The Gerontologist Copyright 2005 by The Gerontological Society of America

Vol. 45, No. 6, 754–763

The Effect of Adult Day Care Services on

Time to Nursing Home Placement in
Older Adults With Alzheimer’s Disease
Judith J. McCann, DNSc,1 Liesi E. Hebert, ScD,1
Yan Li, PhD,1 Fredric D. Wolinsky, PhD,2 David W. Gilley, PhD,1
Neelum T. Aggarwal, MD,3 Joanne M. Miller, PhD,4 and Denis A. Evans, MD1

Purpose: This longitudinal study examined whether
the use of adult day care services delayed time to
nursing home placement in persons with Alzheimer’s
disease. Design and Methods: Two hundred and
eighteen adult day care clients with Alzheimer’s
disease were recruited from 16 adult day programs in
a large metropolitan area. Two hundred and ninety
eight persons with Alzheimer’s disease but not using
adult day care were recruited from a federally funded
Alzheimer’s diagnostic center and frequency
matched to adult day care users on age, gender,
race, and level of cognitive impairment. Participants
were followed at 3-month intervals for up to 48
months. Cox proportional hazards models were used
to examine the effects of adult day care and other
fixed and time-varying factors on risk of nursing home
placement. Results: Risk of nursing home placement
increased significantly with the number of days of
adult day care attendance, with this effect being
substantially greater for men (hazard ratio or HR =
1.33; confidence interval or CI = 1.18–1.49) than
for women (HR = 1.09; CI = 1.00–1.18). Participant
disability and hospitalizations and caregiver age and
This study was supported by Grants R01 AG10315 and R01 AG09966
from the National Institute on Aging, National Institutes of Health. We
thank the study participants and their family members for their time and
commitment to this research project. We also thank Melinda Scheuer and
her staff of research assistants for data-collection activities, Woojeong
Bang and her staff for assistance with analytic programming and analysis,
and George Dombrowski and his staff for data management.
Address correspondence to Judith J. McCann, Rush Institute for
Healthy Aging, Rush University Medical Center, 1645 W. Jackson Blvd.,
Suite 675, Chicago, IL 60612. E-mail: judy_j_mccann@rush.edu
1
Rush Institute for Healthy Aging, Rush University Medical Center,
Chicago, IL.
2 and role overload (Gaugler, Jarrott, et al., 2003;
Department of Health Management and Policy, the Center for
Research in the Implementation of Innovative Strategies in Practice and
the University of Iowa, Iowa City.
3
Department of Neurological Sciences and Rush Alzheimer’s Disease
Center, Rush University Medical Center, Chicago, IL.
4
College of Nursing, Rush University Medical Center, Chicago, IL.
burden were independent predictors, but their in-
clusion in the model did not alter the risk associated
with adult day care. Implications: More severe
disease and greater caregiver burden did not explain
the increased risk of nursing home placement among
adult day care users with Alzheimer’s disease.
Rather, it appears that other unmeasured factors,
such as a proclivity to institutionalize, may account for
the association of adult day care to nursing home risk.
Key Words: Community-based services,
Institutionalization, Longitudinal study,
Survival analysis
Approximately 4.5 million individuals in the
United States have Alzheimer’s disease, the most
common form of dementia. By 2050, this number is
predicted to increase to 13.2 million (Hebert, Scherr,
Bienias, Bennett, & Evans, 2003). Because most
individuals with dementia will be institutionalized
before they die (Smith, Kokmen, & O’Brien, 2000),
and because of the potential cost savings associated
with delaying nursing home placement (Leon, Cheng,
& Neumann, 1998), interventions aimed at prevent-
ing or delaying placement have been the focus of much
research. One such intervention consists of adult day
care services, which are community-based programs
that provide therapeutic and social activities, health
and personal care services, meals, transportation, and
family education and support. Research documents
that adult day care reduces caregiver psychological
distress (Kosloski & Montgomery, 1995; Wimo et al.,
1990; Zarit, Stephens, Townsend, & Greene, 1998)
Gaugler, Kane, Kane, Clay, & Newcomer, 2003),
particularly when it is used on a regular and sustained
basis, and it may facilitate continued family caregiv-
ing. Although anecdotal reports suggest that adult day

