Evolution of Personalized Care

What is the evolution of personalized care?

The evolution of personalized care is the shift from care based on population averages toward care fitted to each patient's risk, history and biology. It runs from guideline medicine, through risk stratification and data-driven care, to precision care that uses genetic results, such as pharmacogenomics, to choose treatments and doses.

None of these stages replaced the one before. Guidelines still set the default, risk models still decide who gets outreach, and genetic results refine specific decisions where the evidence is strong. What changed is how much individual data reaches the clinician at the moment of a decision.

How did care move from population averages to individual patients?

Care moved toward the individual in stages, each made possible by new data and new ways to deliver it. The software problem at each stage was the same: getting the right patient-specific fact in front of the clinician when it changes the decision.

• Population guidelines: Trials established which treatment works best on average, and guidelines turned that into standard care.
• Risk stratification: Scores built from age, history and lab values separated higher-risk patients so prevention could be targeted.
• Data-driven care: EHRs and claims data let organizations find care gaps and tailor programs to groups of similar patients.
• Patient preferences: Shared decision-making tools brought the patient's goals and circumstances into treatment choices.
• Precision care: Genetic and molecular results began to change drug choice, dose and screening for the individual patient.

Oncology moved first, with tumor testing guiding targeted therapies. Pharmacogenomics is how precision care reaches primary care, cardiology, psychiatry and pain management, because it affects common prescriptions rather than rare diseases.

Why does pharmacogenomics come first for most health systems?

Pharmacogenomics comes first because it affects drugs prescribed every day, a result stays valid for the patient's lifetime, and the action it triggers is concrete: choose a different drug or adjust the dose. That makes the clinical value easier to show than for most other genomic uses.

A single panel can cover several genes that influence how patients metabolize or respond to many medications. Once the result is in the record, it can inform prescriptions written years later by clinicians who never ordered the test, provided the EHR stores it as structured data and checks it at order time.

That last condition is where many programs fall down. A result filed as a scanned PDF helps the clinician who ordered it and almost nobody after them.

Pharmacogenomics also gives leaders a manageable first project. The number of actionable gene-drug pairs is limited, the pharmacy team already owns drug rules, and the outcome to track, a changed prescription, is easy to count. Lessons from this first program, especially about lab interfaces and consent, carry over to broader genomic work later.

What does the FDA pharmacogenetic table tell clinicians?

The FDA Table of Pharmacogenetic Associations lists gene-drug pairs where the agency believes the evidence supports a link to drug metabolism, response or safety. It sorts them into three sections by how strong a clinical action the evidence supports.

• Section 1: Associations where the data support therapeutic management recommendations, such as a different drug or dose.
• Section 2: Associations where the data indicate a potential impact on safety or response.
• Section 3: Associations shown to affect pharmacokinetic properties only.

The FDA notes that listing a pair does not mean it advocates testing before prescribing, unless the test is a companion diagnostic. The table is useful for product teams as a scoping tool: rules built on section 1 pairs, such as HLA-B*57:01 and abacavir, rest on firmer ground than rules built on section 3 pairs.

How do genomic results reach the EHR?

Genomic results reach the EHR as either a document, which a person reads, or structured data, which software can act on. Precision care at scale needs the structured route, and the HL7 FHIR Genomics Reporting implementation guide defines how to send it.

The Genomics Reporting guide, version 3.0.0 on FHIR R4, provides profiles for representing genotypes, haplotypes, variants and genomic findings, along with guidance for pharmacogenomic reporting. A lab can send the narrative report and the structured findings together, so the clinician sees a readable report and the EHR stores data rules can query.

• Lab interface: Receive results from each lab partner in FHIR or HL7 v2, then map them to a consistent internal model.
• Normalization: Represent genes, star alleles and phenotypes such as poor metabolizer with standard codes.
• Storage: Keep results on the patient record, separate from the encounter that ordered the test, so they persist.
• Display: Show phenotype and its meaning, not just raw genotype, in a place clinicians will look.

What does pharmacogenomic decision support look like?

