These cases were designated as BCSRT, and those with fewer than two RT claims as BCSO (breast conserving surgery only).
We used multinomial logistic regression analysis to estimate the relationships between the independent variables and the receipt of BCSO, BCSRT, or MST.
The BCS procedure mix has no impact on the relative odds of BOSO, but the MST procedure mix has a highly significant and positive effect on the odds of BCSO relative to MST.
Leaving out the procedure mix variables has virtually no effect on the BCSRT estimates, but reduces the BCSO fee coefficients by about one-third.
Women who are 80 or older, or who had very high Medicare payments in the prior year, are much more likely to receive BCSO and much less likely to receive BCSRT relative to MST.
We addressed many of the potential limitations and sources of bias in the earlier analysis by using data for individual women with confirmed diagnoses of early-stage breast cancer, by isolating the effect of variations in pure fees, by distinguishing between BCSRT and BCSO, and by controlling directly for the effects of prior health condition and disease stage.
We found that Medicare fees were significant factors in the choice between MST and BCSRT, but did not significantly influence the choice of BCSO versus MST.
If this inference is valid, then our earlier study, which could not distinguish BCSO from BCSRT, may have understated the effects of Medicare fees by combining BCSO and BCSRT cases.