JAMDA xxx (2013) e1ee6

JAMDA

journal homepage: www.jamda.com

Original Study

Polypharmacy and Mortality Among Nursing Home Residents With AdvancedCognitive Impairment: Results From the Shelter Study

Graziano Onder MD, PhD a,*, Rosa Liperoti MD, MPH a, Andrea Foebel PhD a,Daniela Fialova PharmD, PhD b,c, Eva Topinkova MDb, Henriëtte G. van der Roest PhD d,Jacob Gindin MD e, Alfonso J. Cruz-Jentoft MD f, Massimo Fini MD g, Giovanni Gambassi MD a,Roberto Bernabei MD a for the SHELTER projectaCentro Medicina dell’Invecchiamento, Università Cattolica Sacro Cuore, Rome, ItalybDepartment of Geriatrics and Gerontology, 1st Faculty of Medicine, Charles University, Prague, Czech RepubliccDepartment of Social and Clinical Pharmacy, Faculty of Pharmacy, Charles University, Prague, Czech Republicd EMGO Institute for Health and Care Research, Department of Nursing Home Medicine, VU University Medical Center, Amsterdam, The Netherlandse Laboratory of Research in Geriatrics and Gerontology, University of Haifa, Haifa, Israelf Servicio de Geriatría, Hospital Universitario Ramón y Cajal, Madrid, SpaingDirezione Scientifica, IRCCS San Raffaele Pisana, Rome, Italy

Keywords:Advanced cognitive impairmentpolypharmacyend of life

The SHELTER study was funded by the SeventhEuropean Union. The work of DF was supported by grand thework of ETwas partly supported by research graCR NT-13238e4/2012.* Address correspondence to Graziano Onder, MD,

chiamento, Università Cattolica del Sacro Cuore, LargoE-mail address: graziano.onder@rm.unicatt.it (G. O

1525-8610/$ - see front matter Copyright � 2013 - Ahttp://dx.doi.org/10.1016/j.jamda.2013.03.014

a b s t r a c t

Introduction: Older adults with advanced cognitive impairment have a limited life expectancy and the useof multiple drugs is of questionable benefit in this population. The aim of the present study was to assessif, in a sample of nursing home (NH) residents with advanced cognitive impairment, the effect of pol-ypharmacy on mortality differs depending on estimated life expectancy.Methods: Data were from the Services and Health for Elderly in Long TERm care (SHELTER) project, a studycollecting information on residents admitted to57NHs in8European countries. Polypharmacywasdefinedas the concomitant use of 10 or more drugs. Limited life expectancy was estimated based on an AdvancedDementia Prognostic Tool (ADEPT) score of 13.5 or more. A Cognitive Performance Scale score of 5 or morewas used to define advanced cognitive impairment. Participants were followed for 1 year.Results: Mean age of 822 residents with advanced cognitive impairment entering the study was 84.6 (SD8.0) years, and 630 (86.6%) were women. Overall, 123 participants (15.0%) had an ADEPT score of 13.5 ormore (indicating limited life expectancy) and 114 (13.9%) were on polypharmacy. Relative to residentswith ADEPT score less than 13.5, those with ADEPT score of 13.5 or higher had a lower use of benzo-diazepines, antidementia drugs, and statins but a higher use of beta-blockers, digoxin, and antibiotics.Polypharmacy was associated with increased mortality among residents with ADEPT score of 13.5 ormore (adjusted hazard ratio [HR] 2.19, 95% confidence interval [CI]: 1.15e4.17), but not among those withADEPT score less than 13.5 (adjusted HR 1.10, 95% CI: 0.71e1.71).Discussion: Polypharmacy is associated with increased mortality in NH residents with advanced cognitiveimpairment at the end of life.Conclusion: These findings underline the need to assess life expectancy in older adults to improve theprescribing process and to simplify drug regimens.

Copyright � 2013 - American Medical Directors Association, Inc.

In Western countries, between 50% and 60% of nursing home (NH)residents have a diagnosis of dementia, and this condition is oftenassociated with other chronic diseases, including hypertension,

Framework Program of theant IGA-MH-CR NS-10029e4,nts VP 25/LF1/2 and IGA-MH-

Centro Medicina dell’Invec-F. Vito 1, 00168, Roma, Italy.nder).

merican Medical Directors Associa

cardiovascular disease, diabetes, and osteoporosis.1,2 As a conse-quence, the concomitant use of multiple drugs is common in thispopulation: data from US and European samples showed that NHresidents with dementia receive an average of 7 to 8 drugs everyday.3e5

Indeed, the use of multiple drugs in this population raises severalrelevant concerns.6,7 First, memory loss, decline in intellectual func-tion, and impaired judgment and language, commonly seen inpatients with dementia, may cause communication difficulties,including a decreased ability to report adverse effects.8 Second, older

tion, Inc.

