University of Groningen

Deprescribing in older people van der Meer, Helene Grietje

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): van der Meer, H. G. (2019). Deprescribing in older people: development and evaluation of complex healthcare interventions. Rijksuniversiteit Groningen.

Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.

Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 30-09-2021 UITNODIGING DEPRESCRIBING IN OLDER PEOPLE DEPRESCRIBING Voor het bijwonen van de openbare verdediging van het proefschrift:

DEPRESCRIBING IN OLDER PEOPLE DEPRESCRIBING Development and evaluation of complex IN OLDER PEOPLE healthcare interventions

Helene Grietje (Heleen) van der Meer was born on Development and evaluation of complex door healthcare interventions 5 August 1990 in Papenburg, Germany, to Wytze Jan HELEEN VAN DER MEER van der Meer, dentist, and Klaaske van der Meer- Jansen, cardiac care nurse. She grew up in Germany together with her younger sister and brother. In 2009 Heleen van der Meer op vrijdag she obtained her Abitur (final exam) at the Gymnasium 2 november om 16.15 Papenburg and started her studies in pharmacy at the in het Academiegebouw University of Groningen. van de Rijkuniversiteit Groningen, Broerstraat 5 Heleen first became acquainted with research in the te Groningen. Heleen van der Meer Heleen van field of pharmacotherapy during her Bachelors studies. The foundation for her doctoral thesis was laid during Na afloop bent u van harte uitgenodigd the project she undertook in Sydney, Australia under voor de receptie supervision of Dr. Lisa Pont and Prof. Dr. Katja Taxis for in het Academiegebouw. her Masters in Pharmacy in 2013/14. On her return to the Netherlands, she accepted a temporary appoint- ment as a researcher with Prof. Taxis and in the same Heleen van der Meer year gave her first podium presentation at an inter- Helmersstraat 36 national scientific conference in Boston, US. She was 2513 RZ Den Haag awarded her Masters in Pharmacy in 2016 and com- 0648897302 pleted her PhD in 2018. Heleen lives in The Hague and [email protected] works as a postdoctoral researcher under supervision of Prof. Taxis on the development and implementation of patient material for deprescribing in older people. PARANIMFEN In addition to her studies and PhD research, Heleen has Karlien Sambell been active within various committees. For example, [email protected] in 2016/17 she organized the PhD Day, a career event for 900 PhD students/postdocs. Furthermore she loves Linda van Eikenhorst tennis and sailing and she has a passion for traveling. [email protected] DEPRESCRIBING IN OLDER PEOPLE Development and evaluation of complex healthcare interventions

Heleen van der Meer Deprescribing in older people Deprescribing in older people Development and evaluation of complex healthcare interventions Development and evaluation of complex healthcare interventions

Proefschrift Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen Colophon ter verkrijging op gezagvan van de de graad van doctor aan de The research presented in this thesis was financially supported by the Royal rector magnificusRijksuniversiteit prof. dr. E. SterkenGroningen Dutch Pharmacists Association (KNMP) and Stichting Stoffels Hornstra. en volgens besluit van hetop College gezag voorvan de Promoties. rector magnificus prof. dr. E. Sterken Cover concept: Heleen van der Meer De openbareen volgens verdediging besluit van zal het plaatsvinden College voor op Promoties. Cover and layout design: Lovebird design. www.lovebird-design.com vrijdagDe openbare 22 maart verdediging2019 om 16.15 zal uur plaatsvinden op Printing: Eikon+ vrijdag 2 november 2018 om 16.15 uur ISBN (e-book): 978-94-034-0954-2 ISBN (printed book): 978-94-034-0955-9 door

Printing of this thesis was financially supported by the Groningen Graduate door School of Science and Engineering (GSSE), the University of Groningen and Stichting Koninklijke Nederlandse Maatschappij ter Bevordering der Pharmacie (KNMP) Fondsen, and is gratefully acknowledged. Helene Grietje van der Meer

© Copyright, 2018, Heleen van der Meer geborenHelene op 5Grietje augustus van1990 der Meer All rights reserved. No part of this publication may be reproduced or transmit- te Papenburg, Duitsland ted in any form or by any means, without written permission of the author. geboren op 5 augustus 1990 te Papenburg, Duitsland

Promotor CONTENTS Prof. dr. K. Taxis Dr. L.G. Pont CHAPTER 1 General Introduction And Thesis Outline 7 Co-promotor Dr. H. Wouters CHAPTER 2 Changes In Prescribing Symptomatic And Beoordelingscommissie Prof. dr. P. Denig Preventive In The Last Year Of Prof. dr. M.L. Bouvy Life In Older Nursing Home Residents 19 Prof. dr. R.H. Vander Stichele

CHAPTER 3 and use in older community-dwelling people: a national population study in the Netherlands 43

CHAPTER 4 Decreasing the load? Is a multidisciplinary multistep medication review in older people an effective intervention to reduce a patient’s burden index? Protocol of a randomised controlled trial 63

CHAPTER 5 Reducing the anticholinergic and sedative load in older patients on polypharmacy by pharmacist-led medication review: A randomised controlled trial 81

CHAPTER 6 Feasibility, acceptability and potential effectiveness of an information technology based, pharmacist-led intervention to prevent an increase in anticholinergic and sedative load among older community-dwelling individuals. 109

CHAPTER 7 General discussion 137

CHAPTER 8 Summary 151 Samenvatting 157 Acknowledgements — Dankwoord 163 List of publications 169 Deprescribing in older people General introduction and thesis outline

1

CHAPTER 1

GENERAL INTRODUCTION AND THESIS OUTLINE

6 7 General introduction and thesis outline

PRESCRIBING IN OLDER PEOPLE

Worldwide, the population of older people is estimated to in- 1 crease from 524 million in 2010 to 1.5 billion in 2050. [1] With ageing, the number of individuals with one or more chronic dis- eases is growing. [2] Medications are the most common interven- tion to cure, prevent or relief symptoms of a disease. Older people aged 65 years and over use more medications than any other age group, 45–75% of this population uses 5 or more medications and 15–30% uses 10 or more medications. [3]

Several important factors complicate medication use in older people. Firstly, use of multiple medications increases the risk to experience adverse drug reactions (ADR). [4] Secondly, age-re- lated changes in pharmacokinetic and dynamic responses to a medication may decrease an older person’s tolerance to medi- cations. [5] Thirdly, scientific evidence on benefits and risks of medications in older people is often absent, as frail older people are rarely included in clinical trials to evaluate medication effi- cacy and safety. [6]

Prescribing of medications that might be inappropriate in older people has been widely studied. A number of definitions of po- tentially inappropriate prescribing (PIP) have been proposed and several criteria have been developed to detect PIP. [7, 8] The screening tool of older people’s prescriptions (STOPP) and screen- ing tool to alert to right treatment (START) criteria [9] and Beers criteria [10] are among the best known. PIP is common among older people [11–14] and has been associated with increased ADRs, morbidity, hospitalisations and decreased quality of life. [15–20]

In this thesis, potentially inappropriate prescribing in two specific patient populations is explored. Firstly, prescribing of preventive medications at the end of life in older nursing home residents. Secondly, prescribing of anticholinergic and sedative medications in older community-dwelling patients.

9 Deprescribing in older people General introduction and thesis outline

PREVENTIVE MEDICATIONS AT THE END OF LIFE DEPRESCRIBING

Toward the end of life, in addition to considerations around The term deprescribing was first introduced in Australia, in 2003. 1 potential medication related benefits and harms, the decision [33] While the term was new, the process of withdrawing inap- to prescribe a medication should also take life expectancy into propriate medications was not. [34] Since the introduction of the consideration. As life expectancy decreases, the goals of care may term deprescribing, several definitions have been proposed. Based change from decreasing mortality and morbidity, to symptom on a systematic literature review on all definitions, deprescrib- control. In light of limited life expectancy toward the end of life, ing was defined as ‘the process of withdrawal of an inappropriate the use of medications to prevent future onset of disease or com- medication, supervised by a health care professional with the goal plications, which need a long time until benefit, might become of managing polypharmacy and improving outcomes.’ [35] less appropriate than medications for symptom management, which have immediate benefits. [21] Few studies have investigated medication use in the last period of life in an older nursing home MEDICATION REVIEW population. [22] Little is known to what extent preventive medi- cations are still used in this phase. In Chapter 2 changes in pre- A widely proposed intervention to facilitate deprescribing is scribing of preventive and symptomatic medication at the end of medication review. [36–38] Medication review is ‘a structured, life in older nursing home residents will be explored. critical examination of a patient’s medicines with the objective of reaching an agreement with the person about treatment, op- timising the impact of medicines, minimising the number of ANTICHOLINERGIC AND SEDATIVE MEDICATIONS medication related problems and reducing waste’. [39] An over- view of systematic reviews showed medication review has the Anticholinergic and sedative medications are commonly identi- potential to improve appropriateness of medications and clinical fied as potentially inappropriate medications for older people. [9, outcomes. [40] To date, the effectiveness of medication review 10] They have negative effects on cognitive and physical function as a deprescribing strategy to reduce the use of anticholinergic/ in older people and increase the risk of falls, dementia, hospital- sedative medications in older people remains unclear. Two small isation and mortality. [23–25] Despite these negative outcomes, Australian studies found that pharmacist-led medication reviews they are frequently used in older people. [26, 27] Use of several were effective in reducing the cumulative anticholinergic/seda- anticholinergic/sedative medications, resulting in a higher an- tive load. However, these studies included a pilot and a retrospec- ticholinergic/sedative load, is associated with increased risk of tive study both based on pharmacist recommendations without negative outcomes. [28–30] To date, most research has focused on investigating actual implementation of recommendations by the quantifying the use of individual anticholinergic/sedative med- general practitioner. [41, 42] Chapter 4 presents the study pro- ications [31] or aggregating use in the form of a total load score. tocol for a randomised controlled trial, which had the aim to [32] Little is known about the prevalence of combinations of evaluate whether medication review is effective in deprescribing multiple anticholinergic/sedative medications used or subgroups anticholinergic/sedative medications in older people with a high of patients based on anticholinergic/sedative medication use. In anticholinergic/sedative load. In Chapter 5 the results of this Chapter 3 these gaps in knowledge will be addressed. randomised controlled trial are shown.

10 11 Deprescribing in older people General introduction and thesis outline

NEW DEPRESCRIBING INTERVENTIONS are effective in older community-dwelling adults with a high an- ticholinergic/sedative load or whether innovative approaches for Given the lack of effective interventions to support deprescrib- deprescribing are more successful. 1 ing of anticholinergic/sedative medications among older popula- tions, there is a critical need for the development and testing of innovative strategies. Information technology (IT) is increasingly THESIS AIM being used to identify patients in need of medication optimis- ation. [43] In Dutch community pharmacy practice pharmacists Development and evaluation of interventions for deprescribing use IT-based drug therapy alerts to monitor safety of a patient’s in older people, by identifying opportunities for deprescribing medication when it is presented in the pharmacy information (chapter 2 and 3), evaluating a current deprescribing intervention system for initial supply. [44] None of these drug therapy alerts (chapter 4 and 5) and developing and evaluating a new depre- focus on anticholinergic/sedative medications. Using IT to iden- scribing intervention (chapter 6). tify older individuals with anticholinergic/sedative medication is worthwhile to explore. This approach can be used in a new depre- scribing intervention.

Best practice for developing and evaluating an intervention is identifying the best available evidence and appropriate theory to develop the intervention, then to test the feasibility and perform an exploratory evaluation, before going on to a definitive evalua- tion followed by eventual implementation. [45]

Therefore, in Chapter 6 the feasibility, acceptability and poten- tial effectiveness of a newly developed IT-based pharmacist-led intervention were explored based on best practice principles for intervention development and evaluation.

GAPS IN KNOWLEDGE

Prescribing patterns of several potentially inappropriate medica- tions in older populations, such as preventive medications at the end of life in older nursing home residents and anticholinergic/ sedative medications in older community-dwelling adults, re- main undiscovered areas. It is unknown whether currently per- formed deprescribing interventions, such as medication reviews,

12 13 Deprescribing in older people General introduction and thesis outline

REFERENCES 16. Lund BC, Carnahan RM, Egge JA, Chrischilles EA, Kaboli PJ. Inappropriate pre- scribing predicts adverse drug events in older adults. Ann Pharmacother. 2010;44(6):957–963. 1. World Health Organization and US National Institute of Aging. Humanity’s Age- 17. Hamilton H, Gallagher P, Ryan C, Byrne S, O’Mahony D. Potentially inappropri- 1 ing. In: Global Health and Ageing. 2011. http://www.who.int/ageing/publications/ global_health.pdf?ua=1. Accessed May 2018. ate medications defined by STOPP criteria and the risk of adverse drug events in older hospitalized patients. Arch Intern Med. 2011;171(11):1013–1019. 2. Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. Epidemiology of mul- timorbidity and implications for health care, research, and medical education: a 18. Price SD, Holman CD, Sanfilippo FM, Emery JD. Association between potentially cross-sectional study. Lancet. 2012;380(9836):37–43. inappropriate medications from the Beers criteria and the risk of unplanned hospitalization in elderly patients. Ann Pharmacother. 2014;48(1):6–16. 3. SIMPATHY. Polypharmacy Management by 2030: a patient safety challenge. 2017. http://www.simpathy.eu/sites/default/files/Managing_polypharmacy2030-web.pdf. 19. Pasina L, Djade CD, Tettamanti M, et al. Prevalence of potentially inappropri- Accessed May 2018. ate medications and risk of adverse clinical outcome in a cohort of hospi- talized elderly patients: results from the REPOSI Study. J Clin Pharm Ther. 4. Atkin PA, Veitch PC, Veitch EM, Ogle SJ. The epidemiology of serious adverse drug 2014;39(5):511–515. reactions among the elderly. Aging. 1999;14(2):141–152. 20. Laroche ML, Charmes JP, Nouaille Y, Picard N, Merle L. Is inappropriate medica- 5. Mangoni AA, Jackson SH. Age-related changes in pharmacokinetics and pharma- tion use a major cause of adverse drug reactions in the elderly? Br J Clin Pharma- codynamics: basic principles and practical applications. Br J Clin Pharmacol. col. 2007;63(2):177–186. 2004;57(1):6–14. 21. Holmes HM, Hayley DC, Alexander GC, Sachs GA. Reconsidering medication ap- 6. McMurdo ME, Roberts H, Parker S, et al. Improving recruitment of older people to propriateness for patients late in life. Arch Int Med. 2006;166(6):605–609. research through good practice. Age Ageing. 2011;40(6):659–665. 22. Poudel A, Yates P, Rowett D, Nissen LM. Use of Preventive Medication in Patients 7. Spinewine A, Schmader KE, Barber N, et al. Appropriate prescribing in elderly With Limited Life Expectancy: A Systematic Review. J Pain Symptom Manage. people: how well can it be measured and optimised? Lancet. 2007;370(9582): 2017;53(6):1097–1110. 173–184. 23. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- 8. O’Connor MN, Gallagher P, O’Mahony D. Inappropriate prescribing: criteria, detec- erties on cognitive function, delirium, physical function and mortality: a sys- tion and prevention. Drugs Aging. 2012;29(6):437–452. tematic review. Age Ageing. 2014;43(5):604–615. 9. O’Mahony D, O’Sullivan D, Byrne S, O’Connor MN, Ryan C, Gallagher P. STOPP/ 24. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls START criteria for potentially inappropriate prescribing in older people: ver- in older people: systematic review of publications from a recent 5-year period. sion 2. Age Ageing. 2015;44(2):213–218. Eur J Clin Pharmacol. 2015;71(12):1429–1440. 10. By the American Geriatrics Society 2015 Beers Criteria Update Expert Panel. Amer- 25. Gray SL, Anderson ML, Dublin S, et al. Cumulative use of strong anticholiner- ican Geriatrics Society 2015 Updated Beers Criteria for Potentially Inappropriate gics and incident dementia: a prospective cohort study. JAMA Intern Med. Medication Use in Older Adults. J Am Geriatr Soc. 2015;63(11):2227–2246. 2015;175(3):401–407. 11. Tommelein E, Mehuys E, Petrovic M, Somers A, Colin P, Boussery K. Potentially in- 26. Bell JS, Mezrani C, Blacker N, et al. Anticholinergic and sedative medicines — appropriate prescribing in community-dwelling older people across Europe: a prescribing considerations for people with dementia. Aust Fam Physician. systematic literature review. Eur J Clin Pharmacol. 2015;71(12):1415–1427. 2012;41(1–2):45–49. 12. Opondo D, Eslami S, Visscher S, et al. Inappropriateness of medication prescrip- 27. Holvast F, van Hattem BA, Sinnige J, et al. Late-life depression and the associa- tions to elderly patients in the primary care setting: a systematic review. PLoS tion with multimorbidity and polypharmacy: a cross-sectional study. Fam Pract. One. 2012;7(8):e43617. 2017;34(5):539–545. 13. Gallagher P, Barry P, O’Mahony D. Inappropriate prescribing in the elderly. J Clin 28. Wouters H, van der Meer H, Taxis K. Quantification of anticholinergic and sedative Pharm Ther. 2007;32(2):113–121. drug load with the Drug Burden Index: a review of outcomes and methodologi- 14. Aparasu RR, Mort JR. Inappropriate prescribing for the elderly: beers criteria-based cal quality of studies. Eur J Clin Pharmacol. 2017;73(3):257–266. review. Ann Pharmacother. 2000;34(3):338–346. 29. Hilmer SN, Mager DE, Simonsick EM, et al. A drug burden index to define 15. Cahir C, Bennett K, Teljeur C, Fahey T. Potentially inappropriate prescribing and ad- the functional burden of medications in older people. Arch Intern Med. verse health outcomes in community dwelling older patients. Br J Clin Pharma- 2007;167(8):781–787. col. 2014;77(1):201–210.

14 15 Deprescribing in older people General introduction and thesis outline

30. Nishtala PS, Narayan SW, Wang T, Hilmer SN. Associations of drug burden index 43. Dreischulte T, Donnan P, Grant A, Hapca A, McCowan C, Guthrie B. Safer Prescrib- with falls, general practitioner visits, and mortality in older people. Pharmaco- ing--A Trial of Education, Informatics, and Financial Incentives. N Engl J Med. epidemiol Drug Saf. 2014;23(7):753–758. 2016;374(11):1053–1064. 31. Taxis K, Kochen S, Wouters H, et al. Cross-national comparison of medication use 44. Heringa M, Floor-Schreudering A, Tromp PC, de Smet PA, Bouvy ML. Nature and 1 in Australian and Dutch nursing homes. Age Ageing. 2017;46(2):320–323. frequency of drug therapy alerts generated by clinical decision support in com- 32. Pont LG, Nielen JT, McLachlan AJ, et al. Measuring anticholinergic drug exposure munity pharmacy. Pharmacoepidemiol Drug Saf. 2016;25(1):82–89. in older community-dwelling Australian men: a comparison of four different 45. Craig P, Dieppe P, Macintyre S, et al. Developing and evaluating complex interven- measures. Br J Clin Pharmacol. 2015;80(5):1169–1175. tions: the new Medical Research Council guidance. BMJ. 2008;337:a1655. 33. Woodward Michael C. Deprescribing: Achieving Better Health Outcomes for Older People through Reducing Medications. Journal of Pharmacy Practice and Re- search. 2003;33(4):323–328. 34. Scott IA, Hilmer SN, Reeve E, et al. Reducing inappropriate polypharmacy: the pro- cess of deprescribing. JAMA Intern Med. 2015;175(5):827–834. 35. Reeve E, Gnjidic D, Long J, Hilmer S. A systematic review of the emerging de fi ni- tion of ‘deprescribing’ with network analysis: implications for future research and clinical practice. Br J Clin Pharmacol. 2015;80(6):1254–1268. 36. Wouters H, Scheper J, Koning H, et al. Discontinuing inappropriate medication use in nursing home residents: A cluster randomized controlled trial. Ann Intern Med. 2017;167(9):609–617. 37. Willeboordse F, Hugtenburg JG, van Dijk L, et al. Opti-Med: the effectiveness of op- timised clinical medication reviews in older people with ‘geriatric giants’ in gen- eral practice; study protocol of a cluster randomised controlled trial. BMC Geri- atr. 2014;14:116-2318-14-116. 38. Clyne B, Smith SM, Hughes CM, et al. Effectiveness of a Multifaceted Interven- tion for Potentially Inappropriate Prescribing in Older Patients in Primary Care: A Cluster-Randomized Controlled Trial (OPTI-SCRIPT Study). Ann Fam Med. 2015;13(6):545–553. 39. NICE Medicines and Prescribing Centre (UK). Recommendations medication re- view. In: Medicines optimisation: the safe and effective use of medicines to en- able the best possible outcomes. National Institute for Health and Care Excel- lence. 2015. https://www.nice.org.uk/guidance/ng5/chapter/recommendations# medication-review. Accessed Mar 2018. 40. Jokanovic N, Tan EC, Sudhakaran S, et al. Pharmacist-led medication review in community settings: An overview of systematic reviews. Res Social Adm Pharm. 2017;13(4):661–685. 41. Gnjidic D, Le Couteur DG, Abernethy DR, Hilmer SN. A pilot randomized clin- ical trial utilizing the drug burden index to reduce exposure to anticholiner- gic and sedative medications in older people. Ann Pharmacother. 2010;44(11): 1725–1732. 42. Castelino RL, Hilmer SN, Bajorek BV, Nishtala P, Chen TF. Drug Burden Index and potentially inappropriate medications in community-dwelling older people: the impact of Home Medicines Review. Drugs Aging. 2010;27(2):135–148.

16 17 Deprescribing in older people General introduction and thesis outline

1

CHAPTER 2

CHANGES IN PRESCRIBING SYMPTOMATIC AND PREVENTIVE MEDICATIONS IN THE LAST YEAR OF LIFE IN OLDER NURSING HOME RESIDENTS

Helene G van der Meer, Katja Taxis, Lisa G Pont

Frontiers in Pharmacology. 2018;8:990.

18 19 Preventive medications at the end of life in older nursing home residents

ABSTRACT INTRODUCTION

Background At the end of life goals of care change from disease At all stages across the life span, the decision to prescribe a medi- prevention to symptom control, however little is known about cation should be based on weighing potential benefits and harms the patterns of medication prescribing at this stage. of the medication considering the individual’s treatment goals. Goals range from decreasing mortality and morbidity, prevention Objectives To explore changes in prescribing of symptomatic of future conditions or complications, or minimisation of symp- 2 and preventive medication in the last year of life in older nursing toms. Toward the end of life, in addition to considerations around home residents. potential medication related benefits and harms, treatment choice should also take life expectancy into consideration. As life ex- Methods A retrospective cohort study was conducted using phar- pectancy decreases, the goals of care may change from decreasing macy medication supply data of 553 residents from 16 nursing mortality and morbidity, to symptom control. [1] Long-term res- home facilities around Sydney, Australia. Residents received 24-h idential aged care or nursing home residents are among the frail- nursing care, were aged ≥ 65 years, died between June 2008 and est of all older populations. They are generally medically complex, June 2010 and were using at least one medication 1 year before using a high number of medications, and this complexity together death. Medications were classified as symptomatic, preventive or with age-related pharmacokinetic- and dynamic puts them at high other. A linear mixed model was used to compare changes in pre- risk of adverse outcomes related to medication. [2–4] scribing in the last year of life. Of all aged care residents, 91% die in the nursing home after an Results 68.1% of residents were female, mean age was 88.0 (SD: average stay of 168 weeks for women and 110 weeks for men, indi- 7.5) years and residents used a mean of 9.1 (SD: 4.1) medications cating that the majority of residents have limited life expectancy 1 year before death. The mean number of symptomatic medica- following nursing home admission. [3] Adjusting prescribing tions per resident increased from 4.6 medications 1 year before according to a decreasing life expectancy involves deprescribing, death to 5.1 medications at death (95% CI 4.4–4.7 to 5.9–5.2, defined as the process of withdrawing inappropriate medications. p = 0.000), while preventive medication decreased from 2.0 to [5, 6] Hence in this population a decrease in preventive and an 1.4 medications (95% CI 1.9–2.1 to 1.3–1.5, p = 0.000). Symptomatic increase in the use of medications for symptom control and palli- medications were used longer in the last year of life, compared ation could be expected. [7] to preventive medications (336.3 days (95% CI 331.8–340.8) versus 310.9 days (95% CI 305.2–316.7), p = 0.000). To date few studies exploring changes in the use of symptom- atic and preventive medications have been conducted in older Conclusions Use of medications for symptom relief increased nursing home populations at the end of life. A recent system- throughout the last year of life, while medications for prevention atic review found that use of preventive medications in patients of long-term complications decreased. But changes were slight with limited life expectancy was common. [8] Only few studies and clinical relevance can be questioned. focused on deprescribing and there was no consensus on how to optimise medication use at the end of life. Of the 15 studies included, three were performed in a nursing home setting. [8]

21 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

These studies included only a small study population [9] or had a used for the study. The dataset included generic name, dose, date cross-sectional study design. [10, 11] In order to consider optimi- of commencement, date of cessation and if the use was regular sation of medication use at the end of life we need to understand or ‘as needed’. The dataset also included limited demographic data the current patterns of use as life expectancy decreases. Therefore, for each resident including age, sex, date of admission, date and the aim of this study was to explore changes in prescribing of reason for discharge and facility. symptomatic and preventive medications in the last year of life among older nursing home residents. Medication classification 2 Medications were coded using the World Health Organization Anatomical Therapeutic Chemical (ATC) code. [12] Medications METHODS were classified into three categories: symptomatic, preventive and other. All medications recommended for symptom control in the Study design and setting Australian national palliative care guidelines were considered as A retrospective cohort study of 3876 nursing home residents liv- symptomatic medications. [13, 14] Medications defined in the lit- ing in 26 residential aged care (RAC) facilities in New South Wales, erature for primary or secondary prevention of all-cause mortality Australia between 1st June 2008 and 10th June 2010. The RAC fa- were defined as preventive medications. [15] Preventive medica- cilities varied from low care to high care. High care facilities pro- tions included antihypertensive medications, [16] antithrombotic vided 24 h nursing care including medication administration. All agents, [17] osteoporosis medication [18] and lipid modifying residents received medical care from the general practitioner of agents. [19] Medications that were not considered as either pre- their choice and were eligible to receive annual medication re- ventive or symptomatic were classified as other. Antibiotics, top- views by a pharmacist. Each facility has a contracted pharmacy for ical preparations, ophthalmological and otological medications medication supply. were excluded due to the episodic nature of the use of these med- ications. Vaccines were also excluded as they were administered Study population by the general practitioner and not supplied by the pharmacy. A Recruitment was done at the facility level. All residents aged 65 list of included medications can be found in the Appendix. years or older who died in one of the 26 RAC facilities between 2nd of June 2008 and 10th of June 2010 were included in the co- Outcomes hort. To allow medication changes in the year prior to death to Three main outcome measures were determined. Firstly, we com- be explored, only those residents who were taking at least one pared the mean number of symptomatic, preventive and other medication 1 year prior to death were included in the cohort. medications per resident at 1 year, 6 months, 1 month and 1 week Residents who were discharged prior to death were excluded (8 days) before death and on the day of death. Secondly, we com- from the study, as medication use could not be ascertained once pared the type of symptomatic, preventive and other medication they left the facility. used 1 year before death versus on the day of death. For this anal- ysis we included all medications, grouped by ATC level 2, which Data source were used by at least 10% of the population either 365 days before Weekly pharmacy medication supply data, including all prescrip- death or on the day of death. Thirdly, we compared the duration tion, non-prescription and complementary medications, were of use of symptomatic, preventive and other medications in the

22 23 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

last year of life. We included all medications used 365 days before RESULTS death, and calculated the days of treatment during the last year of life. Resident characteristics The cohort comprised of 553 residents out of the 3876 residents All medications used 7 or fewer days before death were consid- contained in the dataset (Figure 1). ered to be taken on the day of death. This was done for two rea- sons. Firstly, medication was supplied per week, therefore the Residents were between 65 and 105 years of age and lived in 16 dif- 2 last medication might have been supplied up to 7 days before ferent facilities. The average facilities size was 35 (SD: 21) residents death. Secondly, we assumed some inaccuracies in recording the per facility (range: 5–71) (Table 1). date of death due to a delay in nursing home staff notifying phar- macy staff. Number of symptomatic, preventive and other medications in the last year of life Statistical analysis The total number of medications per resident decreased from Medication changes were analysed with a linear mixed model to account 9.1 (95% CI 8.9–9.3) medications 1 year prior to death to 8.5 for clustering of medications within one resident. Our data did not allow (95% CI 8.5–8.9) medications at death (p = 0.002). Symptomatic clustering for general practitioners. Therefore we performed clustering on the level of facility, to account for possible intra-facility culture of medi- Residents in database cation prescribing. We included a random intercept and a random slope at n = 3876 the level of resident and facility in the analysis. Analyses were adjusted for

Excluded (n = 2904): age, gender, duration of admission and number of medications at 365 days - Age < 65 years (n = 241) or unknown date of birth (n = before death, if the individual p-value in the univariate analysis was 0.25 or 138) less. [20] The number of medication and days of treatment were reported as - Not high care facility (n = 1549), facility with data issues estimated marginal means with their 95% confidence intervals. The second (n = 104), facility with unknown care level (n = 50)

- Did not die in RAC within the study period (n = 2296) outcome was analysed using a McNemar test. We report on proportions and absolute numbers of residents. All analyses were conducted in IBM SPSS 24 Residents satisfying on a significance level of 0.05. demographic

inclusion criteria

n = 972 Ethical approval

This study was approved by the Sydney South West Area Health Excluded (n = 419):

Service Human Research Ethics Committee, the Concord - Stayed < 365 days in nursing home before death (n = 413)

- No recorded symptomatic, preventive or other medication Repatriation General Hospital (CH62/6/2010-49 HREC/10/ history at 365 days before death (n = 6) CGRH/57).

Study cohort n = 553

Figure 1: Flow chart of resident inclusion.

24 25 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

Table 1: Resident characteristics in the type of used was seen, shifting from to , respectively 83.4% to 77.9% (p = 0.005) and 18.1% to Characteristic Residents (n = 553) 44.5% (p = 0.000). Other significant changes in use of symptom- Age, mean years (SD) 88.0 (7.5) atic medications toward death were only seen for diuretics (30.2% * Gender, % female (number) 68.1 (374) to 26.0%, p = 0.009) and medications for gastrointestinal disorders Length of stay in RAC facility, mean weeks (SD) 187.9 (104.4) (17.2% to 22.8%, p = 0.000). In contrast, all preventive medications Number of medications 365 days before death, mean (SD) 9.1 (4.5) decreased significantly from 1 year before death until death. The 2 Number of medications at death, mean (SD) 8.7 (5.1) highest decrease was found in supplements (including cal- *n =549, gender was missing for 4 residents cium), agents acting on the renin-angiotensin-aldosterone-system (RAAS) and lipid modifying agents, those respectively decreased by medication use increased from 4.6 to 5.1 (95% CI 4.4–4.7 to 9.2% (p = 0.000), 8.9% (p = 0.000) and 8.1% (p = 0.000) (Table 2). 5.9–5.2, p = 0.000) medications, while preventive and other med- However, at death about one third of all residents was using at least ication decreased, respectively 2.0 to 1.4 (95% CI 1.9–2.1 to 1.3–1.5, one antihypertensive medication (35.8%), one medication for oste- p = 0.000) and 2.6–2.2 (95% CI 2.4–2.7 to 2.1–2.4, p = 0.000), to- oporosis (32.9%) or an antithrombotic medication (33.1%). ward death (Figure 2). Duration of use of symptomatic, preventive and other Type of symptomatic, preventive and other medication used in medications in the last year of life the last year of life Symptomatic, preventive and other medications were used respec- Analgesics were the most frequently used type of medication over tively for 336.3 [95% CI 331.8–340.8], 310.9 [95% CI 305.2–316.7] the last year of life. use did not change significantly and 320.5 [95% CI 315.2–325.8] days in the last year of life. during the last year of life and was comparable at 1 year before Preventive and other medications were ceased earlier than symp- death and at death, (85.0% to 86.1% of patients, p = 0.610). A shift tomatic medication, respectively 25.4 days earlier [EMM, 95% CI 31.0–19.7, P=0.000] and 15.8 days earlier [EMM, 95% CI 20.9–10.7, P=0.000] (Figure 3). 6

5 †‡ Symptomatic 4 DISCUSSION *†‡ 3 Other

2 Key findings *†‡

Number of medications Number Preventive 1 Throughout the last year of life we saw little change in overall medication use. Medications commonly used for symptom con- 0 365 183 30 8 0 trol slightly increased, while a small decrease in medication for Days before death disease-prevention was seen. However at death, preventive med- Figure 2: Number of symptomatic, preventive, and other medication in the ication such as antithrombotic agents, antihypertensive medica- last year of life. Estimated marginal means (EMMs), adjusted for number of bed days in facility*, age†, and number of medication at 365 days before tions and osteoporosis medications were still prescribed to one death‡. third of all residents.

