Using data from the National Center for Health Statistics (NCHS), John C. Bailar III, M.D., Ph.D., and Heather L. Gornik calculated that after rising for many years, the cancer mortality rate for all cancer sites combined fell 1 percent between 1991 and 1994. NCI calculated last year that the rate fell 2.6 percent between 1991 and 1995.
The difference in these numbers is based on two factors: Bailar and Gornik extended their estimate only to 1994, the most recent year of complete NCHS mortality statistics, while NCI statisticians used preliminary 1995 data from NCHS to extend the estimates to that year, when a continued decline was observed. Also, Bailar and Gornik age-adjusted the rates to the 1990 U.S. population, while NCI, according to its standard practice, used the 1970 population.
When calculating a rate of disease incidence or death for an entire population (all ages combined), the rate is normally age-adjusted to a certain year to compensate for the changing age patterns of the population. The observed or "crude" rate is adjusted to reflect what the rate would be if the age distribution had remained unchanged (for example, the same proportion of people under age 35, over age 65, etc.). Owing to its progressively longer life expectancy, the U.S. population is steadily aging. Because many more older people than younger people die of cancer, the (unadjusted) rates for all ages combined could rise even if the rates for each age group have gone down.
The choice of a "standard population" for age adjustment -- the population age distribution in a particular census year -- is not dictated by any rigid scientific rules. In many cases, historical or geographical context guides usage. For instance, NCHS age-adjusts its health statistics -- including cancer mortality rates -- to 1940 because that was the most recent census year when the agency published in 1943 a major comprehensive mortality analysis covering the years 1900 to 1940. For international comparisons, NCI and many other agencies use the "world standard population," which has a much younger age distribution than the U.S. population.
For its routine data reporting, NCI uses data from NCHS, but age-adjusts the rates to 1970 because the National Cancer Program began in 1971 and NCI's Surveillance, Epidemiology, and End Results (SEER) Program, which compiles cancer incidence data, began in 1973. All cancer rates reported by NCI since then have been age-adjusted to 1970, to allow direct comparison of rates over the entire time period covered by SEER. By contrast, Bailar and Gornik chose 1990, the census year closest in time to the recent decline in mortality.
Cancer rates and trends (how much rates are going up or down) will vary depending on the year chosen as a standard for age adjustment. With regard to the overall cancer mortality trend in the 1990s, the earlier the year chosen for age adjustment, the greater the decline will appear. This is because younger people have benefitted the most from cancer research advances, and the 1940 U.S. population, for instance, was much younger than that of 1970, which in turn was younger than that of 1990. Nevertheless, the downward direction of the mortality trend is still clear regardless of the age adjustment used.
The agencies of the U.S. Department of Health and Human Services continue to discuss a possible update of the standard population used in age adjustment and the potential use of a single standard for all agencies, such as the year 2000 population when census data for that year becomes available.
Bailar and Gornik state that the downturn in cancer mortality rates is likely to continue, and NCI agrees with that assessment. The authors attribute much of the downturn to improved prevention and early detection, particularly past reductions in tobacco use.
NCI funds and conducts a broad range of research in early cancer detection, including studies in breast, prostate, lung, colorectal, ovarian, and cervical cancers. The institute believes that a more enlightened use of chemotherapy combined with improved surgical and radiation treatments for several cancers is also improving outcomes. Naturally, treatment must be effective for early detection to be of value.
NCI has strongly supported recent initiatives to avert the initiation of tobacco use among children and teenagers, and continues to develop a variety of approaches to cessation among those already addicted. The institute also funds programs to promote healthy eating habits and to gain a deeper understanding of the dietary factors that may affect cancer risk.
Cancer prevention is a broad field encompassing basic laboratory research into the causes and biology of cancer, identification and modification of environmental and genetic risk factors, development of cancer-preventing drugs (chemoprevention), and behavioral science aimed at increasing healthy behaviors. NCI supports major research programs in each of these areas.
* The article is titled, "Cancer Undefeated." The authors are John C. Bailar III, and Heather L. Gornik. New England Journal of Medicine 336, May 28, 1997.
Cancer Information from the Office of Cancer Communications
National Cancer Institute news releases are available via the Internet through the World Wide Web (http://rex.nci.nih.gov). Click on Mass Media.
Cancer Information Service
The Cancer Information Service (CIS), a national information and education network, is a free public service of the National Cancer Institute (NCI), the nation's primary agency for cancer research. The CIS meets the information needs of patients, the public, and health professionals. Specially trained staff provide the latest scientific information in understandable language. CIS staff answer questions in English and Spanish and distribute NCI materials.
For NCI information by fax, dial 301-402-5874 from the telephone on a fax machine and listen to
For NCI information by computer:
CancerNet Mail Service (via E-mail):
To obtain a contents list, send E-mail to firstname.lastname@example.org with the word
"help" in the body of the message.
CancerNet is also accessible via the Internet through the World Wide Web at
(http://cancernet.nci.nih.gov) and Gopher (gopher://gopher.nih.gov) servers.