When certainty arrives late, but risk arrives early
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 a rare but catastrophic diagnosis.
The phases are named explicitly to make visible what is usually implicit in practice.
The goal is not to provide a diagnostic checklist or a review of HLH criteria. The goal is to model judgment as it unfolds, under uncertainty, in a setting where waiting for certainty can itself cause harm.
Opening scenario
You are asked to consult on a hospitalized patient.
A 54-year-old man is admitted with fever, cytopenias, and abnormal liver enzymes.
Despite broad-spectrum antibiotics, he continues to worsen.
The primary team asks whether this could be HLH.
No additional details are provided.
A brief note about the sparseness: HLH consults often arrive with a specific worry already raised, and they rarely come as a blank-slate “evaluate cytopenias” page. A small amount of context is part of what makes the scenario feel real.
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 committing to a diagnosis.
In an HLH consult, the first task is not to decide whether the patient has HLH.
It is to decide whether HLH belongs in the clinical universe at all.
This is a permission-granting phase.
Key orienting questions
(not ordered by importance)
How sick is the patient, and how fast are they deteriorating?
HLH is not subtle for long. Progressive instability and momentum matter more than any single lab value.
Is this inflammation behaving like infection, malignancy, or neither?
Persistent fever, worsening cytopenias, rising inflammatory markers, and organ dysfunction despite appropriate therapy should feel different than routine sepsis.
How high is the ferritin, and is it rising?
Ferritin is nonspecific, but extreme elevation or rapid upward trajectory should change how seriously HLH is entertained. A ferritin of 800 does not live in the same universe as a ferritin of 20,000, especially when it is climbing.
Are multiple organ systems drifting in the same direction?
HLH declares itself through convergence. Cytopenias, liver injury, coagulopathy, and neurologic changes accumulating together are more concerning than any one alone.
Is there a context where HLH is even plausible?
Underlying malignancy, immunosuppression, autoimmune disease, or viral triggers do not diagnose HLH. They change whether it deserves active consideration.
Where is the patient right now?
ICU versus floor matters. Monitoring intensity, access to urgent procedures, and tolerance for instability are different in different settings.
At this stage, you are not diagnosing HLH.
You are deciding whether this is a patient in whom missing HLH would be unforgivable, even if the diagnosis ultimately proves wrong.
By the end of Phase 1, the consultant should be able to say:
• HLH is either plausible or implausible in this patient
• I understand how fast this patient is evolving
• I know whether delay itself carries risk
• I know whether this case deserves heightened vigilance
Phase 1 does not apply criteria, order specialized tests, or start therapy. It determines tempo and risk tolerance, and it defines whether HLH deserves space in your thinking.
Phase 2: Diagnostic Framing
(Choosing a direction of reasoning)
Phase 2 begins once HLH has been admitted into the differential universe.
In this consult, you are not asking, “Does this patient meet criteria for HLH?”
You are asking, “What kind of decision problem is this, and has concern crossed a threshold where waiting for certainty becomes dangerous?”
Thresholds, not probabilities
In most consults, Phase 2 is a probability exercise. You rank competing explanations, pursue the most likely, and let discriminators tighten the picture.
HLH does not behave that way.
The base rate is low. The early findings are nonspecific. If you rely on probability alone, the answer will almost always be “no,” right up until the patient declares themselves in multi-organ failure.
So the Phase 2 question shifts.
Here, the consultant’s task is not to ask whether HLH is likely.
It is to decide whether the asymmetry of harm has shifted far enough that inaction has become harder to defend than action.
This is what makes HLH a threshold problem.
Pre-test probabilities, not premature dismissal — adapted
Pre-test probability still exists, but it cannot be the sole governor of action in HLH.
In this setting, premature closure does not mean choosing the wrong diagnosis. It means dismissing concern because no single finding is definitive, or because formal criteria have not yet been met.
The task is not to prove HLH.
The task is to decide whether suspicion itself has become actionable.
This is where experienced consultants get uncomfortable on purpose.
What informs framing in HLH consults
Trajectory
Is the patient stabilizing, plateauing, or worsening despite appropriate therapy? HLH is a disease of momentum. Continued deterioration after 24–48 hours of thoughtful care should carry disproportionate weight.
Accumulation of weak signals
Fever, cytopenias, hyperferritinemia, transaminitis, coagulopathy, neurologic changes. None are diagnostic alone. The concern emerges when multiple weak signals accumulate and intensify in the same direction over time. HLH declares itself through convergence.
Lack of an alternative unifying explanation
Sepsis that behaves oddly. Malignancy that does not fit the pace. Autoimmune disease that fails to explain the severity. When standard explanations strain credibility, HLH moves from background to foreground.
Irreversibility of delay
The longer HLH evolves untreated, the harder it is to rescue. Organ dysfunction accumulates. Inflammatory cascades become self-sustaining. Waiting for certainty is not neutral. This temporal asymmetry weighs heavily in framing.
Supporting tools, not gatekeepers: the role of the H-score
Composite tools such as the H-score can be useful after HLH has been admitted into the differential.
They can help formalize concern, support documentation, and communicate risk across teams.
They should not be used to decide whether HLH deserves consideration in the first place.
Early in the course, the H-score often underestimates risk because key elements are unavailable or still evolving. Later, it can help reinforce a stance that has already been reached through trajectory and pattern recognition.
A sentence worth remembering:
A low H-score early does not mean HLH is unlikely. It often means HLH has not yet declared itself.
Used this way, the H-score supports judgment rather than replacing it.
The role of high-impact discriminators
Some data carry disproportionate weight once available: a rapidly rising ferritin, worsening cytopenias across lineages, hypofibrinogenemia, or evolving organ failure.
