Acetylcholinesterase inhibitors in assisted living: patterns of use and association with retention

INTERNATIONAL JOURNAL OF GERIATRIC PSYCHIATRY

Int J Geriatr Psychiatry 2008; 23: 178–184.

Published online 6 August 2007 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/gps.1859

Acetylcholinesterase inhibitors in assisted living:patterns of use and association with retention

Adam Rosenblatt1,2*, Quincy M. Samus1, Chiadi U. Onyike1, Alva S. Baker2,Matthew McNabney3, Lawrence S. Mayer1, Jason Brandt2,4 and Constantine G. Lyketsos1,2

1Department of Psychiatry and Behavioral Sciences, Division of Geriatric Psychiatry and Neuropsychiatry, Johns HopkinsSchool of Medicine, Baltimore, MD, USA2The Copper Ridge Institute, Sykesville, MD, USA3Department of Medicine, Division of Geriatric Medicine and Gerontology, Johns Hopkins School of Medicine, Baltimore,MD, USA4Department of Psychiatry and Behavioral Sciences, Division of Medical Psychology, Johns Hopkins School of Medicine,Baltimore, MD, USA

SUMMARY

Objectives To describe patterns of Acetylcholinesterase inhibitor (ACI) use in an Assisted Living (AL) population, and theassociation of ACIs with retention in AL.Methods As part of the Maryland Assisted Living Study (MD–AL), 198 residents of 22 ALs were evaluated. Dementia wasdiagnosed in 134, and specifically Alzheimer’s disease (AD) in 79, by an expert consensus panel. Data was collected on ACIagent and dose. Vital status and location were recorded every 6 months. Other data included age, duration of residence,general medical health rating (GHMR), Mini-Mental State Examination (MMSE), Neuropsychiatric Inventory (NPI),Cornell Scale for Depression in Dementia (CSDD) and number of non-psychiatric medications.Results The overall ACI treatment rate was 31%. 34.5% of participants with mild to moderate AD were taking ACIs. Onlytwo in seven participants taking rivastigmine were taking an adequate dose. Participants with AD on ACI’s did not differsignificantly from those not on ACI’s in any of the secondary measures except age and duration of residence, those on theagents being somewhat younger and more recently admitted. For participants with AD, only ACI use was significantlyassociated with retention in AL at 6 months, with a relative risk of death or discharge to higher level care of 0.217. BaselineMMSE was associated with retention for those with non-AD dementia. In a survival analysis ACI use was associated with228.75 days longer retention in participants with AD.Conclusion ACIs have low rates of use in AL and are associated with better retention for residents with AD. Copyright #2007 John Wiley & Sons, Ltd.

key words — assisted living; acetylcholinesterase inhibitors; dementia

INTRODUCTION

Dementia presents a challenge to the goals of AssistedLiving (AL), to provide a residence to elders unable tolive alone, to maximize their quality of life, and tosupport their ability to age in place (Bernard et al.,2001; NCAL, 2001). Indirect prevalence estimates of

*Correspondence to: Dr A. Rosenblatt, Johns Hopkins Hospital,Meyer 2–181, 600 North Wolfe Street, Baltimore, MD 21287, USA.E-mail: [email protected]

Copyright # 2007 John Wiley & Sons, Ltd.

dementia in AL have ranged from as low as 18% to ashigh as 70% (Hendrie, 1998). The Maryland AssistedLiving Study (MD–AL) established, by direct assess-ment, a 68% rate for Central Maryland. When specificdiagnoses were considered, 47% of AL residents, werefound to have Alzheimer’s disease (AD), making thempotential candidates for pharmacotherapy (Rosenblattet al., 2004)

There are five prescription therapies for AD inthe USA: the acetylcholinesterase inhibitors (ACIs)tacrine, donepezil, rivastigmine and galantamine and

Received 13 November 2006Accepted 10 May 2007

acetylcholinesterase inhibitors in assisted living 179

the NMDA antagonist memantine, which was unre-leased at the time of the study. The ACIs have beenshown in various studies to improve measures ofcognition, function (Corey-Bloom et al., 1998; Mohset al., 2001) and behavioral disturbance (Cummingset al., 2000a, 2000b; Tariot et al., 2000; Holmes et al.,2004). However, the clinical significance of theseimprovements is controversial, and it remains unclearwhether the ACIs produce long-term benefits suchas decreased mortality or institutionalization. Someobservational studies suggest better long-term out-comes for Alzheimer’s patients on ACIs (Mohs et al.,2001; Geldmacher et al., 2003; Lopez et al., 2005) butin a controlled study of community dwelling ADpatients in the UK, donepezil did not produce adifference in institutionalization (2000 CollaborativeGroup, 2004).

