How should I estimate age at last follow up and/or year of birth when these data are missing?

We cannot suggest a specific age at last follow up/year of birth to use for individuals where this is unknown. Nevertheless, you should keep the following points in mind when estimating these parameters:

• When you set a person’s age or year of birth to unknown, that individual is effectively excluded from the CanRisk Tool risk calculation

• If you know that a family member was unaffected by cancer and you set that person’s age or year of birth to unknown, then this will lead to an overestimation of risk, as that unaffected person will be excluded from the risk calculation. As a result, we think it is a good idea to estimate age and year of birth for unaffected individuals. However, if you do this, you need to bear in mind that regardless of the values of age and year of birth that you choose, interpolating data in this way may affect the results.

• If you know that a family member was affected by cancer and you set that person’s age or year of birth to unknown, then this will lead to an underestimation of risk, as that person will be excluded from the risk calculation. Therefore, if the year of birth of an affected individual is unknown, then we recommend that you estimate it. Similarly, if the age of an affected individual is unknown, then we recommend that you set it equal to the age at latest cancer diagnosis in the first instance.

• Where possible, it is a good idea to constrain interpolated ages and years of birth using other reliable data in the pedigree. For example, you could perhaps use the year of birth of the oldest child in a nuclear family to constrain the year of birth of that individual’s parents where unknown. When you have uncertainties of this kind in your input pedigree data, we recommend that you carry out multiple processing runs i.e. compute CanRisk Tool risks for each of the different input data scenarios to see the effect on the computed results. For example, you could run a first set of risk calculations using pedigrees where ages and years of birth for the older generations have been set to unknown, and then run a second set of of risk calculations using pedigrees where the same ages and years of birth have been interpolated. This will show how the CanRisk Tool responds to the different data inputs.

• When assessing risk modelling results, it is important to consider the assumptions on which the risk models and input data sets are based