Cancer Mortality

Index of Tables

Introduction

Mortality data reflect how many people died of cancer in a given time period, while incidence data reflect how many people developed cancer in a given time period. Mortality data are gathered from death certificates which are collected, coded, and computerized by the Minnesota Center for Health Statistics at the Minnesota Department of Health. A person who has a diagnosed cancer (for example prostate cancer) may die of another cause (for example heart attack). This person's prostate cancer would not be reflected in mortality statistics (since the cancer is not the underlying cause of death), but would be found in incidence statistics. For a given year, cancer incidence and cancer mortality usually do not reflect the experience of the same individuals. People who are diagnosed with cancer and eventually die from it often do so a number of years after diagnosis.

For a variety of reasons, the picture of cancer occurrence developed from mortality data (deaths due to cancer) will differ from the picture developed from incidence data (newly-diagnosed cancers). Specific reasons for this are:

1. Increasing numbers of people survive longer with their cancer. Incidence closely parallels mortality for the types of cancer with poor survival; for cancers with good survival, incidence is considerably higher than mortality. For example, lung cancer incidence and mortality are fairly similar, while breast and cervical cancer incidence rates are much higher than their corresponding mortality rates. As screening becomes improved and more widespread for specific cancers, the differences in incidence and mortality should widen for these cancers.

2. The differences in survival between the various types of cancer will also affect which types of cancer appear more common. Thus, in males prostate cancer is by far the most common type of cancer when new diagnoses (incidence data) are examined, but lung cancer is the most common cause of cancer death (mortality data). This is due to the much higher survival rates for those with prostate cancer than for those with lung cancer.

3. Cancer incidence data are more accurate than mortality data because of how the data are recorded. The physician filling out a death certificate may have little knowledge of the medical history of the person who died. This can result in a nonspecific or even inaccurate cause of death being recorded on the death certificate. Incidence data collected by the MCSS, in contrast, are based on information recorded at the time the diagnosis is made, are checked for accuracy, and are therefore a more accurate measure of cancer occurrence than are mortality data. Graphs depicting cancer incidence and cancer mortality jointly, thereby illustrating the differences in these two measures of cancer occurrence, are presented in Section 4 of this report.

4. The data for incidence and mortality statistics may contain different definitions of "cancer." Cancer incidence data for 1992-1996 include only those cancers that were microscopically confirmed, that is where a tissue specimen was obtained and examined by a pathologist who determined the cells to be malignant. Death certificates also include cancers that were clinically diagnosed but not microscopically confirmed. A few types of cancer - notably pancreatic cancer - are frequently diagnosed without tissue confirmation, so these cancers would not be included in incidence data but would be included in mortality data when and if the person dies of his or her cancer. Therefore, the number of pancreatic cancer deaths for the years 1992 to 1996 was greater than the number of microscopically-confirmed cases in the same time period. The original quality control study (1986) that established the methods for the MCSS documented that 98.6 percent of all microscopically-confirmed cancers were identified by the MCSS and that more than 95 percent of all cancers in Minnesota residents were diagnosed through tissue specimens. A recent independent assessment (1996) demonstrated that MCSS case ascertainment was 99.6 percent complete for microscopically confirmed diagnoses, and 99.1 percent complete for all (including non-microscopically confirmed) diagnoses.

Table 3.1 provides a profile of the statewide age-adjusted mortality rates for specific cancers for the years 1992-1996. The number of cancer deaths (N) in Minnesota for this five-year period is also given. Lung, prostate, and colon cancer for men, and breast, lung, and colon cancer for women were the leading causes of cancer mortality in Minnesotans, accounting for nearly 50 percent of cancer deaths. Overall, males had a 44 percent higher cancer mortality rate than females. Forty-four percent of that excess can be accounted for by higher lung cancer rates, which are due to higher rates of smoking among older generations of males than females.

