As in previous studies, we found that patients are frequently prescribed branded medications when similarly effective generic medications are available.7
Unlike previous studies, our study also points to targets for changing prescribing patterns. Residents of poor zip codes, those least likely to be able to afford expensive medications, were over 25% more likely to receive branded medications. Consistent with previous findings, Specialists, and Obstetricians were more likely to prescribe branded drugs than Generalists.10
Efforts to provide information to patients and physicians about generic alternatives could focus on these patient and provider groups. Within the single insurer we studied, pharmacy benefit design and type of pharmacy did not significantly influence generic drug use when initial prescriptions are filled. This suggests that benefit design incentives for generic drug use are often not apparent to patients and physicians when medications are initiated.
Few studies have evaluated switches to generic drugs when prescriptions are refilled. In the classes we evaluated, only 1 in 7 patients started on a branded drug changed to a generic in the subsequent year, underscoring the importance of the initial prescription choice. When change did occur, enrollment in tiered pharmacy benefit plans and use of mail-order pharmacies were strongly associated with switching to generic medications. Our findings are consistent with previous research that has shown that tiered pharmacy benefit designs steer patients toward less expensive formulary alternatives,22–25
although the literature is mixed as to generic use.26–29
The fact that pharmacy benefit design had no significant impact on the initial receipt of generic medications adds fuel to the growing concern that patients and physicians do not possess the necessary information about out-of-pocket costs at the time of prescribing to make objective cost-benefit decisions about their medications.30
Patients who are charged higher copayments in tiered plans are more likely to switch to generics, suggesting that higher copayments stimulate requests for less expensive medications. Studies linking generic medication initiation to improved adherence to chronic therapy8
highlight the importance of making thoughtful cost-benefit decisions at the point of initiation.
Mail-order pharmacies may dispense more generics because of different protocols for medication switching, fewer time-constraints when contacting physicians to adjust prescriptions, or because use of insurer’s websites to refill prescriptions by mail-order may educate patients about generic options. Few patients in this sample received their initial prescriptions from mail-order pharmacies, but many more used these pharmacies to refill. While previous findings suggest that mail-order pharmacies are associated with improved medication adherence,31
recent trends towards more frequent mail-order pharmacy use32
and use of financial incentives for patients to use them may increase generic drug use.
Patients from lower-income zip codes were less likely to initiate therapy with generic medications and switched to generic drugs at similar rates to those who live in high-income zip codes. Our findings conflict with a recent study of Medicare beneficiaries taking medication for hypertension that found a modest relationship between income and generic drug use, with lower-income seniors using slightly more generic medications.11
Our study looked at a younger population, all of whom were commercially insured, and they may behave differently than seniors. There are several possible explanations for decreased generic drug use in lower income patients. Low-income patients may be less able or inclined to navigate tiered formularies and select cost-effective options when expensive medications are prescribed. Alternatively, physicians may offer more free samples to lower-income patients, steering them towards more expensive branded drugs, or low-income patients may have stronger preferences about using branded rather than generic medications. We were unable to control for the location of the pharmacy or the clinical encounter, which may have influenced these factors. Further research is necessary to evaluate why lower-income patients are less likely to begin chronic therapy with generic medications, and why these findings may not apply to seniors.
Although our patient sample was relatively young, we found that older patients were more likely to use generic medications, even after controlling for the overall number of prescriptions filled. Older patients may be more experienced medication consumers and more knowledgeable about purchasing options. Nonetheless, further study is needed in a senior population before generalizing these findings to Medicare Part D beneficiaries as the cognitive challenges and financial constraints may lead to different choices.
The limitations inherent in our use of pharmacy claims data may have introduced some bias. We are unable to identify prescriptions written, only prescriptions filled. Some patients, when informed of a high copayment at the pharmacy, may experience “sticker shock” and choose not to fill the initial prescription; others may switch prescriptions before the first prescription is filled. Prescriptions that are switched before the initial fill or that are abandoned at the pharmacy because of cost may lead to some misclassification at the physician level. In addition, low-income patients may be less likely to fill high copayment prescriptions, leading to inflated estimates of the proportion of generic prescriptions written. As a result, low-income patients may be even less likely to receive prescriptions for generic drugs than these findings would suggest. We also are unable to monitor use of free drug samples, which may have led some patients to subsequently fill prescriptions for more expensive medications.33
In addition, we could not account for perceived efficacy of branded versus generic drugs and could not evaluate how those perceptions influenced use.
Our study was also limited by the quality of physician identifier information from pharmacy claims. While scant literature exists to define the accuracy of DEA numbers in claims data, approximately 20% of our claims did not include a DEA number, and some may have been incorrect. It is unclear how this may have biased our study, but all other findings in this study were robust to removal of the physician variables and inclusion of claims with missing DEA numbers, so this limitation did not appear to qualitatively influence our results. Moreover, sensitivity analyses controlling for type of degree (medical doctor vs doctor of osteopathy or physicians assistant) did not indicate significant differences in generic prescribing between groups. In addition, we did not control for individual drug level characteristics, which may have influenced prescribing choices. However, we did control for drug class so individual drug characteristics were unlikely to bias the findings across classes.
Overall, these findings suggest that tiered pharmacy benefit plans and mail-order pharmacies steer patients towards generic drug use after initial brand name prescriptions are filled. However, the initial choice of prescriptions is the strongest determinant of subsequent use, and patients living in the poorest zip codes were least likely to initiate treatment on generic drugs. Efforts to influence patients and physicians to choose similarly effective generics should focus on this initial decision. As the nation struggles with increasing pharmaceutical costs, providing patients and physicians with information about generic alternatives at the time of prescribing may help patients and the nation get the most of out of their drug expenditures.