Medication adherence [MA] is a complex issue depending on both patient and healthcare factors. Non-adherence due to patient-dependent factors is common; analyses of large broad-based clinical trials in the general population demonstrate that 20-30% of study medication doses may be missed. In community-dwelling HF patients with polypharmacy this percentage is likely even higher. One of the factors in poor adherence in elderly non-HF patients has been shown to be cognitive impairment [CI]. Non-adherence to medical regimen is one factor that is associated with worse outcomes in patients with HF, including readmissions. Measures that improve MA are also likely to improve outcomes.
We evaluated a novel intervention incorporating new technology with the goal of improving adherence in patients with HF and CI. Specific objectives of the study were (1) to assess the feasibility of using the Automated Medication Dispensing Device (AMDD) in veterans with HF and CI by measuring patient qualifying rate, consent rate, user rate, patient-level response rate, medication-level response rate, and success rate, (2) to calculate the improvement of MA based on pill counts before and after the introduction of the AMDD, and (3) qualitatively describe patient satisfaction with the AMDD and the reasons for not using the AMDD.
This pilot was designed as a feasibility study testing an intervention using a commercially available, off the shelf Automated Medication Dispensing Device (AMDD) with subjects serving as their own controls. The study was conducted at the VALLHS, Loma Linda, CA which serves a population of 246,000 veterans. The study planned to enroll 50 patients with HF and CI (defined as SLUMS score of <27 in a person with high school education or <25 in a person with less than high school education) from the outpatient HF clinic. Baseline adherence to prescribed medications was measured by an initial 30-day pill count (month 1). Subjects with baseline adherence <88% were given the AMDD to use for 90 days (one month to familiarize the study subjects with the device, two months to collect outcome data). This dispenser was pre-filled by home health nurses once a month; the dispenser is equipped with an alarm mechanism that alerts the patient to take the medications; when the patient pushes the button on the device, it will dispense the medications. 30-day pill counts were performed while patients were using the AMDD on months 3 and 4. The study was powered to detect the success rate of the AMDD, defined as both patient acceptance and a clinical response to the AMDD.
Total of 55 patients were enrolled in the study. Enrollment was poorer than anticipated and after we failed to meet enrollment targets for several months, screening was expanded to all outpatients at VALLHS with the diagnosis of HF. Due to enrollment difficulties, the study was continued for another 6 months past the initial end date.
Objective 1. Feasibility.
Study qualifying rate. Of the 734 patients screened, 364 qualified for the study based on the present diagnosis of stable HF and living arrangements. Qualifying rate: 50%
Consent rate. Of the 364 qualified patients, 122 could not be contacted. Some patients were disqualified at this point because of the lack of land telephone line, or because they had previously had normal screening test for cognitive impairment. Overall, 55 patients consented for the study. The most common reason given by patients for not participating in the study was the belief that they didn't need help with taking their medications, followed by unwillingness to stay in their home for the medication dispensing times. Consent rate: 23%
AMDD qualifying rate. Of the 55 patients who consented for the study, 14 did not have cognitive impairment based on SLUMS test and 10 patients, while cognitively impaired, had medication adherence >88% based on 30-day pill count. 6 patients dropped out of the study in the first phase without finishing the baseline pill count. AMDD qualifying rate: 45%
User rate. Of the 25 patients who agreed to set the AMDD up at their house, 14 ended up using it for the full study period. Reasons for not wanting the AMDD in the house varied without a common theme found. User rate: 56%
Objective 2. Medication adherence (exploratory analytical aim).
In 14 patients who completed the study, overall adherence to medications at baseline was 69.9+/-12% (mean +/- SD) as determined by 30-day pill count at study month 1. The overall adherence to medications at month 3 (after one month was given to the patients to familiarize themselves to the use of the AMDD) was 96.5 +/- 3.5%. This improvement in medication adherence was statistically significant as evaluated by paired t-test (t-statistic 8.34, p<0.001). Medication adherence improved even further at month 4 with overall adherence of 98.4 +/- 2.3, with the difference between month 3 and month 4 of borderline statistical significance (t-statistic 2.12, p=0.05).
Objective 3. Satisfaction with the AMDD (qualitative aim).
Overall, patients who used the AMDD reported high satisfaction with the AMDD (86% very satisfied), and no technical problems with the AMDD. However, 57% of patients still did not think they needed it for their medication compliance.
Average time to set up and fill the AMDD by a visiting nurse was 39 minutes. 27% of the nurses reported technical problems with the AMDD, which were all resolved by calling the company's support line. 64% of the nurses were likely to recommend this device for other patients.
Problems/suggestions reported by both patients and the nurses included inflexibility with the medication dosing schedule, high cost of the device, and the need for electricity (device stops working) and the land line.
Cost for setup per patient was $50 with the monthly service fee while using the AMDD of $65.
Based on the feasibility data, we conclude that the AMDD is likely not going to be widely used in the veteran HF population. Our low enrollment for the study reflects the fact that many patients with HF, while having cognitive impairment (69% in our previous pilot study that included 251 patients in our HF clinic), and poor baseline medication adherence (80% on average in our previous study population, 70% in this study), do not admit the need for assistance with their medications. Patients should first be tested for the presence of cognitive impairment and their baseline medication adherence measured; only then can there be a meaningful discussion about the need for assistance and the usefulness of the AMDD. Additionally, many of the veteran patients do not qualify for the AMDD due to the lack of a land phone line.
For patients who have CI and poor baseline medication adherence, the AMDD could be considered to improve medication adherence. In our study the use of AMDD, while somewhat cumbersome initially, significantly improved medication adherence for patients who continued to use it as prescribed.
Our study shows that it is feasible to use the AMDD in a selected group of veteran patients with chronic HF, in order to improve their medication adherence. In this group, the satisfaction with the device was high. In this selected group of patients, AMDD significantly improved medication adherence.
Because of the cost of the AMDD and the man-hours involved in setting up the AMDD, we believe that it should be offered only to the patients who have clearly demonstrated the inability to follow their medication regimens and have been hospitalized due to medication noncompliance. Feasibility results from our study and lessons learned in patient recruitment will be useful when considering AMDD in other patient populations (such as traumatic brain injury, geriatric patients with multiple comorbidities).
Further improvements in the setup of the AMDD (e.g., use of wireless signal instead of land line, batteries for backup power) would increase the eligible population for this device.
None at this time.
Treatment - Observational