AL residents have high rates of dementia, arefunctionally compromised, and are at risk for transferto nursing homes. They might be sensitive to relativelymodest improvements in cognitive function. Our goalwas to describe patterns of ACI use in AL, and todetermine whether use is associated with retention.We hypothesized that ACI use would be associatedwith retention at an AL level of care, taking otherrelevant factors into account, and that this associationwould be specific to Alzheimer’s disease.

METHODS

Design and procedures

The procedures of the Maryland Assisted LivingStudy have been described elsewhere. (Rosenblattet al., 2004) Informed consent was obtained from 198residents of 22 facilities, or their surrogates. Historywas obtained from the resident, family, and staff. Astudy team examined the resident, and reviewed theAL medical chart and other records, recording alldiagnoses, medications, and other treatments. Thiswas followed by a 1 h cognitive battery including theMini-Mental State Exam (MMSE; Folstein et al.,1975) chosen for its popularity and ease of use, andstandardized measures of psychopathology developedfor use in the elderly and persons with dementia, TheGeneral Medical Health Rating (GMHR; Lyketsoset al., 1999), and the number of non-psychiatricmedications were included as measures of medicalco-morbidity. The Neuropsychiatric Inventory (NPI;Cummings et al., 1994) was a measure of behavioralproblems, and the Cornell Scale for Depression inDementia (CSDD; Alexopoulos et al., 1988) meas-ured depressive symptoms.


At a diagnostic conference, investigators represent-ing psychiatry, neuropsychology, geriatric medicine,and nursing rendered diagnoses using standard clinicalcriteria for probable Alzheimer’s disease (McKhannet al., 1984) probable vascular dementia (NINDS/AIREN, 1993), Lewy body dementia (McKeith et al.,1992), and fronto-temporal dementia (Neary, 1990).Six months after each assessment, status and locationof residents were checked by telephone. The date anddestination of discharge were recorded, or the date ofdeath. The data were entered into a relational databaseusing Paradox 9.

Assessment of outcomes at 6 months after initialassessment

An analysis was performed on the 6-month data to seewhich baseline variables were associated with favor-able outcome, defined as the participant remaining inthe same AL, transferring to a different AL, or going toan independent level of care. Unfavorable outcomesincluded death and transfer to a nursing home,medical, or psychiatric hospital. Three undecideableoutcomes were censored: two participants who wereremoved for economic reasons, and an individual whotransferred to a chronic facility for the handicapped.

The primary statistical method was a t-test, withfavorability of outcome as the grouping variable, or achi-square test arising from a 2� 2 crosstab withfavorability of outcome as one variable of the table.Each analysis was performed separately for residentswith AD and non-AD dementia. Because of theunitary nature of our hypothesis, no statistical controlwas used for multiple comparisons. The small samplesize would render a stepwise or multivariate analysisvirtually uninterpretable. Instead, we attempted toaddress treatment bias by determining whether anyof secondary variables was itself associated withACI use. Residents with milder forms of cognitiveimpairment were not included, but residents were notexcluded for having severe dementia, for whichacetylcholinesterase inhibitors use was not approved,or for being on a sub-therapeutic dose. Data analyseswere performed using SPSS 11.5.

