Picture the scene…
You are about to send a D-Dimer on a patient with chest pain, when your consultant says “I know it’s a sensitive test, but what will you do when it comes back positive?…”
Learning Objective
To learn about diagnostic testing and how we can tell if a test will help us in our decision making.
RCEM Curriculum
CC5
Task 2 – Read
Read this short article1 from the Centre for Evidence Based Medicine to reenforce what you have just listened to.
Task 3 – Discuss
This part of the teaching session should be lead by an experienced clinican. The cases and exercises provided are merely examples and if possible the learners should be encouraged to discuss patients they have seen in their clinical practice.
Case 1 – A patient with shortness of breath
A A 51 year old female presents to the Emergency Department with shortness of breath and mild pleuritic chest pain. She has no risk factors for pulmonary embolism and has a Well’s score of 1. Following your ED pathway you are about to request a D-dimer, when you wonder what the diagnostic characteristics of test are.
1, What is a D-dimer?
Yes, I know this is nothing to do particularly with diagnostic testing, but it’s worth knowing!
D-dimers are protein products of cross-linked fibrin degradation that are present in the blood of most healthy individuals in only negligible amounts. When the body is starting to break down clots, elevated blood concentration of D-dimer is by extension evidence of intravascular coagulation and thrombotic disease.
Although elevation of D-dimer is invariably evident in those with VTE, it can also be evident in many other conditions that are associated with a pro-coagulant state, hence why the specificity is so poor.
2, A D-dimer has a sensitivity of 90% and a specificity of 40%. What does this mean in practice?
A test that is sensitive is good at ruling out disease – in essence the fewer false negatives, the higher the sensitivity, and the more likely a negative test is true.
As the D-Dimer has low specificity it means that a positive test doesn’t really help us at all: a positive test doesn’t mean you have the disease, it just means we cannot say that don’t have it.
3, How will this affect our decision making with this patient?
As our patient has a low pre test probability (with a negative Well’s score) a negative D-dimer will mean that we can say that this patient’s chance of having a venous thromboembolism is low.
If the D-dimer is not negative (note I didn’t say “positive”!) we cannot say that the patient has had a PE, just that we cannot say that she hasn’t.
Case 2 – A new test
The following results were obtained from a randomised controlled trial of a new test “FRB” for the detection of heart failure.
Heart Failure Present | Heart Failure Absent | |
FRB-positive | 80 | 20 |
FRB-negative | 120 | 180 |
1, What is the sensitivity and specificity of FRB for diagnosis of heart failure?
Target disorder Present | Target disorder Absent | |
FRB-positive | a=80 | b=20 |
FRB-negative | c=120 | d=180 |
Sensitivity = true positives/all patients with target disorder
Specificity = true negatives/all patients without target disorder
Sensitivity = a/(a+c) = 80/200 = 40%
Specificity = d/(b+d) = 180/200 = 90%
Although elevation of D-dimer is invariably evident in those with VTE, it can also be evident in many other conditions that are associated with a pro-coagulant state, hence why the specificity is so poor.
2, What is the Positive Predictive Value (PPV) and Negative Predictive Value (NPV) for FRB?
Positive predictive value = true positives/all patients who test positive
Negative predictive value = true negatives/all patients who test negative
PPV = a/(a+b) = 80/100 = 80%
NPV = d/(c+d) = 180/300 = 60%
3, Would you use FRB in the diagnosis of Heart Failure?
Maybe! The specificity looks ok and when converted into likelihood ratios the LR+ = 4 (0.4/(1-0.9)), so a positive test may be helpful.
The negative likelihood ratio is 0.67 (LR- = 1-0.4/0.9 = 0.67), which doesn’t change the probability of the disease being absent significantly, so a negative test doesn’t rule out heart failure as a diagnosis.
Task 4 – Summary
In this session we have explored how we can measure the effectiveness of a diagnostic test using sensitivity and specificity. Not all tests are created equal and must be put into the context of the patient’s pre test probability.
Task 5 – Reflect
In order to embed today’s learning further, reflect on what you have learnt and record in your portfolio whether it has had any impact (or is expected to have any impact) on your performance and practice.
Was this a topic that you were confident you knew already? Which parts were new to you? Were there elements that you will use on your next clinical shift.
Dscuss this session with your colleagues – were there people who missed it who you can share the highlights with?
References
- 1SpPin and SnNout. Centre for Evidence Based Medicine. https://www.cebm.net/2014/03/sppin-and-snnout/ (accessed June 18, 2020).