26 27 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents Δ % 0.2 −1.1 −1.4 −2.9 −2.7* −3.4* −6.0* % 11.4 11.8 25.7 15.9 22.2 14.6 24.2 At death At death 2 including ) including airway disease airway Cardiac therapy Cardiac Thyroid therapy Thyroid Medication group Medication Psychoanaleptics obstructive for Medication Mineral supplements (not Mineral Drugs used in diabetes Anti-anaemic medication A12 C01 A10 B03 R03 ATC N06 H03 code Other

Figure 3: Duration of use of symptomatic, preventive and other medica- Δ % −8.1* −2.7* −9.2* −4.3* −5.6* −6.5* −8.9* −6.0* tion in the last year of life. *We included all medications used 365 days before death. % 7.1 5.6 9.9 33.1 21.3 17.9 14.5 23.9 At death At death Changes in medication use at the end of life The characteristics of our cohort of residents are similar to other studies in this setting, so we believe our sample is representative for the nursing home population in Australia. The residents’ av- erage duration of stay in the RAC facility was slightly higher than blocking agents blocking including calcium including of bone disease agents Calcium channel RAAS system Medication group Medication Lipid modifying treatment Drugs for Beta blockers Mineral supplements Mineral Antithrombotics on the acting Agents the national average, which might be a consequence of selecting patients who stayed at least 1 year in the RAC. [3] code C10 C09 C07 C08 B01 Preventive M05 A11 ATC A12 Δ % 2.2 1.1 5.6* We found an increase in symptomatic medication toward death, −3.1 −1.8 −0.4 −0.4 −4.2* which was also seen in a small study looking at the last 3 months At 8.3

11.2 of life [9] and another study focusing at the last week of life. [21] 50.1 86.1 72.9 22.8 38.3 26.0 death% The increase was very subtle, however, and mostly caused by an increase in gastrointestinal medications. Overall use of analgesics, which are supposed to be the most prominent medication group in palliative care, [13] did not change. But the shift from parac- etamol to use indicates some awareness in the changing needs of residents at the end of life by the GP. systemic use systemic medication related disorders related gastrointestinal gastrointestinal disorders Corticosteroids for for Corticosteroids Medication group Medication Medication for for Medication Medication for acidic for Medication Diuretics Analgesics Antiepileptic Antiepileptic code C03 Symptomatic N02 N05 N03 H02 A06 A03 A02 ATC RAAS = Renin-angiotensin-aldosterone-system. Table 2: Type of symptomatic, preventive and other medication used by residents 1 year before death versus at death at versus death before 1 year residents used by and other medication preventive 2: Type of symptomatic, Table < 0.05. Δ: percentage of residents taking medication at death – percentage of residents taking medication 365 days before death. death. before 365 days medication taking of residents – percentage death at medication taking of residents test (df = 552), P < 0.05. Δ: percentage *McNemar

28 29 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

Despite some deprescribing, the use of antithrombotics, antihy- not be generalisable to residents who died within a few months of pertensives, and osteoporosis medications was very high at the nursing home admission. end of life, similar to other studies. [10, 11, 21] An explanation for this high use could be the lack of consensus on what medications are considered solely preventive and therefore inappropriate at CONCLUSION AND IMPLICATIONS FOR FURTHER RESEARCH the end of life. [22] We included antithrombotics, lipid-modify- ing agents, antihypertensives and osteoporosis medication, but The awareness of deprescribing inappropriate medication at the 2 other studies have also included iron, antibiotics, acid reducers end of life is growing throughout the literature. Recent articles and medications used in diabetes. [8] An exception to preventive have been published guiding the process of deprescribing [5, 23, medications, are lipid-modifying agents. These medications, es- 24] and shared decision making at the end of life. [25] But there pecially statins, were unanimously classified as preventive med- still remains a lack of high quality evidence guiding deprescribing ication and have been explored the most. [8] This attention to at the end of life. [26] For example has a number needed statins might have led to growing awareness of its inappropri- to treat of 120 patients over 6 years to prevent one cardiovascular ateness at the end of life, resulting in early deprescribing by GPs. event and a number needed to harm of 73 for a non-trivial bleed- This could explain the lower use of statins compared to other pre- ings, based on a study population with a mean age of 57 years. ventive medication we found in our study. [27] The figures are likely to be different in an older population. Furthermore, contradictory recommendations and variation in Strengths and limitations interpretations of guidelines leads to clinical uncertainty. [28] An This study is unique in investigating changes in prescribing of example is the most recent discussion on blood pressure control symptomatic and preventive medication in the last year of life in in older patients. [29] Exploring the use of preventive and symp- a relatively large group of residents. We based the classification tomatic medication at the end of life is a first step to improve the of medications on current guidelines. Some limitations need be quality of medication use for these patients. taken into consideration when interpreting our results. Firstly, we were using medication supply data and therefore were not able to ascertain actual medication intake. However, the weekly medica- tion supply ensured that the dataset remained relatively sensitive to change. Secondly, in line with other studies using dispensing data, we had no recorded indication for prescribed medication and therefore our medication classification was an approximation. We used the palliative care guidelines for classification of medica- tion and rely on prescribing following the guidelines for correct classification. Thirdly, we were not able to cluster our data at the level of prescriber since each nursing home resident in Australia has his or her own prescriber. Fourthly, by investigating prescrib- ing in the last year of life we had to exclude residents who stayed in the nursing home facility for a shorter time. Our results may

30 31 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

REFERENCES 17. WHO Collaborating Centre for Drug Statistics Methodology. ATC/DDD In- dex — Antithrombotic agents. 2016. https://www.whocc.no/atc_ddd_index 1. Holmes HM, Hayley DC, Alexander GC, Sachs GA. Reconsidering medication appro- /?code=B01A&showdescription=no. Accessed March 2017. priateness for patients late in life. Arch Int Med. 2006;166(6):605–609. 18. NPS Medicinewise. Medicines for osteoporosis. 2017. http://www.nps.org.au/ 2. Taxis K, O’Sullivan D, Cullinan S, Byrne S. Drug utilization in older people. In: Else- conditions/hormones--and-nutritional-problems/bone-disorders-­and- viers M, Wettermark B, Almarsdóttir A, et al, editors. Drug utilization research: calcium-metabolism/osteoporosis/for-individuals/medicines. Accessed Mar 2017. Methods and applications. London: Wiley-Blackwell; 2016. p. 259–269. 19. WHO Collaborating Centre for Drug Statistics Methodology. ATC/DDD Index — 3. Australian Institute of Health and Welfare. Residential aged care in Austra- Lipid modifying agents. 2016. https://www.whocc.no/atc_ddd_index/?code=C10. lia 2010–11: a statistical overview. 2017. http://www.aihw.gov.au/publication-­ Accessed March 2017. 2 detail/?id=10737422821. Accessed Aug 2017. 20. Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in 4. Taxis K, Kochen S, Wouters H, et al. Cross-national comparison of medication use in logistic regression. Source Code Biol Med. 2008;3:17. Australian and Dutch nursing homes. Age Ageing. 2017;46(2):320–323. 21. Jansen K, Schaufel MA, Ruths S. Drug treatment at the end of life: an epidemiologic 5. Scott IA, Hilmer SN, Reeve E, et al. Reducing inappropriate polypharmacy: the pro- study in nursing homes. Scand J Prim Health Care. 2014;32(4):187–192. cess of deprescribing. JAMA Intern Med. 2015;175(5):827–834. 22. Todd A, Husband A, Andrew I, Pearson SA, Lindsey L, Holmes H. Inappropriate 6. Lee SJ, Leipzig RM, Walter LC. Incorporating lag time to benefit into prevention de- prescribing of preventative medication in patients with life-limiting illness: a cisions for older adults. JAMA. 2013;310(24):2609–2610. systematic review. BMJ Support Palliat Care. 2017;7(2):113–121. 7. Maddison AR, Fisher J, Johnston G. Preventive medication use among persons with 23. Granas AG, Stendal Bakken M, Ruths S, Taxis K. Deprescribing for frail older peo- limited life expectancy. Prog Palliat Care. 2011;19(1):15–21. ple — Learning from the case of Mrs. Hansen. Res Social Adm Pharm. 2017 8. Poudel A, Yates P, Rowett D, Nissen LM. Use of Preventive Medication in Patients [Epub ahead of print]. With Limited Life Expectancy: A Systematic Review. J Pain Symptom Manage. 24. Wouters H, Scheper J, Koning H, et al. Discontinuing inappropriate medication use 2017;53(6):1097–1110. in nursing home residents: A cluster randomized controlled trial. Ann Intern 9. Blass DM, Black BS, Phillips H, et al. Medication use in nursing home residents with Med. 2017;167(9):609–617. advanced dementia. Int J Geriatr Psychiatry. 2008;23(5):490–496. 25. Jansen J, Naganathan V, Carter SM, et al. Too much medicine in older people? De- 10. Heppenstall CP, Broad JB, Boyd M, et al. Medication use and potentially inappro- prescribing through shared decision making. BMJ. 2016;353:i2893. priate medications in those with limited prognosis living in residential aged 26. Tjia J, Velten SJ, Parsons C, Valluri S, Briesacher BA. Studies to reduce unneces- care. Australas J Ageing. 2016;35(2):18–24. sary medication use in frail older adults: a systematic review. Drugs Aging. 11. Onder G, Liperoti R, Foebel A, et al. Polypharmacy and mortality among nursing 2013;30(5):285–307. home residents with advanced cognitive impairment: results from the SHELTER 27. Seshasai SR, Wijesuriya S, Sivakumaran R, et al. Effect of aspirin on vascular and study. J Am Med Dir Assoc. 2013;14(6):450.e7–12. nonvascular outcomes: meta-analysis of randomized controlled trials. Arch In- 12. WHO Collaborating Centre for Drug Statistics Methodology: ATC/DDD Index. tern Med. 2012;172(3):209–216. https://www.whocc.no/atc_ddd_index/ (2018). Accessed Mar 2018. 28. Alhawassi TM, Krass I, Pont LG. Hypertension in Older Persons: A Systematic Re- 13. Australian Government Department of Health. The Pharmaceutical Benefits Scheme for view of National and International Treatment Guidelines. J Clin Hypertens Palliative Care. 2015. https://www.pbs.gov.au/browse/palliative-care. Accessed Mar 2017. (Greenwich). 2015;17(6):486–492. 14. Palliative Care Expert Group. Therapeutic guidelines: palliative care. 3rd ed. Mel- 29. Williamson JD, Supiano MA, Applegate WB, et al. Intensive vs Standard Blood Pres- bourne: Therapeutic Guidelines Limited; 2010. sure Control and Cardiovascular Disease Outcomes in Adults Aged >/=75 Years: A Randomized Clinical Trial. JAMA. 2016;315(24):2673–2682. 15. Hilmer SN, Gnjidic D, Le Couteur DG. Thinking through the medication list — ap- propriate prescribing and deprescribing in robust and frail older patients. Aust Fam Physician. 2012;41(12):924–928. 16. NPS Medicinewise. Blood pressure lowering medicines. 2016. http://www.nps. org.au/conditions/heart-blood-and-blood-vessel-conditions/blood-pressure/ for-health-professionals/managing-blood-pressure-based-on-absolute-risk/ treatment-with-bp-lowering-medicines. Accessed Mar 2017.

32 33 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

Appendix: Classification of medications used by our cohort into symptom- ATC code Name Category atic, preventive or other B03BA01 Cyanocobalamin Other B03BB Folic acid Other ATC code Name Category B03XA02 Darbepoetin alfa Other A01AD11 Various agents for local oral treatment Other C01AA05 Digoxin Other A02AB01 Hydroxide Other C01BC04 Flecainide Other A02AD01 Ordinary salt combinations Other C01BD01 Amiodarone Other A02AF02 Ordinary salt combinations and antiflatulents Other 2 C01CA24 Epinephrine Other A02BA03 Other C01DA02 Glyceryl Trinitrate Other A02BX13 Other C01DA08 Isosorbide Dinitrate Other A03AA04 Mebeverine Other C01DA14 Isosorbide Mononitrate Other A03AX Other drugs for functional gastrointestinal disorders Other C01DX16 Nicorandil Other A05BA03 Silymarin Other C01EB09 Ubidecarenone Other A06AA Softeners, emollients Other G01AF02 Clotrimazole Other A06AC03 Sterculia Other G02CB03 Cabergoline Other A07C Electrolytes with carbohydrates Other G03BA03 Testosterone Other A07EC01 Sulfasalazine Other G03HA01 Cyproterone Other A07EC02 Mesalazine Other G04BX citrotartrate Other A09A Digestives, including Other H03AA01 Levothyroxine sodium Other A10AB Fast-acting insulins Other H03BA02 Propylthiouracil Other A10AC Intermediate-acting insulins Other H03BB01 Carbimazole Other A10AD Intermediate- or long-acting combined with fast-act- Other ing insulins H04AA01 Glucagon Other A10BA02 Metformin Other J05AH02 Oseltamivir Other A10BB01 Glibenclamide Other L01AA02 Chlorambucil Other A10BB07 Glipizide Other L01BC02 Fluorouracil Other A10BB09 Gliclazide Other L01BC06 Capecitabine Other A10BG03 Pioglitazone Other L01XX05 Hydroxycarbamide Other A11DA01 Thiamine Other L02AE02 Leuprorelin Other A11GB Ascorbic acid, combinations Other L02AE03 Goserelin Other A11JD Other products, combinations Other L02BA01 Tamoxifen Other A12BA Potassium Other L02BB02 Nilutamide Other A12BA01 Potassium chloride Other L02BG04 Letrozole Other A12CA01 Sodium chloride Other L02BG06 Exemestane Other A12CB01 Zinc sulfate Other L03AB08 Interferon beta-1b Other A12CC Other L04AX03 Methotrexate Other A12CC05 Magnesium aspartate Other M01AC01 Other B02BA01 Phytomenadione Other M01AC06 Other B03A Iron preparations Other M01AH01 Other

34 35 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

ATC code Name Category ATC code Name Category M01AX05 Other N06AB04 Other M01AX25/ and Glucosamine Other N06AB06 Other M01AX05 N06AB08 Other M03BC01 citrate Other N06AB10 Other M04AA01 Allopurinol Other N06AF03 Other M04AC01 Colchicine Other N06AG02 Moclobemide Other N02AC04 Other 2 N06AX03 Other N02AC54 Dextropropoxyphene, combincations excl. Other psycholeptics N06AX11 Other N02AX02 Other N06AX18 Reboxetine Other N02BA01 Acetylsalicylic acid Other N06AX23 Desvenlafaxine Other N03AA03 Other N06BA07 Modafinil Other N03AX09 Other N06DA02 Donepezil Other N03AX14 Other N06DA03 Rivastigmine Other N04AA01 Trihexylphenidyl Other N06DA04 Galantamine Other N04AA02 Biperiden Other N06DX01 Memantine Other N04BA02 Levodopa and decarboxylase inhibitor Other N07BA01 Other N04BA03 Levodopa, decarboxylase inhibitor and COMT Other N07CA01 Betahistine Other inhibitor P01BA02 Hydroxychloroquine Other N04BB01 Amantadine Other P01BC01 Quinine Other N04BC02 Pergolide Other P02CF01 Ivermectin Other N04BC05 Other P03AC04 Permethrin Other N04BC07 Other R03BA07 Mometasone Other N04BD01 Selegine Other R02AA03 Dichlorobenzyl Other N04BX02 Entacapone Other R02AD02 Other N05AB06 Other R03AC02 Salbutamol Other N05AC01 Pericyazine Other R03AC03 Terbutaline Other N05AC02 Other R03AK06 Fluticasone and Salmeterol Other N05AF01 Flupenthixol Other R03AK07 Formoterol and Budesonide Other N05AH04 Other R03BA01 Beclomethasone Other N05AN Other R03BA02 Budesonide Other N05AX12 Other R03BA05 Fluticasone Other N05BA08 Other R03BB01 Ipratropium Other N05CF01 Other R03BB04 Tiotropium bromide Other N05CF02 Other R03DA04 Theophylline Other N06AA02 Other R05CA12 Hederae helicis folium Other N06AA16 Dosuleptin Other R05CB02 Bromhexine Other N06AB03 Other R05DA04 Other

36 37 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

ATC code Name Category ATC code Name Category R05DA08 Pholcodine Other A02BC02 Symptomatic R05DA09 Dextromethorphan Other A02BC03 Symptomatic R05DA12 Other A02BC04 Symptomatic A02BC05 Symptomatic R06AA02 Other A02BX02 Symptomatic R06AB02 Dexchlorpheniramine Other A03AB02 Glycooyrronium bromide Symptomatic R06AE07 Cetirizine Other A03AB05 Propantheline Symptomatic 2 R06AX02 Other A03BA01 Atropine sulfate Symptomatic R06AX13 Loratadine Other A03FA01 Metoclopramide Symptomatic R06AX26 Fexofenadine Other A03FA02 Cisapride Symptomatic V03AB33 Hydroxycobolamin Other A03FA03 Domperidone Symptomatic V03AE01 Polystyrene Sulfonate Other A04AA01 Ondansetron Symptomatic V03AG Sodium itamine phosphate Other A04AA02 Granisetron Symptomatic A11CC Vitamine D and itamin D analogues Preventive A04AA03 Symptomatic A04AA04 Dolasetron Symptomatic A12A Calcium Preventive A04AD01 Hyoscine hydrobromide Symptomatic B01 Antithrombotic agents Preventive A04AD10 Dronabinol Symptomatic C02 Antihypertensives Preventive A04AD11 Symptomatic C03 Diuretics (except for hydrochlorothiazide, frusemide, Preventive A04AD12 Symptomatic spironolactone) A06AA01 Liquid paraffin Symptomatic C07 Betablocking agents Preventive A06AA02 Docusate Symptomatic C08 Calcium channel blockers (except for nifedipine and Preventive diltiazem) A06AB02 Bisacodyl Symptomatic A06AB06 Senna glycosides Symptomatic C09 Agents acting on the renin angiotensin system Preventive A06AB08 Sodium picosulphate Symptomatic C10A Lipid modifying agents (plain) Preventive A06AB56 Senna glycosides combinations Symptomatic C10B Lipid modifying agents (combinations) Preventive A06AC01 Ispaghula (psylla seeds) Symptomatic G03C Estrogen Preventive A06AC53 Stericula combinations Symptomatic G03F Progesteron Preventive A06AD11 Lactulose Symptomatic G03XC01 Raloxifene hydrochloride Preventive A06AD15 Macrogol Symptomatic G04CA03 hydrochloride Preventive A06AD17 Sodium phosphate Symptomatic H05AA02 Teriparatide Preventive A06AD18 Sorbitol Symptomatic H05BA Calcitonin Preventive A06AG11 Sorbitol Lauryl Sulfoacetate and combinations Symptomatic H05BX01 Cinacalcet Preventive A06AH01 Methylnaltrexone Symptomatic M05BA Bisphosphonates (except for clodronic acid, pamid- Preventive A06AH04 Symptomatic romic acid, ibedronic acid, zoledronic acid) A06AX01 Glycerol Symptomatic M05BB Bisphosphonates combinations Preventive A07DA03 Loperamide Symptomatic M05BX03 Strontium ranelate Preventive A09AA02 Pancrelipase Symptomatic M05BX04 Denosumab (calcium & bone metabolism medicines) Preventive A10AE04 Long acting insulin Symptomatic A01AD02 Symptomatic B02AA02 Tranexamic acid Symptomatic A02BA02 Symptomatic B05B I.V. solutions Symptomatic A02BC01 Symptomatic

38 39 Identifying opportunities for deprescribing Preventive medications at the end of life in older nursing home residents

ATC code Name Category ATC code Name Category C01BB02 Symptomatic N02AA03 Symptomatic C03AA03 Hydrochlorothiazide Symptomatic N02AA05 Symptomatic C03CA01 Furosemide Symptomatic N02AB03 Symptomatic C03DA01 Spironolactone Symptomatic N02AE01 Symptomatic C08CA05 Nifedipine Symptomatic N02BE01 Paracetamol Symptomatic C08DB01 Diltiazem Symptomatic N02BE51 Codeine Symptomatic G04BD04 Symptomatic N03AB02 Symptomatic 2 G04BD08 succinate Symptomatic N03AE01 Symptomatic G04CA02 Symptomatic N03AF01 Symptomatic G04CB01 Symptomatic N03AG01 Sodium Symptomatic H01CB02 Octreotide Symptomatic N03AX12 Symptomatic H01CB03 Lanreotide Symptomatic N03AX16 Symptomatic H02AA02 Fludrocortisone Symptomatic N04AC01 Benzatropine Symptomatic H02AB02 Dexamethasone Symptomatic N05AA01 Symptomatic H02AB04 Methylprednisolone Symptomatic N05AA02 Symptomatic H02AB06 Prednisolone Symptomatic N05AB04 Symptomatic H02AB07 Prednisone Symptomatic N05AD01 Symptomatic H02AB09 Hydrocortisone Symptomatic N05AH03 Symptomatic H02AB10 Cortisone acetate Symptomatic N05AX08 Symptomatic J02AC01 Fluconazole oral Symptomatic N05BA01 Symptomatic J02AC02 Itraconazole oral Symptomatic N05BA04 Symptomatic J05AB01 Aciclovir (i.v.) Symptomatic N05BA06 Symptomatic J05AB09 Famciclovir Symptomatic N05BA12 Symptomatic J05AB11 Valaciclovir Symptomatic N05CD02 Symptomatic M01AB01 Indomethacin Symptomatic N05CD07 Symptomatic M01AB05 Symptomatic N05CD08 Symptomatic M01AB55 Diclofenac combinations Symptomatic N06AA09 Symptomatic M01AE01 Symptomatic N06AA10 Symptomatic M01AE02 Symptomatic N06AA12 Symptomatic M03BX01 Symptomatic N06AB05 Symptomatic M03CA01 Dantrolene Symptomatic N06AX16 Symptomatic M05BA02 Clodronic acid Symptomatic N06AX21 Symptomatic M05BA03 Pamidronic acid Symptomatic N06BA02 Dexamfetamine Symptomatic M05BA06 Ibedronic acid Symptomatic N06BA04 Symptomatic M05BA08 Zoledronic acid Symptomatic N07BC02 Methdadone Symptomatic Mouthwash Bioactive enzymes mouthwash Symptomatic R06AD01 Alimemazine Symptomatic N01AH03 Symptomatic R06AD02 Symptomatic N01AX03 Symptomatic R06AE03 Cyclizine Symptomatic N01BB02 Lignocaine Symptomatic N02AA01 hydrochloride Symptomatic

40 41 Deprescribing in older people General introduction and thesis outline

1 CHAPTER 3

ANTICHOLINERGIC AND SEDATIVE MEDICATION USE IN OLDER COMMUNITY- DWELLING PEOPLE: A NATIONAL POPULATION STUDY IN THE NETHERLANDS

Helene G van der Meer, Katja Taxis, Martina Teichert, AMG Fabienne Griens, Lisa G Pont, Hans Wouters

Submitted

42 43 Anticholinergic/sedative medication use in older community-dwelling people

ABSTRACT INTRODUCTION

Purpose Anticholinergic/sedative medications are frequently Despite their adverse effects on physical and cognitive function, prescribed to older adults, despite their adverse effects on phys- [1, 2] anticholinergic and sedative medications are frequently pre- ical and cognitive function. Most anticholinergic/sedative medi- scribed to older patients. [3, 4] Some medications are deliberately cations act on the central nervous system (CNS). Little is known prescribed for their anticholinergic or sedative effect, for example about prescribing patterns of these medications. inhaled for chronic airway diseases or benzodi- azepines for . However, for most medications the an- Aims To identify the proportion of older adults with a high anti- ticholinergic/sedative effect is a side effect. [5] Anticholinergic/ cholinergic/sedative load and to identify patient subgroups based sedative medications mostly act on the central nervous system on type of CNS-active medication used. (CNS) and include psycholeptics, psychoanaleptics and analgesics. 3 [6] So far, most research has focused on quantifying the cumula- Methods A cross-sectional study of a nationwide sample of pa- tive exposure of multiple anticholinergic/sedative medications in tients with anticholinergic/sedative medications dispensed by older patients with polypharmacy. [7] Little is known about the 1,779 community pharmacies in the Netherlands (90% of all com- prevalence of combinations of multiple anticholinergic/sedative munity pharmacies) in November 2016 was conducted. Patients medications resulting in a high load or whether subgroups of aged ≥65 years with a high anticholinergic/sedative load defined as these patients based on types of anticholinergic/sedative medica- having a Drug Burden Index (DBI) ≥1 were included. Proportion of tions used can be identified. patients with a high anticholinergic/sedative load was calculated by dividing the number of individuals in our study population by Latent class analysis (LCA) is a person-centred approach, which the 2.4 million older patients using medications dispensed from identifies underlying patterns within populations that cannot study pharmacies. Patient subgroups based on type of CNS-active be directly measured or observed. [8] In a population of older medications used were identified with latent class analysis. adults having a high anticholinergic/sedative load, LCA has the potential to identify subgroups of patients based on specific med- Results Overall, 8.7% (209,472 individuals) of older adults using ication patterns or types of anticholinergic/sedative medications medications had a DBI ≥1. Latent class analysis identified four pa- used. This is a novel approach to investigate medication use. In tient subgroups (classes) based on the following types of CNS-active this study, we will firstly determine the proportion of older adults medications used: ‘combined /psychoanaleptic medica- having a high cumulative anticholinergic/sedative load, and tion’ (class 1, 57.9%), ‘analgesics’ (class 2, 17.9%), ’anti-epileptic medi- secondly, we will perform a latent class analysis to identify sub- cation’ (class 3, 17.8%) and ‘anti-Parkinson medication’ (class 4, 6.3%). groups of patients based on the most likely type of CNS-active medications used. Conclusions A large proportion of older adults in the Netherlands had a high anticholinergic/sedative load. Four dis- tinct subgroups using specific CNS-active medication were iden- tified. Interventions aiming at reducing the overall anticholiner- gic/sedative load should be tailored to these subgroups.

45 INTRODUCTION Despite their adverse effects on physical and cognitive function, [1, 2] anticholinergic and sedative medications are frequently prescribed to older patients. [3, 4] Some medications are deliberately prescribed for their anticholinergic or sedative effect, for example inhaled anticholinergics for chronic airway diseases or for insomnia. However, for most medications the anticholinergic/sedative effect is a side effect. [5] Anticholinergic/sedative medications mostly act on the central nervous system (CNS) and include psycholeptics, psychoanaleptics and analgesics. [6] So far, most research has focused on quantifying the cumulative exposure of multiple anticholinergic/sedative medications in older patients with polypharmacy. [7] Little is known about the prevalence of combinations of multiple anticholinergic/sedative medications resulting in a high load or whether subgroups of these patients based on types of anticholinergic/sedative medications used can be identified.

Latent class analysis (LCA) is a person-centred approach, which identifies underlying patterns within populationsIdentifying that opportunities cannot be for deprescribingdirectly measured or observed. [8] In a population of older adults having a Anticholinergic/sedative medication use in older community-dwelling people high anticholinergic/sedative load, LCA has the potential to identify subgroups of patients based on specific medication patterns or types of anticholinergic/sedative medications used. This is a novel METHODS approach to investigate medication use. In this study, we will firstly determine the proportion of older and preparations for which daily dose could not be determined adults having a high cumulative anticholinergic/sedative load, and secondly, we will perform a latent were excluded from the DBI calculation. These comprised derma- Study design & setting tological, gastro enteral-, nasal-, rectal- and vaginal preparations, class analysis to identify subgroups of patients based on the most likely type of CNS-active A cross-sectional study on a nationwide sample of patients with oral fluids, oral- and sublingual sprays, oral drops and parenteral medications used. prescriptions for anticholinergic/sedative medications dispensed medications, but also ‘as needed’ medications. Our database did

by community pharmacies in the Netherlands in November 2016 not include data of medications dispensed ‘over the counter’. METHODSwas conducted. Data were provided by the Dutch Foundation of Pharmaceutical Statistics (Stichting Farmaceutische Kengetallen, We included all medications classified as anticholinergic by Duran Study designSFK), [9] & whichsetting identified 783,540 older patients aged 65 years and et al. [6] Secondly, we systematically reviewed all other medica- A crossover-sectional from 1779 study community on apharmacies nationwide (90% sampleof total Dutchof patients com- with prescriptions for tions used in the Netherlands and included those with anticho- anticholinergic/sedativemunity pharmacies) medications using at leastdispensed one anticholinergic/sedative by community pharmacies in the Netherlands in linergic or sedative properties and those with frequently reported 3 Novembermedication 2016 was in theconducted. study period. Data Thewere SFK provided collects byexhaustive the Dutch data Foundation of Pharmaceutical sedative side effects reported in Dutch pharmacotherapeutic ref- Statisticsabout (Stichting medications Farmaceutische dispensed byKengetallen, more than 95%SFK), of all[9] community which identified 783,540 older patients erence sources.[13, 14] aged 65pharmacies years and in theover Netherlands. from 1779 [9] community Dutch community pharmacies pharmacies (90% of total Dutch community pharmacies)keep completeusing at electronicleast one medicationanticholinergic/sedative records of their medication patients and in the study period. The SFK Following the formula above, the DBI per medication ranged collects patientsexhaustive usually data registerabout medications with a single dispensed pharmacy by morefor medication than 95% of all community pharmacies between 0 and 1, depending on the prescribed daily dose. If the in the Netherlands.supply (a closed [9] Dutch pharmacy community system). pharmacies [10] Our data keep therefore complete pro electronic- medication records of prescribed daily dose was similar to the minimum recommended their patientsvide a goodand patientsapproximation usually of register patients’ with overall a single medication pharmacy use. for medication supply (a closed daily dose, the DBI for that medication would be 0.5. In our study pharmacy system). [10] Our data therefore provide a good approximation of patients’ overall we include patients with a DBI ≥ 1. A DBI above this threshold medicationAnticholinergic use. and sedative load was considered a high anticholinergic/sedative load. Anticholinergic/sedative medication load was quantified with the Drug Burden Index (DBI). [11] Previous studies have identified Study population Anticholinergic and sedative load that a higher DBI was associated with an increased risk of medica- All older adults, aged ≥ 65 years, with a high anticholinergic/sed- Anticholinergic/sedativetion harm among oldermedication populations. load was [12] quantified The DBI was with calculated the Drug Burden Index (DBI). [11] ative load, that is a DBI ≥ 1, were identified from medication dis- Previoususing studies the havefollowing identified formula: that a higher DBI was associated with an increased risk of medication pensing records and included in the study. harm among older populations. [12] The DBI was calculated using the following formula: We excluded 16,498 patients (2,1% of all patients) from 32 pharma- ! cies (1,8% of all pharmacies in database) using a pharmacy infor- DBI = ! !! where D = prescribed daily dose and δ = the minimum recom- mation system with a specific software package, as this software mended daily dose according to Dutch pharmacotherapeutic ref- was known for reporting errors in dispensing dates. We also ex- erence sources. [13, 14] cluded 868 patients with unknown gender and/or age or reported age ≥ 110 years (0.11%). All prescription medications dispensed by the pharmacy with mild or strong anticholinergic and/or sedative (side-) effects with Data source a usage date in the study period (one month) were included in the The dataset contained demographic patient data that were col- DBI calculation. Medications without known prescribed daily dose lected by SFK, such as anonymous patient identification code, age,

46 47 Identifying opportunities for deprescribing Anticholinergic/sedative medication use in older community-dwelling people

gender, anonymous pharmacy code and medication data includ- model with n-classes was better than the previous model with ing generic name, daily defined dose, preparation form and World n-1 classes. Improvement was deemed significant if the associ- Health Organization Anatomical Therapeutic Chemical (ATC) ated p-value was < 0.05. Higher entropy, which is a quality in- code (2016). [15] dicator of classification ranging from 0–1, indicated a better classification. Entropy values > 0.8 were acceptable. To identify Outcomes & statistics clinically relevant classes, alongside goodness of fit, we only con- The proportion of older adults having a high anticholinergic/sed- sidered models with patient subgroups (classes) that consisted of ative load was calculated by dividing the number of individuals at least 5% of the study population. [8] As convergence of local in our study population by the number of older adults (aged ≥ 65 solutions is a common issue of LCA, we increased the number years) who were dispensed at least one medication with a usage of random starts when necessary to get global solutions. We date within the study period from one of the community phar- fixed thresholds of parameter estimates to the observed proba- 3 macies included in our study. bilities if necessary. Following these criteria above, we identified the best fitting model with n-classes and subsequently assigned Identification of subgroups of patients with a high anticholin- patients to their most likely class based on model probabilities. ergic/sedative load was examined with LCA in M-Plus version Demographic descriptives of all patients and of patients within 7.4. [16] Subgroup identification was based on most likely type each class were derived in SPSS version 25. of CNS-active medications (ATC code starting with N) used by a patient within each subgroup (class). We focused on CNS-active medications as these included most anticholinergic/sedative RESULTS medications. CNS-active medications were grouped by ATC code level 2 and were included in the analysis if used by at least 5% Proportion of older adults with high anticholinergic/ of the study population. Use of CNS-active medications per pa- sedative load tient was treated as a categorical variable (dispensed/ not dis- We found 766,174 older adults who were dispensed at least one pensed). LCA was performed in a successive forward manner. anticholinergic/sedative medication from one of 1747 commu- We started with a single class model with the assumption that nity pharmacies in the Netherlands (88% of total). Of this pop- all patients used the same types of CNS-active anticholinergic/ ulation 11,758 patients (1.5%) were excluded, as for these patients sedative medications. This corresponds to a standard descriptive the DBI could not be calculated. A total of 544,944 (71.1%) had a analysis of the medication use of the whole study population. DBI between 0 and 1 and 209,472 (27.3%) had a DBI ≥ 1. Patients Then successive LCA models were performed, adding one class with a DBI ≥ 1 were slightly more female (66.9% versus 62.3 and extra at a time. The most likely number of patient subgroups 60.7%), (Table 1). (classes) was identified by evaluating the statistical ‘goodness of fit’ of the different models with n-classes. Various goodness About 2.4 million older people were dispensed at least one medi- of fit statistics are available for LCA. We inspected the Bayesian cation within the study period from one of the 1747 study pharma- Inspection Criterion (BIC), the Lo-Mendell-Rubin Likelihood cies. Therefore, 31.9% of the Dutch older adults using medication ratio test (LMR) and the entropy. For the best fitting model, the were dispensed at least one anticholinergic/sedative medication BIC value should be lowest. The LMR tested whether the current and 8.7% had a high anticholinergic/sedative load (DBI ≥ 1).