G. Onder et al. / JAMDA xxx (2013) e1ee6e2

adults in general, and especially those with dementia, are usuallyexcluded from clinical trials, and the evidence base is scant.6 Third,drugs used to treat chronic comorbid diseases may further worsencognition or induce delirium.9 Finally, patients with advanceddementia often have a limited life expectancy, which may lowerbenefits from drug treatment.10e12 In particular, drugs used forprimary or secondary prevention may require years of continuoustreatment before their possible benefit is evident, and their usemay beconsidered as inappropriate in the context of a limited life expectancy.

Therefore, use of multiple drugs in patients with advancedcognitive impairment and limited life expectancy may not be of anyclinical benefit, but it might raise the risk of adverse events, ulti-mately leading to poor quality of life and reduced survival.13,14 Theimpact of polypharmacy on mortality in older adults with advancedcognitive impairment has never been evaluated. The aim of thepresent study was to assess if the effect of polypharmacy on mortalitydiffers depending on estimated life expectancy in a sample of NHresidents with advanced cognitive impairment.

Methods

Sample and Study Setting

The Services and Health for Elderly in Long TERm care (SHELTER)study enrolled 4156 residents in 57 NH of 7 European Union (EU)countries (Czech Republic, England, Finland, France, Germany, Italy,The Netherlands) and 1 non-EU country (Israel). The SHELTER studywas designed to validate the use of the interRAI instrument for LongTerm Care Facilities (interRAI LTCF), as a tool to assess the care needsand provision of care to residents in Europe.2

In each country, study partners identified a sample of NHs willingto participate to the study. This sample must not be consideredrandomly selected and it is not intended to be representative of allNHs in each country. Older adults residing in participating NH at thebeginning of the study and those admitted in the 3-month enroll-ment period following the initiation of the study were assessed usingthe interRAI LTCF and reassessed at 6 and 12 months if still in thefacility. If no longer in the facility, reason (death, hospitalization,discharge to home or another institution) and date of death ordischarge were recorded. No exclusion criteria were adopted.

Study researchers, responsible for data collection were trainedfollowing a previously validated procedure.2 In each country, courseswere organized to educate study researchers about the concepts ofcomprehensive geriatric assessment and multidisciplinary teamworkand to train them to the use of interRAI LTCF. Study researchers wereordinary clinical staff, external research staff, or a mixture of both.

Ethical approval for the study was obtained in all countries ac-cording to local regulations. Residents were invited to take part in thestudy and were free to decline participation. Consent was obtainedwith assurance of data confidentiality.

Data Sources

The interRAI LTCF contains more than 350 data elements, includingsociodemographic variables, numerous clinical items about bothphysical and cognitive status, as well as several medical diagnoses.15,16

The interRAI LTCF also includes information about an extensivearray of signs, symptoms, syndromes, and treatments being provided.2

Drug Use

Data about drugs residents received in the 3 days prior to theassessment were collected gathering information from physicianorder sheets and drug administration records. Drug information

included nonproprietary and proprietary name, Anatomical Thera-peutic and Chemical code of the World Health Organization Collab-orating Centre for Drug Statistics Methodology,17 formulation, dosage,frequency (number of times per day, week or month the drug istaken), and route of administration. Polypharmacy status was definedas concurrent use of 10 or more drugs at the baseline assessment.This cutoff was already used by former studies and associated withnegative outcomes, including mortality.3,18,19

Advanced Cognitive Impairment

The Cognitive Performance Scale (CPS), used to assess cognitivestatus, combines information on memory impairment, level ofconsciousness, and executive function.20 The CPS groups residents into7 cognitive performance categories (0 ¼ intact, 1 ¼ borderline intact,2 ¼ mild impairment, 3 ¼ moderate impairment, 4 ¼ moderatelysevere impairment, 5 ¼ severe impairment, and 6 ¼ very severeimpairment with eating problems). A CPS score of 5 corresponds toa mean (SD) Mini-Mental State Examination score of 5.1 (5.3).21 Test-retest and interrater reliability of items included in the CPS scale, aftertranslation into languages of countries participating to the SHELTERproject, was either adequate or excellent (average weighted kappa0.88 and 0.73, respectively).2 In the present study a CPS score of 5 ormore was used to define advanced cognitive impairment.

