08. Testing Thresholds.pdf

Testing Thresholds Foundations of Clinical Medicine CMS100 Objectives 1. Understand the clinical value of testing thresholds 2. Understand the factors influencing where clinicians set testing thresholds 3. Act appropriately based on the relationship between the probability of disease and the testing...

Testing Thresholds Foundations of Clinical Medicine CMS100 Objectives 1. Understand the clinical value of testing thresholds 2. Understand the factors influencing where clinicians set testing thresholds 3. Act appropriately based on the relationship between the probability of disease and the testing threshold 4. Apply testing thresholds to patient cases Clinical Utility We can never be certain of a diagnosis or its absence (i.e. we can’t reach 100% or 0% probability) Every test has a cost: seeking further certainty incurs further harm E.g. time spent, money spent, radiation exposure We must therefore decide (ideally in advance) when to stop investigating based on the balance of the benefit from further information and the harm of further investigation Testing Threshold “Zones” 0% Ruled out Further testing needed “Trash” “Test” ← Testing threshold ← Probability of the condition If the probability of disease is above the testing threshold Further testing (questions, physical exams, labs, etc.) is needed to move the probability past a threshold “Test” zone Differentials here are often called “active alternatives” 100% Testing Threshold “Zones” 0% Ruled out Further testing needed “Trash” “Test” ← Testing threshold ← Probability of the condition Otherwise, if the probability of disease is below the testing threshold Stop doing tests related to that specific differential diagnosis “Trash” zone Differentials here are considered “ruled out” or “excluded” 100% Where Is Your Threshold? Subjective Factors: The Condition Harms of ruling out a diagnosis that may be present The Test Accuracy: How low is the LR; i.e. how much information would it provide? Harms: financial, time spent, other Note: there are many potential harms; consider patient values and preferences I.e. consider the balance of accuracy of the test vs. the harms of the test. Other Considerations Should usually not set the testing threshold below the population prevalence This would suggest that you think someone taken at random from the population is in the “testing” zone, implying at least that you think everyone who walks into your office should be investigated for the condition (also see screening lecture) More holistically: “Would I be comfortable sending the patient home with a x% chance that they had this condition when [some test] is available?” Where Is Your Threshold? The test zone expands (and “trash” zone shrinks) if: Greater harm of ruling out the condition if it is present More favorable balance of test harm relative to information benefit The “trash” zone expands (and test zone shrinks) if: Less harm of ruling out the condition if it is present Less favorable balance of test harm relative to information benefit Source Example A 45-year-old man tells his naturopathic doctor about his 10-year history of occasional headaches which tend to occur predominantly on the right side of his head. He rates the intensity of the pain as grade 5 out of 10 on a scale of 1-10 and finds he is able to continue working during these headaches. They last between 4 to 5 hours. He has no symptoms before the headaches begin. There is no vomiting or light sensitivity. He was diagnosed with diabetes 1 year ago, which is being treated with metformin A physical examination reveals mild weakness (power graded 4/5) and increased reflexes in the right leg. Example Problem List Processed Problem List 45 years old Male 10-year history Occasional Right-sided Pain 5/10 Can continue working Last between 4 to 5 hours No symptoms before the headaches No vomiting No light sensitivity Mild right leg weakness 4/5 Increased reflexes in the right leg Diabetic Taking metformin Middle aged Male Chronic Intermittent 4-5 hrs duration Unilateral Not disabling No prodrome No vomiting No photophobia R leg mild muscle weakness R leg hyperreflexia Diabetes medicated with metformin Example Patient Illness Script Epidemiology 45-year old male Timing Chronic, intermittent, 4-5 hrs duration Syndrome Unilateral, non-disabling headaches without Statement vomiting, photophobia or prodrome. Mild R leg muscle weakness and hyperreflexia. Other history Diabetic, taking metformin Example Migraine Epidemiology Timing Syndrome Mechanism Tension-type headache Brain tumor Very common Any age M~F Rare Typically later life M~F History of cancer Recurring, 4-72 hrs duration Recurring Last hours to days or may be unremitting Progressive, constant, worse at night or early morning Unilateral headache Disabling intensity Nausea Photophobia Phonophobia May be preceded by an aura Bilateral headache Mild-to-moderate intensity Pressing or tightening, Non-pulsating No more than one of photophobia, phonophobia or mild nausea Headache Neurological findings Variable presentation but often meet criteria for tension headaches Classically: nausea and vomiting, worse in am Complex; likely neurogenic muscular origin; related to increased resting muscle tension Usually metastasis; presses on surrounding brain tissue Common Often start in adolescence F>M Example Consider the brain tumor differential. First consider what your testing threshold would be. Consider asking yourself, for different probabilities: “Would I be comfortable sending the patient home with a x% chance that they had a brain tumor when (referral for) MRI is an option?” Example Assuming you find no other notable symptoms during the intake, you might take the probability to be 0.5%. Are we above or below your testing threshold? On this basis, what’s the appropriate action to take? 100% 0% ← Probability of brain tumor Example Another differential was tension-type headache. Again, consider what your testing threshold would be. Consider asking yourself, for different probabilities: “Would I be comfortable sending the patient home with a x% chance that they had a tension-type headache?” Example Based on the findings found so far during the intake, you might take the probability to be approximately 70%. Are we above or below your testing threshold? On this basis, what’s the appropriate action to take? 100% 0% Probability of tension headache → Summary 1. Testing thresholds set a probability below which we stop diagnostic work-up. We do this because complete certainty is impossible 2. Testing thresholds are set according to the harm of inappropriately ruling out a diagnosis and the balance of the harms and benefits of testing 3. When the probability of disease is above a testing threshold, further testing or other management is needed. When it is below a testing threshold, we stop diagnostic work-up for that condition.