Older adults' adherence to medications and willingness to deprescribe: A substudy of a randomized clinical trial

Our study investigated the association between patients' willingness to have medications deprescribed and medication adherence. This longitudinal substudy of the ‘Optimizing PharmacoTherapy In the Multimorbid Elderly in Primary CAre’ (OPTICA) trial, a cluster randomized controlled trial, took place in Swiss primary care settings. Participants were aged ≥65 years and over, with ≥3 chronic conditions and ≥5 regular medications. At baseline, the ‘revised Patient Attitudes Towards Deprescribing’ (rPATD) questionnaire was measured. The A14‐scale measured adherence (self‐report) at the 12‐month follow‐up. Multilevel linear regression analyses adjusted for baseline variables were performed. Of the 298 participants, 45% were women, and the median age was 78. Participants reported a high level of adherence and willingness to have medications deprescribed. We did not find evidence for an association between patients' willingness to deprescribe and medication adherence. Further research is needed to explore the relationship between these concepts and to inform collaborative decisions about medicines in the context of polypharmacy.

to their medication regimen. 7The definitions and measures of adherence vary 8,9 ; however, discontinuing a medication earlier than recommended is an aspect of adherence that relates to deprescribing. 9 A national poll of American adults about deprescribing found that of those who had stopped taking a medication (without a replacement), 36% did so without talking to a healthcare professional. 10Depending on how this is viewed, it could be considered an opportunity for retrospective deprescribing or medication non-adherence.On the other hand, clinicians say deprescribing is a complex process and can be challenging to implement because older patients are often hesitant to stop their medications. 11person's beliefs and attitudes towards their medications play a key role in how willing they are to start, continue or stop taking a medication. 12,13[15][16] It is important to better understand the association between willingness to deprescribe and medication adherence, as this is key to understanding patient attitudes, priorities and preferences in the context of polypharmacy.From here, shared decisions about medicines and attempts to optimize medication use can be made.However, the complex interplay between deprescribing and medication adherence has not been fully explored in the literature.Our study investigates the association between patients' willingness to have medications deprescribed and medication adherence in a sample of older adults with multimorbidity and polypharmacy.

| Overview of the OPTICA trial
The methods and results of the 'Optimising PharmacoTherapy In the multimorbid elderly in primary Care' (OPTICA) trial (clinicaltrials.gov8][19][20] From January 2019 to February 2020, a total of 323 patients from 43 GP practices were recruited into this cluster randomized clinical trial.The last patient completed the 12-month follow-up in February 2021.In the OPTICA trial, 21 GPs with 160 patients were assigned to the intervention group and 22 GPs with 163 patients to the control group.Eligible patients were older adults with multiple chronic conditions and regularly using 5 or more medications.While GPs in the control group continued to provide usual care to their patients, GPs in the intervention group performed a structured medication review centred around an electronic clinical decision support system (eCDSS) called the 'Systematic Tool to Reduce Inappropriate Prescribing'-Assistant (STRIP-Assistant).[23] Due to the pragmatic design, data for the OPTICA trial was partly collected from participants' electronic health records (e.g., medications and diagnoses) and partly from participants or their legal representatives over the phone (e.g., quality of life, medication adherence, willingness to deprescribe and sociodemographic characteristics).The two primary outcomes of the trial were the improvement in the Medication Appropriateness Index (MAI) and the Assessment of Underutilization (AOU) at 12 months.Secondary outcomes included the number of medications, number of falls, fractures and quality of life.The results of the trial on whether the medication review intervention led to an improvement in outcomes were inconclusive.The OPTICA trial was approved by the competent cantonal ethics committee (BASEC-ID: 2018-00914).All participants or their legal representatives provided written informed consent.

| Study design and sample definition
This is a longitudinal substudy of the OPTICA trial.Data from the trial baseline and the 12-month follow-up were used for the analyses.This manuscript adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies. 24All 323 participants of the OPTICA trial were older adults (≥65 years of age), with multimorbidity (≥3 chronic conditions) and polypharmacy (≥5 medications).The present analyses were

What is already known about this subject
• Medication-related problems, high treatment burden and polypharmacy can lead to reduced adherence to medications, and they can be triggers for deprescribing-dose reduction or discontinuation of selected medicines.
• The interplay between deprescribing and medication adherence is influenced by patient-related factors such as attitudes, beliefs, perceived necessity and concerns about medications.
• Understanding patients' willingness to deprescribe or adhere to their medications is crucial for collaborative decision-making in the context of polypharmacy.

