An estimated 182,800 U.S. women will be diagnosed with breast cancer this year, of whom about 40,800 will die. Every time I come across these statistics, I add this missing number: at least 32,000 women will be treated for a cancer that would never have killed them. This is the result of overdiagnosis, an underappreciated byproduct of mammography screening.

In fact, the risk of overdiagnosis accompanies any screening procedure. Subject a large group of symptomless people to mammography, computed tomographic (CT) scanning, or Pap smears, and then biopsy the tiny abnormalities these tests identify-in many cases, the cells will appear to be cancerous under the microscope. But most of these would never become life-threatening, even if left untreated.

Widespread acceptance of mammography screening has caused a dramatic increase in the diagnosis of ductal carcinoma in situ (DCIS), which usually shows up as microcalcifications on a mammogram. Before the advent of mammography screening, this microscopic lesion within the milk duct was rarely seen beyond the autopsy table. According to the American Cancer Society, 39,900 women are now diagnosed with DCIS annually, and research indicates that about 80% of all DCIS (approximately 32,000 cases) will never become invasive, even if left untreated. But DCIS was routinely treated with radical mastectomy in the 1970s, when mammography screening first became available. Surgeons told women, "Be grateful your cancer was found early. Your life has been saved."

Providers now recognize that radical mastectomy for the treatment of DCIS is therapeutic overkill. But overtreatment and uncertainties persist; many lesions are currently treated with lumpectomy plus radiation or simple mastectomy. No test can accurately distinguish the DCIS that would become invasive; what's more, "invasive" is not necessarily synonymous with "fatal." According to the results of a large, ongoing clinical trial from the National Surgical Adjuvant Breast and Bowel Project, invasive breast cancer developed in a small percentage of women whose DCIS was treated either with excision plus radiation or excision alone. Thus far, eight-year results show a breast cancer mortality rate of only 1% in each group. This is the same rate of mortality demonstrated in older, much longer follow-up studies of women whose DCIS was treated with mastectomy. And 13-year follow-up results from the Canadian National Breast Screening Study indicate that early detection of DCIS provides no advantage.

Pick a body part-lung, prostate, cervix, thyroid. Look hard enough, and you'll find a "precancerous" abnormality. In the not-so-distant past, precancerous cervical lesions were almost always treated with hysterectomy because it was assumed they would eventually become malignant. Now it's known that nine out of 10 lesions regress spontaneously. And widely publicized findings of a study by Henschke and colleagues, published in the July 10, 1999, issue of the Lancet, showed increased lung cancer survival as a result of screening with the spiral CT scan. But this study didn't prove a reduced rate of lung cancer mortality, which is the ultimate test of a screening tool's value. Some physicians caution against the premature acceptance of lung scanning, because it may lead to unnecessary surgery; furthermore, survival rates may be artificially inflated by the inclusion of scan-detected cancers that would not have become invasive.

Cancer always kills, we are told, and early detection virtually guarantees cure. These simplistic half-truths-many of them emanating from the American Cancer Society-are at odds with research showing cancer to be heterogeneous, encompassing a broad spectrum of diseases that range from permanently noninvasive to rapidly fatal.

Screening does save lives, but at a far more modest rate than the public has been led to believe. Whenever you find yourself telling patients to be screened, don't forget to give them the whole story.