The Chart Records the Drug. Not What Kept It in Place.
Anticoagulation maintenance decisions in AF — what ten cardiologist conversations reveal about how and why patients stay on therapy
Most chart audits ask physicians to select from a list. Patient Scribe asks them to explain. The result is a conversation — one where the clinical logic behind a prescription, a dose continuation, or a treatment decision after an adverse event becomes visible in the physician’s own terms. The structure comes after. The insight comes from letting the conversation run.
Therapies in scope: Apixaban (Eliquis) · Rivaroxaban (Xarelto) · Warfarin (as prior therapy / comparator reference)
What This Study Was Designed to Answer
Ten cardiologists managing Atrial Fibrillation patients were interviewed on their anticoagulation maintenance decisions. The study focused on four key business questions:
In two of the ten cases, the patient experienced a bleeding event while on anticoagulation. In both cases, the physician maintained the current therapy. The clinical logic in both was the same: the bleed was attributed to a co-precipitant — NSAIDs in one case, Aspirin in another — not to the anticoagulant itself. Both co-precipitants were discontinued. Anticoagulation continued. A traditional PCA would record a bleed and a continuation. It would not record the reasoning that kept the drug in place — or that the physician identified a separate causal agent and resolved it without a switch.
COMMERCIAL IMPLICATION: The switching threshold after a bleeding event is higher than bleed frequency alone suggests. Physicians are not passively continuing therapy — they are actively resolving the bleed by removing the co-precipitant. For any competitor, the entry point is not “fewer bleeds.” It is what happens when no co-precipitant can be identified — that is when the drug itself comes under scrutiny.
Nine of ten patients in this dataset are on Apixaban. But the reasons for staying on it are not uniform. One physician is following a guideline algorithm anchored to CHA₂DS₂-VASc score. Another cited the ARISTOTLE trial by name — specifically lesser bleeding and comparable stroke reduction versus warfarin. A third described Apixaban as the best drug on the market on the basis of accumulated personal clinical experience and outcome conviction. A fourth is maintaining therapy because the patient reports no side effects and is stable. These four physicians would respond to completely different commercial messages. In prescribing data, they are indistinguishable.
COMMERCIAL IMPLICATION: Segmenting by prescribing behavior alone misses the commercial signal. A physician maintained by guideline inertia is a different audience from one held by trial data conviction, and both are different from one operating on accumulated personal experience. Messaging that speaks to one rationale will not land with the others.
In the cases where physicians elaborated on their rationale for continuing anticoagulation after a bleed, the stroke risk override was explicit and conscious — not passive. Physicians named the risk calculation directly: the risk of AF-related stroke was judged greater than the resolved bleeding risk. This is a deliberate clinical decision. A checkbox records a drug continuation. It cannot distinguish between a physician who made an active risk-benefit calculation and one who simply did not reconsider the therapy. The commercial story underneath is entirely different in each case.
COMMERCIAL IMPLICATION: Understanding whether a physician’s continuation decision is active or passive changes the commercial strategy entirely. Active decision-makers are engaged, evidence-responsive, and open to data that refines their risk calculation. Passive maintainers need a different kind of prompt — typically a peer case or a guideline update that creates a reason to reconsider.
Suggested Read: Sickle Cell Disease Patient Charts
P5 — 55M | COMMERCIAL | AF DX JAN 2025 | GI BLEED (NSAIDS) → RIVAROXABAN (XARELTO) MAINTAINED
55-year-old male. Commercial insurance. Moderate activity. AF diagnosed January 2025. History of GI bleed while on anticoagulation.
The patient had a GI bleed. The physician identified NSAIDs as the precipitating cause, not the anticoagulant. NSAIDs were stopped. A proton pump inhibitor was added for GI protection. Xarelto was maintained. The physician’s explicit framing: AF patients are high risk for stroke, and once the co-precipitant was removed, the anticoagulation risk-benefit calculation did not change. The chart records a drug continuation after a bleeding event. It does not record the diagnostic reasoning that identified the cause elsewhere — or that the physician actively resolved the bleed without reconsidering the anticoagulant.
