuticals Corporation, ESPERION Therapeutics, Inc. Advisory Panel; Amgen Inc, Merck Sharp & Dohme Corp, ER Squib & Sons. Research Support; National Institutes of Health, PCORI. S.R. Kashyap: Research Support; Fractyl Health, Inc. Advisory Panel; GI Dynamics. Research Support; Janssen Pharmaceuticals, Inc, Novartis Pharmaceuticals Corporation.

Sequential EHR interventions to increase genetic testing for breast and ovarian cancer predisposition across diverse patient populations in gynecology practices at Penn Medicine.

10589 Background: Genetic testing (GT) identifies individuals who may benefit from increased surveillance and risk reduction strategies. GT is under-utilized, especially in those without a personal history of cancer and in minority populations. Methods: We identified a patient cohort meeting NCCN criteria for genetic testing utilizing electronic health record (EHR) phenotyping in 2 diverse gynecological practices. NCCN criteria included individuals with a personal history of ovarian cancer, early-onset (<50) breast cancer diagnosed before 2021, or a family history (FH) of ovarian cancer or male breast cancer. Participants with prior genetics visits were excluded. Patient nudges and provider messaging strategies were introduced to boost genetic counseling consultation. Nudges included patient portal messaging (PP) followed by texts using the Way To Health platform (WH) in those that did not respond to PP. For non-responders to patient directed nudges, genetic counseling consult orders were placed using Epic’s Pend & Send tool and sent to their gynecologist. Endpoints included the open rate for PP, response rate for the WH text, and the number of genetic counseling appointments completed. Differences between the clinics were calculated by Chi Square. Results: Of 1055 patients identified and who received a PP message regarding genetic counseling, 81% had a FH of ovarian cancer. Characteristics of the patient populations differed across clinics: Clinic D (n=505): 71.3% Black, 18% White and 67% < 45 years. Clinic R (n=550):10% Black, 83.5% White and 63% > 60 years.79% opened PP and 22.1% replied to PP, more in Clinic R (26.7% vs 17.0%, p<0.001). Patient engagement by PP or WH was 59.8% (631/1055), more in Clinic R (67.1% vs 51.9%, p<0.001). Of those that connected by patient nudges, 62.8% (396/631) declined additional follow-up (more in Clinic R, 41.5% then Clinic D, 33.3%, p=0.014), either due to incorrect family history in EHR, prior genetic testing, or, in the majority of cases, because they were not interested (296/631) (46.9%). Provider nudges added little to patient nudges with regard to GT uptake. 25% (266/1055) scheduled and 14.9% (157/1055) of the cohort completed GT appointments with no difference between the two (Clinic D 13.9% vs Clinic R 15.8%, p=NS). Conclusions: Patient directed nudges led to engagement of nearly 60% of patients in two diverse gynecology practices. 25% of individuals scheduled and 14.9% completed appointments, with continued follow-up. Although engagement in PP and WH differed between the two clinics, the number of visits did not. An EHR-based approach to identifying patients and encouraging genetic testing is a relatively low effort, scalable strategy to increase reach and encourage engagement in genetic counseling. However, a majority of patients either did not respond or did not wish to be tested. Clinical trial information: NCT05721326 .

Resident clinical dashboards to support precision education in emergency medicine.

With the move toward competency-based medical education (CBME), data from the electronic health record (EHR) for informed self-improvement may be valuable as a part of programmatic assessment. Personalized dashboards are one way to view these clinical data. The purpose of this concept paper is to summarize the current state of clinical dashboards as they can be utilized by emergency medicine (EM) residency programs. The author group consisted of EM physicians from multiple institutions with medical education and informatics backgrounds and was identified by querying faculty presenting on resident clinical dashboards at the 2024 Society for Academic Emergency Medicine conference. Additional authors were identified by members of the initial group. Best practice literature was referenced; if none was available, group consensus was used. Clinical exposures as well as efficiency, quality, documentation, and diversity metrics may be included in a resident dashboard. Resident dashboard metrics should focus on resident-sensitive measures rather than those primarily affected by attendings or systems-based factors. Implementation of these dashboards requires the technical expertise to turn EHR data into actionable data, a process called EHR phenotyping. The dashboard can be housed directly in the EHR or on a separate platform. Dashboard developers should consider how their implementation plan will affect how often dashboard data will be refreshed and how to best display the data for ease of understanding. Dashboards can provide objective data to residents, residency leadership and clinical competency committees as they identify areas of strength, growth areas, and set specific and actionable goals. The success of resident dashboards is reliant on resident buy-in and creating a culture of psychological safety through thoughtful implementation, coaching, and regular feedback. . Personalized clinical dashboards can play a crucial role in programmatic assessment within CBME, helping EM residents focus their efforts as they advance and refine their skills during training.