Assessment of time to discharge or death

When MD-AL continued into a second phase, we wereable to perform a longer survival analysis. We used theKaplan-Meier method to estimate the probability ofremaining in AL over time and the log-rank test toevaluate differences between residents with AD whowere or were not taking ACI’s at baseline. Duration offollow-up ranged from 2.7 to 4.6 years. Endpoints


DOI: 10.1002/gps

Table 1. ACI treatment rates, by diagnosis and severity

Severity AD (n¼ 79) Non-AD (n¼ 55) All dementia (n¼ 134)

Mild (MMSE> 17) 56.6% 22.2% 38.0%Mild-Mod (MMSE> 10) 34.5% 22.7% 29.9%Mod. (MMSE 10–17) 18.8% 26.7% 20.4%Severe (MMSE< 10) 25.0% 54.5% 34.3%Overall 31.6% 29.1% 30.1%

180 a. rosenblatt ET AL.

were date of death in AL, date of discharge to higher-level care, date of censoring as a result of discharge toanother assisted living, or end of study follow-up.Events for residents who died in a medical or nursingfacility, whose exact date of discharge from AL wasnot known, were estimated as the midpoint of the timelast known in AL and the time of death in the newfacility (n¼ 8). Four cases were excluded because theoutcome was indeterminate, as defined above. Theprogram SAS 9.1 was used for the survival analysis.

RESULTS

ACI use patterns

One hundred and thirty-four out of 198 participantswere found to be suffering from dementia. Of these, 41or 31% were receiving ACI’s. Thirty-four of the 41were taking donepezil, at a mean dose of 8.53 mg/day.One participant was taking 16 mg of galantamine aday. The seven residents taking rivastigmine had a lowrate of adequate dosing. The mean dose was 5.14 mg/day, with only two residents taking the 6–12 mg perday recommended in the package insert.

Of the 41 participants taking ACIs, the consensuspanel determined that 25 had probable Alzheimer’sdisease, five had Vascular dementia, one had Demen-tia with Lewy Bodies, and ten had Mixed or DementiaNot Otherwise Specified. The treatment rate forAlzheimer’s disease was 31.6%, only slightly higher

Table 2. Comparison of residents with AD by treatment group

ACIþ (n¼ 25) A

Mean SD Mea

Sex 11% male — 24% mAge 85.03 5.42 88.3MMSE 15.04 7.83 12.0GMHR 3.20 0.87 2.98No. Meds 4.08 2.55 3.72NPI 11.80 13.23 14.2CSDD 3.92 3.47 5.28Length of Res. 18.64 11.56 27.1


than for Vascular dementia (27.8%) or dementia NOS(27%). The treatment rate for all non-Alzheimer’sforms of dementia together was 29.1%. When parti-cipants were grouped by severity, no particular trendemerged. Treatment rates by diagnosis and severityare displayed in Table 1. For the population in whichACI’s were officially indicated, persons with mildto moderate Alzheimer’s disease, the treatment ratewas 34.5%.

Comparison of treatment groups

Residents taking ACI’s and those not on the drugs werecompared on the basis of sex, age, baseline MMSE,GMHR, number of non-psychiatric medications, NPI,CSDD, and months since admission. The comparisonswere done separately for AD and non-AD dementia.Results are summarized in Tables 2 and 3. For those withAD, treated residents differed significantly from un-treated only by age (85 vs 88 years) and duration ofresidence (19 vs 27 months). For those with non-ADdementia, none of the differences were significant.

6 month outcomes

Residents with AD on ACIs had more favorableoutcomes at 6 months. Of the 54 not taking ACIs 41were still residing in their original facilities, one inanother AL, one in a general hospital, five in nursinghomes, and six had died, for a 77.8% favorableoutcome rate and 11.1% death rate. Of the 25 taking

CI� (n¼ 54) t or x2 df p

n SD

ale — 2.204 1 0.1382 6.41 �2.222 77 0.0297 6.90 1.701 77 0.093

0.63 1.131 36 0.2662.00 0.676 77 0.501

4 16.59 �0.646 77 0.5203.87 �1.498 76 0.138

9 23.36 �2.177 77 0.033


DOI: 10.1002/gps

Table 3. Comparison of residents with non-AD dementia by treatment group

ACIþ (n¼ 16) ACI� (n¼ 39) t or x2 df p

Mean SD Mean SD

Sex 11% male — 38% male — 1.725 1 0.189Age 86.51 5.60 83.52 7.45 1.439 53 0.156MMSE 14.67 8.41 17.92 7.24 �1.416 52 0.163GMHR 2.06 0.44 2.38 0.75 �1.976 46 0.054No. Meds 5.31 1.54 4.69 2.90 1.028 49 0.309NPI 9.13 9.99 13.26 14.18 �1.060 53 0.294CSDD 5.56 4.50 6.44 5.55 �0.558 53 0.579Length of Res. 16.00 14.58 27.30 21.14 �1.950 53 0.057


