Study design
We conducted an 8-week, parallel-arm, randomized trial to evaluate a targeted patient-centered pharmacist–physician team MTM intervention (“targeted MTM intervention”) to improve anticholinergic medication appropriateness and reduce the use of inappropriate anticholinergic medications in older patients enrolled in the ADC cohort. We defined inappropriate anticholinergic medication use in a two-step approach. All anticholinergic medications taken by each participant were labeled “potentially inappropriate” and were subject to review by the study team. The second step was conducted during the targeted MTM intervention when each of the previously flagged medications was evaluated using a risk–benefit approach with final recommendations based on the participant’s input and preference. The study was approved by the Institutional Review Board at UK. All participants provided written informed consent.
Inclusion and exclusion
Patients were considered for inclusion in our study if they met the following eligibility criteria: actively enrolled in the ADC cohort; 65 years of age and older; reporting at least one drug with anticholinergic properties [15,16,17] at their annual ADC visit; and willing to participate in our intervention study. Patients were excluded if they had moderate or severe dementia as measured by a Clinical Dementia Rating (CDR) [18] global score ≥ 2, or lived in a long-term care facility at the time of enrollment.
Participant recruitment
In short, between October 1, 2014 and September 30, 2015 we screened the records for participants actively enrolled in the ADC cohort within 1 week of their scheduled annual visit. We identified study-eligible patients who reported at least one anticholinergic drug and mailed them a letter briefly introducing our study. One to two weeks after mailing the letter, the potential participant received an enrollment call from the study team; once verbal consent for participation was granted, the medication list was finalized and additional details about the medications taken were collected. Between the enrollment call and the first study visit (2 weeks after enrollment), the pharmacist evaluated the appropriateness of the anticholinergic medication(s) and outlined points to address in case the participant was randomized to intervention.
Randomization
After enrollment in the study, patients were randomized to either the intervention or the control group using a simple block randomization scheme (50 subjects randomized into five blocks), which was generated using the web site Randomization.com (http://www.randomization.com). The study statistician prepared 50 sealed opaque envelopes containing the sequential randomization assignments, and these were provided to the study principal investigator, who had no knowledge of the randomization assignments contained within. Each envelope was opened only after the study pharmacist completed the baseline medication review. Because the intervention was educational in nature, complete blinding of the intervention was not possible. However, we attempted to minimize potential bias and achieve the maximum level of blinding possible by this design. Specifically, when reviewing the medication list prior to the intervention, both the study pharmacist and the licensed prescriber at the ADC were unaware of the group allocation. In addition, data analysis was blinded to the intervention.
Study procedures
There were two total study visits for both the intervention and control groups (Fig. 1). At the first study visit, all participants were provided with generic information available from the US Food and Drug Administration, encouraging patients to be proactive in talking to their health care providers about their medications (http://www.fda.gov/drugs/resourcesforyou/ucm079453.htm). In addition, those participants randomized to the intervention group met with the pharmacist–clinician team that conducted the targeted MTM intervention. The end-of-study visit was scheduled 8 weeks after enrollment. At this visit, all participants provided updated information on their medication use to allow us to determine any changes from baseline, and completed the Short Form Health Survey (SF-36) and the end-of-study questionnaire. Those in the control group were given the opportunity to discuss any questions or concerns about their medications with the study pharmacist.
Intervention
The targeted MTM intervention was based on the pharmacist–clinician team drug review between enrollment and visit 1. Typical MTM interventions evaluate all medications used by a patient and determine treatment necessity and potential changes; our intervention only targeted medications known to have anticholinergic properties. For patients randomized to the intervention group, the study pharmacist provided a revised medication plan based on the drug review, which was discussed with the participant and/or their Legally Authorized Representative. Specifically, the proposed plan attempted to recommend discontinuation or replacement of any potentially inappropriate drug with anticholinergic properties, with safer drug alternatives (i.e., with less or no anticholinergic activity). When drug alternatives were unavailable, reduction in dosage was recommended whenever possible to reduce the anticholinergic burden [19]. Similar to routine clinical practice, the study clinician ultimately made the recommendations about prescription changes to the participant, while the study pharmacist was responsible for recommendations and provision of information to educate the participant about medication safety and the importance of patient involvement in medication awareness and oversight. Appropriate changes were determined by the licensed prescriber, but the participant had the freedom to accept or reject the recommendations.
Study endpoints
The coprimary endpoints measured the impact of the targeted MTM intervention on potentially inappropriate anticholinergic use by evaluating change from baseline to end of study in: appropriateness of anticholinergic medication prescribing, as measured by the medication appropriateness index (MAI) [20]; and anticholinergic burden as measured by the number of anticholinergic drugs used and the anticholinergic drug scale (ADS) [16].
