Think About Investigations

Reminder

Do not order tests by ticking boxes.
Think first.

The sequence History and examination come first, and they usually point you to the diagnosis. Investigations confirm it, rule out a dangerous alternative, or measure how severe it is. They support clinical reasoning; they do not replace it.

The standard Good clinicians do not order many tests. They order the right test for a specific question. A "routine" panel sent without one produces borderline results that trigger more tests, delay decisions and worry patients.

The obligation Before you order a test, say what you expect to find and how each possible result would change your plan. If no result would change what you do, you should not order the test.

Core Principle

An investigation must answer a question.
State the question before you order the test. If you cannot, rethink.

Sequence

Investigations follow reasoning, not the other way around

Step 1

History

Step 2

Examination

Step 3

Hypothesis

Step 4

Investigation

Step 5

Reassessment

Remember

Investigation sits in the middle of the clinical process, not at the beginning.
History and examination generate the hypothesis. The investigation tests it. Reassessment interprets the result.

Purposes

Why do we investigate?

Diagnostic Tier Used to establish or rule out a diagnosis

Confirm a Diagnosis

You already suspect the disease. The test verifies it.

Examples Suspected ALL → FBC + peripheral smear → Bone marrow examination.
Suspected nephrotic syndrome → urine protein.

Exclude a Diagnosis

You are actively testing a competing hypothesis.

Example Child with wheeze (?Asthma) + unilateral reduced air entry → chest X-ray to exclude foreign body aspiration.

Management Tier Used to guide decisions after the diagnosis is established

Assess Severity / Detect Complications

You are assessing severity or disease consequences.

Examples Nephrotic syndrome → renal function + lipid profile.
Confirmed ALL → bone marrow trephine + cytogenetics

Plan Management

The result will directly change what you do next.

Examples Before chemotherapy → baseline renal + liver function.
Before surgery → coagulation profile if indicated.

Monitor Progress

Treatment is running. You are checking whether the patient is responding.

Examples TB → clinical and radiological response.
Leukemia → blood counts during chemotherapy.

Diagnostic Decision-Making

Investigate or Not (for diagnostic testing)
When the purpose is to establish or exclude a diagnosis, clinical decision-making can be viewed as three options based on how likely a condition is: do nothing when it seems very unlikely, investigate (test) when there is uncertainty and more information is needed, and treat empirically when the condition appears likely enough to act without waiting for test confirmation.

Note: This threshold model applies only to diagnostic investigations (Purposes 1 and 2: confirm or exclude a diagnosis). For management purposes (plan management, detect complications, monitor progress), investigation is not an alternative to treatment; it guides or accompanies it.

The same model, as a probability line

Do Nothing Below the test threshold

Investigate Between the two thresholds

Treat Empirically Above the treatment threshold

Low pre-test probability → High pre-test probability

The test threshold is the probability below which investigating costs more (risk, delay, expense) than it is likely to be worth. The treatment threshold is the probability above which treating now is more rational than waiting on a result. Investigation is only the right move in the zone between the two — not the automatic default for any degree of uncertainty.

Approach

Two ways to use investigations

Testing Approach

Hypothesis-Driven vs. Exploratory Testing

Approach A: Hypothesis-Driven

You expect a specific result

You suspect a disease. You order a test expecting a particular result. The result either strengthens or challenges your hypothesis. This is scientific clinical reasoning.

Example Suspected iron deficiency → expect low ferritin, low MCV, low MCH.
Result matches → diagnosis strengthened.
Result does not match → rethink the hypothesis.

Approach B: Exploratory Testing

You are gathering information

Sometimes you genuinely do not know. The presentation is undifferentiated. You order baseline investigations to narrow the field. This is legitimate, but must still be logical, not random.

Example Unexplained fever of unknown origin → FBC, CRP, blood culture, urinalysis.
Each test has a reason. Together they map the territory.

The valid approach: hypothesis-driven or exploratory testing

Both are valid.
However, hypothesis-driven testing should be the default, while exploratory testing should be reserved for genuinely undifferentiated presentations. The difference is not in the tests ordered; it is in whether the clinician has a reason for each test.
Exploratory testing that is logical and purposeful is good medicine. Random testing (even if it accidentally finds something) is not.

Checklist

Before You Order

The Pre-Investigation Checklist

Five Questions Before Every Test

Apply these before ordering any investigation.

What am I looking for?

State the specific question this test will answer. If you cannot, do not order it.

Is it indicated?

Does the clinical picture justify this test? Does it serve one of the five purposes of investigation?

Is it safe?

Are there contraindications for this patient?

What if positive?

What will you do if the result confirms the diagnosis or concern?

What if negative?

Will a negative result reassure you or change your management?

Filter

If the result will not change what you do, you should not order the test. This is the most powerful filter. Tests that are ordered "just in case" or "for completeness" fail this criterion every time.

New Addition: A Critical Skill

What if the result surprises you?

When a result does not match your expectation, it is one of the most important moments in clinical reasoning. Do not dismiss it, and do not panic. Work through it systematically.

Check the sample first, but treat this as a brief, systematic checkpoint, not a default explanation. Confirm labeling or processing issues; if no technical error is found, proceed immediately to clinical reconsideration.

Reconsider the diagnosis. Could the unexpected result be telling you something true? Is there a disease you have not yet considered that this result would fit?

Do not dismiss unexpected findings. Incidental findings and unexpected results have led to important diagnoses. Treat them as new information, not as noise to ignore.

Repeat if necessary. A single unexpected result on a test prone to analytical variability may warrant repetition before definitive action, but clinical urgency must take precedence. Do not delay management of a critically abnormal result in order to repeat a test.

Common Errors

What undiscerning investigative thinking looks like

Ordering tests before forming a clinical hypothesis Investigations should answer a specific question: without a hypothesis, you cannot interpret a normal or abnormal result. "Just to check" leads to over-investigation and false positives.

Ordering everything "just in case," without a stated purpose for each Each test has risks (incidentalomas, radiation, cost, false positives). A clear purpose prevents scattergun ordering and teaches resource stewardship.

Not knowing what result to expect before the test is run If you cannot predict a likely result (positive or negative), you do not understand the test's operating characteristics or the pre-test probability. This is the most common error in test ordering.

Not knowing what to do when an abnormal result returns An abnormal result without an action plan is noise. Before ordering, ask: "If this is high or low, will I change management? If not, why am I ordering it?"

Ignoring contraindications to investigation Contraindications include patient anxiety, radiation exposure in pregnancy, contrast nephropathy, or risks of biopsy. A safe clinician always considers harm before yield.

Dismissing an unexpected result rather than reconsidering the diagnosis An unexpected abnormal result is a diagnostic crossroads; it should prompt hypothesis revision, not confirmation bias. "I do not believe this result" without verification is a cognitive error, not clinical judgment.

In short

An investigation is a hypothesis made tangible: state the question you expect it to answer, name the result you anticipate, and decide before you order whether any possible result will genuinely change what you do next.

Key Takeaway

Investigations are tools.
Not decorations. Not routines.
Not replacements for thinking.

Good clinicians do not order many tests.
They order the right tests, at the right time, for the right reason.

Think first Name the question Would it change the plan?

Do not order tests by ticking boxes.Think first.