Tumors Included in Prevalence Estimates
When estimating prevalence there are several methods for determining which tumors to include, depending on the question
of interest.
First Malignant Primary Only (non-reported assumed to
be malignant) - this method would include
only the first malignant tumor. The SEER registries
collect the number (but not the site or behavior) of cancers that
occur prior to the start of the registry, or prior to the person
moving to a SEER catchment area. We make the assumption that these
cancers are malignant, as is true of the majority of SEER cancers.
Thus, if SEER indicates that the first SEER registered tumor is
the person's second tumor (the other was a non-SEER cancer) this
person's cancers are excluded from this method. This is the standard
method used to calculate prevalence statistics at the NCI. It
only counts a person once, and is useful if the user would like
to sum various prevalence estimates across cancer sites without
double counting individuals (i.e., the prevalence of specific cancer
sites adds up to the prevalence of all cancers combined). Prevalence
statistics for different tumor inclusion criteria are available
at Table
I-17 of the Cancer Statistics Review, 1975-2006 (PDF).
This method can be implemented in SEER*Stat by setting the Multiple Primary Selection option on the Selection tab to First Malignant Primary Only (Non-reported Assumed Malignant).
First Malignant Tumor per Site in the Last x Years (observation period) - this method includes the first
malignant tumor per cancer site diagnosed in SEER during the entire observation period. If 20-year prevalence on
January 1, 1999 is being calculated, only the first malignant tumor per cancer site diagnosed from 1979 through 1998 would
contribute to prevalence. For this method, the sum of the prevalence counts across specific sites may be greater than the
prevalence for all sites combined. This is true because a person with multiple cancers may contribute to more than one
estimate.
This method can be implemented in SEER*Stat by following these key steps (only the steps to implement the method are included,
proper settings for the survival corhorts and other options are required to generate meaningful estimates):
- Select malignant cases in the "Other (Case files) selection box" on the Selection tab.
- Set the Multiple Primary Selection option to All Tumors Matching Selection Criteria on the Selection tab.
- Use site recode as a table variable.
First Malignant Tumor per Site and Years since Diagnosis - this method includes the first malignant tumor
diagnosed in SEER for each considered site and each period since diagnosis. The sum of the prevalence counts across
specific sites and across non-overlapping periods since diagnosis may be greater than the prevalence
for all sites combined and for the total period, respectively. This is because a person with multiple
cancers may contribute to more than one site-specific prevalence and to more than one period since
diagnosis for the same site.
This method can be implemented in SEER*Stat by following these key steps (only the steps to implement the method are included,
proper settings for the survival corhorts and other options are required to generate meaningful estimates):
- Select malignant cases in the "Other (Case files) selection box" on the Selection tab.
- Set the Multiple Primary Selection option to All Tumors Matching Selection Criteria on the Selection tab.
- Use site recode as a table variable.
- Select Display By Time Prior to Prevalence Date on the Statistic tab and
- Enter 5,10,15 in the Cut-Points (Number of Years Prior) edit box.
- Select Discrete (Non-Overlapping) Intervals.
- Select First Cancer Per Interval (Intervals May Not Sum to Total).
Examples
The following examples illustrate the situations, using the 3 methods, in which a person would be included in the prevalence
estimates. Suppose we are estimating 20-year prevalence at January 1, 1999.
- A person had a malignant melanoma in 1980 and a malignant breast cancer in 1990.
- First Malignant Tumor Only: That
person's first cancer would contribute to the prevalence of
melanoma and all sites prevalence, but the second cancer would
not contribute to the prevalence of breast cancer.
- First Malignant Tumor per Site in the Last x Years (observation period): The
melanoma would contribute to the melanoma prevalence and, since it was the earlier diagnosis, it would contribute
to the all sites cancer prevalence.
- First Malignant Tumor per Site and Years since Diagnosis: The melanoma would
contribute to the melanoma prevalence and, since it was the earlier diagnosis, it would contribute to the all
sites cancer prevalence. This person's breast cancer would only contribute to the breast cancer prevalence.
- A person had a malignant breast cancer in 1980 and another malignant breast cancer in 1990.
- First Malignant Tumor Only: If prevalence is computed on January 1, 1997, the
person's second cancer would not contribute to the 10-year limited-duration prevalence
of malignant breast cancer. The person's first cancer would contribute to the 15 to < 20 year and the 0 to < 20
year limited-duration malignant breast cancer prevalence.
- First Malignant Tumor per Site in the Last x Years (observation period): If
prevalence is computed for January 1, 1997, then the person's second cancer would contribute to the 10-year limited-duration
prevalence of malignant breast cancer. The person's first cancer would contribute to the 15 to < 20 year limited-duration
prevalence of malignant breast cancer. Only the first cancer would contribute to the 0 to < 20 year limited-duration
prevalence of malignant breast cancer.
- First Malignant Tumor per Site and Years since Diagnosis: If prevalence is computed
in January 1, 1999, then the person's second cancer would contribute to the 5 to < 10 year limited-duration prevalence
of malignant breast cancer. The person's first cancer would contribute to the 15 to < 20 year limited-duration
prevalence of malignant breast cancer. Only the first cancer would contribute to the 0 to < 20 year limited-duration
prevalence of malignant breast cancer.
- A person had a malignant breast cancer in 1990. Although this is the first cancer they had in a
SEER registry, it is indicated that this is their second cancer (the other was a non-SEER cancer). This person would
not be included for the prevalence of malignant breast cancer because we assume that the non-SEER cancer was malignant.
Since this cancer occurs prior to SEER, we do not know the type of cancer or when it occurred.
- First Malignant Tumor Only: Although this is the first cancer they had in a
SEER registry, it is indicated that this is their second cancer (the other was a non-SEER cancer). This person
would not be included for the prevalence of malignant breast cancer because we assume that the non-SEER cancer
was malignant. Since this cancer occurs prior to SEER, we do not know the type of cancer or when it occurred.
- First Malignant Tumor per Site in the Last x Years (observation period): Although
this is the first cancer they had in a SEER registry, it is indicated that this is their second cancer (the other
was a non-SEER cancer). This person would be included for the prevalence of malignant breast cancer because we
do not consider the pre-SEER cancer when using this method.
|