Implementing Pharmacogenomics Clinical Decision Support: A Comprehensive Tutorial on how to Integrate the Epic Genomics Module.

Implementing Pharmacogenomics Clinical Decision Support: A comprehensive tutorial on How to Integrate the Epic Genomics Module - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Clin Pharmacol Ther . Author manuscript; available in PMC: 2026 Jun 1. Published in final edited form as: Clin Pharmacol Ther. 2025 Feb 17;117(6):1670–1681. doi: 10.1002/cpt.3599 Implementing Pharmacogenomics Clinical Decision Support: A comprehensive tutorial on How to Integrate the Epic Genomics Module Bradley T Hall 1, 2, * , Eda Eken 1, 2, * , Larisa H Cavallari 1, 2 , Julio D Duarte 1, 2 , Kristin W Wiisanen 1, 2 , Emily J Cicali 1, 2, # , Khoa A Nguyen 1, 2, # Author information Article notes Copyright and License information 1. Department of Pharmacotherapy and Translational Research, College of Pharmacy, University of Florida, Gainesville, FL, United States 2. Center for Pharmacogenomics and Precision Medicine, University of Florida, Gainesville, FL, United States * Co-first authors # Co-senior authors KW currently at Rosalind Franklin University College of Pharmacy # AUTHOR CONTRIBUTIONS B.T.H. and E.E. contributed equally to this work as co-first authors. E.J.C. and K.A.N. contributed equally as co-senior authors. Writing—original draft preparation, B.T.H. and E.E.; writing—review and editing, all authors; visualization, B.T.H. and E.E.; supervision, L.H.C., J.D.D., K.W.W., E.J.C., and K.A.N.. All authors have read and agreed to the published version of the manuscript. ✉ CORRESPONDING AUTHOR Khoa Nguyen khoanguyen@cop.ufl.edu Issue date 2025 Jun. PMC Copyright notice PMCID: PMC12183801  NIHMSID: NIHMS2083573  PMID: 39960348 The publisher's version of this article is available at Clin Pharmacol Ther Abstract In the past decade, pharmacogenomic (PGx) testing to predict drug response have emerged into clinical care. Clinical decision support (CDS) has and continues to play a key role in educating prescribers and facilitating the integration of pharmacogenomic results into routine clinical practice. The Epic Genomics module, an add-on to Epic’s base clinical software, allows for storage of structured genomic data and provides electronic heath record tools designed with PGx CDS implementation in mind. In early 2022, the University of Florida Health deployed the Genomics module. This tutorial outlines the steps taken by the University of Florida Health Precision Medicine Program to implement Epic’s Genomic Module at University of Florida Health and identifies key factors for a successful implementation. Keywords: Epic genomics module, Genomic Indicators, pharmacogenomics, pharmacogenetics, integration, clinical decision support, implementation INTRODUCTION Pharmacogenomics (PGx) is a field of study focused on how genetic variation affects an individual’s risk for adverse drug effects or drug ineffectiveness[ 1 ]. In the past decade, the translation of PGx associations with drug response has emerged into clinical care[ 2 ]. The application of PGx to drug prescribing has the potential to improve medication-related outcomes and reduce cost burdens for health systems[ 3 – 6 ]. Despite the potential benefits, implementation of PGx into routine clinical practice has been a gradual process. Implementation efforts across the nation have largely been limited to a small number of early adopters, often those involved in PGx research networks or large academic medical centers[ 7 ]. Traditional educational approaches to disseminate PGx knowledge are essential but may not be sufficient given the dynamic and rapid nature of genomic discovery[ 8 – 10 ]. Clinical decision support (CDS) plays a key role in educating prescribers and facilitating the adoption of PGx testing into clinical practice. Effective PGx CDS tools, such as interruptive alerts or consult notes in the medical record, are intended to provide clinicians with clear guidance on how to interpret relevant PGx test results and apply them to pharmacotherapy decisions. Unfortunately, it has taken time for commercial electronic health record (EHR) vendors to respond to the unique infrastructure needed for PGx-specific decision support. In this void, implementation and maintenance of PGx CDS has been primarily built in-house and led by pharmacists[ 7 ]. Many institutions have resorted to using EHR tools originally designed for alternative purposes, such as the problem list, allergy list, or laboratory results, to store information derived from PGx testing, which have significant limitations in how results and CDS are presented and disseminated.[ 11 ] While these alternative methods address certain gaps in PGx clinical implementation, they lack the comprehensive functionality required for a complete solution. Specifically, they fall short in critical areas such as storing and transferring genomic results, translating clinical interpretations from PGx findings, and developing streamlined CDS systems for PGx-guided medication ordering and management. More recently, commercial EHR vendors have taken steps to address limitations with PGx integration into the health record. One such example is the Epic’s Genomics Module, an add-on to Epic’s base clinical software that allows for storage of structured genomic data and provides EHR tools designed with PGx CDS implementation in mind. The Genomic Module was developed to present clinically relevant information, enabling physicians, pharmacists, and other healthcare professionals to access and utilize genomic results for patient care and treatment decisions. Historically, The UF Health genomic information was housed in traditional lab results components (e.g., similar to HgA1c, sodium level, calcium level), resulting in unreliable clinical workflows, missed opportunities for genotype-based care, and impractical outcomes tracking. Epic’s Genomics Module provides an accessible, actionable, and measurable solution in integrating the use of PGx information into normal clinic workflows. In early 2022, the University of Florida UF Health embarked on deploying Epic’s Genomics Module, and herein, we describe our process for integrating the module into our existing PGx infrastructure. The objective of this tutorial is to provide practical guidance to implement Epic’s Genomics Module or similar PGx tools provided by commercial EHRs. It is important to note that while Epic’s Genomics Module allows for implementation of both germline and somatic genetic information, this tutorial focuses solely on implementation of germline genetic information (e.g., PGx genes.) Institutional background. The UF Health Precision Medicine Program (UF Health PMP) consists of a multidisciplinary team, led by the College of Pharmacy, that facilitates the integration of PGx-guided therapy into clinical practice to improve patient outcomes (e.g., optimize medication effectiveness, limit drug toxicities). The majority of PGx testing at UF Health is completed internally by the UF Health Clinical Pathology Laboratory, with a fraction of tests