Life Expectancy: ADEPT Score

To assess life expectancy, the Advanced Dementia Prognostic Tool(ADEPT) score was used. The ADEPT score was created using MDSdata collected from all licensed NHs in the United States.22,23 Thescore consists of 12 items, including length of stay, age, gender,dyspnea, pressure ulcers, total functional dependence, bedfast,insufficient intake, bowel incontinence, body mass index, weight loss,and congestive heart failure. The total score can range between 1.0and 32.5, with higher points indicating a greater risk of death. In thepresent study, baseline interRAI LTCF data were used to calculate theADEPT scores. A cutoff of 13.5 was used to identify residents with anestimated limited life expectancy. This cutoff was already adopted bya study assessing the performance of the ADEPT to estimate 6-monthsurvival in NH residents, showing a high specificity (0.89).23

Independent Variables

The demographic variables included age and gender. To evaluatethe functional status, the 7-point Activities of Daily Living (ADL)Hierarchy scale was used.24 The ADL Hierarchy Scale ranges from0 (no impairment) to 6 (total dependence). ADL disability was cate-gorized as follows: assistance required (ADL Hierarchy Scale score 2 to4) and dependence (ADL Hierarchy Scale score �5). Behavioralsymptoms were present if the participant exhibited one or more of thefollowing symptoms in the 3 days before assessment: wandering,verbally abusive, physically abusive, socially inappropriate behavior,and active resistance of care. Presence of pain and dyspnea in the 3days before assessment was also assessed based on nurse report andreview of medical records. Weight loss was defined as loss of 5% ormore of body weight in past 30 days, or 10% or more in the past 180days. Information on medical diagnoses was collected gatheringinformation from the patient, the general practitioner, and after carefulreview of patient clinical documentation and previous medical history.

Mortality

Residents were followed during their stay in NHs and all deathswere recorded. Time to death was calculated from the date of the

G. Onder et al. / JAMDA xxx (2013) e1ee6 e3

first assessment to the date of death. The maximum length of follow-up was 1 year. Participants were censored at the time of their deathor at discharge from the NH or at the end of follow-up time (1 year).

Statistical Analysis

From the initial sample of 4156 residents, those without advancedcognitive impairment (defined as CPS <5, n ¼ 3142) and those withmissing data on drug use (n ¼ 76) were excluded, leading to a sampleof 938 participants. An additional 116 participants (12.4% of studysample) lacking follow-up assessment were excluded. These did notdiffer from those included in the study with regard to baseline ADEPTscore (10.2 � 3.5 vs 9.8 � 3.6, P ¼ .26) and rate of polypharmacy(11.2% vs 13.9%, P ¼ .43). This approach yielded a final sample size of822 participants.

Baseline characteristics of participants were compared usinganalyses of variance for normally distributed variables, nonpara-metric Mann-Whitney U test for skewed variables, and chi-squareanalyses for dichotomous variables. Cox proportional hazardsregressions were fitted to evaluate the effect of polypharmacy ontime to death, among residents with (ADEPT �13.5) and withoutlimited life expectancy (ADEPT <13.5), adjusting for potentialconfounders. To exclude departure from proportionality assumption,the log-log survival function was examined. Analyses were adjustedfor age, gender, country, and variables associated with polypharmacyat the univariate analysis with a P of less than or equal to .10 either inthe ADEPT score greater than or equal to 13.5 or in the ADEPT scoreless than 13.5 group. The impact of polypharmacy on survival wasalso tested comparing the survival curves obtained with the Kaplan-Meier method in both participants with ADEPT score less than 13.5and greater than or equal to 13.5. All analyses were performed usingSPSS for Windows version 13.0 (SPSS, Inc, Chicago, IL).

