http://www.jstor.org/stable/3765929 .

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Lippincott Williams & Wilkins is collaborating with JSTOR to digitize, preserve and extend access to Medical
Care.

http://www.jstor.org

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

MEDICALCARE
Volume31, Number9, pp SS38-SS49,Supplement
? 1993, J. B. LippincottCompany

Effects of VA Adult Day Health Care on Health Outcomes and

Satisfaction WithCare
MARGARETL. ROTHMAN,PHD,* SUSAN C. HEDRICK,PHD,t
W. ERDLY,PHD,?
KRISA. BULCROFT,PHD,t WILLIAM
MA?
ANDDAVIDG. NICKINOVICH,

It was hypothesized that ADHC would have a positive effect on the health of
patients and their care givers and result in greater satisfaction with care than
customary care. Measurement of health outcomes for patients included assessment of overall, psychological, and social health, and survival. Care giver assessment concentrated on psychosocial health. Findings indicated no difference
in health outcomes between patients assigned to ADHC or their care givers and
their counterparts assigned to customary care. Further analysis of subgroups
found that there were 3 subgroups of patients for whom those assigned to
ADHC had better outcomes (as indicated by lower Sickness Impact Profile
scores) than those assigned to customary care. These subgroups included those
who were 1) not married, 2) most satisfied with their social support network,
and 3) not hospitalized at the time of enrollment in the study. Patients and their
care givers assigned to ADHC were more satisfied with their care than those in
nursing homes, but not more satisfied than those in hospital-based home care.
Care givers reported significantly greater satisfaction with patient care in
ADHC than did care givers of patients receiving care in nursing homes or
ambulatory care clinics.

Adult day health care was expected to

benefit patients and their care givers. Patients were expected to benefit from receipt
* From the Battelle Medical
Technology Assessment
& Policy Research Center, Washington, DC.
t From the HSR&D Field Program, VA Medical
Center, Department of Health Services, University of
Washington, Seattle, Washington.
t From the Department of Sociology, Western Washington University, Bellingham, Washington.
? From the Department of Liberal Studies, University
of Washington, Seattle, Washington.
? From the Department of Sociology, University of
Washington, Seattle, Washington.
Address correspondence to: Margaret L. Rothman,
PHD, Battelle, Medical Technology Assessment and Policy Research Center, 370 L'Enfant Promenade S.W.,
#900, Washington, DC 20024-2115.

of more intensive medical and auxiliary services such as closer supervision of drug therapy and more frequent physical therapy.
Consequently, they were expected to have
higher levels of physical and psychosocial
functioning and greater satisfaction with
care compared with those who did not receive such services. Care givers were expected to benefit directly through respite
care and indirectly through improved functioning of the patients for whom they pro-

Supported by Department of Veterans Affairs Health

Services Research and Development Service, Project
#SDR 85-07 and 071.
The opinions expressed are those of the authors and
do not necessarily reflect the views of the Department.

SS38

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

Vol. 31, No. 9, SupplementEFFECTSOF ADULTDAY HEALTHCAREON HEALTHAND SATISFACTION

vided care. They also were expected to report higher levels of satisfaction with the patient's care.
Methods
Assessment of Bias
Comparison of the groups on all primary
variables available at baseline showed no significant differences, thereby indicating the
randomization was successful. Differences
in attrition, using the Cox proportional hazards model,1 were not significant.2
Procedures for Missing Data
Item mean substitution was used when no
more than 20% of the items were missing
from the Sickness Impact Profile (SIP), Psychological Distress Scale, Satisfaction with
Care Scales, or Caregiver Burden Scale.2
Zero substitution (maximum impairment)
was used for Mini-Mental State Exam items
when no more than 1 item was missing.
Proxy SIPs were collected from care givers
and from long-term care staff if a patient
was in a nursing home at the time of data
collection in addition to patient-generated
data. When no patient-generated responses
were available or when more than 20% of
the patient-generated items were missing,
proxy responses were substituted for the entire scale. Care giver data were substituted
unless no care giver was available, in which
case long-term care staff data were substituted. Although the data provided by patients and proxies may differ, especially in
the psychosocial domain,3 substitution of
proxy data was considered a better alternative than deletion of the case. Proxy responses were not considered valid substitutes for affective or cognitive data, therefore no proxy data were collected for the
Mini-Mental State Exam, Psychological Distress Scale, or Satisfaction With Care Scales.
Satisfaction With Care. The patient satisfaction with care scales were potentially
collected at 3 times: 6 and 12 months, if the