48 49 Identifying opportunities for deprescribing Anticholinergic/sedative medication use in older community-dwelling people

Table 1: Characteristics of patients aged 65 years and over using at least one Table 2: Goodness of fit statistics of latent class analysis anticholinergic/sedative medication, those having a DBI between 0 and 1 and those having a DBI ≥ 1. Lo-Mendell-Rubin Percentage of patients in class Class BIC 2LL p-value Entropy 1 2 3 4 5 Patients aged 65 years and over 1 ------Using at least Having a DBI Having a DBI ≥ 1 one anticholin- between 0 and 1 (Study population) 2 1099816 16928 0.000 0.520 52.1 48.0 - - - ergic/ sedative 3 1093340 11551 0.000 0.691 37.7 37.4 24.9 - - medication 4† 1089137 339641 0.000 0.961 57.9 17.9 17.8 6.3 - Number of patients 766,174 (100%)* 544,944 (71.1%) 209,472 (27.3%) (%) 5 1093671 87912 0.333 0.830 35.7 25.0 13.7 13.0 12.6 Gender (% female) 62.3 60.7 66.9 BIC = Bayesian Information Criterion, 2LL = two log likelihood value. †Best fitting Age (mean (SD)) 75.9 (7.7) 75.8 (7.7) 75.9 (7.8) model. Number anticholiner- 1.5 (0.9) 1.1 (0.3) 2.6 (1.0) gic/sedative medica- 3 tions (mean (SD)) was most likely comprised of four classes. The BIC was lowest Top 5 used anticholin- Antidepressants Antidepressants ergic/ sedative med- (N06A, 26.1) (N06A, 18.6) (N06A, 46.7) for the four-class model, the p-values of the LMR indicated that ications by ATC code & Drugs for obstruc- Hypnotics & the four-class model was better than a three-class and a five- level 3 (ATC code, % tive airway disease sedatives of patients) (N05C, 19.2) (R03B, 15.0) (N05C, 33.4) class model and it had the clearest classification indicated by the Hypnotics & Anxiolytics highest entropy (Table 2). (N05B, 16.5%) sedatives (N05B, 33.2) (N05C, 14.1) Drugs for obstruc- Anxiolytics Opioids tive airway disease (N05B, 10.4) (N02A, 22.6) The four patient subgroups (classes) were described after their (R03B, 16.1) most likely type of CNS-medications used, namely: ‘combined Opioids Opioids Anti-epileptics (N02A, 13.1) (N02A, 9.2) (N03A, 18.6) psycholeptic/psychoanaleptic medication’ (class 1, 57.9%), ‘an- Patients using anti- 543,652 (71.0) 351,502 (64.5) 184,604 (88.1) algesics’ (class 2, 17.9%), ’anti-epileptic medication’ (class 3, cholinergic medica- tions (number (%)) 17.8%) and ‘anti-Parkinson medication’ (class 4, 6.3%), (Figure 1). Probabilities of a patient within each class to use a medication DBI = Drug Burden Index. ATC = Anatomical Therapeutical Chemical. *Of this popu- lation 11,758 patients (1.5%) were excluded, as for these patients the DBI could not be from the five types of CNS-active medications were derived from calculated. the LCA. Estimated probabilities were comparable to observed probabilities. Identification of patient subgroups (classes) using LCA Types of CNS-active medications used by at least 5% of the study Distribution of characteristics across the four identified population were psycholeptics (including , anxio- patient subgroups (classes) lytics, hypnotics/sedatives (ATC N05, 62.6%)), psychoanaleptics The four patient subgroups (classes) differed in age, gender, DBI (including antidepressants, psychostimulants and combinations and mean number of anticholinergic/sedative medications, of psycholeptics/psychoanaleptics (ATC N06, 48.7%)), analgesics (Table 3). Analgesics users (class 2) were oldest (77.3 (SD 8.3) (ATC N02, 23.4%), anti-epileptics (ATC N03, 18.6%) and anti-­ and anti-epileptic medication users (class 3) had the highest Parkinson medication (ATC N04, 8.4%). On these medication number of anticholinergic/sedative medications (3.0 (SD 1.2)). types a LCA for a two-, three-, four- and five-class model was per- Anti-Parkinson medication users (class 4) and anti-epileptic formed. Goodness of fit statistics indicated that the population medication users (class 3) had the lowest proportion of females.

50 51 Identifying opportunities for deprescribing Anticholinergic/sedative medication use in older community-dwelling people

Table 3: Characteristics of the study population and the four identified 1 ······· Observed ̶ ̶ ̶ Estimated classes. 0,8 Characteristic Class 1: Class 2: Class 3: Class 4: 0,6 Psycho- Analgesics Anti-epi- Anti-Parkin- 0,4 leptic and leptic son psycho-ana- 0,2 leptic 0 Number of patients 121,306 37,575 37,343 13,238 Size of class (%) 57.9 17.9 17.8 6.3 Age (mean (SD)) 75.9 (7.7) 77.3 (8.3) 74.7 (7.3) 75.7 (7.0) Gender (% women) 69.4 71.0 60.8 50.4 DBI (mean (SD)) 1.5 (0.5) 1.7 (0.7) 1.7 (0.7) 1.5 (0.5) Class 1: psycholeptic/psychoanaleptic (57.9%) Class 2: analgesics (17.9%) Number of anticholin- 2.4 (0.7) 2.8 (1.0) 3.0 (1.2) 2.7 (0.8) Class 3: anti-epileptic (17.8%) Class 4: anti-parkinson (6.3%) Figure 1: Estimated and observed probabilities of a patient within each ergic/ sedative medica- 3 class to use a medication from the five types of CNS-active medications. tions (mean (SD)) Medications used by at least 5% of study population included in Latent Class Analysis Antidepressants (ATC N06A) were the most commonly used Antidepressants (N06A) 55.8 35.0 36.7 24.3 Hypnotics & sedatives 39.4 33.4 22.4 9.8 medication group across all 4 classes, while the most frequently (N05C) used individual medications in each class were oxazepam (class Anxiolytics (N05B) 41.4 27.6 21.2 7.2 1, 23.9%), fentanyl (class 2, 37.8%), pregabalin (class 3, 40.1%) and Opioids (N02A) 0.0 96.2 26.6 9.0 Anti-epileptics (N03A) 0.0 0.0 100.0 12.2 levodopa with carbidopa or benserazide (class 4, 65.1%). A list Antipsychotics (N05A) 15.8 6.7 10.8 5.3 of the top 10 most used anticholinergic/sedative medications is Dopaminergic anti-Par- 1.7 2.2 3.5 98.7 shown in Appendix Table 1. kinson (N04B) Medications used by at least 5% of study population not included in Latent Class Analysis Anticholinergic inhal- 20.3 15.1 11.6 7.9 ants (R03B) DISCUSSION for 10.0 5.9 5.2 3.4 systemic use (R06A) Key findings Urologicals (G04B) 8.5 6.0 6.8 10.0 Cough suppressants 7.7 4.8 3.7 3.2 Nearly 1 in 10 Dutch older adults using medications had a high an- (R05D) ticholinergic/sedative load. We identified four subgroups (classes) of patients based on their most likely used type of CNS-active A key strength of this study is the use of latent class analysis to medications, described as patients using combined psycholeptic/ explore patterns of anticholinergic/sedative medication use in psychoanaleptic medication, analgesics, anti-epileptic medication a large nation-wide study sample of older adults. The following and patients using anti-Parkinson medications. limitations should be considered when interpreting our find- ings. First, we analysed medications with a usage date within the Strengths and limitations study period of one month. This included medications taken for This was an innovative study identifying patients with high anti- the whole period, but also medications taken for only part of the cholinergic/sedative loads and providing insight into the type of month. This may have overestimated the total daily anticholiner- medication contributing to this high load in individual patients. gic/sedative load for an individual, while medications dispensed

52 53 Identifying opportunities for deprescribing Anticholinergic/sedative medication use in older community-dwelling people

‘over the counter’ were not available. This may have led to an un- Interpretations and other studies, generalizability derestimation of the anticholinergic/sedative load. Second, like We found that in the population of Dutch older adults using in other pharmaco-epidemiological studies using similar data medications, about one third used at least one anticholinergic/ sources, medication-dispensing data are an approximation of ac- sedative medication and one in ten had a high anticholinergic/ tual medication use. [17] Third, we classified medications by its sedative load. This is in line with other studies, [26, 27] but ex- ATC code level 2. We did not have access to details on patient act numbers are difficult to compare due to differences in study comorbidities or the indications for medications included in populations and definitions used. Most patients in our study used our analysis. For example (National-) prescribing guidelines for psycholeptic and psychoanaleptic medications (class 1). While neuropathic pain recommend medications for a range of different the use of these medications may be appropriate for some older therapeutic subgroups, such as antidepressants (ATC code adults, potentially inappropriate use of these medications has N06AA) and anti-epileptics (ATC code N03). [18, 19] Furthermore, been widely reported. [28] We distinguished a subgroup of pa- 3 anti-epileptic medications are prescribed for behavioural disor- tients with pain, using strong opioids (class 2). Yet, anti-epileptic ders. [20] Within the group of anti-epileptic medication users, medications are also prescribed for the management of pain, par- we therefore could not distinguish between patients with epi- ticularly neuropathic pain, [18, 19] but are not used by patients lepsy, neuropathic pain or behavioural disorders. Finally, there is in class 2. As such, the class of anti-epileptic users (class 3) may no international consensus about which medications have anti- also include a considerable number of patients treated for pain. cholinergic/sedative properties. [21] A first attempt has been the In particular, this could be the group of 26.6% of patients in this systematic review on anticholinergic medications where we based class using opioids. But despite this probable overlap, anti-epilep- our list of anticholinergic medications on. [6] For sedative medi- tic users were more likely to be male compared to the total study cations, this is lacking. While anticholinergic effects are a result of population, suggesting that most anti-epileptic medications were muscarinic receptor blocking, [5] different pharmacological path- used to manage epilepsy rather than other symptoms or diseases, ways lead to sedation, of which most pathways are still unknown. as epilepsy is more common in men than women aged 65 years [22] Therefore, we based our list of sedative medications on a sys- and older. [29] We found a high proportion of males in the an- tematic analysis of relevant frequently reported (side-) effects in ti-Parkinson medication class (class 4), which is also in line with relevant reference sources. More work needs to be done, to come the national prevalence of Parkinson’s disease. [30] The small to an evidence-based list of medicines. This may limit the com- number of antipsychotics in this class might indicate that it in- parability of studies using the DBI. [23] Furthermore, although cludes predominantly patients suffering from Parkinson’s disease, anticholinergic and sedative medications are pharmacologically as most antipsychotics are contra-indicated in these patients. [31] different, they have similar negative consequences. [1, 2] This is The small number of antipsychotics in this class may actually why we quantified the combined load of anticholinergic and sed- reflect patients who have drug-induced Parkinsonism caused by ative medications. Other tools are available, which were restricted antipsychotics. [32] to anticholinergic medications, amongst those, one that shows promising results. [24, 25] Implications for practice Our findings give insight into the extent of anticholinergic/ sedative medication use and the different types of medications used that contribute to a high anticholinergic/sedative load. We

54 55 Identifying opportunities for deprescribing Anticholinergic/sedative medication use in older community-dwelling people

found that the majority of patients with a high anticholinergic/ REFERENCES sedative load used combinations of psycholeptic and psychoan- 1. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- aleptic medications. These medications are often inappropriately erties on cognitive function, delirium, physical function and mortality: a sys- used among older adults, increasing the risk of medication re- tematic review. Age Ageing. 2014;43(5):604–615. lated harm, such as falls and hospitalisation, and therefore should 2. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls be considered for deprescribing where appropriate. [33, 34] So in older people: systematic review of publications from a recent 5-year period. Eur J Clin Pharmacol. 2015;71(12):1429–1440. far however, few interventions have been effective in reducing a 3. Holvast F, van Hattem BA, Sinnige J, et al. Late-life depression and the associa- patient’s anticholinergic/sedative load. In our recent randomized tion with multimorbidity and polypharmacy: a cross-sectional study. Fam Pract. controlled trial we found that medication reviews were not effec- 2017;34(5):539–545. tive in reducing a high anticholinergic/sedative load among older 4. Bell JS, Mezrani C, Blacker N, et al. Anticholinergic and sedative medicines — community-dwelling patients. As a consequence different strat- prescribing considerations for people with dementia. Aust Fam Physician. 2012;41(1–2):45–49. 3 egies for identifying those patients who are in greatest need for 5. Nishtala PS, Salahudeen MS, Hilmer SN. Anticholinergics: theoretical and clinical medication optimization and who could benefit from interven- overview. Expert Opin Drug Saf. 2016;15(6):753–768. tion are needed. [35] Targeting specific anticholinergic/sedative 6. Duran CE, Azermai M, Vander Stichele RH. Systematic review of anticholinergic risk medications and tailoring interventions to specific subgroups of scales in older adults. Eur J Clin Pharmacol. 2013;69(7):1485–1496. patients might be the most successful strategy to reduce the over- 7. Taxis K, Kochen S, Wouters H, et al. Cross-national comparison of medication use in all anticholinergic/sedative load. Australian and Dutch nursing homes. Age Ageing. 2017;46(2):320–323. 8. Nylund KL, Asparouhov T, Muthén BO. Deciding on the Number of Classes in La- tent Class Analysis and Growth Mixture Modeling: A Monte Carlo Simulation Study. Struct Equ Modeling. 2007;14(4):535–569. CONCLUSIONS 9. Foundation for Pharmaceutical Statistics. Foundation for Pharmaceutical Statistics. 2018. https://www.sfk.nl/english. Accessed April 2018. A large proportion of older adults in the Netherlands had a high 10. Buurma H, Bouvy ML, De Smet PA, Floor-Schreudering A, Leufkens HG, Egberts anticholinergic/sedative load. Four distinct subgroups were iden- AC. Prevalence and determinants of pharmacy shopping behaviour. J Clin Pharm Ther. 2008;33(1):17–23. tified. Interventions aiming at reducing the overall anticholiner- 11. Hilmer SN, Mager DE, Simonsick EM, et al. A drug burden index to define gic/sedative load should be tailored to these subgroups. the functional burden of medications in older people. Arch Intern Med. 2007;167(8):781–787. 12. Wouters H, van der Meer H, Taxis K. Quantification of anticholinergic and sedative drug load with the Drug Burden Index: a review of outcomes and methodologi- cal quality of studies. Eur J Clin Pharmacol. 2017;73(3):257–266. 13. Farmacotherapeutisch Kompas. Dutch pharmacotherapeutic reference source. 2018. https://www.farmacotherapeutischkompas.nl/. Accessed April 2018. 14. KNMP Kennisbank. Dutch pharmacotherapeutic reference source. 2018. https:// kennisbank.knmp.nl/. Accessed April 2018. 15. WHO Collaborating Centre for Drug Statistics Methodology. ATC/DDD Index. 2017. https://www.whocc.no/atc_ddd_index/. Accessed March 2018. 16. Muthén L, Muthén B. Mplus User’s Guide. 8th ed. Los Angeles: Muthén & Muthén; 2017.

56 57 Identifying opportunities for deprescribing Anticholinergic/sedative medication use in older community-dwelling people

17. Sodihardjo-Yuen F, van Dijk L, Wensing M, De Smet PAGM, Teichert M. Use of 31. Divac N, Stojanovic R, Savi Vujovic K, Medic B, Damjanovic A, Prostran M. The Ef- pharmacy dispensing data to measure adherence and identify nonadherence ficacy and Safety of Medications in the Treatment of Psychosis in with oral hypoglycemic agents. Eur J Clin Pharmacol. 2017;73(2):205–213. Patients with Parkinson’s Disease. Behav Neurol. 2016;2016:4938154. 18. Verenso. Multidisciplinary guideline pain in older people. 2016. https://www.ver- 32. Lopez-Sendon J, Mena MA, de Yebenes JG. Drug-induced parkinsonism. Expert enso.nl/_asset/_public/Richtlijnen_kwaliteit/richtlijnen/database/VER-003-32- Opin Drug Saf. 2013;12(4):487–496. Richtlijn-Pijn-deel1-v5LR.pdf. Accessed May 2018. 33. By the American Geriatrics Society 2015 Beers Criteria Update Expert Panel. Amer- 19. National Institute for Health and Care Excellence (NICE). Neuropathic pain in ican Geriatrics Society 2015 Updated Beers Criteria for Potentially Inappropriate adults. 2018. https://www.nice.org.uk/guidance/cg173/resources/neuropathic-­ Medication Use in Older Adults. J Am Geriatr Soc. 2015;63(11):2227–2246. pain-in-adults-pharmacological-management-in-nonspecialist-settings-pdf- ­ 34. O’Mahony D, O’Sullivan D, Byrne S, O’Connor MN, Ryan C, Gallagher P. STOPP/ 35109750554053. Accessed May 2018. START criteria for potentially inappropriate prescribing in older people: version 20. National Healthcare Institute [Zorginstituut Nederland]. Anti-epileptics. 2018. 2. Age Ageing. 2015;44(2):213–218. https://www.farmacotherapeutischkompas.nl/bladeren/groepsteksten/anti_epileptica. 35. van der Meer HG, Wouters H, Pras N, Taxis K. Reducing the anticholinergic and Accessed September 2018. sedative load in older patients on polypharmacy by pharmacist-led medication 3 21. Pont LG, Nielen JT, McLachlan AJ, et al. Measuring anticholinergic drug exposure review: A randomized controlled trial. BMJ Open. 2018;8(7):e019042. in older community-dwelling Australian men: a comparison of four different measures. Br J Clin Pharmacol. 2015;80(5):1169–1175. 22. Yu X, Franks NP, Wisden W. Sleep and Sedative States Induced by Targeting the Histamine and Noradrenergic Systems. Front Neural Circuits. 2018;12:4. 23. Faure R, Dauphinot V, Krolak-Salmon P, Mouchoux C. A standard international ver- sion of the Drug Burden Index for cross-national comparison of the functional burden of medications in older people. J Am Geriatr Soc. 2013;61(7):1227–1228. 24. Wauters M, Klamer T, Elseviers M, et al. Anticholinergic Exposure in a Cohort of Adults Aged 80 years and Over: Associations of the MARANTE Scale with Mor- tality and Hospitalization. Basic Clin Pharmacol Toxicol. 2017;120(6):591–600. 25. Klamer TT, Wauters M, Azermai M, et al. A Novel Scale Linking Potency and Dos- age to Estimate Anticholinergic Exposure in Older Adults: the Muscarinic Ace- tylcholinergic Receptor ANTagonist Exposure Scale. Basic Clin Pharmacol Toxi- col. 2017;120(6):582–590. 26. Gnjidic D, Hilmer SN, Hartikainen S, et al. Impact of high risk drug use on hospi- talization and mortality in older people with and without Alzheimer’s disease: a national population cohort study. PLoS One. 2014;9(1):e83224. 27. Wilson NM, Hilmer SN, March LM, et al. Associations between drug burden index and physical function in older people in residential aged care facilities. Age Age- ing. 2010;39(4):503–507. 28. Tommelein E, Mehuys E, Petrovic M, Somers A, Colin P, Boussery K. Potentially in- appropriate prescribing in community-dwelling older people across Europe: a systematic literature review. Eur J Clin Pharmacol. 2015;71(12):1415–1427. 29. National Institute of Public Health and the Environment. Prevalence of epilepsy in general practice. 2018. https://www.volksgezondheidenzorg.info/onderwerp/epi- lepsie/cijfers-context/huidige-situatie. Accessed May 2018. 30. National Institute of Public Health and the Environment. Prevalence of Parkinson’s dis- ease in general practice. 2018. https://www.volksgezondheidenzorg.info/­onderwerp/ ziekte-van-parkinson/cijfers-context/huidige-situatie. Accessed May 2018.

58 59 Identifying opportunities for deprescribing

Appendix Table 1: Top 10 used anticholinergic/sedative per class identified with the latent class analysis

Class 1: Class 2: Class 3: Class 4: Psycholeptic and Analgesics Anti-epileptic Anti-Parkinson psycho-analeptic (n=37,575) (n=37,343) (n=13,238) (n=121,306) Oxazepam (23.9%) Fentanyl (37.8%) Pregabaline (40.1%) Levodopa and decarboxylase inhibitor (65.1%) Temazepam (18.8%) Tramadol (35.2%) Valproic acid Pramipexole (13.2%) (36.3%) Tiotropium bro- Oxycodone (34.5%) Gabapentin (12.6%) Ropinirole (17.5%) mide (17.1%) Citalopram (9.6%) Temazepam (17.7%) Clonazepam (12.5%) Amantadine (8.8%)

Paroxetine (9.5%) Oxazepam (15.2%) Carbamazepine Rivastigmine (7.5%) (12.0%) Amitriptyline Amitriptyline Levetirace- Tiotropium bro- (9.4%) (12.4%) tam (11.8%) mide (6.5%) Mirtazapine (9.2%) Tiotropium bro- Amitriptyline Amitriptyline mide (11.9%) (11.3%) (6.1%) Codeine (7.7%) Diazepam (5.7%) Oxycodone (11.2%) Solifenacine (5.8%) Lorazepam (6.9%) Metoclopramide Temazepam (11.1%) Levodopa, decar- (5.2%) boxylase inhibitor and COMT inhibi- tor (5.7%)

COMT = Catechol-O-methyl transferase

60 Deprescribing in older people General introduction and thesis outline

1 CHAPTER 4

DECREASING THE LOAD? IS A MULTIDISCIPLINARY MULTISTEP MEDICATION REVIEW IN OLDER PEOPLE AN EFFECTIVE INTERVENTION TO REDUCE A PATIENT’S DRUG BURDEN INDEX? PROTOCOL OF A RANDOMISED CONTROLLED TRIAL

Helene G van der Meer, Hans Wouters, Rolf van Hulten, Niesko Pras, Katja Taxis

BMJ Open. 2015;5(12):e009213.

62 63 Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

ABSTRACT INTRODUCTION

Introduction Older people often use medications with anticho- Older individuals use more medications than any other age group. linergic or sedative side effects, which increase the risk of falling [1] They typically suffer from multiple acute and chronic diseases, and worsen cognitive impairment. The Drug Burden Index (DBI) which often necessitates the use of multiple concomitant medi- is a measure of the burden of anticholinergic and sedative medi- cations. [2] Polypharmacy in combination with age-related phar- cations. Medication reviews are typically done by a pharmacist in macokinetic and pharmacodynamic changes, such as decrease collaboration with a general practitioner to optimise the medica- in renal function and altered drug responsiveness, predisposes tion use and reduce these adverse drug events. We will evaluate older individuals to an increased risk of drug-drug interactions, whether a Multidisciplinary Multistep Medication Review (3MR) drug-disease interactions, adverse drug events and potentially is an effective intervention to reduce a patient’s DBI. inappropriate prescribing (PIP). [3–5] Many PIP instances are attributable to medication with anticholinergic and/or sedative Methods A randomised controlled trial including 160 patients properties. [6, 7] Those medications increase the risk of falls in from 15 community pharmacies will be conducted. Per pharmacy, older people and worsen cognitive impairment resulting in prob- 1 pharmacist will perform a structured 3MR in close collaboration lems in activities of daily living (ADL). [8–11] Around 600 med- 4 with the general practitioner, including the objective to reduce ications are known to have anticholinergic effects to a greater the DBI. or lesser extent, [12] and many of these are widely used among older people, especially cardiovascular medication [13, 14] and Analysis Primary outcome-the difference in proportion of pa- medicines acting on the central nervous system. Hypnotics and tients having a decrease of the DBI ≥ 0.5 between the intervention sedatives are among the most commonly used psychotropic med- group and control group at follow-up. Secondary outcomes-an- ications, especially in the very old. [15] A Finnish study found ticholinergic and sedative side effects, falls, cognitive function, that almost one-third of adults aged > 75 years used or activities of daily living, quality of life, hospital admission, and medication, and almost one-tenth used mortality. or antipsychotic medicines. [16] Given these findings, decreasing the exposure to anticholinergic and sedative medications is likely Ethics and dissemination The burden of patients will be kept to result in important health benefits for older people. The Drug at a minimum. The 3MR can be considered as usual care by the Burden Index (DBI) calculates an individual patient’s exposure pharmacist and general practitioner. Medical specialists will be to anticholinergic and sedative medications taking into account consulted if necessary. The intervention is specifically aimed at the medicine dosage. [17] A recent literature review shows that older community-dwelling patients in an attempt to optimise the DBI is associated with impairments in physical and cognitive prescribing, in particular to reduce medication with anticholin- functions of older individuals. [18, 19] ergic and sedative properties. Study results will be published in peer-reviewed journals and will be distributed through informa- Medication reviews are seen as a promising strategy to enhance tion channels targeting professionals. the quality of prescribing, although there is still a lack of evi- dence on cost or clinical effectiveness. Medication reviews could Trial Registration number NCT02317666; Pre-results. be more effective by targeting high-risk groups and focusing on

65 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

medicines that could be safely stopped. [20–23] According to control group will receive the 3MR after the study period (post- Dutch guidelines, a medication review is a structured critical ex- poned intervention). Primary and secondary study outcomes will amination of a patient’s medication, done by a pharmacist and be determined for intervention group and control group at base- a general practitioner, to reach an agreement with the patient line and at 3 months follow-up after the intervention. about treatment, optimising the effectiveness of medicines and minimising the number of medication related problems. [24] Participants and setting Annual Multidisciplinary Multistep Medication Reviews (3MR) Our aim is to enrol a minimum of 160 participants from 15 com- are recommended for older chronic polypharmacy patients with munity pharmacies in the region of Groningen, the Netherlands additional risk factors. However, criteria used so far-living in a (see Sample size calculation section). We will approach a total of nursing home, decreased renal clearance (estimated glomerular 400 patients to recruit about 160 participants. One pharmacist filtration rate < 50/ml/min/1.73m2), decreased cognitive function, will conduct the medication reviews in each pharmacy. As one increased risk of falling, signals of decreased medication adher- community pharmacy is mostly associated with several medical ence or unplanned hospital admission [24]-form an inadequate practices, the pharmacist will collaborate with different GPs, but demarcation of the high risk population. Therefore, in the pres- only one GP for each patient. ent study we used the DBI to identify high-risk patients who 4 could benefit from medication reviews. The aim of our study is Pharmacists to evaluate whether a 3MR is an effective intervention to reduce Inclusion criteria a patient’s DBI. • Established collaboration with GP • Experience with medication reviews (accredited commu- nity pharmacist or registered pharmacist in training to be METHODS AND ANALYSIS accredited).

Study design Participants A single-blinded randomised controlled trial will be con- Inclusion criteria ducted in line with the ‘Consolidated Standards of Reporting • Aged ≥ 65 years Trials (CONSORT)’statement (https://www/consort-statement. • Living independently org) and the ‘Standard Protocol Items: Recommendations for • Chronic polypharmacy (≥ 5 medications for ≥ 3 months Interventional Trials (SPIRIT)’criteria. (https://www.spirit-­ [24] and DBI ≥ 1) statement.rug) Patients will be recruited from pharmacies and • Use of at least one medication with ATC N05 or N06 will be randomly allocated into control group and intervention • Written informed consent (IC) group (refer to Selection process, randomisation, intervention allocation and blinding). The intervention consists of a 3MR con- Exclusion criteria ducted by the pharmacist in collaboration with the GP. The main • Palliative care only aim of the medication review is to optimise the patient’s medi- • Limited life expectancy (< 3 months) cation with a focus on lowering the DBI by reducing medication • Urgently in need a medication review with anticholinergic and sedative properties. Participants in the • No Dutch language skills

66 67 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

• Advanced dementia medication and patients’ expectations and preferences about their • Received a medication review within 9 months before the medication will be discussed. study period. Step 2: pharmacotherapeutic medication review Sample size calculation The pharmacist identifies potential pharmacotherapeutic prob- A minimum of ~160 (80 in the control group and 80 in the inter- lems considering the patient’s characteristics and experiences, vention group) will be sufficient to detect a medium effect size life expectancy and preferences using PIP tools such as the with a power of 80%, an α of 5% on the primary outcome and an STOPP and START criteria. [6] The pharmacist will draft recom- intraclass correlation coefficient up to 0.2. [25] To the best of our mendations for the GP. Different problems and/or recommen- knowledge, only one pilot randomised study has been conducted dations will be prioritised. Recommendations could include to that was aimed at decreasing the DBI. [26] We, therefore, cannot start or stop medication, change doses or carry out additional estimate an effect size ‘a priori’ as this should be based on multi- laboratory tests. ple independent studies. Since a small effect size will probably be clinically irrelevant and a large effect size may be unrealistic, we Step 3: multidisciplinary meeting chose a medium effect size. With the aim to include 160 patients The pharmacist discusses the patient’s medication profile with 4 and the expectation of a non-response rate of 60%, we will invite the GP during a face-to-face meeting. Together, they will draft a total of 400 participants. an action plan, including treatment objectives, potential actions and priority of actions (eg, withdrawing medication). Preferences Intervention of the patient, patient characteristics, experience, and life expec- The intervention will be a 3MR carried out by the pharmacist tancy will be central in the decision-making process. If needed, in close collaboration with the GP and if needed, medical spe- the appropriate medical specialists will be included in the medi- cialists. The medication reviews will be based on current Dutch cation review. guidelines with a focus on lowering the load of anticholinergic/ sedative medication following five steps as outlined below. [24] Step 4. pharmaceutical action plan Pharmacy students will be assisting the community pharmacists The pharmacist or GP discusses the action plan, made in step during some steps of the reviews. Participants in the control arm 3, with the patient. An agreement about the action plan will be will receive their medication review after the follow-up measure- made with the patient, preferences, expectations and concerns of ment. All participants have been informed about the possible de- the patient are key points in the decision-making process. Time lay of their medication review as part of the IC procedure. schedule for next intervention will be made and changes in med- ication treatment will be registered. Step 1: pharmacotherapeutic anamnesis The pharmacist collects information about the actual medication Step 5. follow-up use, problems with medication use and experiences, efficacy and Actions made in step 4 are evaluated at an agreed time interval possible side effects of the medication-in particular anticholin- with the pharmacist and/or GP. ergic and sedative medication-during a face-to-face consulta- tion with the patient. Furthermore, the use of ‘over the counter’

68 69 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

Study parameters • Anticholinergic side effects: as measured by the Udvalg for Primary parameter Kliniske Undersøgelser (UKU) side effect rating scale. [31] The key aim of the 3MR is to optimise a patient’s medication and • Sedative side effects derived from a patient-reported ad- Study parameters to lower the DBI, by reducing medications with anticholinergic verse drug event questionnaire. [32] Mainand sedative study parameter properties. The DBI will be measured for all partic- • Risk of falls: as measured by patient-reported fall inci- Theipants key at aimbaseline of the and 3MR follow-up is to optimise using electronic a patient’s pharmacy medication dis- and to lower the DBI, by reducing dents and the ‘Up & Go’ test. [33] medicationspensing records with corrected anticholinergic for actual and medication sedative properties. intake based The on DBI a will be measured for all participants • Cognitive function: as measured by the ‘Seven Minute atdouble baseline check and with follow the- uppatient using by electronic telephone. pharmacy We will dispensincalculate theg records corrected for actual medication Screen’, [34] the ‘Trailmaking Test A & B’ [35] and intakeDBI using based the on following a double formula: check with the patient by telephone. We will calculate the DBI using the the ‘Digit Symbol Coding Test’ of the ‘Wechsler Adult following formula: Intelligence Scale III’. [36] DBIDBI = • ADL: as measured by the ‘Groningen Activiteiten 𝐷𝐷 (D, daily dose of a drug; δ, minimum recommended daily dose as stated in Dutch standard reference Restrictie Schaal’. [37, 38] ∑ 𝐷𝐷+𝛿𝛿 sources(D, daily). dose[27] of a drug; δ, minimum recommended daily dose as • Quality of life: as measured by the EQ-5D-3L question- stated in Dutch standard reference sources). [27] naire. [39] All chronically used (≥ 3 months) medications (excluding dermatological (ATC D) and sensory • Hospital admission: assessed from the patient’s medical 4 medicationAll chronically (ATC used S)) (≥ having 3 months) anticholinergic medications properties (excluding (including derma- dry mouth, constipation and urine records. retention)tological (ATCor sedative D) and properties sensory medicationbased on standard (ATC S)) Dutch having reference anti- sources [27-30] will be included in • Mortality: assessed from the patient’s medical records. thecholinergic calculation. properties For each (including drug the dryvalue mouth, of the constipationDBI will range and from 0 to 1 depending on the δ. The cessationurine retention) of one anticholinergic or sedative properties or sedative based medication on standard would Dutch lower the DBI by about 0.5. We consider Covariates thereference cessation sources of one [27–30] drug willto be be clinically included relevantin the calculation. and therefore For, defined the primary outcome as the All demographic characteristics (sex, age, educational level, mar- differenceeach drug inthe proportion value of the of DBIpatients will havingrange from a decrease 0 to 1 dependingof DBI ≥ 0.5 from baseline to follow-up in the ital status) and number of medications at baseline and follow-up interventionon the δ. The group cessation and in of the one control anticholinergic group. It is or expected sedative thatmedi at- follow-up, the proportion of patients will be included in the analysis. withcation a decreasewould lower of the the DBI DBI ≥ by 0.5 about is significantly 0.5. We consider higher thein the cessa intervention- group in comparison to the Selection process, randomisation, intervention allocation and controltion of group.one drug to be clinically relevant and therefore, defined the primary outcome as the difference in proportion of patients blinding having a decrease of DBI ≥ 0.5 from baseline to follow-up in the A preliminary list of potentially eligible patients will be obtained Secondary parameters intervention group and in the control group. It is expected that at by electronic search in the electronic pharmacy dispensing re- Secondary study parameters are chosen with regard to patient outcomes. All questionnaires and tests follow-up, the proportion of patients with a decrease of the DBI ≥ cords based on a limited set of inclusion criteria (age, chronic will be administered to all participants at baseline and follow-up (see Study procedures section). 0.5 is significantly higher in the intervention group in compari- polypharmacy, use of psychotropic medication (ATC NO5/NO6)).  Anticholinergic side effects: as measured by the Udvalg for Kliniske Undersøgelser (UKU) side son to the control group. Notably, patients in the Netherlands are registered with one effect rating scale. [31] pharmacy, so the pharmacies keep relatively accurate dispensing  Sedative side effects derived from a patient-reported adverse drug event questionnaire. [32] Secondary parameters records of all prescribed medication. Inclusion/exclusion crite-  Secondary Risk of falls:study as parameters measured byare patient chosen-reported with fregardall incidents to pa and- the ‘Up & Go’ test. [33] ria will be checked by the researchers, pharmacists and GPs to tient Cognitive outcomes. function: All questionnaires as measured by and the tests‘Seven will Minute be admin Screen’,- [34] the ‘Trailmaking Test A & B’ obtain a list of eligible patients who will be approached for IC istered[35] to and all the participants ‘Digit Symbol at baseline Coding andTest’ follow-up of the ‘Wechsler (see Study Adult Intelligence Scale III’. [36] as outlined below. Within each pharmacy, all included patients procedures section). ADL: as measured by the ‘Groningen Activiteiten Restrictie Schaal’. [37, 38] will then be matched in pairs by gender, age, DBI and number of  Quality of life: as measured by the EQ-5D-3L questionnaire. [39]  Hospital admission: assessed from the patient’s medical records.  Mortality: assessed from the patient’s medical records. 70 71

Covariates All demographic characteristics (sex, age, educational level, marital status) and number of medications at baseline and follow-up will be included in the analysis.