performed by outside laboratories. Before Epic’s Genomics Module implementation, PGx testing results were electronically interfaced with the EHR as discrete data for internal tests, or as pdf reports if completed by an outside commercial laboratory. To date, these discrete PGx data (i.e., genotype, phenotype, activity score) have been stored in the EHR as a laboratory component result, also known as long read report (LRR) ( see Figure 1 ). UF Health Pathology currently offers several single PGx tests (e.g., CYP2C19 , CYP2D6 , TPMT , NUDT15 ) and a multi-gene panel, GatorPGx, which provides data for seven genes (i.e., CYP2C19, CYP2D6, CYP2C9, CYP3A5, SLCO1B1, VKORC1, CYP4F2 ) and a variant in the CYP2C cluster[ 12 ]. The UF Health PMP works closely with the pathology team to design and validate testing for these pharmacogenes, all of which are associated with clinically actionable drug recommendations as defined by the Clinical Pharmacogenetics Implementation Consortium (CPIC)[ 13 ]. Figure 1. Open in a new tab Genomics Module Summary Abbreviations – VAR: resulted as a genomic variant; LRR: resulted as a traditional lab; Geno: genotype; Pheno: phenotype; AS: activity score; GM: genomics module; GDI: gene-drug interaction; PGx: pharmacogenomics; BPA: Best practice advisories; CDS: clinical decision support The UF Health PMP team was assigned to oversee all PGx-related components of the Epic Genomics Module implementation. In this tutorial, we outline the steps necessary to implement Epic’s Genomics Module and share our institutional experience. These steps can serve as a guide for other institutions seeking to implement and execute Epic’s Genomics Module, helping them anticipate potential challenges and leverage lessons learned during our implementation and troubleshooting processes. Figure 2 outlines all important steps discussed in this manuscript as well as significant items in each step. Of note, the Epic ® functionality described in this tutorial reflects version February 2024. Figure 2. Open in a new tab Overview of steps and actions recommended for PGx implementation Note: Italicized actions are optional depending on the capability of the institution STEP ONE: Assess Existing EHR Infrastructure and Potential Value of Epic’s Genomics Module Implementation If the institution has already implemented PGx, it is important to assess the existing infrastructure to ensure that Epic’s Genomics Module will be of added value as it is not without cost. Although UF Health had already implemented and integrated PGx into our EHR, it was largely home-grown, and the health system determined that the Epic Genomics Module would significantly improve the functionality and effectiveness of our PGx services based on the key benefits described below. For institutions that have not yet implemented PGx, the benefits outlined below offer a compelling rationale for why incorporating PGx within Epic’s Genomics Module might be a prudent choice. Unique EHR data structure for PGx test results. Epic’s Genomics Module provides the capability to store PGx data (e.g., genotype, phenotype) in a location within the EHR specifically designed for such a purpose (e.g., a specific section of the chart dedicated to genetic results and interpretation). This not only has a benefit for visibility of PGx results in the medical record, but also allows for the genotype and phenotype to be linked, which is unique for PGx results and differs from other laboratory results. With Epic’s Genomics Module, Genomic Variants serve as PGx genotype repositories, while Genomic Indicators are phenotype repositories. Discrete PGx test results, whether uploaded as Genetic Variants or historic discrete PGx laboratory results (i.e., LRR), can be linked to the correct Genomic Indicator (e.g., phenotype) automatically by the Epic Translation Engine ( see Figure 1 ). The translation engine populates the correct Genomic Indicator in the patient’s chart based on mapping tables stored in the EHR. In the foundation package offered by Epic, there are prepopulated mapping tables available for many well-developed PGx genes. Institutions can use these tables for the gene(s) they are implementing or build their own mapping tables to meet the specific needs of the institution. One benefit of utilizing the translation engine is that it allows for real-time updates to patient phenotypes or Genomic Indicator. For example, if emerging evidence results in a change in the clinical standard for a phenotype definition (e.g., CYP2D6 *1/*4 was recently reassigned from predicting a normal metabolizer to an intermediate metabolizer phenotype), the mapping table can be updated, and past results will have their associated phenotypes automatically updated, mitigating the need for manual updates. This contrasts with how UF Health has historically stored discrete genotype and phenotype data. Stored as laboratory results, phenotype data was reported by the UF Health Pathology Laboratory at the same time genotype results were returned. If there was an update to how a phenotype should be defined, the laboratory would have to manually re-report the phenotype for each individual patient. To facilitate access the Translation Engine content, Epic created a Turbocharger package for the mapping tables and rules to support CPIC mapping starting in February 2021. Epic outlines the specific mapping process within the Genomics Setup and Support Guide accessible through Epic’s UserWeb. Visibility of PGx results and recommendations for clinicians and patients. Epic’s Genomics Module provides locations within a patient’s record to view PGx results and interpretations for both clinicians and patients. When Genomic Indicator are added to a patient’s chart via the translation engine, clinicians can navigate to the “Genomic Indicators” activity. Here, Genomic Indicator are presented as visual flags to indicate the genetic risk that a patient may have with various medications. They can display information at the gene level or for gene-drug interactions (GDIs). Genomic interactions link to PGx-specific icons and can be assigned a level of severity, which correlates with a color (i.e., red, yellow, green, grey)[ 14 ]. The icons can be displayed on the order entry search window to alert clinicians to medications with potential PGx interactions and match the icons found in the Genomic Indicator activity. A clinician with the appropriate security access can add comments for other clinicians to view. Patients can view their PGx test results and corresponding information by navigating to “My Genetic Profile” within MyChart. At UF Health, clinicians have historically relied on CDS alert pharmacist consultation notes placed in the medical record for interpretation of PGx results. The notes are routed to the clinician through the Epic In-Basket and are useful at the time of result return and note placement. Although static, the consult note contains information that is applicable over the course of a patient’s lifetime. Unfortunately, the consult note gets buried in the medical record over time, and future clinicians