a danish nationwide register-based cohort study

University of Southern Denmark

Association between number of medications and mortality in geriatric inpatients

a Danish nationwide register-based cohort studyBrockhattingen, Kristoffer Kittelmann; Anru, Pavithra Laxsen; Masud, Tahir; Petrovic, Mirko;Ryg, Jesper

Published in:European Geriatric Medicine

DOI:10.1007/s41999-020-00390-3

Publication date:2020

Document version:Accepted manuscript

Citation for pulished version (APA):Brockhattingen, K. K., Anru, P. L., Masud, T., Petrovic, M., & Ryg, J. (2020). Association between number ofmedications and mortality in geriatric inpatients: a Danish nationwide register-based cohort study. EuropeanGeriatric Medicine, 11(6), 1063-1071. https://doi.org/10.1007/s41999-020-00390-3

Go to publication entry in University of Southern Denmark's Research Portal

Terms of useThis work is brought to you by the University of Southern Denmark.Unless otherwise specified it has been shared according to the terms for self-archiving.If no other license is stated, these terms apply:

• You may download this work for personal use only. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying this open access versionIf you believe that this document breaches copyright please contact us providing details and we will investigate your claim.Please direct all enquiries to [email protected]

Download date: 27. May. 2022

https://doi.org/10.1007/s41999-020-00390-3

https://doi.org/10.1007/s41999-020-00390-3

https://portal.findresearcher.sdu.dk/en/publications/b17a6dd5-1492-4d29-b95a-7cf91643a84b

Association between number of medications and mortality in geriatric inpatients – a Danish nationwide register-based cohort study

1

Target journal: European Geriatric Medicine

Title: Association between number of medications and mortality in geriatric inpatients – a Danish nationwide register-based cohort study Author: Kristoffer Kittelmann Brockhattingen1, 2*(orcid.org/0000-0002-3621-5463), Pavithra Laxsen Anru3,4(orcid.org/0000-0002-7046-2556), Tahir Masud2,5,6(orcid.org/0000-0003-1061-2898), Mirko Petrovic7(orcid.org/0000-0002-7506-8646), Jesper Ryg2,6(orcid.org/0000-0002-8641-3062)

Affiliations:

1Department of Geriatrics, Odense University Hospital – Svendborg Hospital, Denmark; 2Geriatric Research Unit, Department of Clinical Research, University of Southern Denmark, Odense, Denmark; 3Research Unit of Clinical Epidemiology, Department of Clinical Research, University of Southern Denmark, Odense, Denmark; 4Center for Clinical Epidemiology, Odense University Hospital, Odense, Denmark; 5Geriatric Department, Nottingham University Hospital, Nottingham, UK; 6Department of Geriatric Medicine, Odense University Hospital, Odense, Denmark; 7Deaprtment of Geriatrics, Ghent University Hospital, Ghent, Belgium

*Corresponding author:

Kristoffer Kittelmann Brockhattingen, MD

Baagøes Alle 31

5700 Svendborg

Denmark

E-mail: [email protected]

Phone: +45 30 89 66 36

Word count: 2,380

Abstract word count: 247

Keywords: Polypharmacy, comorbidity, mortality, geriatric, Barthel, ADL


2

Abstract

Purpose: To explore the association between the number of medications and mortality in geriatric inpatients taking

activities of daily living and comorbidities into account.

Methods: A nationwide population-based cohort study was performed including all patients aged ≥65 years admitted to

geriatric departments in Denmark during 2005-2014. The outcome of interest was mortality. Activities of daily living

using Barthel-Index (BI) were measured at admission. National health registers were used to link data on an individual

level extracting data on medications, and hospital diseases. Patients were followed to the end of study (31.12.2015),

death, or emigration, which ever occurred first. Kaplan-Meier survival curves were used to estimate crude survival

proportions. Univariable and multivariable analyses were performed using Cox regression. The multivariable analysis

were adjusted for age, marital status, period of hospital admission, BMI, and BI (model 1), and additionally either

number of diseases (model 2) or Charlson comorbidity index (model 3).

Results: We included 74,603 patients (62.8% women), with a median age of 83 (interquartile range [IQR] 77-88) years.

Patients used a median of 6 (IQR 4-9) medications. Increasing number of medications was associated with increased

overall, 30-days, and 1-year mortality in all 3 multivariable models for both men and women. For each extra medication

the mortality increased by 3% in women and 4% in men in the fully adjusted model.

Conclusion: Increasing number of medications was associated with mortality in this nationwide cohort of geriatric

inpatients. Our findings highlight the importance of polypharmacy in older patients with comorbidities.


3

Key summary points

Aim: To explore the association between number of medications and mortality in geriatric inpatients when adjusted for

diseases and activities of daily living.

Findings: Increasing number of medications is associated with increased mortality. Every increase in number of

medications by one is associated with a 3% increase in overall mortality.

Message: Evaluation of polypharmacy is important part of geriatric assessment when older adults are hospitalized.