Results

Mean age of the 822 residents was 84.6 (SD 8.0) years, and 630(76.6%) were women. Of the total sample, 123 participants (15.0%)had an ADEPT score greater than or equal to 13.5 (indicating limitedlife expectancy) and 114 (13.9%) were on polypharmacy. Overall, 95 of699 residents in the ADEPT score less than 13.5 group (13.6%) and 19of 123 in the ADEPT score greater than or equal to 13.5 (15.4%) wereon polypharmacy (P ¼ .58) and number of drugs used did not differsignificantly according to ADEPT score groups (ADEPT score <13.5group: 5.9 SD 3.0; ADEPT score �13.5: 6.0 SD 3.0, P ¼ .82). Figure 1presents rate of mortality and mean number of drugs used accord-ing to ADEPT score. One-year mortality rate progressively increasedwith ADEPT score, ranging from 6.9% among residents with score less

Fig. 1. Mortality rate and mean number of drugs according to ADEPT score.

than 4 to 51.0% among those with score greater than or equal to 16,but no difference in the mean number of drugs used was observedwith increasing ADEPT score.

Characteristics of the study population according to polypharmacyare summarized in Table 1. In the group with ADEPT score less than13.5, residents on polypharmacy had a significantly higher rate ofpain and dyspnea and a higher number of comorbid medical diag-noses, in particular, ischemic heart disease and Parkinson disease. Inthe group with ADEPT score greater than or equal to 13.5, none of theexamined variables was significantly associated with polypharmacy.

Table 2 presents pattern of drug use in the study sample. Drugsused to treat gastrointestinal diseases were among the mostfrequently prescribed drugs (laxatives 47.0% of study sample; anti-ulcer drugs 33.9%), but extremely common was also the use ofpsychotropic drugs, with about one-third of participating residentsreceiving antipsychotics (36.5%) or benzodiazepines (32.8%). Also,cardiovascular drugs were commonly used with 34.1% of residentsusing acetylsalicylic acid and antiplatelet drugs, 23.8% diuretics, 16.5%angiotensin-converting enzyme inhibitors, 16.2% beta-blockers, 13.9%calcium channel blockers, 8.3% statins, and 4.7% digoxin.

Use of drugs differs according to ADEPT score. Residents withADEPT score less than 13.5 had a significantly higher use of benzo-diazepines, antidementia drugs and statins and a significantly loweruse of beta-blockers, digoxin, and antibiotics.

In the whole sample, polypharmacy was associated with highermortality rate: 173 (24.4%; crude incident rate 0.30 per person-year[p-y]) of 708 not on polypharmacy and 39 residents on poly-pharmacy (34.2%; crude incident rate 0.45 per p-y) of 114 died duringthe follow up. However, after adjusting for potential confounders,which included CPS score, pain, dyspnea, number of diseases,ischemic heart disease, and Parkinson disease, no significant associ-ation was found between polypharmacy and mortality (adjustedhazard ratio [HR] 1.29, 95% confidence intervals [CI]: 0.90e1.84).

Table 3 presents the association between polypharmacy andmortality according to ADEPT score groups. In the ADEPT score lessthan 13.5 group, mortality rate was 21.4% among residents not onpolypharmacy (crude incident rate 0.26 per p-y) and 27.4% amongthose on polypharmacy (crude incident rate 0.34 per p-y). After ad-justing for potential confounders, the risk of death did not differsignificantly according to polypharmacy (adjusted HR 1.10, 95% CI:0.71e1.71).

In the ADEPT score greater than or equal to 13.5 group, mortalityrate was 42.3% among residents not on polypharmacy (crude incidentrate 0.63 per p-y) and 68.4% among those on polypharmacy (crudeincident rate 1.39 per p-y). After adjusting for potential confounders,the risk of death was significantly higher among residents on poly-pharmacy (adjusted HR 2.19, 95% CI: 1.15e4.17). When the interactionbetween polypharmacy and ADEPT score was formally tested byentering the interaction term in the fully adjusted model, a significantresult was obtained (P ¼ .04), suggesting that ADEPT score maymodify the effect of polypharmacy on mortality.

Figure 2 shows the survival curves according to polypharmacy inparticipants’ ADEPT score less than 13.5 and greater than or equal to13.5. A significantly increased mortality in the polypharmacy groupwas observed among residents with ADEPT greater than or equal to13.5, but not among those with ADEPT score less than 13.5.

Discussion

The present study shows that polypharmacy is common inEuropean NH residents with severe cognitive impairment, withalmost 1 of 6 residents in the study sample receiving 10 or moredrugs and that this condition is associated with increased mortality inpatients with estimated limited life expectancy.