754 The Gerontologist

Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
 care services decrease institutionalization, few studies
have examined this issue; those that have examined it
have reported mixed results.
Of four randomized controlled trials of adult day
care or other forms of respite, two reported no
association (Hedrick et al., 1993; Weissert, Wan,
Livieratos, & Katz, 1980), one found a small delay in
nursing home placement (Lawton, Brody, & Saper-
stein, 1989), and one found increased risk of
placement among spousal caregivers only (Mont-
gomery & Borgatta, 1989). Although randomized
controlled trials are generally most informative, the
interventions in these studies consisted only of
offering adult day care and other forms of respite.
As a result, participants randomized to treatment and
control groups differed little in the amount of respite
or adult day care they used, and services were used at
very low rates. An actual treatment reanalysis of one
randomized controlled trial reported that those with
high respite use were less likely to enter a nursing
home than nonusers (Kosloski & Montgomery, 1995).
Three observational studies examined the effect of
adult day care on nursing home placement in
individuals with dementia. One small study (n = 47;
Wimo et al., 1990) reported a lower risk of institu-
tionalization with adult day care use than without it,
although that study did not include a nonuser group
and used a definition of institutionalization other than
long-term nursing home placement. Two larger
longitudinal studies, one of approximately 400 people
(Gaugler & Zarit, 2001) and one of 3,944 participants
in the Medicare Alzheimer’s Disease Demonstration
Evaluation (Gaugler, Kane, et al., 2003), found that
adult day care use significantly increased the risk of
nursing home placement. Investigators postulated
that families did not use enough adult day care for it to
be effective, and they waited too long in the disease
process to begin adult day services. Thus, adult day
care functioned more as a transition to nursing home
placement than as a form of respite (Gaugler & Zarit;
Zarit, Stephens, Townsend, Greene, & Leitsch, 1999).
Several studies concluded that research on the
efficacy of formal community-based services should
focus on the use of a specific service by a defined
subgroup of elderly persons. The research reported
here closely followed individuals with Alzheimer’s
disease who were adult day care users and compa-
rable nonusers for up to 4 years. By focusing on
a defined service and a group at high risk for nursing
home placement, and by designing the study to
ensure high rates of adult day care attendance, we
intended to provide a rigorous test of the efficacy of
adult day care to prevent or delay institutionaliza-
tion. Using the behavioral model of health services
utilization (Andersen, 1995) as our conceptual
framework, we measured changes in adult day care
use and other predisposing, enabling, and need
variables associated with nursing home placement
every 3 months. We hypothesized that, after we
adjusted for other known predictors of nursing home
Vol. 45, No. 6, 2005
Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
placement in people with Alzheimer’s disease, the
use of adult day care would increase time to nursing
home placement in a dose-response fashion.
Methods
Participants
Study participants were recruited from 16 adult
day care service sites in metropolitan Chicago and
from a federally funded Alzheimer’s diagnostic
center, the Rush Alzheimer’s Disease Center. All
participants were required to be 65 years of age or
older, to meet standard diagnostic criteria for
Alzheimer’s disease, and to reside in a community
setting, excluding assisted living or nursing care
facilities. Pilot data from participants (n = 90) in
seven adult day care programs determined that the
dropout from adult day care was highest during the
first 3 months of enrollment. Therefore, to maximize
exposure to adult day care and to minimize dropout,
we required participants to have used the service for
a minimum of 3 months at baseline and to have
attended the program at least 2 days a week during
the previous month. Of 733 persons screened in adult
day care service sites, 297 (40.5%) were eligible and
218 (73.4%) agreed to participate. We recruited
a cohort of persons with Alzheimer’s disease not
using adult day services from the Rush Alzheimer’s
Disease Center. Because our pilot data revealed
differences between the two groups on some factors
strongly related to nursing home placement, we
frequency matched nonusers to day care users by age
group (65–74, 75–85, and 85þ years), gender, race,
and level of cognitive impairment (mild, moderate,
and severe). Of 698 persons screened, 415 (59.5%)
were eligible and 298 (71.8%) agreed to participate.
Consent procedures were approved by the In-
stitutional Review Board at Rush University Medical
Center, and we obtained signed consent from each
participant with Alzheimer’s disease and a family
member. The participating family member was the
individual who had the most face-to-face contact
with the participant and who provided the most
care. This individual, identified as the caregiver,
provided baseline and follow-up data about the
participant and for the caregiver measures.
Measurement Procedures
The diagnosis of possible, probable, or highly
probable Alzheimer’s disease was made by a board-
certified neurologist using criteria of the joint
working group of the National Institute of Neuro-
logic and Communicative Disorders and Stroke and
Alzheimer’s Disease and Related Disorders Associa-
tion (McKhann et al., 1984). We obtained medical
history as well as data on participant and caregiver
demographic, social, and health variables from an
755
 interview with the caregiver at baseline. Interviews
were repeated either in person or by telephone at
3-month intervals for up to 48 months. Tests of
participant cognitive function were administered in
person at baseline and every 6 months during the
observation period.
Trained research technicians conducted interviews
and cognitive function tests. We specified all data-
collection procedures in a project manual, and we
used weekly staff meetings to address specific
questions about data collection. The research
technicians collected data by using computer-assisted
interview procedures, and we monitored interrater
reliability every 6 months with retraining provided as
needed to maintain agreement at 90% or better.
Average interrater reliability on all measures was
98.6% (range = 97.65–99.78%).
Measures
Time to Nursing Home Placement.—Research
technicians obtained the date of placement from the
caregiver and verified it with the facility. We defined
the time to nursing home placement as the interval
from date of study enrollment to date of nursing
home admission for permanent custodial care. We
did not consider short-term stays for rehabilitative
purposes to be nursing home placements.
Use of Adult Day Care Services.—At each
interview, we recorded any use of adult day services
over the previous 3 months (0–1) and the average
weekly number of days of attendance over the pre-
vious month (0–7). Because of the possibility of
crossover between the adult day care user and
nonuser groups, we obtained this information for
both groups.
Use of Other Community-Based Services.—We
recorded the use of visiting nurse, personal care, and
homemaker services at each follow-up. We summa-
rized each service as any use over the past 3 months
(0–1) and as a categorical variable (1–5) representing
frequency of use over the past 3 months, from less
