SEER*Stat Survival Exercise 3: Cause-Specific Survival

This exercise assumes that you are familiar with using SEER*Stat. If you are just getting started, be sure to do the introductory tutorials first.

Create a table showing 5-year cause-specific survival for regional stage female breast cancer diagnosed between 2004-2010 in the SEER 18 Registries. Include conditional survival. That is, calculate the probability of surviving 3 additional years from 3 conditioning points (Diagnosis, 1-year, and 2-years after). Display the results by age with the following groupings:

Age = All Ages, < 50, 50+

Include the detailed life tables and cumulative summary tables with cause-specific survival and standard errors. Calculate the statistics using monthly intervals, but only include annual intervals in the summary pages. Include the special intervals/conditional survival and median survival in the summary pages.

For more information about cause-specific survival and an overview of cancer survival statistics, see Measures of Cancer SurvivalExternal Web Site Policy on the Surveillance Research Program web site.

Key Points and Reminders

  • Due to inaccuracies in the coding of the cause of death variable, there are various ways to define the appropriate cause of death. SEER provides a user-defined variable with one grouping defining the cause of death for use with cause-specific survival. It is based on a field which was added to the SEER data starting with the November 2008 submission. For information, refer to the SEER Cause-specific Death Classification.
    • This variable seeks to determine whether a patient died from their cancer using site, sequence number and cause of death. Patients who died of something determined to be unrelated are treated as if they were alive. It was designed specifically to be used as the determinant of cause-specific survival analysis. Exclude all patients with missing/unknown cause of death from the analysis.
    • An additional value of "N/A not first tumor" was added to this variable with the November 2010 submission to define cause of death as only first cancers. Because of this, the selection criteria need to be limited to first cancers, which is done on the Selection tab.
    • This field was renamed "SEER cause-specific death classification" with the November 2009 submission, but was originally named "Vital status relative to site at dx".
  • Starting with the November 2009 submission, the default in SEER*Stat is to no longer exclude cases that are Dead Due to Other Causes with No Survival Time. Including these cases allows you to change your definition of cause of death and still work with the same cohort of patients. When these cases were excluded, changing the definition of the cause of death of interest also changed the definition of "other causes", so the number of patients in the cohort would change.
  • Because the survival session does not use census population estimates, you can use any of the age of diagnosis variables when creating the user-defined age variable.
  • For cause-specific survival, the expected survival table is used only for the Expected Survival Table Exclusions on the Selection tab and the Censor When Attained Age Exceeds Expected Table Max on the Parameters tab. These exclusions are optional when calculating cause-specific survival, but are convenient when trying to analyze the same cohort as used when calculating survival requiring expected survival tables.
    • Starting with SEER*Stat 8.1.4, there's a new option on the Parameters tab for censoring cases when attained age exceeds the expected rate table max. This is a default selection starting with the November 2013 data submission because the new expected life tables have a maximum age of 99 years. See the Expected Survival Life Tables on the SEER website for more information.
  • The survival calculation can be based on either pre-calculated survival duration variables or begin and end dates, starting with SEER*Stat 8.0.4. The pre-calculated survival duration option will be the default in SEER databases starting with the November 2012 submission. Begin and end dates are still associated with the pre-calculated duration field, but duration has been calculated using complete dates (day, month, year) rather than just month and year. By using complete dates the pre-calculated duration is more precise than using begin and end dates which do not have day available. For example, a patient with diagnosis = 7/31/2000 and last contact 8/1/2000, the duration would be 1 day, which translates to 0 months of survival. Using begin and end date fields without day, this would be 7/2000 to 8/2000 and would result in 1 month of survival.
  • Specify the Special Interval/Conditional Survival criteria on the Parameters Tab as: 1-36, 13-48, 25-60.

Step 1:  Create a New Survival Session

  • Start SEER*Stat.
  • From the File menu select New > Survival Session or use the Survival button on the toolbar.

Step 2:  Select the Database (Data Tab)

  • On the Data Tab select "Incidence - SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2013 Sub (1973-2011 varying)"

Step 3:  Choose the Statistics (Statistics Tab)

  • In the Cancer Survival Measures box, select Cause-Specific Survival.
    Cause-specific survival is a net survival measure representing cancer survival in the absence of other causes of death. Cause-specific survival estimates the probability of surviving a specific cause of death specified by you using the Definition of Cause of Death.
  • In the Expected Survival Table drop down box, make sure " U.S. 1970-2009 by individual year (White, Black, Other (AI/API), Ages 0-99, All races for Other Unspec 1991+ and Unknown)" is selected. The expected survival table is used only for the Expected Survival Table Exclusions located on the Selection Tab and the Censor When Attained Age Exceeds Expected Table Max on the Parameters Tab when calculating cause-specific survival.
  • Make sure the Definition of Cause of Death is set to "Dead due to cancer using SEER cause-specific death classification". If the database you are using does not have this variable already defined, you can create it using the following steps:
    1. Use the Create button to open the Definition of Cause of Death window.
    2. Using the controls at the top of the window, create a Selection Statement the reads:
      {Cause of Death (COD) and Follow-up.SEER cause-specific death classification} = 'Dead'
    3. Then click the OK button to open the Survival Specific Cause of Death Definition window.
    4. Edit the Name field with this variable name: "Dead using SEER c-s death classification", then click the OK button. Verify that the newly created variable is selected as the Definition of Cause of Death variable.
    5. Set the Missing/Unknown COD to Exclude from Analysis.

