Polypharmacy and rationalisation of medications

Abstract

Growing numbers of people live with multi-morbidity. As patients’ lists of co-morbidities grow, so do the lists of their medications. Polypharmacy presents a significant challenge for both patients and clinicians. Polypharmacy can be appropriate, improving quality of life and life expectancy, but problematic, especially for patients living with frailty, and a growing burden. The task of trying to rationalise an extensive list of medications in a 10-minute appointment can seem like an impossible task. Guidance and support specifically aimed at trainees is limited. This article offers a starting point for GP trainees to stimulate their own exploration of this rapidly evolving subject.

The RCGP curriculum and polypharmacy

The role of the GP in the improving quality, safety and prescribing professional topic guide requires:

Targeted assessment of the patient and other sources

An appropriately detailed understanding of the patient’s history

Wherever possible, an agreement with the patient and their carers on the treatment proposed

That the prescription is in the patient’s best interests, correctly and clearly prescribed, and accurately documented within the patient record

Review by a suitable healthcare professional at an appropriate time

Life stages topic guide: Older adults As a GP, your role is to:

Diagnose, investigate and manage older adults taking into account theories of ageing, differences in epidemiology and risk factors of disease in the elderly population

Communicate appropriately with patients, their families and carers

Coordinate with other organisations and professionals

Review medications and repeat prescriptions effectively, potentially working with pharmacists

Ensure care promotes patients’ sense of identity, independence, personal dignity and that the patient is not discriminated against as a result of their age

Key knowledge and skills: Recognising As a GP, you should recognise that:

Co-morbidity and physical factors disproportionately affect the health of older people and will influence the management of existing disease

A problem-based approach is important, taking in the ‘big picture’, rather than a disease- based approach to the care of older people

Risks of polypharmacy (including the increased risks of significant cross-reactions and side effects)

Polypharmacy: The challenge for trainees

Polypharmacy – the use of multiple medications in a single individual – is now a routine clinical intervention in many patients. Appropriate polypharmacy (see Box 1) can improve health and wellbeing. However, problematic polypharmacy (Box 1) can create challenges for patient and GP alike. To illustrate what we mean by problematic polypharmacy we will consider the case of Bob – a frail patient with complex multi-morbidity and an extensive medication list (Box 2).

Box 1.

Defining appropriate and problematic polypharmacy.

• Appropriate polypharmacy is defined as prescribing for an individual for complex conditions or for multiple conditions in circumstances where medicines use has been optimised and where the medicines are prescribed according to best evidence

• Problematic polypharmacy is defined as the prescribing of multiple medications inappropriately, or where the intended benefit of the medication is not realised

Source: Duerden et al. (2013)

Box 2.

A consultation with Bob.

Bob is an 82-year-old gentleman. He arrives at a busy Friday morning clinic to ask for some antibiotics. He states that he wants these for his ‘dicky tummy’, as his friend is on them. Further probing reveals dyspepsia symptoms. Bob has an extensive medication list that includes three antihypertensive medications, regular co-codamol, senna, lactulose when needed and a repeat low-level oral steroid prescription. He tells you that these are for his gout and for ongoing constipation. He also takes over-the-counter ibuprofen. The notes reveal that Bob saw a rheumatologist several years ago and was diagnosed with polymyalgia rheumatic, for which steroids were started. He has since been lost to follow-up. He has had no blood tests for 12 months. Bob is unfamiliar with the polymyalgia rheumatica diagnosis and asks when he will be prescribed his antibiotics.

This consultation with Bob raises many questions. It appears that Bob is burdened with the use of regular laxatives, due to his on-going opioid use, and possibly by the use of multiple antihypertensive medications. There is a question about whether his use of steroids has contributed to his dyspepsia symptoms. Can the steroids be titrated down in light of the polymyalgia rheumatica (PMR) diagnosis? What effect has the ibuprofen had on both his dyspepsia and his kidney function? How can we educate the patient about the complex effects of these medications on the body? The patient wants antibiotics to treat his abdominal pain, now. How can we navigate all these issues for the patient? Above all else, how do we manage this in whatever is left of the 10-minute appointment slot?

Within this array of questions, a number of core GP principles are highlighted: the importance of exploring and checking a patient’s understanding and expectations, of follow-up and review of both medication and diagnoses in a patient with long-term conditions. Many of the challenges in polypharmacy consultations can be managed using good general practice consultation skills. Managing polypharmacy requires evidence-based knowledge of not only the mechanisms of action of medications, their side effects and interactions, but also an understanding of how multiple medical issues affect our patients.

