Direct answer
How do you distinguish sensitivity, specificity, positive predictive value and negative predictive value for MRCPCH?
Sensitivity and specificity start with the true condition status: sensitivity asks how often a test is positive when the condition is present, while specificity asks how often it is negative when the condition is absent. Positive and negative predictive values start with the test result: PPV asks how often a positive result is a true positive, while NPV asks how often a negative result is a true negative. Predictive values change when prevalence changes.
Key points
- Sensitivity and specificity start with true condition status; predictive values start with the observed test result.
- Use counts in a 2 × 2 table before calculating percentages so denominator errors remain visible.
- As prevalence rises, PPV generally rises and NPV generally falls; the reverse occurs when prevalence falls.
Sensitivity and specificity start with the true condition status: sensitivity asks how often a test is positive when the condition is present, while specificity asks how often it is negative when the condition is absent. Positive and negative predictive values start with the test result: PPV asks how often a positive result is a true positive, while NPV asks how often a negative result is a true negative. Predictive values change when prevalence changes.
This article is for MRCPCH examination education. It explains statistical interpretation, not whether a particular child should be tested or treated.
The direct answer: identify the denominator first
Most confusion disappears when you ask one question before touching a formula:
What group am I starting with?
- Sensitivity: start with everyone who truly has the condition.
- Specificity: start with everyone who truly does not have the condition.
- Positive predictive value (PPV): start with everyone whose test is positive.
- Negative predictive value (NPV): start with everyone whose test is negative.
Sensitivity and specificity describe how a test separates people according to a reference standard. Predictive values describe what a result means within the population being tested. That is why prevalence matters much more directly to PPV and NPV.
The current RCPCH Theory syllabus places statistical testing, evidence-based medicine and screening within Science of Practice. These concepts can therefore appear as calculations, study interpretation or data questions across the MRCPCH pathway.
Build one 2 × 2 table
Use the same table every time:
| Condition present | Condition absent | |
|---|---|---|
| Test positive | True positive (TP) | False positive (FP) |
| Test negative | False negative (FN) | True negative (TN) |
Then choose the denominator from the wording:
| Measure | Question | Formula |
|---|---|---|
| Sensitivity | Of those with the condition, how many test positive? | TP ÷ (TP + FN) |
| Specificity | Of those without the condition, how many test negative? | TN ÷ (TN + FP) |
| PPV | Of all positive results, how many truly have the condition? | TP ÷ (TP + FP) |
| NPV | Of all negative results, how many truly do not have it? | TN ÷ (TN + FN) |
Notice the direction. Sensitivity and specificity move from truth to test. Predictive values move from test to truth.
A compact recall rule
Use state first, result first:
- Sensitivity and specificity: state first - condition present or absent.
- Predictive values: result first - test positive or negative.
This is more dependable than remembering four disconnected formulas.
A worked example
Imagine 100 people. Twenty have the target condition and 80 do not. A test has 90% sensitivity and 80% specificity.
Start with the condition-present group:
- 90% of 20 are true positives: 18 TP.
- The remaining two are false negatives: 2 FN.
Now use the condition-absent group:
- 80% of 80 are true negatives: 64 TN.
- The remaining 16 are false positives: 16 FP.
The table now contains everything:
- Sensitivity = 18 ÷ 20 = 90%.
- Specificity = 64 ÷ 80 = 80%.
- PPV = 18 ÷ 34 = approximately 53%.
- NPV = 64 ÷ 66 = approximately 97%.
The PPV may look surprisingly low. The test produced almost as many false positives as true positives because the condition was uncommon in this sample. A high sensitivity does not guarantee that most positive results are true.
Why prevalence changes predictive values
Keep the same 90% sensitivity and 80% specificity, but imagine a population in which 50 of 100 people have the condition.
- TP = 45 and FN = 5.
- TN = 40 and FP = 10.
- PPV = 45 ÷ 55, approximately 82%.
- NPV = 40 ÷ 45, approximately 89%.
The test characteristics in this simplified example did not change, but PPV rose from about 53% to 82% because the condition became more prevalent. NPV fell because a negative result now came from a population with a higher pre-test probability.
For MRCPCH questions, this leads to a reliable direction rule:
- As prevalence rises, PPV tends to rise and NPV tends to fall.
- As prevalence falls, PPV tends to fall and NPV tends to rise.
Real diagnostic accuracy can also vary with spectrum, setting, thresholds and the reference standard. Sensitivity and specificity should not be treated as universal constants detached from the study population.