Accuracy of Electronic Health Record Phenotypes to Detect Recognition of Hypertension in Pediatric Primary Care.

Accuracy of Electronic Health Record Phenotypes to Detect Recognition of Hypertension in Pediatric Primary Care.

Creating a Learning Health System in a Cancer Center: Generalizability of an Electronic Health Record Phenotype for Advanced Solid Cancer.

To test the generalizability of an electronic health record (EHR) phenotype for patients with advanced solid cancer, which was previously developed in a single cancer center. We compared an algorithm to identify patients with advanced solid cancer from a random sample of patients with active cancer in the Veterans Health Administration (VA) and an academic cancer center with a human-coded reference standard between January 1, 2016, and December 31, 2019. Compared with the human-coded reference standard, the algorithm had high specificity (93%; 95% CI, 87 to 99 and 97%; 95% CI, 93 to 100) and reasonable sensitivity (85%; 95% CI, 76 to 94 and 87%; 95% CI, 77 to 97) in the VA and academic cancer center populations, respectively. Patients with advanced cancer (compared with those with active nonadvanced cancer) had higher mortality at the VA and academic cancer center (29.2% and 17.0% 6-month mortality v 6.8% and 3.5%), respectively. This EHR phenotype can be used to measure and improve the quality of palliative care for patients with advanced cancer within and across health care settings.

Assessment of health conditions from patient electronic health record portals vs self-reported questionnaires: an analysis of the INSPIRE study.

Assessment of health conditions from patient electronic health record portals vs self-reported questionnaires: an analysis of the INSPIRE study.

P-679. Risk Factors Shared Among Common Respiratory Viruses

Abstract Background Upper respiratory infections affect patients of all ages, with severity influenced by demographics, comorbidities, and immunosuppression. Studies often determine risk factors associated with severe disease, but rarely are multiple upper respiratory pathogens compared within the same cohort. Using electronic health record (EHR) phenotyping of 267,031 participants with relevant EHR data enrolled in the National Institute of Health’s All of Us Research Program (All of Us), we identified and validated cohorts of participants, and revealed shared risk factors for severity among upper respiratory viruses. Methods We developed EHR phenotype algorithms for eight respiratory viruses using pathogen-specific billing codes, test results, and treatment-dose antivirals. We identified episodes of infection over 90-day periods and validated these using CDC national detection rates over the same period. We then used multivariable logistic regression to compare demographics and identify predictors of hospital admission, adjusting for age, sex, self-identified race and ethnicity, BMI, smoking status, and EHR record length. Results Scaled episode and testing counts paralleled CDC data, with expected seasonal trends observed for influenza, parainfluenza, respiratory syncytial virus (RSV), rhinovirus, human metapneumovirus (hMPV), SARS-CoV-2, and human coronavirus (hCOV) (RSV, hCOV, influenza, and SARS-CoV-2 shown in Figure 1). Across viruses, hospital admission was associated with older age, male sex, underweight, severe obesity, self-identified Black race, and smoking (Table 1). Notably, significant predictors of admission included COPD, asthma, heart failure, ischemic heart disease, hypertension, diabetes, chronic kidney disease, and immunodeficiency, even after adjusting for confounding variables (Figure 2). Conclusion In this study we established and validated an EHR phenotyping algorithm, and demonstrated that clinical risk factors — particularly cardiopulmonary, metabolic, and immunological comorbidities — are associated with severe outcomes among multiple upper respiratory viruses. These results suggest that universal prevention interventions may mitigate the impact of respiratory viral infections in high-risk populations. Disclosures All Authors: No reported disclosures