Selection process, randomisation, intervention allocation and blinding A preliminary list of potentially eligible patients will be obtained by electronic search in the electronic pharmacy dispensing records based on a limited set of inclusion criteria (age, chronic polypharmacy,

50

Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

medications. Subsequently, this list of participants will be sent to conducted. Analyses will be done ‘per protocol’ and ‘inten- the principal investigator (KT) who is not involved in the recruit- tion-to-treat’. Percentages and frequencies will be calculated for ment and data collection. Within each pair, one participant will nominal variables, median values and IQRs, or frequencies will be randomly assigned to the intervention condition by the princi- be calculated for ordinal data or continuous data with a skewed pal investigator (coin flipping). The principal investigator will in- distribution. Means and SDs will be calculated for continuous form the pharmacists about the patient’s allocation. Pharmacists data that follow a normal distribution. Missing data will be kept and participants cannot be kept blind. All researchers involved in at a minimum by standardising and monitoring data collection. data collection will be kept blind to the allocation. Therefore this In case of missing data, sensitivity analyses will be conducted to is a single blinded study. This method will ensure that we have examine the influence of missing data on the study findings. All balanced groups within each pharmacy, random allocation, and statistical tests will be one sided. P-values ≤ 0.05 will be consid- concealment of allocation from the researchers involved in data ered significant. collection. Pharmacies will be enrolled continuously, and all par- ticipants of one pharmacy will be enrolled at the same time. This Primary study parameters excludes others methods such as stratified randomisation of all Generalised linear mixed models will be employed to account for participants at the same time. dependence of data (patients within pharmacy). Consequently, a 4 random intercept and a random slope at the level of pharmacies Quality of data will be entered into the linear mixed model. Furthermore, we will We will collect data in a standardised manner using data collec- adjust for significant covariates. tion sheets. All researchers will be trained by an experienced neu- ropsychologist. We will assess patients’ cognitive function using Secondary study parameters objective and validated neuropsychological tests (see Study pa- Secondary study parameters will be examined in a similar way. rameters section). Validated questionnaires will be used to assess Depending on whether they are continuous variables and their anticholinergic and sedative side effects, loss of ADL, and quality distribution is normally or Poisson distributed, we will employ of life (see Study parameters section). Medication data will be col- standard linear mixed models or Poisson linear mixed models. lected from the pharmacy information system and actual use will be verified by the patient. All data will be entered in a Microsoft Study procedures Access database by a research assistant. Baseline data will be col- The flowchart of Figure 1 provides a schematic overview of the lected before the intervention. Follow-up data will be collected study phases along with the participant flow at each study phase. 3 months after the intervention has taken place, assuming that within these 3 months the maximum effect of possible medica- tion changes made during the 3MR are reached. All data entries ETHICS AND DISSEMINATION will be double-checked against hardcopy source data. The ethics and dissemination are in line with a similar study. Statistical analysis [40] The study will be conducted according to the Declaration of All data will be analysed in IBM SPSS version 22. Descriptive Helsinki regarding the Ethical Principles for Medical Research statistics of the intervention group and control group will be Involving Human Subjects (amended by the 64th World Medical

72 73 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

rest with the treating GP. We, therefore, argue that our interven- 1. Assessing eligibility No. potential P-GP participants tion is one of usual care based on the latest evidence-based prin- Excluded (n = ) ciples and recommendations made in the guidelines by the Dutch 2. Asking IC No. society of General Practitioners. [24] The intervention is specifi- P approached Decline of cally aimed at older community-dwelling pharmacy patients in an participation (n = ) Written IC no. attempt to optimise prescribing and for this study, in particular, patients to reduce medication with anticholinergic and sedative proper- 3. Randomisation IR ties. [41] Data will be handled and stored. To ensure participants’ confidentiality, research data and participants’ personal data will Intervention Control arm arm (n = ) (n = ) be stored in two different files. Data records from both files will 4. Baseline measure No baseline data and CR-RA reasons why (n = ) be linked with an identification number that cannot be traced to Intervention Control arm arm (n = ) (n = ) the individual patient and their characteristics. The file with pa- Control arm lost to 5. Intervention follow-up data and tients’ personal data will be password protected and will be safe- P-GP reasons why (n = ) guarded by the investigators. To avoid scientific fraud or miscon-

Intervention arm lost duct, all investigators will have full access to the data. 6. Follow-up measure 4 to follow-up and CR-RA reasons why (n = ) Intervention Control arm arm (n = ) (n = ) Finally, study results will be published in peer-reviewed journals and in newsletters for pharmacists, news messages for the public, 7. Analysis CR and on websites for professionals. If possible, data will be pub- Intention to treat lished in open-access articles or as full text post prints in order to Intervention (n = ) Control (n = ) Excluded from make them available to the public. Duplicate publication will be analysis and reasons why (n = ) avoided. Per protocol Intervention (n = ) Control (n = ) In addition, this study has been registered at http://www.

Figure 1: Flowchart study phases and participants (CR, coordinating re- ClinicalTrials.gov (trial registration number: NCT02317666). searcher; GP, general practitioner; IC, informed consent; IR, investiga- tional researcher; P, pharmacist; RA, research assistant).

Association’s General Assembly, Fortaleza, Brazil, October 2013) and in accordance with Dutch medical-ethical legislation. The community pharmacist will ask her/his patients to participate. Prior to participation, written IC will be asked. The 3MR will be based on expert consensus and the medical literature. Moreover, the 3MR will result in high-quality treatment recommendations that will be attained by the pharmacist and treating GP working in close collaboration. Final treatment decisions, however, always

74 75 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — protocol of a RCT

REFERENCES 16. Hartikainen S, Rahkonen T, Kautiainen H, Sulkava R. Kuopio 75+ study: does ad- vanced age predict more common use of psychotropics among the elderly? Int 1. Taxis K, O’Sullivan D, Cullinan S, Byrne S. Drug utilization in older people. In: Else- Clin Psychopharmacol. 2003;18(3):163–167. viers M, Wettermark B, Almarsdóttir A, et al, editors. Drug utilization research: 17. Hilmer SN, Mager DE, Simonsick EM, et al. A drug burden index to define Methods and applications. London: Wiley-Blackwell; 2016. p. 259–269. the functional burden of medications in older people. Arch Intern Med. 2. Boyd CM, Darer J, Boult C, Fried LP, Boult L, Wu AW. Clinical practice guidelines 2007;167(8):781–787. and quality of care for older patients with multiple comorbid diseases: implica- 18. Ruxton K, Woodman RJ, Mangoni AA. Drugs with anticholinergic effects and cog- tions for pay for performance. JAMA. 2005;294(6):716–724. nitive impairment, falls and all-cause mortality in older adults: A systematic re- 3. Davies EA, O’Mahony MS. Adverse drug reactions in special populations — the el- view and meta-analysis. Br J Clin Pharmacol. 2015;80(2):209–220. derly. Br J Clin Pharmacol. 2015;80(4):796–807. 19. Kouladjian L, Gnjidic D, Chen TF, Mangoni AA, Hilmer SN. Drug Burden Index in 4. McLean AJ, Le Couteur DG. Aging biology and geriatric clinical pharmacology. Phar- older adults: theoretical and practical issues. Clin Interv Aging. 2014;9:1503–1515. macol Rev. 2004;56(2):163–184. 20. Vinks TH, Egberts TC, de Lange TM, de Koning FH. Pharmacist-based medication 5. Shi S, Morike K, Klotz U. The clinical implications of ageing for rational drug ther- review reduces potential drug-related problems in the elderly: the SMOG con- apy. Eur J Clin Pharmacol. 2008;64(2):183–199. trolled trial. Drugs Aging. 2009;26(2):123–133. 6. O’Mahony D, O’Sullivan D, Byrne S, O’Connor MN, Ryan C, Gallagher P. STOPP/ 21. Teichert M, Luijben SN, Wereldsma A, et al. Implementation of medication reviews START criteria for potentially inappropriate prescribing in older people: ver- in community pharmacies and their effect on potentially inappropriate drug use sion 2. Age Ageing. 2015;44(2):213–218. in elderly patients. Int J Clin Pharm. 2013;35(5):719–726. 7. American Geriatrics Society 2012 Beers Criteria Update Expert Panel. American Geri- 22. Hatah E, Braund R, Tordoff J, Duffull SB. A systematic review and meta-analy- 4 atrics Society updated Beers Criteria for potentially inappropriate medication sis of pharmacist-led fee-for-services medication review. Br J Clin Pharmacol. use in older adults. J Am Geriatr Soc. 2012;60(4):616–631. 2014;77(1):102–115. 8. Landi F, Russo A, Liperoti R, et al. Anticholinergic drugs and physical function 23. Blenkinsopp A, Bond C, Raynor DK. Medication reviews. Br J Clin Pharmacol. among frail elderly population. Clin Pharmacol Ther. 2007;81(2):235–241. 2012;74(4):573–580. 9. Mulsant BH, Pollock BG, Kirshner M, Shen C, Dodge H, Ganguli M. Serum anticho- 24. Nederlands Huisartsen Genootschap (NHG). Multidisciplinary guideline on poly- linergic activity in a community-based sample of older adults: relationship with pharmacy in older individuals. https://www.nhg.org/sites/default/files/content/ cognitive performance. Arch Gen Psychiatry. 2003;60(2):198–203. nhg_org/uploads/polyfarmacie_bij_ouderen.pdf. Accessed Jun 2015. 10. Cumming RG, Le Couteur DG. Benzodiazepines and risk of hip fractures in older 25. Cohen J. A power primer. Psychol Bull. 1992;112(1):155–159. people: a review of the evidence. CNS Drugs. 2003;17(11):825–837. 26. Gnjidic D, Le Couteur DG, Abernethy DR, Hilmer SN. A pilot randomized clinical 11. Huang AR, Mallet L, Rochefort CM, Eguale T, Buckeridge DL, Tamblyn R. Medica- trial utilizing the drug burden index to reduce exposure to anticholinergic and tion-related falls in the elderly: causative factors and preventive strategies. Drugs sedative medications in older people. Ann Pharmacother. 2010;44(11):1725–1732. Aging. 2012;29(5):359–376. 27. Farmacotherapeutisch Kompas. Dutch pharmacotherapeutic reference source. 12. Cancelli I, Gigli GL, Piani A, et al. Drugs with anticholinergic properties as a risk https://www.farmacotherapeutischkompas.nl/ (2018). Accessed Apr 2018. factor for cognitive impairment in elderly people: a population-based study. J 28. Expertisecentrum pharmacotherapie bij ouderen (EPHOR). Dutch reference source Clin Psychopharmacol. 2008;28(6):654–659. for pharmacotherapy in older people. http://www.ephor.nl/pdf/Ephors-rap- 13. Hiitola PK, Enlund H, Sulkava RO, Hartikainen SA. Changes in the use of cardiovas- porten (2018). Accessed Apr 2018. cular medicines in the elderly aged 75 years or older--a population-based Kuopio 29. KNMP Kennisbank: Dutch pharmacotherapeutic reference source. https://kennis- 75+ study. J Clin Pharm Ther. 2007;32(3):253–259. bank.knmp.nl/ (2018). Accessed Apr 2018. 14. Flory JH, Ky B, Haynes K, et al. Observational cohort study of the safety of digoxin 30. Duran CE, Azermai M, Vander Stichele RH. Systematic review of anticholinergic use in women with heart failure. BMJ Open. 2012;2(2):e000888. risk scales in older adults. Eur J Clin Pharmacol. 2013;69(7):1485–1496. 15. Wastesson JW, Parker MG, Fastbom J, Thorslund M, Johnell K. Drug use in cente- 31. Lingjaerde O, Ahlfors UG, Bech P, Dencker SJ, Elgen K. The UKU side effect rating narians compared with nonagenarians and octogenarians in Sweden: a nation- scale. A new comprehensive rating scale for psychotropic drugs and a cross-sec- wide register-based study. Age Ageing. 2012;41(2):218–224. tional study of side effects in neuroleptic-treated patients. Acta Psychiatr Scand Suppl. 1987;334:1–100.

76 77 Evaluating a current deprescribing intervention

32. de Vries ST, Haaijer-Ruskamp FM, de Zeeuw D, Denig P. Construct and concurrent validity of a patient-reported adverse drug event questionnaire: a cross-sectional study. Health Qual Life Outcomes. 2014;12:103. 33. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39(2):142–148. 34. Solomon PR, Hirschoff A, Kelly B, et al. A 7 minute neurocognitive screening bat- tery highly sensitive to Alzheimer’s disease. Arch Neurol. 1998;55(3):349–355. 35. Reitan RM. The relation of the trail making test to organic brain damage. J Consult Psychol. 1955;19(5):393–394. 36. Wechsler D. Wechsler Adult Intelligence Scale. Third ed. San Antonio: The Psycho- logical Corporation; 1997. 37. Kempen GI, Suurmeijer TP. The development of a hierarchical polychotomous ADL- IADL scale for noninstitutionalized elders. Gerontologist. 1990;30(4):497–502. 38. Kempen GI, Miedema I, Ormel J, Molenaar W. The assessment of disability with the Groningen Activity Restriction Scale. Conceptual framework and psycho- metric properties. Soc Sci Med. 1996;43(11):1601–1610. 39. EuroQol. EQ-5D-3L questionnaire. www.euroqol.org. Accessed Aug 2014. 40. Wouters H, Quik EH, Boersma F, et al. Discontinuing Inappropriate Medication in Nursing Home Residents (DIM-NHR Study): protocol of a cluster randomised controlled trial. BMJ Open. 2014;4(10):e00 41. Pont LG, Nielen JT, McLachlan AJ, et al. Measuring anticholinergic drug exposure in older community-dwelling Australian men: a comparison of four different measures. Br J Clin Pharmacol. 2015;80(5):1169–1175.

78 Deprescribing in older people General introduction and thesis outline

1 CHAPTER 5

REDUCING THE ANTICHOLINERGIC AND SEDATIVE LOAD IN OLDER PATIENTS ON POLYPHARMACY BY PHARMACIST-LED MEDICATION REVIEW: A RANDOMISED CONTROLLED TRIAL

Helene G van der Meer, Hans Wouters, Lisa G Pont, Katja Taxis

BMJ Open. 2018;8(7):e019042

80 81 Reducing the anticholinergic/sedative load by medication review — a RCT

ABSTRACT Conclusions Pharmacist-led medication review as currently performed in the Netherlands was not effective in reducing the Objective To evaluate if a pharmacist-led medication review is ef- anticholinergic/sedative load, measured with the DBI, within the fective at reducing the anticholinergic/sedative load, as measured time frame of 3 months. Preventive strategies, signalling a rising by the Drug Burden Index (DBI). load and taking action before chronic use of anticholinergic/seda- tive medication is established, may be more successful. Design Randomised controlled single blind trial. Trial registration Clinical trials NCT02317666. Setting 15 community pharmacies in the Northern Netherlands.

Participants 157 community-dwelling patients aged ≥ 65 years who used ≥ 5 medicines for ≥ 3 months, including at least one psycholeptic/psychoanaleptic medication, and who had a DBI ≥ 1.

Intervention A medication review by the community pharma- cist in collaboration with the patient’s general practitioner and patient.

Primary and secondary outcome measures The primary out- 5 come was the proportion of patients whose DBI decreased by at least 0.5. Secondary outcomes were the presence of anticholin- ergic/sedative side effects, falls, cognitive function, activities of daily living, quality of life, hospital admission, and mortality. Data were collected at baseline and 3 months follow-up.

Results Mean participant age was 75.7 (SD: 6.9) years in the in- tervention arm and 76.6 (SD: 6.7) years in the control arm, the majority were female (respectively 69.3% and 72.0%). Logistic regression analysis showed no difference in the proportion of patients with a ≥ 0.5 decrease in DBI between intervention arm (17.3%) and control arm (15.9%), (OR 1.04, 95% CI 0.47 to 2.64, p = 0.927). Intervention patients scored higher on the digit symbol substitution test, measure of cognitive function, (OR 2.02, 95% CI 1.11 to 3.67, p = 0.021), and reported fewer sedative side effects (OR 0.61, 95% CI 0.40 to 0.94, p = 0.024) at follow-up. No significant difference was found for other secondary outcomes.

83 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

BACKGROUND health benefits on the patient. [19] The most effective method for medication review remains unknown. Focusing on specific Older people suffer from many medical conditions and use more subgroups such as older people with multiple comorbidities and medication than any other age group. Multiple medication use in polypharmacy, [20] or patients suffering from pain [21] may be combination with age-related physiological changes increase the one strategy to optimise medication review associated benefits. risk of medication related harm including adverse drug events, To date, there is no consensus on the effectiveness of medication drug-drug- and drug-disease-interactions. [1] Medications with review as a strategy to reduce anticholinergic and sedative load anticholinergic and/or sedative properties are of particular con- as measured by the DBI. Therefore, the primary aim of this study cern in older people, because they worsen cognitive impairment was to evaluate if a medication review is an effective strategy to and physical functioning, increase the risk of falls and negatively reduce anticholinergic and sedative load as measured by the DBI. impact activities of daily living, hospitalisation, and mortality. [2, Secondarily, we evaluated the effect of a medication review on pa- 3] Despite the risks, these medications are commonly prescribed tient outcomes including cognitive function, risk of falls, activi- to older individuals. [4, 5] Different measures have been devel- ties of daily living and quality of life. oped to quantify the anticholinergic load in patients. [6] The Drug Burden Index (DBI) determines an individual’s exposure to anticholinergic and sedative medication taking into account the METHODS dose. [7, 8] A high DBI has been associated with impairments in both physical- and cognitive function among older individuals. Study design, setting & participants [9, 10] Hence, decreasing exposure to anticholinergic and sed- We conducted a randomised controlled, single blind trial in 15 5 ative medication, as measured by the DBI, may have important community pharmacies from December 2014 until October 2015 health benefits in older people. in the Northern Netherlands. Pharmacies were recruited via the regional association of pharmacists and participation was volun- Two small Australian studies suggest that medication review could tary. One pharmacist per pharmacy was involved in the study. In be a promising strategy in reducing the DBI in community-dwell- Dutch community pharmacy practice, all registered pharmacists ing older people. [11, 12] Medication review is ‘a structured critical are allowed to perform medication reviews. Furthermore, phar- examination of a person’s medicines with the objective of reach- macists collaborate with GPs in their area. This includes local ing an agreement with the person about treatment, optimising regular meetings of pharmacists and GPs in pharmacotherapy the impact of medicines, minimising the number of medication- counselling groups. [22] In the Netherlands, each individual is related problems and reducing waste’. [13] While meta-analyses registered with a single pharmacy. [23] Pharmacies hold a com- of studies in different settings show a lack of effectiveness on plete electronic medication history for each patient registered outcomes such as mortality or hospital (re-) admissions, [14–16] with them. When undertaking a medication review it is routine these studies included different types of medication review. Well- practice of pharmacists to obtain an extensive summary of the structured medication review with good cooperation between electronic patients’ medical records, including latest recorded ep- pharmacist and general practitioner (GP) and involvement of the isodes and lab-values, from the GP. [24] At the time of the study, patient were most likely to be successful. [17, 18] Furthermore fee- all Dutch community pharmacists were required to perform for-pharmacist-led medication review seemed to have positive medication reviews in cooperation with the GP for high-risk

84 85 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

patients according to the guidelines. [25] Patients who were aged pharmacotherapeutic problems taking into account the patient’s ≥ 65 years, living independently, using ≥ 5 medications for ≥ 3 medical records, including latest recorded episodes and lab-val- months, including at least one psycholeptic or psychoanaleptic ues. Accordingly, the pharmacist drafted written recommenda- medication (Anatomic Therapeutic Classification (ATC) code N05 tions for medication optimisation to discuss with the patients’ GP. or N06), [26] and with a DBI ≥ 1 were identified by the pharma- Third, a multidisciplinary meeting, between pharmacist and GP cist and invited to participate in the study. Exclusion criteria were was held. At this meeting, the potential medication problems of limited life expectancy (< 3 months), non-Dutch language speaker the patient were discussed and draft of a pharmacotherapeutic ac- or advanced dementia. Patients who had received a medication tion plan was decided. Fourth is a discussion of the draft pharma- review within the past 9 months before the study period and cotherapeutic action plan between patient and pharmacist and/ patients who needed a medication review urgently were also ex- or GP. The patients’ expectations and wishes were key elements cluded. Exclusion criteria were identified by the pharmacist with in the decision-making process and were included in the final whom the patient was registered. [27] action plan. Fifth, a follow-up of the final pharmacotherapeutic action plan was undertaken. Further detail of the medication Randomisation, allocation & blinding review process and the Dutch guideline underpinning the study Eligible patients were approached by the pharmacist and asked can be found in our previously published study protocol. [27] The to provide written informed consent. In each pharmacy, patients pharmacists participating in the study all undertook regular med- willing to participate were then matched in pairs by gender, age, ication reviews as part of their practice and as such were familiar DBI and number of medications. One patient of each pair was with the guideline. Nonetheless, we provided the guidelines to randomly assigned to the intervention condition. All participants the pharmacists with the request to focus on anticholinergic and 5 gave written consent prior to the intervention allocation. The sedative medications. No additional educational material on an- randomisation process was conducted by the principal investiga- ticholinergic and sedative medication was provided. In order to tor, who was not involved in recruitment or data collection. The get a reflection of ‘real world’ practice, we let the pharmacists per- researchers who enrolled the patients and collected the data were form the medication reviews according to their routine practice, kept blind to the allocation. Pharmacists and patients could not but we did check whether all five steps were conducted. The med- be kept blind, but were explicitly asked not to reveal study alloca- ication review took place within days after the baseline measure- tion for individual patients to the researchers who collected the ment for the intervention patients. In the control arm, patients data. Therefore, this was a single blind study. received the medication review after the study period.

Intervention Outcomes The intervention was a medication review conducted by the The primary outcome was defined as the difference in proportion community pharmacist in close collaboration with the patients’ of patients having a decrease of DBI ≥ 0.5 at 3-month follow-up. GP and, if needed, other medical specialists. In the Netherlands We chose a 3-month follow-up because this was a reasonable time medication review consisted of five steps. [25] Step one was a frame to detect medication changes by the medication review. A face-to-face consultation between the pharmacist and patient longer follow-up would have increased the chance of medication to discuss medication use. Second, the pharmacist undertook a changes due to other reasons, such as changes in disease status. pharmacotherapeutic medication review, identified potential Our hypothesis was that the proportion with a 0.5 decrease in

86 87 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT Study parameters Main study parameter TheDBI keywould aim be ofhigher the 3MRin the is intervention to optimise arm a patient’s compared medication with the and to lower the DBI, by reducing admission, and mortality was assessed based on patient/relative medicationscontrol arm. withWe chose anticholinergic 0.5, as this andequals sedative the cessation properties. of one The drug, DBI will be measured for all participants reporting. atwhich baseline we consideredand follow -aup clinically using electronic relevant pharmacydecrease. Thedispensin DBI wasg records corrected for actual medication intakecalculated based using on thea double following check formula with the[8]: patient by telephone. We will calculate the DBI using the Sample size calculation following formula: To the best of our knowledge, only one randomised pilot study DBIDBI = has been conducted assessing the DBI. [12] We therefore could 𝐷𝐷 (D, daily dose of a drug; δ, minimum recommended daily dose as stated in Dutch standard reference not calculate the sample size ‘a priori’. However, we estimated a ∑ 𝐷𝐷+𝛿𝛿 sourcesD = daily). [27] dose, δ = minimum recommended daily dose were sample size based on a power of 80% at a significance of 0.05 and derived for the study from Dutch standard reference sources. an intraclass correlation coefficient up to 0.2 to detect a medium All[28, chronically29] Except for used sensory (≥ 3 months)and dermatological medications preparations, (excluding alldermatological (ATC D) and sensory effect size on the primary outcome. [40] We chose a medium ef- medicationchronic medications (ATC S)) (i.e.having those anticholinergic used for ≥ 3 propertiesmonths) with (including anti- dry mouth, constipation and urine fect size as we considered a small effect size to be not clinically retention)cholinergic or properties sedative properties (dry mouth, based constipation on standard and Dutch urine reference reten- sources [27-30] will be included in relevant and a power to detect a medium effect size also to be thetion) calculation. and sedative For properties each drug based the value on Dutch of the standard DBI will reference range from 0 to 1 depending on the δ. The capable of detecting a large effect size. For this calculation around cessationsources [28–30] of one wereanticholinergic included in or the sedative calculation. medication Medication would data lower the DBI by about 0.5. We consider 160 participants (80 in control arm and 80 in intervention arm) thewere cessation derived fromof one electronic drug to bepharmacy clinically dispensing relevant dataand thereforeand were , defined the primary outcome as the were needed. We expected a non-response rate of 60% and there- differenceverified with in proportionthe patient. of patients having a decrease of DBI ≥ 0.5 from baseline to follow-up in the fore aimed to invite 400 patients to participate in the study. intervention group and in the control group. It is expected that at follow-up, the proportion of patients Statistical analysis withWe included a decrease the of following the DBI ≥secondary 0.5 is significantly outcomes: higheranticholinergic in the intervention group in comparison to the controlside effects, group. measured by the Udvalg for Kliniske Undersøgelser We performed two analyses. In the first analysis we included all 5 side effect rating scale, [31] sedative side effects, derived from a pa- patients with a baseline measurement. In the second analysis, we tient-reported adverse drug event questionnaire, [32] and risk of included all patients who were not lost to follow-up, and who Secondary parameters falls, determined by the Up & Go test. [33] Cognitive function was received the intervention as allocated. Descriptive statistics were Secondary study parameters are chosen with regard to patient outcomes. All questionnaires and tests measured using validated tests for memory and executive func- calculated for both allocation arms at baseline. For the analysis will be administered to all participants at baseline and follow-up (see Study procedures section). tion, namely the Seven Minute Screen (7MS), [34] the Trailmaking of the primary outcome, we initially considered a generalised  Anticholinergic side effects: as measured by the Udvalg for Kliniske Undersøgelser (UKU) side Test A & B, [35] and Digit Symbol Substitution Test (DSST). [36] linear mixed effects model to adjust for dependence of observa- effect rating scale. [31] The latter has also previously been used to examine the validity of tions (ie, clustering of patients within pharmacies). However, as  Sedative side effects derived from a patient-reported adverse drug event questionnaire. [32] the DBI. [8] Activities of daily living were derived using the val- the intraclass correlation was not significant and no significant  idated Risk Groningen of falls: as Activity measured Restriction by patient Scale-reported (GARS), fall [37, incidents 38] and and the ‘Up & Go’ test. [33] clustering was observed, extension of the model with random quality Cognitive of life wasfunction: measured as measured by the Euroqol-5 by the ‘Seven Dimension-3 Minute Screen’,Level [34] the ‘Trailmaking Test A & B’ effects at the level of pharmacies was not necessary. Therefore, questionnaire,[35] and the including ‘Digit Symbol visual analogueCoding Test’ scale. of [39] the All ‘Wechsler tools were Adult Intelligence Scale III’. [36] only fixed effects were considered and standard fixed effects lo- administered ADL: as measured in Dutch byand the data ‘Groningen were collected Activiteiten in a standardised Restrictie Schaal’. [37, 38] gistic regression model were applied. Most secondary outcomes manner, Quality using of life:data as collection measured sheets,by the EQby -researchers5D-3L questionnaire. who were [39] were examined with standard regression models. Variables with trained Hospital by a admission:psychologist. assessed Data collection from the patient’stook place medical at baseline records. a skewed distribution were transformed before analysis. For di- and Mortality: 3-month follow-up assessed fromfor both the patient’sallocations. medical Patients records. with the in- chotomous variables we reported percentages and numbers of ability to walk were excluded from the Up&Go test and the GARS patients in the best scoring group, for skewed variables we report Covariatesquestionnaire. At follow-up the number of fall incidents, hospital the median and IQR and for normally distributed data we report All demographic characteristics (sex, age, educational level, marital status) and number of medications at baseline and follow-up will be included in the analysis. 88 89 Selection process, randomisation, intervention allocation and blinding A preliminary list of potentially eligible patients will be obtained by electronic search in the electronic pharmacy dispensing records based on a limited set of inclusion criteria (age, chronic polypharmacy,

50

Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

the mean and SD. Further detail on the analysis of secondary out- Missing data in cognitive tests due to inability of the patient to come tests and — questionnaires data can be found in Appendix complete the task were replaced with the worst score for that Table 1. Reported falls, hospitalisation and mortality were only specific group. Missing data of patients who could not be tested assessed from patients with a follow-up measurement. These vari- at follow-up within the study period, or who had missing data ables were dichotomized, reported as number and percentages of for other reasons were replaced by multiple imputation (five patients and analysed using Fisher’s exact test. A sensitivity analy- times) in SPSS 24. In this paper we report on the imputed data- sis was conducted on outliers (Appendix Table 2) and all analyses set. Sensitivity analysis showed no difference between the dataset were adjusted for gender, age, and number of medication at base- with and without missing data. line. Secondary outcomes were also adjusted for baseline scores. Analyses were done in SPSS 24 and MLwiN 2.36, and statistical Patient and Public involvement tests were two-sided and conducted at the 5% significance level. Patients and or public were not involved in the design or con- duct of the study. After the study period all participants received a Missing data thank you letter including a brief summary of the overall results. Few data were missing for the primary outcome. Of the two pa- tients, who were lost to follow-up, the baseline observation for medication use was carried forward to follow-up. For eight pa- RESULTS tients, medication use could not be verified with the patient, as they could not be reached by telephone despite several attempts. Participant flow For these patients, the medication data from the pharmacy dis- Overall, 498 patients were approached for participation, 164 pa- 5 pensing system were used. For secondary outcomes, 5.3% of data tients provided informed consent (32.9% response rate), and were missing in the complete dataset, mostly at follow-up (4.8%). 157 patients completed at least the baseline measurement and In the intervention arm, 7.0% of data was missing (6.1% at fol- were included in the first analysis (Figure 1). The drop-out rate low-up) across 18 patients, whereas in the control arm 3.7% was was 4.3%. missing (3.4% at follow-up) across 12 patients. In total 30 patients had missing data, of whom two were lost to follow-up. Eight Participant characteristics patients were not able to complete one or more cognitive tests The average participant age was 75.7 (SD: 6.9) years in the inter- (0.5% of all data). Eleven patients could not be tested at follow-up vention arm and 76.6 (SD: 6.7) years in the control arm, and the within the study period, six patients due to sickness, four patients majority were female (respectively 69.3% and 72.0%). Participants due to practical reasons (despite numerous attempts we were un- in the control arm used slightly more medicines at baseline successful to arrange an appointment for the follow-up measure- (9.3 (SD: 3.2) to 8.4 (SD: 2.4)), and more control patients were liv- ment), and one patient had died two days before the follow-up ing with a partner (53.6% to 44.0%) (Table 1). appointment. A few data were missing for other reasons across nine patients, for example patients forgetting their glasses, due to Primary outcome time constraints, or other reasons. In the first analysis, which included all patients with a base- line measurement, the proportion of patients with a decrease of DBI ≥ 0.5 did not differ between patients in intervention arm

90 91 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

Screening for eligibility Table 1: Demographic characteristics at baseline.

