Table 4.
Sensitivity Analysis of Physician Self-Referral Indicator (Magnetic Resonance Imaging [MRI] Acquisition Status)
Dependent Variable: Low Back Surgery Receipt | ||||
---|---|---|---|---|
Orthopedist | Primary Care | |||
Definition of Physician MRI Acquisition | OLS | IV | OLS | IV |
≥1 claim, global or technical fee for MRI | 0.082 (0.001)* | 0.341 (0.139)** | 0.098 (0.000)* | 0.056 (0.071) |
≥3 claims | 0.082 (0.001)* | 0.374 (0.161)** | 0.098 (0.001)* | 0.039 (0.065) |
≥5 claims | 0.082 (0.001)* | 0.343 (0.177)+ | 0.098 (0.000)* | 0.056 (0.060) |
≥10 claims | 0.082 (0.001)* | 0.318 (0.170)+ | 0.098 (0.000)* | 0.039 (0.082) |
Notes. Standard errors in parentheses.
Regression estimates and standard errors for the association between patient's low back MRI use and surgery use, using physician self-referral status as an instrument for patient MRI receipt. Orthopedist model includes 78,914 patient episodes. Primary care model includes 661,553 patient episodes. Number of MRI claims billed by physicians is over the 1999–2005 study period. Estimates based on standard OLS or two-stage instrumental variables (IV) regressions that control for patient demographics, Medicaid status, prior year health spending, Elixhauser comorbidities, and year, month, and physician fixed effects. Regressions include physicians who bill for MRI (acquirers) and traditional MRI users.
Significant at 10%.
Significant at 5%.
Significant at 1%.