Bidirectional Risk Modulator and Modifier Variant of Dilated and Hypertrophic Cardiomyopathy in BAG3

The genetic factors that modulate the reduced penetrance and variable expressivity of heritable dilated cardiomyopathy (DCM) are largely unknown. BAG3 genetic variants have been implicated in both DCM and hypertrophic cardiomyopathy (HCM), nominating BAG3 as a gene that harbors potential modifier variants in DCM. To interrogate the clinical traits and diseases associated with BAG3 coding variation. This was a cross-sectional study in the Penn Medicine BioBank (PMBB) enrolling patients of the University of Pennsylvania Health System's clinical practice sites from 2014 to 2023. Whole-exome sequencing (WES) was linked to electronic health record (EHR) data to associate BAG3 coding variants with EHR phenotypes. This was a health care population-based study including individuals of European and African genetic ancestry in the PMBB with WES linked to EHR phenotypes, with replication studies in BioVU, UK Biobank, MyCode, and DCM Precision Medicine Study. Carrier status for BAG3 coding variants. Association of BAG3 coding variation with clinical diagnoses, echocardiographic traits, and longitudinal outcomes. In PMBB (n = 43 731; median [IQR] age, 65 [50-76] years; 21 907 female [50.1%]), among 30 324 European and 11 198 African individuals, the common C151R variant was associated with decreased risk for DCM (odds ratio [OR], 0.85; 95% CI, 0.78-0.92) and simultaneous increased risk for HCM (OR, 1.59; 95% CI, 1.25-2.02), which was confirmed in the replication cohorts. C151R carriers exhibited improved longitudinal outcomes compared with noncarriers as assessed by age at death (hazard ratio [HR], 0.85; 95% CI, 0.74-0.96; median [IQR] age, 71.8 [63.1-80.7] in carriers and 70.3 [61.6-79.2] in noncarriers) and heart transplant (HR, 0.81; 95% CI, 0.66-0.99; median [IQR] age, 56.7 [46.1-63.1] in carriers and 55.6 [45.2-62.9] in noncarriers). C151R was associated with reduced risk of DCM (OR, 0.42; 95% CI, 0.24-0.74) and heart failure (OR, 0.27; 95% CI, 0.14-0.50) among individuals harboring truncating TTN variants in exons with high cardiac expression (n = 358). BAG3 C151R was identified as a bidirectional modulator of risk along the DCM-HCM spectrum, as well as an important genetic modifier variant in TTN-mediated DCM. This work expands on the understanding of the etiology and penetrance of DCM, suggesting that BAG3 C151R is an important genetic modifier variant contributing to the variable expressivity of DCM, warranting further exploration of its mechanisms and of genetic modifiers in DCM more broadly.

Automated Medical Records Review for Mild Cognitive Impairment and Dementia.

Objectives : Unstructured and structured data in electronic health records (EHR) are a rich source of information for research and quality improvement studies. However, extracting accurate information from EHR is labor-intensive. Here we introduce an automated EHR phenotyping model to identify patients with Alzheimer's Disease, related dementias (ADRD), or mild cognitive impairment (MCI). Methods : We assembled medical notes and associated International Classification of Diseases (ICD) codes and medication prescriptions from 3,626 outpatient adults from two hospitals seen between February 2015 and June 2022. Ground truth annotations regarding the presence vs. absence of a diagnosis of MCI or ADRD were determined through manual chart review. Indicators extracted from notes included the presence of keywords and phrases in unstructured clinical notes, prescriptions of medications associated with MCI/ADRD, and ICD codes associated with MCI/ADRD. We trained a regularized logistic regression model to predict the ground truth annotations. Model performance was evaluated using area under the receiver operating curve (AUROC), area under the precision-recall curve (AUPRC), accuracy, specificity, precision/positive predictive value, recall/sensitivity, and F1 score (harmonic mean of precision and recall). Results : Thirty percent of patients in the cohort carried diagnoses of MCI/ADRD based on manual review. When evaluated on a held-out test set, the best model using clinical notes, ICDs, and medications, achieved an AUROC of 0.98, an AUPRC of 0.98, an accuracy of 0.93, a sensitivity (recall) of 0.91, a specificity of 0.96, a precision of 0.96, and an F1 score of 0.93 The estimated overall accuracy for patients randomly selected from EHRs was 99.88%. Conclusion : Automated EHR phenotyping accurately identifies patients with MCI/ADRD based on clinical notes, ICD codes, and medication records. This approach holds potential for large-scale MCI/ADRD research utilizing EHR databases.

