By using rigorous statistical methods and data on the outcomes of a large population of people with cancer, this calculator provides survival estimates that are useful for people with cancer, clinicians, researchers, and policymakers.
Like all prediction tools, however, the calculator has some limitations that create uncertainties in the survival estimates it produces. Users of the calculator need to understand these limitations and uncertainties when using the tool in clinical decision making.
Comparability of Treatment Outcomes
The calculator cannot compare the effectiveness of different cancer treatments, that is, the effects of different treatments on the person’s survival. This is because the calculator does not use data from randomized clinical trials. In randomized trials, people with cancer are randomly assigned to, and thus have an equal chance of receiving, different treatments. This enables a fair comparison of treatment outcomes.
The calculator does not give information on the effectiveness of different cancer treatments. It was built using cancer registry data, which are observational. People in the registry received their treatments based on a complex mix of factors (both observed and unobserved) and likely did not have an equal chance of receiving different treatments, making it hard to compare outcomes between different treatment approaches. Data about the effectiveness of different cancer treatments are best found by examining the results of randomized clinical trials. Guidelines for oral cancer care (PDF) are based on these and other types of studies that can suggest best practices for treating oral cancer.
The strength of the calculator is its ability to provide estimates of survival and risks of death from other causes. Because the data include a wide variety of people, not just those in randomized trials, the estimates are more likely to reflect typical survival. Read more about how the data used by the calculator differ from randomized clinical trial data.
Accuracy and Precision
The accuracy and precision of survival estimates produced by the calculator are limited by the following methodological circumstances:
- Data Limitations. Limitations of the SEER cancer registry data on which the calculator is based include missing and inaccurate data regarding patient characteristics and causes of death. However, the SEER Program uses rigorous quality control procedures that help minimize any problems with the data.
- Model Validation. The variables included in the model were determined with the full oral cancer SEER 18 registry data from 2000-2011. Validation was performed by estimating the model using a portion of the data and was validated (confirmed) using the remainder.
Applicability of Survival Estimates to Individual People with Cancer
No statistical model includes all of the unique factors (variables) that affect an individual person’s future survival, and since there is some randomness inherent in the progression of cancer and the ultimate outcome, these estimates do not tell us what might happen to any individual person. However, population-based survival estimates such as provided by this calculator can provide useful context for individuals who have a new cancer diagnosis, as long as these limitations are understood.
For example, when considering the 5-year survival estimate of 70% for people with a specific set of characteristics (e.g., age, cancer type and stage, coexisting conditions), this estimate refers to 100 similar people, of whom 70 will be alive in 5 years.
It is also important to understand which anatomic areas are included and not included in oral cancer. See the Oral Cancer: Key Things to Know page to learn more.