ACIs, 22 were in their original facilities, one inanother AL, one had gone home with family, one wasin a nursing home, and none had died, a 92% favorablerate. The association with mortality was of marginalsignificance ( p¼ 0.093 by Fisher’s exact test).

For residents with non-AD dementia, those onACIs actually did somewhat worse. At 6 months thefavorable rate for untreated residents was 76.3%, witha 10.5% death rate. For those on the medications thefavorable rate was 62.5% with a 25% death rate.Gender and facility size showed no significant asso-ciation with outcome for either type of dementia. Weconsidered eight other factors as potential predictorsof a outcome, cholinesterase inhibitor use and sevenpotential confounds: General Medical Health Rating,age, baseline MMSE, NPI, CSDD, number ofnon-psychiatric medications, and duration of resi-dence.

The results of a chi-square or t-test of each of thesepredictors for AD and non-AD dememtia at 6 monthsare presented in Table 4. For those with AD, GMHRshowed an association that was near significant(t¼�1.959, df¼ 76, sig¼ 0.054), but only ACI usewas significantly associated with favorable outcome at6 months (chi-square¼ 3.900 sig¼ 0.048). The rela-tive risk of unfavorable outcome for those on ACIs

Table 4. Association between individual variables and favorable ou

Subset Alzheimer’sDementia

Predictor x2* or t df Significa

ACI 3.900* 1 0.048MMSE �1.216 76 0.228GMHR �1.959 76 0.054NPI �1.318 45 0.194CSDD 1.076 75 0.285AGE 1.624 76 0.109LOS 1.509 13 0.155No. Meds �.885 76 0.379


was 0.217. Neither of the two variables by whichAD residents differed according to treatment group,namely age and duration of residence, was itselfsignificantly associated with favorable outcome at6 months. For those with non-AD dementia, ACIsshowed no statistical effect (chi-square¼ 1.071, sig¼0.301). Instead only baseline MMSE was significantlyassociated with 6-month outcome (t¼�3.48 df¼ 51sig¼ 0.001).

Time to discharge

In the survival analysis, 65% (49/75) of residents withAD discharged to higher level care or died. The meansurvival from the time of assessment was 1.76 years.ACI use was associated with retention in residentswith AD. Thirty-eight of 52 (73%) residents with ADwho were not taking ACIs at baseline discharged tohigher level care or died during the observation periodcompared to 11/23 (48%) who were taking ACIs (p¼0.03). Residents with AD taking ACIs at baseline alsoremained in assisted living significantly longer thanthose not on the agents (Log-rank test, p¼ 0.0095). Themedian number of days remaining in assisted living forthose on ACIs was 666.00 days (SD¼ 438.80) vs 437.25(SD¼ 346.01) in those who were not, with a differenceof 228.75 days between groups. These results are

tcome at 6 months

Non-Alzheimer’sDementia

nce x2* or t df Significance

1.071* 1 0.301�3.48 51 0.001�0.514 52 0.609

0.133 52 0.8941.165 52 0.2490.552 52 0.5970.924 52 0.360

�0.702 52 0.486


DOI: 10.1002/gps

Figure 1. Survival analysis of Assisted Living residents with Alzheimer’s disease.


depicted in Figure 1. ACIs were not associated withincreased AL retention in the non-AD residents.

DISCUSSION

Individuals with AD who were taking ACIs showedbetter retention at 6 months than did those not takingthem. ACIs were also associated with almost 8additional months of AL residence from studybaseline. A ‘preservation of function effect’ for ACI’sof between 6 and 12 months would be in keeping withthe results of published studies (Mohs et al., 2001;Geldmacher et al., 2003).

