Older adults’ attitudes towards deprescribing and medication changes: a longitudinal sub-study of a cluster randomised controlled trial

Objective To investigate the association between older patients’ willingness to have one or more medications deprescribed and: (1) change in medications, (2) change in the appropriateness of medications and (3) implementation of prescribing recommendations generated by the electronic decision support system tested in the ‘Optimising PharmacoTherapy In the Multimorbid Elderly in Primary CAre’ (OPTICA) trial. Design A longitudinal sub-study of the OPTICA trial, a cluster randomised controlled trial. Setting Swiss primary care settings. Participants Participants were aged ≥65 years, with ≥3 chronic conditions and ≥5 regular medications recruited from 43 general practitioner (GP) practices. Exposures Patients’ willingness to have medications deprescribed was assessed using three questions from the ‘revised Patient Attitudes Towards Deprescribing’ (rPATD) questionnaire and its concerns about stopping score. Measures/analyses Medication-related outcomes were collected at 1 year follow-up. Aim 1 outcome: change in the number of long-term medications between baseline and 12 month follow-up. Aim 2 outcome: change in medication appropriateness (Medication Appropriateness Index). Aim 3 outcome: binary variable on whether any prescribing recommendation generated during the OPTICA medication review was implemented. We used multilevel linear regression analyses (aim 1 and aim 2) and multilevel logistic regression analyses (aim 3). Models were adjusted for sociodemographic variables and the clustering effect at GP level. Results 298 patients completed the rPATD, 45% were women and 78 years was the median age. A statistically significant association was found between the concerns about stopping score and the change in the number of medications over time (per 1-unit increase in the score the average number of medications use was 0.65 higher; 95% CI: 0.08 to 1.22). Other than that we did not find evidence for statistically significant associations between patients’ agreement with deprescribing and medication-related outcomes. Conclusions We did not find evidence for an association between most measures of patient agreement with deprescribing and medication-related outcomes over 1 year. Trial registration number NCT03724539.


Discussion
Its worth mentioning that the lack of an association could also be because our interventions are not effective.Patients can have high willingness (i.e. the rPATD could be valid) and are thus a captive audience but will still have barriers to address before saying yes that need to be addressed by clinicians and by extension interventions.The qual literature says that inherently patients do not want to take medicines they don't need, which supports the high rPATD scores.Was there anything in the OPTICA process evaluation to comment on this?Either why, what could this gap be due to?
Also, what you've done here is useful and I wonder whether one of your recommendations should be that more trials do this so that we can add to the evidence regarding the link (or lack of) between willingness and outcomes, and possibly types of interventions and whether this matters.

GENERAL COMMENTS
Thanks for the opportunity to review your manuscript!Just a few comments: Page 6 -inclusion criteria included taking 5 or more meds regularly -I assuming you did NOT include "as needed" medications Page 6 -the STRIP-Assistant tool -was this developed by these authors?Was it validated?Page 6 -might be useful to reader to understand a bit better the "Medication Appropriateness Index" and "Assessment of Underutilizaton" tools (without having to pull the references)

Dr. Sion Scott, University of Leicester College of Life Sciences
Thanks for inviting me to review this longitudinal sub-study of the OPTICA trial linking rPATD data with medicines outcomes.The manuscript is very well written and methods are robust.I have only a few comments to polish.RESPONSE: We would like to thank the reviewer for this positive feedback.
1. Abstract -not sure why it mentions rPATD at baseline without also mentioning follow-up.I think you mean it was given alongside OPTICA baseline outcome data collection.Perhaps just rephrase so its clear there isn't a follow-up to 'follow'.RESPONSE: Yes, that is what we mean.Reference to baseline has now been deleted from the Exposures and Results sections of the Abstract.3. Discussion -Its worth mentioning that the lack of an association could also be because our interventions are not effective.Patients can have high willingness (i.e. the rPATD could be valid) and are thus a captive audience but will still have barriers to address before saying yes that need to be addressed by clinicians and by extension interventions.The qual literature says that inherently patients do not want to take medicines they don't need, which supports the high rPATD scores.Was there anything in the OPTICA process evaluation to comment on this?Either why, what could this gap be due to? RESPONSE: We agree the lack of association could also be due to the effectiveness of deprescribing interventions.Qualitative literature does support the notion that many older adults would prefer to take fewer medications, however, older adults also often believe their medicines are important and necessary with coexisting, contradictory beliefs about medicines.We looked into the process evaluation of the OPTICA trial, but unfortunately, we do not have any evidence explaining this gap.

