St Emlyn’s Lesson Plans

Diagnostic testing: decide before you request

Use a D-dimer case and a worked table to understand test accuracy, pre-test probability and the decision after a result.

Early-career Emergency Medicine · 45-minute session

For learners

Before: Allow about 17 minutes in total. With the reading alternative: about 13 minutes.

When we meet: We will discuss a PE investigation pathway and calculate accuracy measures for a fictional test at different starting probabilities.

After: choose a point to follow up in practice

For facilitators

Send: the learner preparation link before teaching.

Bring: Whiteboard or paper, calculator, four labelled table cells and current PE pathway. No statistical software is required.

Run: 45-minute session plan · 30-minute option

Before the session

If you are attending this session, complete the core preparation below before you come. Allow about 17 minutes in total. With the reading alternative: about 13 minutes.

Core preparation

Core listening · 11 min 30 sec

St Emlyn’s Podcast: Ep 255 - Likelihood Ratios: Critical Appraisal Nugget 12

An explanation of why the same test result can mean different things in different patients.

Listen to the episode

Search for this episode by title in MedPod Learn if you want to record your listening there.

Prefer to read? Use Oxford CEBM: likelihood ratios instead of listening. Allow about 8 minutes.

Read the Definition and Calculation sections, then complete the numerical preparation task below. The separate question to think about asks about your experience of ordering tests.

https://www.cebm.ox.ac.uk/resources/ebm-tools/likelihood-ratios

Listen for the conversion between probability and odds, then complete the preparation task below using its fictional numbers. The episode's ECG calculator example applies only within its validated scope; use the current local PE pathway during the clinical case.

https://www.stemlynspodcast.org/e/ep-255-lielihood-ratios-critical-appraisal-nugget-12/

Prepare (allow 5 minutes): For a fictional test, start with a pre-test probability of 10% and a positive likelihood ratio of 4. Calculate the pre-test odds, post-test odds and post-test probability. Bring your working and one sentence explaining why a positive result does not automatically establish the diagnosis.

Think about: Can you think of a test you ordered whose result did not help you make a decision?

Picture the scene

Your consultant asks, “What will you do if that D-dimer is positive?” Before ordering it, have you worked out what either result would change?

By the end of the session

Explain sensitivity, specificity and predictive values using a two-by-two table.
Show why the same test result can mean different things at different pre-test probabilities.
Choose testing within an appropriate clinical pathway and state what each possible result will change.

After the session

Before your next low-yield investigation, write down what a positive and a negative result would each change.

Facilitator guide

Send learners the preparation link several days before teaching. Allow about 17 minutes in total. With the reading alternative: about 13 minutes.

Before learners arrive

Check the preparation example: pre-test odds 0.10/0.90=1/9; post-test odds 4/9; post-test probability 4/13, about 31%. Recalculate the case table too. Bring the current local PE pathway and keep its clinical thresholds separate from these fictional numbers.
Choose an example from your practice where testing created an incidental finding or extra investigation, while keeping the patient unidentifiable.

Equipment: Whiteboard or paper, calculator, four labelled table cells and current PE pathway. No statistical software is required.

Setup: Ask pairs to calculate independently, then compare answers. Have learners explain the denominators in words so any arithmetic errors can be worked through together.

If the preparation has not been completed

Replace the five-minute opening with the fictional FRB counts in case two. Draw four boxes and ask pairs to label disease status, test result and the relevant totals before introducing formulas. Use the PE pathway together during case one. In the 30-minute plan, use five minutes here and reduce the D-dimer case from eight to six minutes; omit likelihood-ratio calculations if the denominators still need practice.

Suggested session plan

0–5 min — What will the result change? Use the D-dimer hook to expose unplanned testing.
5–17 min — Clinical pathway. Complete the probability assessment and identify the next action for each result.
17–32 min — Worked table. Calculate the four measures and likelihood ratios with visible denominators.
32–40 min — Change the baseline probability. Compare the 10% example with a different prior probability and discuss decision thresholds.
40–45 min — Close. Each learner names one question to ask before their next test request.

Start the discussion

Read the opening scenario together. Give everyone a moment to think, then ask what concerns them and what they need to know next.

Cases and discussion

Use these cases, or invite an anonymised case from the group that addresses the same questions.

A D-dimer needs a question

A 51-year-old has mild pleuritic pain and breathlessness. They are stable. A clinician is about to order D-dimer before completing the history or deciding how likely PE is.

Add this information as the discussion develops:

Ask learners to identify what information is needed to choose the appropriate pathway.
After assessment, PE remains a possible diagnosis but is unlikely by the two-level Wells score.
The D-dimer exceeds the applicable threshold. Ask the group to state what has changed and what has not.