We did not see an association of ACIs with retentionfor residents who had non-AD dementia. This is notsurprising, although reports suggest these agents mayhelp some patients with vascular dementia (Wilkinsonet al., 2003). Baseline MMSE was associated withfavorable outcome only in the non-AD cases. Thismight be because the MMSE scores in the AD caseswere modified by the ACIs, but we can only speculatein the absence of longitudinal data.


The fact that this study was part of a larger projectintroduced several methodological limitations. ACItreatment was neither randomized nor controlled.Residents treated with AD on ACIs were younger thanthose not prescribed the agents, which could havecontributed to a more favorable outcome, although agewas not independently associated with outcome in ouranalysis. Residents with AD on ACIs had also beenmore recently admitted to their facilities. This mayreflect the growing popularity of these agents, or thetendency of physicians to discontinue them after agiven amount of time, but once again, duration ofresidence was not associated with outcome. Treatedand untreated residents did not differ significantly onGMHR or number of non-psychiatric medications, butresidents with AD showed a trend toward differentialmortality on the basis of treatment. This could reflectdifferences in health status not captured by the othervariables.

The relatively small numbers of participants did notpermit a multivariate analysis. On the other hand,because of the institutional setting, it was unlikely thatindividuals on ACI’s would have better compliance or


DOI: 10.1002/gps


receive more healthcare than other residents, a limi-tation of another study that has examined long-termACI use (Geldmacher, 2003). While there was noplacebo arm, residents with dementia were typicallyunaware that they were being given a medication forcognition. We also chose an outcome measure, ratesof death or discharge, which seems unlikely to beinfluenced by the expectations of the prescribingphysician or the experimenter.

We do not have information on continued ACI usebeyond the baseline visit. Furthermore, to preservepower, and to use a more naturalistic cohort, we didnot limit the study population to those taking anadequate dose of an ACI or those in the FDA indicatedrange of mild to moderate dementia. These threelimitations could actually obscure an effect of ACI useon retention, and thus the agents may have been moreadvantageous than they appear.

The Maryland Assisted Living Study is currently inits second phase. This phase involves recruiting ALresidents within 6 months of entry into their facility,with follow-up at 6 months intervals, allowing us togather longitudinal data. We will be able to validatethe findings in this report in a new cohort and to test forthe ongoing effects of factors such as dose andduration of treatment. A placebo-controlled trial in anAL population would be the gold standard.

Our findings have a number of potential clinicalimplications. Treatment rates with ACIs were low, evenfor the population for whom these agents are FDAindicated. There is, as yet, no widespread consensus onhow to use these agents, but our findings would seem tosupport the idea of routine use in AL, in cases of AD. Wedid not demonstrate beneficial effects in dementias otherthan AD, which suggests that clinicians should paycloser attention to diagnostic criteria. The existence ofAL as a lower cost alternative to a nursing home, theprevalence of dementia, and the tendency of ALresidents to transition to a higher level of care makes thisfertile ground for the study of agents which mitigate theeffects of AD.

CONFLICT OF INTEREST

Dr Rosenblatt has served as consultant for Pfizer andNovartis, has received speaking honoraria from Pfizerand Eisai, and has received research support fromForest Laboratories, Amarin, and Astra-Zenica.Dr Lyketsos has served as a consultant for Pfizer,Eisai, Novartis, and Forest Laboratories, has receivedspeaker’s honoraria from Pfizer, Eisai, Novartis, ParkDavis (Warner-Lambert), and Forest Laboratories, andhas received research support from Pfizer/Eisai, Park


Davis, and Forest Laboratories. The other authorshave no conflicts of interest.

ACKNOWLEDGEMENTS

This study was supported by a grant from the NationalInstitutes of Mental Health and National Institute onAging, RO1 MH 60626.

We are grateful to the study participants and theirfamilies for their participation, to the management andstaff of the participating assisted living facilities,and to the staff at Copper Ridge for their dedicationand assistance in the development of the study.

REFERENCES

2000 Collaborative Group. 2004. Long-term donepezil treatment in565 patients with Alzheimer’s disease (AD2000): randomizeddouble-blind trial. Lancet 363: 2105–2115.

Alexopoulos GS, Abrams RC, Young RC, et al. 1988. CornellScale for Depression in Dementia. Biol Psychiatry 23: 271–284.