The MAI rates each medication based on 10 different criteria related to indication, effectiveness, dosing and medication-taking behavior, potential for drug–drug or drug–disease interactions, duplication of therapy, patient acceptability of the medication, and whether the medication is the least expensive option for the specific indication. Each criterion has explicit instructions and examples to guide evaluation, and the study pharmacist rates whether the particular medication was “appropriate”, “marginally appropriate”, or “inappropriate”; the final MAI score can range from 0 (totally appropriate) to 18 (completely inappropriate) [20]. MAI assessments made by a clinical pharmacist and a physician (i.e., internist and geriatrician) demonstrate high inter-rater (κ = 0.83) and intra-rater (κ = 0.92) reliability [20]. Our focus was on drugs with anticholinergic properties, either obtained by prescription or over the counter. Medication appropriateness at baseline was assessed in a blinded manner, before randomization.
Anticholinergic burden (and change from baseline to end of study) was measured using the updated version of the ADS score. The updated version of the scale was obtained from the lead author; changes from the original scale include reassigned scores for some medications (e.g., change from ADS = 0 to ADS = 1 for loratadine, or from ADS = 1 to ADS = 0 for oxazepam), and new scores for medications that were unavailable or unassessed at the time the original scale was published. The ADS has four levels for each included drug, ranging from 0 (no known anticholinergic activity) to 3 (markedly anticholinergic activity) by comparison with serum anticholinergic activity [16]. Of the existing anticholinergic scales, the ADS categorized the largest number of medications for anticholinergic activity [21]. The summation of anticholinergic activity level for all of the drugs taken by a patient reflected the total anticholinergic burden for the participant, with higher scores indicating higher burden [16].
In addition, we created a composite binary outcome measure to incorporate a change in dose when treatment discontinuation was not possible. Specifically, a reduction in anticholinergic medications (yes/no) was defined in the case of either discontinuation or dose reduction.
The secondary endpoint included the change in perceived health status from baseline to the end-of-study visit as measured using the SF-36, a validated instrument that evaluates eight health domains categorized into three major health attributes: functional status (i.e., physical functioning, social functioning, role limitations due to physical problems, role limitations due to emotional problems); well-being (mental health, vitality, bodily pain); and general health perception (an overall evaluation of health) [22]. In addition, these eight health domains can be grouped into two component scores: physical and mental health [23]. Previous research reported that the SF-36 correlated well with the Sickness Impact Profile scores, a more thorough health status evaluation [24, 25]. SF-36 summary scores range from 0 to 100, with lower scores indicating poorer perceived health. Thus, a positive change from baseline to end of study indicates improvement in perceived health. All participants completed the baseline SF-36 as part of their annual ADC assessment within the 2 weeks prior to enrollment and again at the end-of-study visit.
At the end-of-study visit, we also asked participants to complete a questionnaire asking about their experience as part of the study and how they perceived the intervention. We also investigated reasons for participation, and whether our study impacted the pattern of communication between study participants and their health care providers (i.e., physician and or pharmacist). The additional file includes the end-of-study questionnaire as used by our participants (see Additional file 1).
Statistical analysis
We used Student’s t tests (or Wilcoxon rank-sum tests for nonnormally distributed variables) and chi-square or Fisher’s exact tests to assess the comparability of study groups after randomization. To examine the effect of the intervention on prescribing appropriateness measured using the MAI and ADS, we performed analysis of covariance (ANCOVA), with the dependent variables being the difference in scores between baseline and the end-of-study measure. The impact of the intervention on reducing the number of anticholinergic medications from baseline to the end of study was assessed using Poisson regression that also accounted for the number of medications the participant was taking at baseline. For our composite binary measure of reducing anticholinergic load, we conducted logistic regression; to evaluate the robustness of this measure of change, we also conducted a sensitivity analysis using logistic regression restricted to participants using moderate or strong anticholinergic medications. For the perceived health status measure, because the SF-36 does not produce a single overall measure, analysis of covariance was used to estimate the effect of the intervention on the eight SF-36 health concepts and the two component scores. For all of the analyses, our a-priori statistical analysis plan considered controlling for any variable that might have been significantly different between the two groups after randomization.
Power calculation
Based on previous studies using the MAI as the outcome of interest, we calculated the sample size to detect a clinically relevant mean difference of 1.0 between baseline and the end-of-study assessment for the intervention group. We estimated that 34 subjects in total (17 per group) would be sufficient to detect this difference with 80% power at a significance level of 0.05. This was a rather conservative approach because previous studies showed that medication reconciliation interventions can determine a mean MAI change ranging between 1.9 and 17 [26]. In order to account for the potential loss to follow-up, we planned to enroll 25 participants for each group, for a total of 50 participants.