than once a month to 5 or more times a week. We
also created summary variables of the total number
of services used (0–3) and the total frequency of
service use (Sum of Service 3 Frequency; range = 0–
15) over the past 3 months.
Financial Measures.—At baseline, caregivers
reported the participants’ total annual family income
using 10 income categories and the show-card
method employed in the Established Populations
for Epidemiologic Studies of the Elderly projects
(Cornoni-Huntley, Brock, Ostfeld, Taylor, & Wal-
lace, 1986). We also recorded ownership of long-
term-care insurance as a dichotomous variable.
756
Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
Participant Measures.—At baseline, we obtained
data on participant age, gender, marital status, race,
education, and living arrangement. We analyzed age
as a continuous variable, and we analyzed marital
status as married or not married. We summarized
race as Black or White (and others) because less than
2% of the participants were some other race. We
assessed education in terms of years of completed
schooling and analyzed it as a continuous variable.
We characterized living arrangement as living alone
or living with others, and we determined it at
baseline and each follow-up.
At each 3-month follow-up with the caregiver, we
collected data on participant physical function by
using three standard physical function or disability
scales (Katz & Akpom, 1976; Nagi, 1976; Rosow &
Breslau, 1966). For each measure, we summed items
to produce a score ranging from 0 (no disability) to
the maximum score indicating disability on all items
(6 for the Katz, 5 for the Nagi, and 3 for the Rosow–
Breslau). Baseline internal consistency reliability
coefficients for the three disability scales ranged
from .74 to .89. We also collected data on number of
hospitalizations and the occurrence of positive (0–5)
and negative (0–5) behaviors (Gilley et al., 2004) and
urinary or bowel incontinence. At baseline and every
6 months, we assessed cognitive impairment by using
the Mini-Mental State Examination (MMSE; Fol-
stein, Folstein, & McHugh, 1975), a 20-item measure
of orientation, memory, attention, language, and
visual-spatial abilities. The baseline internal consis-
tency reliability coefficient for this measure was .92.
Caregiver Measures.—At baseline, we collected
data on caregiver age, gender, marital status, race,
education, relationship to the participant with Alz-
heimer’s disease, and the number of years the
caregiver had been providing care (duration of care-
giving). We analyzed years of age, education, and
caregiving as continuous variables. The coding of
marital status and race was identical to the partici-
pant data. We coded the relationship to the partic-
ipant as spouse, child, or other.
Research technicians interviewed caregivers every
3 months by using standard measures. We computed
time spent in caregiving each week by multiplying the
average weekly days of care by the average daily
hours of care (range=0–168). We coded employment
status as currently employed or not employed, and if
caregivers were employed, we recorded the average
weekly hours of work. We also administered
a modified 7-item measure (Pearlin, Mullan, Semple,
& Skaff, 1990) of the extent to which caregiving
interfered with the caregivers’ work, with higher
scores indicating more interference. We assessed
caregiver negative and positive affect with self-report
measures that used the past week as the reference
period. We measured depressive symptoms with the
10-item short form of the Center for Epidemio-
logic Studies Depression scale (Kohout, Berkman,
The Gerontologist
 Evans, & Cornoni-Huntley, 1993), with higher scores
indicating greater depressive symptoms. We mea-
sured positive affect with the 10-item positive mood
scale of the Positive and Negative Affect Scale
(Watson, Clark, & Tellegen, 1988), with higher
scores indicating more positive affect. We measured
caregiving specific negative and positive appraisals
with a 10-item Subjective Caregiving Burden scale
and a 5-item Caregiving Satisfaction scale (Lawton,
Moss, Kleban, Glicksman, & Rovine, 1991). The
burden and satisfaction scales used a 4-point Likert
response format, with higher scores indicating greater
subjective stress and greater personal satisfaction
associated with providing care.
We assessed self-reported health with three items
from the Health-Related Quality of Life measure
(Hennessy, Moriarty, Zack, Scherr, & Brackbill,
1994). Caregivers rated their overall health as poor,
fair, good, or excellent, and they estimated the
number of days in the previous month that their
physical health was not good (0–30) and their mental
health was not good (0–30). We combined responses
to these last two questions to calculate a summary
index of overall unhealthy days (Centers for Disease
Control and Prevention, 2000), with a logical
maximum of 30 unhealthy days. We measured
physical function and disability with the same Nagi
(1976) and Rosow-Breslau (1966) disability scales
used with the participant.
We measured social support with the Perceived
Social Support Scale (Zimet, Powell, Farley, Werk-
man, & Berkoff, 1990), a 12-item measure of
perceived availability and satisfaction with support
received from family, friends, and a ‘‘special person.’’
Higher scores on this measure indicate greater
perceived support. We assessed spirituality with 6
items that measure religious coping, support, beliefs,
and practices (Fetzer Institute, 1999). Although the
measure taps various aspects of spirituality, we
treated the scale as unidimensional, and it had a
Cronbach’s coefficient of a = 0.92 in our sample.
Higher scores on this measure indicate greater
spirituality. The baseline internal consistency reli-
ability coefficients for the caregiver measures used
in this study ranged from .74 to .92, with one ex-
ception of .57 for the scale measuring caregiving–
work interference.
Statistical Analysis
We compared time to nursing home placement
between participants who used adult day care and
those who did not. We used Kaplan–Meier survival
curves to graphically display the association between
adult day use and nursing home placement, and we
used Cox proportional hazards modeling to examine
the simultaneous effects of adult day care use and
other fixed and time-varying predictors on institu-
tionalization risk. Our model development was
Vol. 45, No. 6, 2005
Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
based on the behavioral model of health services
utilization (Andersen, 1995) and proceeded in four
steps. We first examined the relation of institution-
alization risk to participant age, gender, race, and
time-varying MMSE score, and in separate models
adjusting for the main effects of these variables, we
examined all interactions among these variables.
Second, we examined the effect of adult day care use,
adjusting for age, gender, race, and time-varying
MMSE score, and in separate models controlling for
the main effects of these variables, we examined the
interaction of adult day care use with each of these
variables. We quantified adult day care use in two
ways: (a) as a dichotomous variable reflecting adult
day care user and nonuser groups as constituted at
baseline (0–1), and (b) as a continuous and time-
varying variable reflecting the weekly average days of
adult day care received over the previous month (0–
7). The results reported in this article are based on
the latter approach, which allowed us to adjust for
the initiation of adult day care by participants who
were nonusers at baseline and to examine the dose
effect of adult day care.