patient had been in a specific care environment for 30 days or more (i.e., ADHC, nursing home, or home care); and at discharge
from ADHC. A substitution scheme was implemented to ensure that only the most recent value entered into the analysis and that
only 1 value was entered for each respondent. For the patients in VA-ADHC, the following scheme was used: 1) discharge interviews were selected first; 2) if there was no
discharge interview, the 12-month satisfaction scores were entered, 3) if there was no
12-month score, the 6-month score was entered.
For the customary care group, the same
substitution scheme was used, first for nursing home, then for hospital-based home
care. Selection on care environment was necessary because the customary care group
may have responded to multiple questionnaires based on their situation at the time of
the interview. A similar substitution scheme
that included postbereavement was implemented for the care giver data.
Group Comparisons
Mean health outcomes for the VA-ADHC
and customary care groups were compared
at 6 and 12 months using analysis of covariance with the baseline value of the independent variable as the covariate. Satisfaction
scores were compared using a t-test of independent means. The conventional level of
0.05 (two-tailed test) was used to denote statistical significance. The actual probability
values are shown in the tables for all values
of 0.10 or less. Survival curves were constructed using the Kaplan-Meier method
and tested for significance using the Cox
proportional hazard model.
An analysis of 6-month outcomes for only
those respondents who completed the study
(i.e., for whom 12-month scores were available) was also conducted for the health outcome data, allowing a comparison of 6- and
12-month scores that was unbiased by dropouts during the last 6 months of participaSS39

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

ROTHMAN ET AL.

MEDICALCARE

tion. The results of these analyses are presented when they differ from the analysis of
all people who completed the 6-month interview.
Six primary outcome variables as well as a
number of subscales were analyzed in this
portion of the study, thus increasing the
probability of making a type 1 error. To decrease the likelihood of making such an
error, we examined the pattern of findings
for other similar outcomes and whether the
finding occurred at both 6 and 12 months, as
well as the P value for each individual outcome.
Subgroup Analyses
The study research questions include a determination of whether ADHC outcomes
differ across sites and among patients with
different characteristics. A limited number
of subgroup variables (n = 24 variables)
based on the results of other studies of community-based long-term care alternatives4
were selected before data analysis. Subgroups were defined in an identical manner
for the ADHC and customary care groups
and formed using baseline variables that
could not have been affected by the treatment. Data were analyzed by analysis of covariance in which condition (VA-ADHC
versus customary care) and the subgrouping
variables were considered independent variables. Those subgroupings that yielded a significant treatment-by-subgroup interaction
were further examined using t-tests within
each level of the subgroup (levels for each
subgroup are defined below).
Seven types of subgroups were considered: 1) patients with specified characteristics at study intake, 2) patients at the 4 VAADHC sites, 3) patients who entered the
study during the first or second half of the
enrollment period, 4) patients who were
likely to be high utilizers of VA-ADHC, 5)
patients who were at higher risk of going to
a nursing home, 6) patients who were at risk
of using greater amounts of ambulatory

care, and 7) patients who were at greater risk

of hospital admission. The methods for
forming each of these types of subgroups are
discussed below.
Subgroups Based on Intake Characteristics. Whenever possible, established
categories were used in the analysis. Those
continuous variables that did not have established categories, e.g., SIP scores, were
divided into 2 or 3 categories to achieve
equivalent numbers of patients in each category.
Patients by Site. Although efforts were
made by VA Central Office to standardize
the clinical care provided across site by mandated staffing levels and training sessions,
there were differences in such factors as the
philosophy and goals of the ADHC staff and
Medical Center case mix. These variations
offered some opportunity to assess the extent to which they were associated with program effectiveness; with only 4 programs,
however, the relationship of site characteristics to outcome can be at most suggestive.
Patients Entering Study in First or Second Half of Enrollment Period. Outcomes were compared for subgroups entering the study in either the first or last 9
months of enrollment for 2 reasons: 1) to
determine whether there was a differential
effect based on the experience of the program in providing care, i.e., did programs
become more effective over time?; and 2) to
assess whether there was a differential effect
caused by any changes in program case mix
over time. For example, programs might
have initially admitted the patients thought
to be most appropriate for ADHC, and later
switched to less appropriate patients as the
available pool decreased.
Patients Who Were High ADHC Utilizers. Patients differed considerably in
the number of days they attended ADHC.
Some patients did not attend at all or attended so few days that one could not reasonably expect to see an effect on patient
outcomes. For this reason, we examined cost
and efficacy for patients who were likely to

