To determine whether the prescribing patterns for nonindicated androgen suppression therapy (AST), using neoadjuvant AST as the model, changed according to the prevailing clinical evidence, changes in reimbursement, or evidence of increased harm from treatment.

We identified 34,976 men with prostate cancer who had undergone radical prostatectomy within 12 months of diagnosis from the Surveillance, Epidemiology, and End Results-Medicare data set (1992-2007), and their clinical and demographic parameters were assessed. We measured the Medicare claims for receipt of AST before radical prostatectomy and calculated the annual rates of neoadjuvant AST, which were adjusted for confounding variables using multivariate logistic regression analysis, and compared them with the prevailing published clinical data on the outcomes of neoadjuvant AST, changes in reimbursement, or published data on clinical harm from treatment.

The use of neoadjuvant AST increased from 7.8% in 1992 to a peak of 17.6% in 1996 and then decreased steadily to 4.6% in 2007. This rate change was significant on multivariate regression analysis, with a single join point in 1996 (P <.001), and corresponded to published data showing improved surgical margin rates and pathologic downstaging in the early 1990s and data showing no improvement in disease recurrence or overall survival beginning in 1997. Changes in reimbursement and evidence of harm from AST were not associated with the decreased use of neoadjuvant AST.

Using neoadjuvant AST as the model for the nonindicated use of AST, physicians reduced AST use in response to high-level evidence showing a lack of benefit, despite the high reimbursement. This suggests that physicians adapt to emerging evidence and use evidence-based practice.