We mostly don't have detailed patient histories. Most of the relevant clinical data that would be needed for such a rigorous statistical analysis isn't recorded as discrete data elements, it's just unstructured text. NLP can be used to extract concept codes but the error rate is high so you still need an experienced (expensive) human to manually fix the errors. No one wants to pay for that.

The codes we do have are mostly CPT4 and ICD-10 for billing purposes. Those are generally pretty accurate, but not detailed enough to reliably assess whether one surgeon is better than another at a particular procedure.