What this study adds
• Our study examined the association between patients' reported willingness to have medications deprescribed and medication adherence in older adults with multimorbidity and polypharmacy.
• The longitudinal study design allowed for a clear temporal distinction between patients' willingness to have medications deprescribed assessed at baseline and medication adherence at 12-month follow-up.
• Participants reported a high level of medication adherence and willingness to have medications deprescribed; however, we did not find any evidence for an association between willingness to deprescribe and adherence.limited to the participants for whom the patient version of the 'revised Patient Attitudes Towards Deprescribing' (rPATD) 25 questionnaire was assessed at baseline.

| Assessment of patients' willingness to have medications deprescribed
Patients' attitudes towards having medications deprescribed were measured using the rPATD 25 at baseline (Box 1).Several rPATD questions were chosen as independent variables to measure the variation in patients' self-reported attitudes towards deprescribing, rather than the global question alone that tends to have a ceiling effect. 26The variables were dichotomized creating a category with patients reporting to be agreeing or strongly agreeing with the rPATD statements and another group with patients reporting to be unsure, disagreeing or strongly disagreeing.This approach was chosen because there were very few events in some of the categories of the original 5-point Likert scale.

| Assessment of medication adherence
The A14-scale 27 was used to measure patients' medication adherence and individual barriers to adherence.This questionnaire contains 14 Likert-scale questions ranging from never (4 points) to very often (0 points).The total score ranges from 0 to 56 where a higher score represents better adherence.The score was used as a continuous outcome in our analyses and medication adherence was assessed the 12-month follow-up.

| Statistical analyses
First, the demographics and main clinical characteristics of study participants were described, by willingness to have medications deprescribed.Second, patients' baseline willingness to have medications deprescribed was described.Third, then multilevel linear regression analyses were performed using a two-stage approach with a minimally and a fully adjusted model.All models were adjusted for the clustering effect at the level of general practitioner, the group allocation and a selection of covariates.The fully adjusted model was adjusted for the following covariates: patient age, gender, educational status, living situation, number of chronic conditions, number of chronic medications, informal care received (yes/no), number of GP visits, number of specialist visits, number of hospitalizations and proxies for patient satisfaction with and financial burden caused by medication use (measured with rPATD questions 'I spend a lot of money on my medicines' and 'Overall, I am satisfied with my current medicines').9][30] Including them in the model allows to control for potential confounding.The minimally adjusted model was adjusted for the following covariates: patient age and gender.

| RESULTS
Table 1 shows participants' baseline characteristics by medication adherence.Forty-five percent of participants in the sample were women (n = 133), and median age was 78 (interquartile range: 74-83).Most participants reported to be satisfied with their current medications (93%), and 88% of patients agreed or fully agreed with the statement 'If my doctor said it was possible, I would be willing to stop one or more of my regular medicines'.More than two-thirds of participants were classified as having equal or higher than median medication adherence.Table 2 shows patients' willingness to have medications deprescribed by medication adherence.
The association between patients' willingness to have medications deprescribed and medication adherence from both the minimally and fully adjusted models are shown in Table 3.We did not find evidence for a statistically significant association between adherence and patients' reported willingness to have medications deprescribed as measured by the rPATD questions 'If my doctor said it was possible, I would be willing to stop one or more of my regular medicines', 'I would like my doctor to reduce the dose of one or more of my medicines', 'I would like to try stopping one of my medicines to see how I feel without it' or the concerns about stopping score.
Box 1 Independent variables: rPATD 25 questions about attitudes towards deprescribing.Concerns about stopping questions score, ranging from 1 to 5, calculated based on the following rPATD questions: 'I have had a bad experience when stopping a medicine before' 'I would be reluctant to stop a medicine that I had been taking for a long time' 'If one of my medicines was stopped I would be worried about missing out on future benefits' 'I get stressed whenever changes are made to my medicines' 'If my doctor recommended stopping a medicine I would feel that he/she was giving up on me'