P7 — FEMALE | MEDICARE | AF DX 2018 | SEDENTARY | CHA₂DS₂-VASC → APIXABAN MAINTAINED
Female. Medicare. Sedentary lifestyle. AF diagnosed 2018. Elevated CHA₂DS₂-VASc score.
Guideline logic drove the maintenance decision. CHA₂DS₂-VASc above 2 in a female patient means continued full-dose anticoagulation in the absence of bleeding — that is the algorithm, and the physician followed it. There was no adverse event, no competing clinical factor, no patient preference that entered the decision. The physician is not evaluating Apixaban; the physician is executing a guideline. A competitor entering this conversation has to change the algorithm, not the physician’s clinical opinion.
“It’s guideline recommended with a CHA₂DS₂-VASc score above 2 and the female to continue on full-dose anticoagulation in the absence of bleeding.”
— Cardiologist, AF patient audit (P7) — on maintenance rationale
P8 — 70F | PPO | AF DX 6 MONTHS AGO | ARISTOTLE TRIAL → APIXABAN MAINTAINED
70-year-old female. PPO insurance. AF diagnosed 6 months prior. No comorbidities specified.
The ARISTOTLE trial was named unprompted as the evidence anchor for the maintenance decision. The physician cited lesser bleeding and comparable stroke reduction versus warfarin as the operative findings. This is a physician reasoning from a specific trial, not a general safety impression. The prescribing record shows Apixaban continued at current dose. It does not show that a named trial, a specific endpoint comparison, and an active evidence review were the basis for that decision.
P9 — 62F | BCBS | AF DX JAN 2022 | EPISTAXIS (ASPIRIN) → APIXABAN MAINTAINED
62-year-old female. Blue Cross Blue Shield. Sedentary. AF diagnosed January 2022. Comorbidities: Hypertension and Diabetes. History of epistaxis while on Aspirin + Apixaban.
The patient experienced a nosebleed. The physician attributed it to Aspirin — “I thought it was just increasing her risk of bleeding” — discontinued the Aspirin, and maintained Apixaban without dose modification. The rationale for keeping Apixaban was stated directly: it has the best safety data, there was zero advantage to switching, and if she were to bleed again, the next step would be left atrial appendage closure, not a drug switch. The prescribing record captures a continuation. The clinical conviction behind it — and the defined escalation path if the current approach fails — is invisible.
“Eliquis has the best safety data, so I saw zero advantage to change to a different anticoagulant in her case.”
— Cardiologist, AF patient audit (P9) — on post-bleed maintenance rationale
Suggested Read: Decoding Prescribing Decisions in Ulcerative Colitis
The physicians who are maintaining Apixaban on guideline autopilot, on personal conviction, on trial evidence, and on patient stability are four different commercial audiences. The physicians who held the line after a bleeding event are making explicit risk calculations that would tell a brand team exactly where the therapy is vulnerable — and exactly what would have to change for a switch to become a real possibility. None of that is in the Rx field.
The question a brand team in AF most needs answered is not who is prescribing Eliquis — it is why they are, and under what conditions that could change. A physician maintaining on guideline autopilot does not have a near-term switch trigger. A physician who has already defined what comes next if the patient bleeds again is one adverse event away from a different prescription. Those are not the same commercial situation. Aggregate continuation data cannot distinguish them.
Patient Scribe recovers the reasoning from the prescription — the logic behind continuation decisions, the conditions physicians have already set for escalation or switch, the clinical rationale that makes the Rx field legible rather than just countable. In chronic disease management, where maintenance looks like inertia but often is not, the conversation transforms a prescribing count into a commercial strategy.
AF prescribing involves risk tolerance, patient preference, and adherence anxiety that structured surveys flatten into categories.
Beyond What Was Prescribed shows how conversational AI-powered chart audits recover that clinical reasoning — across UC, HS, and Multiple Myeloma — and what the intelligence gap costs brand teams in real commercial decisions.
ZoomRx is offering a zero-cost pilot wave — same rigor as a full Patient chart audits study, no commitment required.