Clinical reasoning does not begin when the history is written.
It begins before the first question is asked.
This section focuses on problem-based clinical reasoning, the cognitive work clinicians do when they are presented with a defined problem and must decide what information to seek, which questions matter, and how to prioritize them.
If Reasoning Foundations addresses how information is organized once it is known, Problem-Based Clinical Reasoning addresses the step before that: how clinicians generate targeted questions in the first place.
Reasoning begins with a problem, not a template
Many learners are trained to “take a complete history” using fixed subsections and broad review-of-systems checklists. This approach protects completeness, but it rarely teaches selectivity, the ability to ask the right questions rather than all questions.
Expert clinicians work differently. They begin by naming the problem they are trying to solve and then allow that problem to guide their questioning. The history is not exhaustive, it is purposeful.
A simple contrast shows the difference.
Template-driven approach: ask broadly because that is what the template expects.
Problem-driven approach: ask selectively because the problem demands it.
A patient referred for iron deficiency anemia does not require an indiscriminate tour through every system. The early questions are chosen to answer a small number of high-value clinical questions: how sick is the patient, what is the biology, and why is it happening.
Disease scripts as reasoning tools
Clinicians often reason using disease scripts, structured mental models that link:
- enabling conditions and risk factors
- underlying mechanisms
- expected consequences (symptoms, signs, laboratory patterns)
Experts use these scripts intuitively. Novices are often taught them as lists to memorize.
In this section, disease scripts are treated differently. They are not endpoints. They are tools for generating questions.
Rather than recalling every feature of a disease, problem-based clinical reasoning shows how clinicians use scripts to decide:
- which historical features are most discriminating
- which questions can be asked once rather than repeatedly
- which answers meaningfully raise or lower the likelihood of competing diagnoses
This transforms disease scripts from static knowledge into active reasoning engines.
Diagnostic signals vs. physiologic threats
Not all abnormal numbers are trying to answer the same question.
Some values function as diagnostic signals. They may be asymptomatic, but they narrow the differential and demand explanation.
Other values function as physiologic threat signals. They correlate with symptoms, complications, or instability and require urgency assessment.
Confusing these roles is a common source of both overreaction and false reassurance.
Before generating questions, experienced clinicians implicitly ask:
Is this number pointing to an underlying process, a current physiologic risk, or both?
That decision determines which symptoms matter, which questions are prioritized, and how urgently the evaluation proceeds.
Avoid Adjective-Driven Reasoning
Terms like mild, borderline, or significant compress reasoning into a single word.
Problem-based reasoning works best when qualifiers are replaced with:
- numbers
- thresholds
- trajectories
- context
Ask not “Is this mild?”
Ask “Where does this value sit relative to the decision points that matter?”
Adjectives should conclude reasoning, not replace it.1
From differential to questions: the pre-encounter sequence
Problem-based clinical reasoning typically follows a sequence:
- Name the problem clearly.
Examples: iron deficiency anemia, thrombocytopenia, elevated ferritin, macrocytosis. - Organize the differential using a simple model.
Different problems call for different organizing frameworks:- mechanism-based: iron deficiency (intake, absorption, loss)
- context-based: thrombocytopenia (outpatient vs inpatient vs ICU context)
- pattern-based: macrocytosis (reticulocytosis vs non-reticulocytosis pattern)
- Translate each branch into targeted questions.
Each branch generates questions that distinguish competing explanations. - Recognize overlap and prioritize.
Many symptoms and risk factors appear in multiple scripts. Overlapping questions are asked once, then used to inform multiple branches. High-yield discriminators take precedence over exhaustive lists.
This process occurs before and during the encounter, and it directly shapes the history of presenting problems that follows.
Not all symptoms serve the same purpose
A common learner mistake is to treat symptoms as if they all do the same job. In reality, symptoms often function in different ways, even within a single condition.
Once a problem is named, symptoms tend to fall into three roles:
- severity and physiologic impact (how sick is the patient right now)
- biology of the disease state (what processes are active, even if etiology is unchanged)
- etiologic discrimination (which cause is most likely, and which causes become less likely)
Problem-based reasoning makes these roles visible. It prevents symptoms from collapsing into an undifferentiated checklist and helps the clinician gather information deliberately.
Parsimony over exhaustiveness
A central goal of problem-based clinical reasoning is parsimony.
Good histories are not defined by the number of questions asked, but by how effectively questions narrow the differential. Many symptoms and risk factors appear in multiple scripts. Asking them repeatedly adds noise rather than clarity.
Problem-based reasoning teaches clinicians to:
- ask overlapping questions once
- use answers to inform multiple branches of the differential
- resist low-yield lines of inquiry prematurely
This is not shortcutting. It is expert efficiency.
Parsimony Without Blindness: Occam and Hickam in Practice
Clinical reasoning often sits between two competing impulses.
Occam’s Razor encourages us to seek a unifying explanation whenever possible.
It protects against diagnostic sprawl and helps clinicians prioritize coherence over accumulation.
Hickam’s Dictum reminds us that patients are not obligated to read textbooks.
Multiple processes can coexist, particularly in older adults, patients with chronic disease, or those with complex exposures.
Problem-based clinical reasoning does not choose one over the other.
Instead, it uses parsimony as a starting hypothesis, not a conclusion.
Clinicians begin by asking:
Is there a single explanation that reasonably accounts for the findings?
If that explanation strains — requires special pleading, ignores discordant data, or fails to explain key features — the reasoning expands.
Hickam’s Dictum is not an excuse to abandon structure.
It is a reminder to reopen the differential when the unifying story breaks.
Expert reasoning is not choosing Occam or Hickam.
It is knowing when to stop unifying and when to allow multiplicity.
Hematology provides frequent examples of this tension.
A patient with pancytopenia may have:
- a single unifying process (for example, marrow failure or infiltration), or
- multiple concurrent contributors (nutritional deficiency, medication effect, and hypersplenism).
Problem-based clinical reasoning allows both possibilities to remain live — initially favoring coherence, but revising when the data demand it.
How this section fits with the others
Problem-based clinical reasoning is intentionally upstream of documentation.
It informs what questions are asked.
It feeds directly into the history of presenting problems.
It makes the assessment easier to construct.
It simplifies oral presentation, because the logic is already ordered.
In this way, problem-based reasoning bridges foundational thinking and clinical expression.
Disease scripts, redefined
Disease scripts are often taught as lists to remember.
Here, they are used as tools to decide what to ask, and why.
Why this helps you
Problem-based clinical reasoning helps clinicians:
- ask fewer, better questions
- avoid unfocused or repetitive histories
- prioritize discriminating features
- enter the encounter with a clear plan
The payoff is not less curiosity. It is more deliberate thinking.