may not be aware that the note exists. Through the completion of several pilot studies, we found that clinicians indicated a PGx section of the chart would be useful for this life-long information[ 15 ]. The Epic Genomic Module addresses this need through the Genomic Indicator activity, improving the accessibility and utility of PGx information. As one clinician noted, “This [Genomic Indicator activity] would be super helpful...I have patients who state, ‘I don’t tolerate statins’...it’d be helpful to know if there were certain ones they may tolerate better so they can reduce their cardiovascular risk.” Another clinician in palliative care commented, “This would be useful in determining what pain medications I can use for my cancer patients.” Multiple PGx CDS capabilities. In addition to the Genomic Indicator activity, which passively displays patient-specific PGx information in the chart, the Epic Genomic Modulesupports additional CDS capabilities to warn clinicians of specific drug interactions in real time. Real time CDS can appear when clinicians are ordering medications, providing relevant information at the ideal time within a clinician’s workflow. Epic has long supported interruptive CDS in the form of Our Practice Advisories (OPAs) which is formerly known as Best Practice Advisory, but with the Epic Genomic Modulethese alerts can now be triggered by a Genomic Indicator ( see Figure 3 for a screenshot of a PGx OPA ). At UF Health, our legacy PGx OPAs were developed over several years by a range of individuals, resulting in inconsistent alert triggers (e.g., genotype, phenotype). By streamlining our trigger criteria to rely on Genomic Indicator for our existing 50 OPAs we could reduce the overall maintenance required. Additionally, a new CDS tool provided in the Epic Genomic Module is PGx In-line Medication Warnings ( see Figure 4 as an example ). This new tool employs a novel approach by integrating alerts directly into the clinician’s current order entry workflow, rather than using disruptive pop-up notifications. At UF Health, this in-line medication warning system will replace certain conventional alerts that solely recommend dose adjustments, streamlining the process for clinicians to accept recommendations with fewer clicks. By implementing this in-line warning method, the risk of alert fatigue from PGx-CDS notifications may be significantly reduced, potentially enhancing overall clinical efficiency and decision-making. Figure 3. Open in a new tab Clinician view of UF Health Our Practice Advisory Figure 4. Open in a new tab UF Health In-line medication warning at clopidogrel order entry informing clinician of actionable CYP2C19 phenotype results STEP TWO: Clinical Language Development Once the relevant gene-drug pairs are identified for implementation (i.e., gene-drug pairs with strong evidence for clinical utility), institutions must determine how they will customize the display of the Epic Genomic Moduleand revise or develop accompanying language. As described above, we implemented Genomic Indicator to represent the phenotype interpretation and genomic interactions to describe GDI implications and recommendations. As such each gene has multiple Genomic Indicator to reflect the individual’s predicted gene activity. For example, CYP2C19 has five indicators: poor metabolizer, intermediate metabolizer, normal metabolizer, rapid metabolizer, and ultrarapid metabolizer. The specific Genomic Indicator that populates within an individual’s profile is dependent on their specific genotype result. Standard GDI will populate as interactions under the related Genomic Indicator. For example, citalopram will always appear under CYP2C19 Genomic Indicator but the specific language depends on the patient’s CYP2C19 Genomic Indicator. Overall, we established four areas of language to develop for each possible gene activity (i.e., phenotype); (1) clinician-facing Genomic Indicator, (2) patient-facing Genomic Indicator, (3) clinician-facing genomic interactions, and (4) patient-facing genomic interactions. Review of existing PGx clinical language. Epic’s Genomic Module foundation includes clinical content templates for use in PGx CDS. Epic recommends reviewing for clinical accuracy and appropriateness and revising as needed prior to implementing. After our program reviewed the language, we determine that the content as written for genes intended for implementation was not ideal for our institution and preferred to use language that had been developed and refined over the last decade-plus by the UF Health PMP in close consultation with prescribers. For example, the stock language for the CYP2C19 poor metabolizer - clopidogrel interaction content was “Genetic results for this patient indicate a high sensitivity to clopidogrel (PLAVIX)[ 16 ], based on CYP2C19 genotype.” Our group determined this to be misleading, as poor metabolizers would have impaired activation of clopidogrel, leading to reduced drug effectiveness. Because of concerns about inaccuracies, such as this one, and given that our institution had an existing and robust archive of clinical language and recommendations to populate large portions of the Genomic Indicator and other PGx CDS, we elected to use out existing content. For institutions that do not have existing language, we recommend review of the stock language for accuracy or leveraging suggested language from reputable resources (e.g., CPIC) in addition to working with clinicians involved in the implementation to modify the language to best fit clinician preferences for receiving result interpretation and genotype-based recommendations[ 17 ]. Our institutional process for developing and refining language for GDIs has been previously described and can be used as a guide[ 18 ]. Genomic Indicator language. The purpose of Genomic Indicator is to provide gene-level information to clinicians and patients. Hence, we standardized formatting for both clinician- and patient-facing Genomic Indicator language, which includes four components: (1) Phenotype with general description of enzyme or transporter activity, (2) medications impacted by variability and relevant inhibitors/inducers if applicable, (3) contact information of the UF Health PMP and (4) instructions on how to use the indicator if discordant results appear in a patient’s profile. These components align with American College of Medical Genetics and Genomics (ACMG) recommendation of reporting genotype and metabolizer phenotypes (where applicable), providing a list of medications that may be affected by the identified genotype, clearly stating that the accuracy of phenotypic prediction is dependent on the variants detected, and providing a list of resources that could inform actionable decisions (e.g., CPIC guidelines, FDA tables)[ 19 ]. We created Genomic Indicator for all possible predicted phenotypes for each gene (e.g., poor metabolizer, normal metabolizer), as well as Genomic Indicator to indicate assay failure or if the phenotype was unknown. Our institution has range phenotypes for certain genes, which required creating a unique Genomic Indicator. We developed gene-independent standard language for “assay failure” (i.e., the analysis failed to yield an informative result and thus no genotype is reported) and for “unknown phenotype” (i.e., individual is carrying at least one allele with uncertain/unknown function and the phenotype cannot be determined at the time of result report). Unique language was developed for clinicians and patients following the same described standardized format. Templated language was developed to describe the phenotype interpretation appropriate for both audiences. For example, clinician-facing reports state, “very increased CYP2C19 enzyme activity” while patient-facing reports state, “based on your results, you have very increased CYP2C19 enzyme activity.” Medications impacted by the result are included in both clinician and patient-facing reports, but for clinicians, language is included about impacting blood levels as compared with patient language, which says “the enzyme breaks down or activates medications such as...”