This week, I talked by email with physician and researcher Dr. Joseph Ross about the importance of using unique identifiers in pharma disclosure data, and other ways to take the guess work out of the Physician Payments Sunshine database.
PS: Last week, a commentary in JAMA (Carpenter/Joffe) suggested that unique investigator identifiers should be used in the Physician Payments Sunshine database to make the data most usable, accurate, and suitable for cross-reference with other databases.
You’ve spent a lot of time with data collected under similar disclosure laws in Minnesota and Vermont. From your point of view as a researcher, what aspects of an identifier are important or helpful to looking at these data?
JR: From our experience examining the data in Minnesota and Vermont, I can only say that some unique identifier is necessary. Neither the Minnesota or Vermont data that we used included a unique identifier, just the name of a physician, clinician, or hospital/clinic that accepted the payment from the pharmaceutical company.
We found that trying to determine exactly who accepted which payment was virtually impossible, because one could not be sure whether a “Dan Smith” in Minneapolis was the same or a different “Dan Smith” in St. Paul, for example. Without a unique identifier, doing research to examine which physicians or clinicians are accepting payments can only be done in the aggregate, not at the individual level.
However, what is most interesting about Carpenter and Joffe’s proposal is not the unique identifier in itself, but the development of an identifier that could be used across databases, linking payment disclosure data, to grant applicants, to manuscript submissions and publications, to service on NIH or FDA committees. I could imagine this identifier being an expanded use of the NPI (national provider identifier), or another number, but it would certainly make identifying potential conflicts of interest far simpler.
PS: One of the reasons these authors suggest a unique identifier is to capture payments made to non-physician investigators. Did you deal with this issue in your work with the Minnesota and Vermont data?
JR: Yes, examining the data in Minnesota and Vermont, there were a substantial number of payments made to non-physicians, although these clinicians were not necessarily investigators. I agree that if the PPSA is broadened to include payments made to non-physicians, an alternative identifier to the NPI might be necessary. NPIs are required by all health care providers, including non-physicians who bill insurance companies or write prescriptions. But some non-physicians who receive payments from industry may not fall into this category.
However, to me, the real issue is a shortcoming of the PPSA, which requires companies to include physicians’ NPIs in their annual submissions, but prevents these numbers from being made publicly available.
PS: Why is that a problem?
JR: This is a problem because, as it stands, the payment disclosure data cannot be confidently attributed to an individual without using the NPI as a unique identifier. Our prior experience suggests that individual physicians will appear in the disclosure data multiple times, potentially with different spellings of their names, different practice addresses and so forth. It becomes a bit of guess work to determine if a payment to Daniel Smith from Lilly in September should be combined with the payment to Dan Smith from Lilly in April.
PS: In your 2007 JAMA article, you wrote about difficulty you had obtaining the Vermont/Minnesota data, and the poor quality of the information once you received it. Out of this experience, what should some of the priorities be in developing Sunshine systems and payment categories to produce a meaningful public disclosure scheme?
JR: Well, we’re still more than two years away from seeing how the PPSA plays out in practice. The legislation was fairly specific in terms of what information is to be collected.
Of all the categories pre-specified, the “nature” of the payment is most subject to interpretation.
My hope is that they interpret it as specifically as possible, so that, for instance, its disclosed when a payment is for a) speaking at a ACCME accredited educational event, b) speaking at a non-ACCME accredited educational event, c) attending a ACCME accredited educational event, or d) attending a non-ACCME accredited educational event. If the “nature” of a payment is disclosed vaguely, all four of those activities could be listed under the umbrella of “education.”
Otherwise, I agree with the Carpenter/Joffe commentary: Something needs to be done to ensure that a unique identifier is required for disclosure by companies and made available to the public. Without this, these payments may be misattributed, particularly within larger markets, limiting the effectiveness of this law within those communities.
Joseph S. Ross, MD, MHS, is an Assistant Professor in the Section of General Internal Medicine at the Yale University School of Medicine in New Haven, CT. He holds a medical degree from the Albert Einstein College of Medicine, Bronx, NY, and completed his post-graduate training in primary care internal medicine at Montefiore Medical Center in Bronx. As a fellow in the Robert Wood Johnson Clinical Scholars program at Yale University, Dr. Ross earned a Master’s degree in health sciences research. Using health services research methods, Dr. Ross’s research focuses on examining factors which affect the use or delivery of recommended ambulatory care services for older adults and other vulnerable populations, evaluating the impact of state and federal policies on the delivery of appropriate and higher quality care, and issues related to conflicts of interest, medical professionalism, and drug safety.
Interviewed by Kate Petersen, PostScript blogger