Systematic reviews were the perceived most important source of information for updating a potentially inappropriate medication list for the elderly: An online survey

Abstract

Polypharmacy in the elderly is a recognized public health problem. For many diseases several drugs are available. Thus, it is important to rule out drugs with an unfavorable risk-benefit profile in older patients, and to suggest safer alternatives. This is often done by developing and publishing lists of potentially inappropriate medications (PIM).

Making evidence-based decisions for the elderly is often challenging as randomized controlled trials frequently do not include older adults with multimorbidity and are not sufficiently powered for analyzing harms. Despite many PIM lists having been published, there is no methodological guidance on how they should be developed, what we recognized while preparing our update of the German PRISCUS list.¹ Most PIM lists use the Delphi method to find consensus.^{2, 3}

When using a Delphi study to develop a PIM list it is crucial to consider what information will be provided to the participants of the Delphi study to formulate recommendations. There is no research on what type of information experts need or prefer to support rating of PIM criteria or derive recommendations for PIMs. Our aim was to investigate the process of providing information and the importance of different sources of information in the context of developing a PIM list.

Our study was carried out as part of updating the Potentially inappropriate medications list in the elderly in Germany (henceforth termed “the PRISCUS list”).¹ This included a three-round Delphi process to derive the list of PIMs. In total, 95 experts from clinical practice and research from general medicine, geriatrics, clinical pharmacy, pharmacology, psychiatry, palliative medicine, and internal medicine evaluated whether selected substances are PIM for the elderly. Due to limited resources, we were only able to conduct 13 de novo SRs. Altogether, 187 substances were classed as PIM.⁴

We conducted a closed follow-up survey of all experts who took part in the assessment of PIM for the update of the PRISCUS list. The survey was administered in German via SurveyMonkey (https://de.surveymonkey.com/). We sent out an invitation email including a hyperlink to all experts who took part in the assessment for the update of the PRISUCS list. The follow-up survey took place between May and June 2021. One reminder was sent via email to all experts.

Participants were told that taking the survey would not take longer than 5 min. There was no incentive for participating in the survey. Collected data was imported into an Excel spreadsheet where all analyses were conducted. The number of participants taking part in at least one Delphi round was 59. These participants were invited to take part in our follow-up study. Of the 59 invitees, 40 (68%) responded, that is, accessed the survey. Of those (n = 40), 36 (90%) completed the entire survey.

Systematic reviews including summary of finding tables were valued most by the respondents (mean 4.19) (Table 1). The vast majority of participants (86%; 30/36) either agreed or fully agreed with the statement that it was easy to understand the provided information, while only one person did not agree with that. All but one participant agreed with the statement that the information was provided in a way that helped to inform decision-making of the participants. Only 22% (8/36) agreed with the statement that they would have given the same ratings without having obtained the provided information. Most participants (44%; 16/36) partly agreed with this statement (Table S1).

Our findings suggest that most of the provided information was very well perceived and had a high impact on rating the PIM. SRs including summary of findings tables were found to be the most valued information source out of six different sources, interestingly, followed by information about anticholinergic effects. We are not aware of any study we could directly compare our results with. Our study is the first showing results of a follow-up survey of Delphi participants in the development of a PIM list.

The most striking finding of our study is that SRs were the most valued source of information. Conflicting with this, SRs are usually not used when developing PIM lists. Some PIM lists exist claiming that there performed a literature review.⁷ However, it is mostly unclear what the literature review was about as no details are provided. We cannot preclude that in some cases SRs were searched for, while it seems very unlikely that SRs were performed de novo to inform the development of the PIM list as this would have been reported. The most recent update of the Beers list from 2019 highlights the evolvement of methodological approaches to develop PIM lists.⁸ An updated literature review was performed with much more details reported, but it cannot be reproduced. Also, important to note that for the first time GRADE assessment was mentioned to have been performed when developing the Beers list. To the best of our knowledge no other PIM list was developed using the GRADE criteria. However, no further details or assessments of GRADE are provided along with the published Beers list. This is a pity as we found the usual GRADE criteria to be in need of modifications to meet the purpose of developing a PIM list.⁶

The second most valued information was about anticholinergic effects, which is a specific indicator for potential side effects affecting cognition and resulting in a diverse range of symptoms all of them of particular relevance for older adults.^{9, 10} Thus, it is not surprising, that this information was also highly appreciated and is considered to be particularly relevant in addition to SRs.

PIM lists provide recommendations for practice. In this perspective, they are very similar to clinical practice guidelines (CPGs). However, CPGs tend to be based on SRs including GRADE assessments. Transparency is called for when developing CPGs to ensure trustworthiness. Along with the recommendations all details such as literature reviews and GRADE assessments are usually reported. We do not see any reason why this should not also be the case for future PIM lists because they are very similar in their intended use. This appears particularly important because of their high impact on practice and ability to reduce the number of PIMs prescribed.^11-13

Overall, the respondents were very positive how the information was presented to them. Nevertheless, there still seems to be room for improvement. There is no knowledge how experts derive decisions based on the information provided to them. The clinical expertise of the experts is also of importance and plays a major role. There was no one participating in the survey who fully agreed that they would have made the same decisions without the provided information. Understanding GRADE SoF tables can be challenging for readers. Thus, ways of presenting SoF tables are also subject of research.^{14, 15}

Our results might be limited by a small number of respondents. However, a comparison of PIM lists found that the number of experts ranged from 4 to 62, and approximately one in two PIM lists had only up to 20 experts.³ Thus, our results need to be treated cautiously, although the sample size can still be regarded high when compared to other PIM lists.

Investigators should in particular pay attention to the question what information is offered to experts. Currently, the development of PIM lists rather follows a “one size fits all approach.” Given that SRs will not be feasible to be conducted for all potential drugs, future research needs to focus on what question SRs would be most important.

We conclude that SRs including SoF tables are good to understand and provide a valuable source of information in the development of PIM lists. Developers of PIM lists and comparable outputs should aim to base their recommendations on SRs of adverse events but need to balance resources, timelines, and methodological rigor.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The study was embedded in the update of the “Potentially inappropriate medications list in the elderly” funded by the German Ministry for Education and Research (BMBF 01KX1812).

DP and TM are members of the GRADE working group.