Electronic Health Record Phenotyping of Pediatric Suicide-Related Emergency Department Visits

Suicide is a leading cause of death among young people. Accurate detection of self-injurious thoughts and behaviors (SITB) underpins equity in youth suicide prevention. To compare methods of detecting SITB using structured electronic health information and measure algorithmic performance across demographics. This cross-sectional study used medical records among youths aged 6 to 17 years with at least 1 mental health-related emergency department (ED) visit in 2017 to 2019 to an academic health system in Southern California serving 787 000 unique individuals each year. Analyses were conducted between January and September 2023. Multiexpert electronic health record review ascertained the presence of SITB using the Columbia Classification Algorithm of Suicide Assessment. Random forest classifiers with nested cross-validation were developed using (1) International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes for nonfatal suicide attempt and self-harm and chief concern and (2) all available structured data, including diagnoses, medications, and laboratory tests. Detection performance was assessed overall and stratified by age group, sex, and race and ethnicity. The sample comprised 2702 unique youths with an MH-related ED visit (1384 youths who identified as female [51.2%]; 131 Asian [4.8%], 266 Black [9.8%], 719 Hispanic [26.6%], 1319 White [48.8%], and 233 other race [8.6%]; median [IQR] age, 14 [12-16] years), including 898 children and 1804 adolescents. Approximately half of visits were related to SITB (1286 visits [47.6%]). Sensitivity of SITB detection using only codes and chief concern varied by age group and increased until age 15 years (6-9 years: 59.3% [95% CI, 48.5%-69.5%]; 10-12 years: 69.0% [95% CI, 63.8%-73.9%]; 13-15 years: 88.4% [95% CI, 85.1%-91.2%]; 16-17 years: 83.1% [95% CI, 79.1%-86.6%]), while specificity remained constant. The area under the receiver operating characteristic curve (AUROC) was lower among preadolescents (0.841 [95% CI, 0.815-0.867]) and male (0.869 [95% CI, 0.848-0.890]), Black (0.859 [95% CI, 0.813-0.905]), and Hispanic (0.861 [95% CI, 0.831-0.891]) youths compared with adolescents (0.925 [95% CI, 0.912-0.938]), female youths (0.923 [95% CI, 0.909-0.937]), and youths of other races and ethnicities (eg, White: 0.901 [95% CI, 0.884-0.918]). Augmented classification (ie, using all available structured data) outperformed classification with codes and chief concern alone (AUROC, 0.975 [95% CI, 0.968-0.980] vs 0.894 [95% CI, 0.882-0.905]; P < .001). In this study, diagnostic codes and chief concern underestimated SITB prevalence, particularly among minoritized youths. These results suggest that priority on algorithmic fairness in suicide prevention strategies must extend to accurate detection of youths with suicide-related emergencies.

A Quantitative Bias Analysis Approach to Informative Presence Bias in Electronic Health Records.

Accurate outcome and exposure ascertainment in electronic health record (EHR) data, referred to as EHR phenotyping, relies on the completeness and accuracy of EHR data for each individual. However, some individuals, such as those with a greater comorbidity burden, visit the health care system more frequently and thus have more complete data, compared with others. Ignoring such dependence of exposure and outcome misclassification on visit frequency can bias estimates of associations in EHR analysis. We developed a framework for describing the structure of outcome and exposure misclassification due to informative visit processes in EHR data and assessed the utility of a quantitative bias analysis approach to adjusting for bias induced by informative visit patterns. Using simulations, we found that this method produced unbiased estimates across all informative visit structures, if the phenotype sensitivity and specificity were correctly specified. We applied this method in an example where the association between diabetes and progression-free survival in metastatic breast cancer patients may be subject to informative presence bias. The quantitative bias analysis approach allowed us to evaluate robustness of results to informative presence bias and indicated that findings were unlikely to change across a range of plausible values for phenotype sensitivity and specificity. Researchers using EHR data should carefully consider the informative visit structure reflected in their data and use appropriate approaches such as the quantitative bias analysis approach described here to evaluate robustness of study findings.