Table 1Sample Characteristics According to Polypharmacy (Use of �10 Drugs)

ADEPT Score < 13.5 ADEPT Score � 13.5

No Polypharmacyn ¼ 604 (%)

Polypharmacyn ¼ 95 (%)

P No Polypharmacyn ¼ 104 (%)

Polypharmacyn ¼ 19 (%)

P

DemographicsAge, y, mean � SD 83.8 � 8.6 83.8 � 8.3 .98 89.4 � 7.5 87.2 � 7.6 .24Female gender 482 (79.8) 72 (75.8) .37 66 (63.5) 10 (52.6) .37Length of stay �6 mo 507 (83.9) 81 (85.3) .74 59 (56.7) 13 (68.4) .34

Geriatric conditionsADL disability* .32 .39Assistance required 215 (35.6) 36 (37.9) 4 (3.8) 0 (0)Dependent 385 (63.7) 57 (60.0) 100 (96.2) 19 (100)CPS score, mean � SD 5.4 � 0.5 5.3 � 0.4 .06 5.7 � 0.5 5.8 � 0.4 .21Behavioral symptoms 364 (60.3) 62 (65.3) .35 47 (45.2) 7 (36.8) .50Weight loss 57 (9.4) 11 (11.6) .51 22 (21.2) 6 (31.6) .32

SymptomsPain 129 (21.4) 40 (42.1) <.001 27 (26.0) 5 (26.3) .97Dyspnea 24 (4.0) 14 (14.7) <.001 17 (16.3) 5 (26.3) .30

ComorbidityNo. of diseases, mean � SD 2.5 � 1.5 3.0 � 1.3 .003 3.1 � 1.6 3.5 � 1.9 .36Ischemic heart disease 125 (20.7) 29 (30.5) .03 38 (36.9) 11 (57.9) .09Heart failure 66 (10.9) 14 (14.7) .28 25 (24.0) 6 (31.6) .49Parkinson disease 41 (6.8) 14 (14.7) .01 10 (9.6) 2 (10.5) .90Stroke 98 (16.2) 21 (22.1) .16 36 (34.6) 5 (26.3) .48Diabetes 113 (18.7) 23 (24.2) .21 16 (15.4) 3 (15.8) .96Cancer 52 (8.6) 8 (8.4) .95 16 (14.8) 3 (15.8) .60ADEPT, mean � SD 8.7 � 2.6 8.9 � 2.5 .58 15.8 � 1.9 16.3 � 1.7 .24

ADEPT, Advanced Dementia Prognostic Tool; ADL, activities of daily living; CPS, Cognitive Performance Scale.*Assistance required is defined by ADL hierarchical scale score 2 to 4, dependent by ADL hierarchical scale score 5 to 6.

G. Onder et al. / JAMDA xxx (2013) e1ee6e4

Rate of polypharmacy in the present study is in line with a formerstudy performed in 323 residents enrolled in 22 NHs in the UnitedStates (mean number of drugs used 5.9),5 but it differs from the oneobserved among 2665 community-dwelling adults aged 65 or olderwith dementia in the United States, included in the NationalAlzheimer’s Coordinating Center Uniform Data Set (mean number ofdrugs used was 5).25 Such a difference could be explained by the fact

Table 2Patterns of Drug Use According to ADEPT Score

Drug Class Whole Samplen ¼ 822 (%)

Gastrointestinal drugsLaxatives 386 (47.0)Antiulcer drugs 279 (33.9)

Psychotropic drugsAntipsychotics 300 (36.5)Benzodiazepines 270 (32.8)Antidepressants 242 (29.4)Antidementia drugs 109 (13.3)

Cardiovascular drugsAcetylsalicylic acid and antiplatelet drugs 280 (34.1)Diuretics 196 (23.8)Ace inhibitors 136 (16.5)Beta blockers 133 (16.2)Calcium channel blockers 114 (13.9)Statins 68 (8.3)Digoxin 39 (4.7)Angiotensin Receptor Blockers 22 (2.7)

Musculoskeletal system drugsAnalgesics 234 (28.5)Antiosteoporosis drugs (including vitamin D) 99 (12.0)

AntidiabeticsOral hypoglycemic agents 51 (6.2)Insulin 47 (5.7)

OthersAntibiotics 36 (4.4)Vitamin supplements (not including vitamin D) 22 (2.7)Corticosteroids 20 (2.4)

ADEPT, Advanced Dementia Prognostic Tool.

that demented patients in NH may be sicker and more in need ofpharmacological treatments as compared with those living in thecommunity.

