Venue: The Fuqua School of Business, Duke University, 1 Towerview Drive, Durham, NC 27708-0120
Presentation
Physician Response to Financial Incentives and FDA Labeling: the Case of Anti-cancer Drugs
Physicians remain the key decision makers in the vast majority of healthcare transactions, and their economic relations with patients and payers are characterized by asymmetric information that can lead to agency problems. Specifically, physicians may act as imperfect agents when treating their patients by choosing treatments that maximize profits instead of optimally improving their patients' health.
Our analysis is based on choice of a therapy conditional on treatment, and the focus permits the exploration of physician response to differences in incentives between technologies. Physician-administered drugs, which include most anti-cancer drugs and biologics, have explicit financial incentives for the administering physicians, with profit margins ranging from pennies to hundreds of dollars across different drugs. This research explores physicians' response to these financial incentives as well as other factors like new clinical information when choosing drugs to treat metastatic breast cancer.
We also measure the difference in choice probabilities for observed attributes of drugs that have been approved by the FDA for the treatment of a specific disease versus those that have not been approved for such use but are used in an "off-label" capacity. Specifically, we examine how much FDA approval matters to a physician treating breast cancer with anti-cancer drugs.
SEER-Medicare and Medispan pricing data are formed into a panel of 4,503 patients who were diagnosed with metastatic breast cancer and treated with anti-cancer drugs from 1992 to 2002. Discrete choice models are employed to analyze the effects of product attributes, including financial margin, volume of randomized controlled trial publications, FDA-labeled indications, years since launch, generic status, and other covariates, on odds of choice of anti-cancer drug. We use anti-cancer drug-level fixed effects to control for average unobserved heterogeneity related to a drug's quality. We also use the pharmaceutical preparations producer price index to instrument for time-related endogeneity related to acquisition cost and profit margin of the drugs.
The odds ratio (OR) for financial incentives for a patient's first anti-cancer drug administration is positive and significant on choice of drug. A $100 increase in reimbursement per day increases the odds of choosing a drug by 5 to 30 percent. The financial incentives effects are robust to multiple specification and model changes. An individual physician's odds of using a drug are 26 to 79 percent lower if the drug is off-label for breast cancer treatment versus a drug that is on-label depending on the number of administrations a patient has had. However, when employing the mixed logit model, statistical significance disappears in three of four specifications.
This research provides evidence that financial incentives had a positive but modest effect on the average physician's choice of anti-cancer drug when treating metastatic breast cancer. Coefficient estimates are all positive and statistically significant, but marginal effects on drug market share and a simulation model reveals a total budgetary impact for Medicare of less than one million dollars a year for metastatic breast cancer. We find that there is evidence that physicians seem to prefer on- to off-label drugs, especially for first administrations.