The Effect of Specialist Cost Information on Primary Care Physician Referral Patterns

Economists have long argued that cost information plays an important role in the efficient allocation of resources (e.g., Hayek, 1945), but this argument relies on the assumption that cost or price information is available to decision makers at the time of their decision. In the American health care system, this assumption is quite often not true. In the course of everyday practice, physicians often make decisions regarding drugs to prescribe, tests to order, and referrals to make in the near total absence of the comparative costs of the options available. This phenomenon has, to some extent, been documented in the medical literature (e.g., Shulkin, 1988; Tierney et al., 1990; Reichert S et al., 2000; Feldman et al., 2013).

In this paper, I examine the importance of cost information in the physician referral process. In modern health care provision, physicians are highly specialized workers who are often loosely organized in teams when a patient's medical care requires input from physicians in more than one area of specialized knowledge. Such a team is typically led by a primary care physician (PCP), who refers the patient to specialists with relevant expertise (Starfield, 1998). Since specialists may not all sell their services at the same rate, the patient's health care will only be cost efficient if the referral is made to the specialist who provides the required service (at the required level of quality) at the lowest price. In practice, however, the PCP usually does not have access to (and so cannot consider) information on the relative cost of available specialists, and so referrals are likely made inefficiently with respect to cost.

To investigate these issues, I partner with a group of physician practices -- an Independent Practice Association (``the IPA'') -- to perform a field experiment testing whether providing information on the costliness of specialist physicians to PCPs alters the PCP’s referral behavior. The IPA's primary care practices are assigned randomly to treatment or control groups, and the PCPs of the treatment group practices are provided a list of average costs for several ophthalmologists that are part of the IPA. Using data collected by the IPA, I compare changes between the treatment and control group referral rates to the ophthalmologists of interest. My results suggest that receipt of the cost list induced PCPs to reallocate referrals towards the least expensive Ophthalmology practice by more than 45%. At the same time, my analysis also takes advantage of the fact that the IPA services two broad types of patients – one for which the ultimate cost to the IPA varies with the intensity that the Ophthalmologist treats with, and another for which the cost is fixed (i.e. capitated). For capitated type patients, the response to the treatment is much more muted than for those for which cost depends on treatment intensity. This suggests that the motivation for the observed response is cost reduction, rather than an alternative theory like the cost information carries some sort of quality signal.