Does disclosure change the way doctors write prescriptions? And are there any other unintended consequences?

The answer to both is yes, according to new University of Michigan research. Ross School of Business doctoral student Tong Guo and professors Puneet Manchanda and S. Sriram studied the effects of a 2009 sunshine law in Massachusetts in three common drug categories.

They found that doctors prescribed significantly fewer name–brand drugs after the law was passed, and that doctors wrote fewer overall prescriptions. The implication is that while the decrease in name–brand prescriptions may have lowered costs, the decrease in generic prescriptions – while not as dramatic – suggests possible underprescribing.

"What got us excited about this research was that the effect of payment disclosures on physician behavior was an open question," said Manchanda, the Isadore and Leon Winkelman Professor of Marketing. "Legislators were convinced this was the right thing to do, that once you have disclosure, the costs will go down. But it's also possible the law has other nuances they didn't take into account. So we wanted to dig into the data to learn what happened."

The authors examined prescriptions written by physicians in Massachusetts because it created a public database of payment disclosures and because bordering states of New York and Connecticut didn't enact such a law, creating a good comparison group. It differs in a few ways from a sunshine provision in the Affordable Care Act, which covers physicians and teaching hospitals. The Massachusetts law covers anyone writing a prescription.

Guo, Manchanda and Sriram used outpatient prescription information from January 2008 and December 2011, tracking the number of new prescriptions and refills written by each medical professional for three types of drugs – statins, antidepressants and antipsychotics. This information was made available to the authors by the U–M Institute for Healthcare Policy and Innovation.

"Drugs in these three categories treat chronic conditions with patients taking them over long time periods and are marketed heavily," Guo said.

Comparing Massachusetts physicians' prescriptions with the prescriptions written by physicians in the control group(s), they found a 48–59 percent decrease for name–brand statins, a 46–54 percent decrease for branded antidepressants and a 40–45 percent decrease for branded antipsychotics.

"These are economically significant effects," Sriram said. "We were initially concerned about the size of these effects, but the fact that we find them consistently across different control groups and drug classes gave us a lot of confidence."

The study only measured the effects of disclosure, not in any changes in payments the law may have triggered. But another result suggests the change in prescription behavior is driven by disclosure – doctors in Massachusetts also prescribed fewer generic drugs after the law passed. Generic drug companies generally do not pay physicians.

Generic statins dropped by 38–46 percent, generic antidepressants by 32–41 percent and generic antipsychotics by 38–40 percent.

"There was a much larger relative decrease in name–brand drug prescriptions, which means some doctors shifted toward generics as a result of the disclosure and that fits with the intent of the law," Manchanda said. "But we also have this decrease in generics, which suggests self–monitoring could be leading to underprescribing. This paper opens up many other questions that we and other researchers need to study to get the full picture."