Potential participants (n = 567) Intervention Control (n = 75) (n = 82) Excluded (n = 403) Age (years) mean (SD) 75.7 (6.9) 76.6 (6.7) Not eligible : 69 Declined to participate: 334 Sex (female) (n (%)) 52 (69.3) 59 (72.0)

Number of medicines (mean (SD) 8.4 (2.4) 9.3 (3.2) Participants included in the study (n = 164) DBI (mean (SD) 3.1 (1.0) 3.2 (1.0) Marital status (n (%)) Randomisation of participants to allocation Partner 33 (44.0) 44 (53.6) Widow/widower/Divorced/single 34 (45.3) 32 (39.0) Intervention arm (n = 80) Control arm (n = 84) Unknown 8 (10.6) 6 (7.3) Withdrawn before baseline: 5 Withdrawn before baseline: 2 Level of education (n (%)) Loss to follow-up or nonadherence to allocation No/ low/ middle 58 (77.3) 64 (78.0) High 9 (12.0) 13 (15.8)

Intervention arm (n = 10) Control arm (n = 2) Unknown 8 (10.6) 5 (6.0) Did not receive intervention: 10 Died: 1 Moved to another pharmacy: 1 Medication use at baseline (top 5 (n (%))) ATC code nervous system 75 (100) 82 (100) ATC code cardiovascular 70 (93.3) 74 (90.2) Data analysis ATC code alimentary tract 64 (85.3) 71 (86.6) ATC code blood/ blood forming organs 49 (65.3) 46 (56.1) Intervention arm Control arm ATC code respiratory tract 20 (26.7) 38 (46.3) First analysis* (n = 75) First analysis* (n = 82) Second analysis† (n = 65) Second analysis† (n = 80) ATC = Anatomical Therapeutical Chemical.

Figure 1: Participant flow 5 *All patients who had a baseline measurement. Table 2: Proportion of patients having a decrease in DBI ≥ 0.5 by analysis †All patients who were not lost to follow-up and received the intervention type as allocated. Analysis type Proportion with decrease Odds ratio p-value of DBI ≥ 0.5 (%, n) (95% CI) * Intervention Control and control arm (17.3% to 15.9%, OR 1.04, 95% CI 0.47 to 2.64, p = 0.927). Similar results were obtained in the second analysis, First analysis (n = 157) 17.3 (13) 15.9 (13) 1.04 (0.47 to 2.64) 0.927 Second analysis (n = 145) 18.5 (12) 16.3 (13) 1.09 (0.45 to 2.63) 0.857 which included all patients who were not lost to follow-up, and who received the intervention as allocated (Table 2). Descriptive * Binary logistic regression, adjusted for age, gender, number of medication at baseline. First analysis: all patients with a baseline measurement. Second analysis: all patients analysis showed medication changes (starting, stopping, dos- who were not lost to follow-up, and who received the intervention as allocated age change) of DBI medications on ATC code level 1 in 53.8% of patients from intervention arm and in 45.0% of patients from Secondary outcomes control arm. For cardiovascular DBI medications, dose increases Secondary outcome tests and questionnaires were analysed in- and dose decreases of different medications occurred in 10.8% pa- cluding all patients who were not lost to follow-up and who re- tients from intervention arm compared to 1.3% of patients from ceived the intervention as allocated (Table 3). A difference was control arm (Appendix Table 3). seen in the DSST and reporting of sedative side effects between allocation arms. Patients in the intervention arm scored higher at follow-up on average (3 (SD: 1) to 1 (SD: 0) point (s), OR 2.02, 95%

92 93 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

CI 1.11 to 3.67, p = 0.021) and reported less sedative side effects at Table 3: Secondary outcome tests and — questionnaires at follow-up follow-up compared to the control arm (−1 (IQR: −2) to 1 (IQR: 0) point(s), OR 0.61, 95% CI 0.40 to 0.94, p = 0.024). For all other Outcome Intervention (n = 65) Control (n = 80) Treatment dif- ference at FU secondary outcomes no difference was found between interven- BL score Δ with BL score Δ with (95% CI) FU FU tion arm and control arm. Trailmaking Test A, 59.0 −8.4 (−4.8) 61.0 (27.8) −6.0 (1.6) −0.01 median (IQR) (36.9) (−0.11–0.09)†

Reported falls and hospitalisation could be assessed from 136 pa- Trailmaking Test B, 149.0 −3.9 (24.1) 152.0 1.0 (19.0) −0.01 tients who were included in the second analysis. No significant median (IQR) (103.0) (103.0) (−0.14–0.11)† difference was found in reported falls between control arm and DSST, mean (SD) 36.4 (12.2) 2.6 (1.2) 36.4 (13.2) 1.0 (−0.3) 0.70 † intervention arm, respectively 15 patients (19.5%) versus 18 pa- (0.11–1.30) * tients (30.5%), (p = 0.100). There was also no difference found 7MS enhanced 85 (55) 0 (0) 84 (71) 5 (4) 0.54 cued recall, % (n) (0.15–1.90) ‡ between control arm and intervention arm in hospitalisation, best scoring 9 (11.7%) versus 3 (5.1%) patients reported unplanned hospital 7MS Benton tem- 95 (62) −3 (−2) 99 (79) −4 (−3) 1.38 admission, (p = 0.149). Of all patients who were included in the poral orientation, (0.28–6.88) ‡ % (n) best scoring study, 2 died, one (1.2%) in control arm to one (1.3%) intervention arm, (p = 0.732). 7MS clock drawing, 80 (52) −8 (−5) 86 (69) −6 (−5) 0.67 % (n) best scoring (0.28–1.62) ‡

7MS category 16.1 (5.5) 0.1 (−0.6) 15.9 (5.0) 0.4 (−0.3) −0.18 fluency, mean (SD) (−1.55–1.20)† DISCUSSION 5 GARS, % (n) best 72 (46) 2 (−1) 69 (54) 0 (0) 1.73 scoring (0.62–4.84) ‡⁰

In our study, pharmacist-led medication review did not reduce Sedative side 3.0 (5.0) −1.0 (−2.0) 2.0 (4.0) 1 (0) 0.61 the anticholinergic and/or sedative medication load in older peo- effects, median (0.40–0.94) §* (IQR) ple within the first 3 months following review. In addition, medi- UKU, median 17.0 (22.0) −3.0 (1.0) 18.0 (27.0) −1.6 (−2.4) 0.97 cation review did not improve cognitive function, apart from the (IQR) (0.67–1.39) § DSST. We also found that medication review had no effect on an- EQ-5D-3L, % (n) 74 (48) 9 (6) 76 (61) 4 (3) 1.43 ticholinergic side effects, quality of life, activities of daily living, best scoring (0.51–4.03)‡ risk of falls, hospitalisation and mortality. However, intervention VAS, mean (SD) 6.6 (1.6) −0.2 (0.0) 6.8 (1.4) −0.1 (0.1) −0.09 patients reported fewer sedative side effects. (−0.50–0.32)†

Up&Go, % (n) best 66 (42) 0 (0) 64 (50) 4 (3) 1.37 Strengths and limitations scoring (0.60–3.14) ‡⁰

This randomised controlled trial was the first to focus on chang- BL = Baseline, FU = Follow-up, DSST = Digit Symbol Substitution Test; 7MS = Seven ing anticholinergic and sedative medication load by medication Minute Screen; GARS = Groningen Activities Restriction Scale; UKU = Udvalg for Kliniske Undersøgelser (measuring anticholinergic side effects); VAS = visual analogue review. The trial was completed successfully, allocation arms scale (part of EQ-5D-3L). †Linear regression analysis (reporting unstandardised b), ‡lo- gistic regression analysis (reporting odds ratio), §negative binomial regression analysis were comparable and we achieved a medium response rate. We (reporting incident rate ratio) used, all adjusted for age, gender, number of medication at baseline. *Statistically significant difference (p < 0.05). ⁰Deviation of number of also believe our study was appropriately powered to detect a clin- patients: n = 64 for intervention, n = 78 for control, 3 patients were excluded from ically relevant medium difference between intervention arm and this test/questionnaire.

94 95 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

control arm. Yet there are some methodological limitations that may require more time, for example withdrawing of medication should be considered when interpreting our findings. Firstly, our by step-wise reduction of dosing. However, there did not seem to study design might have introduced a risk of contamination be- be a difference in dosage changes between intervention arm and tween intervention arm and control arm, as pharmacists and GPs control arm. Finally, one third of all eligible patients were will- could have been triggered to optimise medication use also for pa- ing to participate in the study. Given the frailty of this population tients in the control arm during the study period. We know from and the time consuming nature of participation, we think this is the pharmacists that no structured medication reviews were per- a very reasonable response rate. Nevertheless, our results may not formed for control patients during the study period. Therefore we be generalisable to the total population. believe that changes we observed in control patients were due to usual care. Cluster randomisation may have prevented the chance Comparison with other studies of contamination, but this method has other disadvantages. [41] The medication changes in both arms were comparable. Small Secondly, although we did check whether all steps of the medica- changes in different therapeutic medication groups suggest fluc- tion review were conducted, it was outside the scope of our study tuations of medication use over time as prescribing is a dynam- to investigate to what extent pharmacists adhered to methods ic-rather than a static process. We do not know the pattern of recommended by the guideline on performing the medication fluctuations in anticholinergic and sedative medication prescrib- review. Informal conversations with pharmacists suggested that ing; this should be explored in longitudinal studies powered to although the guidelines recommend a face-to-face meeting be- detect changes at medication level. Our results are in line with a tween the pharmacist and GP, some pharmacists contacted the number of meta-analyses, which also reported a lack of effect of GP by phone, fax, or email due to lack of time. This might have medication reviews on a variety of patient outcomes. [14–16] Our 5 had an effect on the implementation of medication suggestions. results are in contrast to a number of studies, which found med- [18] Furthermore, while as part of the established collaboration ication reviews to be effective in specific subgroups of patients between pharmacists and GPs in Dutch primary care, Dutch with multiple comorbidities, polypharmacy and pain. [20, 21] The pharmacists routinely request an extensive summary of the elec- medication reviews in these studies, however, were not specifi- tronic patient’s medical records from the GP to perform a medi- cally focusing on medication with anticholinergic and/or sedative cation review, it is possible that some pharmacists did not do this. properties as we did. Two small Australian studies suggest that the We performed a pragmatic trial and therefore our results reflect DBI can be lowered, but these studies were based on pharmacist ‘real-world’ practice of how medication reviews were carried out recommendations and did not investigate actual implementation in Dutch health care practice at the time of the study. Thirdly, we of these by the GP. [11, 12] Although some lowering of the DBI was followed patients for 3 months after the intervention. Possibly, seen, the latter study did find that GPs had difficulties in chang- more time may have been necessary to determine the effect of the ing medications, for example with those medications initiated by intervention. We were not able to collect data about timing of the specialists. A recent study also showed that while it was possible medication review steps, so in some cases there may have been to optimise use for a number of medication classes, psychotropic delay in performing all steps. But in Dutch primary care, pharma- medications were among the most difficult to adjust. [42] So, de- cists and GP’s have an established close collaboration and there- spite guidance how to reduce anticholinergic and sedative medi- fore we believe that long delays were unlikely. Another argument cation, [43–45] as highlighted by our findings, there seem to be for a longer follow-up could be that changes in medication use important barriers preventing reduction in clinical practice.

96 97 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

CONCLUSIONS AND IMPLICATIONS REFERENCES

1. Taxis K, O’Sullivan D, Cullinan S, Byrne S. Drug utilization in older people. In: Else- Using the DBI, a highly vulnerable population group in need viers M, Wettermark B, Almarsdóttir A, et al, editors. Drug utilization research: of medication optimisation can be identified. Pharmacist-led Methods and applications. London: Wiley-Blackwell; 2016. p. 259–269. medication review as currently performed in the Netherlands 2. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- did not appear effective in reducing the DBI. While our study erties on cognitive function, delirium, physical function and mortality: a sys- tematic review. Age Ageing. 2014;43(5):604–615. was powered to detect a difference in medication use, it should 3. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls be acknowledged that other patient outcomes, like geriatric in older people: systematic review of publications from a recent 5-year period. syndromes (eg, risk of falls) and adverse events (eg, drug-re- Eur J Clin Pharmacol. 2015;71(12):1429–1440. lated hospital admission) are very important for the evaluation 4. Holvast F, van Hattem BA, Sinnige J, et al. Late-life depression and the associa- of medication review in older patients. Further studies should tion with multimorbidity and polypharmacy: a cross-sectional study. Fam Pract. 2017;34(5):539–545. ensure sufficient sample sizes to study these outcomes. [46, 47] 5. Bell JS, Mezrani C, Blacker N, et al. Anticholinergic and sedative medicines — Despite some practical issues with the DBI, such as the lack of prescribing considerations for people with dementia. Aust Fam Physician. an international consensus-based list of anticholinergic/sedative 2012;41(1–2):45–49. medication including minimum doses, [10] we suggest to use the 6. Pont LG, Nielen JT, McLachlan AJ, et al. Measuring anticholinergic drug exposure in DBI as a tool to identify harmful medication users. This depre- older community-dwelling Australian men: a comparison of four different mea- sures. Br J Clin Pharmacol. 2015;80(5):1169–1175. scribing approach may be suitable for other patient groups and in 7. Cardwell K, Hughes CM, Ryan C. The Association Between Anticholinergic Medica- other settings such as nursing homes or GP practice with co-lo- tion Burden and Health Related Outcomes in the ‘Oldest Old’: A Systematic Re- cated pharmacist. [4, 48–50] Enlarging the multidisciplinary view of the Literature. Drugs Aging. 2015;32(10):835–848. 5 team should also be considered, for example psychiatrists advis- 8. Hilmer SN, Mager DE, Simonsick EM, et al. A drug burden index to define ing GPs on lowering or ceasing medication and psychologists as- the functional burden of medications in older people. Arch Intern Med. 2007;167(8):78–787. sisting patients during withdrawal. Furthermore, signalling a ris- 9. Ruxton K, Woodman RJ, Mangoni AA. Drugs with anticholinergic effects and cog- ing load and taking action before chronic use of medication with nitive impairment, falls and all-cause mortality in older adults: A systematic re- anticholinergic and/or sedative properties is established may be view and meta-analysis. Br J Clin Pharmacol. 2015;80(2):209–220. the preferred approach. 10. Wouters H, van der Meer H, Taxis K. Quantification of anticholinergic and sedative drug load with the Drug Burden Index: a review of outcomes and methodologi- cal quality of studies. Eur J Clin Pharmacol. 2017;73(3):257–266. 11. Castelino RL, Hilmer SN, Bajorek BV, Nishtala P, Chen TF. Drug Burden Index and potentially inappropriate medications in community-dwelling older people: the impact of Home Medicines Review. Drugs Aging. 2010;27(2):135–148. 12. Gnjidic D, Le Couteur DG, Abernethy DR, Hilmer SN. A pilot randomized clinical trial utilizing the drug burden index to reduce exposure to anticholinergic and sedative medications in older people. Ann Pharmacother. 2010;44(11):1725–1732. 13. NICE Medicines and Prescribing Centre (UK). Recommendations medication re- view. In: Medicines optimisation: the safe and effective use of medicines to enable the best possible outcomes. National Institute for Health and Care Excellence. 2015. https://www.nice.org.uk/guidance/ng5/chapter/recommenda- tions#medication-review. Accessed Mar 2018.

98 99 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

14. Christensen M, Lundh A. Medication review in hospitalised patients to reduce 28. Farmacotherapeutisch Kompas. Dutch pharmacotherapeutic reference source. morbidity and mortality. Cochrane Database Syst Rev. 2016;2:CD008986. https://www.farmacotherapeutischkompas.nl/ (2018). Accessed Apr 2018. 15. Holland R, Desborough J, Goodyer L, Hall S, Wright D, Loke YK. Does pharma- 29. KNMP Kennisbank: Dutch pharmacotherapeutic reference source. https://kennis- cist-led medication review help to reduce hospital admissions and deaths in bank.knmp.nl/ (2018). Accessed Apr 2018. older people? A systematic review and meta-analysis. Br J Clin Pharmacol. 30. Expertisecentrum pharmacotherapie bij ouderen (EPHOR). Dutch reference source 2008;65(3):303–316. for pharmacotherapy in older people. http://www.ephor.nl/pdf/Ephors-rap- 16. Wallerstedt SM, Kindblom JM, Nylen K, Samuelsson O, Strandell A. Medication re- porten (2018). Accessed Apr 2018. views for nursing home residents to reduce mortality and hospitalization: sys- 31. Lingjaerde O, Ahlfors UG, Bech P, Dencker SJ, Elgen K. The UKU side effect rating tematic review and meta-analysis. Br J Clin Pharmacol. 2014;78(3):488–497. scale. A new comprehensive rating scale for psychotropic drugs and a cross-sec- 17. Kwint HF, Bermingham L, Faber A, Gussekloo J, Bouvy ML. The relationship be- tional study of side effects in neuroleptic-treated patients. Acta Psychiatr Scand tween the extent of collaboration of general practitioners and pharmacists and Suppl. 1987;334:1–100. the implementation of recommendations arising from medication review: a sys- 32. de Vries ST, Haaijer-Ruskamp FM, de Zeeuw D, Denig P. Construct and concurrent tematic review. Drugs Aging. 2013;30(2):91–102. validity of a patient-reported adverse drug event questionnaire: a cross-sectional 18. Geurts MM, Talsma J, Brouwers JR, de Gier JJ. Medication review and recon- study. Health Qual Life Outcomes. 2014;12:103. ciliation with cooperation between pharmacist and general practitioner and 33. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility the benefit for the patient: a systematic review. Br J Clin Pharmacol. 2012; for frail elderly persons. J Am Geriatr Soc. 1991;39(2):142–148. 74(1):16–33. 34. Solomon PR, Hirschoff A, Kelly B, et al. A 7 minute neurocognitive screening bat- 19. Hatah E, Braund R, Tordoff J, Duffull SB. A systematic review and meta-analy- tery highly sensitive to Alzheimer’s disease. Arch Neurol. 1998;55(3):349–355. sis of pharmacist-led fee-for-services medication review. Br J Clin Pharmacol. 2014;77(1):102–115. 35. Reitan RM. The relation of the trail making test to organic brain damage. J Consult Psychol. 1955;19(5):393–394. 20. Jokanovic N, Tan EC, Sudhakaran S, et al. Pharmacist-led medication review in community settings: An overview of systematic reviews. Res Social Adm Pharm. 36. Wechsler D. Wechsler Adult Intelligence Scale. Third ed. San Antonio: The Psycho- 2016;13(4):661–685. logical Corporation; 1997. 5 21. Hadi MA, Alldred DP, Briggs M, Munyombwe T, Closs SJ. Effectiveness of pharma- 37. Kempen GI, Suurmeijer TP. The development of a hierarchical polychotomous ADL- cist-led medication review in chronic pain management: systematic review and IADL scale for noninstitutionalized elders. Gerontologist. 1990;30(4):497–502. meta-analysis. Clin J Pain. 2014;30(11):1006–1014. 38. Kempen GI, Miedema I, Ormel J, Molenaar W. The assessment of disability with 22. Teichert M, Schoenmakers T, Kylstra N, et al. Quality indicators for pharmaceutical the Groningen Activity Restriction Scale. Conceptual framework and psycho- care: a comprehensive set with national scores for Dutch community pharma- metric properties. Soc Sci Med. 1996;43(11):1601–1610. cies. Int J Clin Pharm. 2016;38(4):870–879. 39. EuroQol. EQ-5D-3L questionnaire. www.euroqol.org. Accessed Aug 2014. 23. Buurma H, Bouvy ML, De Smet PA, Floor-Schreudering A, Leufkens HG, Egberts 40. Cohen J. A power primer. Psychol Bull. 1992;112(1):155–159. AC. Prevalence and determinants of pharmacy shopping behaviour. J Clin Pharm 41. Torgerson DJ. Contamination in trials: is cluster randomisation the answer? BMJ. Ther. 2008;33(1):17–23. 2001;322(7282):355–357. 24. Kwint HF, Faber A, Gussekloo J, Bouvy ML. Completeness of medication reviews 42. Guthrie B, Kavanagh K, Robertson C, et al. Data feedback and behavioural change provided by community pharmacists. J Clin Pharm Ther. 2014;39(3):248–252. intervention to improve primary care prescribing safety (EFIPPS): multicentre, 25. Nederlands Huisartsen Genootschap (NHG). Multidisciplinary guideline on poly- three arm, cluster randomised controlled trial. BMJ. 2016;354:i4079. pharmacy in older individuals. 2012. https://www.nhg.org/sites/default/files/ 43. Gould RL, Coulson MC, Patel N, Highton-Williamson E, Howard RJ. Interventions content/nhg_org/uploads/polyfarmacie_bij_ouderen.pdf. Accessed Aug 2015. for reducing use in older people: meta-analysis of randomised 26. WHO Collaborating Centre for Drug Statistics Methodology: ATC/DDD Index. controlled trials. Br J Psychiatry. 2014;204(2):98–107. https://www.whocc.no/atc_ddd_index/ (2018). Accessed Mar 2018. 44. Tay HS, Soiza RL, Mangoni AA. Minimizing anticholinergic drug prescribing in 27. van der Meer HG, Wouters H, van Hulten R, Pras N, Taxis K. Decreasing the load? older hospitalized patients: a full audit cycle. Ther Adv Drug Saf. 2014;5(3):121–128. Is a Multidisciplinary Multistep Medication Review in older people an effective 45. Bpac nz. A practical guide to STOPPING MEDICINES in Older People. BPJ. intervention to reduce a patient’s Drug Burden Index? Protocol of a randomised 2010;27(27):10–23. controlled trial. BMJ Open. 2015;5(12):e009213.

100 101 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

46. Beuscart JB, Pont LG, Thevelin S, et al. A systematic review of the outcomes re- ported in trials of medication review in older patients: the need for a core out- come set. Br J Clin Pharmacol. 2017;83(5):942–952. scoring scoring mean median median in best group % (n) Reporting (SD) (IQR) (IQR) 47. Granas AG, Stendal Bakken M, Ruths S, Taxis K. Deprescribing for frail older peo- ple — Learning from the case of Mrs. Hansen. Res Social Adm Pharm. 2017 [Epub ahead of print]. type linear linear linear logistic 48. van der Meer HG, Taxis K, Pont LG. Changes in Prescribing Symptomatic and Pre- Regression ventive Medications in the Last Year of Life in Older Nursing Home Residents. Front Pharmacol. 2018;8:990. points off Cut- 15 N/A N/A N/A 49. Wouters H, Scheper J, Koning H, et al. Discontinuing inappropriate medication use in nursing home residents: A cluster randomized controlled trial. Ann Intern Med. 2017;167(9):609–617. 50. Hazen AC, Sloeserwij VM, Zwart DL, et al. Design of the POINT study: Pharmaco- logarithmic logarithmic dichotomized 5% classes therapy Optimisation through Integration of a Non-dispensing pharmacist in a Transformation primary care Team (POINT). BMC Fam Pract. 2015;16:76. effects Ceiling Ceiling Normal Left skewed Left skewed Distribution score score worst worst measured 75–7 26–202 55–439 16–5 Best- 5 score score worst worst achievable 1–300 1–600 133–0 16–0 Best- seconds to seconds to symbols scale recalled of items complete complete correct Time in Time in Number of Measurement Measurement Number - presented by examiner (e.g. I show I show (e.g. examiner by presented by followed piece of furniture?) the correct increasing order. increasing the correct multiple for number the correct the correct increasing order while order increasing the correct show you? show ticipants encoded using cues encoded using ticipants letters e.g. 1-A-2-B-3-…etc. letters e.g. displayed above. displayed cued recall using these cues (e.g. these cues (e.g. using cued recall arrays of numbers using a legend a legend using of numbers arrays alternating between numbers and numbers between alternating Connecting a series of numbers in a series of numbers Connecting Connecting numbers and letters in numbers Connecting you four pictures, which one is a pictures, four you Recalling 16 pictures that par that 16 pictures Recalling Matching of the correct symbol to of the correct Matching Description of test “what piece of furniture did I just piece of furniture “what enhanced cued recall Cognitive function Cognitive 7MS Outcome Trailmaking Trailmaking A Test Trailmaking Trailmaking B Test DSST Appendix Table 1: Secondary in our study. outcomes distribution and treatment Table Appendix

102 103 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT - best scor scoring scoring scoring scoring scoring scoring scoring scoring in best median in best mean mean in best ing group ing median in best group group group group % (n) % (n) in % (n) % (n) % (n) Reporting Reporting (IQR) (SD) (SD) (IQR) binomial binomial type type logistic logistic logistic logistic logistic linear linear logistic negative negative logistic negative negative Regression Regression Regression points points 0,5 off off 36 Cut- Cut- 5 6 N/A N/A N/A N/A N/A dichotomized dichotomized dichotomized dichotomized Transformation Transformation N/A N/A N/A N/A ≥15 binomial binomial effects Ceiling Ceiling Floor effects Floor Distribution Distribution Negative Negative Dichotomous Normal Right skewed Left skewed Normal Negative Negative score score score 0–106 0–12 0–84 worst worst measured worst measured 32–5 7–1 10–2 1–(−)0.204 18–66 6–43 Best- Best- 5 score score score 0–113 0–14 0–144 worst worst worst achievable achievable 7–0 10–0 1–0 18–72 Best- Best- 45–0 <15 – ≥15 problems problems produced side effects seconds scale scale scale side effects number of number number of number names of animal error points error correct items correct drawn Severity/ Severity/ Severity of Severity Utilities Time in Number of Points on Measurement Measurement Measurement Number of Number - - - ble in 60 seconds. side effects. side effects. turn around and sit down. turn around stairs. ting the hands to twenty to four. the hands to twenty ting including all the numbers and set all the numbers including linergic side effects. side effects. linergic with activities in daily living e.g. e.g. living with activities in daily regard to mobility, self-care, usual self-care, to mobility, regard orientation. on a vertical visual analogue scale. visual analogue on a vertical dressing oneself and climbing the oneself and climbing dressing activities, pain discomfort and discomfort activities, pain anxiety/depression. Stand up from a chair, walk 3m, walk a chair, Stand up from Questionnaire assessing sedative sedative assessing Questionnaire Questionnaire assessing anticho assessing Questionnaire Questionnaire assessing problems problems assessing Questionnaire Drawing a circle with a clock face face with a clock a circle Drawing Naming as many animals as possi as many Naming Description of test Description of test Assessing patient’s time patient’s Assessing Assessing self-related health state health state self-related Assessing Assessing quality of life with quality of life Assessing side effects temporal temporal orientation orientation drawing category fluency Cognitive function Cognitive 7MS clock 7MS clock 7MS 7MS Groningen Groningen Side effects Sedative Scale Scale UKU Up&Go test Outcome Outcome Quality of life Benton Risk of falls Restriction Restriction EQ-5D-3L: EQ-5D-3L: EQ-5D-3L Visual Analogue Analogue Activities of daily living Activities of daily Activity

104 105 Evaluating a current deprescribing intervention Reducing the anticholinergic/sedative load by medication review — a RCT

Appendix Table 2: Sensitivity analysis of proportion of patients having a ATC code class Intervention (n = 65) Control (n = 80) decrease in DBI ≥ 0.5 DBI med- All medi- DBI med- All medi- ication cation ication cation (%, n) (%, n) (%, n) (%, n) Proportion with decrease Odds ratio (95% p-value Stopped of DBI ≥ 0.5, n/N (%) CI) * R (respiratory system) 0 (0) 7.7 (5) 3.7 (3) 13.8 (11) Intervention Control M (musculo-skeletal system) 3.1 (2) 4.6 (3) 5.0 (4) 5.0 (4) G (genito and sex 3.1 (2) 4.6 (3) 1.3 (1) 1.3 (1) DBI > 1.5 12/64 (18.8) 13/78 (16.7) 1.15 (0.49 to 2.74) 0.746 hormones) DBI < 6 12/64 (18.8) 13/79 (16.5) 1.17 (0.49 to 2.78) 0.720 D (dermatologicals) 0 (0) 1.5 (1) 0 (0) 0 (0) H (systemic hormonal preparations) 0 (0) 0 (0) 0 (0) 2.4 (2) Number of medica- 12/61 (19.7) 12/78 (15.4) 1.35 (0.56 to 3.25) 0.508 L (antineoplastic and immune-mod- 0 (0) 0 (0) 1.3 (1) 1.3 (1) tions at baseline >5 ulating agents) Number of medica- 12/65 (18.5) 13/79 (16.5) 1.15 (0.48 to 2.73) 0.752 S (sensory organs) 0 (0) 0 (0) 0 (0) 1.3 (1) tions at baseline < 20 Total* 30.8 (20) 46.2 (30) 22.5 (18) 41.3 (33) Dose change Age > 66 12/65 (18.5) 12/77 (15.6) 1.23 (0.51 to 2.95) 0.649 N (nervous system) 21.5 (14) 23.1 (15) 21.3 (17) 22.5 (18) Age < 93 12/64 (18.8) 13/79 (16.5) 1.17 (0.49 to 2.78) 0.720 C (cardiovascular system) 10.8 (7) 15.4 (10) 1.3 (1) 2.5 (2) A (alimentary tract and metabolism) 1.5 (1) 4.6 (3) 3.7 (3) 3.8 (3) * Binary logistic regression, unadjusted, according to second analysis. R (respiratory system) 0 (0) 1.5 (1) 1.3 (1) 1.3 (1) B (blood and blood forming organs) 0 (0) 1.5 (1) 0 (0) 0 (0) M (musculo-skeletal system) 0 (0) 1.5 (1) 0 (0) 0 (0) Appendix Table 3: Patients who had medications started, stopped and H (systemic hormonal preparations) 0 (0) 0 (0) 0 (0) 3.8 (3) changed in dose at follow-up in intervention arm and control arm. J (antiinfectives for systemic use) 0 (0) 0 (0) 1.3 (1) 1.3 (1) Total* 27.7 (18) 38.5 (25) 23.8 (19) 28.0 (23) 5 ATC code class Intervention (n = 65) Control (n = 80) Total interventions 53.8 (35) 72.3 (47) 45.0 (36) 66.3 (53) DBI med- All medi- DBI med- All medi- ication cation ication cation ATC = Anatomical Therapeutic Chemical, classification by the WHO Collaborating (%, n) (%, n) (%, n) (%, n) Centre for Drug Statistics Methodology. Based on second analysis. *Not sum of subto- tals, as some patients had several interventions. Started A (alimentary tract and metabolism) 7.7 (5) 20.0 (13) 3.7 (3) 11.3 (9) R (respiratory system) 4.6 (3) 12.3 (8) 3.7 (3) 7.5 (6) N (nervous system) 4.6 (3) 6.2 (4) 3.7 (3) 7.5 (6) M (musculo-skeletal system) 0 (0) 6.2 (4) 3.7 (3) 5.0 (4) C (cardiovascular system) 4.6 (3) 6.2 (4) 1.3 (1) 2.5 (2) B (blood and blood forming organs) 0 (0) 10.8 (7) 0 (0) 7.5 (6) L (antineoplastic and immune-mod- 1.5 (1) 3.1 (2) 0 (0) 0 (0) ulating agents) H (systemic hormonal preparations) 0 (0) 1.5 (1) 0 (0) 0 (0) G (genito urinary system and sex 0 (0) 0 (0) 1.3 (1) 1.3 (1) hormones) S (sensory organs) 0 (0) 0 (0) 0 (0) 1.3 (1) Total* 20.0 (13) 43.1 (28) 13.8 (11) 33.8 (27) Stopped A (alimentary tract and metabolism) 9.2 (6) 21.5 (14) 2.5 (2) 6.3 (5) N (nervous system) 13.8 (9) 15.4 (10) 12.5 (10) 15.0 (12) C (cardiovascular system) 6.2 (4) 9.2 (6) 7.5 (6) 10.0 (8) B (blood and blood forming organs) 0 (0) 9.2 (6) 0 (0) 6.3 (5)

106 107 Deprescribing in older people General introduction and thesis outline

1 CHAPTER 6

FEASIBILITY, ACCEPTABILITY AND POTENTIAL EFFECTIVENESS OF AN INFORMATION TECHNOLOGY BASED, PHARMACIST-LED INTERVENTION TO PREVENT AN INCREASE IN ANTICHOLINERGIC AND SEDATIVE LOAD AMONG OLDER COMMUNITY-DWELLING INDIVIDUALS

Helene G van der Meer, Hans Wouters, Martina Teichert, AMG Fabienne Griens, Jugoslav Pavlovic, Lisa G Pont, Katja Taxis

Therapeutic Advances in Drug Safety [In press]

108 109 IT-based intervention to prevent an increase in anticholinergic/sedative load

ABSTRACT and less likely for pharmacies with lower level of collaboration with GPs (OR 0.15;95% CI 0.02–0.97). Background Anticholinergic/sedative medications are frequently used by older people, despite their negative impacts on cognitive Conclusion This innovative IT-based intervention was feasible, and physical function. We explore the feasibility, acceptability acceptable and potentially effective. In one third of patients an and potential effectiveness of an innovative information technol- increase in anticholinergic/sedative load was prevented within ogy (IT)-based intervention to prevent an increase in anticholin- reasonable time investment. ergic/sedative load in older people.