Third, in separate stepwise selection models that
adjusted for participant age, gender, race, time-
varying MMSE score, and time-varying adult day
care use, we examined groups of predisposing, en-
abling, and need variables from the behavioral model
known to be associated with nursing home placement,
and we tested for interaction of adult day care use with
the significant predictors in each group. Fourth, we
used one stepwise selection model adjusted for
participant age, gender, race, time-varying MMSE
score, and time-varying adult day care use to select the
significant predictors from among all the significant
predictors in the separate stepwise selection models.
This provided a parsimonious final model of pre-
dictors of nursing home placement in this cohort of
people with Alzheimer’s disease.
Sensitivity Analysis for Differences Between
Adult Day Care User and Nonuser Groups.—A
major concern in any observational study is the pos-
sibility of selection bias (D’Agostino, 1998; Rubin,
1979). That is, despite our matching efforts, differ-
ences in the comparison groups might still affect the
study findings. We addressed this in two ways. First,
following standard procedures (D’Agostino; Rubin),
we used multivariable logistic regression to model
whether the participant was in the adult day care user
versus nonuser group at baseline. This model in-
corporated all measured variables at baseline. We
then included the predicted probability of group
status (i.e., the propensity score) as a predictor of risk
of nursing home placement in a Cox proportional
hazards model. This facilitated a simple and direct
comparison of the two groups in this observational
study. Second, we reran our final model, including 11
additional variables for which there were significant
group differences at baseline.
757
 Results
We followed study participants for up to 48
months, with an average of 85% of survivors
participating at each follow-up (range = 77% to
100%). At the close of the study, 181 participants
(35%) had transitioned from the community to
a nursing home facility; 107 participants (21%) died
while living in the community; and 228 participants
(44%) were alive and still living in the community.
The average time on study for the 228 community-
living participants was 3.6 years (SD = 0.56) for
those using adult day care at baseline, and 3.4 years
(SD = 0.70) for those not using day care at baseline.
At baseline, adult day care users (n = 218) had
been enrolled in adult day care for an average of 2.0
years (SD = 1.9) and were attending adult day care
an average of 3.9 days per week (SD = 1.2). Over the
entire 48-month observation period, the average
weekly attendance was 3.6 days (SD = 1.4). Among
participants not using adult day care at baseline (n =
298), 13% reported prior use and 16% used adult
day care during the study period.
Although we frequency matched adult day care
nonusers to users by age and MMSE groups, we
found small but statistically significant differences
for the continuous measures of these two variables
(Table 1). However, there was sufficient distribution
of age and MMSE scores in each group to permit
adjustment in analyses. There were no significant
differences in the matching variables of participant
gender or race. The distributions of the other
analytic variables are shown in Table 1. Compared
with nonusers, adult day care users had significantly
lower education and household income, and they
were less likely to be married or living alone. Adult
day care users had higher functional limitation and
disability scores, but they used significantly fewer
community-based services at baseline.
Caregivers of participants using adult day care
were younger, less likely to be married or male, and
less likely to be spouses of care recipients than those
not using this service. They also had been providing
care significantly longer and reported greater burden
at baseline than did caregivers of nonusers. There
were no significant differences between the two
groups of caregivers with respect to the percentage
employed, the number of hours worked each week,
or weekly hours of care provided.
Table 2 displays descriptive data on the 16 adult
day care programs. All programs were nonprofit, and
private pay and the Illinois Department on Aging
were the largest payment sources. All programs
provided assistance with personal care, oral medica-
tion administration, and blood pressure and weight
monitoring, and about half of the programs provided
more extensive health and rehabilitation services. Pro-
grams did not vary in terms of the types of activities
offered to clients, but they did vary on services offered
to families. About half provided support groups,
758
Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
counseling, and case management; none offered
overnight or weekend respite. Admission and dis-
charge criteria were similar across programs. The
most common reasons for not admitting or for
discharge were uncontrollable disruptive behavior
and the need for one-to-one supervision. Programs
differed most on whether they could handle partic-
ipants who were incontinent or were highly de-
pendent in activities of daily living.
Cox Proportional Hazards Models
In a model that included only baseline group status
(user or nonuser) and the selection bias propensity
score, the risk of nursing home placement was
significantly greater in the adult day care group
(hazard ratio or HR = 2.3; p = .001), and the
propensity score was not significant (HR = 0.83; p =
.56). In subsequent models, we used a continuous
variable of actual days of day care attendance
measured at each follow-up rather than a dichoto-
mous indicator of baseline group status. In a model
(Table 3, Model A) that included participant age,
gender, race, time-varying MMSE score, and days of
adult day care, the risk of nursing home placement
was significantly greater for men than for women,
and it increased significantly with older age and more
days of attendance at adult day care. Risk was lower
for Black participants and those with better cognitive
function (higher MMSE score) than it was for non-
Black participants and those with lower cognitive
function. There were no interactions among the
demographic and cognitive variables (data not
shown). In tests for interaction of demographic and
cognitive variables with adult day care attendance,
only gender was significant, indicating that nursing
home risk increased more for men who used more
days of adult day care than for women who did so
(Table 3, Model B).
We also considered other characterizations of
adult day care use. These included a yes–no indicator
of day care use at each follow-up and adult day care
exposure of 3 months and 6 months before the
observation time. These were somewhat less power-
ful predictors but yielded similar results. We
examined the effect of duration of adult day care
use at baseline and its interaction with days of use at
follow-up, but neither of these was a significant
predictor of nursing home risk, nor was a history of
adult day care use among the nonusers at baseline.
We considered Model B in Table 3 our base model
and tested the effect of additional participant and
caregiver variables on risk of nursing home place-
ment by adding them to this base model. We grouped
these variables into categories reflecting predisposing,
enabling, and need variables specified in the behav-
ioral model of health services utilization (Andersen,
1995). Predisposing variables influence the tendency
to use services; enabling variables influence access to
The Gerontologist
 Table 1. Distribution of Baseline Variables by Day Care Use