Step 4:  Define the Analysis Cohort (Selection Tab)

  • Use the default settings for the Standard Case Selections. These default selections represent the standard selections most commonly used for a survival analysis. Ensure that "Alive with No Survival Time" is selected, and "Dead Due to Other Causes with No Survival Time" is not selected.
  • Based on the Problem Statement, we want to make selections based on sex, year of diagnosis, cancer site, and stage.
  • Open the Case Selection window to create the search statement. When complete, the Selection Statement should read:
    {Race, Sex, Year Dx, Registry, County.Sex} = ' Female'
    AND {Race, Sex, Year Dx, Registry, County.Year of diagnosis} = '2004','2005','2006','2007','2008','2009','2010'
    AND {Site and Morphology.Site recode ICD-O-3/WHO 2008} = ' Breast'
    AND {Stage - LRD (Summary and Historic).Summary stage 2000 (1998+)} = 'Regional'

Step 5:  Set the Parameters

  • Use the default Survival Calculation setting for "Pre-calculated Duration".
  • The "Censor When Attained Age Exceed Expected Table Max" should be checked by default.
  • By default, SEER*Stat calculates statistics using monthly intervals. To show 5-years of survival, in the Intervals box, set the Number to 60. This indicates a maximum number of 60 monthly survival intervals to be displayed in the output survival tables.
  • Check the box "Include 0th Interval" to generate statistics for time zero.
  • Set the Special Intervals/Conditional Survival to 1-36, 13-48, 25-60.
  • Check "Cumulative Summary" and "Standard Life" in the Display Box.
  • Specify the Interval in the Cumulative Summary box as 12,24,36,48,60.
  • Check the boxes to "Include Special Intervals/Conditional Survival" and "Include Median Survival" in the Cumulative Summary pages.

Learn More...

  • These parameters establish how case survival time will be defined and the intervals that will be used in the analysis. The survival time for each case is calculated and then partitioned into the defined survival duration intervals. For every life table, SEER*Stat calculates the number of cases alive at the beginning of each survival interval.

Step 6:  Set the Table Variables (Table Tab)

  • Set age as a Row variable. You need to create a user-defined variable for age.
  • First, open the Data Dictionary.

  • Select the "Age recode with <1 year olds" variable from the "Age at Diagnosis" category and use the Create button to open the Edit Variable window.
    • Edit the Name field with this variable name, "Age (All, <50, 50+)".
    • Delete all the groupings from the Groupings box.
    • Create an age grouping for "< 50" using the values "00 years" through "45-49 years" in the Values box.
    • Create the "50+" grouping using the Add Rest button to add a grouping with the values "50-54 years" through "85+".
    • Select Added as one grouping (all values combined) on the Add Rest dialog.
    • Use the Add All button to create a combined grouping for all the values and name it "All ages". Move the "All ages" grouping to the top of the list using the Up button.
    • When you are finished, click the OK button.
  • Expand the "User-defined" category in the Available Variables box using the "+" symbol.
  • Add the age variable to the row dimension. The dimensions specified on the Table Tab only apply to the summary tables.

Step 7:  Edit Setting on the Output Tab

Enter the following title for your results matrix:
5 Year Cause-Specific Survival
SEER 18, Malignant Regional Female Breast Cancer
Includes Cases Diagnosed in 2004-2010
Survival Exercise 3

Step 8:  Execute SEER*Stat

  • Use the Execute button or select Execute from the Session menu to execute the session. (Execute Offline is a 3rd option available and has been explained in previous exercises.)
  • A dialog will display the progress of the job. When the job completes a new window will open containing the output table or matrix.

Step 9:  The Results Matrix

  • Use the Save As command on the File menu to save the matrix. Enter "Survival Exercise 3" as the filename. SEER*Stat will assign the "ssm" extension to indicate that this is a "SEER*Stat Survival Matrix" file.
  • Compare your results to this SEER*Stat matrix file: Exercise Matrix 3 Results.
  • The results matrix consists of multiple pages of output since the Standard Life tables use the variables you set on the Table Tab as page variables. Use the drop down list on the toolbar to select a different page to view.
  • There are no values displayed in the Median Cause-Specific column in the Summary Table. This is because the cumulative cause-specific survival did not go below 50% within the first 60 months of survival for any age group. If we calculated survival for distant instead of regional female breast cancer, values for median survival would be available within the first 60 months. See key.survival3_distant.ssm.
  • The Survival Results Matrix section of the help system contains more information about the SEER*Stat matrix and its features.