Critical Appraisal Nugget 10
Long-term listeners to the St Emlyn’s podcast may remember our ‘Critical Appraisal Nuggets’ (CANs). In February, we re-invigorated the series with an episode on semi-structured interviews, learning all the pros and cons of this qualitative method with Dr. Laura Howard.
Sensitivity & Specificity
The latest CAN is one of our brand-new ‘revision editions’ — brief podcasts aimed at covering the essentials of critical appraisal for medical students and junior doctors preparing for exams.
With the help of Gregory Yates, an academic doctor based in Manchester, this episode introduces two core concepts: sensitivity and specificity. These are two ways of thinking about the accuracy of a diagnostic test. Knowing the sensitivity and specificity of an investigation will give you a decent idea of how it should be used in the emergency department.
Sensitivity (Sn) describes the chance that a test will be positive if your patient has the condition you’re testing for. Some people call it the ‘true positive rate’ or alternatively the positivity in disease (PID) rate. If you need a hand remembering it, you can always remember that PID is a sensitive issue.
Meanwhile, specificity (Sp) considers the chance of a test being negative if the patient doesn’t have the condition you’re testing for. It’s the ‘true negative rate’ or alternatively the negativity in health (NIH) rate. There are times when we particularly need a test to have a high sensitivity. This is generally when we want to be particularly confident that a test accurately identifies everyone with the relevant condition because we really don’t want to miss it. We need a high sensitivity to rule out disease. (Sn-uff it out). At other times, we need to be confident that a patient with a positive test actually has the disease – for example, if the treatment is unpleasant or involves exposing patients to risk. In that case, we want a high specificity to rule in disease. (Sp-in it in).
In this CAN, we use D-Dimer as an example of a very sensitive investigation: it’s positive in nearly 100% of cases of venous thromboembolism. Specificity describes the likelihood that the test will be negative if your patient does not have the disease. We use HbA1c as an example of a highly specific investigation: it’s rarely used in the emergency department, but if it’s elevated, we can be almost certain that the patient is diabetic. HbA1c is almost never (<1%) raised in non-diabetics.
The trouble is, many patients with a positive D-Dimer do not have a venous clot, and the majority of diabetics will have a normal HbA1c! No test is perfect, and we discuss how emergency physicians weigh up sensitivity and specificity when choosing which investigations are the best “fit” for clinical decision-making. By the end of this CAN, you will be ready to do the same — in your exams, and on the shop floor.
Gregory Yates and Rick Body