M (SD) Percent

Day Care No Day Care Day Care No Day Care

Variable (Range) n ¼ 218 n ¼ 298 n ¼ 218 n ¼ 298 pa
Participant Characteristics
Age 81.5 (6.4) 79.3 (6.5) .0001
MMSE score (0–30) 12.2 (7.5) 14.2 (8.3) .004
Education 10.9 (3.8) 12.2 (3.4) .0001
Katz disability score (0–6) 2.2 (1.8) 1.8 (2.2) .04
Nagi disability score (0–5) 2.3 (1.7) 1.8 (1.8) .002
Rosow-Breslau disability score (0–3) 1.7 (.93) 1.5 (1.1) .03
Negative behavior score (0–5) .70 (.85) .81 (.97) .16
Positive behavior score (0–5) 3.5 (1.6) 3.3 (1.7) .22
Male gender 30.3 32.2 .64
Black race 31.7 25.2 .10
Married 24.4 53.7 .0001
Living alone 6.4 13.1 .01
Any urinary incontinence 57.1 49.0 .07
Any bowel incontinence 29.9 24.3 .24
Caregiver Characteristics
Age 58.2 (12.3) 62.2 (14.3) .0009
Education 14.0 (2.7) 14.5 (2.9) .06
Hours of paid work each week (0–80) 20.9 (21.3) 17.9 (21.6) .12
Hours of care each week (0.5–168) 51.4 (51.8) 49.5 (58.7) .71
Years of caregiving (1–29) 4.4 (4.0) 3.4 (3.3) .003
Days of poor health (0–30) 9.1 (10.6) 8.1 (10.6) .27
Nagi disability score (0–5) .63 (1.1) .56 (1.1) .51
Rosow-Breslau disability score (0–3) .27 (.67) .32 (.70) .44
Depressive symptoms (0–10) 2.0 (2.1) 2.0 (2.1) .63
Positive affect (10–40) 28.5 (6.5) 29.4 (5.9) .07
Caregiving satisfaction (5–20) 14.5 (4.1) 15.0 (3.9) .14
Caregiving burden (10–40) 17.0 (6.3) 15.4 (5.5) .002
Social support (12–60) 24.8 (6.9) 25.5 (7.5) .31
Spirituality (6–30) 22.0 (6.8) 21.7 (6.6) .53
Male gender 12.4 33.6 .0001
Black race 32.3 24.9 .07
Married 65.4 74.8 .02
Relation to participant .0001
Spouse 20.6 42.3
Child 66.1 48.3
Other 13.3 9.4
Overall health rating .61
Excellent 28.2 32.7
Good 48.8 48.5
Fair 18.8 15.8
Poor 4.2 3.0
Working part time or full time 55.1 47.0 .07
Financial measures
Income category (1–10) b 3.7 (2.3) 5.1 (2.9) .0001
Own long-term-care insurance 7.4 10.4 .24
Community-based services
Homemaker services 10.8 20.3 .004
Personal care attendant 14.4 21.3 .045
Visiting nurse 1.4 9.5 .0002
Total no. of services used (0–3) .26 (.57) .51 (.77) .0001
Total frequency of service use (0–15) 1.1 (2.5) 2.0 (3.3) .0001
a
Fordifference between groups.
b
Income data missing for 10% of day care group and 15% of no day care group.

Vol. 45, No. 6, 2005 759

Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
 Table 2. Descriptive Data on 16 Adult Day Service Programs Table 3. Effect of Demographic Variables, MMSE Score, and
Day Care Use on Risk of Nursing Home Placement
Program Characteristic M (SD) Range
Model A Model B
Years of operation as a day
services program 14.1 (6.3) 1–23 Hazard Ratio Hazard Ratio
Years of service for current (95% Confidence p (95% Confidence p
director 7.4 (5.1) 1–15 Variable Interval) Interval)
Square footage of facility 4,120 (2,308) 1,355–9,500
Charge per day in dollars 40.82 (10.0) 25.00–52.00 Participant age 1.03 (1.00–1.05) .04 1.02 (1.00–1.05) .06
Total number of clients Male
enrolled in program 52.6 (19.3) 18–85 participant 1.45 (1.05–2.01) .03 0.95 (0.60–1.49) .81
Daily client capacity of Black
the program 43.6 (20.0) 18–75 participant 0.68 (0.46–0.96) .03 0.63 (0.44–0.90) .01
Average daily attendance 28.9 (13.7) 12–55 Time-varying
No. of different activities MMSE
offered from 7 am to 12 pm 5.1 (2.6) 2–12 scorea 0.98 (0.96–0.99) .01 0.97 (0.96–0.99) .005
No. of different activities offered 1 additional day
from 12 pm to 5 pm 4.6 (2.3) 2–10 of day care 1.14 (1.06–1.22) .0003
1 additional day
of day care
for women 1.09 (1.00–1.18) .05
services; and need variables refer to the health care 1 additional day
requirements of the individual with Alzheimer’s of day
disease and the primary caregiver. care for men 1.26 (1.13–1.42) , .00001
a
Higher scores indicate better cognitive function.
Predisposing Variables
care, and homemaker services at each follow-up by
We examined the influence of relation of the
using the individual and summary variables described
caregiver to the participant and the marital status of
in the Methods section. These service-use variables
both, using grouping variables of spouse (reference
neither significantly predicted the risk of nursing
category), married child, unmarried child, and other.
home placement nor altered the risk associated with
None of the latter three groups had a significantly
adult day care attendance.
different risk of nursing home placement than the
spouse category, and none affected the risk associ-
ated with adult day care attendance. We character-
ized living arrangement in two ways: (a) living alone Need Variables
versus living with others, and (b) living alone for all
follow-up observations, living with others previously We tested for Alzheimer’s disease participant
and then changing to living alone, or living with condition by using several time-varying measures. In
others. None of these characterizations of living stepwise analyses, with the Katz, Nagi, and Rosow–
arrangement was a significant predictor of institu- Breslau functional status and disability scales as
tionalization risk, and they did not affect the risk candidate variables, only the Rosow–Breslau scale
associated with adult day care. We also did not find entered the model. The risk associated with adult day
a difference in risk for the 10% of participants whose care attendance did not change with the Rosow–
caregiver changed during the course of follow-up. Breslau scale in the model. Hospitalization in the
previous 3 months and depressed mood significantly
increased the risk of nursing home placement, and
Enabling Variables higher levels of positive behavior significantly de-
creased the risk of placement. However, none of
About 10% of participants had long-term-care these variables altered the risk associated with adult
insurance. Neither this nor a 10-category measure of day care.
income group analyzed as a continuous variable We examined the effect of duration of caregiving
significantly predicted risk of nursing home place- on risk of nursing home placement. We also
ment. We then grouped income as $25,000 or more examined two other measures of time at risk: time
versus less than $25,000, which was the cutoff for since onset of any cognitive symptoms and time since
those whose adult day care was paid by Medicaid or onset of memory problems. None of these variables
Veterans Affairs (VA) benefits. We used a second in- was significant. Time spent providing care each
dicator variable for those whose income was un- week, employment and hours worked each week, the
known. Neither low income nor missing income was extent to which caregiving interfered with work, and
significant. To control for the potential effect of measures of caregiver depression, burden, positive
community-based services on nursing home place- affect, satisfaction, social support, and spirituality
ment, we examined the use of visiting nurse, personal were all considered as candidates in a stepwise

