I would like to know how to adjust agreement measures to estimate candidate method agreement with a gold-standard method, given results of a candidate vs silver-standard test, where the silver-standard has some prior, known agreement with the gold-standard. I am looking for specific procedures to use with Bland-Altman, Passing-Bablok regression, and Arms methods.
Typical comparisons between medical test methods utilize Bland-Altman procedures, Passing-Bablok regression, and/or Accuracy rms error, between a candidate method and a gold-standard reference. Sometimes the gold-standard method is problematic, due to expense, difficulty, pain or inconvenience to patients or volunteers, etc. This is often the motivation for development of a silver-standard method, less vulnerable to these problems, at the cost of some degree of disagreement with the gold-standard results. The degree of disagreement (for example, limits of agreement) are known as a result of the original testing validating the silver-standard.
In developing newer methods, a candidate might be assessed against the silver-standard rather than the gold-standard, to avoid the problems associated with the gold-standard. An example would be evaluation of a pulse oximeter for SpO2 measurement; the ISO 80601-2-61 (2017) standard provides testing methodologies for either an invasive blood sample oxygen saturation reference, or alternatively, to another non-invasive pulse oximeter system with known error relative to blood samples. In the latter case, ISO say that the Accuracy rms error should be “expressed relative to the ‘gold-standard’ CO-OXIMETER and includes the error of the secondary-standard PULSE OXIMETER EQUIPMENT.” However, ISO does not provide any guidance on how exactly to do this.