Technical Notes: Definitions
Several technical terms are used in presenting the data in this report. Their definitions are presented here to clarify them for the reader.
Incidence rate
The cancer incidence rate is the number of new cancers of a specific site/type occurring in a specified population during a year, usually expressed as the number of cancers per 100,000 persons at risk. That is, Incidence rate = (New cancers / Population) * 100,000.
The numerator of the incidence rate is the number of new cancers; the denominator of the incidence rate is the size of the population. The number of new cancers may include multiple primary cancers occurring in one patient. The primary site reported is the site of origin and not the metastatic site. In general, the incidence rate would not include recurrences. The population used depends on the rate to be calculated. For cancer sites that occur in only one sex, the sex-specific population (e.g., females for cervical cancer) is used.
The incidence rate can be computed for a given type of cancer or for all cancers combined. Except for 5-year age-specific rates, all incidence rates in this report are age-adjusted (see below) to the 2000 US standard population (or, where appropriate, to the world standard million population). (In some previous editions of the CSR, the 1970 US standard million population was used; therefore, incidence rates in this edition cannot be compared to rates published in those editions.) Incidence rates are for invasive cancer only, unless otherwise specified. (Exceptions are the incidence rate for cancer of the urinary bladder (where both in situ and invasive cancers are counted) and breast cancer in situ, which is shown separately.)
Death rate
The cancer death (or mortality) rate is the number of deaths with cancer given as the underlying cause of death occurring in a specified population during a year, usually expressed as the number of deaths due to cancer per 100,000 persons. That is, Death Rate = (Cancer Deaths / Population) * 100,000.
The numerator of the death rate is the number of deaths; the denominator of the death rate is the size of the population. As with the incidence rate, the population used depends on the rate to be calculated. The death rate can be computed for a given cancer site or for all cancers combined. Except for 5-year age-specific rates, all death rates in this report are age-adjusted (see below) to the 2000 US standard million population (or, where appropriate, to the world standard million population). (In some previous editions of the CSR, the 1970 US standard million population was used; therefore, death rates in this edition cannot be compared to rates published in those editions.)
Age distribution
A table showing a partition of the entire lifespan into disjoint age intervals, along with the proportion of the population in each interval.
Median age
The age at which half of a population is younger and half is older.
Standard population
A standard population for a geographic area, such as the US or the world, is a table giving the proportions of the population falling into the age groups 0, 1-4, 5-9, ..., 80-84, and 85+. A standard million population for a geographic area is a table giving the number of persons in each age group 0, 1-4, ... , 85+ out of a theoretical cohort of 1,000,000 persons that is distributed by age in the same proportions as the standard population. Table A-7 shows the US 2000 standard population and the world standard million population. (Some World Health Organization mortality publications use a different world standard million population.)
Age-adjusted rate
An age-adjusted incidence or mortality rate is a weighted average of the age-specific incidence or mortality rates, where the weights are the counts of persons in the corresponding age groups of a standard million population. The potential confounding effect of age is reduced when comparing age-adjusted rates based on the same standard million population. For this report, the 2000 US standard million population (or, where appropriate, the world standard million population) is used in computing age-adjusted rates, unless otherwise noted.
Percent change
The percent change (PC) in a statistic over a given time interval is Percent change = (Final value - Initial value) / Initial value * 100. A positive PC corresponds to an increasing trend, a negative PC to a decreasing trend.
Annual percent change
The annual percent change (APC) is calculated by first fitting a regression line to the natural logarithms of the rates (r) using calendar year (x) as a regressor variable. In this report the method of weighted least squares is used to calculate the regression equation. If ln(r) = mx + b is the resulting regression equation (with slope m), then
APC = 100 * (em - 1). A positive APC corresponds to an increasing trend, a negative APC to a decreasing trend.
Because the methods used in their calculation are mathematically different, the signs of the PC and the APC for a given statistic and time interval may differ, as occurs in a few of the tables presented. That is, one of these statistics may show an increasing trend, the other a decreasing trend.
Testing the hypothesis that the actual mean annual percent change is 0 is equivalent to testing the hypothesis that the theoretical slope estimated by the slope m of the line representing the equation ln(r) = mx + b is 0. The latter hypothesis is tested using the t distribution of m / SEm with n-2 degrees of freedom. The standard error of m, called SEm, is obtained from the fit of the regression (Kleinbaum et al., 1988). (This calculation assumes that the rates increased or decreased at a constant rate over the entire calendar year interval; the validity of this assumption was not assessed.) In those few instances where at least one of the rates was 0, the linear regression was not calculated.
Average Annual Percent Change
The average annual percent change (AAPC) is a summary measure of a trend over a pre-specified fixed interval based on an underlying joinpoint model. It allows us to use a single number to describe the average trend over a period of multiple years. It can be estimated even if the joinpoint model indicates that there were changes in trends during those years, since it is estimated as a weighted average of the joinpoint APCs, with the weights equal to the lengths of each segment over the pre-specified fixed interval.
Life table
A table for a given population listing, for each sex and each age from 0 to 120, how many members die at that age and how many survive one more year.
Observed survival rate
The observed survival rate represents the proportion of cancer patients surviving for a specified time interval after diagnosis. Note that some of those not surviving died of the given cancer and some died of other causes.
Relative survival rate
The relative survival rate is calculated using a procedure (Ederer et al., 1961) whereby the observed survival rate is adjusted for expected mortality. The relative survival rate approximates the likelihood that a patient cohort will not die from causes associated specifically with the given cancer before some specified time after diagnosis. It is always larger than the observed survival rate for the same group of patients.
Standard error
The standard error of a rate is a measure of the sampling variability of the rate.
Person-years of life lost
The person-years of life lost (PYLL) is calculated as follows: For each individual who dies of the cancer of interest, the number of years of expected additional life for an average person of that age, race, and sex is obtained from life tables for the US population (available from the NCHS). The PYLL in the general population associated with a particular cancer for a given year is simply the sum of this expectation over all those individuals who died of that cancer in that year.
Average years of life lost
The average years of life lost (AYLL) associated with a particular cancer for a given year is the PYLL associated with that cancer in the general population divided by the number of deaths from that cancer in the general population in that year.
Prevalence
Prevalence is defined as the number or percent of people alive on a certain date in a population who previously had a diagnosis of the disease. It includes new (incident) and pre-existing cases and is a function of past incidence, past survival, and the size and age structure of the population. Limited-Duration Prevalence represents the proportion of people alive on a certain day who had a diagnosis of the disease within the past x years (e.g. x = 5, 10, or 20 years). Complete prevalence is an estimate of the number of persons (or the proportion of the population) alive on a specified date who had been diagnosed with the given disease, no matter how long ago that diagnosis was. Vist the Statistical Research & Applications Branch website for more details on cancer prevalence definitions and methods.
Stage of disease at diagnosis
Extent-of-disease information determines stage of disease at diagnosis. The SEER summary stage presented has four levels. An invasive neoplasm confined entirely to the organ of origin is said to be localized. A neoplasm that has extended beyond the limits of the organ of origin, either directly into surrounding organs or tissues or into regional lymph nodes, is said to be regional. A neoplasm that has spread to parts of the body remote from the primary tumor, either by direct extension or by discontinuous metastasis, is said to be distant. When information is not sufficient to assign a stage, a neoplasm is said to be unstaged. In situ tumors (except those of the cervix uteri) are also collected by SEER but generally are not published in this series. For some cancers and diagnosis years, the extent of disease information can also be converted to Stages 0-IV as defined by the American Joint Committee on Cancer (Greene et al, 2002).