760 The Gerontologist

Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
 Table 4. Final Model of Effect of Day Care on Risk of
Nursing Home Placement

95%
Hazard Confidence
Variable Ratio Interval
Participant age 1.01 0.99–1.03
Male participant 0.73 0.45–1.19
Black participant 0.69 0.48–1.00
Participant MMSE scorea 0.98 0.96–1.00
Participant Rosow-Breslau
disabilitya 1.32 1.11–1.56
Participant hospitalizationsa 1.54 1.00–2.37
Caregiver age 1.01 1.00–1.03
Caregiver burdena 1.05 1.03–1.08
1 additional day of day care
for womena 1.09 1.00–1.18
1 additional day of day care
for mena 1.33 1.18–1.49
a
Time-varying measure at each follow-up.
analysis. Only caregiver burden entered the model.
Greater burden was associated with a significant
increase in risk of nursing home placement, but the
risk associated with adult day care remained the same.
We also examined caregiver age and several measures
of caregiver physical health as candidates in a stepwise
model. Only caregiver age was significant, and it did
not alter the risk associated with adult day care.
In individual models adjusting for age, gender,
race, time-varying MMSE score, and adult day care
use, we examined the interaction of day care use with
each of the significant predictors from the previous
models (participant Rosow–Breslau disability score,
depressed mood, positive behavior, and hospital-
izations, as well as caregiver age and burden).
Interactions of adult day care use with participant
hospitalizations and caregiver age were significant;
however, neither of these interactions was retained in
our final stepwise model. In our final stepwise model,
we forced in age, gender, race, time-varying MMSE
score, and adult day care use, and we considered as
candidates for selection all of the significant main and
interaction effects from previous models. Only
participant disability and hospitalizations and care-
giver age and burden entered the model as indepen-
dent predictors of nursing home placement. The
interaction of gender and adult day care use remained
significant in this final model (Table 4). Figure 1
graphically displays the results from this final model.
To test the robustness of our findings to potential
selection bias, we reran our final model and included
the propensity score as a covariate. The propensity
score was not significant and did not alter the relation
of adult day care use to nursing home placement. In
a separate analysis, we also reran our final model,
including 11 variables for which there were significant
group differences at baseline. Two variables were
significant (care recipient married and living alone),
but the increased risk of nursing home placement
p
.35
.20
.05
.09
.002
.05
.02
, .0001
.06
Figure 1. Risk of nursing home placement for men and
, .0001 women by days of day care attendance.
associated with adult day care use remained
(for women, HR = 1.10, p = .06; for men, HR =
1.20, p = .02).
Discussion
Contrary to our hypothesis, participants with
Alzheimer’s disease who used more days of adult
day care each week had an increased risk of nursing
home placement. This risk persisted despite consid-
eration of multiple indicators of disease duration and
severity and of caregiver burden and workload.
Previous studies have reported mixed results. One
large observational study of people with dementia
also found an increased risk of nursing home
placement among most adult day care users (Gaugler,
Kane, et al., 2003), but not in the dose-response
relation we found. Adjusting for patient and
caregiver characteristics, that study found increased
risk with both low and high amounts of day care use,
whereas intermediate use was not significantly
different from nonuse. Two randomized controlled
trials offering adult day care to one group found no
difference in risk of nursing home placement between
the treatment and control groups, but the average
attendance in the treatment group was low, only
about 1 day a week (Weissert et al., 1980; also see
Hedrick et al., 1993).
Some investigators have suggested that the reason
adult day care does not delay nursing home placement
is because caregivers wait too long to begin using it.
According to Gaugler and Zarit (2001) and Zarit and
colleagues (1999), the caregivers are so burdened and
the patient so severely impaired that the relief that
adult day care provides may actually expedite nursing
home placement. Zarit and colleagues found that
caregivers who used adult day care for 3 months or
less had higher baseline levels of role captivity and
were caring for relatives with more severe functional
and behavioral problems. More than one third of