Polypharmacy, however well intentioned, is far from benign and is associated with increased risks of adverse drug reactions (ADRs). A team using data from the UK Biobank project (Hanlon et al., 2018), identified that patients with chronic obstructive pulmonary disease (COPD) who were also suffering from other co-morbidities were more likely than similar patients without COPD to be prescribed medication contributing towards falls, constipation, urinary retention, central nervous system depression, bleeding and renal injury. We prescribe opioids for pain relief, which causes constipation, and in turn, we then prescribe laxatives for the treatment of constipation. However, elderly patients on laxatives are twice as likely to fall (Bloch et al., 2010). Thus, we have increased the patients’ risk of injury and hospital admission. It is not only the use of medicines associated with ADRs that is of concern to the clinician. Evidence indicates that 5–20% of prescriptions contain errors (Avery et al., 2012). The more medication a patient is taking, the higher the chance that one of the prescriptions will contain an error. Iatrogenic illness in patients with multi-morbidity can cause considerable harm. We must be mindful of how we educate patients about the risks and benefits of on-going medication, so that they are able to make informed decisions about the management of their own health conditions.

Applying the principles of evidence-based practice for appropriate polypharmacy (Box 1), using the patient-centred consultation skills of the excellent GP therefore offers us a good starting point for managing this consultation with Bob. We must not find ourselves becoming complacent with repeat medication prescribing and ensure that regular follow-ups are booked for patients to evaluate if long-term prescribed medications are still in the patient’s best interests. The value of continuity of care comes to the fore in management of polypharmacy in general practice. Box 3 offers some suggestions for the medication review process.

Box 3.

Optimising medicines: A medication review check list.

Medication reviews can be done either face-to-face with a patient, or through a review of the patient record. When you have completed a medication review you should clearly document in the patients notes for each medicine:

1. Why is this person on this medication?

Make sure you know when and why each medicine was started. Consider whether this person still needs this medication now and/or whether it will need to be stopped in the future? Ensure the reason for the medicine and planned duration of use is appropriately recoded in the patient record. Ask the patient if they know why they are on the medicine.

2. Is each individual medicine correctly prescribed?

If it is clinically appropriate for the medicine to be used, check: the dose, formulation, timing and duration of the prescription (ongoing or time bounded). Are there any errors in the writing of the prescription? Are alerts in place for blood tests and monitoring?

3. Are all medicines prescribed showing on the patients repeat prescription list?

For example, hospital-issued medication (e.g. when a patient starts methotrexate) may not show on the GP prescription list unless specifically entered

4. Red flags - do you have concerns about the collection of medicines prescribed for this individual?

Check for drug interactions (which will usually show up in alerts). Are there any high-risk medicines in the list (e.g. medicines that need monitoring, addictive medicines e.g. opiods etc.)?

Are there any potentially medically inappropriate medications? Do you need to see the patient face-to-face to explore any of these issues?

5. Is the patient using the medication as prescribed?

Is the patient over or under using their medication? Both would prompt a request for the patient to come in to discuss their medication

6. Are there things I am not sure about?

Maybe unfamiliar medication or conditions; red flags associated with complex or high-risk issues?

Do I need to discuss this with an experienced GP, a pharmacist, the hospital team? And most of all the patient?

Remember - you do not have to do all of this in 10 minutes!

Bob was fortunate to see an excellent GP trainee who reviewed his PMR diagnosis, was able to wean him off his steroid and so discontinue many of the associated medication. However, Bob still had raised blood pressure and was prescribed multiple antihypertensive medications. He came back to see the trainee (whom he now regarded as his regular GP) and reported still feeling burdened by all his tablets. ‘Doctor, do I really need all of these tablets?’ he asked.

Bob’s question now comes into the complex territory of problematic polypharmacy (Box 2). The intended benefit from a medical perspective is to control blood pressure, but Bob’s perception of benefit may differ. There may be a need for compromise between the goals of medicine and the goals of the patient (Duerden et al., 2013).

Polypharmacy: The challenge for patients

Appropriate polypharmacy applies evidence-based medical practice to improve outcomes for patients. However, a growing evidence base highlights what can be appropriate from a medical perspective, may be problematic from a patient/personal perspective (Denford et al., 2014).

