Person Selection Using SEER*Stat
Each record in a SEER*Stat incidence database represents one tumor. Therefore, it is possible for a person to have more than one record in a database. SEER*Stat allows you to select records for an individual within Frequency and Case Listing sessions. To do so:
- Turn Person Selection ON from the Session Menu.
- Specify the selection criteria for the index record (if a selection statement is blank then all records meet the criteria). Only individuals with a tumor matching the index record criteria will be considered in the session. The first tumor (based on diagnosis sequence) matching the index record will be used as the index record for that individual.
- You may narrow your selection further. That is, you may want to select
people only if they have a tumor matching the index record and:
- Do not have any other tumors
- Have at least one other tumor diagnosed at any time, at a time prior to the index record, or at a time after the index record. You may define criteria for the other tumors (for example, you could specify that the other tumor is lung cancer).
In Case Listing Sessions, all records for a selected person are always included in the table. In Frequency Sessions, you may choose whether to include all records for a selected person, only the index record, or all records other than the index record.
Multiple Primary Ready Databases
Person Selection can only be used with databases that are "multiple primary ready", that is, databases in which the records are grouped by person and are in order of diagnosis. A database is multiple primary ready if
- each record represents a tumor or case, and
- the records are sorted such that a person's records are grouped together and are in chronological order. That is, the database must be sorted by patient identifiers and a variable such as date of diagnosis or sequence number. For example, the SEER incidence databases are sorted by SEER registry, Patient ID, and Sequence number. Patient ID is only unique within a single registry, therefore, SEER Registry and Patient ID are used together to identify a person. Sequence number defines the order of diagnosis of the tumors for an individual.
The following are Case Listing results that illustrate different Person Selection options. Open these 3 results matrix files in SEER*Stat and compare the tables. Selection used for the index record: malignant female breast cancer cases diagnosed from 1996-1999 in Iowa (ages 50+).
- ps.off.slm: Person Selection is OFF. This table only contains records matching the index record selection (look at the site recode and behavior recode columns, all records are malignant breast cancer cases).
- ps.on.slm: Person Selection is ON. The only change made in the session was checking "Person Selection" in the Session menu. No changes were made to the Selection Tab. Therefore, all records in the first matrix (ps.off.slm) are also included here. If the woman with the breast cancer case in 1996-1999 had another diagnosis of any kind, that other case is also shown. Look at the data columns. Notice that the "other" records may have a diagnosis date earlier than 1996 or later than 1999; they may be malignant or in situ. In other words, the other records are not restricted to any of the search criteria used for the index record.
- ps.on.narrow.slm: Person Selection is ON and "Narrow the Selection of Persons" was checked on the Selection Tab. The default setting for the other records was used, that is, "Select only records with a matching index record and at least 1 additional record matching the following:". No selection statement was added to the box that follows, but because Malignant Behavior and Known age was selected in the Select Only box (default settings), each person in the results has at least one other malignant cancer. There are far fewer records in this matrix than in #2. The breast cancer records for women with only one tumor record, or whose other tumor records do not include at least one malignant tumor, are not shown.
Note: In all examples, the selections for malignant cases, known ages, and cases only in the research database are made using the checkboxes at the top of the Selection Tab. When Person Selection is ON and "Narrow the Selection of Persons" is checked, the subset of records will match any selections made in the Select Only box.
Try to answer the following questions using Case Listing or Frequency sessions in SEER*Stat before viewing the answers.
- How many people included in the Nov 2013 data submission for the SEER 18 registries have 10 or more tumors?
- What is the total number of tumor records for those people?
- How many of these tumors are primary tumors?