Methods Prospective study in 51 Dutch community pharmacies. Pharmacists used an IT-tool to identify patients aged ≥65 years, with existing high anticholinergic/sedative loads (Drug Burden Index ≥2) and a newly initiated anticholinergic/sedative medica- tion. We determined the following. Feasibility: number of eligible patients identified. Acceptability: pharmacists’ satisfaction with the intervention, pharmacists’ time investment and patients’ will- ingness to reduce medication use. Potential effectiveness: number of recommendations, rate of agreement of GPs with proposed recommendations and factors associated with agreement. To eval- uate the latter, pharmacists conducted medication reviews and proposed recommendations to GPs for 5–10 patients selected by the IT-tool. 6 Results We included 305 patients from 47 pharmacies. Feasibility: a mean of 17.0 (SD 8.8) patients were identified per pharmacy. Acceptability: 43 pharmacists (91.5%) were satisfied with the in- tervention. Median time investment per patient was 33 minutes (range 6.5–210). Of 35 patients, 30 (85.7%) were willing to reduce medication use. Potential effectiveness: pharmacists proposed 351 recommendations for 212 patients (69.5%). GPs agreed with recommendations for 108 patients (35.4%). Agreement to stop a medication was reached in 19.8% of recommendations for newly initiated medications (37 of 187) and for 15.2% of recommenda- tions for existing medications (25 of 164). Agreement was more likely for recommendations on codeine (OR 3.30;95%CI 1.14–9.57) or medications initiated by a specialist (OR 2.85;95%CI 1.19–6.84)

111 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

BACKGROUND using information technology to identify older individuals with newly initiated anticholinergic/sedative medication is worth- Medications with anticholinergic and/or sedative properties are while to explore. of great concern in older people. They have a negative impact on cognitive and physical function and increase the risk of falls, de- We built an innovative IT-based pharmacist-led intervention to mentia, hospitalization, and mortality. [1–3] Despite these risks, prevent an increase in anticholinergic/sedative load among older anticholinergic and sedative medications are frequently pre- Dutch community-dwelling individuals. In line with best practice scribed to older people. [4, 5] Interventions to reduce the anti- for the development and evaluation of such a complex healthcare cholinergic/sedative load among older individuals are urgently intervention, [12] in this study we test the feasibility, acceptabil- needed. One strategy that has been proposed for reducing this ity and potential effectiveness of this IT-based pharmacist-led load is pharmacist-led medication review. This is ‘a structured, intervention. critical examination of a patient’s medicines with the objective of reaching an agreement with the person about treatment, op- timising the impact of medicines, minimising the number of METHODS medication related problems and reducing waste’. [6] A few stud- ies evaluated the effect of pharmacist-led medication review on Study design & setting chronically used anticholinergic/sedative medications. While two The study was conducted in 51 community pharmacies located small Australian studies found positive effects, [7, 8] we found throughout the Netherlands in both rural and urban areas be- pharmacist-led medication review to have no effect on depre- tween September and December 2017. At each pharmacy, one scribing chronically used anticholinergic/sedative medications in pharmacist participated in the study. All participating pharmacists a recent randomized controlled trial across 15 Dutch community were enrolled in the national 2-year post-graduate programme pharmacies. [9] to become a pharmacist specialized in community pharmacy. Participation in this study was part of their specialization train- 6 Information technology (IT)-based interventions targeting newly ing. Pharmaceutical care is well established in the Netherlands. initiated medications are another approach that potentially may This includes patient counselling for newly initiated medication, reduce anticholinergic/sedative load. Since deprescribing chron- drug-drug interaction monitoring, and performing medication ically used anticholinergic/sedative medications is difficult, using reviews. Pharmacies operate a pharmacy information system IT to identify patients with newly initiated medications and per- with a complete electronic medication history of their patients, forming a medication review to prevent an increase in anticholin- as each individual patient is registered with a single pharmacy. ergic/sedative load may be more successful. The use of IT-based [13] Furthermore, Dutch pharmacists routinely collaborate with approaches to identify patients with potentially ineffective or the general practitioners (GPs) in the area. This includes routine harmful medication use is increasing. [10] In Dutch community contact (phone or face-to-face meeting) to discuss individual pharmacy practice, pharmacists already use IT-based drug ther- patients and regular pharmacotherapy audit meetings. [14] The apy alerts to monitor the safety of medication use in electronic Medical Ethical Committee of the University Medical Centre of patient records (e.g. detecting drug-drug interactions, contra-in- Groningen confirmed that the study did not fall under the scope dications, dosing in patients with renal impairment). [11] Thus, of the Medical Research Involving Human Subjects Act.

112 113 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

IT-based tool to identify eligible patients were included in the calculation. As there is no consensus inter- Each pharmacist ran an online report module containing an al- nationally regarding which medications are considered to have gorithm based on the patient inclusion criteria, described below, anticholinergic properties, [22] we derived a medication list in their pharmacy information system to obtain a list of eligible based on the anticholinergic medication classification by Duran practitionerspatients. The module(GPs) in was the developed area. This by theincludes Dutch routineFoundation contact for (phone or face-to-face meeting) to et al. [23] We also included all medications with reported mild Pharmaceuticaldiscuss individual Statistics patients (SFK). and The regular SFK haspharmacotherapy access to anonymous audit meetings. [14] The Medical Ethical or strong anticholinergic/sedative properties and side effects in Committeepharmacy dispensing of the University data from Medical the pharmacy Centre of informationGroningen confirmed sys- that the study did not fall under Dutch Pharmacotherapeutic reference sources in the DBI cal- thetem scope of more of thethan Medical 95% of Researchthe Dutch Involving community Human pharmacies; Subjects they Act. culation. [20, 21] Topical preparations, ‘as needed’ medications collect these data to analyse national drug utilization and to pro- and medications which lacked a specified dosing regimen in videIT-based pharmaceutical tool to identify services. eligible [15] patients the electronic dispensing records were excluded from the DBI calculation. As the DBI per medication ranges between 0 and 1, Each pharmacist ran an online report module containing an algorithm based on the patient inclusion Eligible patients were aged ≥ 65 years and received a newly pre- depending on the daily dose, our chosen DBI threshold suggests criteria, described below, in their pharmacy information system to obtain a list of eligible patients. scribed potentially inappropriate anticholinergic/sedative med- that the patient is prescribed at least 3–4 anticholinergic/sedative The module was developed by the Dutch Foundation for Pharmaceutical Statistics (SFK). The SFK ication in the past month. A newly prescribed medication was medications. In a previous study we found a frail older patient has access to anonymous pharmacy dispensing data from the pharmacy information system of more defined as a medication, or a medication with a similar action population using about 3–4 anticholinergic/sedative medications than 95% of the Dutch community pharmacies; they collect these data to analyse national drug (World Health Organisation Anatomical Therapeutical Chemical being at risk of medication related harm and in need of medica- utilization and to provide pharmaceutical services. [15] (ATC) code level 3 or 4) [16] dispensed for the first time in a tion optimisation. [9] 12-month period. We screened for those newly prescribed anti- Eligiblecholinergic/sedative patients were medications aged ≥ 65 that years were and known received to be a potennewly- prescribed potentially inappropriate IT-based pharmacist-led intervention anticholinergic/sedativetially inappropriate in older medication people, in including the past mobenzodiazepines,nth. A newly prescribed medication was defined as The intervention consisted of five steps. First, the pharmacist abladder medication, antimuscarinics, or a medication tricyclic antidepressants,with a similar opioids, action classic(World Health Organisation Anatomical obtained a list of eligible patients as described above. For each Therapeuticalantihistamines, Chemical antipsychotics, (ATC) codesecond-generation level 3 or 4) [16]antidepres dispensed- for the first time in a 12-month patient identified with the algorithm, age, gender, DBI, medica- period.sants and We a screened few cardiovascular for those newly medications. prescribed For anticholinergic/sedative these medica- medications that were known tion profile and medication history were displayed to the phar- 6 totions be evidence-basedpotentially inappropriate guidance on in prescribing older people, in older including people benzodiazepines, was bladder antimuscarinics, macist. Medications that contributed to the patients’ DBI, as well tricyclicavailable. [17,antidepressants, 18] Furthermore, opioids, patients neededclassic toantihistamines, have a total cu - antipsychotics, second-generation as newly initiated medications along with their date of prescrip- antidepressantsmulative anticholinergic/sedative and a few cardiovascular load above medications. a predefined For threshthese -medications evidence-based guidance tion were highlighted. Second, from the list of displayed patients, onold prescribingvalue of 2, according in older topeople the Drug was Burden available. Index [17, (DBI). 18] TheFurthermore, DBI patients needed to have a total pharmacists selected 5 to 10 patients whom they wished to in- cumulativeis a measure anticholinergic/sedativeof total cumulative anticholinergic/sedative load above a predefined load and threshold value of 2, according to the clude in this study. Third, the pharmacists evaluated the medi- Drugwas calculated Burden Index as (DBI). The DBI is a measure of total cumulative anticholinergic/sedative load and cation use, both newly initiated and existing medications, and was calculated as drafted recommendations to reduce the anticholinergic/sedative load for each of the selected patients. For this evaluation we pro- ! where D = daily dose and δ = the minimum recommended daily dose. [19] The recommended daily vided pharmacists an evidence-based guidance document outlin- DBI = !! ! dosewhere was D = dailydetermined dose and according δ = the minimumto Dutch recommended Pharmacotherapeutic daily reference sources. [20, 21] All ing information on rational prescribing for those anticholinergic/ medicationsdose. [19] The with recommended potential anticholinergic/sedative daily dose was determined properties accord -were included in the calculation. As sedative medications that are known to be potentially inappro- thereing to Dutchis no Pharmacotherapeuticconsensus internationally reference regarding sources. [20,which 21] Allmedications are considered to have priate in older people, including all newly initiated medications anticholinergicmedications with properties, potential [22] anticholinergic/sedative we derived a medication properties list based on the anticholinergic medication we screened for. Information in the document was based on classification by Duran et al. [23] We also included all medications with reported mild or strong anticholinergic/sedative properties and side effects in Dutch Pharmacotherapeutic reference sources in the114 DBI calculation. [20, 21] Topical preparations, ‘as needed’ medications and medications which 115 lacked a specified dosing regimen in the electronic dispensing records were excluded from the DBI calculation. As the DBI per medication ranges between 0 and 1, depending on the daily dose, our chosen DBI threshold suggests that the patient is prescribed at least 3-4 anticholinergic/sedative medications. In a previous study we found a frail older patient population using about 3-4 anticholinergic/sedative medications being at risk of medication related harm and in need of medication optimisation. [9]

IT-based pharmacist-led intervention The intervention consisted of five steps. First, the pharmacist obtained a list of eligible patients as described above. For each patient identified with the algorithm, age, gender, DBI, medication profile

4

Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

recent Dutch guidelines and also included recommendations on if it was considered practical, clear and educational. In addition, non-pharmaceutical options. [17] Fourth, pharmacists discussed the pharmacists were asked if they would like to continue using recommendations with the GP and if needed medical specialists. the intervention in the future following completion of the study. Pharmacists could choose their preferred communication method with the GP, but we advised a face-to-face meeting. Pharmacist Data on all medications dispensed between June 2017 and and GP agreed who would discuss recommendations for medica- December 2017 for patients who were selected by the pharma- tion changes with the patient, which would be the last step of the cists were provided by SFK. The pharmacists authorized SFK to intervention. provide these data. For each patient the medications used on the dispensing date of the newly initiated medication were identified Data collection from the dataset and used for the analysis. Data were collected by various methods. Data on the pharmacists, participating pharmacies, patients identified with the algorithm, We aimed at conducting a structured telephone interview with patients selected for medication review, time taken for each step 1–2 patients per pharmacy to explore the patients’ perspective on in the process and the medication review changes proposed were reducing their medication use. Each pharmacist asked his/her pa- collected via an online questionnaire completed by the participat- tients included in this study whether they were willing to partic- ing pharmacists. We checked for consistency and completeness of ipate in a telephone interview. Patients who gave verbal consent data reported by the pharmacists and based on this we excluded to the pharmacist received information about the telephone in- four pharmacists from the analysis. terview and an informed consent form. Only patients who signed the informed consent form were interviewed. Each patient inter- For each selected patient, the pharmacists reported age, gender, view lasted about 10 minutes. reasons for selection and details of recommendations proposed to the GP. The latter were reported per medication and included Feasibility type of recommendation (stop/substitute/start medication or Patient identification with IT-tool 6 change dose, checking indication for use, monitoring lab-values We assessed the number of potentially eligible patients identified or giving other advice), type of prescriber (GP or medical special- with the IT-tool per pharmacy and the number of falsely identi- ist), communication-method with GP to discuss recommenda- fied patients. False identification occurred if the calculated DBI tions (face-to-face, phone, fax/email or none), whether agreement by the module was ≥ 2, while in fact the real DBI was < 2. This on the recommendation was reached and if so, who would com- happened due to two problems. First, we detected an error in the municate the recommendations to the patient. If pharmacists had online report module, which appeared if the pharmacist ran an- no recommendations for selected patients, they were asked to other algorithm within the online report module. The SFK solved provide the reasons for this. the error within the first month of data collection, but until this time for these pharmacies the online report module did not only The online questionnaire also included structured questions re- include currently used chronic medication in the DBI calculation, garding the acceptability of the intervention. Structured ques- but also some anticholinergic/sedative medications that were tions on a 3-point Likert-scale were used to assess if pharmacists already stopped. Second, the dispensing data on which the DBI were satisfied with the intervention, if they found it meaningful, was calculated could include pseudo double medication records.

116 117 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

These were records of the same ATC code (level 5), strength and of agreement was assessed per patient and type of recommenda- daily dose as another record within one patient with overlapping tion. Number of patients without recommendations and reasons treatment dates. Pseudo double medication records were a result why were counted and the number of recommendations per ATC of early medication dispenses, e.g. a patient had not yet finished a level 2 and type of communication to the patient was assessed. medication package but a new package was already dispensed. We Agreement was only counted as such if there was a discussion be- reduced all pseudo double medication records to single medica- tween pharmacist and GP or specialist. If the prescriber did not tion records. The DBI was recalculated by hand after adjustment respond to the pharmacist’s recommendation, e.g. if recommen- of the medication data and compared with the DBI calculated by dation was sent via email or fax, this was counted as no agreement. the module. All demographic characteristics and description of medication use were based on the adjusted dataset. Secondly, we assessed whether patient characteristics, type of medication, type of communication between pharmacist and GP, Acceptability type of initiating prescriber and level of pharmacotherapy audit Pharmacist and patient acceptability of the intervention was meeting with GPs were associated with agreement of the GP with assessed. Pharmacists’ acceptability was assessed by asking the pharmacists’ recommendations. Pharmacotherapy audit meet- pharmacists whether they found the IT-based intervention mean- ings were nationally classified into four categories: no structured ingful, practical, clear, and educational. We also assessed their meetings (level 1), regular meetings without concrete agreements willingness to use the intervention in the future. Pharmacists (level 2), regular meetings with concrete agreements (level 3) and were also asked to report the mean time needed per intervention regular meetings with evaluating concrete agreements (level 4). step per patient. [24] Agreements focused on the prescribing and dispensing of medications. For the patient perspective on reducing medication use, we deter- mined the number of patients interviewed who expressed a desire Statistical analysis to reduce their medication use, who were willing to reduce med- Descriptive statistics of all data were derived with IBM SPSS 6 ication use if the GP would advise this and who were not willing Statistics version 25. Factors associated with agreement were an- to reduce their medication use even if the GP would advise this. alysed with logistic mixed effects models in MLwiN version 3. Random effects on the level of pharmacy and patient were applied. Potential effectiveness A univariate analysis on all variables was applied first. Variables Potential effectiveness was assessed in two ways. First, the num- with a univariate p-value < 0.1 were included in the multivariate ber of recommendations proposed by the pharmacist and rate analysis. P-values of < 0.05 were considered significant. of agreement of GP with proposed recommendations was de- termined. We categorized recommendations into medication changes (stopping, substituting and starting a medication or dos- RESULTS age change) and medication monitoring (checking indication for use, monitoring lab-values or giving other advice). We also cat- Study population egorized recommendations for all medications, newly initiated In total 47 pharmacists from 47 community pharmacies were and existing medications. Number of recommendations and rate included in the study. Overall, 305 patients were selected with a

118 119 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

Table 1: Demographic characteristics pharmacists and patients Characteristic Outcome Patients N = 305 Characteristic Outcome Top 5 newly initiated anticholinergic/sedative medications Pharmacists N = 47 (n patients (%): Oxycodone 51 (16.7) Age (mean (SD)) 28.3 (2.3) Codeine 43 (14.1) Gender (% female) 68.1 Tramadol 38 (12.5) Temazepam 26 (8.5) Working experience (mean years (SD)) 2.0 (1.0) Amitriptyline 25 (8.2) Working hours per week (median hours (range)) 40.0 (24–45) Top 5 used medications per ATC level 1 (n patients (%): Cardiovascular system 285 (93.4) Pharmacies N = 47 Alimentary tract and metabolism 274 (90.0) Nervous system 260 (85.2) Pharmacists FTE (mean (SD)) 2.3 (1.0) Blood and blood forming organs 164 (53.8) Number of patients per pharmacy (n pharmacies per category Respiratory system 83 (27.5) (%)): 3 (6.4) < 8000 FTE = fulltime equivalent. *Multiple reasons per patient could be selected. 172 pa- 11 (23.4) 8000–10,000 tients were selected for 1 reasons, 95 for 2 reasons, 38 for > 2 reasons. GP = general 13 (27.7) 10,000–12,000 practitioner. 11 (23.4) 12,000–14,000 9 (19.1) > 14,000 Percentage of patients aged 65+ per pharmacy (n pharmacies per median of 6 patients (range 3–10) per pharmacy selected for med- category, (%)): < 20% 11 (23.4) ication review. The demographic characteristics of pharmacists, 20–50% 24 (51.1) > 50% 10 (21.3) pharmacies and patients are shown in Table 1. Unknown 2 (4.3) Number of collaborating GPs per pharmacy (mean (SD)) 12.1 (6.3) Feasibility Level of pharmacotherapy audit meetings with GPs (n pharma- cies per category, (%)): Patient identification with IT-tool Level 1: no structured meetings 0 (0.0) On average, 17.0 (SD 8.8.0, range 3–32) patients per pharmacy Level 2: regular meetings without concrete agreements 2 (4.3) Level 3: regular meetings with concrete agreements 30 (63.8) were identified with the IT-tool. With calculation of the DBI by Level 4: regular meetings with evaluating concrete agreements 13 (27.7) None 2 (4.3) hand, we found that 13 selected patients (4.3%) had a real DBI < 2. 6 Patients N = 305 These patients were included due to the error we found in the on- Age (mean (SD)) 76.5 (8.0) line report module (n=11) and pseudo double medication records Gender (% female) 64.0 (n=2). In addition, we detected pseudo double medication records DBI value at identification (mean (SD)) 3.6 (1.3) for 85 patients (27.9%), but these patients had a DBI ≥ 2 even after Number of anticholinergic/sedative medications used 5.8 (2.1) removing the double medication records. Without adjusting these (mean (SD)) pseudo double medication records to single records, the mean Number of medications used (mean (SD)) 9.2 (3.3) Reasons for patient selection for medication review (n per DBI of patients selected in this study would have been 4.2 (SD 2.0) category (%))*: versus 3.6 (SD 1.3) after adjustment. Newly initiated medication 142 (46.6) Risk factors (high age, high DBI, risk medication) 159 (52.1) Good collaboration with GP 98 (32.1) Other/ no specific reason 88 (28.9) Acceptability A large majority of pharmacists (n=43, 91.5%) were satisfied with the intervention (17 completely, 26 partly), 41 pharmacists (87.2%) found it meaningful (19 completely, 22 partly), 41 pharmacists

120 121 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

(87.2%) found it practical (16 completely, 25 partly), 46 pharma- For 93 patients (30.5%) no recommendations were proposed. cists (97.9%) found it clear (34 completely, 12 partly) and 44 phar- Reasons for not proposing an intervention were that no medica- macists (93.6%) found it educational (30 completely, 14 partly). tion optimisation was possible, e.g. the medication being for short Almost three quarters of pharmacists (n=33, 70.2%) wanted to term use or patient was already on tapering scheme (62 patients), keep using the intervention in the future. the pharmacist knew beforehand that either patient or GP would not accept any medication recommendation (15 patients), med- The median time investment per patient was 33 minutes (range ication recommendations were difficult as medication was of a 6.5–210). Most time was needed for medication evaluation and specialist nature (9 patients) or due to other reasons, e.g. patient drafting of recommendations (median 15 minutes, range 8–120), had died (6 patients). For 1 patient the pharmacist did not report then for discussion of recommendations with the GP (median the reason for not proposing a recommendation. 10, range 5–60), for patient selection a median of 5 minutes was needed (range 2–60) and the least time was needed to identify In total 351 recommendations were proposed, of which 148 patients with the IT-tool (median 2 minutes, range 1–25). (48.5%) were agreed with by the GP. For 13 of 351 recommen- dations (4.3%) the medical specialist was contacted. Stopping a Telephone interviews were conducted with 35 patients (10.7%). medication or substitution by a safer alternative were the most One in five patients (n=8, 22.9%) reported that they wished to commonly proposed recommendations, respectively 41.3 and stop one or more medications or would stop on GP’s advice (n=22, 32.5% of the total recommendations. The rate of agreement for 62.9%). There were 5 patients (14.3%) who did not want to stop stopping or substituting a medication was higher for newly initi- any medication, even if advised by the GP. ated medications (57.8% and 35.3% agreement) than for existing medications (30.9% and 24.1% agreement). Agreement to stop a Potential effectiveness medication was reached in 17.7% of recommendations (62 of 351), Recommendations were proposed for 212 patients (69.5%), a mean in 19.8% of recommendations for newly initiated medications (37 of 1.7 (SD 0.9) per patient. Recommendations included medica- of 187) and in 15.2% of recommendations for existing medications 6 tion changes (169 patients), medication monitoring (24 patients) (25 of 164), Table 2. or both (19 patients). Overall, the GP agreed with pharmacists’ recommendations in 108 patients (35.4%) and with recommenda- Of the 148 recommendations with agreement, discussion with tions to change medications in 97 patients (31.8%). the patient was done by the GP (n=54, 36.5%), pharmacist (n=46, 31.1%) or someone else (n=9, 6.1%). In some cases there was no Most recommendations were proposed for opioids (ATC N02A, communication with the patient as he/she was not reachable by 16.8%), such as oxycodone and tramadol (respectively 40.7% and phone (n=7, 4.7%) or no discussion was needed (e.g. lab-value 42.4%), antidepressants (ATC N06A, 13.1%), such as amitriptyline check; n=15, 10.1%). For 17 recommendations (11.5%) the pharma- (52.2%), anxiolytics (ATC N05B, 10.3%), such as oxazepam (58.3%), cists did not report who contacted the patient. and sedatives (ATC N05C, 9.7%), such as temazepam (67.6%). A detailed overview of all recommendations proposed on medica- GP agreement with proposed recommendations was more likely tion grouped by ATC level 2 can be found in additional file 1. for recommendations on cough and cold preparations (codeine) (OR 3.30; 95%CI 1.14–9.57) or medication initiated by a medical

122 123 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

Table 2: Type of recommendations by pharmacist and rate of agreement by Table 3: Factors associated with GP agreement with recommended medi- general practitioner* cation changes

Univariate Multivariate Newly initiated Existing medica- Total (n = 351) medications tions (n =164) Factor OR (95% CI) p-value OR (95% CI) p-value (n = 187) Patient characteristics Proposed Agreed Agreed Agreed n (% of Proposed Proposed n (% n (% n (% Age 0.98 (0.95–1.01) 0.264 NA NA Recommendation total n (% of n (% of of pro- of pro- of pro- pro- total) total) Gender 0.67 (0.40–1.14) 0.142 NA NA posed) posed) posed) posed) DBI 0.99 (0.82–1.19) 0.885 NA NA Medication changes Number of medications 1.01 (0.94–1.09) 0.844 NA NA Stop 145 (41.3) 62 (42.8) 64 (34.2) 37 (57.8) 81 (49.4) 25 (30.9) Type of medication Substitute 114 (32.5) 37 (32.5) 85 (45.5) 30 (35.3) 29 (17.7) 7 (24.1) Newly initiated medication 1.75 (1.02–2.99) 0.042* 1.47 (0.84–2.58) 0.182 Dose adjustment 32 (9.1) 15 (46.9) 14 (7.5) 5 (35.7) 18 (11.0) 10 (55.6) Drugs for acid related disor- 1.07 (0.39–2.97) 0.898 NA NA Start 9 (2.6) 5 (55.6) 0 (0) - 9 (5.5) 5 (55.6) ders (ATC code A02) Subtotal 300 (85.5) 119 (39.7) 163 (87.2) 72 (44.2) 137 (83.5) 47 (34.3) Urologicals (ATC code G04) 0.86 (0.28–2.57) 0.794 NA NA

Medication monitoring Analgesics (ATC code N02) 1.60 (0.85–3.00) 0.142 NA NA

Check lab-values 13 (3.7) 12 (92.3) 0 (0) - 13 (7.9) 12 (92.3) Psycholeptic 0.52 (0.29–0.93) 0.027* 0.58 (0.32–1.07) 0.082 Additional infor- (ATC code N05) mation on med- 38 (10.8) 17 (44.7) 24 (12.8) 11 (45.8) 14 (8.5) 6 (42.9) Psychoanaleptic 0.55 (0.26–1.17) 0.119 NA NA ication use (e.g. (ATC code N06) advice or check indication) Cough and cold prepara- 4.71 (1.64–13.50) 0.004* 3.30 (1.14–9.59) 0.028* tions 0(ATC code R05) Subtotal 51 (14.5) 29 (56.9) 24 (12.8) 11 (45.8) 27 (16.5) 18 (66.7) Type of communication between pharmacist and GP Total 351 (100) 148 (42.2) 187 (100) 83 (44.4) 164 (100) 65 (39.6) Face-to-face 2.10 (0.58–7.60) 0.262 NA NA

*for 13 of 351 recommendations (4.3%) a medical specialist was contacted. Telephone 1.41 (0.39–5.05) 0.613 NA NA Fax/Email 1.43 (0.37–5.55) 0.617 NA NA 6 specialist (OR 2.85; 95%CI 1.19–6.84). Furthermore, less estab- None Ref Ref NA NA

lished working collaboration between pharmacist and GP’s re- Initiating prescriber

sulted in less agreement with recommendations compared to Medical specialist 2.44 (1.03–5.79) 0.042* 2.85 (1.19–6.84) 0.019* well-established collaboration (OR 0.15; 95% CI 0.02–0.97), Table 3. GP Ref Ref Ref Ref

Pharmacotherapy audit meeting pharmacist/GPs

None to level 2a 0.13 (0.02–0.82) 0.030* 0.15 (0.02–0.97) 0.047* DISCUSSION Level 3b 1.10 (0.55–2.19) 0.809 NA NA

Level 4c Ref Ref Ref Ref Key findings The innovative IT-based pharmacist-led intervention targeting Variables with a univariate p-value < 0.1 were included in the multivariate analysis. ATC = Anatomical Therapeutical Chemical. NA = not applicable, not included in newly initiated anticholinergic/sedative medications was feasi- the multivariate analysis. GP = general practitioner. Ref: Reference. *Statistically significant. a: no (structured) meetings (level 1) or regular meetings without concrete ble, acceptable and potentially effective. Pharmacists were able agreements (level 2). b: regular meetings with concrete agreements (level 3). c: regular to identify a considerable number of older patients in need of meetings with concrete agreements and evaluation (level 4).

124 125 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

medication optimisation with the IT-tool. Acceptability of the recommendations compared to primary care physicians. [28] intervention was high both among pharmacists and patients. The However, most recommendations in our study focused on psy- potential effectiveness of the intervention appears high with one chotropic medication and contacting a medical specialist for or more recommendations being proposed for over two thirds of these medications might be preferable. patients and agreement of GP with pharmacists’ recommenda- tions for one third of all patients. Agreement was more likely for Strengths and limitations recommendations on codeine use, for medications initiated by a We developed and evaluated an innovative intervention per- medical specialist, and when pharmacist and GP had a well-estab- formed in a relatively large homogenous group of motivated, lished working collaboration. early-career pharmacists who had access to the full medication records for their patients and who were trained in pharmaceutical Comparison with other studies patient care. The evaluation conducted was robust, following ac- The fragile older population with a high anticholinergic/sedative cepted guidance for the development and evaluation of complex load included in this study was comparable to the population health care interventions and included feasibility, acceptability selected in our previous randomized controlled trial on pharma- and potential effectiveness in a large number of pharmacists and cist-led medication review in terms of age, gender, DBI and medi- patients. [12] This evaluation provides valuable information for cation use. [9] While the previous study found that pharmacist-led further development and testing of the intervention. The inter- medication review was not effective in reducing anticholinergic/ vention was designed for the convenience of the pharmacist and sedative load associated with chronic medication, our new ap- pharmacists could adapt it to fit his or her practice in the real proach targeting newly initiated anticholinergic/sedative med- world setting. Analysing the impact of the intervention, we iden- ications appears more successful, especially for newly initiated tified the number of recommendations and classified those in a medications including anxiolytics, hypnotics and antidepressants. meaningful way, distinguishing between medication changes and While our approach is innovative in the pharmacy setting, our re- medication monitoring. sults are comparable with a study in the general practice setting, 6 which found that newly initiated benzodiazepines and tricyclic Some limitations need to be taken into consideration when in- antidepressants, were more likely to be successfully reduced by terpreting our results. First, due to the nature of our study it was GPs than long term used hypnotics. [25] GP agreement with phar- not possible to perform a follow-up meeting. We therefore do macists’ recommendations in our study was comparable to others. not know whether all planned medication changes were imple- In line with these studies, agreement seemed higher when GP and mented. We report on the agreement of the GP with pharmacists’ pharmacist had a well-established working collaboration. [26] recommendations, which may overestimate actual implemented medication changes. Also, it was outside the scope of our study to The GP was more likely to agree with recommendations for med- explore to what extent pharmacist or GP communicated recom- ications with unknown or questionable efficacy and a high side mendations with the patient. Second, we do not know whether effect profile, such as codeine. [27] We found that agreement was all steps of the intervention were followed in the proposed order, higher for medications initiated by a medical specialist compared e.g. some patients could have been contacted before discussion to GP. This was surprising as previous literature found that med- with the GP. However, this was a result of the real-world nature ical specialists in general are less likely to agree with pharmacist of the intervention and allowing some flexibility in the order of

126 127 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

steps is likely to make a strategy more pragmatic in clinical prac- therapy alerts in Dutch pharmacy practice, the algorithm should tice. Third, using dispensed medication records has disadvan- be fully integrated in the pharmacy information system. This way tages, such as pseudo double medication records which resulted it will operate prospectively with the system deploying an alert for in a small number of false positive identification of patients a newly initiated anticholinergic/sedative medication that would with pseudo double medication records with the same ATC level increase the patient’s total anticholinergic/sedative load above 5 medications. There could have also been patients with pseudo the specific threshold at the time the prescription is presented double medication records with different ATC codes, e.g. patients for initial supply. Further therapeutic advice for reducing the load who switched to another medication with similar effects. Fourth, should be directly displayed alongside the alert. This way, the wide confidence intervals suggest that our data sample was not pharmacist is able to propose and discuss recommendations with large enough to draw strong conclusions about the factors associ- the GP prior to dispensing the medication. These refinements will ated with GP agreement with recommended medication changes. likely increase the rate of implementation of recommendations, as But we believe that our findings are a basis for further refinement the medication change is being implemented before the patient of the intervention. Finally, this project was part of the pharma- has commenced treatment with the newly prescribed medication. cists’ post-graduate training and all pharmacists should have been Secondly, as no consensus based list of anticholinergic/sedative able to perform the intervention. However, we had to exclude medication is available, [29] we included a broad range of medi- four pharmacists as data provided from these pharmacies was in- cations with mild and strong anticholinergic/sedative properties consistent. We think that this is a reflection of real world prac- and/or reported side effects. Most recommendations in our study tice, in which practicalities, e.g. building renovations, changing were proposed for medications with strong anticholinergic/seda- IT-system, sickness, holidays, lack of personnel, but perhaps also tive properties, such as psychotropic and bladder antimuscarinics, lack of motivation may affect the performance of interventions. only a few recommendations were proposed for medications with Furthermore, there might be a difference in motivation of using mild or unknown anticholinergic/sedative properties, like cardio- the IT-tool between the young pharmacists in our study com- vascular medication. We suggest a refinement of our list, including pared to more experiences pharmacists in practice. only medications with known anticholinergic/sedative proper- 6 ties and frequently reported anticholinergic/sedative side effects, Conclusions and implications for practice and further research this may reduce alert fatigue. [30] Furthermore, while we used The pharmacist-led IT-based intervention as performed in this the Drug Burden Index to calculate the anticholinergic/sedative study, appears feasible, acceptable and potentially effective. load, other tools have been developed, amongst those, one that Pharmacists needed on average about half an hour to perform the shows promising results. [31, 32] Finally while the feasibility, ac- intervention and in 1 out of 3 patients the GP agreed with phar- ceptability and potential effectiveness of the intervention appears macists’ recommendations to change medication. Therefore, when high, the cost-effectiveness and implementation of medication extrapolating, about 1.5 hours was needed to prevent an increase recommendations and long-term medication changes in com- in anticholinergic/sedative load in one patient. Our results sug- bination with relevant patient outcomes, like geriatric outcomes gest some refinements of the intervention should be considered (e.g. fall risk, frailty and cognitive function) and adverse events prior to upscaling. Our study used the algorithm retrospectively to (e.g. drug-related hospital admission) [33] should be evaluated in a identify patients over the past month who could be considered for real-world randomized controlled trial in community pharmacies a medication review. In line with the current use of IT-based drug preferably with high level collaboration with GPs.