Despite moving away from the age of paternal medicine and towards shared decision making, insufficient attention has been given to the demands placed on patients when trying to comply with the prescription of multiple medications. There is growing recognition of the burden for patients of taking multiple medicines, even when prescribed appropriately. The average number of items dispensed per head of population has risen from 15.5 (2007) to 20.0 (2016) (NHS Digital, 2019). This increase in prescribing is particularly high in the elderly, with 10% of over 65 year olds being prescribed over 10 simultaneous medications (Guthrie and Makubate, 2012). Patients report feeling burdened by the work of taking their medicines: 40% of people taking five or more medicines a day report feeling burdened by their medication (Krska et al., 2018), which may explain why 30–50% of medication is not taken as intended. When we start to think about rationalisation and optimisation of medication to support individual patient health and wellbeing we may need to recognise that the compliance of the patient may be more of a key to delivering optimised care than, for example, adding a second antihypertensive medication.

The patient’s perspective on their medication is an important consideration for clinicians. A patient with too many medications may frequently miss doses. Patients who become fed up with taking too many pills many choose to omit the ones that they feel offer no benefit. This may be particularly true for medication used in primary prevention. If such medication is effective, the patient will be none the wiser and never experience symptomatic benefit. Compliance can also be impaired by side effects, more likely with polypharmacy. Just because it is biomedically appropriate to prescribe, does not mean it is right for the patient. We may sometimes forget that there is more to our patients’ lives than taking pills! Patient experience is essential in developing appropriate prescribing guidelines, but often overlooked. A review of reports on medicines optimisation published by the Royal Pharmaceutical Society, The King's Fund and National Institute for Health and Social Care Excellence between 2013 and 2017 (Heaton et al., 2017) showed a paucity of attention to the issue of patient experience in published guidelines on the use of medication. In fact, only The King’s Fund considered evidence from qualitative studies of people's use of medicines. This highlights an important issue with evidence-based guidelines. Guidelines aim to inform the best care for the broad majority of the cohort in question, but this can be at the expense of truly individualised care. The good clinician knows when to deviate from guidelines to provide optimal care for the individual patient.

Finding solutions: Rethinking how we use medicines for person-centred care

So how do we find a way to optimise medication use for our patients? Individualisation of prescribing is the key to safe and effective prescribing in multi-morbidity. However, there is a lack of clarity in the literature over what direction individualisation should take (Denford et al., 2014). There is a considerable amount of research that is focused on the exciting field of pharmacogenetics, but little attention on studies involving how patients and clinicians can work together to create optimised person-centred treatment plans. Denford et al. (2014) suggest ‘mutually agreed tailoring’ as a method of developing treatment plans with patients. These plans encourage patients to take a greater role in their own care and include patients having more knowledge about side-effect profiles (including advice on which side effects may be tolerable), how to monitor medication effects at home, and having an input on what times are best for patients to take their medication.

Trainees – and indeed the wider profession – need more help and guidance to IDENTIFY those burdened rather than benefited by their medicines and so to MANAGE deprescribing when patients want to discuss stopping tablets. However, the evidence on what to do and how to do this is limited (Avery and Bell, 2019). This may explain why clinicians find it hard, in practice, to tailor medication to the individual needs of patients. A recent study examined professional barriers to tailoring medication (Reeve et al., 2018). GPs recognised the objective of starting medication, in line with guidelines and patient expectation. Stopping medication requires more cognitive input than is required when following well-used guidelines for commencing medication. Clinicians reported not having the opportunity (or 'the headspace’) to engage with the rationalisation of medication as often as they would have liked. GPs expressed concern about the lack of professional training given to deprescribing. Finally, physicians expressed anxiety about working outside of (and in some cases directly against) guidelines. Specifically, in light of growing litigation, GPs felt vulnerable and concerned about the consequences of not following guidelines.

There are now a number of tools being developed to support clinicians in rationalising medication. Evidence of an impact on medication load and on patient mortality remains limited (Avery and Bell, 2019). One meta-analysis suggests a decrease in medication of only 0.2 drugs per patient, on average, following use of such tools. Another meta-analysis showed no overall reduction in patient mortality from such interventions (Page et al., 2016). However, the tools may help individual clinicians and patients to work together safely and achieve person-centred decisions about medication.

This is a rapidly evolving area. Our aim is to draw your attention to the issue and to encourage you to tackle each of these challenges. Table 1 suggests some tools for you to look at and consider their use.

Table 1.

Tools available to identify and manage problematic polypharmacy.