Pharmacogenomic decision support checks a patient's stored phenotype when a drug is ordered and, if the gene-drug pair calls for action, shows a specific recommendation. Good alerts appear only for relevant pairs, explain the result in plain terms and offer the alternative as a one-click order.

• Post-test alerts: Fire when a drug is ordered for a patient whose result calls for a change.
• Pre-test alerts: Suggest testing when a high-impact drug is ordered and no result exists, where local policy supports that.
• Order-time services: Use CDS Hooks so an external service can return recommendations inside the EHR's ordering workflow.
• Feedback loop: Track overrides by gene-drug pair so pharmacists can refine rules that clinicians routinely reject.

Testing strategy shapes the design. Reactive testing orders a gene when a drug is about to be prescribed, so the result arrives days later. Preemptive testing runs a panel in advance, so results are already on file when the prescription is written, which makes order-time alerts far more useful.

What data and consent questions come with genomic care?

Genomic care raises questions other data does not: results are permanent, they say something about relatives, and they may be reused for research. Programs need clear answers on consent, access, storage and reuse before results start flowing into the record.

• Consent scope: Record whether the patient agreed to clinical use only or also to research reuse.
• Access: Decide which roles see full genomic data and which see only the phenotype needed for prescribing.
• Incidental findings: Agree with the lab which findings are reported and how they reach the care team.
• Retention: Keep raw data and interpreted results under separate policies, since interpretation changes as evidence grows.
• Reinterpretation: Plan how updated interpretations from the lab replace older results without losing the history.

Write these decisions down before the first result arrives. Changing consent or access rules after thousands of results are on file means reviewing every one of them, which few teams have time to do.

What does a precision care rollout look like?

A precision care rollout usually runs in phases: choose the gene-drug pairs, connect a lab, store results as data, switch on decision support for a pilot group, then measure and expand. Each phase has a clear owner and a clear test of whether it worked.

• 1. Governance: A pharmacy and therapeutics group selects the first gene-drug pairs and writes the recommendations clinicians will see.
• 2. Lab partner: Agree the panel, turnaround time, report format and whether structured FHIR results are available.
• 3. Intake: Build the interface, map results to internal codes and test with real reports covering normal and unusual genotypes.
• 4. Storage and display: Put phenotypes in a visible place in the chart, with a plain-language explanation of each.
• 5. Pilot: Turn on alerts for one clinic or service line and have pharmacists review every alert for the first weeks.
• 6. Measure: Track how many prescriptions changed, how often alerts were overridden and why, then tune rules before expanding.

Budget for the interpretation layer as well as the interface. Translating a genotype into a phenotype and a phenotype into a prescribing recommendation is clinical content that needs an owner, a review cycle and version control, just like any other order set.

Clinician education belongs in the plan too. Short guidance on what a phenotype means and when to call the pharmacist does more for adoption than a long training session before go-live.

Where precision care programs stall

• PDF results: Reports stored as documents that decision support cannot read.
• One lab only: Interfaces built for a single lab, so switching or adding labs means rebuilding.
• Alert noise: Rules that fire on weak associations until clinicians stop reading them.
• No pharmacist workflow: Recommendations with no one assigned to follow up when clinicians need advice.
• No outcome tracking: No measure of changed prescriptions or avoided adverse events to justify the program's cost.

Start small and measurable. Pick a handful of strong gene-drug pairs, build structured intake and order-time support for them, and track how often recommendations change prescriptions before adding more.

How Agnotic can help

The step from genetic test to changed prescription is mostly an integration problem. Agnotic Technologies builds precision medicine data integrations that receive results from lab partners, map them to the HL7 Genomics Reporting profiles, store them on the patient record and expose them to decision support.

Agnotic also builds the order-time services, including CDS Hooks endpoints for pharmacogenomic recommendations, with consent capture, role-based access and audit logging, plus the reporting that shows which recommendations changed care.

Sources

• FDA: Table of Pharmacogenetic Associations
• HL7 FHIR Genomics Reporting Implementation Guide