Does financial self-interest in imaging genuinely give rise to increased costs?

Motivation: Financial self-interest (FSI) in imaging is controversial, partly because imaging has grown especially rapidly. Proponents say FSI in imaging has a number of advantages, such as patient convenience and speedier initiation of definitive treatment. They also say previous research that finds FSI in imaging is associated with elevated utilization has multiple limitations that render its findings unconvincing. These purported limitations include: • With one exception, there has not been an adjustment for disease severity. • Multiple important issues never studied, such as the possibility that: o Physicians with FSI have a naturally imaging-intensive practice style and do no more imaging than if they did not have FSI.

To ascertain if the relationship between FSI and high utilization is current and genuinely causal, our research addresses the limitations that are claimed.

Methods: Using Medicare's 2004-07 “Research Identifiable Files” (a 5% sample of fee-for-service beneficiaries) and Symmetry's Episode Grouper, we study 12 broad medical conditions (such as heart disease) and a total of 25 combinations of medical conditions with imaging modalities. We use both multivariate logistic regression and difference-in-difference analyses.

Results: Unadjusted, the odds of imaging by a physician with FSI averaged 2.15 times those of a physician without FSI. Adjusted for disease severity, the odds ratio still averaged 2.15. Fully adjusted (for other patient characteristics, physician specialty, etc.), the odds ratio averaged 1.87. Prior to acquiring FSI, physicians who eventually acquired FSI had average odds of imaging 1.29 times those of physicians who did not acquire FSI. After acquiring FSI, the relative odds ratio averaged 1.82. On average, physicians who acquired FSI increased their odds of imaging by 49% more than did physicians who did not acquire FSI. Physicians with FSI in a modality averaged more imaging in other modalities than did physicians without FSI in the modality.

Conclusions: Incontrovertibly, FSI in imaging is associated with, and engenders, high utilization, and this remains currently true. After adjusting for plausible explanatory factors, imaging by physicians with FSI remains much higher than imaging by those without. Acquisition of FSI by itself generates a nearly 50% increase in imaging. Those with FSI in a modality make more use, not less use, of alternative modalities. Recent large payment reductions have had very little effect on the situation. Payers should consider banning—or, at the least, severely limiting--FSI in imaging unless FSI's claimed advantages can be shown to be both real and large in their effect.