SS40

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

Vol. 31, No. 9, SupplementEFFECTSOF ADULTDAY HEALTHCAREON HEALTHAND SATISFACTION

be high utilizers of ADHC using a regression

equation to predict use based on baseline values. In the analyses of high utilizers of
ADHC or VA clinics, and the analyses of
those most likely to go to a nursing home or
hospital, described below, the subgroups
were composed of patients with a high versus low probability of use, i.e., propensity to
use these services rather than patients who
actually used high or low levels of these services. Estimates of service use were based on
the propensity to use services, i.e., regression equations predicting the probability of
high or low service use based on baseline
characteristics. This procedure was followed
because if the patient's actual use of services
was used to select patients for analysis, the
findings could be biased because the 2
groups being compared would not represent
individuals with an equal a priori likelihood
of being in either group. Because patients
with high probabilities to use services were
selected in the same way in the 2 groups, as
a function of baseline characteristics only,
there should be no bias in comparing these
subgroups.
Predictors of the likelihood of being in the
top one-third (55 or more visits) and the top
two-thirds (12 or more visits) of attenders
were identified. Being among the top third
of ADHC utilizers was predicted by previous admission to ADHC, not being hospitalized at entry into the study, not being in a
nursing home during the 6 months before
entry into the study, lower SIP Physical Dimension Scores (i.e., being less physically
impaired), being married, and better care
giver psychosocial health. Being among the
top two-thirds of ADHC utilizers was predicted by 2 variables: previous admission to
ADHC and lower SIP Physical Dimension
scores.
Patients at Higher Risk of Nursing
Home Placement. Patients at the highest
risk of going to a nursing home were identified in the customary care group by comparing those patients who were admitted to a
nursing home with those patients who did

not enter a nursing home during the year of

follow-up. Five characteristics assessed at
intake into the study predicted nursing
home entry: residency in a nursing home
any time during the 6 previous months, being in a nursing home at entry into the study,
higher SIP Physical Dimension Scores (i.e.,
being more physically impaired), being
white, and exhibiting more behavioral problems as reported by the care giver at baseline. Patients with the highest one-third of
the utilization scores were selected because
that is approximately the proportion who actually entered a nursing home during the
study period.
Patients at Risk of High Use of Ambulatory Care. Patients likely to be high utilizers of VA clinics were identified in the customary care group by a regression predicting
whether a patient would be in the top onethird of VA clinic utilizers. Having a greater
than 50% service connected disability (i.e.,
severely disabled patients whose condition
is recognized by VA as related to their military service, and who have high priority for
VA services) was the only variable significantly related to clinic utilization.
Patients at High Risk of Hospital
Admission. Only the patient's SIP physical score at study entry predicted entry into a
hospital during the year of follow-up for the
customary care group. Patients having a SIP
physical score of 23 or greater were designated as being at highest risk of hospital admission.
Results
Patient Health Outcomes
Overall Health. The primary health
outcome measure used in the ADHC study
was the SIP.5 Scores on the SIP may range
from 0 to 100, with a high score indicating
greater dysfunction. Outcomes for the 2 dimensions (physical and psychosocial) and
the total score of the SIP for the ADHC and
customary care groups at baseline, 6, and 12
months are shown in Table 1. Although cusSS41

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

ROTHMAN ET AL.

MEDICALCARE

TABLE1. Comparisona of Meanb Sickness Impact Profile Scoresc at Baseline,

6 and 12 Months BetweenVA-ADHCand CustomaryCare

Baseline

Sickness
ImpactProfile

VA-ADHC

367

Six Months

Twelve Months

Customary
Customary
Customary
Care
Care
VA-ADHC
VA-ADHC
Care
362

310

296

251

259

Physical

Mean
SD

31.7
18.8

33.8
18.4

30.2
21.6

32.6
21.8

29.0
18.6

32.1
18.8

Psychosocial

Mean
SD

34.7
21.3

35.7
21.7

34.3
21.6

35.3
21.8

34.1
22.2

34.5
20.6

Total

Mean
SD

33.3
16.9

35.0
16.5

32.3
17.2

34.1
17.0

31.8
17.3

33.2
16.4

-5.1 to 3.3

Confidenceinterval

-4.8 to 5.8

a Multipleregressionused to comparemeans aftercontrollingfor the baselinevalue of the dependentvariable.