. Contact information for clinicians lets them know how to get in touch with the PMP for questions and provides links to educational resources. We refer patients to our clinical service, MyRx[ 20 ], where they can opt in for an educational visit with a pharmacist[ 21 ]. The fourth component about discordant results is only relevant for clinician-facing, and we felt was necessary to include because of the existence of both LRR and VAR (i.e., resulted as a Genomic Variant) results in our medical records ( see Figure 1 ). The translator will update the indictors in real time, but the laboratory report will not be updated and as such may result in discordant results. The UF Health PMP is educating our clinicians to use the Genomic Indicator as the primary place to look for phenotype result and interpretation. For patient-facing language, it is important to ensure it is developed in a way that allows for patients to better understand PGx. Many studies have demonstrated that patient education can increase knowledge, decrease anxiety, and improve clinical decision-making[ 22 – 24 ]. It is recommended that the communication of PGx test results to patients focus on the interpretation or meaning of the results (e.g., phenotype) instead of the scientific description of results (e.g., genotype). It has also been recommended to avoid stiGenomic Moduleatizing terms such as “mutant,” “defective,” or “abnormal” since these terms can have negative connotations. Instead, use of neutral terms such as “working” or “not working,” is advised to describe the function of enzymes. Our PMP team decided to utilize patient-friendly language for phenotype interpretation that refer to “enzyme activity”. Additionally, while terms such as drug metabolism and gene expression are reported in research, these may not be ideal when considering patient understanding of these topics. Addressing patient health literacy and genetic literacy in PGx is crucial to empowering patients for widespread precision medicine adoption. Genomic interaction language. The purpose of genomic interactions is to provide specific GDI interpretations and recommendations. To avoid providing medication-related recommendations to patients, we developed separate standardized formats for clinician-facing and patient-facing language. For clinician-facing language, we maintained the format of our existing CDS that was implemented after many revisions based on clinician feedback to be concise and to provide a specific drug recommendation. This format is split into a 1) problem statement, 2) recommendation, and 3) drug-specific link to additional educational information ( see Figure 3 ). The problem statement indicates genotype implications for drug response and potential clinical impact. The recommendation statement is specific and includes alternatives if applicable. For situations where we recommend using the standard dose, we opted to keep “problem” for consistency but wrote “N/A”. When possible, we combined multiple medications within the same class to limit the number of unique interactions that needed to be created. For example, PPIs have one interaction. For patient-facing language, we developed a standard bank of options that can be used across all gene-drug pairs. The standard bank options include, “normal response expected”, “increased risk of treatment failure”, and “increased risk of side effects”. There are exceptions to when these are more detailed for specific gene-drug pairs (e.g., specifying with chronic use for proton pump inhibitors) ( see Supplementary Table 2 ). STEP THREE: IT builds within the EHR Following the development of the language to be used, Epic analysts within the institution must build and maintain the technical infrastructure within the EHR. Genomic Indicators and genomic interactions. For Genomic Indicator, we developed and implemented a standardized naming convention, listing the PGx enzyme/transporter name followed by the PGx phenotype (e.g., CYP2C19 Poor Metabolizer). While Genomic Indicator gene-level content was input directly into the Genomic Indicator build as rich text, the content for the GDIs was populated using SmartText, a text repository that can be reused at multiple locations. This allows for language to easily be re-used and maintained, for example, as described above the interaction text mimics our OPAs when applicable. We also leverage a clickable hyperlink within each Genomic Indicator to bring the user to the complete PGx laboratory report if desired. Each institution can select the user types who can view or edit Genomic Indicator. The UF Health PMP decided it was best practice for our institution to provide prescribers, pharmacists, and select PGx research study staff read-only access of the Genomic Indicator activity. These individuals are who we expect to use the Epic GENOMIC MODULE in their typical workflow. You can also select the users, if any, who will be able to edit Genomic Indicator. As described above, Genomic Indicator will populate based on the translation engines, but there are certain scenarios that will require manual addition of Genomic Indicator from PDF laboratory reports, or based on untranslated results, as described further below. As such, we opted to have our PMP PGx pharmacists be the only users able to edit Genomic Indicator. These individuals are responsible for maintaining our institution’s policy on acceptable PGx results to be used in clinical care and as such it made sense to only have these individuals as editors of the content. An example of the Genomic Indicator activity is shown in Figure 5 . Figure 5. Open in a new tab Navigating the genomic indicator activity In developing patient-accessible genetic indicators (Genomic Indicator) and interactions, the tool provide options to automatically share results with patients or allow clinicians to selectively release them. The Cures Act mandates that our institution release all laboratory results, including pharmacogenomic (PGx) findings, to patients. Consequently, we chose to make patient-facing Genomic