Response items (Likert-scale):
Strongly agree, agree, unsure, disagree and strongly disagree In this substudy of a randomized clinical trial, we did not find evidence for an association between patients' reported agreement with deprescribing and medication adherence.Participants in the sample generally reported a high level of medication adherence and high agreement with hypothetical deprescribing if their doctor said it was possible.
Studies assessing adherence and self-reported agreement with deprescribing using the (r)PATD questionnaires have been conducted in Australia [31][32][33] (n = 3) and Jordan 34 (n = 1).Of the four studies, three different self-reported measurements of adherence were used: the Morisky Medication Adherence Score, 35 the Tool for Adherence Behaviour Screening (TABS) 36 and the Adherence Attitude Inventory (AAI). 37No association was found between adherence and deprescribing outcomes in two of the Australian studies of deprescribing  Number of specialists visits in the 6 months prior to trial enrolment Hospital visits in the 6 months prior to trial enrolment 0 (0-1) 0 (0-0) 0 (0-0) rPATD Q10.'I spend a lot of money on my medicines' Strongly agree 42 ( 14 interventions (one used the PATD 31 and the other the rPATD 33 ).A cross-sectional study conducted in Jordan 34 (n = 501) found that participants with high commitment to adherence (domain of the AAI) and high self-efficacy were less likely to agree with the idea of deprescribing a medication.Similar to ours, this study used the rPATD question: 'I would like to try stopping one of my medicines to see how I feel without it' to assess deprescribing.However, our study was longitudinal and theirs cross-sectional, which must be considered when comparing the results.
There are different potential reasons for why we did not find evidence for an association between agreement with deprescribing and adherence.First, the methods used to measure agreement with deprescribing, and medication adherence may lack sensitivity or accuracy.Second, it can be hypothesized that a person with high selfefficacy and/or health literacy may be more likely to be both adherent and be willing to reduce or stop unnecessary medicines, recognizing both the benefits of adhering to medicines but also the potential harms of unnecessary or inappropriate medicines.
Our study was strengthened by its longitudinal design, which allows for a clear temporal distinction between patients' agreement with deprescribing assessed at baseline and medication adherence at 12-month follow-up.In future research, it would be interesting to examine the association between these two concepts in an opposite temporal order.Furthermore, it would be useful to study the association between those two concepts not only using selfreported questionnaires but also using real-world adherence data (e.g., proportion days covered and use of electronic pill bottles) and data related to the implementation of specific deprescribing recommendations to determine if that would show the same results.
Limitations were that participants who agreed to take part in the OPTICA trial may have been more motivated about their medications than their peers.The present analyses were a substudy of the OPTICA trial, during which a medication review intervention was tested.Since the main trial findings did not show any differences in medication-related outcomes at the 12-month follow-up and since we adjusted the analyses for the group allocation, however, we think it is unlikely that the intervention has affected medication adherence at the 12-month follow-up.Agreement with b The concerns about stopping score were calculated based on five questions from the rPATD (Box 1).For factors, a higher total score indicates greater perceived concerns about stopping medicines (strongly agree = 5, agree = 4, unsure = 3, disagree = 2, strongly disagree = 1). 25A B L E 3 Multivariate associations between patients' medication adherence and their willingness to have medications deprescribed.Note: Multilevel linear regression models were performed.The fully adjusted model was adjusted for the clustering effect at the level of the general practitioner, the group allocation and for the following covariates: patient age, gender, educational status, living situation, number of chronic conditions, number of chronic medications, informal care received (yes/no), number of GP visits, number of specialist visits, number of hospitalizations and rPATD questions 'I spend a lot of money on my medicines' and 'Overall, I am satisfied with my current medicines'.The minimally adjusted model was adjusted for the clustering effect at the level of the general practitioner, the group allocation and for the following covariates: patient age and gender. a The concerns about stopping score was calculated based on five questions from the rPATD (Box 1).For factors, a higher total score indicates greater perceived concerns about stopping medicines (strongly agree = 5, agree = 4, unsure = 3, disagree = 2, strongly disagree = 1). 25eprescribing was measured hypothetically meaning participants may feel differently about real-life deprescribing decisions.Adherence was assessed using a self-reported questionnaire, which may have led to an overestimation of adherence due to social desirability and memory biases.38 It seems that the adherence measure we used (A14-scale) and the rPATD global question have a ceiling effect; therefore, misclassification bias cannot be excluded.Further, self-reportedagreement with deprescribing may have been overestimated due to social desirability bias.

AUTHOR CONTRIBUTIONS
Global question:'If my doctor said it was possible, I would be willing to stop one or more of my regular medicines' Appropriateness questions: 'I would like to try stopping one of my medicines to see how I feel without it' 'I would like my doctor to reduce the dose of one or more of my medicines'

T A B L E 1
Baseline characteristics, by willingness to have medications deprescribed.
Based on the rPATD question 'If my doctor said it was possible, I would be willing to stop one or more of my regular medicines'.(Stronglyagree/agree categorized as high willingness to have medications deprescribed; unsure/disagree/strongly disagree categorized as lower willingness to have medications deprescribed).Information on medication adherence was available for 276 patients at the 12-month follow-up.The score ranges from 0 to 56 points.
Note: Categorical variables: frequencies and percentages are presented.Continuous outcomes: median and the interquartile range (IQR) are presented.a Among the 298 patients, 146 patients were then randomized to the control group and 152 patients to the intervention group.b c T A B L E 2 Patients' willingness to have medications deprescribed.
Note: Categorical variables: frequencies and percentages are presented.Continuous outcomes: median and the interquartile range (IQR) are presented.a Among the 298 patients, 146 patients were then randomized to the control group and 152 patients to the intervention group.