b95%confidenceintervalsare for differencesbetween adjustedmeans.

c HigherSIP scoresindicateworse function.Scoresrangefrom 0 to 100.

tomary care patients had slightly higher

scores, the mean level of impairment was
not significantly different between the
groups and appears to be very similar across
time.
The survival curves shown in Figure 1 indicate that 77% of patients who completed
the first interview in the ADHC group and
81% in the customary care group survived.
This difference is not significant.
Health perceptions were assessed by selfreport of overall health and health compared with one's age-peers. There were no
significant differences between the VAADHC and customary care groups at 6 or 12
months, as shown in Table 2.
Psychological Health. Psychologicalhealth
was assessed by the Psychological Distress
Scale6 and the Mini-mental State Exam.7Patient scores on the 3 subscales of the Psychological Distress Scale (anxiety, depression,
and behavioral/emotional control) as well
as the total score are shown in Table 2. The
mean responses show almost no difference
between baseline and 12 months and no significant differences between the groups at
any point.
The mean Mini-mental State Exam score
for the combined sample at baseline was

23.5, with 36% of the sample scoring below

23, the score generally considered indicative
of need for further evaluation for cognitive
impairment.8'9The percentage of the sample
scoring below 23 dropped to 33% by the 12month assessment, but there were no significant differences in group means at 6 or 12
months (Table 2).
Social Health. Six dimensions of the social support network were measured: importance of individual members, satisfaction
with the help/support received from network members, size of the support network,
degree of upset and degree of helpfulness
experienced by patient when network
members were providing assistance, and
number of confidants. The mean size of the
social support network at baseline was 5
members. Twenty-eight percent of the respondents reported having 2 or less network
members, and 29% reported no confidants.
The overall level of satisfaction with the
help and support received from network
members was positive, with only 13% of the
sample expressing dissatisfaction. There
were no significant differences between the
groups at 6 or 12 months (not shown).
Results for Subgroups of Patients.
Significant differences were observed on the

SS42

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

Vol. 31, No. 9, Supplement EFFECTS OF ADULT DAY HEALTH CARE ON HEALTH AND SATISFACTION

PROP ,QRTION
2-

lf.Di

0.9
2

0.8

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

VA-ADHC

7
6
MONTH

CUSTOMARY CARE

10

11

12

CONTRACT ADHC

FIG. 1. Comparison of survival of VA-ADHC, Customary Care, and Contract ADHC groups.

SIP for only 3 groups of patients. Table 3

shows that among unmarried patients, those
assigned to ADHC had lower mean total SIP
scores (less dysfunction) at 6 months than
those in the customary care group, and
lower mean physical, psychosocial, and total
SIP scores at 12 months.
Among those patients most satisfied with
the help and support they received from
their social support network at the initial interview, the ADHC group showed significantly lower SIP psychosocial and total
scores than the customary care group at 6
months, as shown in Table 4. Differences
between the groups were significant for
physical and total SIP scores, but not for
psychosocial function at 12 months. No
such differences were observed for patients
who were less satisfied with their social networks.
Table 5 shows that among patients not in
the hospital at enrollment, those assigned to

ADHC had significantly lower physical,

psychosocial, and total SIP scores at 6
months, but only the physical dimension
difference remained significant at 12
months. In contrast, the adjusted mean differences for those in hospital at enrollment
were quite small and nonsignificant.
The level of significance for the 6-month
comparisons for only those persons who
completed the study varied slightly from
those that included persons who dropped
out or died during the second 6 months for 2
subgroups. For patients who were very satisfied with their social support, and for those
not hospitalized at enrollment, the difference between groups in psychosocial function became nonsignificant when dropouts
were deleted, indicating that the change in
level of significance from 6 to 12 months
may have been due to differences in sample
composition rather than to change in mean
level for the entire group.
SS43

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

ROTHMAN ET AL.

MEDICALCARE

TABLE2.

