When every option carries harm
This post walks through a real inpatient hematology consult, step by step, to show how expert consultants organize their thinking over time when faced with competing risks.
The phases are named explicitly to make visible what is usually implicit in practice.
The goal is not to provide an algorithm or a set of management rules. The goal is to model judgment as it unfolds, under imperfect information, in a situation where there is no neutral choice.
Opening scenario
You are asked to consult on a hospitalized patient.
A 72-year-old woman was admitted with acute shortness of breath and was found to have a pulmonary embolism. She was started on therapeutic anticoagulation.
Two days later, she develops melena and a falling hemoglobin.
You are asked to help manage anticoagulation.
No additional details are provided.
Companion resources (in development)
Cause-based frameworks and quick-reference tools for common inpatient hematology problems are being developed as part of TBP’s consult reasoning series. They are meant to be used after initial orientation and framing, not in place of them.
How to use this post when you get paged
This is not a diagnostic guide, and it is not meant to be read linearly at the bedside.
Instead, use it as a cognitive checklist at three moments:
- When the page comes in, use Phase 1 to orient to danger, urgency, and asymmetric risk before naming diagnoses.
- When you make your first recommendation, use Phase 2 and 3 to decide what stance you’re taking and how to communicate it clearly under uncertainty.
- When new information arrives, use Phase 4 to recalibrate without rewriting history.
The goal is not to tell you what to think, but to help you recognize what kind of thinking the situation demands.
Different consults demand different kinds of thinking
Some hinge on thresholds, where the central question is whether inaction has become more dangerous than action. Others require balancing harms, where no option is safe and the work lies in choosing which risk to accept. Still others require proportionality, where the diagnosis is known and the challenge is matching the mechanism of an intervention to how the disease is behaving over time.
The disease examples that follow are not exhaustive. They are illustrations meant to help you recognize these patterns when you are in the middle of one.
Phase 1: Initial Orientation
(Often begins at the time of the page)
Phase 1 involves rapid, provisional thinking under time pressure, aimed at defining danger and scope rather than solving the problem.
In this consult, the first task is not to decide what to do with anticoagulation. It is to understand what kind of danger you are being asked to manage.
At this stage, the consultant is orienting to competing risks, not adjudicating them.
Key orienting questions
(not ordered by importance)
Is the bleeding active, ongoing, or accelerating?
A stable hemoglobin with old blood in the stool lives in a different universe than brisk bleeding with hemodynamic consequences.
How anticoagulated is the patient in effect right now?
PT and PTT values, when relevant, help answer this question — not to assign blame or mechanism, but to understand how much margin you actually have.
Where is the bleeding, and how controllable is it likely to be?
Diffuse mucosal bleeding, procedural bleeding, and focal GI bleeding carry very different implications for reversibility.
How recent and how dangerous is the clot?
A subsegmental PE discovered incidentally is not the same problem as a recent PE with right-sided strain. Time since diagnosis matters as much as clot location.
Why is the patient anticoagulated, and how narrow is the margin for error?
Some indications tolerate interruption poorly. Others tolerate it better than we sometimes admit.
Where is the patient right now?
ICU versus floor matters. Monitoring intensity, access to urgent procedures, and tolerance for risk are different in different settings.
What would be hardest to undo tomorrow?
Holding anticoagulation and restarting it later is not symmetric with continuing anticoagulation and managing bleeding. The errors are not equally reversible.
At this stage, you are not deciding whether to stop anticoagulation.
You are deciding which risk must dominate the next few hours, and which risk you are willing to tolerate for now.
By the end of Phase 1, the consultant should be able to say:
- I understand how real the bleeding is
- I understand how unforgiving the clot risk might be
- I understand whether any decision carries irreversible consequences
- I know which risk would make me lose sleep if I underestimated it
Phase 1 does not weigh anticoagulant classes, reversal agents, or endoscopy timing. It determines tempo and risk tolerance.
Phase 2: Diagnostic Framing
(Choosing a direction of reasoning)
Phase 2 begins once the patient has been seen and the initial urgency has been contained, even though uncertainty remains.
In this consult, you are not asking, “Why is this patient bleeding?”
You are asking, “What kind of decision problem am I being asked to manage?”
This phase is about constraints, not causes.
What informs framing in competing-risk consults
Unlike diagnostic consults, framing here is driven by three interlocking considerations.
Reversibility
How easily can the harm be undone if you get it wrong?
Bleeding that can be controlled endoscopically or surgically carries different stakes than bleeding that is diffuse or inaccessible.
Substitutability
Is there a way to modify rather than abandon therapy?
Dose reduction, temporary interruption, switching to shorter-acting agents, or mechanical protection can sometimes reduce harm without pretending the risk has disappeared.
Time
Is there a way to buy hours or a day without committing to a brittle decision?
Many anticoagulation decisions are safest when framed as intentionally provisional.
This is where experienced consultants think in gradients.
How anticoagulated is the patient in effect right now?
How much protection is the anticoagulant actually providing at this moment, and how much is it contributing to ongoing harm?
Gradient, not binary
Many teams think “continue” versus “stop.” Consultants often think in partial moves: how to reduce harm while preserving some protection, and how to make that choice intentionally provisional.
The role of high-impact discriminators
Some information carries disproportionate weight once available.
