For the most basic example, we queried a 42 year old female with an age at menarche of 13. Hughes returns a lifetime (85) Tyrer Cuzick score of 13.1%. However if you put the same patient in IBIS, just age and menarche, the lifetime risk score returned is 11.1%. This happens in both simple (as in this case) and more complex histories.
As a general principle it is not uncommon for different applications that use these models to have varying results within a few percentage points based on the nature of the data and how is collected. We are always happy to work with you to explore individual cases in order to understand any issue and ameliorate any concerns. That being said, we have a pretty simple explanation for the case you provided:
In IBIS, by default when you open the application, the parity data point is set to ‘unknown’. In RiskApps we have traditionally considered the patient as parous if they provided an age of first live birth or in the presence of children on the pedigree, otherwise RiskApps considers the patient as nulliparous (not ‘unknown’). We have had this question before, and it is part of a larger conversation about how to best model patient provided data, given that we have a variety of patient surveys capable of collecting this information in different levels of detail based on the clinical workflow in question.
We are always interested in hearing from our users about their concerns and we are always working on ways to improve the software, and so I am very open to talking with your clinical staff about this or other issues that come up.