4

Introduction

Geriatric inpatients are characterized and challenged by the presence of multimorbidity, polypharmacy, and often

functional decline [1-3]. Between multiple factors associated with mortality among older inpatients [4-6] it is

acknowledged that one of the most important is activities of daily living (ADL) [7]. ADL can e.g. be measured by the

Barthel Index, which focuses on the patient's level of dependency in basic ADL. Strong correlations between Barthel

Index and mortality in geriatric patients have previously been reported [8, 9]. Multi-morbidity and chronic diseases in

general are also associated with mortality [10, 11], particularly the Charlson comorbidity Index which focuses on

certain severe diseases [12, 13].

Recent studies have highlighted the importance of considering polypharmacy as an independent risk of

hospitalization and mortality because of adverse drug reactions and drug-drug interactions [14, 15]. Some studies have

tried to distinguish polypharmacy from multi-morbidity when assessing the independent influence on mortality [16, 17].

Although a few of these studies provided evidence that polypharmacy is associated with increased mortality [18, 19],

only a limited number of them took functional decline into account when assessing the association between

polypharmacy and overall mortality [18, 20, 21].

This study aims to explore the association between the number of medications and mortality in older

geriatric inpatients when adjusted for diseases and activities of daily living.

Methods

This was a nationwide register-based cohort study in geriatric inpatients with all-cause mortality as the main outcome.

In Denmark, each citizen is given a unique ten-digit social security number at birth or upon immigrating to the country.

This number serves as a link to all information (e.g. marital status, migration, death etc.) about the individual [22]. We

used the identification number to link data on an individual level from the following four Danish health registers: The

Danish National Database of Geriatrics [23], The Danish National Patient Register [24], The Danish Civil Registry

System [25], and the Danish National Database of Reimbursed Prescriptions [26]. Setting and participants have

previously been described in detail elsewhere [8]. In short, the study population was identified through the Danish

National Database of Geriatrics and included patients 65 years or older with their first registration in the database

during 2005-2014. The inclusion date was defined as the first date of admission to a geriatric department registered in

the database during the study period. Patients were followed from the individual index date and until death,


5

immigration, or end of study on December 31st 2015, allowing up to 10 years of follow-up, depending on the time of

admission.

Data on prescription medicine purchased 120 days prior to index date were extracted from the Danish

National Database of Reimbursed Prescriptions. A cut-off of 120 days was chosen since the majority of medications for

long-term treatment are administered in 100-pill packages in Denmark [27]. Anatomical therapeutic chemical (ATC)

codes were counted at the fourth level to classify and quantify medications and numbers of prescribed medications were

grouped into four categories: 0, 1-4, 5-9, or >10.

Information regarding any prior hospitalizations, dates of admission and discharge, and diagnoses

where collected from the Danish National Patient Register. The total number of diseases was calculated by using the

ICD-10 hospital discharge- and outpatient diagnoses 10 years prior to the index dates. The ICD-10 diagnosis was

discriminated at the third level of the ICD-10 code. To avoid excessive scores for disease, the same diagnoses only

counted once. Diagnoses were grouped in four categories 0, 1-9, 10-19 or >20 points. Charlson Comorbidity Index

(CCI) focuses on 19 chronic diseases and is well known to predict mortality. Points are graded by the severity of

number and severity of the disease [13]. It has, since its first validation, been adjusted several times in accordance to the

ICD-9 and ICD-10 [28]. The score is often used in geriatric research for examining comorbidity [29]. We calculated

CCI using ICD-10 codes from 10 years prior to individual index admission.

Variables such as height, weight, and the assessment of basic ADL were extracted from the Danish

National Database of Geriatrics. The body mass index (BMI) was calculated as weight in kilograms divided by the

height in meters squared and divided into categories according to the WHO definition. In Denmark, assessment of basic

ADL using Barthel Index is performed upon admission to the geriatric departments [8, 9]. The Barthel Index score used

in the database is the modified version introduced by Shah et al.: the Barthel Index-100 [30, 31]. This gathers

information across ten areas of basic ADL and summarizes it in a total score. The highest score indicates a high level of

independence, whereas a low score indicates a low level of independence The Barthel Index scores were grouped into

the four standard categories used in Denmark: 0-24, 25-49, 50-79, and 80-100 [8].

Finally, data on death, migration, and marital status were extracted from the Danish Civil Registry

System.

Statistics


6

Descriptive statistics are reported as mean with corresponding standard deviation (SD) or median with corresponding

inter quartile range (IQR) (25–75% percentile) as appropriate. Kaplan-Meier survival curves [32] were used to estimate

crude survival proportions according to each of the number of medications subcategories.

Univariable and multivariable analyses were performed using Cox regression [33]. The multivariable

analysis included adjustment for age, marital status, time period of hospitalization, BMI, and Barthel-Index (model 1),

and further adding either number of diseases (model 2) or CCI (model 3). All variables except marital status were

included as continuous variables in the models. Hazard ratios (HR) and corresponding 95% confidence interval (CI)

were performed for 30-days, 1-year, and overall mortality.