Vol. 45, No. 6, 2005 761

Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
 these brief users immediately placed their relative in
a nursing home. This scenario is not a likely explana-
tion for our findings because our inclusion criteria
required a minimum of 3 months of adult day care use
at baseline. We also examined a large number of
measures of disease severity and caregiver burden,
and they did not change the risk associated with day
care attendance.
An alternative explanation is that some caregivers
may not view adult day care as a reasonable
alternative and may institutionalize their family
members at a higher rate earlier in their caregiving
careers, thus taking them out of the risk pool and
leaving adult day care nonusers who are particularly
resistant to nursing home placement. However, we do
not believe this explanation accounts for our study
findings: first, because three measures of time at risk
(duration of caregiving, time since onset of any
cognitive problems, and time since onset of memory
problems) were not significantly associated with nurs-
ing home placement and did not alter the effect of
adult day care; and second, because more days of day
care attendance increased the risk of nursing home
placement, suggesting that the increased risk is not due
solely to differences between users and nonusers.
Considering our results in light of the behavioral
model of health service utilization, we found that few
of the predisposing or enabling variables were
associated with nursing home placement, and that
need variables (participant disease characteristics and
caregiver burden) were the most important predictors
with adult day care. We found that the risk of nursing
home placement increased with caregiver age and that
nearly all the increased risk associated with adult day
care use was limited to male participants. We are not
sure why the risk associated with day care was greater
for men than for women; it might be that more men
have access to VA nursing home care. We found no
effect of income or long-term-care insurance on
nursing home risk, but we did not ask specifically
about eligibility for VA benefits. Other studies that
examined the use of respite or day care reported
contradictory findings regarding the effect of financial
resources on nursing home placement (Gaugler,
Kane, et al., 2003; Kosloski & Montgomery, 1995).
Results of this study suggest that more frequent
use of adult day care is not a proxy for greater disease
severity or caregiver burden, and that there remain
unmeasured differences between users and nonusers.
That is, there appear to be important risk factors for
nursing home placement that cannot be characterized
by commonly used measures and that are captured
by adult day care use. Perhaps the risk has to do
with the willingness of a particular type of caregiver
to relinquish care when it becomes too difficult.
Willingness to use a service such as adult day care
may be indicative of a greater proclivity to choose
institutional care, signifying a self-selection bias. The
idea that there may be fundamental differences
between caregivers who actively seek out services
762
Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
and those who do not has been proposed as an
explanation for lack of treatment effects in studies of
respite and adult day care (Kosloski & Montgomery,
1995; Lawton et al., 1989). Indeed, research has
shown significant associations between family care-
giver attitudes, beliefs, and perceptions and the use of
various clinical and community services (Kosloski,
Montgomery, & Karner, 1999; Kosloski, Schaefer,
Allwardt, Montgomery, & Karner, 2002; Pedlar &
Biegel, 1999), and these factors tend to vary by race or
ethnicity of the caregiver (Kosloski et al., 2002; Miller
& Mukherjee, 1999). We attempted to account for
this by adjusting for race and for the use of other
community-based services in our analysis. However,
none of the service-use variables significantly pre-
dicted the risk of nursing home placement, and
neither race nor service use altered the risk associated
with adult day care attendance. This suggests that
service use in general does not increase nursing home
placement but that something associated with adult
day care does—either characteristics of people who
use adult day care or something about the experience
itself. As suggested by others (Gaugler & Zarit,
2001), caregivers who use adult day care or other
respite services may become more aware of their level
of stress and more willing to consider nursing home
placement as an acceptable option, especially if the
service experience is positive or if the caregiver re-
ceives encouragement to institutionalize from profes-
sionals or other caregivers. Future researchers need to
develop and incorporate measures of these potential
factors in order to explore these complex issues.
The strengths of this study include specificity,
both in the service used and in the condition of the
person receiving the service. We limited the study to
people with Alzheimer’s disease using an adequate
amount of adult day care at baseline to be considered
truly exposed. We also measured day care use in all
participants at each follow-up to capture changes in
both the user and nonuser groups. We adjusted for
many known predictors of nursing home placement
in persons with Alzheimer’s disease, including the
use of other community-based services. Finally, we
created a dynamic model by using time-varying
measures, which evaluated changes in important
predictors every 3 months, and we had high baseline
and follow-up participation rates.
An important limitation of this study is that it was
observational rather than a randomized controlled
trial. This introduced the possibility of sample
selection bias because we obtained our adult day
care users and nonusers from different sources. The
most comparable group of nonusers came from an
Alzheimer’s diagnostic center that provided a com-
mon factor of caregivers seeking clinical services.
Despite careful matching on some variables and
statistical control for a large number of factors in our
analysis, we probably did not account for all the
differences between individuals who do and do not
use adult day care. However, results from our
The Gerontologist
 sensitivity analyses revealed that our findings were
robust when all baseline group differences were
considered. Even randomized controlled trials seem
unable to obtain a clear comparison, because adult
day care has only been offered, not forced, and
individuals differ in ways we have yet to measure in
their interest in using this service. Thus it is likely that
the effect of adult day care use was not caused by the
different sample sources but rather by true differ-
ences between users and nonusers in the underlying
population. A different source of lack of representa-
tiveness is that we used a convenience sample of adult
day care users, and our nonuser sample was drawn
from a single source, so users and nonusers may not
be representative of all individuals with Alzheimer’s
disease. In addition, the 16 participating adult day
care sites are not necessarily representative of sites in
the region.
In summary, we found that risk of nursing home
placement increased with increasing use of adult day
care. A detailed analysis suggests differences between
caregivers who use adult day care and those who do