Bone marrow findings and specialized assays may help later, but their absence early is expected. Unlike consults where key discriminators arrive within hours or a day, HLH confirmation often lags the decision point.
HLH decisions are frequently made before definitive confirmation is possible.
The output of Phase 2
The output of Phase 2 is not a diagnosis.
It is a stance.
For example:
“This does not yet prove HLH, but the pattern and trajectory cross a threshold where we need to act as if it could be. We should escalate evaluation, align the teams, and prepare for empiric therapy if deterioration continues.”
Or:
“There are inflammatory features here, but the trajectory and context do not support HLH as a leading concern right now. We should continue to treat the most plausible explanation and monitor closely, with explicit triggers for reassessment.”
One stance mobilizes resources.
The other preserves restraint.
What Phase 2 does—and does not—do
Phase 2 does:
• determine whether suspicion has crossed a treatment-relevant threshold
• clarify what findings would escalate or de-escalate concern
• justify urgency without declaring certainty
Phase 2 does not:
• guarantee the diagnosis
• eliminate uncertainty
• replace judgment with criteria
That stance now needs to be communicated.
Phase 3: Communicating the Consult
(Expressing judgment clearly)
Phase 3 begins once you have decided that HLH deserves either foreground or background attention.
In this consult, communication is the intervention, because it is what prevents the system from defaulting to false reassurance or endless deferral.
Two audiences, two purposes
Internal communication (within the consult team)
Before speaking externally, the consult team aligns on the level of concern and the threshold for action.
Is this “possible but unlikely,” meaning we can watch and revisit, or is this “unlikely but too dangerous to ignore,” meaning we should escalate now? Those sound similar. They behave very differently.
This internal calibration protects against both overreaction and dangerous delay.
External communication (to the primary team)
The goal is to convey seriousness without overclaiming.
Effective HLH communication names uncertainty honestly while explaining why that uncertainty does not justify inaction.
This often includes:
- stating that HLH is not proven
- explaining why waiting for proof may be unsafe
- clarifying what actions are being taken now, and what they are meant to accomplish
- naming what would make you change course
HLH consults fail when concern is either dramatized or minimized. The middle ground is hard, and that is where expertise lives.
Phase 4: Recalibration Over Time
(Revising judgment as new information arrives)
Phase 4 begins as the clinical picture evolves.
In HLH, recalibration is driven less by the arrival of a single decisive test than by whether the trajectory continues to converge—or begins to loosen.
Ferritin may rise rapidly or plateau. Cytopenias may worsen across lineages or begin to stabilize. Organ dysfunction may accumulate or soften. An alternative explanation may finally declare itself—or fail to.
These changes do not simply add information.
They reshape meaning.
Recalibration is about timing, not better reasoning
Earlier phases rely on what can reasonably be inferred in real time. Phase 4 acknowledges that HLH tempts premature dismissal because confirmation often arrives late, sometimes after the decision point has already passed.
What matters is not whether the initial stance was provisional—it almost always was—but whether it is revisited honestly as time adds clarity.
Only then does recalibration move from concept to practice.
A recalibration narrative
When I first saw this patient, he had persistent fevers, bicytopenia, rising liver enzymes, and a ferritin in the low thousands. He was not in shock, but he was clearly worsening despite appropriate antimicrobial therapy.
At that point, HLH was not established. The question was not whether the diagnosis was proven, but whether the pattern and momentum had crossed a threshold where waiting carried more risk than acting.
My initial stance was one of guarded escalation: to broaden evaluation urgently, involve the relevant teams early, and prepare for empiric therapy if the trajectory continued, while still treating the most plausible alternative explanations.
Over the next 24 hours, the picture intensified. Cytopenias worsened, ferritin rose rapidly, and coagulation parameters deteriorated. No alternative diagnosis emerged that could plausibly account for the pace, convergence, and direction of the findings.
At that point, the balance shifted.
Internally, we confronted the central question explicitly: had delay itself become the unacceptable risk? We agreed that it had. Externally, we explained that HLH was not yet definitively confirmed, that our decision was driven by trajectory and convergence rather than a single test, and that diagnostic evaluation would continue in parallel, with a clear plan to reassess if the picture evolved differently.
What had initially been concern became conviction that continued delay posed greater danger than empiric treatment.
In other cases, recalibration moves in the opposite direction. Early concern softens as the patient stabilizes or an alternative diagnosis declares itself. The HLH workup is halted. Treatment is deferred. That is not retreat. It is fidelity to the evolving biology.
What matters is not the direction of recalibration, but that it is explicit.
Communication revisited
As the picture evolves, internal discussion recalibrates concern, and external communication adjusts urgency and recommendations.
Escalation and de-escalation should feel intentional rather than reactive.
Changing one’s mind in HLH consults is not a failure of judgment.
It is the work.
What Phase 4 demands of the consultant
Phase 4 requires comfort with revising concern without rewriting history.
Starting treatment does not mean certainty was achieved. Stopping evaluation does not mean concern was misguided. Good consult practice explains why earlier decisions were appropriate at the time and why a different stance is warranted now.
Closing reflection
HLH is not difficult because it is rare. It is difficult because it punishes hesitation and resists proof.
Urgency is defined before diagnosis.
Direction is chosen before certainty.
Judgment is communicated before confirmation.
And conclusions are revised as reality evolves.
What distinguishes experienced consultants is not that they always identify HLH correctly, but that they know when suspicion itself has crossed a threshold that demands action.
Good consulting in HLH is not about certainty.
It is about recognizing when waiting becomes the most dangerous choice.