An aim of the study, said lead author Dr. James S. Yeh, was to determine and reduce any industry influence that could produce bad behavior.
“You want your doctors to be objective rather than doing something because there is a financial gain, be it subconscious or conscious,” Yeh said. “That’s not the way we should be doing medicine.”
Yeh added that not all industry relationships with physicians are problematic. Often pharmaceutical companies fund large research projects that could produce advances in medical care.
The Harvard study concluded that physicians’ rate of prescribing brand-name statins — the category of drugs that treat high cholesterol — increased by 0.1 percent for every $1,000 in industry money received. Under $2,000 in payments, there was no significant increase in brand-name prescribing.
A 0.1 percent increase in brand-name prescribing might not seem like much, but the study notes that brand-name statins cost two to four times wholesale as much as generics and that cost is an important factor in patient health outcomes. Patients are less likely to keep taking drugs that are more expensive.
Both ProPublica and the Harvard researchers were careful not to assign a causal relationship between payments and brand-name prescribing because there’s a lot that the data alone can’t show — for instance, why a doctor chose a certain drug, or the list of preferred drugs used by a patient’s insurer. It’s also possible that pharmaceutical companies market their drugs more to doctors who are already high brand-name prescribers.
“I think the causal relationship is really difficult to determine because there’s so many other factors that impact the physician’s prescribing behavior,” said Yeh, a research fellow at Brigham and Women’s Hospital and Harvard Medical School.
The Harvard study differs from ProPublica’s analysis in a few ways. ProPublica looked at doctors nationwide in the five most populous specialties, using data from the U.S. Centers for Medicare and Medicaid Services’ Open Payments program, which included all payments to doctors from drug and device makers in 2014. The Harvard researchers only looked at Massachusetts doctors, using a 2011 physician payment database kept by the state’s department of health.
Also, ProPublica focused on overall brand-name prescribing rates, not just for statins. The researchers chose statins because they are frequently prescribed and because there are both generic and brand-name options.
Both analyses used prescribing data from Medicare’s Part D program to determine physicians’ brand-name prescribing rates — in fact, the Harvard researchers purchased Part D data prepared for analysis by ProPublica. Part D currently provides drug coverage to nearly 41 million seniors and disabled people.
One of Yeh’s co-authors, Dr. Aaron Kesselheim, provided feedback on the design of ProPublica’s analysis.