not that are not accounted for with the standard
measures of demographic characteristics, disease
severity, and caregiver burden. An investigation of
factors involved in willingness to obtain help with
care may be fruitful.
References
Andersen, R. M. (1995). Revisiting the behavioral model and access to
medical care: Does it matter? Journal of Health and Social Behavior,
36, 1–10.
Centers for Disease Control and Prevention. (2000). Measuring healthy days.
Atlanta, GA: Author.
Cornoni-Huntley, J., Brock, D. B., Ostfeld, A. M., Taylor, J. O., & Wallace,
R. B. (1986). Established populations for epidemiologic studies of
the elderly: Resource data book (NIH Publication No. 86-2443).
Washington, DC: U.S. Department of Health and Human Services.
D’Agostino, R. B., Jr. (1998). Propensity score methods for bias reduction in
the comparison of a treatment to a non-randomized control group.
Statistics in Medicine, 17, 2265–2281.
Fetzer Institute. (1999). Multidimensional measurement of religiousness/
spirituality for use in health research: A report of the Fetzer Institute/
National Institute on Aging Working Group. Kalamazoo, MI: Health
Research Program Area of the John E. Fetzer Institute.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). ‘‘Mini-mental
state’’: A practical method for grading the cognitive state of patients for
the clinician. Journal of Psychiatric Research, 12, 189–198.
Gaugler, J. E., Jarrott, S. E., Zarit, S. H., Stephens, M. A., Townsend, A., &
Greene, R. (2003). Respite for dementia caregivers: The effects of adult
day service use on caregiving hours and care demands. International
Psychogeriatrics, 15, 37–58.
Gaugler, J. E., Kane, R. L., Kane, R. A., Clay, T., & Newcomer, R. (2003).
Caregiving and institutionalization of cognitively impaired older people:
Utilizing dynamic predictors of change. The Gerontologist, 43, 219–229.
Gaugler, J. E., & Zarit, S. H. (2001). The effectiveness of adult day services
for disabled older people. Journal of Aging and Social Policy, 12, 23–47.
Gilley, D. W., Bienias, J. L., Wilson, R. S., Bennett, D. A., Beck, T. L., &
Evans, D. A. (2004). Influence of behavioral symptoms on rates of
institutionalization for persons with Alzheimer’s disease. Psychological
Medicine, 34, 1129–1135.
Hebert, L. E., Scherr, P. A., Bienias, J. L., Bennett, D. A., & Evans, D. A.
(2003). Alzheimer disease in the U.S. population: Prevalence estimates
using the 2000 census. Archives of Neurology, 60, 1119–1122.
Hedrick, S. C., Rothman, M. L., Chapko, M., Ehreth, J., Diehr, P., Inui,
T. S., et al. (1993). Summary and discussion of methods and results of
Vol. 45, No. 6, 2005 763
Downloaded from https://academic.oup.com/gerontologist/article-abstract/45/6/754/553039
by Airlangga University user
on 28 February 2018
the Adult Day Health Care Evaluation Study. Medical Care, 31, SS94–
SS103.
Hennessy, C. H., Moriarty, D. G., Zack, M. M., Scherr, P. A., & Brackbill,
R. (1994). Measuring health-related quality of life for public health
surveillance. Public Health Reports, 109, 665–672.
Katz, S., & Akpom, C. A. (1976). A measure of primary sociobiological
functions. International Journal of Health Services, 6, 493–508.
Kohout, F. J., Berkman, L. F., Evans, D. A., & Cornoni-Huntley, J. (1993).
Two shorter forms of the CES-D (Center for Epidemiologic Studies
Depression) depression symptoms index. Journal of Aging and Health,
5, 179–193.
Kosloski, K., & Montgomery, R. J. (1995). The impact of respite use on
nursing home placement. The Gerontologist, 35, 67–74.
Kosloski, K., Montgomery, R. J. V., & Karner, T. X. (1999). Differences
in the perceived need for assistive services by culturally diverse care-
givers of persons with dementia. The Journal of Applied Gerontology,
18, 239–256.
Kosloski, K., Schaefer, J. P., Allwardt, D., Montgomery, R. J., & Karner,
T. X. (2002). The role of cultural factors on clients’ attitudes toward
caregiving, perceptions of service delivery, and service utilization. Home
Health Care Services Quarterly, 21, 65–88.
Lawton, M. P., Brody, E. M., & Saperstein, A. R. (1989). A controlled study
of respite service for caregivers of Alzheimer’s patients. The Gerontol-
ogist, 29, 8–16.
Lawton, M. P., Moss, M., Kleban, M. H., Glicksman, A., & Rovine, M.
(1991). A two-factor model of caregiving appraisal and psychological
well-being. Journal of Gerontology, 46, 181–189.
Leon, J., Cheng, C. K., & Neumann, P. J. (1998). Alzheimer’s disease care:
Costs and potential savings. Health Affairs (Millwood), 17, 206–216.
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., &
Stadlan, E. M. (1984). Clinical diagnosis of Alzheimer’s disease: Report of
the NINCDS-ADRDA Work Group under the auspices of Department of
Health and Human Services Task Force on Alzheimer’s Disease.
Neurology, 34, 939–944.
Miller, B., & Mukherjee, S. (1999). Service use, caregiving mastery, and
attitudes toward community services. The Journal of Applied Geron-
tology, 18, 162–176.
Montgomery, R. J., & Borgatta, E. F. (1989). The effects of alternative
support strategies on family caregiving. The Gerontologist, 29, 457–464.
Nagi, S. Z. (1976). An epidemiology of disability among adults in the United
States. Milbank Memorial Fund Quarterly: Health and Society, 54,
439–467.
Pearlin, L. I., Mullan, J. T., Semple, S. J., & Skaff, M. M. (1990). Caregiving
and the stress process: An overview of concepts and their measures. The
Gerontologist, 30, 583–594.
Pedlar, D., & Biegel, D. (1999). The impact of family caregiver attitudes on
the use of community services for dementia care. The Journal of Applied
Gerontology, 18, 201–221.
Rosow, I., & Breslau, N. (1966). A Guttman health scale for the aged.
Journal of Gerontology, 21, 556–559.
Rubin, D. B. (1979). Using multivariate matched sampling and regression
adjustment to control bias in observational studies. Journal of the
American Statistical Association, 74, 318–324.
Smith, G. E., Kokmen, E., & O’Brien, P. C. (2000). Risk factors for nursing
home placement in a population-based dementia cohort. Journal of the
American Geriatrics Society, 48, 519–525.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation
of brief measures of positive and negative affect: The PANAS scales.
Journal of Personality and Social Psychology, 54, 1063–1070.
Weissert, W., Wan, T., Livieratos, B., & Katz, S. (1980). Effects and costs of
day-care services for the chronically ill: A randomized experiment.
Medical Care, 18, 567–584.
Wimo, A., Wallin, J. O., Lundgren, K., Ronnback, E., Asplund, K.,
Mattsson, B., et al. (1990). Impact of day care on dementia patients—
Costs, well-being and relatives’ views. Family Practice, 7, 279–287.
Zarit, S. H., Stephens, M. A., Townsend, A., & Greene, R. (1998). Stress
reduction for family caregivers: Effects of adult day care use. Journal of
Gerontology: Social Sciences, 53B, S267–S277.
Zarit, S. H., Stephens, M. A. P., Townsend, A., Greene, R., & Leitsch, S. A.
(1999). Patterns of adult day service use by family caregivers: A
comparison of brief versus sustained use. Family Relations, 48, 355–361.
Zimet, G. D., Powell, S. S., Farley, G. K., Werkman, S., & Berkoff, K. A. (1990).
Psychometric characteristics of the Multidimensional Scale of Perceived
Social Support. Journal of Personality Assessment, 55, 610–617.
Received November 4, 2004
Accepted June 6, 2005
Decision Editor: Linda S. Noelker, PhD