Comparisona of Meanb Patient Psychological Distressc Cognitive Functiond and Health

Perceptionsc at Baseline, 6 and 12 Months Between VA-ADHC and Customary Care
6 Months

Baseline
Scale

VA-ADHC

12 Months

Customary
Customary
Customary
Care
Care
VA-ADHC
Care
VA-ADHC

OverallHealthPerception
n
Mean
SD

380
3.9
1.2

371
3.9
1.2

315
3.7
1.2

Confidenceinterval

311
3.7
1.1

275
3.6
1.2

273
3.6
1.2

-.3 to .1

-.3 to .1

PsychologicalDistress
Anxiety
Depression
Control
Total

294
n
Mean 29.5
SD
10.7
Mean 14.4
5.9
SD
Mean 22.2
8.0
SD
Mean
SD

66.2
22.6

271
28.1
10.5
14.0
6.0
21.3
8.4

223
27.7
10.4
13.1
6.0
21.0
8.2

210
28.5
10.4
13.7
6.0
20.6
7.9

157
27.9
10.8
13.7
6.1
21.1
8.7

125
28.7
10.9
13.7
6.3
20.7
8.9

63.4
23.0

62.0
22.2

62.9
22.2

62.8
23.2

63.3
24.1

Confidenceinterval
Mini-mentalStateExam

n
329
Mean 23.8
SD
4.7

-5.1 to 1.0
305
23.3
5.2

Confidenceinterval

250
23.8
4.9

229
23.7
5.1
-.8 to 1.0

-4.8 to 5.8
193
23.7
5.3

175
24.3
5.0
-4.8 to 5.8

Multipleregressionused to comparemeans aftercontrollingfor the baselinevalue of the dependentvariable.

b95%confidenceintervalsare for differencesbetween adjustedmeans.
c A high scoreindicatesgreaterpsychologicaldistress.Totalscoresrangefrom 24 to 144.
dA
high scoreindicatesbettercognitivefunctioning.Scoresrangefrom 0 to 30.
'A high scoreindicatesgreaterimpairment.Scoresrangefrom 1 to 5.

In an effort to further understand the outcomes described above, we examined the

characteristics of the patients who composed those subgroups at entry into the
study. For each group we compared the
mean values of the subgroup that showed
significant differences with those who did
not, e.g., characteristics of married versus
unmarried patients. Results of the univariate
comparisons (analysis of variance was used
to compare group means) showed that patients who were not married had significantly lower SIP scores (on both dimensions
and total scores), smaller network sizes, reported less upset with the help they received
from network members, and were younger

(mean age, 71.3 years) than those who were

married (mean age, 73.7) at study entry.
Those most satisfied with their social support networks had lower Psychological Distress Scale scores (all subscales) and psychosocial SIP scores. Other aspects of their network also appeared more positive in that the
"most satisfied" group of patients reported
having significantly more confidants and
perceived their support network as more
helpful and less upsetting when providing
support. The only difference in health outcomes among those hospitalized at enrollment and those who were not was that the
former group were more satisfied with their
support network. Multivariate analysis (dis-

SS44

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

Vol. 31, No. 9, Supplement EFFECTS OF ADULT DAY HEALTH CARE ON HEALTH AND SATISFACTION

3.
TABLE

Comparison of Meana SIP Scores at 6 and 12 Months Between VA-ADHC and Customary
Care By Initial Marital Status
VA-ADHC

Customary
Care

Adjusted
MeanDifference

pb

Not Married
6 Months
Physical'
Psychosocial
Total
n

24.7
26.7
26.0
127

28.9
31.6
30.7
116

-2.5
-3.9
-3.2

NS
0.06
0.04

12 Months
Physical
Psychosocial
Total
n

22.9
25.5
25.3
108

27.9
32.7
30.8
108

-4.0
-6.0
-4.7

0.05
0.01
0.01

Married
6 Months
Physical
Psychosocial
Total
n

34.0
39.7
36.8
175

34.9
37.7
36.2
181

.2
.8
.8

NS
NS
NS

12 Months
Physical
Psychosocial
Total
n

33.4
40.3
36.7
154

34.5
36.3
35.2
172

-.2
2.6
1.3

NS
NS
NS

a Multile

used to comparemeans aftercontrollingfor the baselinevalue of the dependentvariable.

bP valuesregression
are based on adjustedmean differences.
A high scorerepresentsgreaterdysfunction.Scoresrangefrom 0 to 100.

