We talked about diagnostic testing in our biostatistics/epi class yesterday, and I really think it was the first time that I realized how important statistics can be. It was probably the first time that I really saw how statistics affect patients–individuals–not big masses of people.
We talked about prevalence, pre- and post-test probabilities, predictive values, and likelihood ratios. Reading it now, it sounds pretty dry and boring, but it absolutely affects how likely a patient is to have a certain disease. It’s these kind of decisions and analyses that are absolutely vital to medical decision-making, and vital to being able to provide the most information and the best information to a patient; they also say a lot about how efficiently we spend health dollars in the US.
The big boom in “spiral CTs” or “whole body scans” can be pretty worthless if there’s no clinical suspicion of illness. Even a test that’s sensitive and specific 95% of the time will be a worthless test if the prevalence of a disease is extremely rare; one may increase their risk of some disease from 1% to 2%, but 98 times out of 100 you won’t have the disease, and you’ve just spent $6,000 to tell you that.