Identifying prehospital trauma patients from ambulance patient care records; comparing two methods using linked data in New South Wales, Australia

BackgroundLinked datasets for trauma system monitoring should ideally follow patients from the prehospital scene to hospital admission and post-discharge. Having a well-defined cohort when using administrative datasets is essential because they must capture the representative population. Unlike hospital electronic health records (EHR), ambulance patient-care records lack access to sources beyond immediate clinical notes. Relying on a limited set of variables to define a study population might result in missed patient inclusion. We aimed to compare two methods of identifying prehospital trauma patients: one using only those documented under a trauma protocol and another incorporating additional data elements from ambulance patient care records. MethodsWe analyzed data from six routinely collected administrative datasets from 2015 to 2018, including ambulance patient-care records, aeromedical data, emergency department visits, hospitalizations, rehabilitation outcomes, and death records. Three prehospital trauma cohorts were created: an Extended-T-protocol cohort (patients transported under a trauma protocol and/or patients with prespecified criteria from structured data fields), T-protocol cohort (only patients documented as transported under a trauma protocol) and non-T-protocol (extended-T-protocol population not in the T-protocol cohort). Patient-encounter characteristics, mortality, clinical and post-hospital discharge outcomes were compared. A conservative p-value of 0.01 was considered significant ResultsOf 1 038 263 patient-encounters included in the extended-T-population 814 729 (78.5 %) were transported, with 438 893 (53.9 %) documented as a T-protocol patient. Half (49.6 %) of the non-T-protocol sub-cohort had an International Classification of Disease 10th edition injury or external cause code, indicating 79644 missed patients when a T-protocol-only definition was used. The non-T-protocol sub-cohort also identified additional patients with intubation, prehospital blood transfusion and positive eFAST. A higher proportion of non-T protocol patients than T-protocol patients were admitted to the ICU (4.6% vs 3.6 %), ventilated (1.8% vs 1.3 %), received in-hospital transfusion (7.9 vs 6.8 %) or died (1.8% vs 1.3 %). Urgent trauma surgery was similar between groups (1.3% vs 1.4 %). ConclusionThe extended-T-population definition identified 50 % more admitted patients with an ICD-10-AM code consistent with an injury, including patients with severe trauma. Developing an EHR phenotype incorporating multiple data fields of ambulance-transported trauma patients for use with linked data may avoid missing these patients.

Large language models facilitate the generation of electronic health record phenotyping algorithms

Large language models facilitate the generation of electronic health record phenotyping algorithms

Creating a next-generation phenotype library: the health data research UK Phenotype Library.

To enable reproducible research at scale by creating a platform that enables health data users to find, access, curate, and re-use electronic health record phenotyping algorithms. We undertook a structured approach to identifying requirements for a phenotype algorithm platform by engaging with key stakeholders. User experience analysis was used to inform the design, which we implemented as a web application featuring a novel metadata standard for defining phenotyping algorithms, access via Application Programming Interface (API), support for computable data flows, and version control. The application has creation and editing functionality, enabling researchers to submit phenotypes directly. We created and launched the Phenotype Library in October 2021. The platform currently hosts 1049 phenotype definitions defined against 40 health data sources and >200K terms across 16 medical ontologies. We present several case studies demonstrating its utility for supporting and enabling research: the library hosts curated phenotype collections for the BREATHE respiratory health research hub and the Adolescent Mental Health Data Platform, and it is supporting the development of an informatics tool to generate clinical evidence for clinical guideline development groups. This platform makes an impact by being open to all health data users and accepting all appropriate content, as well as implementing key features that have not been widely available, including managing structured metadata, access via an API, and support for computable phenotypes. We have created the first openly available, programmatically accessible resource enabling the global health research community to store and manage phenotyping algorithms. Removing barriers to describing, sharing, and computing phenotypes will help unleash the potential benefit of health data for patients and the public.