Interestingly in our sample, no difference was observed in the rateof polypharmacy according to life expectancy. This finding may berelated to the fact that discontinuation of drugs in end-of-life care isnot standard practice and there are no clear indications available

ADEPT <13.5n ¼ 699 (%)

ADEPT �13.5n ¼ 123 (%)

P

333 (47.6) 53 (43.1) .35229 (32.8) 50 (40.7) .09

261 (37.3) 39 (31.7) .23247 (35.3) 23 (18.7) <.001214 (30.6) 28 (22.8) .30104 (14.9) 5 (4.1) .001

238 (34.0) 42 (34.1) .98164 (23.5) 32 (26.0) .54117 (16.7) 19 (15.4) .72104 (14.9) 29 (23.6) .02101 (14.4) 13 (10.6) .2566 (9.4) 2 (1.6) .00428 (4.0) 11 (8.9) .0219 (2.7) 3 (2.4) .86

193 (27.6) 41 (33.3) .2089 (12.7) 10 (8.1) .15

47 (6.7) 4 (3.3) .1441 (5.9) 6 (4.9) .66

23 (3.3) 16 (13.0) <.00116 (2.3) 6 (4.9) .1019 (2.7) 1 (0.8) .21

Table 3Polypharmacy and Mortality

No. of Events (%) Crude Incident Rateper Person-Year

Crude Hazard Ratio(95% CI)

Adjusted* HazardRatio (95% CI)

ADEPT <13.5 (n ¼ 699)No polypharmacy 129/604 (21.4) 0.26 1 (Reference) 1 (Reference)Polypharmacy 26/95 (27.4) 0.34 1.29 (0.85e1.97) 1.10 (0.71e1.71)

ADEPT �13.5 (n ¼ 123)No polypharmacy 44/104 (42.3) 0.63 1 (Reference) 1 (Reference)Polypharmacy 13/19 (68.4) 1.39 2.09 (1.12e3.89) 2.19 (1.15e4.17)

ADEPT, Advanced Dementia Prognostic Tool; CI, confidence interval.P for interaction between polypharmacy and ADEPT category in the fully adjusted model ¼ .04.

*Adjusted for age, gender, country, Cognitive Performance Scale score, pain, dyspnea, number of diseases, ischemic heart disease, Parkinson disease.

G. Onder et al. / JAMDA xxx (2013) e1ee6 e5

about drug use to treat chronic and acute diseases at the end of life.Avoiding or discontinuing drugs aimed at prolonging life or pre-venting clinical events may seem reasonable when the time neededto obtain the expected benefits from the drugs is longer than theestimated life expectancy of a particular individual, and when they donot offer symptomatic benefits.26,27 For example, in the SHELTERsample, use of statins and antidementia drugs was reduced inparticipants with limited life expectancy. Benefits from statins may bequestionable in a cohort of patients with end-stage dementia andlimited life expectancy when quality of life is the main focus of care

Fig. 2. Survival according to polypharmacy in residents with ADEPT score <13.5(upper panel) and �13.5 (lower panel).

and the benefits of these drugs are irrelevant.28 Also, use of anti-dementia drugs may raise several concerns in patients with advanceddementia,28 despite recent results from the DOMINO trial suggestingthat in patients with moderate to severe Alzheimer dementia,continuing donepezil, as compared with discontinuing it, was asso-ciated with better scores on measures of cognitive ability and activ-ities of daily living.29 Noticeably, only half of the patients who wereassigned to continue donepezil in this trial maintained their treat-ment for the entire study period, suggesting that many patients mayhave perceived that continuing the drug was not effective.30

However, drugs used to treat some chronic (ie, heart failure) oracute diseases (ie, pneumonia) may improve both life span andsymptoms, and treatment optimization of the underlying disease maycontribute to an increased control of symptoms.26 In this context,drugs commonly used in the treatment of heart failure (beta-blockersand digoxin) and antibiotics were more commonly prescribed inresidents with limited life expectancy in the present study.

