Discussion
The intervention used in this study is unique in that it expanded roles and lines of communication for hospital-based and community pharmacists in a traditional, nonintegrated US health care system. Although the primary end point of patient self-reported adherence was not statistically significant, the prespecified secondary end point using prescription refill records documented both overestimation of adherence by patients' self-report and a trend toward improvement in adherence to β-blockers and statins with the intervention. These results fulfill a tenet of recently proposed healthcare reforms that require improved continuity of care transcending traditional institutional and practice boundaries. Widespread implementation of this type of intervention would require practice and policy changes in the health care system. This study tested the principle that hospital and community pharmacists can assist in improving continuity of treatment between the hospital and community setting and in monitoring and supporting adherence to life-saving medications. Further corroboration of this finding would be necessary before health system changes could be recommended to support changes in the pharmacist's role.
Community pharmacists are uniquely positioned to interact with patients at more frequent intervals than most other health care providers and can promote effective medication-taking behavior through a variety of mechanisms including education, adherence aids, support, and feedback/reminders. They also have immediate access to medication refill information to ascertain medication adherence. However, because they are typically outside of closed health care systems, they do not usually have access to a patient's medical history, which may limit the extent of service that they can provide. In addition, community pharmacists do not typically engage prescribers nor are they engaged by prescribers to identify and resolve adherence issues. Thus, the pharmacist has access to medication adherence patterns but not medical history, and the prescriber has access to medical history but not medication adherence patterns, leading to a less-than-optimal system. This study sought to bridge the gaps by establishing virtual networks among hospital and community physicians and pharmacists for each patient. Previous studies using pharmacists to improve patient medication adherence have been typically limited to a single health care setting (hospital or community), small number of participating pharmacies, or single pharmacy chain or were conducted outside the United States in health care systems with preexisting integration between hospital and community providers. To our knowledge, this is the first intervention of this type and magnitude to be implemented and tested to improve medication adherence outside a closed health care system.
In considering replication and expansion of this intervention, there are several lessons about process and evaluation that may be useful. The establishment of the virtual network required some health care providers, such as the study pharmacist and community pharmacist, to alter their scope of work or the delivery of services. Within the context of a research study, these new or expanded activities were encouraged, implemented, and funded outside the existing health care infrastructure. To embed these new activities within current health care processes, changes would be required to obviate the need both for continual re-engagement/education to account for pharmacist turnover in community settings and for external payments for "extra" services such as provision of medication therapy management. Participating health care providers may be more receptive to new relationships with other health care providers and new processes if the relationships became part of standard operating procedures rather than a component of a research study.
Another important finding was the discrepancy between proportion of patients determined to be adherent using the 2 assessment methods. Adherence as determined by prescription refill data was substantially lower than that determined by patient self-report of use at one point in time. Although these are not necessarily new findings, the substantial discrepancy that was documented is important to communicate to those about to embark on research on medication adherence. Although neither assessment method measured actual day-to-day use, achieving adherence as measured by refill records required more evidence of effort by the patients than affirming verbally that they had drug available to them at a single point in time. Adherence to cardiovascular medications was found to be very low in this population of patients, which was very diverse and had substantial rates of comorbid conditions, including hypertension, diabetes, renal disease, cerebrovascular disease, heart failure, obesity, depression, and smoking. These results not only highlight the need for interventions to improve adherence but also emphasize the need to use robust methods to assess adherence in research studies and clinical practice.
There are several limitations to this study. The choice of patient self-report of adherence as the primary end point and the noted discrepancy between self-reported adherence and adherence determined by refill records have already been described. The study also was not able to recruit the targeted number of patients, resulting in lower power to detect a difference between the study arms. In addition, because follow-up interviews were not available for all patients with prescription refill records, physician-directed medication discontinuations were not able to be included in the assessment of adherence using refill data. Thus, patients would have been considered nonadherent if their physician had purposely discontinued β-blocker or statin, resulting in an overestimate of nonadherence by that measure in our analysis.
Despite the lack of a statistically significant difference in self-reported adherence between the intervention and control arms, the results of the prespecified secondary end point using prescription refill data as the method for assessing adherence provide encouraging evidence to support further evaluation of such a multifaceted, combined hospital and community-based strategy to improve medication adherence.