Furthermore, univariable and multivariable HRs and corresponding 95% CI were performed for overall mortality

according to each increase of one medication. Wald statistics was used to test for significance of the categorical

variables in the multivariable Cox regression model. The fully adjusted models were conducted as complete case

analysis. Analyses were stratified by sex at birth according to the health registers. P-values <0.05 indicated statistical

significance. Analyses were performed by a biostatistician using STATA (version 14.2; StataCorp LP, College Station,

TX, USA).

Ethics

The Danish data protection agency approved the study (2012580018, J.nr. 16/23359). Approval by ethical committee

and informed consent was not necessary according to the Danish law on medical ethics due to the register-based study

design.

Results

A total of 74,603 patients (46,823 women and 27,780 men) with a median age of 83 (interquartile range [IQR] 77-88)

years were included in the study. Follow-up ranged from 1 day-11 years and 51,197 patients died while none were lost

to follow-up corresponding to 192,012 person-years. Median (IQR) survival for the total cohort was 2.66 (0.64-5.65)

years.

The baseline demographics of the study population are presented in Table 1. The median (IQR) number of medications

was 6 (4-9), the median (IQR) number of diseases was 12 (7-18), and the median (IQR) CCI was 2 (1-3). Most patients

had low- to moderate reduced ADL with a median (IQR) Barthel-Index of 54 (29-77).

Survival decreased with increasing number of medications (Figure 1) as well as increasing number of

diseases (Supplementary Figure 1) and disease severity as expressed by increasing CCI (Supplementary Figure 2).


7

In the univariable analysis, increasing number of medications was associated with an increase in

overall, 30-days, and 1-year mortality (Table 2). The association was higher in men compared to women (p<0.0001).

The association remained significant in all three models in the multivariable analysis (Table 2).

Using number of medications as a continuous variable and “no medications” as the reference, the

overall risk of dying (HR (95%CI)) during follow-up in the total cohort increased by 3.3% (3.0-3.5) for each extra

medication in the univariable analyses (Table 3). In the multivariable model adjusting for age, Barthel index, marital

status, period of hospitalized admission, and BMI each extra medication was associated with an increased risk of 5.0%

(4.7-5.3) (model 1). When adding either the number of diseases or CCI to the multivariable model each extra

medication was associated with an increased risk of 3.8% (3.5-4.2) (model 2) and 2.7% (2.4-3.0) (model 3),

respectively. The risk was significantly higher in men than women (p<0.0001). In a sensitivity analysis patients with

very high number of medications (20+) were excluded. This did not change the estimates (data not shown).

Discussion

In the present study, we found that increasing number of medications is associated with 30-day, 1-year, and overall all-

cause mortality in geriatric inpatients. To our knowledge, this is the largest study assessing the association between

number of medications and mortality when adjusting for ADL and comorbidities.

The present study has several strengths. We were able to combine data from the nationwide population-

based Danish health registers with data on all geriatric inpatients during 10 years allowing for longitudinal cohort

design with no patients lost to follow-up. Due to the design, the study had no issue with recall bias when addressing

prior diseases, use of medication, or the objective measurements of ADL. All this strengthens the validity of the present

study.

However, our study also has some limitations. The study did not account for “over-the-counter” medications or the

patient's adherence to their medical treatments. We had no way to address this further and this could have under- or

overestimated the demonstrated association between number of medications and mortality. Also, the number of diseases

and CCI was calculated using hospital ICD-10 diagnosis. We had no data on ICPC diagnoses [34], which are used by

the GPs in Denmark as a diagnosis code system for the primary healthcare sector. Using hospital ICD-10 diagnoses

only to define diseases could have underestimated the actual number of diseases among the studied population. And in

this way, we cannot rule out that some of the medications are a marker of non-recorded diseases. Also, other important

risk factors of mortality might have been present at time of index admission such as delirium. CCI is known to be


8

associated with mortality and a prior study has shown high validity when using the Danish National Registers to

calculate CCI [35]. We did adjust for number of known diseases in our model 2 and disease severity by using CCI in

model 3. However, the present dataset holds no information on actual cognitive status at admission and can therefore

not be address further. Finally, our study along with other similar studies, shares the risk of confounding by indication

giving the difficulty understanding whether medications are associated with increased mortality or a proxy for the

number and severity of diseases. Our approach by using different adjustment models to explore the impact of

medications showed a persistent association with increasing number of medications. Despite this, we cannot rule out

that residual effect might still be present.

Other studies have also reported an association between increasing number of medications and

mortality [17-20]. Most of these studies used the CCI as a measure of disease burden [18, 20, 21]. That is, most of the

current literature used a weighted score to examine the association between disease burden and mortality [5]. Only few

studies used the accumulated unweighted number of diseases, i.e. diagnosis count [36, 37]. In a recent systematic

review, it was concluded that the Geriatric Index of Comorbidity had a better predictive capacity on mortality when

compared to other comorbidity scores such as CCI and diagnosis count [5]. On the other hand, another study showed

that diagnosis count was a better predictor than the Geriatric Index of Comorbidity [37], although, this might be due to a

lesser disease complexity and younger mean age among the studied population [37, 38]. In our study, we included

adjustment for either overall disease burden (diseases count) or disease severity (CCI) in the multivariable models and

showed that use of medications still was associated with mortality.

