Health at a Glance
Updating the standard population and its effect on cancer incidence and mortality rates Health at a Glance
Updating the standard population and its effect on cancer incidence and mortality rates

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

by Lawrence Ellison

Release date: October 20, 2016

More information PDF version

On this page

• Highlights
• Introduction
• Cancer incidence and mortality rates increase with age
• Canadian population older in 2011 than 1991
• Overall age-standardized cancer incidence rates higher with new (2011) standard population
• Overall age-standardized cancer mortality rates higher with new (2011) standard population
• Changes in age-standardized rates greater for mortality than incidence
• Impact of updating standard population varies by cancer type
• Conclusion
• Data sources, methods and definitions
• References and notes
• Related material for this article

Start of text box 1

End of text box 1

Introduction

Rates of newly diagnosed cases of cancer (incidence) and deaths attributed to cancer (mortality) may be presented as either unstandardized (crude) or age-standardized rates. Crude cancer incidence rates provide a measure of the risk of being diagnosed with cancer. For example, the crude cancer incidence rate for all ages combined was 520.5 per 100,000 persons in Canada in 2012. Crude cancer mortality rates provide information on the risk of death attributable to cancer. The crude cancer mortality rate for all ages in 2012 was 214.0 per 100,000 persons.

Quite often there is interest in comparing cancer rates in a population over time, across populations from different geographic areas, or across populations defined by both time and geography. In these circumstances, the use of the crude rate for all ages combined is not recommended. This is because the chance of being diagnosed with, or dying from, cancer, varies with age, and different populations may have different age-structures.^{Note 1} The age-structure of the Canadian population, for example, has been changing over time as the proportion of older people has been increasing.^{Note 2}

What is desired is a measure that permits a “fair” comparison of the risk of being diagnosed with, or dying from, cancer, across different populations. The age-standardized rate provides such a measure by providing the rate that would have been observed in each population of interest if each of these populations had the same age-structure (a common standard population). It allows us to compare apples to apples.

In March 2016, the standard population used to derive age-standardized rates in Canada was updated to one based on the 2011 Canadian population.^{Note 3Note 4} Previously, it had been based on the 1991 Canadian population.^{Note 4} As a result of this change, age-standardized cancer incidence and mortality rates for each year have also changed—often substantially. The changes in age-standardized rates that accompany the change in standard population are, however, artificial in nature, and do not imply that there has been a change in the risk of being diagnosed with, or dying from, cancer. Nonetheless, the changes are worth exploring as they may vary by province or cancer type.

This article examines differences in age-standardized cancer incidence and mortality rates resulting from a change in the standard population used to derive them. Differences are examined over time, and by province and cancer type for 2012. The data sources are the Canadian Cancer Registry (1992 to 2012), the Canadian Vital Statistics – Death database (1992 to 2012) and Canada’s population estimates by age and sex, (1992 to 2012) all from Statistics Canada.

Cancer incidence and mortality rates increase with age

For all cancers combined, incidence and mortality rates increased with age^{Note 5} (Chart 1). In 2012, increases in cancer incidence rates with age were largest over the range between the 15 to 19 and 60 to 64 age groups. While the rate of increase declined in older age groups, the incidence rate itself continued to rise until it reached a peak in the 85 to 89 age group. During the same year, rates of increase in cancer mortality rates with age were largest over the range between the 25 to 29 and 55 to 59 age groups. Though the rate of increase declined thereafter, the mortality rate itself continued to rise through to the 90 and older age group.

Canadian population older in 2011 than 1991

The age-structure of the Canadian population in 2011 was older than it was in 1991 (Chart 2). For every age group from 45 to 49 onwards, the percentage of the population in a particular age group was greater in 2011 than in 1991. The absolute difference was largest among those 50 to 54. For every age group up to and including those aged 40 to 44, the percentage of the population in a particular age group was smaller in 2011. In this case the difference was largest in the 30 to 34 age group.

Overall age-standardized cancer incidence rates higher with new (2011) standard population

Age-standardized incidence rates based on the 2011 standard population are higher than those based on the 1991 standard (Chart 3). This does not mean that there has been a sudden increase in the risk of developing cancer. Rather, the increase is artificial in nature and is explained by a combination of two factors. First, cancer incidence rates increase with age (see Chart 1). Second, rates for age groups comprised of older people are more heavily weighted (that is, given greater emphasis) with the 2011 standard than with the 1991 standard as the age-structure of the 2011 population is older than that of 1991 (see Chart 2).^{Note 4} While the rates are higher, the trend in age-standardized cancer incidence rates is nearly the same whether age-standardization is performed using the 2011 or 1991 population standard.