Results -
We have added to the Discussion: "There is a complex interplay of factors, such as clinical decisionmaking and patients' attitudes, that are behind acceptance (or not) of deprescribing.It is possible that the lack of association between the rPATD and medication-related changes in our study was due to the inconclusive effectiveness of the OPTICA deprescribing intervention, which is similar to other deprescribing interventional studies.While it is useful to quantify attitudes towards deprescribing to get a sense of older adults' general thoughts about their medications, it may be unfair to expect selfreported attitudes to equate to actual medication changes."Also added to the Discussion: "Further exploration is needed into the link between attitudes towards medicines and actual medication changes, possibly through process evaluations of deprescribing trials."4. Also, what you've done here is useful and I wonder whether one of your recommendations should be that more trials do this so that we can add to the evidence regarding the link (or lack of) between willingness and outcomes, and possibly types of interventions and whether this matters.RESPONSE: We fully agree with the statement that currently little is known about the association between willingness to have medications deprescribed and outcomes.If deprescribing trials consistently measured willingness to deprescribe at baseline, this would shed more light on the relationship between willingness to deprescribing and outcomes.Added to the Discussion: "Further exploration is needed into the link between attitudes towards medicines and actual medication changes, possibly through process evaluations of deprescribing trials."

Prof. Mary McPherson, University of Maryland Baltimore
Thanks for the opportunity to review your manuscript!Just a few comments: RESPONSE: We selected the covariates included in the analyses based on the literature of the factors associated with number of medications/polypharmacy and the factors associated with potentially inappropriate medication use/medication appropriateness and the variables that were available in the data collected during the OPTICA trial.We would like to emphasize though that the literature shows mixed results regarding some of these variables with the direction of associations varying across studies.
We have added to the manuscript: "The included variables were based on the literature of the factors associated with number of medications/polypharmacy and the factors associated with potentially inappropriate medication use/medication appropriateness considering the data available from the OPTICA trial."7. Page 11 -Just making sure I understand this correctly -per 1-unit increase in the concerns about stopping score (and the higher the score, the more reluctant to change) is correlated with the change in number of meds from baseline to 12 mo follow up (meaning every unit higher [less excited about med changes] correlated with one additional medication prescribed at 12 months.Right?RESPONSE: Indeed, the coefficient can be interpreted as the expected change in the outcome (change in the number of medications) per unit change in the exposure variable (concerns about stopping score).The higher the score, the higher the number of medications at baseline.8. Last, just confirming there was no attempt to query the PRESCRIBER's interest in deprescribing, correct?The patient can say yay or nay all day long, but the prescriber is the one who pulls the trigger!Thanks.RESPONSE: Some data from the main OPTICA trial was collected regarding general practitioners' reasons for not implementing the prescribing implementations.We have added this point to the discussion: "Deprescribing in clinical practice and interventional studies may not occur for many reasons, such as if the general practitioner chooses not to initiate it.From the main OPTICA trial, the most common reasons for not implementing prescribing recommendations were that general practitioners thought that patients' current medications were beneficial and that the recommended change was not suitable."The reviewer provided a marked copy with additional comments.Please contact the publisher for full details.related outcomes over time.Additionally, the intervention to optimize medication was offered randomly." This sentence is a bit vague, could benefit from additional specification (or re-wording) RESPONSE: Original sentence: "Due to challenges with how data from the electronic health records of participating patients were recorded, there was some missing data on medication, which is why some participants were excluded from the analyses." This was a bit vague, we have adjusted the sentence: "Some patients were excluded from the analyses due to missing data on their medication." Formatting off (repeated words in citation?)RESPONSE: Thank you for spotting this, we have fixed it.
review was conducted in tandem with Derek Edwards, PharmD, MS, PGY-2 Pain Management and Palliative Care Pharmacy Resident at the University of Maryland School of Pharmacy.Derek has no relevant competing interests that may impact this review.Please see the attached document for the full review comments.

Table 2 -
it's quite hard to follow with the Likert scales presented vertically.Could you have them going across e.g. in other rAPTD papers: https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2706177 table 1 RESPONSE: Thank you for this suggestion, we have updated the table accordingly.