To our knowledge, three other studies have taken ADL in to account when assessing the role of

polypharmacy on mortality [18, 20, 21]. One study [18] used the SF-36 physical component score [39] and found that

both over-and-under medication was harmful to the patients. Whereas the two others [20, 21] used the Katz index [40].

Wimmer et al. found that medication regime was a better predictor of all-cause mortality then polypharmacy [20]. Iwata

et al. found that polypharmacy, defined as >6 medications, was a significant predictor of post-discharge mortality after

1 year [21]. The two studies differ in their use of the Katz index whereas Wimmer et al. uses all six basic activities, and

Iwata et al. uses five. It is unclear whether the evaluation of ADL was performed via interview/self-reporting or by

healthcare workers. SF-36 is reported by the patients themselves and therefore subject to recall bias. The Katz index

covers six areas of ADL, whereas Barthel index as used in our study is more comprehensive covering ten areas of ADL.


9

The present study uses ATC classification as a tool to measure the number of medications whereas the

three previously mentioned studies differ on how they obtained data on medication (self-reporting and medical records).

Although, Beer et al. mentioned that the medication used by the participants in their studies were ATC classified, it is

unclear whether they used ATC as a tool to measure the number of medications. Interestingly, Iwata et al. examined

different drugs and their association with 1-year mortality and found that benzodiazepines and NSAIDs where

associated with an increased risk in mortality. Our study did not examine different types of drugs and their association

with mortality. We did examine the burden of one added drug and the risk of mortality. This was not assessed by Iwata

et al. or the two other studies.

Other studies assessing medication and mortality have focused on specific cut-off points for polypharmacy (≥5

medications) or excessive polypharmacy (≥10 medications) [41, 42]. Our study also assessed the usual cutpoints and

found an association both with short and longer-term survival in the multivariable analysis. In this way, we were able to

show that burden of medication has both an acute and more prolonged negative effect. In the present study we also

addressed polypharmacy as a continuous variable and found that each medication added to the total burden. The overall

mortality increased by 3% per medication in the fully adjusted model and in this way was associated with an increased

risk on top of the risk of quantitative number of diseases or severe comorbidities. To our knowledge, this association

has not previously been reported. One explanation of these findings is that drug burden may be a better proxy for

physiological age than diseases. Another is that the results could question whether the geriatric patients in this cohort

were subject to overtreatment. Whatever the reason our finding highlights the importance of evaluating the medications

prescribed to the geriatric patient when admitted to hospital. The results of the current study also emphasize the

importance of ensuring continuously assessment/monitoring of the medication burden put upon many older patients.

More in-depth research examining the potential role of deprescribing and whether this could affect all-cause survival is

warranted and we recommend that future studies should focus on this area.

Conclusion

Increasing number of medications was associated with 30-days, 1-year, and overall mortality in this nationwide cohort

of geriatric inpatients. For each extra medication the mortality increased 3% in women and 4% in men in the fully

adjusted model including ADL and diseases. Our findings highlight the importance of the burden of polypharmacy on

older patients with comorbidities. Future studies are needed to address the potential effect of deprescribing on survival

in these patients.


10

Disclosure and Funding

All authors have completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and declare no personal

conflict of interest regarding the present study. The study was supported by the Program for Clinical Research

Infrastructure (PROCRIN) established by the Lundbeck Foundation and the Novo Nordisk Foundation. The funders were

not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data;

preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data sharing statement

According to the Danish Law on personal data, we are not allowed to make the dataset publicly available. Access to

data from the Danish Health Data Authority requires association with a Danish research unit and approval from the

Danish Data Protection Agency (www.datatilsynet.dk/english/) and the National Health Service Register (www.

sundhedsdatastyrelsen.dk/da/forskerservice (in Danish)).


11

References

1. Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. Epidemiology of multimorbidity and implications

for health care, research, and medical education: a cross-sectional study. Lancet (London, England).

2012;380(9836):37-43. doi:10.1016/s0140-6736(12)60240-2.

2. Cooper R, Kuh D, Cooper C, Gale CR, Lawlor DA, Matthews F et al. Objective measures of physical capability and

subsequent health: a systematic review. Age and ageing. 2011;40(1):14-23. doi:10.1093/ageing/afq117.

3. Devi J. The scales of functional assessment of Activities of Daily Living in geriatrics. Age and ageing.

2018;47(4):500-2. doi:10.1093/ageing/afy050.

4. Fried TR, O'Leary J, Towle V, Goldstein MK, Trentalange M, Martin DK. Health outcomes associated with

polypharmacy in community-dwelling older adults: a systematic review. Journal of the American Geriatrics Society.

