Association for Public Policy Analysis & Management

Research Objective: The United States in the midst of an opioid epidemic, with 20,101 overdose deaths from opioid pain relievers in 2015. There is widespread consensus among medical and public health experts that increasing rates of opioid dependence and overdose are driven by high-risk opioid prescribing practices, such as long-term and high-dose opioid prescribing, by well-intentioned healthcare providers attempting to lessen patients’ pain. One potentially important driver of high-risk opioid prescribing, direct-to-physician marketing or “detailing”, has received little prior study.

Pharmaceutical firms spent $29.3 billion on marketing in 2010, with the majority coming from direct-to-physician marketing or “detailing.” There are approximately 80,000 pharmaceutical sales representatives in the United States with “detailing” taking the form of direct visits, consulting, gifts/meals, and speaking engagements. In order to combat potentially improper detailing, many hospitals have begun adopting stricter conflict of interest (COI) policies to better regulate relationships between physicians and manufactures, though, the impact of marketing or marketing restricting policies have not been examined in the opioid class. Understanding this relationship is crucial given that 1 in 12 physicians in the United States has received opioid marketing dollars between 2013 and 2015.

In this paper, we examine the potential impact of opioid marketing on physician prescribing patterns by leveraging plausibly exogenous changes in hospital COI policies. We match 2013-2015 data on physician prescribing patterns (Medicare Part D prescriber file) and physician characteristics (Physician Compare) to data on academic medical center COI policies (American Medical Student Association COI scorecard).

Study Design: Our intent-to-treat analysis at the physician-year level employs a difference-in-differences framework to estimate the effects of academic medical center COI policies on opioid prescribing behavior. Outcome measures include physician-year level opioid claim counts, total cost of opioid prescriptions, days supplied, and number of beneficiaries prescribed an opioid. Academic medical center COI policy measures include policies (and strength of policies) regulating consulting, disclosure, gifts and meals, interactions with sales representatives, and speaking engagements.

Population Studied: Our matched dataset includes 97,313 physicians who are primarily affiliated with 114 academic medical centers.

Principal Findings: We find that stronger disclosure, sales representative interaction, and speaking engagement policies decrease opioid prescribing. Implementing at least 3 marketing restricting policies is associated with 2.3 percent fewer opioid days supplied, and, implementing all 5 marketing restricting policies is associated with 6.8 percent fewer opioid days supplied. Effects are concentrated in disclosure and sales representatives policies. We find weak or no evidence that consulting, curriculum development, or gift and meal policies impact prescribing.

Conclusions: Our findings contribute to the literature on direct-to-physician detailing and suggest potentially economically relevant effects for marketing restricting policies on physician prescribing behavior. In determining conflict of interest policies and considering limits on pharmaceutical company-physician interactions, hospital administrators and policy makers should consider stricter policies, especially with regards to disclosure.