Indicator available to enhance patients’ understanding of their PGx results. As a result, all Genomic Indicator and related interactions were automatically shared with patients through their patient portal, specifically in the “My Genetic Profile” section. This approach aims to empower patients with direct access to their genetic information and its potential impact on their healthcare. Figure 6 provides an example on how PGx information can be shown to patient in MyChart. Figure 6. Open in a new tab MyChart patient portal view of “My Genetic Profile” displaying PGx test results and corresponding clinician information Unresolved indicator workflow. Due to variations in PGx reporting formats and lack of standardization, automatic translation of all PGx results is challenging, particularly for those from external laboratories. For instance, some results may be presented in non-discrete formats (e.g., PDF files) or use terminology not included in our mapping tables (e.g., reported as “negative”). Additionally, human errors during data entry can lead to incorrect genotype inputs (e.g., “1/*1” instead of “*1/*1”). In such cases, automatic mapping of genotype results is not possible. To address this issue, we implemented an “Unresolved Indicator” system. This serves as a verification gateway for genetic laboratory results that couldn’t be matched to existing Genomic Indicator. “Unresolved Indicator” appear on a patient’s Genomic Indicator list until resolved but are not visible to patients. When the translation engine assigns an “Unresolved Indicator” to a genotype, it triggers an In-Basket message to our Precision Medicine Pharmacist team. A pharmacogenomics specialist then manually reviews the genotype associated with this indicator, determines the appropriate Genomic Indicator, removes the “Unresolved Indicator”, and links the laboratory result to the correct Genomic Indicator. This process integrates seamlessly with our institution’s workflow, as our pharmacists routinely monitor a shared In-Basket pool for result notifications and other important alerts. Genomic Indicators as a repository for activity score. An Activity Score in pharmacogenomics is a numerical value used to quantify the relative enzymatic activity of genetic variants compared to the normal or wild-type allele, typically assigning values of 0 (no function), 0.5 (decreased function), or 1 (normal function) to each allele[ 25 ]. This scoring system facilitates the translation of complex diplotype data into a patient’s phenotype to guide drug therapy and is central to many guidelines issued by the CPIC[ 26 ]. Certain medication recommendations are based on activity score (AS) rather than just the phenotype. For example, recommendations for NSAIDs for intermediate metabolizers differ depending on if the CYP2C9 AS is 1 or 1.5[ 27 ]. During our implementation (version February 2024), Epic only has only one option to enter AS as a part of the result in the VAR report similar to the genotype. However, this AS data did not interface with any other Epic clinical data, and therefore had no functional capabilities. This means that CDS could not utilize the AS data entered into Epic. To solve this, we used Genomic Indicator as the repository for the AS. This resulted in creating multiple Genomic Indicator for one phenotype. For example, for CYP2C9, we created two separate Genomic Indicator records for CYP2C9 Intermediate Metabolizer; those with an AS of 1.0 versus an AS of 1.5. It is important to note that Epic has updated its AS properties at the time of this writing. With the new properties, AS can be associated with Genomic Indicator as either a number or a range in addition to VAR results. This update will eliminate the need to create multiple Genomic Indicator entries for different activity score values. OPA can also evaluate results based on both Genomic Indicator and AS values. Translation engine mapping table validation. It is important to ensure accuracy in the mapping tables used by the translation engine ( see Figure 1 ). We reviewed every table provided by Epic and compared it to our institutional genotype to phenotype mapping. This was an important step as the mapping tables provided by Epic included genotypes that we did not test for as well as genotype-phenotype mappings that were not used by our institution. We also updated these tables for the use of Genomic Indicator as activity score repositories, such as in the case of the CYP2D6 mapping table. PGx Icons. As described previously, PGx icons are linked to Genomic Interactions and display at the order entry search window to clue in prescribers to medications with potential PGx interactions. These icons colors (i.e., red, yellow, green, grey) are visually appealing to the prescriber, but the corresponding severity language is linked to the same patient-facing interactions. For example, if we used the red version of the PGx icon to alert the clinician to a severe GDI at order entry, the patient would see in their patient portal that the interaction was severe enough to be labeled with “Alert.” Similar language occurred with the yellow and green icons linking to the language of “Warning” and “Safe” respectively. Language such as this was deemed potentially dangerous as it could risk a patient stopping their medication inappropriately. As such, our institution opted to only use the grey version of the PGx icons that appeared at order entry, which provided the neutral label of “Info.” This challenge, however, will be resolved in the future upgrade of Epic’s GENOMIC MODULE, as Epic is planning to remove the connection between color-coded PGx icons with “concerned” languages for patient-facing CDS. Interruptive alert updates. The Epic GENOMIC MODULE allows for OPAs to trigger based off the Genomic Indicator. Our institution already had interruptive alerts in place that used genotype or phenotype results as the trigger criteria, so we had to update the trigger criteria. For institutions initiating CDS with GENOMIC MODULE, it is important to start out by building Genomic Indicator as the trigger criteria as alerts will remain current and relevant for the patient even if result translations are updated in the translation engine. Epic’s GENOMIC MODULE also provides a unique PGx symbol to be used in interruptive alerts that makes these alerts stand out and ideally help with alert fatigue. We added this when we updated our alerts to assist clinicians in quickly recognizing PGx alerts when they appeared. STEP FOUR: Develop Standards of Practice As PGx is a field of rapidly emerging evidence, continuous review of the evidence landscape is warranted. As such, it should be anticipated that new genes and GDIs will be implemented over time as well as updates needed to the translation tables. We recommend creating a Standard of Practice (SOP) to ensure the institution maintains their approach for steps two-four. UF Health PMP’s SOP is summarized in Supplementary Information . Another important benefit for the SOP is to allow standard procedures to be implemented over time even with new personal. It