If it weren't for routine PSA prostate cancer screening, an extra 17,000 Americans each year would learn that they had the worst form of the disease, a new study suggests.

That kind of prostate cancer -- metastatic prostate cancer, in which the cancer spreads to the bone or other parts of the body -- is rapidly fatal, usually within two years or less.

The new study seems to make a powerful argument in favor of PSA testing. It finds that in the three years before widespread PSA testing (1983-1985), men getting their first diagnosis of prostate cancer were three times more likely to learn they had very late-stage cancer than men diagnosed in the most recent three years for which data is available (2006-2008).

"By not using PSA tests in the vast majority of men, you have to accept you are going to increase very serious metastatic disease threefold," says study leader Edward Messing, MD, chief of urology at the University of Rochester Medical Center.

In 2008, about 8,000 U.S. men were diagnosed with metastatic prostate cancer. By projecting data from the pre-PSA era forward, Messing calculates that without routine PSA tests, 25,000 men would have been diagnosed in 2008 -- an extra 17,000 cases of deadly disease.

But it's not that simple, says Barnett Kramer, MD, MPH, associate director for disease prevention at the National Institutes of Health. Studies looking back at cancer trends in a population are very unreliable when it comes to showing what caused those trends.

For example, Kramer says, what if a man got a PSA test when his prostate cancer was in the very earliest stages of metastatic disease? Microscopic cancers already have seeded his body. But he would not yet have symptoms or detectable metastatic disease, so he'd be diagnosed with earlier-stage disease. He's subtracted forever from the men whose first diagnosis was late-stage disease, even though screening could not save his life.

Messing says this misses the point of his study, which appears in the online July 30 issue of the journal Cancer.

"The reason our study has some meaning is that all a screening test can give you is a shift to lower-stage disease," he says. "It can't cure the disease and it can't prevent the disease. All it can do is allow you to catch it earlier and give appropriate treatment for the stage of disease you have caught it at."

Kramer notes that another study using National Cancer Institute data recently found that late-stage prostate cancers declined in men over age 75 after it was recommended that they stop getting PSA tests. It's highly unlikely that not getting screened prevented these cancers. But it does show how easily false results creep into look-back studies.