Ask the group:

What is the clinical question, and is testing needed at all?
What information is lost when someone calls a Wells score “negative”?
Does an abnormal D-dimer confirm PE?
How will you explain the next step and residual uncertainty to the patient?

Discussion notes

Points to draw out:

Use the validated clinical pathway. NICE supports PERC in selected people with low overall clinical suspicion and feasible alternative diagnoses, then probability-based testing where PE is suspected.
For an appropriate low-probability pathway, a negative D-dimer can avoid imaging; a positive result generally moves the pathway towards imaging rather than confirming PE. Consider the guideline’s age-adjusted threshold advice in people over 50.
D-dimer performance varies with the assay and population. It is not invariably raised in every case of VTE.

Common misconceptions:

Ordering first and estimating probability afterwards.
Calling a low-specificity test “useless when positive” instead of explaining its pathway consequence.

For a more experienced group: Change the history to high clinical suspicion. Ask why a reassuring D-dimer cannot simply be transplanted from the low-probability pathway.

A fictional test called FRB

FRB is a fictional test used for this calculation exercise. Of 400 people, 200 have the target condition. FRB is positive in 80 with disease and 20 without disease; it is negative in 120 with disease and 180 without disease.

Add this information as the discussion develops:

Draw four boxes and check that row and column totals agree.
Ask one pair for sensitivity and specificity and another for predictive values.
Now imagine the test used in a much lower-prevalence population, while holding sensitivity and specificity fixed solely for the arithmetic exercise.

Ask the group:

Which denominator belongs to sensitivity, and which belongs to positive predictive value?
Can this test exclude the condition when negative?
What do LR+ and LR− contribute?
Why might accuracy itself change when the patient spectrum changes?

Discussion notes

Points to draw out:

The supplied arithmetic gives sensitivity 80/200=40%, specificity 180/200=90%, PPV 80/100=80% and NPV 180/300=60%.
LR+=0.40/(1−0.90)=4. LR−=(1−0.40)/0.90≈0.67. Keep the brackets in the calculation so the numerator and denominator are clear. A negative result makes only a modest downward probability shift.
For a fictional 10% pre-test probability, pre-test odds are 0.10/0.90. Multiplying by LR+ 4 gives post-test probability about 31%; using LR− 0.67 gives about 7%. Neither result alone dictates every management choice.
Predictive values depend on prevalence. Real studies also need scrutiny of patient selection, reference standard, verification and spectrum; treating sensitivity and specificity as fixed was a teaching simplification.

Common misconceptions:

Reversing disease and test denominators.
Saying a positive test means an 80% chance of disease in every future setting.
Assuming the fictional test's performance has been established in real patients.

For a more experienced group: Use 1% and 50% pre-test probabilities and ask where a result might cross a testing or treatment threshold.

Important clinical points

The fictional FRB numbers are for learning the calculations. Clinical decisions need the appropriate assay and validated pathway.
A negative result can be insufficient when the pre-test probability remains high.

Closing the session

Use these questions to check what the group will take away. Go back over any clinical points that remain unclear.

How is sensitivity different from the probability of disease after a positive result?
Why does a positive D-dimer usually lead to another step?
What would make you distrust an apparently excellent accuracy study?

Key teaching points:

Before requesting a test, decide what each result would change.
Use the two-by-two table to check the denominator in each accuracy measure.
Interpret the result alongside the probability of disease before testing.

Before leaving, ask each person to choose one action from After the session and say when they will try it. They complete that follow-up in practice, with supervision where needed.

Adapting the discussion

Uncertainty and local practice

Accuracy estimates have confidence intervals and may not transfer unchanged between settings. The arithmetic exercise isolates one concept and does not validate FRB or replace a clinical pathway.

If the discussion drifts

Which patients are in your denominator, and what action follows this result?

For a more experienced group

Add partial verification or an imperfect reference standard and ask how it could distort the apparent test performance.

A 30-minute version

30-minute option: question 3, D-dimer 8, table 12, probability and closing 7. Omit likelihood-ratio calculations if the group first needs secure denominators.

Evidence and guidance

NICE NG158: Venous thromboembolic diseases — Clinical probability, PERC in selected low-suspicion patients, D-dimer and imaging pathways.https://www.nice.org.uk/guidance/ng158/chapter/Recommendations
Oxford CEBM: Likelihood ratios — Definition and Calculation sections support the preparation arithmetic. The fictional example is not a validated clinical threshold.https://www.cebm.ox.ac.uk/resources/ebm-tools/likelihood-ratios