The solid line in Chart 3 represents the crude incidence rate due to cancer. The line indicates that the crude rate increased 26.9% from 1992 to 2012.^{Note 6} However, once the aging of the population that occurred during this time period is adjusted for by age-standardization, a decline in the rate of cancer incidence of 2.6% is seen.^{Note 7}

Crude rates for the most recent data years are more similar to the corresponding age-standardized rates for those years when the newer, rather than the older, standard population is used. The reverse is true for the earliest years. This highlights the primary reason for switching to a standard population based on more recent population data. That is, to bring age-standardized rates more in line with the crude rates for the most recent data years.

Overall age-standardized cancer mortality rates higher with new (2011) standard population

Age-standardized mortality rates based on the 2011 standard population are higher than those based on the 1991 standard (Chart 4). This was observed above with the cancer incidence and the explanation is the same—cancer mortality rates increase with age (see Chart 1) and the older age groups are more heavily weighted (that is, given greater emphasis) with the 2011 standard (see Chart 2).^{Note 4} The trend in age-standardized cancer mortality rates is nearly the same whether age-standardization is performed using the 2011 or 1991 population standard. This is similar to what was observed above with the corresponding trend in cancer incidence.

The solid line in Chart 4 represents the crude mortality rate due to cancer. The line indicates that the crude rate increased 10.7% from 1992 to 2012.^{Note 8} However, once the aging of the population that occurred during this time period is adjusted for by age-standardization, the data indicate that there has actually been a decline in the rate of cancer mortality of 19.0%.^{Note 7}

Once again, the crude rates for more recent data years more closely resemble the corresponding age-standardized rates for those years when the newer standard population is used. In fact, the crude and age-standardized cancer mortality rates are identical for 2011.

Start of text box 2

End of text box 2

Changes in age-standardized rates greater for mortality than incidence

For all cancers combined, updating the standard population to one based on 2011 rather than 1991 data resulted in the 2012 Canadian age-standardized incidence rate changing from 395.3 to 512.2 per 100,000, a 29.6% increase (Chart 5). Provincially, increases ranged from 28.5% for Prince Edward Island to 31.1% for Quebec.^{Note 11} As the changes are not the same across the provinces, comparisons of provincial age-standardized rates for all cancers combined may, in theory, be affected by the change in the standard population used. In practice, however, any discrepancies are very slight and not meaningful. (See text box entitled “Age-standardization in practice”).

Updating the standard population also resulted in a 36.2% change in the age-standardized mortality rate, from 154.2 to 210.0 per 100,000, in 2012. Increases were largest for British Columbia (36.4%), Ontario (36.4%) and Quebec (36.3%), and smallest for Prince Edward Island (33.0%).

Increases in age-standardized cancer rates for 2012 that resulted from the update in the standard population used were greater for mortality than for incidence for every province. The magnitude of the difference was highest in Ontario (7.7 percentage points) and lowest in Quebec (5.3). The consistently larger increases for mortality may be explained by the greater proportion of cancer deaths, relative to the proportion of newly diagnosed cases, in older ages (see Chart 1).

Impact of updating standard population varies by cancer type

The impact of updating the standard population from 1991 to 2011 on age-standardized incidence rates varied considerably by cancer type (Table 1). For over two-thirds of the cancer types examined, age-standardized incidence rates were at least 28% higher using the 2011 standard. The cancers associated with the largest increases were esophageal (34.3%), pancreatic (33.8%), bladder (33.6%) and multiple myeloma (33.6%).^{Note 12} In contrast, the updated age-standardized rate for testicular cancer was 11.5% lower while for Hodgkin lymphoma it was virtually unchanged. Increases were also relatively low for cervical cancer (7.4%), thyroid cancer (12.0%), and cancers of the brain and other nervous system (16.9%).

The magnitude of the change in cancer-specific age-standardized incidence rates resulting from the use of the updated population standard was associated with the median age at diagnosis. Specifically, larger increases tended to be associated with cancers with a relatively high median age at diagnosis while smaller increases and declines were associated with cancers having a lower median age. These results are as one might expect given the greater weighting of older age groups in the updated standard.^{Note 9}

Conclusion

Age-standardization allows for comparisons of incidence or mortality rates over time that take into consideration changes in the age distribution of the population. The standard population used to age-standardize cancer incidence and cancer mortality rates has recently been updated to one based on the 2011, rather than 1991, Canadian population age structure. As a direct result, age-standardized cancer rates are now higher for each year of reported data for all cancers combined and for most, but not all, individual cancers. By definition, crude rates are unaffected.

While differences in the magnitude of the change are interesting to examine by province or by cancer type, it is stressed that the changes are artificial in nature and do not represent a change in the risk of developing or dying from cancer as such. The main advantage of the update to the standard population is that age-standardized rates for recent years now more closely reflect actual (crude) rates. While age-standardized rates are a hypothetical construct, they are sometimes mistakenly interpreted as reflecting actual risk. Use of the updated standard will lessen the extent of any such misinterpretation for recent years.

Start of text box 3

End of text box 3

Start of text box 4