Cancer Survival Statistics: Cohort Definition Using Diagnosis Year

Survival estimates from cancer registry data are usually dated measures of current-year survival, because of the time needed to observe survival and lag between available data and the current year. There are different approaches of grouping survival experience with respect to year of diagnosis and follow-up to obtain more up-to-date estimates of patients recently diagnosed. The following figures demonstrate these approaches. Figure 1 uses the minimum number of years required to define each method. Figure 2 uses additional years for stability, which is the method implemented by SEER* for most published survival statistics.

Cohort - Includes calendar years for which all cases have potential follow-up for the survival duration. For example, in Figures 1 and 2 above, the cohort method can only include patients diagnosed in years 2008 and 2006-2008 respectively. The Cohort method is useful in communicating survival trends.

Complete Analysis - Includes all patients diagnosed in the most recent years spanning the maximum duration to be estimated. For example, in the figures above, the complete analyses would include at a minimum patients diagnosed from 2008-2012, or all patients diagnosed from 2006-2012. Because the complete analysis cohort includes patients diagnosed more recently, it gives a more up-to-date estimate of recent survival.

Period - Uses only the most recent interval survival estimate of cases diagnosed in different calendar years (cross-sectional estimate of survival). In Figure 1, the estimate of period 5-year survival from cases diagnosed between 2008 and 2012 uses the first year interval survival from patients diagnosed in 2012, the two-year interval survival from patients diagnosed in 2011, and so on. Because period uses only the most recent survival experience, when there is an increasing trend in survival it provides a more up-to-date measure of recent survival (Brenner et al. 2002). The method implemented in SEER*Stat differs slightly from Brenner et al., (See Cronin et al., 2003 (PDF, 504 KB) for more information).

The five SEER*Stat matrices used to obtain the percentages for the Observed Survival by Year of Follow-up and Year of Diagnosis* figures shown above are available for download. You must have the SEER*Stat software in order to open these files.

• Individual Year Values (2006-2012) – individual.year.values.ssm
• Cohort (2006-2008 and 2008) - cohort.ssm
• Complete Analysis (2006-2012 and 2008-2012) - complete.ssm
• Period (2012, 1 Year per Cohort) – period.1year.ssm
• Period (2012, 3 Years per Cohort) – period.3year.ssm

Projection Method - Models and extrapolates survival using all available information. Newly diagnosed patients may desire an estimate of their prospects for long-term survival. Standard estimates of survival may be outdated since they do not reflect recent advances. The projection method fits a regression model to interval relative survival and includes a parameter associated with a trend on diagnosis year. The cumulative relative survival in a target year is calculated by multiplying the projected interval survival for that year. Because recent trends in survival are estimated and projected, the projection method may provide more up-to-date estimates of survival for newly diagnosed patients. (Mariotto et al., 2006)

*Years of Diagnoses Included in SEER Survival Analyses

In the examples above, survival is calculated through 2012 using SEER research databases that include cancer diagnoses and follow up through 2013. SEER excludes the last year of diagnoses in survival analyses because the patients all have less than one year of potential follow-up.