Part of the usefulness of the census that the federal government undertakes every decade is the ability to see how the populations of the individual states and the population of United States as a whole have changed. The government uses the information generated by the census to decide, among other issues, how much representation each state should have in Congress and how much federal funding each state should receive for infrastructure, health, and welfare programs.

Epidemiologists also use these numbers to track the US population along with diseases and health issues affecting it. In recent years, the census has told us there are increasing number of minorities in certain states, that the country as a whole is aging, and that certain states are losing people, while others are gaining.

Once epidemiologists know the raw numbers of who lives where and some facts about the individuals, such as ethnicity, race, and age, they need to be able to compare different states or regions.

Comparisons cannot be done on raw, or crude, numbers of cancer cases or deaths because the populations may not be comparable with respect to age. Much like finding the common denominator when working with fractions, epidemiologists must calculate age-adjusted incidence or death rates so that the populations of different states or regions are similar enough for comparison.

For example, consider two states, Florida and Alaska. Florida has a relatively old population and Alaska has a relatively young population. If you look only at each state’s crude cancer death rate (total number of cancer deaths divided by the total population) it appears as if Florida has a cancer death rate almost three times that of Alaska. That is because the population of Florida is older and the risk of most cancers increase with age. The age difference makes the overall cancer death rate in Florida appear higher than in Alaska.

Therefore, epidemiologists use a strategy called age-adjustment so that the rates of different states or regions can be compared among people of similar age. Indeed, when the cancer death rates for Florida and Alaska are age-adjusted, they are almost identical. This means that cancer mortality rates are actually similar in the two states, not separated by a threefold difference that appeared at first glance. Getting to the truth in the numbers is important for understanding risk and for planning programs and allocating resources appropriately.

Up until this year, epidemiologists used the a projection of the standard populations from 1970 or even the 1940 to standardize the numbers. Starting with next year’s Cancer Facts and Figures (2003), and carrying over into all of the Society’s publications, epidemiologists will begin to use a projection of the 2000 population as the standard.

This change will occur during 2003 as new data become available. The shift in age adjustment to the 2000 population is an important change that will continue to allow comparisons over time and among different parts of the population, but it will more accurately reflect the US population as it is now, not as it was 30 or more years ago.

Overall, the new calculations will show that the age-adjusted rates of cancer are higher than previously reported when we used the 1970 standard population. This is because Americans are living longer and more people are surviving into the ages where cancer becomes more common. Longevity has increased because of the decline in infectious diseases, such as polio, tuberculosis, and childhood infections, and the more recent decrease in death rates from heart disease and stroke. Additionally, some of the greatest successes we have had in cancer control are with younger people. Some childhood cancers, which were uniformly fatal 30 years ago, are now curable 75% of the time. But cancers of the lung, breast, prostate, and colorectum still largely occur among older people.

So, as our statistics are adjusted for the 2000 population, a few things will appear to occur:

• The age-adjusted incidence and death rates from all cancers combined will increase by about 20%, even though the actual risk of being diagnosed with or dying from cancer will not change at any given age;
• The new approach will increase the age-adjusted rate of cancers that occur at older ages, and decrease the rate of childhood cancers;
• Some major improvements in cancer, which have occurred in young and middle- aged people, will be diluted;
• Racial and ethnic differences will be amplified in some cases and diminished in others.

Some specific examples of the changes that will result from adjusting to a new age standard are as follows:

• Cancer incidence rates will increase from about 18% to 21%;
• The median (halfway) increase for all cancer registries will show an approximately 20% increase;
• There will be a statistical 25% increase in pancreatic cancer rates, a 22% increase in prostate cancer rates, a 21% increase in rectal cancer, and an 18% increase in female breast cancer rates.

Another example of the change can be found by recalculating the 1973 cancer mortality rate. Using the 1970 population, the rate was 150 per 100,000. Adjusting for the 2000 population, the rate rises to over 200 per 100,000. This is the type of change that will appear in the all American Cancer Society documents, including internal documents, such as our 2015 goals.

It is important to note that the recalculation using the 2000 population does not indicate changes in disease incidence or death – only a new system for reporting them. The change will affect only the age-adjusted rates, not the number of cases or deaths or the rate at any given age.

The new age-adjustment will also affect the rates of cancer incidence and mortality on different races. For example, cancer mortality rates for whites will change from 127 per 100,000 people to 208 per 100,000 because of changing the standard from 1940 to 2000. For African Americans, the rate will change from 173 per 100,000 to 273 per 100,000.

An example that shows how rates can change with the changing population standard is this: during the period 1992-1996, cancer mortality rates for blacks were 36%, 33%, or 31% higher than whites, depending on whether the rates were adjusted to the 1940, 1970, or 2000 population standard, respectively.

Similarly, depending on the population standard chosen, whites had a 57%, 59%, or 63% higher mortality rate during this same period compared to Native Americans.

Explaining the difference in age-adjustment is an important, if difficult, process. No one is trying to manipulate statistics for any gain. Everyone, including the federal government and non-profit health care organizations, is using new information to make their statistics more relevant to the population of today. However, the public could easily become alarmed about the increases in cancer incidence and mortality that will appear in our documents, even though the cancer death rates have declined during the 1990s. Similarly, the public may be confused by the shrinking differences in rates between whites and African Americans, for example. Following are some talking points to consider if asked about this change:

About Age-Adjusting:

• Epidemiologist use age-adjustment in order to compare cancer incidence and mortality rates across different states or regions. Crude rates show the number of cases of cancer or cancer deaths in a certain population, but cannot be used to compare different populations due to the many differences in the populations. Other adjustments epidemiologists may make include sex, race, and perhaps occupation, when comparing cancer rates.

About Changing the Standard Population:

• The standard population that epidemiologist will use for age adjusting purposes is the 2000 population, derived from the a projection of the 2000 population.
• Changing the standard population to which epidemiologist adjust their numbers does not mean a change in cancer risk at any particular age, but only that people are living longer so that the average rate in the population is higher.
• The updated standard more accurately reflects the aging population of the US and provides more realistic statistics.
• This change will bring all government branches and non-governmental organizations into alignment with regard to their published cancer statistics. Previously, for example, the National Center for Health Statistics has been using the 1940 US population and the National Cancer Institute has been using the 1970 population as their standards. Switching to the 2000 population as the standard will allow comparison between documents and better communication.

About Timing:

• Starting with calculations of the deaths from cancer in 1999, the federal government and the American Cancer Society will switch to the 2000 standard population. This is derived from a projection of the 2000 population.
• This change will not occur all at once. Starting with Cancer Facts and Figures 2003, many American Cancer Society charts and graphs will use the 2000 standard population. However, some statistics, which are derived from branches of the federal government, may still use earlier standard populations. The standard populations for all statistics will be noted.

In summary, this is an important change to our statistics. What is sensible and accurate on the part of epidemiologists may be easily misconstrued by people trying to compare documents from this year with those of the past.

May 3, 2002