Evaluation of a pharmacist-led pilot service based on the anticholinergic risk scale

To determine primary care physician (PCP) acceptance rates of electronic medication therapy recommendations based on anticholinergic burden for high-risk elderly patients, and to evaluate potential associations between recommendation acceptance and patient-provider characteristics.

Two medical clinics within Dean Health System, an integrated health care organization comprising ambulatory surgery centers, medical clinics, community pharmacies, specialty pharmacies, a health plan, and a pharmacy benefits management company.

In this pilot service, the medical records of patients at least 60 years old who met the following criteria were evaluated bimonthly: 1) PCP visit within 2 weeks; (2) three or more inpatient hospitalizations or emergency department visits in the past year; and (3) ten or more active medications. Anticholinergic Risk Scale (ARS) scores of eligible patients were calculated, and medication therapy recommendations were sent electronically to PCPs for patients with an ARS score greater than 3. Post-visit recommendation outcomes were recorded.

Descriptive statistics were utilized to characterize patients, physicians, and recommendations. A generalized linear mixed effects model with physician specific random effects was employed to evaluate recommendation acceptance rates, and odds ratios were calculated to quantify associations between baseline patient/provider characteristics and the likelihood of recommendation acceptance. Changes in aggregate ARS scores were evaluated with the use of a paired t test.

Fifty-nine patients were included in this pilot, with 89 medication therapy recommendations made to 21 PCPs. An overall recommendation acceptance rate of 50% (95% confidence interval [CI] 37%-63%) was observed. There were no significant associations identified between baseline patient/provider characteristics and medication recommendation acceptance.

High recommendation acceptance rates were achieved with the combination of objective anticholinergic risk assessment and algorithm-driven medication therapy recommendations. The lack of identified associations between patient/provider characteristics and recommendation acceptance supports the future scalability of this novel service.