Large Language Models Facilitate the Generation of Electronic Health Record Phenotyping Algorithms

Objectives:Phenotyping is a core task in observational health research utilizing electronic health records (EHRs). Developing an accurate algorithm demands substantial input from domain experts, involving extensive literature review and evidence synthesis. This burdensome process limits scalability and delays knowledge discovery. We investigate the potential for leveraging large language models (LLMs) to enhance the efficiency of EHR phenotyping by generating high-quality algorithm drafts.Materials and Methods:We prompted four LLMs—GPT-4 and GPT-3.5 of ChatGPT, Claude 2, and Bard—in October 2023, asking them to generate executable phenotyping algorithms in the form of SQL queries adhering to a common data model (CDM) for three phenotypes (i.e., type 2 diabetes mellitus, dementia, and hypothyroidism). Three phenotyping experts evaluated the returned algorithms across several critical metrics. We further implemented the top-rated algorithms and compared them against clinician-validated phenotyping algorithms from the Electronic Medical Records and Genomics (eMERGE) network.Results:GPT-4 and GPT-3.5 exhibited significantly higher overall expert evaluation scores in instruction following, algorithmic logic, and SQL executability, when compared to Claude 2 and Bard. Although GPT-4 and GPT-3.5 effectively identified relevant clinical concepts, they exhibited immature capability in organizing phenotyping criteria with the proper logic, leading to phenotyping algorithms that were either excessively restrictive (with low recall) or overly broad (with low positive predictive values).Conclusion:GPT versions 3.5 and 4 are capable of drafting phenotyping algorithms by identifying relevant clinical criteria aligned with a CDM. However, expertise in informatics and clinical experience is still required to assess and further refine generated algorithms.

EHR Phenotyping Methods for Measuring Treatment Adherence Among People Living With HIV in All of Us: Towards Disparities and Inequalities in HIV Care Continuum.

HIV treatment adherence is among the most important determinants of HIV outcomes. However, only 50% of people living with HIV in the US were retained in care. Measuring HIV treatment adherence in the clinical settings is feasible but when it comes to the growing number of multi-site Electronic Health Records (EHR), there has been a dearth of research for adequate informatics methods to handle EHR. We sought to address this gap by developing a cluster of metrics for measuring HIV treatment adherence via EHR phenotyping methods. Our methods were developed and tested in the All of Us research program. We also performed preliminary analyses to explore disparities in HIV treatment adherence and demographic factors contributing to poor adherence. This study paves the way for systematic data mining and analyses for the HIV care continuum, disparities, and inequality research on All of Us and other EHR normalized with the OMOP Common Data Model.

Developing and evaluating pediatric phecodes (Peds-Phecodes) for high-throughput phenotyping using electronic health records.

Developing and evaluating pediatric phecodes (Peds-Phecodes) for high-throughput phenotyping using electronic health records.

Prognosis, characteristics, and provision of care for patients with the unspecified heart failure electronic health record phenotype: a population-based linked cohort study of 95262 individuals