In our study polypharmacy was associated with increasedmortality among patients with limited life expectancy but not amongthose without limited life expectancy, suggesting that intensemedical treatment might be unrewarding and potentially dangerousin older adults with terminal illness. Indeed, polypharmacy isa consistent independent predictor of adverse drug events, inparticular among frail older adults living in NHs, and several obser-vational studies have associated this condition with increasedmortality.13,14,19 In particular, among older adults with cognitiveimpairment and limited life expectancy, clinical benefits derived fromuse of multiple drugs and in particular those aimed at prolonging lifeor preventing clinical events, are negligible and do not counterbal-ance the risk of iatrogenic illness.26e28 In addition, this populationmay have feeding problems, which make the oral administration ofseveral drugs difficult and increase the risk of aspiration pneumonia.6

Interestingly O’Mahony and O’Connor recently suggested theconversion of major polypharmacy to “oligopharmacy,” (ie, �5 dailydrugs) to avoid the onset of serious adverse drug events in patientswith limited life expectancy.31 Different approaches were tested toreduce the burden of polypharmacy in older adults.32,33 Interestingly,one small study from Israel showed that a deliberate policy of oli-gopharmacy in frail, disabled older people in nursing care unitssignificantly reduced mortality, hospitalization, and drug costs.34 Thisstructured approach to palliative pharmacotherapy has also beenshown to be applicable in community-dwelling frail older people, aswell as nursing unit residents (the Garfinkel method).35

Some limitations of the present study need to be recognized. Thecriteria used to define severe cognitive impairment (CPS �5), do notallow identification of the cause of cognitive impairment. Althoughthe interRAI LTCF is a standardized and validated assessmentinstrument, the recording of diseases causing cognitive impairment isnot its specific focus. In addition, the results we observed refer toa group of older adults with advanced cognitive impairment in NHand they are not generalizable to populations in different settings.

G. Onder et al. / JAMDA xxx (2013) e1ee6e6

In conclusion, the present study shows that polypharmacy isassociated with increased mortality in NH residents with advancedcognitive impairment at the end of life. These findings underline theneed to assess life expectancy in older adults to improve theprescribing process and to simplify drug regimens among those withlimited life expectancy.36,37 Future studies should assess if interven-tions aimed at avoiding or discontinuing drugs prescribed to prolonglife or prevent clinical events and focused to relieve symptoms mayimprove quality of life in older adults with dementia and limited lifeexpectancy.

References

1. Lee PG, Cigolle C, Blaum C. The co-occurrence of chronic diseases and geriatricsyndromes: The Health and Retirement Study. J Am Geriatr Soc 2009;57:511e516.

2. Onder G, Carpenter I, Finne-Soveri H, et al. Assessment of nursing home resi-dents in Europe: The Services and Health for Elderly in Long TERm care(SHELTER) study. BMC Health Serv Res 2012;12:5.

3. Onder G, Liperoti R, Fialova D, et al, SHELTER Project. Polypharmacy in nursinghome in Europe: Results from the SHELTER study. J Gerontol A Biol Sci Med Sci2012;67:698e704.

4. Morley JE. Polypharmacy in the nursing home. J Am Med Dir Assoc 2009;10:289e291.

5. Bronskill SE, Gill SS, Paterson JM, et al. Exploring variation in rates of poly-pharmacy across long term care homes. J Am Med Dir Assoc 2012;13:309.e15e309.e21.

6. Brauner DJ, Muir JC, Sachs GA. Treating nondementia illnesses in patients withdementia. JAMA 2000;283:3230e3235.

7. American Geriatrics Society Expert Panel on the Care of Older Adults withMultimorbidity. Patient-centered care for older adults with multiple chronicconditions: a stepwise approach from the American Geriatrics Society: Amer-ican Geriatrics Society Expert Panel on the Care of Older Adults with Multi-morbidity. J Am Geriatr Soc 2012;60:1957e1968.

8. Onder G, Lattanzio F, Battaglia M, et al. The risk of adverse drug reactions inolder patients: Beyond drug metabolism. Curr Drug Metab 2011;12:647e651.

9. Huey ED, Taylor JL, Luu P, et al. Factors associated with use of medications withpotential to impair cognition or cholinesterase inhibitors among Alzheimer’sdisease patients. Alzheimers Dement 2006;2:314e321.

10. Holmes HM. Rational prescribing for patients with a reduced life expectancy.Clin Pharmacol Ther 2009;85:103e107.

11. Currow DC, Abernethy AP. Frameworks for approaching prescribing at the endof life. Arch Intern Med 2006;166:2404.

12. Fusco D, Lattanzio F, Tosato M, et al. Development of CRIteria to assessappropriate Medication use among Elderly complex patients (CRIME) project:Rationale and methodology. Drugs Aging 2009;26:3e13.

13. Cherubini A, Eusebi P, Dell’Aquila G, et al. Predictors of hospitalization in Italiannursing home residents: the U.L.I.S.S.E. project. J Am Med Dir Assoc 2012;13:84.e5e84.e10.