Identifying patients
From the clinician’s perspective: STOPP START criteria	www.ncbi.nlm.nih.gov/pmc/articles/PMC4339726/	• Contains 114 criteria (80 Stop and 34 Start) that either encourage the clinician to cease or begin medications in patients >64 years old • This would be useful as a tool to run periodically within a practice to audit/identify problematic polypharmacy
From the patient’s perspective: Living with Medicines Questionnaire	www.dovepress.com/development-and-validation-of-a-revised-instrument-to-measure-burden-o-peer-reviewed-article-PROM	• This survey can be given to patients and asks them to rate a variety of factors relating to their experience of medication use. Questions are related to how medicines interfere with the patient’s life, side-effect profile, perceived effectiveness of medications and quality of communication with the clinician • A particularly useful tool when medicines seem not be to having the anticipated effect (i.e. when compliance may be an issue to consider)
Managing prescribing
NHS Scotland Polypharmacy Guidance Realistic Prescribing	www.therapeutics.scot.nhs.uk/wp-content/uploads/2018/04/Polypharmacy-Guidance-2018.pdf	• A very detailed resource that contains information on individual drugs (and problems associated with polypharmacy of which to be mindful), but also contains resources for methodology of how the trainee should approach the area of rationalisation of medications • The seven-step approach to medicines review (p. 14) is particularly useful for the clinician as a quick-reference crib sheet
Oxfordshire Clinical Commissioning Group, Prescribing Rationalisation Clinical Tool	https://clinox.info/clinical-support/local-pathways-and-guidelines/Clinical%20Guidelines/Prescribing%20 Rationalisation% 20Clinical%20Tool.pdf	• A condensed tool, focusing on key areas of problematic prescribing found in the Oxfordshire area. Similar tools may exist within other Clinical Commissioning Groups • By focusing the guidance to align with local issues the tool is simple to utilise (it is only 10 pages long) and allows the clinician to direct their time towards the key problems in their locality
Tower Hamlets Clinical Commissioning Group: Managing Medicines in the Last Years of Life Decision Support Guidance for Health Professionals in North East London	www.stjh.org.uk/system/files_force/files/MANAGING%20 MEDICINES%20IN%20 THE%20LAST%20 YEARS%20OF% 20LIFE.pdf?download =1	• A tool focusing on rationalising medications as patients approach the end of life. Broken down into 10 broad pathological categories (Diabetes, Respiratory disease, Heart failure, Cancer, Last days of life, etc.) it offers guidance to the clinician on which medications to stop, continue and even which medications to consider introducing. • Very detailed and useful guide. By focusing on pathological domains rather than individual classes of drugs, commonly associated drugs are considered together, making the guidance instinctively more applicable to practice
Gold Standards Framework	www.goldstandards framework.org.uk/	• Although rationalisation of medications at the end of life may be done quite well in the very acute setting (where the patient has days or small number of weeks to live), it is often overlooked in the patients who may not formally be on a palliative pathway. Use of the question: ‘Would I be surprised if this person were to die in the next 12 months?’ (Boyd and Murray 2010), is useful for prompting clinicians to consider engaging the patient in a discussion surrounding discontinuation of medications. • Tools are available on the Gold Standards Framework website to aid clinicians in holding these difficult discussions.

Gaps still to fill: A rapidly evolving field of practice

Education around deprescribing is being rolled out as part of junior doctor training (Poots et al., 2017). Tools are available to aid clinicians with overcoming problems associated with polypharmacy. These tools are varied and with interrogation of the literature clinicians are likely to find tools that fit their own practice. These tools are far from perfect, however. Reviewing a patent’s medication in a systematic way is a significantly time-consuming task; it may be complicated by patients’ cognitive impairment or poor understanding of how their medication works. It may be necessary to involve carers or nursing home staff in effective rationalisation of medication and to ensure that medication is optimised to individual patient’s needs.

It would be neglectful to consider rationalisation of medication without recognition of the on-going GP workforce crisis. According to the survey undertaken by Reeve et al. (2018), GPs want to engage more with deprescribing and individualisation of prescribing, but feel constrained by the lack of time. In addition, 32% of GPs report burnout (Patel, 2019). How can GPs incorporate new methods of working into practice? Neural networks and deep learning may offer assistance to the over-burdened GP. A team in Stanford, USA has developed a neural network named Decagon to look at documented side-effect profiles of medications alongside protein–protein interactions that can predict side effects associated with polypharmacy (Zitnik et al., 2018). Such tools may be used to alert prescribers to interactions that would otherwise be missed in high-pressure short-duration consultations; they will become invaluable as polypharmacy continues to increase in prevalence.