2014;62(12):2261-72. doi:10.1111/jgs.13153.

5. Soh CH, Ul Hassan SW, Sacre J, Maier AB. Morbidity measures predicting mortality in inpatients: a systematic

review. Journal of the American Medical Directors Association. 2020;21(4):462-8.e7. doi:10.1016/j.jamda.2019.12.001.

6. Rizzuto D, Melis RJF, Angleman S, Qiu C, Marengoni A. Effect of chronic diseases and multimorbidity on survival

and functioning in elderly adults. Journal of the American Geriatrics Society. 2017;65(5):1056-60.

doi:10.1111/jgs.14868.

7. Thinggaard M, McGue M, Jeune B, Osler M, Vaupel JW, Christensen K. Survival Prognosis in Very Old Adults.

Journal of the American Geriatrics Society. 2016;64(1):81-8. doi:10.1111/jgs.13838.

8. Ryg J, Engberg H, Mariadas P, Pedersen SGH, Jorgensen MG, Vinding KL et al. Barthel Index at hospital admission

is associated with mortality in geriatric patients: a Danish nationwide population-based cohort study. Clinical

epidemiology. 2018;10:1789-800. doi:10.2147/clep.s176035.

9. Mahoney FI, Barthel DW. Functional evaluation: the Barthel index. Maryland state medical journal. 1965;14:61-5.

10. Johnston MC, Black C, Mercer SW, Prescott GJ, Crilly MA. Prevalence of secondary care multimorbidity in mid-

life and its association with premature mortality in a large longitudinal cohort study. BMJ open. 2020;10(5):e033622.

doi:10.1136/bmjopen-2019-033622.

11. Storeng SH, Vinjerui KH, Sund ER, Krokstad S. Associations between complex multimorbidity, activities of daily

living and mortality among older Norwegians. A prospective cohort study: the HUNT Study, Norway. BMC geriatrics.

2020;20(1):21. doi:10.1186/s12877-020-1425-3.


12

12. Quinones AR, Markwardt S, Botoseneanu A. Multimorbidity combinations and disability in older adults. The

journals of gerontology Series A, Biological sciences and medical sciences. 2016;71(6):823-30.

doi:10.1093/gerona/glw035.

13. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in

longitudinal studies: development and validation. Journal of chronic diseases. 1987;40(5):373-83. doi:10.1016/0021-

9681(87)90171-8.

14. Muhlack DC, Hoppe LK, Weberpals J, Brenner H, Schottker B. The association of potentially inappropriate

medication at older age with cardiovascular events and overall mortality: a systematic review and meta-analysis of

cohort studies. Journal of the American Medical Directors Association. 2017;18(3):211-20.

doi:10.1016/j.jamda.2016.11.025.

15. Gnjidic D, Hilmer SN, Blyth FM, Naganathan V, Waite L, Seibel MJ et al. Polypharmacy cutoff and outcomes: five

or more medicines were used to identify community-dwelling older men at risk of different adverse outcomes. Journal

of clinical epidemiology. 2012;65(9):989-95. doi:10.1016/j.jclinepi.2012.02.018.

16. Wastesson JW, Morin L, Tan ECK, Johnell K. An update on the clinical consequences of polypharmacy in older

adults: a narrative review. Expert Opin Drug Saf. 2018;17(12):1185-96. doi:10.1080/14740338.2018.1546841.

17. Leelakanok N, Holcombe AL, Lund BC, Gu X, Schweizer ML. Association between polypharmacy and death: A

systematic review and meta-analysis. Journal of the American Pharmacists Association : JAPhA. 2017;57(6):729-

38.e10. doi:10.1016/j.japh.2017.06.002.

18. Beer C, Hyde Z, Almeida OP, Norman P, Hankey GJ, Yeap BB et al. Quality use of medicines and health outcomes

among a cohort of community dwelling older men: an observational study. British Journal of Clinical Pharmacology.

2011;71(4):592-9. doi:10.1111/j.1365-2125.2010.03875.x.

19. Sganga F, Landi F, Ruggiero C, Corsonello A, Vetrano DL, Lattanzio F et al. Polypharmacy and health outcomes

among older adults discharged from hospital: Results from the CRIME study. Geriatrics & gerontology international.

2015;15(2):141-6. doi:10.1111/ggi.12241.

20. Wimmer BC, Bell JS, Fastbom J, Wiese MD, Johnell K. Medication regimen complexity and polypharmacy as

factors associated with all-cause mortality in older people: a population-based cohort study. The Annals of

pharmacotherapy. 2016;50(2):89-95. doi:10.1177/1060028015621071.

21. Iwata M, Kuzuya M, Kitagawa Y, Suzuki Y, Iguchi A. Underappreciated predictors for postdischarge mortality in

acute hospitalized oldest-old patients. Gerontology. 2006;52(2):92-8. doi:10.1159/000090954.


