What Are Preliminary Estimates, and Why Use Them?
Each April, the NCI releases official cancer statistics using data submitted by the registries the prior November. For example, in April of 2024 data submitted in November 2023 includes cases diagnosed through 2021, with registries aiming to capture at least 98% of expected 2021 cases in the November submission to ensure completeness. These statistics are also used as part of the U.S. Cancer Statistics, the Annual Report to the Nation on the Status of Cancer, and in many other reports.
In February, updated data becomes available, adding an additional year of diagnoses (e.g., February 2024 includes cases through 2022). This data does not need to meet the same completeness standards, as the registries have a shorter time to submit. The Surveillance Research Program (SRP) combines this February data with statistical modeling to adjust for the undercount of cases in the most recent years to produce preliminary estimates of age-adjusted rates and trends through 2022, previewing patterns expected in the official April release.
Preliminary estimates can help to identify these impacts earlier than one could otherwise and can help guide research and facilitate more prompt attention to areas of concern. SRP continues to work on minimizing the delay in reporting to allow for official statistics to be available more quickly; the availability of these preliminary estimates demonstrates progress on such efforts.
SEER*Explorer and this website provide preliminary estimates of age-adjusted rates and trends (2000-2022). The website also includes information on the interpretation and limitations of preliminary estimates as well as comparison with the NCI official trends through 2021.
What Cautions Should Be Considered in Using Preliminary Estimates?
Please note that because SRP is releasing these early estimates based on less complete data, they are preliminary and therefore subject to change. The following points should considered as someone uses these estimates.
- Case counts based on the February submission represent a much larger undercount than the counts in the subsequent November submission.
A Reporting Delay Adjustment Model (based on historical data) adjusts for the undercount of cases in both submissions of data (November and February). However, the adjustments for November data are about three times smaller than those for February data. This larger adjustment for the preliminary counts using the February data submission, means a greater potential for error in estimating precisely how large the adjustment should be. - Preliminary estimates include a limited set of registries representing less of the U.S. population.
The NCI official trend estimates include data from 22 registries covering about 50% of the U.S. population. Preliminary estimates, however, are based on a smaller set of registries. Some registries are not included in the estimates because either: (1) they have not been part of the SEER program long enough to have a history of preliminary estimates, or (2) their case count completeness for preliminary data is insufficient to ensure the highest quality preliminary estimates. Users should be aware of this geographic limitation of preliminary estimates. - Validation studies have demonstrated generally good agreement between the adjusted preliminary incidence rates and the ones later released in April of each year.
While past publications have generally shown good agreement between the adjusted preliminary estimates based on the February submission and the official estimates submitted in November of that year and released in April of the next year, appropriate caution is needed when interpreting these preliminary estimates. For further background and publications on preliminary estimates, see Historical Background and References for Preliminary Estimates.
This website contains the following sections:
- Preliminary Incidence Rates and Trends 2000- 2022
- Methodology for Deriving Preliminary Incidence Rates
- Validation
- Historical Background and References for Preliminary Estimates
When using these estimates, please refer to them as preliminary and reference them as suggested below.
Suggested Citation
SEER Preliminary Cancer Incidence Rate Estimates for diagnosis years 2000 to 2022, National Cancer Institute, Bethesda, MD, https://seer.cancer.gov/statistics/preliminary-estimates/, based on the February 2024 SEER data submission. Posted to the SEER web site and SEER Explorer, November 2024.