This paper employs a nationally representative, household-based dataset in order to test how the compensation method of both specialists and primary care providers affects surgery rates. After controlling for adverse selection, I find that when specialists are paid through a fee-for-system scheme rather than on a capitation basis, surgery rates increase 78%. The impact of primary care physician compensation on surgery rates depends on whether or not referral restrictions are present.
You can read the full text of a working paper version here. Other research by Jason Shafrin can be found at JasonShafrin.com.