Polypharmacy prescribing guidance - draft: consultation

19. Appendix G: Methods for deriving evidence-based guidance

Study design

Due to the extensive scope of the Polypharmacy Guidance, a rapid guideline development approach was used to derive evidence summaries and evidence-informed recommendations. This adopted an overview of systematic reviews (SR) and guidelines (GL), to accelerate the efficient and reliable synthesis of the evidence base.

The process below lists all steps in the process for one hot topic, from defining the research question to completing the guidance hot topic section. This process was replicated for each hot topic.

Process:

• Devise PICO and research questions
• Devise search strategy
• Devise generic eligibility criteria
• Run searches
• Study selection
• Assess quality
• Data extraction
• Evidence summary
• Considered judgement
• Final report and recommendations
• Guidance Hot Topic selection

The following details the above steps in the process. This uses the section of Polypharmacy in Frailty as an example of the general approach taken for all other Hot Topics.

Search strategy

Iterative systematic literature searches were carried out to identify guidelines and systematic reviews, from October 2021 to December 2022, using the following databases:

• BMJ Best Practice
• Cochrane Library
• DynaMed
• Emergency Care Research Institute (ECRI)
• Embase
• Google
• Guidelines International Network
• Medline
• National Institute for Health and Care Excellence (NICE)
• NICE Evidence Search
• National Institute for Health and Care Research (NIHR) Journals (Health Technology Assessment - HTA, Health and Social Care Delivery Research, and International HTA database)
• Scottish Intercollegiate Guideline Network (SIGN)
• The Knowledge Network
• Turning Research into Practice (TRIP) Medical Database

An example of the search strategy for polypharmacy and frailty is shown in Figure 15. The reviewer conferred with subject experts to seek advice on relevant articles and checked reference lists.

We ran a further update search to the end of October 2023 – see Appendix H.

In addition, in conjunction with the forthcoming update to SIGN’s Chronic Pain Guideline, we reviewed the additional evidence used by SIGN and applied this to the chronic pain hot topic.

Generic eligibility criteria

For each section the PICOSS criteria was used to determine the search parameters. Generic eligibility criteria guided sifting of references for inclusion/exclusion.

Inclusion (using polypharmacy and frailty example):

• Population: adults at risk of polypharmacy, by virtue of having one or more long-term conditions, e.g. hypertension and diabetes. Older adults were specified as 65 years and older
• Intervention: medication management, (e.g. a medication review or deprescribing), which could relate to a range of medicines: e.g. antihypertensives, diabetic medication, statin, medication for mental health
• Control: usual treatment
• Outcomes could include:

• effectiveness of treatment, e.g. for blood pressure; mortality; or cardiovascular events
• harm of treatment, i.e. medication safety and adverse drug reactions
• related to medication management, i.e. medication review or deprescribing
• evidence-based recommendations (for guidelines)

• Settings: any setting, but for specific questions, care homes or private households
• Study design: guidelines or systematic reviews

• Timespan: 2016 or later. For Chronic Pain it was 2018 or later as building on SIGN 136156

Exclusion:

• Non-English language

Assessment of quality

For the appraisal of systematic reviews, the SIGN Methodology Checklist: Systematic reviews and meta-analyses^[363] was used. This first requires an analysis of systematic reviews in relation to 12 questions. Based on these judgements, the methodological quality of systematic reviews was assessed; ranging from high quality, through acceptable, low quality to unacceptable. Where SR and GL were considered poor quality, these were not excluded but, instead this was noted, and less weight was given to such reviews in the analysis.

The quality of guidelines was assessed using the Appraisal of Guidelines for Research and Evaluation (AGREE) II tool.^[364] This makes a judgement on the guidelines in relation to 23 questions across six domains. To expedite the process, scoring was only completed across two domains - scope and purpose, and rigour of development. The other domains’ questions were less relevant to how the GL authors conducted the methods to derive their evidence and conclusions. The overall quality of guidelines are rated on a scale of poor, mixed or good – see Table 40.

Table 40: Rating overall Guidelines

Poor quality

Poor qualityDevelopment of the guideline was not well reported or not conducted

Mixed quality

Some aspects of rigour of development were missing, but no further flaws

Good quality

All aspects of rigour of development reported, and no other obvious faults

When necessary, a judgement was made on the quality of the primary research that was reported by the SR, or in part formed the basis of the guideline evidence (which might have been a Randomised Control Trial (RCT)). The assessment of the quality was reported as free text in the body of the evidence summary.

Study selection, quality assessment, and data extraction

One reviewer sifted the titles and abstracts, and full texts of the retrieved articles. Two reviewers independently assessed the quality of the retrieved articles and reconciled differences. Reviewer one extracted information and data and a second Health Services Researcher checked, with any differences reconciled.

Analysis

All included reports were categorised in accordance with the section of the guideline. A narrative synthesis was conducted, as the outputs/conclusions of the GL and SR were too diverse to produce a meaningful summary of effects, using a meta-analysis.

Considered judgement

The considered judgement method as set out in SIGN 50^[365] was adapted to provide a guideline recommendation from the evidence. The short life working groups were provided with a summary report detailing:

• the quality of evidence of each guideline and systematic review
• a summary of the evidence extracted
• draft recommendations
• evidence tables

• quality assessment documentation

Based on these summary reports, the groups accepted, adapted, or added recommendations. The strength of recommendations was then decided based on the evidence. The short life working groups agreed a specific set of wording for writing the judgements using should, could, or consider - see table below.

When there was no evidence enabling recommendations, e.g. to identify the optimal blood pressure level for hypertension in frailty, the short-life working group derived good practice points. These good practice points were derived by reaching consensus based on the expertise and experience of the group.

Write up of final guidance

With the evidence tables, quality assessment reports, and evidence summaries, provided by the Health Service Researcher and the recommendations as derived above, subject experts then drafted each section of the guidance. The wider short-life working group reviewed each section and agreed recommendations. An editorial group then produced the final draft.