14. Onder G, Petrovic M, Tangiisuran B, et al. Development and validation ofa score to assess risk of adverse drug reactions among in-hospital patients 65years or older: The GerontoNet ADR risk score. Arch Intern Med 2010;170:1142e1148.

15. Mor V. A comprehensive clinical assessment tool to inform policy and practice:Applications of the minimum data set. Med Care 2004;42:III50eIII59.

16. Bernabei R, Landi F, Onder G, et al. Second and third generation assessmentinstruments: The birth of standardization in geriatric care. J Gerontol A Biol SciMed Sci 2008;63:308e313.

17. WHO Collaborating Centre for Drug Statistics Methodology. Available at:http://www.whocc.no/. Accessed on June 29, 2011.

18. Kaufman DW, Kelly JP, Rosenberg L, et al. Recent patterns of medication use inthe ambulatory adult population of the United States: The Slone survey. JAMA2002;287:337e344.

19. Jyrkkä J, Enlund H, Korhonen MJ, et al. Polypharmacy status as an indicator ofmortality in an elderly population. Drugs Aging 2009;26:1039e1048.

20. Morris JN, Fries BE, Mehr DR, et al. The MDS Cognitive Performance Scale.J Gerontol 1994;49:174e182.

21. Hartmaier SL, Sloane PD, Guess HA, et al. Validation of the Minimum Data SetCognitive Performance Scale: Agreement with the Mini-Mental State Exami-nation. J Gerontol A Biol Sci Med Sci 1995;50:M128eM133.

22. Mitchell SL, Miller SC, Teno JM, et al. The advanced dementia prognostic tool: Arisk score to estimate survival in nursing home residents with advanceddementia. J Pain Symptom Manage 2010;40:639e651.

23. Mitchell SL, Miller SC, Teno JM, et al. Prediction of 6-month survival of nursinghome residents with advanced dementia using ADEPT vs hospice eligibilityguidelines. JAMA 2010;304:1929e1935.

24. Morris JN, Fries BE, Morris SA. Scaling ADL’s within the MDS. J Gerontol 1999;4:M546eM553.

25. Lau DT, Mercaldo ND, Harris AT, et al. Polypharmacy and potentially inap-propriate medication use among community-dwelling elders with dementia.Alzheimer Dis Assoc Disord 2010;24:56e63.

26. Cruz-Jentoft AJ, Boland B, Rexach L. Drug therapy optimization at the end oflife. Drugs Aging 2012;29:511e521.

27. Vetrano DL, Tosato M, Colloca G, et al, SHELTER Study. Polypharmacy in nursinghome residents with severe cognitive impairment: Results from the SHELTERStudy. [published online ahead of print December 8, 2012]. Alzheimers Dementdoi: 10.1016/j.jalz.2012.09.009.

28. Holmes HM, Sachs GA, Shega JW, et al. Integrating palliative medicine into thecare of persons with advanced dementia: Identifying appropriate medicationuse. J Am Geriatr Soc 2008;56:1306e1311.

29. Howard R, McShane R, Lindesay J, et al. Donepezil and memantine formoderate-to-severe Alzheimer’s disease. N Engl J Med 2012;366:893e903.

30. Schneider LS. Discontinuing donepezil or starting memantine for Alzheimer’sdisease. N Engl J Med 2012;366:957e959.

31. O’Mahony D, O’Connor MN. Pharmacotherapy at the end-of-life. Age Ageing2011;40:419e422.

32. Tamura BK, Bell CL, Lubimir K, et al. Physician intervention for medicationreduction in a nursing home: The polypharmacy outcomes project. J Am MedDir Assoc 2011;12:326e330.

33. Little MO, Morley A. Reducing polypharmacy: Evidence from a simple qualityimprovement initiative. J Am Med Dir Assoc 2013;14:152e156.

34. Garfinkel D, Zur-Gil S, Ben-Israel J. The war against polypharmacy: A new cost-effective geriatric-palliative approach for improving drug therapy in disabledelderly people. Isr Med Assoc J 2007;9:430e434.

35. Garfinkel D, Mangin D. Feasibility study of a systematic approach for discon-tinuation of multiple medications in older adults: Addressing polypharmacy.Arch Intern Med 2010;170:1648e1654.

36. Gill TM. The central role of prognosis in clinical decision making. JAMA 2012;307:199e200.

37. Morley JE. End-of-life care in the nursing home. J Am Med Dir Assoc 2011;12:77e83.