13

22. Schmidt M, Pedersen L, Sørensen HT. The Danish Civil Registration System as a tool in epidemiology. European

journal of epidemiology. 2014;29(8):541-9. doi:10.1007/s10654-014-9930-3.

23. Kannegaard PN, Vinding KL, Hare-Bruun H. National database of geriatrics. Clinical epidemiology. 2016;8:731-5.

doi:10.2147/clep.s99473.

24. Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sorensen HT. The danish national patient

registry: a review of content, data quality, and research potential. Clinical epidemiology. 2015;7:449-90.

doi:10.2147/clep.s91125.

25. Schmidt M, Pedersen L, Sorensen HT. The danish civil registration system as a tool in epidemiology. European

journal of epidemiology. 2014;29(8):541-9. doi:10.1007/s10654-014-9930-3.

26. Johannesdottir SA, Horvath-Puho E, Ehrenstein V, Schmidt M, Pedersen L, Sorensen HT. Existing data sources for

clinical epidemiology: the danish national database of reimbursed prescriptions. Clinical epidemiology. 2012;4:303-13.

doi:10.2147/clep.s37587.

27. Larsen MD, Rosholm JU, Hallas J. The influence of comprehensive geriatric assessment on drug therapy in elderly

patients. European journal of clinical pharmacology. 2014;70(2):233-9. doi:10.1007/s00228-013-1601-x.

28. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson

comorbidity index predicted in-hospital mortality. Journal of clinical epidemiology. 2004;57(12):1288-94.

doi:10.1016/j.jclinepi.2004.03.012.

29. Bannay A, Chaignot C, Blotiere PO, Basson M, Weill A, Ricordeau P et al. The best use of the Charlson

Comorbidity Index with electronic health care database to predict mortality. Medical care. 2016;54(2):188-94.

doi:10.1097/mlr.0000000000000471.

30. Shah S, Vanclay F, Cooper B. Improving the sensitivity of the Barthel Index for stroke rehabilitation. Journal of

clinical epidemiology. 1989;42(8):703-9. doi:10.1016/0895-4356(89)90065-6.

31. Maribo T, Lauritsen JM, Waehrens E, Poulsen I, Hesselbo B. [Barthel Index for evaluation of function: a Danish

consensus on its use]. Ugeskrift for laeger. 2006;168(34):2790-2.

32. Bland JM, Altman DG. Survival probabilities (the Kaplan-Meier method). BMJ. 1998;317(7172):1572-.

doi:10.1136/bmj.317.7172.1572.

33. Mehrotra DV, Zhang Y. Hazard ratio estimation and inference in clinical trials with many tied event times. Statistics

in medicine. 2018;37(25):3547-56. doi:10.1002/sim.7843.


14

34. Botsis T, Bassoe CF, Hartvigsen G. Sixteen years of ICPC use in Norwegian primary care: looking through the

facts. BMC medical informatics and decision making. 2010;10:11. doi:10.1186/1472-6947-10-11.

35. Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sørensen HT. The predictive value of ICD-10 diagnostic

coding used to assess Charlson comorbidity index conditions in the population-based Danish National Registry of

Patients. BMC medical research methodology. 2011;11:83. doi:10.1186/1471-2288-11-83.

36. Satariano WA, Ragland DR. The effect of comorbidity on 3-year survival of women with primary breast cancer.

Annals of internal medicine. 1994;120(2):104-10. doi:10.7326/0003-4819-120-2-199401150-00002.

37. Di Bari M, Virgillo A, Matteuzzi D, Inzitari M, Mazzaglia G, Pozzi C et al. Predictive validity of measures of

comorbidity in older community dwellers: the Insufficienza Cardiaca negli Anziani Residenti a Dicomano Study.

Journal of the American Geriatrics Society. 2006;54(2):210-6. doi:10.1111/j.1532-5415.2005.00572.x.

38. Martínez-Velilla N, Cambra-Contin K, Ibáñez-Beroiz B. Comorbidity and prognostic indices do not improve the 5-

year mortality prediction of components of comprehensive geriatric assessment in hospitalized older patients. BMC

geriatrics. 2014;14(1):64. doi:10.1186/1471-2318-14-64.

39. Ware JE, Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and

item selection. Medical care. 1992;30(6):473-83.

40. Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies of illness in the aged. The index of ADL: a

standardized measure of biological and psychosocial function. Jama. 1963;185:914-9.

doi:10.1001/jama.1963.03060120024016.

41. Porter B, Arthur A, Savva GM. How do potentially inappropriate medications and polypharmacy affect mortality in

frail and non-frail cognitively impaired older adults? A cohort study. BMJ open. 2019;9(5):e026171.

doi:10.1136/bmjopen-2018-026171.

42. Díez-Manglano J, Giménez-López M, Garcés-Horna V, Sevil-Puras M, Castellar-Otín E, González-García P et al.

Excessive polypharmacy and survival in polypathological patients. European journal of clinical pharmacology.

2015;71(6):733-9. doi:10.1007/s00228-015-1837-8.


