Swallowing a Bitter Pill: Expensive Prescriptions Mean Low Adherence

Insurance companies shift costs to patients as prescription drug costs rise. Studies show that increasing drug prices result in reduced patient use of prescribed drugs. This decreased utilization can cause higher overall costs through increased medical costs and complications resulting from non-compliance with physician recommendations.

A recent NBER working paper by Mariana Carrera, Dana Goldman, and Geoffrey Joyce examines the effect of prescription drug choice by physicians with respect to patient income level utilizing a large and well publicized price change of Zocor, a widely prescribed cholesterol drug.

As physicians do not have access to patients’ copays for prescription drugs or explicit knowledge of patient income level, the authors theorize that a physician who prescribes more drugs to a poorer, and therefore more price sensitive patient, will have better knowledge of the patient’s willingness to pay. This may happen as a patient feels more comfortable with the physician, or as a physician learns about a patient’s adherence pattern to differently priced prescription drugs. Utilizing this information, a doctor will be more likely to prescribe a drug they know is cheaper, such as a generic drug, even if the drug may be slightly inferior to a more expensive drug, as the patient will be more likely to adhere to the prescribed therapy. A higher-income patient will not be as price sensitive, and the prescriber will not be as likely to prescribe a generic as this will not improve adherence and may not be as effective a treatment.

To test their theory, the researchers utilize data from full medical and pharmaceutical claims from 150 benefit plans offered by 29 Fortune 500 firms from 2005-2007. The ages of the beneficiaries are restricted to those between 30 and 64. Yearly income for beneficiaries is included in increments of $10,000, from less than $50,000 to more than $250,000.

The researchers take advantage of the 2006 change in patent status for the prescription drug Zocor. Zocor is in the class of drugs known as statins for which there are a number of close substitutes. Alternative drugs in this category, still under patent protection, are more efficient; that is, the same treatment effect occurs with lower doses of the drug, decreasing the risk of side effects. Therefore, switching a low-income patient’s prescription to generic Zocor around the time of patent status would primarily be driven by cost considerations.

The authors find that the change in patent status for Zocor was associated with an increase in physicians’ responsiveness to drug cost and an increase in the likelihood of prescribing a cheaper alternative, especially for low-income patients. The paper’s regression model represents this by a change in the significance of an interaction term between an indicator for high-income and the national average copay using a log likelihood model of prescribing behavior. Prior to the patent change, there was no statistically significant difference in prevalence in prescribing a cheaper statin drug to a high- or low-income patient, and physicians were only slightly responsive to the average national copay; each 10 dollar increase in the average national copay was associated with a 0.14 decrease in probability of a drug’s prescription. Upon Zocor’s loss of patent protection, however, this probability decreased to -0.42 for each 10 dollar rise in cost under the most restrictive model. An interaction term between an indicator for high-income (greater than $50,000) and an increase in price was +0.20 in the most restrictive model. This implies that while physicians became more likely to prescribe cheaper drugs to all patients, the effect was more pronounced for low-income patients.

The paper also examines the mechanism through which physicians might learn more about patient sensitivity to price. The model indicates an association between price differences of an individual’s copay and the national average copay, but only for low-income individuals with more than five recently prescribed prescriptions. For low-income patients with a prescription history with the same physician, there is also a correlation between prescription history with one prescriber and prescribing a less expensive drug. Combined, these results indicate that physicians can, in fact, learn about low-income patient’s preferences for cheaper drugs. With more than five prescriptions the patient may better understand drug formulas and costs. Physicians may also be able to see the types of drugs a patient has been prescribed, and infer their preference for less expensive drugs. In the second case, where patients have a history with a specific physician, a patient may be more comfortable expressing their preferences, or a physician may be more familiar with a patient’s financial restrictions.

A model quantifies the six month adherence to a prescribed statin, as measured by filling enough scripts to take the drug at least 80 percent of the time. The results indicate that an increase in copay by ten dollars results in a decrease in the probability of adherence by 0.04 for low-income patients, but reduces the probability of adherence by only 0.006 for high-income patients. This indicates that lower-income patients are much more price sensitive, and would benefit most from cheaper prescription drug costs.

As patients increasingly shoulder the burden of higher drug costs, physicians should be supplied with information about patient copays in order to improve adherence for low-income patients, patient outcomes, and lower long-term costs. Additionally, for some patients, insurance companies may want to encourage “stepping”, in which a lower-tier drug must be prescribed before moving to a similar, higher-tier drug. By encouraging the use of cheaper drugs as a front-line, patients with less knowledge about their prescription prices are more likely to adhere initially, and continue to adhere, rather than getting “sticker shock” from a more expensive drug and deciding that they cannot afford to continue treatment. Both of these policies would have the effect of improving patient adherence, especially for low-income patients.

Feature Photo: cc/(Charles Williams)