criminant analysis) failed to significantly discriminate between the ADHC and customary care groups.
Satisfaction With Care. Satisfaction
with care was measured by an instrument
developed for the ADHC study.2 It was designed to elicit the respondent's opinion regarding the presence of specific aspects of a
care environment that are generally considered indicators of high-quality care.
The respondents in customary care received their care in either a nursing home,
ambulatory care clinic, or home care (although in some cases they reported receiving no care from any source). The satisfaction questionnaire was not applicable to respondents in the ambulatory care clinic;
thus, patients receiving this type of care are
not included in the analysis.2
The findings for patient satisfaction with

care shown in Table 6 indicate that the VAADHC group was significantly more satisfied than customary care patients in nursing
homes, but not more satisfied than patients
in home care. Overall, patients assigned to
ADHC were more satisfied with their care
than those assigned to customary care.
These differences remained even after controlling for level of illness (total SIP score at
baseline).
Care Giver Health Outcomes. Almost
three-fourths (71%) of the patients in the
VA-ADHC study reported having care
givers. Of this number, 76% were spouses,
11% were adult children, and 3% were siblings. The remaining care givers consisted of
parents, long-term care staff, and others.
Care giver ages ranged from 20 to 100, with
a mean age of 62.5 years (standard deviation
[SD], 12.9). Spouse care givers were slightly
SS45

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

ROTHMAN ET AL.
TABLE4.

MEDICALCARE

Comparison of Meana SIP Scores at 6 and 12 Months Between VA-ADHC and Customary
Care by Initial Patient Satisfaction with Social Support

Very satisfied
6 Months
Physical'
Psychosocial
Total

VA-ADHC

Customary
Care

26.7
28.5
28.2

32.3
31.6
32.4

151

12 Months
Physical
Psychosocial
Total

23.4
27.9
26.0

131

Adjusted
Mean Difference

pb

-2.9
-4.2
-3.5

0.08
0.03d
0.02

-5.2
-3.1
-4.2

0.005
NS
0.01

143
31.6
21.1
31.7
127

Verydissatisfiedto
moderatelysatisfied
6 Months
Physical
Psychosocial
Total
n

31.0
36.3
33.7
35

27.5
37.4
32.9
69

1.4
-0.8
0.4

NS
NS
NS

12 Months
Physical
Psychosocial
Total
n

33.7
37.1
35.7
70

27.8
38.4
33.3
68

3.3
-1.0
1.9

0.06
NS
NS

a Multile

regression used to compare means after controlling for the baseline value of the dependent variable.
P values are based on adjusted mean differences.
cA high score represents greater dysfunction. Scores range from 0 to 100.

d Thedifferencebetween

are deleted.

groupsis not significantwhen those patientswho droppedout between6 and 12 months

older (mean, 65.8; SD, 9.4). The length of

time care givers reported providing the same
level of care to the patient ranged from 0 to
41 years (mean, 3.0; SD, 4.7).
Care givers in both groups reported very
little impairment at baseline in activities of
daily living (98% reported no activity limitations), but 40% rated their overall health as
fair to poor, and 35% reported their health
as worse than others their own age (not
shown). Psychological distress was assessed
by the Psychological Distress Scale described in a previous section. The results
shown in Table 7 indicate a high level of
psychological distress (comparable to that of
the study patients), but no significant differences between VA-ADHC and customary
care at 6 or 12 months.

The Caregiver Burden Questionnaire10

was used to assess subjective and objective
burden related to caring for the patient. The
subscale and total scores shown in Table 7
indicate very little change in perception of
burden related to caregiving between the
baseline and 12 month interview and no significant differences between the VA-ADHC
and customary care groups at any time.
A subgroup consisting of only those care
givers who were spouses of patients was examined separately. Although spouse care
givers were consistently more impaired on
all health outcomes, we found no significant
differences between the VA-ADHC and customary care groups.
Satisfaction With Care. Care givers responded to the same satisfaction with care

SS46

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

Vol. 31, No. 9, Supplement EFFECTS OF ADULT DAY HEALTH CARE ON HEALTH AND SATISFACTION

5.
TABLE

Comparison of Meana SIP Scores at 6 and 12 Months Between VA-ADHC and Customary
Care by Patient's Hospital Status at Enrollment
Adjusted
MeanDifference