Whether the accuracy of the phenotype ascribed to patients in electronic health records (EHRs) is associated with variation in prognosis and care provision is unknown. We investigated this for heart failure (HF, characterised as HF with preserved ejection fraction [HFpEF], HF with reduced ejection fraction [HFrEF] and unspecified HF). We included individuals aged 16 years and older with a new diagnosis of HF between January 2, 1998 and February 28, 2022 from linked primary and secondary care records in the Clinical Practice Research Datalink in England. We investigated the provision of guideline-recommended diagnostic investigations and pharmacological treatments. The primary outcome was a composite of HF hospitalisation or all-cause death, and secondary outcomes were time to HF hospitalisation, all-cause death and death from cardiovascular causes. We used Kaplan-Meier curves and log rank tests to compare survival across HF phenotypes and adjusted for potential confounders in Cox proportional hazards regression analyses. Of a cohort of 95,262 individuals, 1271 (1.3%) were recorded as having HFpEF, 10,793 (11.3%) as HFrEF and 83,198 (87.3%) as unspecified HF. Individuals recorded as unspecified HF were older with a higher prevalence of dementia. Unspecified HF, compared to patients with a recorded HF phenotype, were less likely to receive specialist assessment, echocardiography or natriuretic peptide testing in the peri-diagnostic period, or receive angiotensin-converting enzyme inhibitors, beta blockers or mineralocorticoid receptor antagonists up to 12 months after diagnosis (risk ratios compared to HFrEF, 0.64, 95% CI 0.63-0.64; 0.59, 0.58-0.60; 0.57, 0.55-0.59; respectively) and had significantly worse outcomes (adjusted hazard ratios compared to HFrEF, HF hospitalisation and death 1.66, 95% CI 1.59-1.74; all-cause mortality 2.00, 1.90-2.10; cardiovascular death 1.77, 1.65-1.90). Our findings suggested that absence of specification of HF phenotype in routine EHRs is inversely associated with clinical investigations, treatments and survival, representing an actionable target to mitigate prognostic and health resource burden. Japan Research Foundation for Healthy Aging and British Heart Foundation.

Electronic Phenotype for Advanced Chronic Kidney Disease in a Veteran Health Care System Clinical Database: Systems-Based Strategy for Model Development and Evaluation.

Identifying advanced (stages 4 and 5) chronic kidney disease (CKD) cohorts in clinical databases is complicated and often unreliable. Accurately identifying these patients can allow targeting this population for their specialized clinical and research needs. This study was conducted as a system-based strategy to identify all prevalent Veterans with advanced CKD for subsequent enrollment in a clinical trial. We aimed to examine the prevalence and accuracy of conventionally used diagnosis codes and estimated glomerular filtration rate (eGFR)-based phenotypes for advanced CKD in an electronic health record (EHR) database. We sought to develop a pragmatic EHR phenotype capable of improving the real-time identification of advanced CKD cohorts in a regional Veterans health care system. Using the Veterans Affairs Informatics and Computing Infrastructure services, we extracted the source cohort of Veterans with advanced CKD based on a combination of the latest eGFR value ≤30 ml·min-1·1.73 m-2 or existing International Classification of Diseases (ICD)-10 diagnosis codes for advanced CKD (N18.4 and N18.5) in the last 12 months. We estimated the prevalence of advanced CKD using various prior published EHR phenotypes (ie, advanced CKD diagnosis codes, using the latest single eGFR <30 ml·min-1·1.73 m-2, utilizing two eGFR values) and our operational EHR phenotypes of a high-, intermediate-, and low-risk advanced CKD cohort. We evaluated the accuracy of these phenotypes by examining the likelihood of a sustained reduction of eGFR <30 ml·min-1·1.73 m-2 over a 6-month follow-up period. Of the 133,756 active Veteran enrollees at North Florida/South Georgia Veterans Health System (NF/SG VHS), we identified a source cohort of 1759 Veterans with advanced nondialysis CKD. Among these, 1102 (62.9%) Veterans had diagnosis codes for advanced CKD; 1391(79.1%) had the index eGFR <30 ml·min-1·1.73 m-2; and 928 (52.7%), 480 (27.2%), and 315 (17.9%) Veterans had high-, intermediate-, and low-risk advanced CKD, respectively. The prevalence of advanced CKD among Veterans at NF/SG VHS varied between 1% and 1.5% depending on the EHR phenotype. At the 6-month follow-up, the probability of Veterans remaining in the advanced CKD stage was 65.3% in the group defined by the ICD-10 codes and 90% in the groups defined by eGFR values. Based on our phenotype, 94.2% of high-risk, 71% of intermediate-risk, and 16.1% of low-risk groups remained in the advanced CKD category. While the prevalence of advanced CKD has limited variation between different EHR phenotypes, the accuracy can be improved by utilizing two eGFR values in a stratified manner. We report the development of a pragmatic EHR-based model to identify advanced CKD within a regional Veterans health care system in real time with a tiered approach that allows targeting the needs of the groups at risk of progression to end-stage kidney disease.