Lung cancer mortality accounted for 24 percent of cancer deaths in Minnesotans. Lung cancer mortality, which had been increasing over the past four decades in males, is now starting to decline (Section 4). For females, the more recent increases in lung cancer mortality continue. Between 1950 and 1990 breast cancer and leukemia mortality were virtually unchanged; prostate and brain cancer mortality have increased very slightly; and cervical cancer mortality has continued to decrease dramatically due to extensive screening. Since 1992 a significant decline in breast cancer mortality has been documented which is probably the result of years of early detection programs (Section 4). Since 1988 significant declines in colorectal cancer mortality have been seen in both sexes. Likewise for both sexes significant increases in non-Hodgkin's lymphoma continue, a trend which began in 1950. The dramatic decrease in testicular cancer mortality, which began two decades ago as a result of advances in treatment, continues.

Age-Specific Cancer Mortality

Tables 3.2 and 3.3 contain cancer mortality rates for males and females for 18 separate age groups. The strong effect of aging on cancer mortality is demonstrated with these data. Cancers of the prostate and colon are good examples of this relationship. Starting around age 40, mortality rates for these cancers generally increase by 50 to 100 percent every five years. Other cancer mortality rates, such as those for brain and lung, increase substantially through middle age and then level off and even decrease among the very elderly.

Risk of Dying From Cancer

Table 3.4 contains estimates of the number of deaths due to specific cancers expected in 1,000 Minnesotans born today. Because this expression of cancer mortality (lifetime risk of dying of cancer) is widely used in the media and is intuitively easier to understand for the general public, it is important to understand the assumptions used when making these estimates. The idea of lifetime risk of cancer (risk of developing cancer in the lifetime of a Minnesotan who is born today) is covered extensively in Section 2. The methods and assumptions used to estimate these risks are similar to those used to predict how many individuals will die from specific cancers. These assumptions will only briefly be reiterated here. The estimates assume that (1) people born today will have the same life expectancy that currently exists in Minnesota (1996), (2) that people born today will experience the same age-specific cancer mortality rates that currently exist in Minnesota (1996).

Using the above-stated assumptions for Minnesotans born today, 245 of every 1,000 males and 215 of every 1,000 females are expected to die of cancer. Therefore, an estimated one-fifth to one-quarter of Minnesotans born today will die of cancer. This is in contrast to nearly double that number who will be diagnosed with cancer of any kind at some point in their lives. Cancers of the lung, prostate, and colon account for half (52 percent) of expected cancer deaths in males; cancers of the lung, breast, and colon account for nearly half (47 percent) of expected cancer deaths in females. While cancer mortality rates were noted to be 44 percent higher for males than females (Table 3.1), the lifetime risk of dying of cancer is only 14 percent higher for males than for females. This is due to the longer life expectancy of females.

Childhood Cancer Mortality

While cancer incidence in children is rare, the disease is still the second leading cause of childhood mortality in the United States. During the five years 1992-1996, 122 deaths from cancer were recorded in Minnesota children younger than 15 years of age, for an age-adjusted annual mortality rate of 2.4 per 100,000 (Table 3.5). Cancer mortality in Minnesota children was 17 percent lower than the national rate, which was 2.9 deaths per 100,000. For each of the major childhood cancers, Minnesota death rates are equal to or lower than the U.S. death rates. Both in Minnesota and nationally, acute lymphocytic leukemia (ALL) was the most common type of newly-diagnosed childhood cancer. In contrast, mortality rates from brain cancers were two times higher than for ALL in Minnesota children, and 1.6 times higher than ALL mortality rates in this age group nationally. This is due to the poorer prognosis of brain cancers relative to that of ALL. In general, though, survival rates for many childhood cancers have improved during the past several decades. The National Cancer Institute reports that the five-year survival rate in this age group for all cancers combined jumped from just over 55 percent in 1974-1976 to almost 74 percent during 1989-1996.

Although cancer incidence in Minnesota children was slightly higher (Section 2F) than national estimates based on SEER data, cancer mortality was approximately 17 percent lower than the nation as a whole. This differential might be explained by better access to medical care or greater success in treatment in Minnesota children, or - just as likely - by the SEER incidence data (based on 10 percent of the U.S. population) failing to represent the true national incidence of pediatric cancers.