Bernard SL, Zimmerman S, Eckert JK. 2001. Aging in place. InAssisted Living, Zimmerman S, Sloane PD, Eckert JK (eds).Johns Hopkins University Press: Baltimore, MD; 224–241.

Corey-Bloom J, Anand R, Veach J, for the ENA 713 B352 Study.1998. A randomised trial evaluating the efficacy and safety ofENA 713 (rivastigmine tartrate), a new acetylcholinesteraseinhibitor, in patients with mild to moderately severe Alzheimer’sdisease. Int J Ger Psychopharmacol 1: 55–65.

Cummings JL, Mega M, Gray K, et al. 1994. The NeuropsychiatricInventory: comprehensive assessment of psychopathology indementia. Neurology 44: 2308–2314.

Cummings JL, Anand R, Koumaras B. 2000a. Rivastigmineprovides behavioral benefits to Alzheimer’s disease patientsresiding in a nursing home: findings from a 26- week trial.Neurology 54: A468.

Cummings JL, Donohue JA, Brooks RL. 2000b. The relationshipbetween donepezil and behavioral disturbances in patients withAlzheimer’s disease. Am J Geriatr Psychiatry 8: 134–140.

Folstein MF, Folstein SE, McHugh PR. 1975. ‘Mini-Mental State’:a practical method for grading the cognitive state of patients forthe clinician. J Psychiatric Res 2: 189–198.

Geldmacher DS, Provenzano G, McRae T, et al. 2003. Donepezilis associated with delayed nursing home placement in patientswith Alzheimer’s disease. J Am Geriatr Soc 51: 937–944.

Hendrie HC. 1998. Epidemiology of dementia and Alzheimer’sdisease. Am J Geriatr Psychiatry 6(2 Suppl. 1): S3–S18.

Holmes C, Wilkinson D, Dean C, et al. 2004. The efficacy ofdonepezil in the treatment of neuropsychiatric symptoms inAlzheimer disease. Neurology 63: 214–219.

Lopez OL, Becker JT, Saxton J, et al. 2005. Alteration of aclinically meaningful outcome in the natural history of Alzhei-mer’s disease by cholinesterase inhibition. J Am Geriatr Soc 53:83–87.

Lyketsos CG, Galik E, Steele C, et al. 1999. The ‘General MedicalHealth Rating’ (GMHR): a bedside global rating of medicalcomorbidity in patients with dementia. J Am Geriatr Soc 47:487–491.

McKeith IG, Perry RH, Fairbairn AF, et al. 1992. Operationalcriteria for senile dementia of Lewy body type. Psychol Med 22:911–922.


DOI: 10.1002/gps


Mohs RC, Doody RS, Morris JC, et al. 2001. A 1-year,placebo-controlled preservation of function survival study ofdonepezil in AD patients. Neurology 57: 481–478.

McKhann G, Drachman D, Folstein MF, et al. 1984. Clinicaldiagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA work group. Neurology 34: 939–949.

NINDS/AIREN International Workshop. 1993. Vascular dementia:diagnostic criteria for research studies. Neurology 43: 250–260.

National Center for Assisted Living. 2001. Facts and Trends: TheAssisted Living Sourcebook. NCAL.

Neary D. 1990. Dementia of the Frontal Lobe type. J Am GeriatrSoc 38: 71–72.


Rosenblatt A, Samus QM, Steele C, et al. 2004. The MarylandAssisted Living Study: prevalence, recognition and treatment ofdementia and other psychiatric disorders in the assisted livingpopulation of central Maryland. J Am Geriatr Soc 52: 1618–1625.

Tariot PN, Solomon PR, Morris JC, et al. 2000. A 5-month,randomized, placebo-controlled trial of galantamine in AD. TheGalantamine USA-10 Study Group. Neurology 54: 2269–2276.

Wilkinson D, Doody R, Hekme R, et al. and the Donepezil 308Study Group. 2003. Donepezil in vascular dementia: a random-ized, placebo-controlled study. Neurology 61: 479–486.


DOI: 10.1002/gps

Documents

Acetylcholinesterase inhibitors in assisted living: patterns of use and association with retention