VA-ADHC

Customary
Care

Not in hospital
6 Months
Physical'
Psychosocial
Total
n

30.4
34.9
29.8
117

37.7
37.3
34.8
96

-2.5
-3.0
-3.7

0.00

12 Months
Physical
Psychosocial
Total
n

29.8
35.5
32.4
96

39.8
38.0
38.0
74

-5.7
-1.4
-3.2

0.01
NS
0.10

In the hospital
6 Months
Physical
Psychosocial
Total
n

29.9
33.8
32.0
185

30.3
34.7
32.7
203

0.5
-0.6
0.0

NS
NS
NS

12 Months
Physical
Psychosocial
Total
n

28.9
33.4
31.8
164

28.8
33.7
31.6
181

-0.2
-1.1
-0.3

NS
NS
NS

Pb

0.04d

0.01

Multileregressionused to comparemeans aftercontrollingfor the baselinevalue of the dependentvariable.

bP values are based on adjustedmean differences.
cA
high scorerepresentsgreaterdysfunction.Scoresrangefrom 0 to 100.
d Thedifferencebetween
groupsis not significantwhen those patientswho droppedout between6 and 12 months
are deleted.

questionnaire as the patients, with only

slight modifications (the subject was the patient rather than the respondent). The find-

Comparisona of Satisfactionb with

Care for Patient Assigned to VA-ADHC
and Customary Care

TABLE6.

VA-ADHC
Care
Environment Mean SD n

CustomaryCare
Mean SD n

VA-ADHCvs.
Nursing
Home
32.9 6.0 199 25.3 8.0 27 0.01
VA-ADHCvs.
Home Care 32.9 6.0 199 32.8 5.3 25 NS
a

ings, shown in Table 8, indicate that, similar

to patients, care givers of ADHC patients
were more satisfied than care givers of patients in nursing homes but not more satisfied than care givers of patients in home
care. Controlling for patient health status
(total SIP score) at baseline did not change
these results.

T-testsused to comparemeans.
bA high scorerepresentsgreatersatisfaction.Scores
rangefrom 8 to 40.

Discussion
There were no significant differences between the VA-ADHC and customary care
groups on any of the patient or care giver
health outcome variables. This indicates
that, overall, the ADHC program did not significantly improve or even slow the mean
rate of decline in any measured domain of
patient health relative to the customary care
SS47

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

ROTHMAN ET AL.

7.
TABLE

MEDICALCARE

Comparisona of Meanb Caregiver Psychological Distressc and Burdendat Six and Twelve
Months Between VA-ADHC and Customary Care
Baseline
VA-ADHC

Scale
Psychological Distress

264

Twelve Months

Six Months

Customary
Care
255

VA-ADHC

Customary
Care

216

203

VA-ADHC

Customary
Care
169

180

Physical

Mean
SD

29.5
9.9

29.8
9.9

29.1
9.2

28.9
9.7

28.9
9.7

28.8
10.1

Depression

Mean
SD

14.8
5.2

14.8
5.7

14.5
5.1

14.6
5.7

14.3
5.6

14.9
5.6

Control

Mean
SD

19.8
7.1

19.7
7.4

19.9
6.9

20.0
7.9

19.2
7.4

19.7
8.7

Total

Mean
SD

64.1
20.8

64.3
20.9

63.5
19.8

63.6
22.5

62.4
22.7

63.4
23.9

Caregiver Burden

-2.4 to 1.8

-2.3 to 1.3

Confidence interval
n

262

255

214

202

169

180

Subjective

Mean
SD

12.4
4.3

12.4
4.4

12.5
4.3

12.8
4.4

12.9
4.4

13.1
4.7

Objective

Mean
SD

18.8
6.2

18.9
6.2

21.1
6.4

21.2
6.5

20.8
6.7

21.0
6.9

Total

Mean
SD

33.9
9.3

34.0
9.5

33.6
9.3

34.1
9.3

33.7
9.8

34.0
9.8

-4.5 to .9

Confidence interval
a

-4.5 to 1.5

Multile regression used to compare means after controlling for the baseline value of the dependent variable.

b Confidence intervals are for differences between adjusted means.

cA
high score indicates greater distress. Total scores range from 44 to 144.
d

A high score indicates greater burden. Total scores range from 12 to 60.

