Here’s another example: In scenario X, we have no screening method and 1000 people are diagnosed with progressive breast cancer. 600 of them die over the course of five years, which works out to a 40 percent survival rate. In scenario Y, those same 1000 people get progressive cancer, but another 2000 people get a screening test that shows they have those abnormal but ultimately harmless cells we talked about earlier. Those people all live, and the same 600 people with progressive tumors die. Now our survival rate is 80 percent. But we haven’t saved any more lives, we just told more people they have cancer.
Let’s use an example. In scenario A, doctors diagnose a 67-year-old man with prostate cancer when he shows up to a physical with an enlarged prostate. The cancer kills him three years later. The five-year survival rate of our single-man study is zero percent. In scenario B, that same man gets a PSA test when he’s just 60, but he still dies at 70. Now our five-year survival rate is 100 percent, even though the man didn’t live any longer — he just spent an additional seven years of his life as a cancer patient. This example is extreme (no one bases their survival rates on a single person), but you can imagine how having a whole bunch of cases similar to this one in the mix could skew statistical results. Finding cancer earlier doesn’t always change your prognosis, but it always changes your length of survival as a cancer patient. That makes it seem like a test helps prolong life, even when it doesn’t.