This study assessed the association between the presence of pharmacists and the proportion of diabetes patients adherent to OHAs within VA primary care clinics. The results indicate that adherence scores are not significantly better within clinics that have a pharmacist present, regardless of the amount of FTE pharmacist staff available. However, the perception of pharmacy services as a bottleneck was associated with lower clinic-level medication adherence. While the percent of variation in clinic-level adherence explained by each pharmacy measure was generally small (< 5%), relative to pharmacist presence measures, the perception of the pharmacy as a bottleneck had substantially greater explanatory power.
Our clinic-level findings generally differ from prior patient-level studies demonstrating the value of having pharmacists present in primary care clinics
]. Most often, these studies evaluated a specific function or task related to the pharmacist as a mechanism for improving diabetes care. In contrast, the current study evaluated pharmacist presence and therefore characterizes their impact on adherence in more “real world” settings.
As the structure of clinics changes towards a multi-disciplinary team approach using the PCMH, there will be an increasing component of healthcare provided by the non-physician care management team
]. The PCMH calls for pharmacists to be part of this team and pharmacists have been included in prior PCMH demonstration projects
]. The goals of including pharmacists in the PCMH are to optimize medication regimens, improve management of chronic diseases, and lower barriers to adherence through assisting in refilling and renewal of medications
]. VA is working towards complete implementation of a PCMH within all primary care clinics and has routinely included pharmacists in primary care clinics since 2010
]. Our results suggest that without a clearly defined role, simply including pharmacists in clinics may not improve adherence to OHAs.
As in other healthcare systems, several organizational factors in VA that are beyond pharmacists influence may contribute to adherence problems. For example, barriers in refilling medications could potentially mitigate the benefits of pharmacists. These barriers include trouble getting through on phone lines
], lower health related internet use among veterans living in rural locations
], and difficulty getting to clinics
]. Although the role of a pharmacist in the PCMH model should ideally include helping to decrease barriers to accessing or refilling medications, pharmacists may currently have much different job functions. For example, if pharmacists are staffing anti-coagulation clinics, spending time dispensing medications, or completing other such tasks, then they will not have the ability to engage in activities such as medication education and counseling which have been shown to positively impact medication adherence for patients with diabetes.
The association between the degree of bottleneck in pharmacy services and clinic-level medication adherence suggests factors related to pharmacy services as perceived by the primary care director impair the timely dispensing of medications to patients. This result is consistent with a prior study showing pharmacy processes such as communication between inpatient discharge team, outpatient physicians and pharmacists were associated with a reduction in health care visits and readmission
]. Although prior studies have shown the value of including pharmacists on general medicine teams
], our results suggest that the mere presence of a pharmacist is not enough to improve medication adherence. Instead, organization resources, policies and procedures must be in place in order for pharmacists to effectively perform job duties.
Our study has several strengths. Data were drawn from a large, national administrative data system that includes information on several important confounding patient-level variables. The power for our study was excellent, with a greater than 99% ability to detect a 1% difference in clinic-level adherence between clinics with and without a pharmacist. Also, we were able to link administrative data to a nationwide clinic-level survey of pharmacist availability that is rarely available in other health systems and had a 93% response rate from 250 clinic representatives.
Several limitations exist in our study. First, we are unable to distinguish pharmacists based on variation across clinics in their training, duties, or activities. If pharmacists have more clinically-oriented roles at some locations, but are limited to only dispensing medications at other locations, then our findings are biased towards the null hypothesis, masking potential pharmacist benefits in facilities that have a more clearly defined adherence-support role for their pharmacists. Future research should also examine the exact roles of pharmacists within VA and whether there is any association with medication adherence based on specific clinical functions. Second, the measures of pharmacist availability and FTEs used in the study are reported by primary care directors, but have not been validated. However, these measures are the best that are currently available because administrative data do not indicate whether pharmacists are specifically assigned to a primary care clinic. Third, the number of community-based clinics reporting the presence of a pharmacist was higher than we expected, and some clinics may have reported on pharmacists at parent VA facilities who may have little day to day involvement in that community clinic. However, the results remain the same when we conducted the analysis including hospital-based clinics only, so we cannot attribute the observed null findings exclusively to reporting on off-site pharmacists. Finally, this study was conducted prior to the implementation of PCMH in VA. Further studies should examine the functions of pharmacists in that new primary care model.