15

Legends

Figure 1: Kaplan-Meier survival curves total cohort (A), women (B), and men (C) stratified by number of medications


16

Table 1 Baseline characteristics of the study population

Total cohort

N=74,603

Women

N=46,823

Men

N=27,780

Number of drugs purchased (120 days) median [IQR] ‡ 0 (%)

1-4 (%) 5-9 (%) ≥10 (%)

Missing (%)

6 (4-9) 3.83

30.08 44.58 19.89 1.62

6 (4-9) 3.46

29.36 45.49 20.25 1.43

6 (4-9) 4.45

31.29 43.04 19.29 1.94

Diseases [IQR] * 12 (7-18) 12 (7-18) 13 (8-19) 0 (%) 0.85 0.87 0.81

1-4 (%) 11.22 11.85 10.16 5-9 (%) 24.52 25.46 22.93 ≥10 (%) 63.41 61.82 66.10

Charlson Comorbidity Index median [IQR] 0 (%) 1 (%) 2 (%)

≥3 (%)

2 (1-3) 17.66 26.31 21.16 34.87

2 (1-3) 20.96 28.26 21.24 29.53

2 (1-4) 12.08 23.02 21.02 43.87

Barthel Index median [IQR] 54 (29-77) 55 (30-77) 52 (26-77) 80-100 (%) 21.18 21.29 21.00 50-79 (%) 30.17 31.36 28.18 25-49 (%) 22.09 22.16 21.98 0-24 (%) 20.34 19.18 22.29

Missing (%) 6.21 6.01 6.56 Age (years) median [IQR] 83 (77-88) 84 (79-89) 81 (76-86)

65-74 (%) 16.19 13.07 21.45 75-84 (%) 41.03 39.22 44.08 85-94 (%) 38.86 42.92 32.03 ≥95 (%) 3.92 4.79 2.44

Marital status (%) Unmarried 6.50 5.84 7.62

Married 29.01 17.66 48.14 Divorced 12.34 12.31 12.40 Widowed 52.12 64.17 31.82 Missing 0.02 0.01 0.03

Period of admission (%) 2005-2009 42.04 43.64 39.34 2010-2014 57.96 56.36 60.66

BMI (kg/m2) mean (SD) 23.93 (5.07) 23.60 (5.26) 24.50 (4.65) <18.5 (%) 9.50 11.64 5.91

18.5-24.9 (%) 40.37 40.54 40.08 25-29.9 (%) 20.19 18.50 23.04

≥30 (%) 8.90 8.75 9.14 Missing (%) 21.04 20.58 21.83

Notes: * Number of unique diseases was calculated based on hospital discharge diagnoses during ten years before

baseline. The ICD-10 diagnosis was discriminated at the third level of the ICD-10 code ‡ All redeemed prescriptions

were included, except from the following ATC codes: B05x (Blood substitutes and perfusion solutions), B06x (Other

haematological agents), D09x (Medicated dressings), J07x (Vaccines), N01x (Anaesthetics) and Vx (Various).

Medications were counted at the third level of the ATC code, i.e. including the first 4 digits of the ATC code (e.g.

Salicylic acid and derivates: N02B). Normal distributed data are presented with mean (SD) whereas non-normal


17

distributed data are presented with both median (IQR) and mean (SD) Abbreviations: ATC, Anatomical Therapeutic

Chemical; BMI, Body Mass Index; IQR, Inter Quartile Range


18

Table 2: Univariable and multivariable Hazard Ratios and corresponding 95% confidence intervals for overall mortality, 30 days mortality, and 1-year mortality for total cohort (A), Women (B), and Men (C)

A Total cohort

Univariable HR (95 % CI)

Multivariable HR (95% CI)

Model 1 Model 2 Model 3 Overall Mortality

Medications 0 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1-4 1.06 (1.01,1.12) 1.06 (1.01,1.13) 1.04 (0.98,1.10) 1.04 (0.98,1.10) 5-9 1.26 (1.20,1.32) 1.31 (1.24,1.38) 1.21 (1.15,1.28) 1.17 (1.11,1.24) ≥10 1.46 (1.39,1.54) 1.69 (1.59,1.78) 1.47 (1.38,1.55) 1.35 (1.27,1.42) 30 days mortality

Medications 0 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1-4 1.04 (0.99,1.10) 1.03 (0.97,1.08) 1.02 (0.96,1.07) 1.01 (0.95,1.07) 5-9 1.14 (1.08,1.19) 1.16 (1.09,1.22) 1.12 (1.06,1.19) 1.09 (1.03,1.15) ≥10 1.20 (1.14,1.26) 1.32 (1.25,1.40) 1.25 (1.18,1.32) 1.17 (1.10,1.24) 1-year mortality

Medications 0 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1-4 1.05 (1.00,1.11) 1.03 (0.98,1.09) 1.02 (0.96,1.08) 1.01 (0.96,1.07) 5-9 1.19 (1.13,1.25) 1.22 (1.15,1.29) 1.17 (1.10,1.23) 1.11 (1.05,1.17) ≥10 1.31 (1.24,1.38) 1.48 (1.40,1.57) 1.37 (1.29,1.45) 1.23 (1.16,1.30)