can refer to templated language, directions for trouble shooting, protocols to put into production, how to update translator files, etc., If possible, we recommend creating the SOP prior to drafting the Epic GENOMIC MODULE language as standardized formatting may be beneficial in streamlining the language development process to ensure timely and uniform implementation. STEP FIVE: Post-Implementation Maintenance PGx clinical practice guidelines (e.g., CPIC guidelines; Dutch Pharmacogenetics Working Group)[ 28 ] and the FDA Table of Pharmacogenetic Associations[ 13 ] are periodically updated and can result in major changes to prescribing guidance. However, guideline revisions may lag several months to years after landmark studies are published as they require a thorough evidence evaluation and establishment of consensus from expert authors. Therefore, it is important to plan for regular literature review and discussion with institutional stakeholders to determine continued appropriateness of developed PGx guidance. Recent examples of changing guidance from CPIC guideline updates include revised dose adjustment recommendations (SSRIs and CYP2D6 ), expansion of recommendations to other medications within a drug class (NSAIDs and CYP2C9 ), and consideration of additional genes in clinical decision-making (e.g. antidepressants and CYP2B6 [ 29 ]). We recommend implementing the following review schedule: Quarterly literature scans for new PGx studies and guideline updates. Biannual stakeholder meetings (e.g. clinical pharmacists, geneticists, relevant medical specialists) to evaluate the implications of recent findings and pontential needs for guideline revisions. Annual comprehensive review to assess all implemented PGx guidance, comparing it with the latest published guidelines and significant studies. Ad hoc reviews when major studies or guideline updates are published that could substantially impact current practice. Following these reviews, components of the Genomic Module and associated OPAs should be promptly updated to maintain consistency with the latest evidence-based recommendations. This systematic approach ensures that the institution’s PGx guidance remains current and clinically relevant. DISCUSSION The UF Health PMP, in close collaboration with the UF Pathology Laboratory and the UF IT department, successfully implemented the PGx components of the Genomic Module. The following key components were identified as critical for a successful implementation: Interdisciplinary team is key to a successful implementation. An interdisciplinary team comprising of PGx clinical experts, clinical informatics specialists, and IT analysts was critical to guarantee a successful implementation of Epic’s Genomic Modulein our healthcare system. PGx clinical experts played an essential role given their knowledge of PGx principles, GDIs, and clinical guidelines. As the incorporation of PGx components into Epic’s Genomic Moduleis a relatively new concept with limited guidance from clinical and technical sources, the expertise of PGx clinicians was invaluable. They supported the development of language tailored to both healthcare clinicians and patients, who are the primary users of Genomic Module, ensuring that the information was accurate, clear, and actionable. Clinical informatics specialists were instrumental in integrating PGx data into the EHR system and CDS tools. Their expertise was crucial in developing workflows and processes for incorporating new Genomic Indicator, as well as converting existing alert systems to accommodate these indicators as essential factors. Furthermore, they ensured interoperability and seamless data exchange between laboratory results and CDS systems within the Epic EHR platform. Finally, clinical informatics specialists with expertise in pharmacogenomics played a vital role in the successful integration. Their deep understanding of both clinical informatics and PGx allowed them to design user-friendly interfaces that presented complex genetic information in an easily interpretable format for clinicians. IT analysts were responsible for the technical implementation and integration of Epic’s Genomic Modulewithin our healthcare system. Given the thousands of legacy PGx results that needed to be incorporated, the IT analysts played a vital role in ensuring the smooth integration and conversion of PGx tools with the EHR. Their technical expertise was essential for overcoming any challenges related to data storage, security, and privacy requirements. By bringing together the diverse perspectives and skills of this multidisciplinary team, our healthcare organization was able to address various practical aspects of PGx implementation, including clinical knowledge, informatics workflows, and technical integration. This collaborative approach helped overcome challenges and ensured that PGx was effectively incorporated into routine clinical practice, ultimately improving patient care and medication safety. Providing passive clinical support to end-users was a resource-intensive endeavor, but it was essential for the successful implementation of the genomics module. The lack of data standardization for legacy PGx results, novel genetic variants, and newly defined phenotypes posed significant challenges in building and maintaining Genomic Indicator. Since 2011, when discrete PGx results were first implemented in our EHR system, the absence of standardized data formats and terminologies for representing PGx results hindered the accurate conversion and mapping of old results to the new Genomic Indicator. Additionally, inconsistencies in naming conventions or annotations (e.g., *1 vs. 1, or *2+ vs. *2 dup) led to inconsistent interpretations and errors in the PGx CDS. Furthermore, as new genetic variants are discovered and their clinical significance is explored, there is a constant need to update the clinical contents and decision support rules within the Genomic Indicator. The identification of newly defined phenotypes or GDIs may necessitate modifications to existing language and indicators or the creation of entirely new ones. At our institution, these critical roles are currently led and maintained by the PGx clinical expert team, which included a clinical informatics specialist. While the foundation languages developed by Epic were intended to foster the adoption of standardized formats, collaboration, and data sharing among healthcare organizations, the lack of patient specificity and accuracy in these languages required us to develop completely new languages tailored to our needs. Ideally, clinician-facing and patient-facing languages related to Epic’s Genomic Module should be shared and adapted across institutions utilizing Genomic Module, ultimately leading to the development of a universal library. This collaborative approach would not only reduce redundant efforts but also promote consistency and accuracy in the information provided to healthcare professionals and patients. Comparison with other large-scale pharmacogenomic implementations. Our experience in implementing PGx CDS aligns