America isn't winning the war on cancer after all.

Contrary to optimistic reports from the National Cancer Institute showing the incidence of several devastating cancers has leveled off or even declined in recent years, rates for at least some of those cancers has been rising, according to a new analysis by NCI scientists.

Previous indications of a decline reflected significant delays in reporting cancer cases, the researchers report Wednesday in the Journal of the National Cancer Institute. More accurate information about cancer rates presents a grimmer picture.

"Maybe we were a little too eager to declare the effectiveness of our intervention and prevention programs," says Brenda Edwards, who is associate director for the surveillance research program at NCI, of Bethesda, Md., but wasn't among the authors of the new study.

The revised estimates present a dispiriting picture of the nation's progress in preventing cancer. Breast-cancer rates in white women had been almost flat since 1987, according to the original NCI figures, which the American Cancer Society also uses as the basis for the popular "facts and figures" on its Web site.

The reanalysis shows that breast-cancer rates actually have been rising 0.6% a year since 1987. That prompted the NCI scientists to call for research "to explain the cause for the recent rise in breast cancer incidence."

Lung cancer in women also had been believed to be flat; the re-analysis shows it has been rising 1.2% a year since 1996. Melanoma rates in white males had reportedly been flat or even falling. The new analysis finds it has been soaring 4.1% a year since 1981, suggesting that prevention strategies that focus on staying out of the sun are falling short. Prostate-cancer rates in white males, rather than falling since 1995, have in fact been rising 2.2% a year. For white men, 1998 prostate-cancer rates are actually 12% higher than originally reported; for black men they are 14% higher.

Colorectal cancer cases for both genders and all races are 3% higher than first reported, suggesting that early-screening techniques (which focus on discovering precancerous polyps through colonoscopies) aren't as powerful or widely used as hoped. The rate of colorectal cancer in white women, for instance, has been rising 2.8% annually since 1996, rather than the originally calculated 0.9%.

National incidence data are based on reports from 10 registries in the SEER (Surveillance, Epidemiology, and End Results) program at NCI, which samples 14% of the U.S. population by collecting cancer reports from hospitals, doctors and clinics. The registries have 19 months to report cases to NCI.

Scientists had long suspected that the original numbers were skewed. "It was well known that reports of new cancer cases dribbled in over the years, long after the 19-month reporting deadline," says Benjamin Hankey, the senior author of the study. So, researchers wondered, just how sharply did late reporting affect the final cancer-rate statistic for a specific year?

Using data from 1981 to 1998, scientists led by Mr. Hankey analyzed reporting delays by counting how many additions nine registries made to their original count over the years. Based on that, but allowing for improvement in the timeliness and accuracy of the reports, NCI statistician Limin Clegg estimated the under-reports from each registry for five types of cancer. The delays are such that initial reports account for only 88% to 97% of the actual cancer cases, depending on the type, finds Dr. Clegg. That has left a "false impression of a recent decline in cancer incidence," write the NCI scientists.

NCI's cancer-incidence rates are the basis for decisions by policy makers and clinicians alike: The numbers are used to allocate research and clinical resources, to give people a sense of their risk for various cancers and to offer hints about environmental causes of cancer ranging from use of sunblock to changes in diet and cumulative exposure to toxic chemicals.

Now researchers feel a renewed urgency to study why the rates of several cancers are still on the rise. "This tells us something we didn't know about whether our intervention and prevention programs are working," says Ahmedin Jemal, director of the surveillance program for the American Cancer Society.

Write to Sharon Begley at sharon.begley@wsj.com⁸

Updated October 16, 2002 2:58 p.m. EDT