On this episode of “Tell Me Where IT Hurts,” Anna Dover, PharmD, Senior Director of Editorial Content at FDB (First Databank), joins host Dr. Jay Anders, Chief Medical Officer of Medicomp Systems.
A PharmD with more than 20 years of experience spanning clinical pharmacy, decision support, and healthcare informatics, Dover began in direct patient care before moving into informatics during an early EHR implementation, where she first worked with FDB data. She joined FDB more than nine years ago on the product management team before moving to the editorial side, where she now oversees the team that identifies, structures, and curates the drug information powering FDB’s decision support.
The conversation opens on a problem every clinician recognizes: alert fatigue. Dover frames clinical decision support as a balance between guiding, nudging, and stopping a clinician, and argues that the best decision support is often passive rather than interruptive. Getting it right means understanding the workflow and the specialty, since an oncology clinic and a cardiology clinic want very different things. She shares an early-career lesson: an alert she designed was clinically sound but ignored, simply because it fired at the wrong point in the workflow. The takeaway, echoed by Anders, is that hard stops breed workarounds, and clinicians want the information without being blocked from moving forward.
The discussion turns to pharmacogenomics, the practice of using a patient’s genetic information to determine whether a given drug may be less effective or cause harm. Rather than chasing the most granular genetic detail, FDB focuses on the phenotype, the physiologic expression of a genetic variation, and its intersection with a specific drug. Dover notes that the science is still evolving but has reached a point where a number of genetic markers carry compelling evidence and should be reviewed before a patient starts therapy, with the potential to safely speed time to efficacy and improve quality of life.
“What makes pharmacogenomics exciting is that it is so highly tailored to that particular patient that it shouldn’t feel like noise. It should feel valuable in the clinical workflow.”
Anna Dover, PharmD
Senior Director of Editorial Content at FDB
But there remains a data problem. Genetic results are often stashed in a PDF inside the EHR, unstructured and unavailable to decision support. There is no single comprehensive vocabulary for transmitting the information across the standards in use. Interpretations can also go stale as the science evolves or as additional genes are tested. And the testing panel itself varies by lab.
To help clinicians keep pace, FDB embeds hyperlinks to source evidence such as CPIC directly in its content. Dover also raises an ethical issue that separates genetic data from ordinary lab value: it is durable and revealing. Even the fact that a test was ordered can disclose sensitive context, which can have implications for things like life insurance and employment. Both she and Anders agree that patient data belongs to the patient. She also flags a gap on the dispensing side, where community pharmacists often lack access to the pharmacogenomic information available within a health system.
On data quality and AI, Anders cites concern about ambient documentation tools, and notes that a large share of AI-generated summaries contain errors. He asks how bad data gets into the record. Dover counters that inaccurate records are not new, citing abstraction errors from the paper-to-EHR conversion era, and sees opportunity in newer technology to surface anomalies for clinician review. She closes by detailing FDB’s direction: making decision support more patient-specific by incorporating genetic tests, lab values, and diagnoses, and exploring AI, including a recently launched MCP that grounds large language models in evidence-based, clinically curated information rather than scraped web content.
Asked about one thing she would change in healthcare, Dover points to reducing the cognitive burden technology places on clinicians, so the right choice for the patient is also the easiest one to make.
Show Notes
Among the topics covered:
- Balancing clinical decision support: guiding and nudging versus hard stops
- Alert fatigue and why the best decision support is often passive
- Designing alerts for the right point in the clinical workflow
- What pharmacogenomics is and how it tailors therapy to the patient
- Focusing on phenotype rather than star alleles in genetic decision support
- Genetic results locked in PDFs and the lack of a standard transmission vocabulary
- Embedding source evidence (CPIC and others) directly in clinical content
- Privacy and ethics: genetic data as durable, revealing patient information
- The community pharmacy gap in pharmacogenomic information
- Data accuracy, ambient documentation, and using AI to surface anomalies
- FDB’s direction: patient-specific decision support and a new MCP for grounding LLMs
- And more…
Show Links
- Anna Dover, PharmD on LinkedIn
- FDB (First Databank)
- Dr. Jay Anders on LinkedIn
- Medicomp Systems on LinkedIn
- Tell Me Where IT Hurts on Healthcare NOW Radio