128 129 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

REFERENCES 14. Teichert M, Schoenmakers T, Kylstra N, et al. Quality indicators for pharmaceutical care: a comprehensive set with national scores for Dutch community pharma- 1. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- cies. Int J Clin Pharm. 2016;38(4):870–879. erties on cognitive function, delirium, physical function and mortality: a sys- 15. Foundation for Pharmaceutical Statistics. Foundation for Pharmaceutical Statistics. tematic review. Age Ageing. 2014;43(5):604–615. 2018. https://www.sfk.nl/english. Accessed April 2018. 2. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls 16. WHO Collaborating Centre for Drug Statistics Methodology. ATC/DDD Index. 2017. in older people: systematic review of publications from a recent 5-year period. https://www.whocc.no/atc_ddd_index/. Accessed March 2018. Eur J Clin Pharmacol. 2015;71(12):1429–1440. 17. Expertisecentrum pharmacotherapie bij ouderen (EPHOR). Dutch reference 3. Gray SL, Anderson ML, Dublin S, et al. Cumulative use of strong anticholiner- source for pharmacotherapy in older people. 2018. http://www.ephor.nl/pdf/ gics and incident dementia: a prospective cohort study. JAMA Intern Med. Ephors-rapporten. Accessed April 2018. 2015;175(3):401–407. 18. O’Mahony D, O’Sullivan D, Byrne S, O’Connor MN, Ryan C, Gallagher P. STOPP/ 4. Holvast F, van Hattem BA, Sinnige J, et al. Late-life depression and the associa- START criteria for potentially inappropriate prescribing in older people: version tion with multimorbidity and polypharmacy: a cross-sectional study. Fam Pract. 2. Age Ageing. 2015;44(2):213–218. 2017;34(5):539–545. 19. Hilmer SN, Mager DE, Simonsick EM, et al. A drug burden index to define the 5. Bell JS, Mezrani C, Blacker N, et al. Anticholinergic and sedative medicines — functional burden of medications in older people. Arch Intern Med. 2007;167(8): prescribing considerations for people with dementia. Aust Fam Physician. 781–787. 2012;41(1–2):45–49. 20. Farmacotherapeutisch Kompas. Dutch pharmacotherapeutic reference source. 2018. 6. NICE Medicines and Prescribing Centre (UK). Recommendations medication review. https://www.farmacotherapeutischkompas.nl/. Accessed April 2018. In: Medicines optimisation: the safe and effective use of medicines to enable the best possible outcomes. National Institute for Health and Care Excellence. 2015. 21. KNMP Kennisbank. Dutch pharmacotherapeutic reference source. 2018. https:// https://www.nice.org.uk/guidance/ng5/chapter/recommendations#medication-­ kennisbank.knmp.nl/. Accessed April 2018. review. Accessed March 2018. 22. Pont LG, Nielen JT, McLachlan AJ, et al. Measuring anticholinergic drug exposure 7. Castelino RL, Hilmer SN, Bajorek BV, Nishtala P, Chen TF. Drug Burden Index and in older community-dwelling Australian men: a comparison of four different potentially inappropriate medications in community-dwelling older people: the measures. Br J Clin Pharmacol. 2015;80(5):1169–1175. impact of Home Medicines Review. Drugs Aging. 2010;27(2):135–148. 23. Duran CE, Azermai M, Vander Stichele RH. Systematic review of anticholinergic 8. Gnjidic D, Le Couteur DG, Abernethy DR, Hilmer SN. A pilot randomized clini- risk scales in older adults. Eur J Clin Pharmacol. 2013;69(7):1485–1496. cal trial utilizing the drug burden index to reduce exposure to anticholiner- 24. Van Dijk L, Barnhoorn H, De Bakker DH. [Pharmacotherapy audit meetings in gic and sedative medications in older people. Ann Pharmacother. 2010;44(11): 1999: the state of affairs] Dutch. Utrecht: NIVEL, 2001. 6 1725–1732. 25. Tamblyn R, Huang A, Perreault R, et al. The medical office of the 21st century 9. van der Meer HG, Wouters H, Pras N, Taxis K. Reducing the anticholinergic and sed- (MOXXI): effectiveness of computerized decision-making support in reducing ative load in older patients on polypharmacy by pharmacist-led medication re- inappropriate prescribing in primary care. CMAJ. 2003;169(6):549–556. view: A randomized controlled trial. BMJ Open. 2018;8(7):e019042. 26. Kwint HF, Bermingham L, Faber A, Gussekloo J, Bouvy ML. The relationship be- 10. Dreischulte T, Donnan P, Grant A, Hapca A, McCowan C, Guthrie B. Safer Prescrib- tween the extent of collaboration of general practitioners and pharmacists and ing—A Trial of Education, Informatics, and Financial Incentives. N Engl J Med. the implementation of recommendations arising from medication review: a sys- 2016;374(11):1053–1064. tematic review. Drugs Aging. 2013;30(2):91–102. 11. Heringa M, Floor-Schreudering A, Tromp PC, de Smet PA, Bouvy ML. Nature and 27. Bolser DC, Davenport PW. Codeine and cough: an ineffective gold standard. Curr frequency of drug therapy alerts generated by clinical decision support in com- Opin Allergy Clin Immunol. 2007;7(1):32–36. munity pharmacy. Pharmacoepidemiol Drug Saf. 2016;25(1):82–89. 28. Perera PN, Guy MC, Sweaney AM, Boesen KP. Evaluation of prescriber responses 12. Moore GF, Audrey S, Barker M, et al. Process evaluation of complex interventions: to pharmacist recommendations communicated by fax in a medication therapy Medical Research Council guidance. BMJ. 2015;350:h1258. management program (MTMP). J Manag Care Pharm. 2011;17(5):345–354. 13. Buurma H, Bouvy ML, De Smet PA, Floor-Schreudering A, Leufkens HG, Egberts 29. Wouters H, van der Meer H, Taxis K. Quantification of anticholinergic and sedative AC. Prevalence and determinants of pharmacy shopping behaviour. J Clin Pharm drug load with the Drug Burden Index: a review of outcomes and methodologi- Ther. 2008;33(1):17–23. cal quality of studies. Eur J Clin Pharmacol. 2017;73(3):257–266.

130 131 Developing and evaluating a new deprescribing intervention IT-based intervention to prevent an increase in anticholinergic/sedative load

30. Heringa M, van der Heide A, Floor-Schreudering A, De Smet PAGM, Bouvy ML. Appendix 1: Type of recommendations proposed by pharmacists on medi- Better specification of triggers to reduce the number of drug interaction alerts in cation grouped by ATC level 2 and rate of agreement by general practitioner primary care. Int J Med Inform. 2018;109:96–102. All medications Newly initiated Existing medications 31. Wauters M, Klamer T, Elseviers M, et al. Anticholinergic Exposure in a Cohort of medications Adults Aged 80 years and Over: Associations of the MARANTE Scale with Mor- ATC Pro- Agreed ATC Pro- Agreed ATC Pro- Agreed tality and Hospitalization. Basic Clin Pharmacol Toxicol. 2017;120(6):591–600. code posed n (% of code posed n (% of code posed n (% of 32. Klamer TT, Wauters M, Azermai M, et al. A Novel Scale Linking Potency and Dos- n (% of pro- n (% of pro- n (% of pro- total) posed) total) posed) total) posed) age to Estimate Anticholinergic Exposure in Older Adults: the Muscarinic Ace- Medication changes: stop tylcholinergic Receptor ANTagonist Exposure Scale. Basic Clin Pharmacol Toxi- N05 45 (12.8) 14 (31.1) N02 27 (14.4) 17 (63.0) N05 24 (14.6) 4 (16.7) col. 2017;120(6):582–590. N02 29 (8.3) 19 (65.5) N05 21 (11.2) 10 (47.6) A02 14 (8.5) 5 (35.7) 33. Beuscart JB, Pont LG, Thevelin S, et al. A systematic review of the outcomes re- G04 15 (4.3) 6 (40.0) G04 6 (3.2) 3 (50.0) G04 9 (5.5) 3 (33.3) ported in trials of medication review in older patients: the need for a core out- A02 14 (4.0) 5 (35.7) R05 4 (2.1) 3 (75.0) N06 9 (5.5) 2 (22.2) come set. Br J Clin Pharmacol. 2017;83(5):942–952. N06 13 (3.7) 4 (30.8) N06 4 (2.1) 2 (50.0) N07 4 (2.4) 0 (0) C03 4 (2.4) 3 (75.0) Subtotal 145 (41.3) 62 (42.8) Total 64 (34.2) 37 (57.8) Total 81 (49.4) 25 (30.9) Medication changes: substitute N05 30 (8.5) 6 (20.0) N06 20 (10.7) 6 (30.0) N05 10 (6.1) 3 (30.0) N06 27 (7.7) 8 (29.6) N02 21 (11.2) 9 (42.9) N06 7 (4.3) 2 (28.6) N02 23 (6.6) 9 (39.1) N05 20 (10.7) 3 (15.0) C07 4 (2.4) 0 (0) R05 15 (4.3) 9 (60.0) R05 15 (8.0) 9 (60.0) N02 2 (1.2) 0 (0) R06 5 (1.4) 2 (40.0) R06 4 (2.1) 2 (50.0) C09 3 (1.8) 1 (33.3) Subtotal 114 (32.5) 37 (32.5) Total 85 (45.5) 30 (35.3) Total 29 (17.7) 7 (24.1) Medication changes: dose adjustment N05 10 (2.8) 4 (40.0) N05 5 (2.7) 2 (40.0) N05 5 (3.0) 2 (40.0) N02 6 (1.7) 1 (16.7) N02 4 (2.1) 0 (0) A02 5 (3.0) 2 (40.0) A02 5 (1.4) 2 (40.0) N06 3 (1.6) 1 (33.3) N02 2 (1.2) 1 (50.0) N06 3 (0.9) 1 (33.3) R05 2 (1.1) 2 (100) A11 2 (1.2) 2 (100) R05 2 (0.6) 2 (100.0) C07 1 (0.6) 1 (100) 6 A11 2 (0.6) 2 (100.0) Subtotal 32 (9.1) 15 (46.9) Total 14 (7.5) 5 (35.7) Total 18 (10.9) 10 (55.6) Medication changes: start C10 3 (0.9) 1 (33.3) C10 3 (1.8) 1 (33.3) A12 2 (0.6) 2 (100) A12 2 (1.2) 2 (100) A06 2 (0.6) 1 (50.0) A06 2 (1.2) 1 (50.0) C03 1 (0.3) 0 (0) C03 1 (0.6) 0 (0) A02 1 (0.3) 1 (100) A02 1 (0.6) 1 (100) Subtotal 9 (2.6) 5 (55.6) Total 0 (0) - Total 9 (5.5) 5 (55.6) Medication monitoring: check lab-values C10 5 (1.4) 5 (100) C10 5 (3.0) 5 (100) C03 3 (0.9) 3 (100) C03 3 (1.8) 3 (100) B03 2 (0.6) 1 (50.0) B03 2 (1.2) 1 (50.0) C09 2 (0.6) 2 (100) C09 2 (1.2) 2 (100) C07 1 (0.3) 1 (100) C07 1 (0.6) 1 (100) Subtotal 13 (3.7) 12 (92.3) Total 0 (0) - Total 13 (7.9) 12 (92.3)

132 133 Developing and evaluating a new deprescribing intervention

All medications Newly initiated Existing medications medications ATC Pro- Agreed ATC Pro- Agreed ATC Pro- Agreed code posed n (% of code posed n (% of code posed n (% of n (% of pro- n (% of pro- n (% of pro- total) posed) total) posed) total) posed) Medication monitoring: additional information on medication use Un- 19 (5.4) 0 (0) Un- 11 (5.9) 0 (0) Un- 8 (4.9) 0 (0) known known known G04 6 (1.7) 5 (83.3) G04 5 (2.7) 4 (80.0) A02 2 (1.2) 2 (100) N05 4 (1.1) 4 (100) N05 4 (2.1) 4 (100) C08 2 (1.2) 2 (100) N06 3 (0.9) 3 (100) N06 2 (1.1) 2 (100) G04 1 (0.6) 1 (100) C08 3 (0.9) 3 (100) C08 1 (0.5) 1 (100) N06 1 (0.6) 1 (100) A02 2 (0.6) 2 (100) N02 1 (0.5) 0 (0) Subtotal 38 (10.8) 17 (44.7) Total 24 (12.8) 11 (45.8) Total 14 (8.5) 6 (42.9) Total recommendations N05 89 (25.4) 28 (31.5) N02 53 (28.3) 26 (49.1) N05 39 (23.8) 9 (23.1) N02 59 (16.8) 29 (49.2) N05 50 (26.7) 19 (38.0) A02 22 (13.4) 10 (45.5) N06 46 (13.1) 16 (34.8) N06 29 (15.5) 11 (37.9) N06 17 (10.4) 5 (29.4) G04 23 (6.6) 12 (52.2) R05 21 (11.2) 14 (66.7) C10 11 (6.7) 7 (63.6) A02 22 (6.3) 10 (45.5) G04 13 (7.0) 8 (61.5) G04 10 (6.1) 4 (40.0) R05 21 (6.0) 14 (66.7) C03 8 (4.9) 6 (75.0) Total 351 (100) 148 (42.2) Total 187 (100) 83 (44.4) Total 164 (100) 65 (39.6)

ATC = Anatomical Therapeutical Chemical.

134 Deprescribing in older people General introduction and thesis outline

1 CHAPTER 7

GENERAL DISCUSSION

136 137 General discussion

MAIN FINDINGS

This thesis addresses different stages in the process of develop- ing and evaluating deprescribing interventions. Deprescribing interventions usually are complex, involving various stakehold- ers, such as doctors, pharmacists, patients and patients’ care- givers. The UK Medical Research Council (MRC) has published guidance regarding best practice for developing and evaluating complex healthcare interventions. [1] Thus, we positioned the chapters of this thesis in the context of the MRC guidance (Figure 1).

In the first part of this thesis, two methods for the identification of potentially inappropriate prescribing of specific medications within defined populations were explored. Using a retrospective cohort study we identified high use of preventive medications at the end of life in older nursing home residents. The study also highlights the use of routinely collected data for the identifica- tion of potentially inappropriate prescribing (Chapter 2). The study showed that deprescribing remained limited, with little change in medication prescribing throughout the last year of life. Small increases in the prescribing of symptomatic medications indicated some awareness of changed need, but high prescribing of preventive medications suggested treatment goals were not be- ing revised when life expectancy changed. Using a cross-sectional study design in a national population of community-dwelling older persons, we identified individuals with a high anticholiner- 7 gic/sedative load (Chapter 3). Our results showed that a large pro- portion of older community-dwelling patients in the Netherlands had a high anticholinergic/sedative medication load. According to their medication use, four distinct subpopulations with high anticholinergic/sedative loads were identified using latent class analysis. Both studies provide evidence for deprescribing oppor- tunities, and therefore are positioned in the development stage of the MRC development and evaluation process.

139 Deprescribing in older people General discussion

While it is best practice to evaluate an intervention before im- plementing it in practice, sometimes interventions are imple- mented without first evaluating effectiveness. In this case, the effectiveness of the intervention might need to be evaluated in practice, for example in specific circumstances (defined target populations/specific medications). The deprescribing interven- tion, pharmacist-led medication review as currently performed in the Netherlands, was evaluated in a randomised controlled trial to examine its effectiveness on deprescribing chronically used an- Figure 1: Position of the studies described in this thesis in the figure ‘Key ticholinergic/sedative medications in older community-dwelling elements of the development and evaluation process’ by the UK Medical Research Council. Adapted by permission from BMJ Publishing Group patients with a high anticholinergic/sedative load. The results of Limited. Developing and evaluating complex interventions: the new this study showed that while vulnerable older people in need of Medical Research Council guidance, Craig P, Dieppe P, Macintyre S, et al., medication optimisation were targeted, pharmacist-led medica- 337:a1655, ©2008. tion reviews were not effective in deprescribing anticholinergic/ sedative medications in this population (Chapter 4 and 5). This WHAT CAN WE LEARN FROM THIS THESIS? study showed the need to go back to the development and fea- sibility/piloting stage in order to target the intervention to this This thesis shows a robust approach of developing and evaluating population. The study also suggests the need for de-implementa- deprescribing interventions, by performing a variety of studies tion of ineffective medication review activities in this population. investigating different stages in the process. All studies in this thesis focused on older people at high risk for medication related Based on the results of Chapter 3, 4 and 5, a new deprescribing in- harm. Within each study, specific subpopulations of older people tervention was developed and its feasibility, acceptability and po- were targeted, such as nursing home residents at the end of life or tential effectiveness were tested in a prospective study (Chapter 6). community-dwelling older patients. Furthermore, different med- This study design was more sophisticated than a feasibility/pilot, ication groups were studied (preventive medications, anticholin- but not yet a definite effectiveness evaluation (randomised con- ergic/sedative medications) using different study designs (retro- trolled trial) and is therefore positioned between these two stages spective, cross-sectional, randomised controlled trial, feasibility/ in the MRC development and evaluation process. The new depre- acceptability/potential effectiveness study). 7 scribing intervention signalled initiation of a new anticholinergic/ sedative medication in older people with an already high anticho- This thesis highlights that pharmacist-led medication review as linergic/sedative load. The results of the study showed that this currently conducted in the Netherlands is not a suitable interven- intervention was feasible, as a considerable number of patients in tion for deprescribing chronically used anticholinergic/sedative need of medication optimisation could be identified. Acceptability medications among older community-dwelling patients. While of the intervention was high both among pharmacists and pa- effectiveness of pharmacist-led medication review for optimising tients and time investment was reasonable. Also, the intervention medication was found in other population groups, such as users of was potentially effective, as in one third of patients an increase in cardiovascular medication [2] and nursing home residents, [3] the anticholinergic/sedative load was prevented. lack of effectiveness of pharmacist-led medication review in our

140 141 Deprescribing in older people General discussion

population may be due to a number of factors. Firstly, it may be medication use and a good collaboration with the patient’s GP is related to differences in patient populations and risk assessment needed. [10] of their medications, e.g. long term risks for cognitive- and phys- ical decline in anticholinergic/sedative medications, [4, 5] which While for anticholinergic/sedative medications we developed and may be difficult to assess, versus measurable hard outcomes, such evaluated interventions, it was outside the scope of this thesis to as blood pressure, for cardiovascular medication. [6] Secondly, perform this for preventive medications used at the end of life in differences in the level of collaboration between the pharmacist older nursing home residents. Our study showed prescribing of and physician. Close collaboration of physicians specialised in preventive medication in this population is high, suggesting the aged care with pharmacists in nursing homes, might have been potential need for deprescribing. Since the initiation of our study accountable for the positive effects of medication review seen in in 2013, several studies have been performed in this area, includ- this setting. [3] Several barriers to deprescribing have been iden- ing an exploration of patients’, relatives’, nurses’ and physicians’ tified, including the collaboration between different health care perspective on medication management at the end of life [11] and professionals. [7] Based on our findings we would like to add an identification of barriers of physicians to deprescribe medications important barrier/enabler, which is targeting the deprescribing to at the end of life. [12] An efficacy study on the use of antipsy- the right subpopulation. chotics for delirium in palliative care [13] indicates the growing awareness of the need for evidence-base prescribing in individu- Considering these barriers, an innovative deprescribing inter- als at the end of life. A first pragmatic trial showed positive effects vention on anticholinergic/sedative medications was developed of deprescribing statins at the end of life. [14] Suggestions were targeting newly initiated, instead of chronic, anticholinergic/ even made for best choices on study designs for the evaluation of sedative medications. Evaluation of the intervention showed it deprescribing in an older population with limited life expectancy. was a feasible, acceptable and potentially effective approach to re- [15] As highlighted already, multidisciplinary medication reviews ducing anticholinergic/sedative load in older community-dwell- have been found to be effective to discontinue inappropriate ing adults. The intervention targeted specific medications rather medication in nursing home residents. [3] This body of research than all chronic anticholinergic/sedative medications, it was less is a good basis to develop specific interventions optimising med- time consuming than current medication review processes [8, 9] ication use at the end of life in older people, following the MRC and the newly initiated anticholinergic/sedative medication and development and evaluation process. other anticholinergic/sedative medications were highlighted for 7 the pharmacist when selecting a patient. Furthermore, guidance on prescribing of relevant anticholinergic/sedative medications IMPLICATIONS FOR PRACTICE AND FURTHER RESEARCH was provided, helping the pharmacist to propose evidence-based recommendations for medication optimisation. While showing Our studies support the concept of the development and eval- promise in terms of effectiveness, this study also suggests some uation process of complex healthcare interventions following refinements of the intervention, such as electronic integration of best practice guidance. [1] Good evidence on real world practice the intervention, including most updated evidence-based phar- is required before implementing an intervention and scaling up. macotherapeutic advice and a more specific focus on relevant [16] Interventions that are implemented without evidence-based medications. Furthermore, a complete overview of a patient’s development and extensive evaluation can lead to ineffective

142 143 Deprescribing in older people General discussion

clinical practice and unnecessary time- and financial investment. skills might improve the interprofessional collaboration. [25] [17] The MRC framework provides useful guidance in the process More specific guidelines are needed to help healthcare profes- of development and evaluation of interventions, but based on this sionals in the deprescribing process. The underlying pharma- thesis we would suggest an amendment to this process. The MRC cotherapeutic evidence remains weak, although some work has process does not include the step from implementation back to been done, e.g. the study about efficacy of antipsychotics for de- evaluation, while this thesis shows that sometimes this step may lirium in palliative care, described above. [13] Deprescribing in- be needed. Thus we propose the red arrow in Figure 1. If an al- terventions should be patient-centred and recommendations for ready implemented intervention is to be used for different medi- deprescribing should be based on shared decision-making with cations or population groups, it should be re-evaluated and if not the patient. [26] More work is needed to understand the patient effective, the intervention should not be used for this population. perspective in this process. This may include helping patients This may lead to de-implementation of the intervention. [18] understand the benefits of deprescribing medication and taking away their fears for adverse effects of deprescribing. [27] Despite the extensive work in this thesis, a number of questions remain. Our study about preventive medication at the end of life raises the question about the timing of deprescribing. At which CONCLUSIONS point do risks outweigh the benefits and when is a patient at his/her end of life? Our study on medication reviews leaves the Opportunities for deprescribing exist in older populations, such question which patient population may benefit from medica- as the high prescribing of preventive medications at the end of tion reviews. What is the best way to identify this patient group? life in older nursing home residents and anticholinergic/seda- Algorithms seem to be an efficient way to identify target popu- tive medications in older community-dwelling adults. To reduce lations for an intervention. Furthermore, how do we know a pa- a high anticholinergic/sedative load in community-dwelling­ tient is benefiting from deprescribing? Robust meta-analyses have older adults, medication reviews, as currently performed in the not been able to show significant effects of medication review Netherlands, are not effective. An innovative deprescribing in- on hard outcomes, such as hospitalisation and mortality. [19–22] tervention using information technology to target newly pre- Outcome reporting of studies evaluating medication review is scribed anticholinergic/sedative medication in this population heterogeneous. A core set of relevant patient outcomes, like geri- seems more successful. Future deprescribing strategies should be atric outcomes (e.g. fall risk, frailty and cognitive function) and patient-­centred, targeted to the right populations and medica- 7 adverse events (e.g. side effects, drug-related hospital admission) tions, tailored to patient's needs, and should have a high degree of should be evaluated in real world randomised controlled trials. interprofessional collaboration. [23] Furthermore, cost-effectiveness evaluations should be per- formed to advise policy makers about potential implementation of interventions. [16] Interprofessional collaboration of health- care providers, such as pharmacists and general practitioners, but also medical specialists should be improved, as poorly developed interprofessional relationships are an important barrier in the de- prescribing process. [24] Improving pharmacists’ communication

144 145 Deprescribing in older people General discussion

REFERENCES 14. Kutner JS, Blatchford PJ, Taylor DH,Jr, et al. Safety and benefit of discontinuing sta- tin therapy in the setting of advanced, life-limiting illness: a randomized clinical 1. Craig P, Dieppe P, Macintyre S, et al. Developing and evaluating complex interven- trial. JAMA Intern Med. 2015;175(5):691–700. tions: the new Medical Research Council guidance. BMJ. 2008;337:a1655. 15. Geijteman EC, Tiemeier H, van Gelder T. Selecting the Optimal Design for Drug 2. Jokanovic N, Tan EC, Sudhakaran S, et al. Pharmacist-led medication review in com- Discontinuation Trials in a Setting of Advanced, Life-Limiting Illness. JAMA In- munity settings: An overview of systematic reviews. Res Social Adm Pharm. tern Med. 2015;175(10):1724–1725. 2017;13(4):661–685. 16. Guthrie B, Gillies J, Calderwood C, Smith G, Mercer S. Developing mid- 3. Wouters H, Scheper J, Koning H, et al. Discontinuing inappropriate medication use dle-ground research to support primary care transformation. Br J Gen Pract. in nursing home residents: A cluster randomized controlled trial. Ann Intern 2017;67(664):498–499. Med. 2017;167(9):609–617. 17. Verkerk EW, Tanke MAC, Kool RB, van Dulmen SA, Westert GP. Limit, lean or lis- 4. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- ten? A typology of low-value care that gives direction in de-implementation. Int erties on cognitive function, delirium, physical function and mortality: a sys- J Qual Health Care. 2018 [Epub ahead of print]. tematic review. Age Ageing. 2014;43(5):604–615. 18. Norton WE, Kennedy AE, Chambers DA. Studying de-implementation in health: an 5. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls analysis of funded research grants. Implement Sci. 2017;12(1):144. in older people: systematic review of publications from a recent 5-year period. 19. Wallerstedt SM, Kindblom JM, Nylen K, Samuelsson O, Strandell A. Medication re- Eur J Clin Pharmacol. 2015;71(12):1429–1440. views for nursing home residents to reduce mortality and hospitalization: sys- 6. Tan EC, Stewart K, Elliott RA, George J. Pharmacist services provided in general tematic review and meta-analysis. Br J Clin Pharmacol. 2014;78(3):488–497. practice clinics: a systematic review and meta-analysis. Res Social Adm Pharm. 20. Holland R, Desborough J, Goodyer L, Hall S, Wright D, Loke YK. Does pharmacist-­ 2014;10(4):608–622. led medication review help to reduce hospital admissions and deaths in 7. Anderson K, Stowasser D, Freeman C, Scott I. Prescriber barriers and enablers to older people? A systematic review and meta-analysis. Br J Clin Pharmacol. minimising potentially inappropriate medications in adults: a systematic review 2008;65(3):303–316. and thematic synthesis. BMJ Open. 2014;4(12):e006544. 21. Thomas R, Huntley AL, Mann M, et al. Pharmacist-led interventions to reduce un- 8. Willeboordse F, Schellevis FG, Meulendijk MC, Hugtenburg JG, Elders PJM. Imple- planned admissions for older people: a systematic review and meta-analysis of mentation fidelity of a clinical medication review intervention: process evalua- randomised controlled trials. Age Ageing. 2014;43(2):174–187. tion. Int J Clin Pharm. 2018;40(3):550–565. 22. Hohl CM, Wickham ME, Sobolev B, et al. The effect of early in-hospital medi- 9. Mast R, Schouten G, van Woerkom M. Niveau van medicatiebeoordeling initiatieven cation review on health outcomes: a systematic review. Br J Clin Pharmacol. in Nederland kan beter. (Room for improvement in medication review initiatives in the 2015;80(1):51–61. Netherlands). Pharmaceutisch Weekblad. 2010;4(11/12):189–194. 23. Beuscart JB, Pont LG, Thevelin S, et al. A systematic review of the outcomes re- 10. Kwint HF, Bermingham L, Faber A, Gussekloo J, Bouvy ML. The relationship be- ported in trials of medication review in older patients: the need for a core out- tween the extent of collaboration of general practitioners and pharmacists and come set. Br J Clin Pharmacol. 2017;83(5):942–952. the implementation of recommendations arising from medication review: a sys- 24. Anderson K, Foster M, Freeman C, Luetsch K, Scott I. Negotiating “Unmeasur- tematic review. Drugs Aging. 2013;30(2):91–102. able Harm and Benefit”: Perspectives of General Practitioners and Consultant 7 11. Dees MK, Geijteman ECT, Dekkers WJM, et al. Perspectives of patients, close rel- Pharmacists on Deprescribing in the Primary Care Setting. Qual Health Res. atives, nurses, and physicians on end-of-life medication management. Palliat 2017;27(13):1936–1947. Support Care. 2017 [Epub ahead of print]. 25. Luetsch K, Rowett D. Developing interprofessional communication skills for phar- 12. Geijteman ECT, Huisman BAA, Dees MK, et al. Medication Discontinuation at the macists to improve their ability to collaborate with other professions. J Interprof End of Life: A Questionnaire Study on Physicians’ Experiences and Opinions. J Care. 2016;30(4):458–465. Palliat Med. 2018 [Epub ahead of print]. 26. Jansen J, Naganathan V, Carter SM, et al. Too much medicine in older people? De- 13. Agar MR, Lawlor PG, Quinn S, et al. Efficacy of Oral Risperidone, Haloperidol, or prescribing through shared decision making. BMJ. 2016;353:i2893. Placebo for Symptoms of Delirium Among Patients in Palliative Care: A Ran- 27. Reeve E, To J, Hendrix I, Shakib S, Roberts MS, Wiese MD. Patient barriers to and domized Clinical Trial. JAMA Intern Med. 2017;177(1):34–42. enablers of deprescribing: a systematic review. Drugs Aging. 2013;30(10):793–807.

146 147 Deprescribing in older people General introduction and thesis outline

1 CHAPTER 8

SUMMARY

SAMENVATTING

ACKNOWLEDGEMENTS — DANKWOORD

INTERNATIONAL PUBLCIATIONS AND PRESENTATIONS

148 149 Summary

SUMMARY

Older people aged 65 years and over use more medications than any other age group. [1] Three important factors complicate med- ication use in this population. First, multiple medication use in- creases their risk to experience adverse drug reactions. [2] Second, age-related changes in the body decrease an older person’s tol- erance to medications. [3] Third, scientific evidence on benefits and risks of medications in older people is often lacking. [4] Inappropriate prescribing in older people is common [5–8] and has been associated with increased adverse drug reactions, mor- bidity, hospitalisations and decreased quality of life. [9–14]

Deprescribing is the solution to reduce potentially inappropriate prescribing in older people. It has been defined as ‘the process of withdrawal of an inappropriate medication, supervised by a health care professional with the goal of managing polypharmacy and improving outcomes’, [15] Deprescribing is a complex healthcare intervention as it involves various stakeholders, such as doctors, pharmacists, patients and patients’ caregivers. Complex health- care interventions need to be developed and evaluated following best practice guidance before implemented in practice. This guid- ance includes identifying evidence and appropriate theory to de- velop the intervention, then to test the feasibility and perform an exploratory evaluation, before going on to a definitive evaluation followed by eventual implementation. [16]

The aim of this thesis was the development and evaluation of in- terventions for deprescribing in older people, by identifying op- portunities for deprescribing (chapter 2 and 3), evaluating a cur- rent deprescribing intervention (chapter 4 and 5) and developing 8 and evaluating a new deprescribing intervention (chapter 6).