Endoscopy findings that identify and control a bleeding source, serial hemoglobin measurements that show stabilization or continued loss, or imaging that clarifies clot burden can rapidly shift the balance.
Their absence early is not neglect. It reflects the reality that many anticoagulation decisions must be made before definitive information arrives.
The output of Phase 2
The output is not a plan. It is a stance.
For example:
“The bleeding is real, but not yet catastrophic. The clot risk still dominates. We will preserve anticoagulation while pursuing urgent source evaluation, with a low threshold to recalibrate.”
Or:
“The bleeding has crossed a threshold where clot prevention must temporarily take second place. We will hold anticoagulation, prioritize hemostasis, and reassess within the next 12–24 hours.”
The word temporarily is not a hedge. It is a commitment to revisit the decision.
Phase 3: Communicating the Consult
(Expressing judgment clearly)
Phase 3 begins once you have a working stance.
In this consult, communication is not about correctness. It is about alignment.
Two audiences, two purposes
Internal communication (within the consult team)
Before speaking externally, the consultant aligns internally.
This is where uncertainty is surfaced rather than hidden. Are we underreacting to bleeding? Overestimating clot danger? Comfortable with the amount of risk being accepted?
External communication (to the primary team)
The goal is to guide action under uncertainty.
Effective communication does not promise safety. It explains tradeoffs.
This often includes:
• which risk is being prioritized right now
• which risk is being consciously accepted
• what the plan is designed to accomplish
• what would make us stop and recalibrate
Recommendations without rationale are brittle in situations of true equipoise.
Timing matters. This communication often happens before endoscopy, before definitive imaging, and before the bleeding has fully declared itself.
That is appropriate.
Phase 4: Recalibration Over Time
(Revising judgment as new information arrives)
Phase 4 begins once the immediate crisis has been contained enough for time to add information.
In bleeding-and-clotting consults, recalibration is driven by movement of risk, not by diagnostic clarity. One risk recedes. The other advances. The consultant’s task is to notice when that balance has shifted enough that the original stance no longer fits.
This is not a failure of earlier judgment.
It is the point of holding a stance rather than a fixed plan.
Recalibration is about timing, not better reasoning
Earlier phases rely on what can reasonably be inferred under uncertainty. Phase 4 acknowledges that in competing-risk problems, biology and events, not insight, do the decisive work.
Bleeding declares itself through persistence, control, or escalation.
Thrombotic risk declares itself through time since the event, stability, and recovery.
What matters is not whether the initial decision was provisional, it almost always was, but whether it is revisited deliberately as the balance of risk evolves.
Good recalibration recognizes that a decision that was protective yesterday can become hazardous today, and that restraint can become riskier than action once circumstances change.
A recalibration narrative
When I first saw this patient, she was two days out from a diagnosed pulmonary embolism and had developed melena with a hemoglobin that had fallen from 11 to 8. She was tachycardic but hemodynamically stable. The PE had been submassive, with evidence of right-sided strain on echocardiography.
At that point, the bleeding was real but not catastrophic. The clot was recent and potentially unforgiving. The dominant question was not what the final plan would be, but which risk needed to anchor the next several hours.
My initial stance was to preserve anticoagulation while pursuing urgent evaluation of the bleeding source.
Over the next 18 hours, endoscopy identified and treated a bleeding duodenal ulcer. The melena stopped. Hemoglobin stabilized after transfusion.
At that point, the balance shifted.
The risk that had dominated the first day receded. The clot risk, which had been temporarily subordinated, moved back to the foreground. Restarting full-intensity anticoagulation was not a reflex. It required a new orientation to timing, stability, and tolerance for rebleeding.
Internally, we discussed how much confidence we had in hemostasis and what degree of clot protection was acceptable at that moment. Externally, we explained why restarting anticoagulation now made sense, even though bleeding control was recent and not absolute.
In other cases, recalibration moves in the opposite direction. Bleeding that initially seems manageable proves persistent or dangerous. What was once an acceptable risk becomes unacceptable, and anticoagulation remains interrupted longer than anyone anticipated.
What matters is not the direction of recalibration, but that it is explicit.
Communication revisited
As the balance shifts, communication must shift with it.
Internal discussion recalibrates concern and risk tolerance before recommendations change. External communication then updates the primary team, not just with a new plan, but with an explanation of why the plan has changed.
Urgency may escalate. It may de-escalate. Either movement should feel intentional rather than reactive.
Changing one’s mind in these cases is not a failure of judgment.
It is the work.
What Phase 4 demands of the consultant
Phase 4 requires the ability to change course without defensiveness.
New information does not invalidate earlier decisions. It reframes them. Good consult practice includes explaining why a prior stance was reasonable at the time, and why a different stance is warranted now.
Closing reflection
Bleeding and clotting on anticoagulation is not a problem with a stable solution. It is a moving balance.
Urgency is defined before diagnosis, or in this case, before knowing which risk will dominate.
Direction is chosen before certainty, which risk to prioritize right now.
Judgment is communicated before completion, often before endoscopy or stabilization.
And conclusions are revised as reality evolves.
What distinguishes experienced consultants is not that they eliminate risk, but that they decide, consciously and transparently, which risk they are accepting right now, and why.
That decision must be revisited as reality evolves. Good consulting in this setting is about staying aligned with the problem over time.