group. Significant differences were observed, however, for 3 subgroups of patients. Patients who, at study enrollment,
were not married, were most satisfied with
their social support network, or were in hospital had significantly more positive outcomes (lower SIP scores) when assigned to
ADHC than those assigned to customary
care, after controlling for baseline health
status. Examination of the characteristics of
these patient groups, however, did not show
any consistent patterns that might explain
these findings.
The results of the subgroup analyses must
be taken as tentative for 2 reasons. First, because of the large number of subgroup analy-

ses involved, there is a high probability of

finding some comparisons significant by
chance; and second, there is little opportunity of corroboration of these findings in the
current literature because few previous studies of ADHC analyzed subgroups. For these
reasons, it is suggested that these findings be
considered as suggestions for future research.
Patients and their care givers did report
greater satisfaction with care in the ADHC
programs than their counterparts in nursing
homes. These differences were large and remained significant even after controlling for
overall level of health (total SIP score). It
should be noted that these comparisons vio-

SS48

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions

Vol. 31, No. 9, Supplement EFFECTS OF ADULT DAY HEALTH CARE ON HEALTH AND SATISFACTION

TABLE8. Comparisona of Mean Caregiver

Satisfaction with Care VA-ADHC and
Customary Care
VA-ADHC
Comparison
VA-ADHC vs.
Nursing
Homeb

Mean SD

Customary Care
n

Mean SD

35.6

4.0 178

28.5

7.9 33 0.00

VA-ADHC vs.
Home Careb 35.6

4.0 178

34.3

6.5 36

NS

T-tests used to compare means.

Scores range from 8 to 40. A high score indicates
greater satisfaction.
b

late the assumptions of the randomized trial

because a large portion of the customary
care group patients did not receive care in
the 2 environments that were evaluated (i.e.,
satisfaction with care in ambulatory care
clinic was not evaluated with this measure).
It is possible that the observed differences
could be attributed to uncontrolled differences in the type of patients admitted to
nursing homes rather than to an effect of
ADHC.
Care givers reported psychological distress levels as high as those reported by patients, although they indicated almost no
impairment in activities of daily living. The
perceived burden of care giving was also
high, exceeding that of care givers of patients with Alzheimer's disease and other
types of cognitive impairment.11 There were
no differences in these measures at any time
for those care givers of patients assigned to

ADHC versus customary care; thus, the VAADHC programs were not effective in reducing the care giver's psychological distress
or burden.
References
1. Breslow NE. The proportional hazard model: Applications in epidemiology. Comm Statist A Theory
Methods 1978;A7:315.
2. Chapko MK, Rothman ML, Ehreth JL, et al. Data
collection in the adult day health care evaluation study.
Med Care 1993;31(Suppl.):SS15.
3. Rothman ML, Hedrick SC, Bulcroft KA, et al. The
validity of proxy-generated scores as measures of patient health status. Med Care 1991;29:115.
4. Weissert WG, Cready M, Pawelak JE. Home and
community care: Three decades of findings. In: Petersen MD, White DL. Health Care of the Elderly: An Information Sourcebook. Newbury Park: Sage Publications, 1989.
5. Bergner M, Bobbitt RA, Carter WB, et al. The sickness impact profile: Development and final revision of
a health status measure. Med Care 1981; 19:787.
6. Veit C, Ware J. The structure of psychological distress and well-being in general populations. J Consulting Clin Psychol 1983;31:730.
7. Folstein MF, Folstein S, McHugh PR. Mini-mental
state: A practical method for grading the cognitive state
of patients for the clinician. J Psychiatr Res
1975; 12:189.
8. Folstein MF, Anthony J, Parhad I, et al. The meaning of cognitive impairment in the elderly. J Am Geriatr
Soc 1985;33:228.
9. Kramer M, German PS, Anthony JC, et al. Patterns of mental disorders among the elderly residents of
eastern Baltimore. J Am Geriatrics Soc 1985;33:236245.
10. Hooyman N, Gonyea J, Montgomery R. The impact of in-home services termination on family caregivers. The Gerontologist 1985;25:141.
11. Montgomery RJV,Kosloski K, BorgattaE. Service
use and the caregiving experience: Does Alzheimers's
disease make a difference? In: Biegel DE and Blum A
(Eds) Aging and caregiving. Sage Publications: Newbury Park, 1990.

SS49

This content downloaded from 137.151.141.100 on Fri, 7 Nov 2014 17:01:18 PM

All use subject to JSTOR Terms and Conditions