with the implementation described by Caraballo et al[ 30 ] at Mayo Clinic. Their approach parallels ours in key aspects, including the involvement of a multidisciplinary team, integration of PGx test results into the EHR, and provision of point-of-care CDS. They also emphasized clinician education and encountered comparable challenges, particularly the need for ongoing updates to keep pace with evolving pharmacogenomic knowledge. However, our implementation uniquely focused on converting legacy PGx results into a new, standardized format within the Epic EHR platform. We also developed new, tailored languages for both clinicians and patients, addressing the limitations of foundation languages provided by Epic. Furthermore, our approach emphasized the importance of seamless data exchange between laboratory results and CDS systems, and we tackled the challenge of incorporating thousands of legacy PGx results. These distinctive aspects of our implementation contribute valuable insights to the growing body of knowledge on large-scale pharmacogenomic CDS integration into healthcare systems. OPPORTUNITY FOR IMPROVEMENT While the implementation of Epic’s Genomic Module has been a significant step forward, we identified several limitations and suggestions for improvement to accelerate its success further. First, the Epic Genomic Module version implemented (version February 2024) lacks a sufficient reporting system to facilitate the integration of relevant genomic results into clinical notes or other reporting documents. As discussed previously, the Epic EHR system offers a feature called SmartTexts and SmartLinks, which allows users to insert predefined lab results, text or templates into clinical notes and documentation. However, at this time, SmartLink is not available to automatically pull both VAR results and Genomic Indicator from patient profiles, making it challenging for healthcare clinicians to incorporate genomic results and recommendations into their documentation seamlessly. This limitation creates a burden of toggling between different reports to check for genomic profiles and increases the risk of human errors when manually transcribing genomic results into documents. Secondly, PGx results in Genetic Variants format are currently only available within the Genomic Indicator section ( see Figure 5 ), but not in the Results Review ( see Figure 7 ), a common location where healthcare clinicians’ access and review all laboratory and procedure results. This limitation may potentially reduce the awareness and visibility of PGx results among new clinicians or those unfamiliar with the Epic Genomic Module system. Figure 7. Open in a new tab Legacy view of patients’ PGx test results within Epic’s “result review” activity. The third challenge lies in monitoring provider acceptance and feedback for passive alerts, specifically in-line medication warnings. Unlike active alerts utilizing OPA, where clinician acceptance and override rates can be tracked through workbench reports, SlicerDicer, or other retrospective analysis tools[ 31 ], assessing provider responses to in-line medication warnings proves more challenging. These passive alerts, integrated into the medication order process, do not generate a traceable “breadcrumb” trail, making it difficult to monitor provider interactions with the alert system. This lack of trackability significantly hinders our ability to evaluate the effectiveness and impact of these passive alerts on clinical decision-making. Rather we will need to make assumptions when evaluating and will depend on concordance of medications with patient’s phenotype. Fourthly, the GDIs are currently organized by gene rather than by drug. For example, all GDIs associated with the CYP2D6 gene, such as opioids- CYP2D6 and SSRI- CYP2D6 , are grouped under the CYP2D6 indicator. The system lacks an option to group and display all GDIs to a specific medication. This is particularly important for medication classes that may be affected by two genes (e.g. SSRIs). Consequently, if a clinician needs information about what the best SSRI is, they must review all relevant genomic interactions (e.g. citalopram, escitalopram, sertraline, fluvoxamine paroxetine listed under the respective CYP2D6 and CYP2C19 indicators). To work around this current limitation, we added a statement under both sertraline genomic interactions that refers the clinician to the other relevant interaction to make a comprehensive therapeutic decision. However, this gene-based organization method can potentially decrease efficiency in clinical decision-making as clinicians are directed to view two separate locations to formulate a decision on one medication, therefore, we suggest the following improvements: (a) Develop a reporting system or integrate the clinical content within Genomic Indicator, enabling healthcare clinicians to easily incorporate genomic results and recommendations into their clinical notes and documentation. (b) Display PGx results in variant format within the Results Review section, increasing their visibility and accessibility to all healthcare clinicians. (c) Implement an additional grouping option that allows users to view all GDIs specific to a particular medication, where the information and recommendations for two genes-one drug can be shown effortlessly. Finally, while the overall steps (1 through 5) outlined in this guideline can be adapted to various EHR systems, the specific implementation details are tailored for the Epic ® Genomic Module. Elements such as OPAs, genomic indicators, and foundation languages are unique to the Epic ® system. It’s worth noting that, to our knowledge, the Epic ® Genomic Module is currently the only commercially available tool with the capability to store and display comprehensive genomic profiles. Therefore, while the general principles may be broadly applicable, the technical specifics of this guideline are primarily relevant to institutions using Epic ® EHR. By addressing these limitations and implementing the suggested improvements, the implementation of Epic Genomic Module can be further enhanced, facilitating more efficient and seamless integration of genomic information into clinical practice, ultimately improving patient care and medication management. Supplementary Material Supplemental Materials NIHMS2083573-supplement-Supplemental_Materials.docx (25.2KB, docx) Footnotes CONFLICT OF INTEREST The authors declared no competing interests for this work. REFERENCES 1. Devine B, Boyce RD, and Wiisanen K, Clinical decision support for pharmacogenomic precision medicine: Foundations and Implementation. 1st edition ed. 2022, London: Academic Press. 296. [ Google Scholar ] 2. Relling MV, et al. , The Clinical Pharmacogenetics Implementation Consortium: 10 Years Later. Clin Pharmacol Ther, 2020. 107(1): p. 171–175. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Solomon HV, Cates KW, and Li KJ, Does obtaining CYP2D6 and CYP2C19 pharmacogenetic testing predict antidepressant response or adverse drug reactions? Psychiatry Res, 2019. 271: p. 604–613. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Bradley P, et al. , Improved efficacy with targeted pharmacogenetic-guided treatment of patients with depression and anxiety: A randomized clinical trial demonstrating clinical utility. J Psychiatr Res, 2018. 96: p. 100–107. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Aldrich SL, et al. , Influence of CYP2C19 Metabolizer Status on Escitalopram/Citalopram Tolerability and Response in Youth With Anxiety and Depressive Disorders. Front Pharmacol, 2019. 10: p. 99. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Greden JF, et al. , Impact of pharmacogenomics on clinical outcomes in major depressive disorder in the GUIDED trial: A large, patient- and rater-blinded, randomized, controlled study. J Psychiatr Res, 2019. 111: p. 59–67. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Haidar CE, et al. , ASHP Statement on the Pharmacist’s Role in Clinical Pharmacogenomics. Am J Health Syst Pharm, 2022. 79(8): p. 704–707. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Giri J, et al. , Education and Knowledge in Pharmacogenomics: Still a Challenge? Clin Pharmacol Ther, 2018. 103(5): p. 752–755. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Formea CM, et al. , Implementation of a pharmacogenomics education program for pharmacists. Am J Health Syst Pharm, 2018. 75(23): p. 1939–1946. [ DOI ] [ PubMed ] [ Google Scholar ] 10. Luczak T, et al. , Strategies and settings of clinical pharmacogenetic implementation: a scoping review of pharmacogenetics programs. Pharmacogenomics, 2021. 22(6): p. 345–364. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Peterson JF, et al. , Electronic health record design and implementation for pharmacogenomics: a local perspective. Genet Med, 2013. 15(10): p. 833–41. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Marrero RJ, et al. , How to Transition from Single-Gene Pharmacogenetic Testing to Preemptive Panel-Based Testing: A Tutorial. Clin Pharmacol Ther, 2020. 108(3): p. 557–565. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. FDA. US Food and Drug Administration (FDA) Table of pharmacogenetics association. Section 1: Pharmacogenetic Associations for which the Data Support Therapeutic Management Recommendations. October 26, 2022; Available from: https://www.fda.gov/medical-devices/precWoSon-medicine/table-pharmacogenetic-associations . 14. O’Donnell PH, et al. , Adoption of a clinical pharmacogenomics implementation program during outpatient care--initial results of the University of Chicago “1,200 Patients Project”. Am J Med Genet C Semin Med Genet, 2014. 166C(1): p. 68–75. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Cicali EJ, et al. , Challenges and lessons learned from clinical pharmacogenetic implementation of multiple gene-drug pairs across ambulatory care settings. Genet Med, 2019. 21(10): p. 2264–2274. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Plavix (Clopidogrel) [Prescribing Information], Bristol-Myers Squibb/Sanofi. 2020. [ Google Scholar ] 17. Kabbani D, et al. , Pharmacogenomics in practice: a review and implementation guide. Front Pharmacol, 2023. 14: p. 1189976. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Cicali EJ, et al. , Implementation of CYP2C19 genotyping to guide proton pump inhibitor use at an academic health center. Am J Health Syst Pharm, 2023. 80(15): p. 994–1003. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Tayeh MK, et al. , Clinical pharmacogenomic testing and reporting: A technical standard of the American College of Medical Genetics and Genomics (ACMG). Genet Med, 2022. 24(4): p. 759–768. [ DOI ] [ PubMed ] [ Google Scholar ] 20. UFHealth. MyRx. [cited 2024 10/17]; Available from: https://ufhealth.org/myrx/ . [ Google Scholar ] 21. Cicali EJ, et al. , How to Implement a Pharmacogenetics Service at your Institution. J Am Coll Clin Pharm, 2022. 5(11): p. 1161–1175. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Chen J-C, Tsai S-F, and Liu S-A, A retrospective study of differences in patients’ anxiety and satisfaction between paper-based and computer-based tools for “Shared Decision-Making”. Scientific Reports, 2023. 13(1): p. 5187. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Zemni I, et al. , The effectiveness of a health education intervention to reduce anxiety in quarantined COVID-19 patients: a randomized controlled trial. BMC Public Health, 2023. 23(1): p. 1188. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Medel D, et al. , Clinical decision making: validation of the nursing anxiety and self-confidence with clinical decision making scale (NASC-CDM ©) into Spanish and comparative cross-sectional study in nursing students. BMC Nursing, 2024. 23(1): p. 265. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Moyer AM, Langman LJ, and Black JL III, A Novel Approach to Improve Accuracy of CYP2D6 Enzyme Activity and Drug Response Predictions. Clinical Chemistry, 2022. 68(5): p. 616–618. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Gaedigk A, et al. , Ten Years’ Experience with the CYP2D6 Activity Score: A Perspective on Future Investigations to Improve Clinical Predictions for Precision Therapeutics. J Pers Med, 2018. 8(2). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Caudle KE, et al. , Clinical pharmacogenetics implementation consortium guidelines for CYP2C9 and HLA-B genotypes and phenytoin dosing. Clin Pharmacol Ther, 2014. 96(5): p. 542–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Bank PCD, et al. , Comparison of the Guidelines of the Clinical Pharmacogenetics Implementation Consortium and the Dutch Pharmacogenetics Working Group. Clin Pharmacol Ther, 2018. 103(4): p. 599–618. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Bousman CA, et al. , Clinical Pharmacogenetics Implementation Consortium (CPIC) Guideline for CYP2D6, CYP2C19, CYP2B6, SLC6A4, and HTR2A Genotypes and Serotonin Reuptake Inhibitor Antidepressants. Clin Pharmacol Ther, 2023. 114(1): p. 51–68. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Caraballo PJ, et al. , Integrating pharmacogenomics into the electronic health record by implementing genomic indicators. J Am Med Inform Assoc, 2020. 27(1): p. 154–158. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Lemke L, et al. , Clinician Response to Pharmacogenetic Clinical Decision Support Alerts. Clinical Pharmacology & Therapeutics, 2023. 114. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplemental Materials NIHMS2083573-supplement-Supplemental_Materials.docx (25.2KB, docx) ACTIONS View on publisher site PDF (1.5 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• The evolving landscape of pharmacogenomics: Current achievements and future directions.#28271
• Public Transcriptomic Data Mining for SCLC: From Candidate Ma rkers to Therapeutic Exploration.#483411
• CPGPrompt: translating clinical guidelines into large language model-executable decision support.#375475
• From Utopia to Science: Challenges of Personalised Genomics Information for Health Management and Health Enhancement#779047
• Metadata of metagenomic projects and disease cohorts used in this study#353640