In chapter 2 we explored changes in prescribing of preventive and symptomatic medication at the end of life in older nursing home residents (retrospective cohort study). In light of limited

151 Summary Summary

life expectancy toward the end of life, the use of medications to Based on the results of chapters 3, 4 and 5, we developed a new prevent future onset of disease or complications, which need a deprescribing intervention. This intervention targeted newly long time until benefit, might become less appropriate than prescribed (instead of only chronic) anticholinergic/sedative medications for symptom management, which have immediate medications. Using information technology, patients with a benefits. [17] However, the study showed that deprescribing of newly prescribed medication that increased their already existing preventive medications was limited. Small increases in the pre- anticholinergic/sedative load were detected. For these patients scribing of symptomatic medications indicated some awareness a medication review was performed. In chapter 6 we showed of changed need, but high prescribing of preventive medications that this intervention was feasible, acceptable and potentially suggested treatment goals were not being revised when life ex- effective (prospective intervention study). A considerable num- pectancy changed. ber of patients in need of medication optimisation was detected. Pharmacists and patients were satisfied with the intervention and In chapter 3 we explored high use of anticholinergic/sedative the time investment was reasonable. In one third of patients, an medications in the national population of community-dwelling increase in anticholinergic/sedative load was prevented. Results older adults (cross sectional study). Anticholinergic/sedative can be used to refine the deprescribing intervention, then to test medications are potentially inappropriate in older people as they it in a randomized controlled trial. have negative effects on cognitive and physical function and in- crease the risk of falls, dementia, hospitalisation and mortality. This thesis shows a robust approach of the development and eval- [18–20] Our results showed that a large proportion of older com- uation of deprescribing interventions, by performing a variety of munity-dwelling adults in the Netherlands had a high anticholin- studies investigating different stages in the process. All studies ergic/sedative medication load. According to their medication use, in this thesis focused on specific subpopulations of older people four distinct subpopulations were identified. Deprescribing inter- at high risk for medication related harm, investigated different ventions need to be targeted and tailored to these subpopulations. potentially inappropriate medications and used different study designs. Opportunities for deprescribing exist in older popula- In chapter 5 and 6 we evaluated whether pharmacist-led medica- tions, such as the high prescribing of preventive medications at tion review is effective at deprescribing anticholinergic/sedative the end of life in older nursing home residents and anticholiner- medications in older community-dwelling adults with a high gic/sedative medications in older community-dwelling adults. To chronic anticholinergic/sedative load (randomized controlled reduce a high anticholinergic/sedative load in community-dwell- trial). While effectiveness of pharmacist-led medication review ing older adults, medication reviews, as currently performed in for optimising medication was found in population groups, such the Netherlands, are not effective. An innovative deprescribing as users of cardiovascular medication [21] and nursing home intervention using information technology to target newly pre- residents, [22] results of this study showed no effect of pharma- scribed anticholinergic/sedative medication in this population 8 cist-led medication review on deprescribing chronically used an- seems more successful. Future deprescribing strategies should ticholinergic/sedative medications in our population. There is a be patient-centred, targeted to the right population, tailored need to go back to the development and feasibility/piloting stage to patient’s needs and have a high degree of interprofessional in order to target the intervention to this population. collaboration.

152 153 Summary Summary

REFERENCES 16. Moore GF, Audrey S, Barker M, et al. Process evaluation of complex interventions: Medical Research Council guidance. BMJ. 2015;350:h1258. 1. SIMPATHY. Polypharmacy Management by 2030: a patient safety challenge. 2017. 17. Holmes HM, Hayley DC, Alexander GC, Sachs GA. Reconsidering medication ap- http://www.simpathy.eu/sites/default/files/Managing_polypharmacy2030-web.pdf. propriateness for patients late in life. Arch Int Med. 2006;166(6):605–609. Accessed May 2018. 18. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- 2. Atkin PA, Veitch PC, Veitch EM, Ogle SJ. The epidemiology of serious adverse drug erties on cognitive function, delirium, physical function and mortality: a sys- reactions among the elderly. Drugs Aging. 1999;14(2):141–152. tematic review. Age Ageing. 2014;43(5):604–615. 3. Mangoni AA, Jackson SH. Age-related changes in pharmacokinetics and pharma- 19. Gray SL, Anderson ML, Dublin S, et al. Cumulative use of strong anticholiner- codynamics: basic principles and practical applications. Br J Clin Pharmacol. gics and incident dementia: a prospective cohort study. JAMA Intern Med. 2004;57(1):6–14. 2015;175(3):401–407. 4. McMurdo ME, Roberts H, Parker S, et al. Improving recruitment of older people to 20. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls research through good practice. Age Ageing. 2011;40(6):659–665. in older people: systematic review of publications from a recent 5-year period. 5. Tommelein E, Mehuys E, Petrovic M, Somers A, Colin P, Boussery K. Potentially in- Eur J Clin Pharmacol. 2015;71(12):1429–1440. appropriate prescribing in community-dwelling older people across Europe: a 21. Jokanovic N, Tan EC, Sudhakaran S, et al. Pharmacist-led medication review in systematic literature review. Eur J Clin Pharmacol. 2015;71(12):1415–1427. community settings: An overview of systematic reviews. Res Social Adm Pharm. 6. Opondo D, Eslami S, Visscher S, et al. Inappropriateness of medication prescrip- 2017;13(4):661–685. tions to elderly patients in the primary care setting: a systematic review. PLoS 22. Wouters H, Scheper J, Koning H, et al. Discontinuing inappropriate medication use One. 2012;7(8):e43617. in nursing home residents: A cluster randomized controlled trial. Ann Intern 7. Gallagher P, Barry P, O’Mahony D. Inappropriate prescribing in the elderly. J Clin Med. 2017;167(9):609–617. Pharm Ther. 2007;32(2):113–121. 8. Aparasu RR, Mort JR. Inappropriate prescribing for the elderly: beers criteria-based review. Ann Pharmacother. 2000;34(3):338–346. 9. Cahir C, Bennett K, Teljeur C, Fahey T. Potentially inappropriate prescribing and ad- verse health outcomes in community dwelling older patients. Br J Clin Pharma- col. 2014;77(1):201–210. 10. Lund BC, Carnahan RM, Egge JA, Chrischilles EA, Kaboli PJ. Inappropriate pre- scribing predicts adverse drug events in older adults. Ann Pharmacother. 2010;44(6):957–963. 11. Hamilton H, Gallagher P, Ryan C, Byrne S, O’Mahony D. Potentially inappropri- ate medications defined by STOPP criteria and the risk of adverse drug events in older hospitalized patients. Arch Intern Med. 2011;171(11):1013–1019. 12. Price SD, Holman CD, Sanfilippo FM, Emery JD. Association between potentially inappropriate medications from the Beers criteria and the risk of unplanned hospitalization in elderly patients. Ann Pharmacother. 2014;48(1):6–16. 13. Pasina L, Djade CD, Tettamanti M, et al. Prevalence of potentially inappropriate med- ications and risk of adverse clinical outcome in a cohort of hospitalized elderly pa- tients: results from the REPOSI Study. J Clin Pharm Ther. 2014;39(5):511–515. 8 14. Laroche ML, Charmes JP, Nouaille Y, Picard N, Merle L. Is inappropriate medica- tion use a major cause of adverse drug reactions in the elderly? Br J Clin Pharma- col. 2007;63(2):177–186. 15. Reeve E, Gnjidic D, Long J, Hilmer S. A systematic review of the emerging de fi ni- tion of ‘deprescribing’ with network analysis: implications for future research and clinical practice. Br J Clin Pharmacol. 2015;80(6):1254–1268.

154 155 Samenvatting

SAMENVATTING

Ouderen (mensen van 65 jaar en ouder) gebruiken meer genees- middelen dan andere leeftijdsgroepen. [1] Een drietal factoren compliceren het geneesmiddelgebruik bij ouderen. Ten eerste, meervoudig gebruik van geneesmiddelen door ouderen vergroot het risico op bijwerkingen. [2] Ten tweede, leeftijdsgebonden veranderingen in het lichaam van ouderen, zoals verminderde lever- en nierfunctie, verminderen de tolerantie voor genees- middelen. [3] Tot slot ontbreekt het vaak aan wetenschappelijk bewijs over de voor- en nadelen van geneesmiddelen bij oude- ren. [4] Het voorschrijven van risicovolle geneesmiddelen bij ouderen is veelvoorkomend [5–8] en wordt geassocieerd met een toename in bijwerkingen, ziekenhuisopname en verminderde kwaliteit van leven. [9–14]

Deprescribing is de oplossing om risicovol voorschrijven bij oude- ren te verminderen. Dit wordt gedefinieerd als het ‘verminderen van risicovolle geneesmiddelen onder toezicht van een zorgver- lener met als doel het beheer van polyfarmacie en het verbete- ren van patiëntgebonden uitkomsten’, [15]. Deprescribing is een complexe interventie in de gezondheidszorg omdat verschillende belanghebbenden hierbij betrokken zijn zoals artsen, apothekers, patiënten en mantelzorgers. Dergelijke interventies moeten zorg- vuldig ontwikkeld en geëvalueerd worden. Dit gebeurt in vier stappen. Eerst dient de noodzaak voor de interventie te worden aangetoond. Vervolgens dient de haalbaarheid getest te worden in een verkennende evaluatie, gevolgd door een definitieve evaluatie. Indien de interventie effectief is gebleken kan deze uiteindelijk geïmplementeerd worden in de praktijk. [16] 8 Het doel van dit proefschrift was de ontwikkeling en evaluatie van deprescribing interventies, door het aantonen van de noodzaak (hoofdstuk 2 en 3), het definitief evalueren van een bestaande de- prescribing interventie (hoofdstuk 4 en 5) en het ontwikkelen en evalueren van een nieuwe deprescribing interventie (hoofdstuk 6).

157 Samenvatting Samenvatting

In hoofdstuk 2 zijn veranderingen in het voorschrijven van pre- en verpleeghuisbewoners. [22] Wij vonden echter dat medicatie- ventieve en symptomatische geneesmiddelen aan het levenseinde beoordelingen geen effect hadden op het verminderen van chro- van oudere verpleeghuisbewoners onderzocht (retrospectieve nisch gebruikte anticholinerge/sederende geneesmiddelen bij cohortstudie). Preventieve geneesmiddelen, die gebruikt worden onze patiëntengroep. Deze resultaten toonden aan dat het nodig om ziekte of complicatie op de lange termijn te voorkomen zijn was om terug te gaan naar de ontwikkelfase om de interventie toe potentieel minder geschikt bij patiënten met een korte levens- te spitsen op deze patiëntengroep. verwachting dan geneesmiddelen voor symptoombestrijding, die direct werken. [17] Wij vonden echter dat er weinig deprescribing Op basis van de resultaten van hoofdstuk 3, 4 en 5 hebben wij een plaatsvond van preventieve geneesmiddelen. Een lichte toename nieuwe deprescribing interventie ontwikkeld. Deze interventie in het voorschrijven van symptomatische geneesmiddelen duidde richtte zich op nieuw voorgeschreven (in plaats van alleen chro- op enige bewustwording van veranderde behoefte van patiënten nische) anticholinerge/sederende geneesmiddelen. Met behulp aan het levenseinde. Het veelvoudig voorschrijven van preven- van informatietechnologie werden patiënten waarbij het nieuwe tieve geneesmiddelen suggereerde echter dat behandeldoelen geneesmiddel de reeds bestaande anticholinerge/sederende be- niet werden herzien wanneer de levensverwachting afnam. lasting verhoogde opgespoord. Vervolgens werd een medicatiebe- oordeling bij deze patiënten uitgevoerd. In hoofdstuk 6 toonden In hoofdstuk 3 onderzochten wij het veelvoudig gebruik van anti- wij aan dat deze interventie haalbaar, acceptabel en potentieel ef- cholinerge/sederende geneesmiddelen bij thuiswonende ouderen fectief was (prospectieve interventiestudie). Er werden voldoende in Nederland (cross-sectionele studie). Anticholinerge/sederende relevante patiënten opgespoord. Apothekers en patiënten waren geneesmiddelen hebben een negatief effect op de cognitieve en tevreden met de interventie en de tijdsinvestering was redelijk. fysieke functie en vergroten het risico op vallen, dementie, zie- Bij één derde van de patiënten kon een toename van de anticholi- kenhuisopname en overlijden. [18–20] Onze studie toonde aan nerge/sederende belasting voorkomen worden. De resultaten van dat een groot deel van de thuiswonende ouderen in Nederland deze studie kunnen worden gebruikt om de interventie te verfij- meerdere anticholinerge/sederende geneesmiddelen gebruikte. nen en vervolgens te testen in een definitieve evaluatie. Op basis van het geneesmiddelgebruik konden wij deze ouderen in vier klinische subgroepen verdelen. Interventies ter verminde- Dit proefschrift beschrijft een robuuste benadering van de ont- ring van anticholinerge/sederende geneesmiddelen moeten wor- wikkeling en evaluatie van deprescribing interventies. Alle stu- den toegespitst op elke van deze vier groepen. dies richtten zich op specifieke oudere patiëntengroepen met een hoog risico op geneesmiddel gerelateerde problemen, on- In hoofdstuk 5 en 6 onderzochten wij of medicatiebeoorde- derzochten verschillende potentiele risicogeneesmiddelen en lingen door de apotheker effectief zijn in het verminderen van gebruikten verschillende onderzoeksmethoden. De noodzaak anticholinerge/sederende geneesmiddelen bij thuiswondende voor deprescribing werd aangetoond voor preventieve genees- 8 ouderen met een chronische hoge anticholinerge/sederende middelen bij oudere verpleeghuisbewoners aan het levenseinde belasting (gerandomiseerde gecontroleerde interventiestudie). en anticholinerge/sederende geneesmiddelen bij thuiswonende Medicatiebeoordelingen zijn eerder effectief gebleken in het ouderen. Medicatiebeoordelingen door de apotheker waren niet verbeteren van geneesmiddelgebruik bij andere patiëntengroe- effectief in het verminderen van een hoge chronische anticholi- pen, zoals gebruikers van cardiovasculaire geneesmiddelen [21] nerge/sederende belasting bij thuiswonende ouderen. Een nieuwe

158 159 Samenvatting Samenvatting

deprescribing interventie die zich met behulp van informatie- REFERENTIES technologie richt op nieuw voorgeschreven anticholinerge/sede- 1. SIMPATHY. Polypharmacy Management by 2030: a patient safety challenge. 2017. rende geneesmiddelen in deze patiëntengroep leek succesvoller http://www.simpathy.eu/sites/default/files/Managing_polypharmacy2030-web. te zijn. Bij toekomstige deprescribing interventies moet de pati- pdf. Accessed May 2018. ënt centraal staan, de interventie gericht zijn op een specifieke 2. Atkin PA, Veitch PC, Veitch EM, Ogle SJ. The epidemiology of serious adverse drug patiënten- en/of geneesmiddelengroep en moet er een goede in- reactions among the elderly. Drugs Aging. 1999;14(2):141–152. terprofessionele samenwerking bestaan. 3. Mangoni AA, Jackson SH. Age-related changes in pharmacokinetics and pharma- codynamics: basic principles and practical applications. Br J Clin Pharmacol. 2004;57(1):6–14. 4. McMurdo ME, Roberts H, Parker S, et al. Improving recruitment of older people to research through good practice. Age Ageing. 2011;40(6):659–665. 5. Tommelein E, Mehuys E, Petrovic M, Somers A, Colin P, Boussery K. Potentially in- appropriate prescribing in community-dwelling older people across Europe: a systematic literature review. Eur J Clin Pharmacol. 2015;71(12):1415–1427. 6. Opondo D, Eslami S, Visscher S, et al. Inappropriateness of medication prescrip- tions to elderly patients in the primary care setting: a systematic review. PLoS One. 2012;7(8):e43617. 7. Gallagher P, Barry P, O’Mahony D. Inappropriate prescribing in the elderly. J Clin Pharm Ther. 2007;32(2):113–121. 8. Aparasu RR, Mort JR. Inappropriate prescribing for the elderly: beers criteria-based review. Ann Pharmacother. 2000;34(3):338–346. 9. Cahir C, Bennett K, Teljeur C, Fahey T. Potentially inappropriate prescribing and ad- verse health outcomes in community dwelling older patients. Br J Clin Pharma- col. 2014;77(1):201–210. 10. Lund BC, Carnahan RM, Egge JA, Chrischilles EA, Kaboli PJ. Inappropriate pre- scribing predicts adverse drug events in older adults. Ann Pharmacother. 2010;44(6):957–963. 11. Hamilton H, Gallagher P, Ryan C, Byrne S, O’Mahony D. Potentially inappropri- ate medications defined by STOPP criteria and the risk of adverse drug events in older hospitalized patients. Arch Intern Med. 2011;171(11):1013–1019. 12. Price SD, Holman CD, Sanfilippo FM, Emery JD. Association between potentially inappropriate medications from the Beers criteria and the risk of unplanned hospitalization in elderly patients. Ann Pharmacother. 2014;48(1):6–16. 13. Pasina L, Djade CD, Tettamanti M, et al. Prevalence of potentially inappropriate med- ications and risk of adverse clinical outcome in a cohort of hospitalized elderly pa- tients: results from the REPOSI Study. J Clin Pharm Ther. 2014;39(5):511–515. 8 14. Laroche ML, Charmes JP, Nouaille Y, Picard N, Merle L. Is inappropriate medica- tion use a major cause of adverse drug reactions in the elderly? Br J Clin Pharma- col. 2007;63(2):177–186. 15. Reeve E, Gnjidic D, Long J, Hilmer S. A systematic review of the emerging de fi ni- tion of ‘deprescribing’ with network analysis: implications for future research and clinical practice. Br J Clin Pharmacol. 2015;80(6):1254–1268.

160 161 Samenvatting Acknowledgements – Dankwoord

16. Moore GF, Audrey S, Barker M, et al. Process evaluation of complex interventions: ACKNOWLEDGEMENTS – DANKWOORD Medical Research Council guidance. BMJ. 2015;350:h1258. 17. Holmes HM, Hayley DC, Alexander GC, Sachs GA. Reconsidering medication ap- propriateness for patients late in life. Arch Int Med. 2006;166(6):605–609. Vijf jaar geleden stapte ik in een vliegtuig op weg naar Australië, 18. Fox C, Smith T, Maidment I, et al. Effect of medications with anti-cholinergic prop- niet wetende welke avonturen en mooie vriendschappen er op erties on cognitive function, delirium, physical function and mortality: a sys- tematic review. Age Ageing. 2014;43(5):604–615. mijn pad zouden komen, maar vooral niet dat het onderzoek 19. Gray SL, Anderson ML, Dublin S, et al. Cumulative use of strong anticholiner- dat ik daar zou gaan doen het begin zou zijn van een (soms on- gics and incident dementia: a prospective cohort study. JAMA Intern Med. stuimige) reis die heeft mogen resulteren in dit proefschrift. Ik 2015;175(3):401–407. kijk terug op een hele bijzondere periode, die niet mogelijk was 20. Park H, Satoh H, Miki A, Urushihara H, Sawada Y. Medications associated with falls geweest zonder een aantal personen die ik hieronder graag wil in older people: systematic review of publications from a recent 5-year period. Eur J Clin Pharmacol. 2015;71(12):1429–1440. bedanken. 21. Jokanovic N, Tan EC, Sudhakaran S, et al. Pharmacist-led medication review in community settings: An overview of systematic reviews. Res Social Adm Pharm. Prof. dr. Katja Taxis, beste Katja, we kennen elkaar al sinds het 2017;13(4):661–685. voorjaar van 2012, toen ik mijn Bacheloronderzoek onder begelei- 22. Wouters H, Scheper J, Koning H, et al. Discontinuing inappropriate medication use ding van jou en Willem van der Veen ging uitvoeren. Een jaar later in nursing home residents: A cluster randomized controlled trial. Ann Intern Med. 2017;167(9):609–617. mocht ik via jou naar Australië om mijn masteronderzoek bij dr. Lisa Pont uit te voeren en toen ik terug kwam stelde jij me aan als onderzoeker, student assistent, promovendus en sinds september 2018 als postdoc. Beste Katja, al die jaren kon ik mij geen betere begeleider wensen. Ik heb veel van jou geleerd en heb genoten van onze discussies tijdens de vele afspraken, of dat nou was in jouw kantoor, via de telefoon, Skype of op het terras in Den Haag. Jij gaf mij de vrijheid om projecten zelfstandig op te zetten en uit te voeren, maar wist altijd precies wat er speelde en ondersteunde mij waar nodig. Ik ben ontzettend dankbaar voor de kansen die jij mij hebt gegeven en hoop dat ook ons volgende project een succes wordt.

Dr. Lisa Pont, dear Lisa, ‘thank your for your great advice and guid- ance... your enthusiasm inspired me... your positive attitude and patience made working under your supervision a great pleasure...’ 8 This is what I wrote in April 2014 in the preface of my Masterthesis. It still is true and I am very happy that you became my second pro- motor. Thank you for your great ideas and quick thinking, which made our Skype meetings very lively, whether they were planned in the early morning or at midnight. Despite the distance I hope

162 163 Acknowledgements – Dankwoord Acknowledgements – Dankwoord

we stay in touch and have more great talks over coffee or Guinness Dr. Niesko Pras, dank voor jouw interesse in mijn onderzoeken en like we had in Dublin, Glasgow and Prague. de farmacotherapeutische adviezen. Jugoslav Pavlovic, veel dank voor jouw hulp bij het verwerken van de giga hoeveelheid aan Dr. Hans Wouters, beste Hans, ook jou leerde ik al vrij vroeg data en alle andere IT-ondersteuning. in mijn promotietraject kennen en ik ben heel blij dat jij mijn co-promotor bent geworden. Dank voor al jouw advies, met name Graag wil ik alle toenmalige studenten farmacie bedanken die op het gebied van statistiek heb ik veel van je geleerd, maar ook tijdens hun (onderzoek-) stage in de openbare apotheek gehol- voor je humor en relativeringsvermogen. We hebben samen veel pen hebben bij de uitvoering van de medicatiebeoordelingen. gelachen en dat maakte onze samenwerking erg prettig. Bedankt Marjolein en Floor, dank voor jullie hulp bij de dataverzameling. dat ik mee mocht werken aan jouw projecten. Ik vind het een eer Laura en Nienke, dank voor de inzet tijdens jullie masteronder- jouw eerste promovenda te mogen zijn. zoek en Geanne en Esther, dank voor de inzet tijdens jullie ba- chelorproject, ik vond het erg leuk om jullie te begeleiden. Graag wil ik de leden van de beoordelingscommissie prof. dr. P. Denig, prof. dr. M.L. Bouvy en prof. dr. R.H. Vander Stichele har- I thank all my colleagues at the department of Pharmacotherapy, telijk danken voor het lezen en beoordelen van mijn proefschrift. -epidemiology and -economy for the international lunches and interesting research meetings. In particular I want to thank my Alle apothekers en patiënten die deelgenomen hebben aan de on- former roommates, Doti and Aizati, thank you for your great com- derzoeken in dit proefschrift heel hartelijk dank voor uw bijdrage. pany and for tolerating the many phone calls I had to conduct to Zonder uw inzet waren de praktijkonderzoeken niet mogelijk pharmacists and patients. Aizati and Thang, it was an honour to geweest. be your paranymph. Jannie en Bert dank voor jullie ondersteu- ning vanuit het secretariaat. Dr. Martina Teichert, beste Martina, hartelijk dank voor jouw enthousiasme en betrokkenheid bij verschillende studies in dit The PhD Day 2018 crew, thank you for a great year with many proefschrift. Jouw inspanningen bij de KNMP waren cruciaal bij meetings, dinners and drinks, all for the sake of organizing a ca- het opzetten van deze studies. Dank ook voor jouw kritische blik reer event for 900 PhDs/postdocs. It was awesome! op mijn artikelen. Lieve paranimfen, Karlien en Linda, dank voor jullie steun en hulp Fabienne Griens en Ton Schalk, veel dank voor jullie inspannin- de afgelopen jaren. Ik voel me vereerd dat jullie op 2 November gen vanuit de SFK. De data die jullie hebben aangeleverd en de naast mij staan! Lieve Karlien, wij zijn vriendinnen sinds wij el- ontwikkelde webrapportage waren essentieel voor het uitvoeren kaar op het vliegveld van Abu Dhabi onderweg naar Sydney heb- van twee studies in dit proefschrift. Fabienne, veel dank ook voor ben ontmoet. We hebben prachtige avonturen samen beleefd in 8 de vele telefonische overlegmomenten. Australië en Nieuw Zeeland en deze zonder moeite bij terugkomst in Nederland voortgezet. Over onderzoek doen konden wij van Dr. Caroline van de Steeg veel dank voor de inbedding van mijn alles delen en jij was dan ook vaak mijn steun als het even te- studie in de opleiding tot openbaar apotheker specialist en de or- genzat. Dank je wel voor alle gezellige momenten, het lachen en ganisatie die daarbij kwam kijken. huilen, het luisteren en meeleven. Lieve Linda, dank je wel voor

164 165 Acknowledgements – Dankwoord Acknowledgements – Dankwoord

de gezelligheid op kantoor en onze overleggen over onderzoeks- Lieve Martijn, toen wij elkaar drie jaar geleden ontmoetten veran- resultaten, het uitvoeren van praktijkonderzoek en de ontwikke- derde mijn leven op slag. Naast het feit dat ik ineens ieder week- ling van farmaceutische patiëntenzorg. Ik vind het jammer dat we end in Den Haag was (en vervolgens met de tijd steeds een dagje elkaar sinds mijn verhuizing minder zien, maar ik hoop dat we langer tot we helemaal gingen samenwonen) vond ik een gevoel blijven bellen en zo onze discussies voortzetten. Ik ben heel be- van rust en geluk bij jou. Dank je wel voor alle mooie momenten nieuwd naar hoe jouw proefschrift er over een jaar uitziet! die we samen al hebben beleefd, jouw liefde en onvermoeibare steun. Zonder jou was dit proefschrift er misschien niet gekomen, Al mijn andere lieve vrienden en vriendinnen, studiegenootjes, het is daarom ook een beetje van jou. commissiegenootjes, clubgenootjes en oud-huisgenootjes be- dankt voor jullie gezelligheid de afgelopen jaren.

Lieve schoonfamilie, bedankt dat jullie mij met open armen heb- ben ontvangen. Ik vind het altijd ontzettend gezellig met jullie. Margreet en Wouter, jullie promotie heeft mij geïnspireerd en ik hoop dat mijn verdediging ook net zo goed zal verlopen als die van jullie.

Lieve Anna, mijn jongere zus, ik vind het heel leuk dat wij onze passie voor de zorg kunnen delen, jij als (bijna) arts en ik als apo- theker. Wat was het leuk dat we een tijdje samen in Groningen studeerden. Dank je wel voor alle keren dat ik bij je mocht blijven logeren toen ik al in Den Haag woonde.

Lieve Siebrand, mijn jongere broer, als enige van het gezin ben jij niet in de zorg werkzaam maar heb jij voor de bouw en de studie architectuur gekozen. Ik vind het ontzettend stoer dat jij met een paar vrienden een huis kan bouwen in een paar weken tijd. Je doet het ontzettend goed met je studie en ik ben heel blij dat je het in Oldenburg zo naar je zin hebt.

Lieve papa en mama, mijn passie voor de zorg komt door jullie. Ik 8 geniet van onze discussies over patiëntenzorg en jullie kijk daarop inspireert mij. De afgelopen jaren zijn jullie mijn grote steun ge- weest, of het nou gaat om de zoveelste verhuizing of de kaft van dit boekje. Overal weten jullie wel raad mee. Dank je wel dat jullie er altijd voor mij zijn.

166 167 International publications and presentations

INTERNATIONAL PUBLICATIONS AND PRESENTATIONS

Within this thesis van der Meer HG, Wouters H, Pont LG, Taxis K. Reducing the anticholinergic and sedative load in older patients on polyphar- macy by pharmacist-led medication review: A randomised con- trolled trial. BMJ Open. 2018;8(7):e019042. • Oral presentation at the 32nd International Conference on Pharmacoepidemiology & Therapeutic Risk Management, 25–28 August 2016, Dublin, Ireland.

van der Meer HG, Taxis K, Pont LG. Changes in Prescribing Symptomatic and Preventive Medications in the Last Year of Life in Older Nursing Home Residents. Front Pharmacol. 2018;8:990. • Oral presentation at the European Drug Utilisation Research Group — Conference, 27–29 August 2014, Groningen, the Netherlands. • Oral presentation at the 31st International Conference on Pharmacoepidemiology & Therapeutic Risk Management, 24–26 August 2015, Boston, United States of America.

van der Meer HG, Wouters H, van Hulten R, Pras N, Taxis K. Decreasing the load? Is a Multidisciplinary Multistep Medication Review in older people an effective intervention to reduce a pa- tient’s Drug Burden Index? Protocol of a randomised controlled trial. BMJ Open. 2015;5(12):e009213.

van der Meer HG, Wouters H, Teichert M, Griens AMGF, Pavlovic J, Pont LG, Taxis K. Feasibility, acceptability and poten- tial effectiveness of an information technology based, pharma- cist-led intervention to prevent an increase in anticholinergic 8 and sedative load among older community-dwelling individuals. Ther Adv Drug Saf. In press. • Poster presentation at the the 34th International Conference on Pharmacoepidemiology & Therapeutic Risk Management, 22–26 August 2018 in Prague, Czech Republic.

169 International publications and presentations

van der Meer HG, Taxis K, Teichert M, Griens AMGF, Pont LG, Wouters H. Anticholinergic and sedative medication use in older community-dwelling people: a national population study in The Netherlands. Submitted. • Oral presentation at the 34th International Conference on Pharmacoepidemiology & Therapeutic Risk Management, 22–26 August 2018 in Prague, Czech Republic. • Oral presentation at the European Drug Utilisation Research Group — Conference, 15–17 November 2017, Glasgow, Scotland.

Not within this thesis Wouters H, Scheper J, Koning H, Brouwer C, Twisk, J, van der Meer H, Boersma F, Zuidema S, Taxis K. Discontinuing inappro- priate medication use in nursing home residents: A cluster ran- domized controlled trial. Ann Intern Med. 2017;167(9):609–617.

Wouters H, van der Meer H, Taxis K. Quantification of anticho- linergic and sedative drug load with the Drug Burden Index: a re- view of outcomes and methodological quality of studies. Eur J Clin Pharmacol. 2017;73(3):257–266.

170 UITNODIGING DEPRESCRIBING IN OLDER PEOPLE DEPRESCRIBING Voor het bijwonen van de openbare verdediging van het proefschrift:

DEPRESCRIBING IN OLDER PEOPLE DEPRESCRIBING Development and evaluation of complex IN OLDER PEOPLE healthcare interventions

Helene Grietje (Heleen) van der Meer was born on Development and evaluation of complex door healthcare interventions 5 August 1990 in Papenburg, Germany, to Wytze Jan HELEEN VAN DER MEER van der Meer, dentist, and Klaaske van der Meer- Jansen, cardiac care nurse. She grew up in Germany together with her younger sister and brother. In 2009 Heleen van der Meer op vrijdag she obtained her Abitur (final exam) at the Gymnasium 2 november om 16.15 Papenburg and started her studies in pharmacy at the in het Academiegebouw University of Groningen. van de Rijkuniversiteit Groningen, Broerstraat 5 Heleen first became acquainted with research in the te Groningen. Heleen van der Meer Heleen van field of pharmacotherapy during her Bachelors studies. The foundation for her doctoral thesis was laid during Na afloop bent u van harte uitgenodigd the project she undertook in Sydney, Australia under voor de receptie supervision of Dr. Lisa Pont and Prof. Dr. Katja Taxis for in het Academiegebouw. her Masters in Pharmacy in 2013/14. On her return to the Netherlands, she accepted a temporary appoint- ment as a researcher with Prof. Taxis and in the same Heleen van der Meer year gave her first podium presentation at an inter- Helmersstraat 36 national scientific conference in Boston, US. She was 2513 RZ Den Haag awarded her Masters in Pharmacy in 2016 and com- 0648897302 pleted her PhD in 2018. Heleen lives in The Hague and [email protected] works as a postdoctoral researcher under supervision of Prof. Taxis on the development and implementation of patient material for deprescribing in older people. PARANIMFEN In addition to her studies and PhD research, Heleen has Karlien Sambell been active within various committees. For example, [email protected] in 2016/17 she organized the PhD Day, a career event for 900 PhD students/postdocs. Furthermore she loves Linda van Eikenhorst tennis and sailing and she has a passion for traveling. [email protected]