With many prescription medications available to treat a particular health condition, drug choice is a joint decision of physician and patient and depends on many factors. In particular, drug choice may be influenced by the drug choice of peers who share the same health condition. Observing the experience of others with a drug may reveal information about a patient’s own preferences (for example, drug side effects) and lead to a better patient-to-drug match. On the other hand, the experiences of others in the peer group may lead patients to demand or, instead, to reject a specific drug regardless of physician's recommendations, perhaps leading to a worse patient-to-drug match. Both impacts have implications for health outcomes and lifetime health care costs. In this paper, I document the existence and magnitude of peer effects among household members who share the same medical condition on the choice of prescription medications. Despite the growing literature on the effect of social networks on health care use, there is currently very little research on peer influence in drug-taking behavior. To my knowledge, this is the first paper to study household peer influences in prescription medications.

Existence of peer influence on prescription drug choice has important consequences for information dissemination, impact of drug advertising, drug adherence, and understanding the role of patients in clinical decision-making. It may also allow states to bargain more aggressively with drug manufacturers for rebates and discounts on medications on state drug formularies. When peer effects are present, the demand for drugs listed on formulary lists is amplified. Previous literature found that drug formularies in state Medicaid programs impact the choice of drugs not only for people covered by Medicaid but also for privately-insured patients. The proposed explanation for the spillover is that doctors tend to prescribe using the drug plan of the average patient. My paper suggests peer effects among households in which members are covered by different insurance plans is an alternative explanation for observed drug formulary spillovers.

Identification of peer effects is challenging due to selection of household members, shared household environment, and simultaneity of peer influences. I address each of these issues using data from the Medical Expenditure Panel Survey from 1996 to 2011. I use two alternative strategies to identify the effects. First, I control for observable common factors at the household level and use the timing of the first prescription for a drug. I estimate the effects separately for each type of relationship among household members because of differences in the degree of self-selection in groups. For example, married couples are more likely to suffer from the household selection compared to in-laws. In the second strategy I propose a new instrument, receipt of a free drug sample, to address the endogeneity of peers’ use of the drug.

I find significant interaction effects within households in their use of prescription medications and the effects are robust across alternative specifications. I also explore asymmetries in the flow of peer influences among household members.