GPs no longer work alone in practices. The rise of the multi-disciplinary team is beginning to ease the pressure on GP workloads across the country. The work associated with rationalising medications and identifying those at-risk from the perils of polypharmacy need not fall entirely on the shoulders of GPs. Practice pharmacists will be able to help tackle this problem. There is NHS funding for 20 000 new staff (NHS England, 2019), and an opportunity for rationalisation of medications and identification protocols for problematic polypharmacy to be part of everyday practice. This aim and opportunity is recognised by our pharmacist colleagues (McClay and Gyimah, 2019).

Conclusion

Managing polypharmacy – especially problematic polypharmacy – is a challenge for trainees and qualified clinicians alike. Many of the issues arising can be dealt with using good evidence-based medicines management, and the core principles of good general practice including whole-person assessment, good communication skills and continuity of care.

However, problematic polypharmacy requires medical practice to pay greater attention not only to the disease that people have, but to the delivery of whole-person care based on the principles and practice of expert generalist medicine. Tools and ways of working are evolving, but the needs of patients are also evolving and present a challenge to evidence-based practice. Good team work is needed to define, deliver and educate in all aspects of this important, emerging area of practice.

KEY POINTS

When considering polypharmacy and rationalisation of medication remember the principles of good general practice and person-centred care

Explore the ideas, concerns and expectations that patients have about their medication

Ask patients if they feel burdened by their medication and invite them to work with you to identify possible areas for compromise

Identify and use tools to help with the management of polypharmacy and rationalisation of medication; with your trainer reflect and learn from use of these tools

Complex management and decisions deserve time, attention and teamwork

Remember you are part of a team: share complex decisions to benefit from the critical, collaborative and thoughtful involvement of other GPs and pharmacists

ORCID iD

Joanne Reeve https://orcid.org/0000-0002-3184-7955

References

Avery A, Barber N, Ghaleb M, et al. (2012) Investigating the prevalence and causes of prescribing errors in general practice: The PRACtICe study. Available at: www.gmc-uk.org/about/research/12996. asp (accessed 30 August 2019).

Avery

Bell

(2019) Rationalising medications through deprescribing. BMJ 364: l570DOI: 10.1136/bmj.I570.

Barber

Alldred

Raynor

, et al.(2009) Care homes’ use of medicines study: Prevalence, causes and potential harm of medication errors in care homes for older people. BMJ Quality & Safety 18(5): 341–346. DOI: 10.1136/qshc.2009.034231.

Bloch

Thibaud

Dugué

, et al.(2010) Laxatives as a risk factor for iatrogenic falls in elderly subjects. Myth or reality? Drugs & Aging 27(11): 895–901. DOI: 10.2165/11584280-000000000-00000.

Boyd

Murray

(2010) Recognising and managing key transitions in end of life care. British Medical Journal 341: c4863DOI: 10.1136/bmj.c4863.

Denford

Frost

Dieppe

, et al.(2014) Individualisation of drug treatments for patients with long-term conditions: A review of concepts. BMJ Open 4(3): e004172DOI: 10.1136/bmjopen-2013-004172.

Duerden M, Avery T and Payne R (2013) Polypharmacy and medicines optimization: Making it safe and sound. Available at: www.kingsfund.org.uk/sites/default/files/field/field_publication_file/polypharmacy-and-medicines-optimisation-kingsfund-nov13.pdf (accessed 29 August 2019).

Gold Standard Framework (2013) Resources. Available at: www.goldstandardsframework.org.uk/ (accessed 29 August 2019).

Guthrie

Makubate

(2012) The rising tide of polypharmacy and potentially serious drug interactions 1995–2010: Repeated cross sectional analysis of dispensed prescribing in one region. Primary Health Care Research & Development 13(S1): 45 SE.2 DOI: 10.1186/s12916-015-0322-7.

10.

Hanlon

Nicholl

Jani

, et al.(2019) Examining patterns of multimorbidity, polypharmacy and risk of adverse drug reactions in chronic obstructive pulmonary disease: A cross-sectional UK Biobank study. BMJ Open 8(1): e018404DOI: 10.1136/bmjopen-2017-018404.

11.

Heaton

Britten

Krska

, et al.(2017) Person-centred medicines optimisation policy in England: An agenda for research on polypharmacy. Primary Health Care Research and Development 18(1): 24–34. DOI: 10.1017/S1463423616000207.

12.

