Today's unrelated tidbits:
- A great tool for helping men with a decision for/against PSA testing. This thing is buried deep in the guts AAFP website – so deep that I couldn't find it. Here it is.
- A very good review article on dx/work-up of fatigue.
- Excellent guidelines on Otitis, pharyngitis and bronchitis
- Baysean Calculator.
The last one is the best of many calculators out on the web. Here's why – it combines the calculations of likelihood ratios with the calculation of PPV and NPV. Huh? Terms for normal humans:
a) How good is the test that I'm doing for this disease? (sensitivity) let's take one that is VERY common: the "rapid strep" test. It's pretty good, but not perfect. Let's say it's 85% sensitive. This means that 15% of the time it will miss strep throat. So … does a patient in my office with a sore throat and a negative "rapid strep" have a 15% chance of having strep throat? Nope. Read on.
b) How accurate is the test? (specificity). If I get a positive result – how sure can I be that the patient really has it? In the case of the rapid strep – darn sure. Let's say 99%.
c) How likely is THIS patient to have this disease given their symptoms? Long story – Kids are more likely to have it than adults. Let's say an adult has a 10% chance of having a sore throat caused by strep – and a kids have a 25% chance.
So now we can use the calculator. Enter a prevalence of 25% (it's a kid with a sore throat) and a sensitivty of 80% and a specificty of 99%.
PTL- .. The post-test likelihood of a negative test = .048 So a kid with a negative rapid strep has only a 5% chance of having strep throat – not 15% like we thought above.
But we can play with this plot even more. We know from the long story above (click it – it's a very good article) that the kid with a cough and no adenopathy is much less likely to have strep than the kid with no cough and positive anterior adenopathy. So let's say our patient has a cough and runny nose and not much of a fever. Hmm .. doesn't sound much like strep. But he does have a sore throat. Let's move his "prevalence" score to 10%.
Yikes. Now the PTL- is .01 — not very likely that this kid has strep.
So now you see why I'm not compelled to send a follow-up culture (with sensitivity of 90% rather than 85%) if I get a negative rapid strep.