False-positive and false-negative rates
Two complementary rates are often tested:
- False-negative rate = 1 − sensitivity. It is the proportion of people with the condition who test negative.
- False-positive rate = 1 − specificity. It is the proportion of people without the condition who test positive.
Do not calculate the false-positive rate as 1 − PPV. That gives the proportion of positive results that are false positives, which uses a different denominator.
Where likelihood ratios fit
Likelihood ratios combine sensitivity and specificity:
- LR+ = sensitivity ÷ (1 − specificity)
- LR− = (1 − sensitivity) ÷ specificity
For the worked example, LR+ is 0.90 ÷ 0.20 = 4.5, while LR− is 0.10 ÷ 0.80 = 0.125. A positive result raises the odds of the condition; a negative result lowers them. Likelihood ratios can convert pre-test odds into post-test odds, but do not perform that conversion unless the question asks for it.
How the examiner may test the distinction
Use a four-step method:
- Underline the starting group. "Among children with the condition" signals sensitivity; "among positive results" signals PPV.
- Draw the table. Label TP, FP, FN and TN before entering numbers.
- Use counts before percentages. Counts make denominator errors visible.
- Check the direction. If prevalence falls, a claimed rise in PPV should make you pause.
For data-interpretation practice, pair this method with revision that matches the MRCPCH exam format. If formulas remain mixed up, a small comparison grid is usually more useful than a large concept map.
Common MRCPCH traps
"A sensitive test confirms the diagnosis"
Not by itself. Sensitivity begins with people already known to have the condition. The meaning of a positive result depends on false positives and pre-test probability as well.
"Predictive values belong permanently to the test"
They do not. PPV and NPV vary with prevalence and setting. Quote them with the population in which they were estimated.
"Screening always means maximum sensitivity"
Screening often values sensitivity, but the appropriate threshold also depends on harms, costs, follow-up testing and the intended role of the test. Treat "screening equals sensitivity" as a prompt, not a complete principle.
"Changing the cut-off improves everything"
For many continuous tests, lowering a positive threshold increases sensitivity but reduces specificity; raising it usually does the reverse. A new threshold trades one type of error against another.
Conclusion
Start with the denominator. Sensitivity and specificity begin with true condition status; PPV and NPV begin with the observed test result. Build the 2 × 2 table, calculate counts, and remember that prevalence moves predictive values. This single workflow is safer than relying on isolated mnemonics.
Frequently asked questions
What is the easiest way to remember sensitivity and specificity?
Ask which true-state group you start with. Sensitivity starts with people who have the condition and asks how many test positive. Specificity starts with people without the condition and asks how many test negative.
What is the difference between sensitivity and PPV?
Sensitivity is the proportion of people with the condition who test positive. PPV is the proportion of positive results that are true positives. They use different denominators.
How does prevalence affect PPV and NPV?
When prevalence rises, PPV generally rises and NPV generally falls. When prevalence falls, PPV generally falls and NPV generally rises, assuming the other test characteristics remain comparable.
Is the false-positive rate 1 minus sensitivity?
No. The false-positive rate is 1 minus specificity. One minus sensitivity is the false-negative rate.
Sources
- RCPCH - Theory exams: structure and syllabi
- RCPCH - MRCPCH Theory examination syllabi, version 3
- FDA - Statistical guidance on reporting diagnostic-test studies
- STARD 2015 explanation and elaboration
- Oxford Centre for Evidence-Based Medicine - Likelihood ratios
- Understanding diagnostic tests: sensitivity, specificity and predictive values
- Royal Statistical Society - Diagnostic tests report
- PaedVault Acceptable Use Policy
Quick answers
Frequently asked questions
What is the easiest way to remember sensitivity and specificity?
Ask which true-state group you start with. Sensitivity starts with people who have the condition and asks how many test positive. Specificity starts with people without the condition and asks how many test negative.
What is the difference between sensitivity and PPV?
Sensitivity is the proportion of people with the condition who test positive. PPV is the proportion of positive results that are true positives. They use different denominators.
How does prevalence affect PPV and NPV?
When prevalence rises, PPV generally rises and NPV generally falls. When prevalence falls, PPV generally falls and NPV generally rises, assuming the other test characteristics remain comparable.
Is the false-positive rate 1 minus sensitivity?
No. The false-positive rate is 1 minus specificity. One minus sensitivity is the false-negative rate.