Katusiime

Corlett

Krska

(2018) Development and validation of a revised instrument to measure burden of long-term medicines use: The Living with Medicines Questionnaire version 3. Patient Related Outcome Measures Journal 9: 155–168. DOI: 10.2147/PROM.S151143.

13.

Krska

Katusiime

Corlett

(2018) Patient experiences of the burden of long-term medicines use and factors affecting burden: A cross-sectional survey. Health and Social Care in the Community 26(6): 946–959. DOI: 10.1111/hsc.12624.

14.

McClay E and Gyimah C (2019) Pharmacists must work with others to roll back polypharmacy culture. Available at: www.pharmaceutical-journal.com/opinion/insight/pharmacists-must-work-with-others-to-roll-back-polypharmacy-culture/20206704.article?firstPass=false (accessed 30 August 2019).

15.

NHS Digital (2019) Prescriptions dispensed in the community - statistics for England, 2007-2017. Available at: https://digital.nhs.uk/data-and-information/publications/statistical/prescriptions-dispensed-in-the-community/prescriptions-dispensed-in-the-community-england –-2007 –-2017#key-facts (accessed 29 August 2019).

16.

NHS England (2019) Five-year deal to expand GP services and kick start NHS Long Term Plan implementation. Available at: www.england.nhs.uk/2019/01/five-year-deal-to-expand-gp-services-and-kick-start-nhs-long-term-plan-implementation/ (accessed 30 August 2019).

17.

NHS Scotland (2018) Polypharmacy guidance realistic prescribing, third edition. Available at: www.therapeutics.scot.nhs.uk/wp-content/uploads/2018/04/Polypharmacy-Guidance-2018.pdf (accessed 30 August 2019).

18.

O’Mahony

O’Sullivan

Byrne

, et al.(2015) STOPP/START criteria for potentially inappropriate prescribing in older people: version 2. Age Ageing 44(2): 213–218. DOI: 10.1093/ageing/afu145.

19.

Oxfordshire Clinical Commissioning Group (2017) Prescribing rationalisation clinical tool. Available at: https://clinox.info/clinical-support/local-pathways-and-guidelines/Clinical%20Guidelines/Prescribing%20Rationalisation%20Clinical%20Tool.pdf (accessed 29 August 2019).

20.

Page

Clifford

Potter

, et al.(2016) The feasibility and effect of deprescribing in older adults on mortality and health: A systematic review and meta-analysis. British Journal of Clinical Pharmacy 82: 583–623. DOI: 10.1111/bcp.12975 pmid: 27077231.

21.

Patel A (2019) One in three GPs suffer from burnout due to workload, figures show. Available at: www.pulsetoday.co.uk/news/hot-topics/war-on-workload/one-in-three-gps-suffer-from-burnout-due-to-workload-figures-show/20038268.article (accessed 30 August 2019).

22.

Poots

Jubraj

Barnett

(2017) Education around deprescribing: ‘spread and embed’ the story so far. European Journal of Hospital Pharmacy 24(1): 7–9. DOI: 10.1136/ejhpharm-2016-001153.

23.

RCGP (2019) Professional topic guide: Improving quality, safety and prescribing. Available at: www.rcgp.org.uk/training-exams/training/gp-curriculum-overview.aspx (accessed 1 November 2019).

24.

Reeve

Britten

Byng

, et al.(2018) Identifying enablers and barriers to individually tailored prescribing: A survey of healthcare professionals in the UK. BMC Family Practice 19(1): 17DOI: 10.1186/s12875-017-0705-2.

25.

Szczepura

Wild

Nelson

(2011) Medication administration errors for older people in long-term residential care. BMC Geriatrics 11: 82DOI: 10.1186/1471-2318-11-82.

26.

Tower Hamlets Clinical Commissioning Group (2014) Managing medicines in the last years of life. Available at: www.stjh.org.uk/system/files_force/files/MANAGING%20MEDICINES%20IN%20THE%20LAST%20YEARS%20OF%20LIFE.pdf?download=1 (accessed 29 August 2019).

27.

Wennberg D, Siegel M, Darin B, et al. (2006) Combined predictive model: Final report and technical documentation. Available at: www.kingsfund.org.uk/sites/default/files/field/field_document/PARR-combined-predictive-model-final-report-dec06.pdf (accessed 30 August 2019).

28.

Zitnik

Agrawal

Leskovec

(2018) Modeling polypharmacy side effects with graph convolutional networks. Bioinformatics 34(13): 457–466. DOI: 10.1093/bioinformatics/bty294.