Model 1: Adjusted for Barthel Index, age, marital status, period of index admission and BMI Model 2: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and disease count Model 3: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and Charlson comorbidity index n= 56,572


19

B Women

Univariable HR (95% CI)



Medications 0 1 (reference) 1 (reference) 1 (reference) 1 (reference) 1-4 1.03 (0.96,1.10) 1.07 (0.99,1.15) 1.04 (0.96,1.12) 1.04 (0.96,1.12) 5-9 1.21 (1.14,1.29) 1.33 (1.24,1.43) 1.24 (1.15,1.34) 1.20 (1.12,1.29) ≥10 1.41 (1.32,1.51) 1.74 (1.61,1.88) 1.53 (1.42,1.66) 1.41 (1.30,1.52) 30 days mortality



Model 1: Adjusted for Barthel Index, age, marital status, period of index admission and BMI Model 2: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and disease count Model 3: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and Charlson comorbidity index n=35,818


20

C Men

Univariable HR (95% CI)



Medications 0 1 (reference) 1 (reference 1 (reference) 1 (reference) 1-4 1.15 (1.07,1.24) 1.13 (1.04,1.22) 1.11 (1.02,1.21) 1.10 (1.01,1.20) 5-9 1.39 (1.30,1.50) 1.42 (1.31,1.54) 1.33 (1.23,1.45) 1.26 (1.16,1.37) ≥10 1.64 (1.52,1.77) 1.88 (1.72,2.05) 1.67 (1.53,1.82) 1.50 (1.37,1.64) 30 days mortality



Model 1: Adjusted for Barthel Index, age, marital status, period of index admission and BMI Model 2: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and disease count Model 3: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and Charlson comorbidity index n=20,754

21

Table 3: Univariable and multivariable Hazard Ratios and corresponding 95% confidence intervals for overall mortality according to each increase of one medication

Univariable

Hazard Ratio (95%CI)

Multivariable

Hazard Ratio (95%CI)

Model 1 Model 2 Model 3 Total cohort 1.033 (1.030,1.035) 1.050 (1.047,1.053) 1.038 (1.035,1.042) 1.027 (1.024,1.030) Women 1.031 (1.028,1.035) 1.053 (1.049,1.057) 1.043 (1.039,1.047) 1.032 (1.028,1.036) Men 1.038 (1.034,1.042) 1.057 (1.052,1.061) 1.047 (1.042,1.052) 1.035 (1.030,1.040)

Model 1: Adjusted for Barthel Index, age, marital status, period of index admission and BMI Model 2: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and disease count Model 3: Adjusted for Barthel Index, age, marital status, period of index admission, BMI, and Charlson comorbidity index n=54,308 (women 34,535; men 19,773)

22

Figure 1

A Total cohort

0.00

0.25

0.50

0.75

1.00

0 1 2 3 4 5 6 7 8 9 10 11Analysis time (years)

0 1-45-9 ≥ 10

Total population, n=73,394Kaplan-Meier survival curves stratified by number of medications

23

B Women

0.00

0.25

0.50

0.75

1.00


0 1-45-9 ≥ 10

Women, n=46,152Kaplan-Meier survival curves stratified by number of medications

24

C Men

0.00

0.25

0.50

0.75

1.00


0 1-45-9 ≥ 10

Men, n=27,242Kaplan-Meier survival curves stratified by number of medications

25

Appendix

Supplement Figure 1: Kaplan-Meier survival curves total cohort (A), women (B), and men (C) stratified by number of diseases

Supplement Figure 2: Kaplan-Meier survival curves total cohort (A), women (B), and men (C) stratified by Charlson comorbidity index

26

Supplement Figure 1

A Total cohort

0.00

0.25

0.50

0.75

1.00


0 1-45-9 ≥ 10

Total population, n=74,603Kaplan-Meier survival curves stratified by number of diseases

27

B Women

0.00

0.25

0.50

0.75

1.00


0 1-45-9 ≥ 10

Women, n=46,823Kaplan-Meier survival curves stratified by number of diseases

28

C Men

0.00

0.25

0.50

0.75

1.00


0 1-45-9 ≥ 10

Men, n=27,780Kaplan-Meier survival curves stratified by number of diseases

29

Supplement Figure 2

A Total cohort

0.00

0.25

0.50

0.75

1.00


0 12 ≥ 3

Total population, n=74,602Kaplan-Meier survival curves stratified by Charlson comorbidity index

30

B Women

0.00

0.25

0.50

0.75

1.00


0 12 ≥ 3

Women, n=46,822Kaplan-Meier survival curves stratified by Charlson comorbidity index

31

C Men

0.00

0.25

0.50

0.75

1.00


0 12 ≥ 3

Men, n=27,780Kaplan-Meier survival curves stratified by Charlson comorbidity index

a danish nationwide register-based cohort study

Documents