Methods used by the halls.md Breast Cancer Risk Calculator.

Documenting the methods used by the Risk Calculator is essential, so the results can be trusted.  The basis for risk calculation methods is described in references 1 through 5. See the references page, below.

The internal workings of my Gail model 1 calculator start by calculating the "summary relative risk" using the answers from questions 1 through 6.

Then you have the option of adding several extra relative risk factors (from questions 7 through 12)  which were not part of model 1. These extra factors are:

• a) mammographic density (ref 6),
• b) age-specific tamoxifen effect (ref 7),
• c) dose-specific alcohol effect (ref 9),
• d) age-specific correction factors to adjust for Black race (using data from ref 5),
• e) presence of LCIS on a biopsy (ref 10), and
• f) oral contraceptives effects (ref 11)

The relative risks of these extra factors are multiplied with the summary relative risk.

Then, the absolute risks at 10-year, 20-year, 30-year and lifetime(90) are calculated using the graphical interpolation method described in reference 1.  Polynomial equations are used in place of the curves graphed in reference 1.

However, model 1 is known to underestimate absolute risks in older women particularly,  so age-specific correction factors are applied (from ref 5).   Then, a correction is applied for not participating in screening mammography (ref 8).  I have not yet implemented a correction factor for one problem:  The model 1 calculator can significantly overestimate the chance of cancer in women with many risk factors.  This is a known problem with model 1 and it can be made worse by including extra risk factors.

The internal workings of my NSABP model 2 calculator also start by calculating the summary relative risk, using the basic Gail model method.  However, my model 2 calculator uses slightly different values for relative risks than those published in  reference 5 for model 2.   The Appendix (below) lists the relative risk figures used by my calculator, in comparison to the published values from reference 5.  To explain the difference, if you try to read the text of reference 5, you'll see that the mathematics described are very complicated, and unfortunately, not all the mathematical terms used are fully defined in reference 5.  I suspect that the relative risk values used in my calculator, have merged the effects of mathematical terms "h*I(t)" and "F(t)", whereas these terms are presumably used in discrete  mathematical operations in the NCI's downloadable "Breast Cancer Risk Assessment Tool", (which implements model 2).  Model 2 does not include the risk of developing Ductal Carcinoma In-Situ (DCIS).  Therefore, the NSABP model 2 results will be lower than Gail model 1 results.  In my risk calculator, there is a checkbox in the "Fine Print" section, that will add in the risk of in-situ cancers (such as DCIS) into the model 2 estimate.

Reminder:  Because this calculator includes extra risk modifiers and correction factors that were not originally part of  model 1 or model 2, it is not possible to demonstrate that the results are accurate from published population-based studies.  Therefore, please consider these results to be an estimate.

Sincerely,
Steven B. Halls, MD, FRCPC.

Nagivation Options:

• Click here to examine the Breast cancer risk References.
  (links to medical journals).
• Click here to read the Appendix.
  (details about modifications to the standard Gail Model).

Back to the Breast Cancer Risk Calculator. or back to Body Mass Index and other calculators.