CONFERENCE DRAFT
DRAFT: Do not quote or circulate
without permission from the author or
AcademyHealth, June 25, 2005
CONSUMER-DIRECTED CARE AND
ITS IMPLICATIONS FOR
STATE AND FEDERAL
LONG-TERM CARE POLICY
Randall Brown, Ph.D.
Mathematica Policy Research, Inc.
This paper was commissioned by The Commonwealth Fund for use at its 2005 Colloquium, Building Bridges:
Making a Difference in Long-Term Care, held June 25, 2005.
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
INTRODUCTION
Imagine that you are an 87-year old woman with limited mobility who lives alone, and have
been receiving 20 hours per week of assistance from Medicaid (the maximum the state allows)
with getting out of bed, bathing, housework, and meal preparation. The care is provided by a
woman who works for a certified provider agency. Your daughter comes by every evening and
weekend to help you with dinner and getting to bed. Without the agency’s help you would have
to enter a nursing home (assisted living might provide enough help, but you can’t afford that, and
Medicaid doesn’t pay for it). But occasionally your daughter has to work on the weekend, or is
not feeling well herself, and it’s hard to find someone to take over then—sometimes you just
have to fend for yourself, though the family worries that you’ll fall again or burn yourself. And
sometimes the agency aide doesn’t show up, or comes late, so you have to lay in bed until you
can get someone to come over. Even when she’s there, she just watches television most of the
time. You’ve complained, and they sent you a different aide last time, but you’ve just gotten
comfortable with the previous one doing such intimate care. The aides don’t know how to cook
the spicy Mexican food that you are used to, and aren’t willing to try, so meals are a drudgery
too. All this makes you feel miserable, powerless, and guilty as well, because your daughter is a
wreck from worrying about you all the time and has to give up so much of her scarce free time to
help you.
Care provided through states’ Medicaid personal care benefits and their home and
community based services (HCBS) waivers enables many people to remain in the community
who would otherwise have to move to a nursing home or live alone in danger. While agencyprovided care is a godsend for many frail older people or younger adults with disabilities, and
their families, for a sizable minority it leaves much to be desired, as the example above
illustrates. For these consumers, the traditional agency-based approach to providing care is too
rigid, allowing them too little choice about when, how, and by whom their care is provided, and
what type of help is given. Furthermore, agencies frequently have trouble finding enough
reliable, competent, and sensitive staff at the wages they can afford under Medicaid, especially in
remote or high-crime areas.
As a consequence, Medicaid spending on these services is
inefficient, and leaves too much burden on unpaid family caregivers. When unpaid caregivers
burn out from the emotional and physical toll of caregiving, consumers often have to enter a
nursing home.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 1 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
To address these shortcomings, advocates for people with disabilities have pushed for, and
states have moved toward, more consumer control over this care. States were prompted to
consider this option in part by a 1999 Supreme Court decision (Olmstead), which ordered states
to provide care to people with disabilities in a community setting if the consumer preferred that
setting, the state found it appropriate, and the resources were available. Federal system change
grants and the New Freedom Initiative gave states further incentive and resources to offer
consumers more choice over their care and control over their lives. Every state now offers some
form of consumer direction to at least some subgroup of Medicaid consumers.1 Infield (2004) in
a report prepared for NASUA and CNCOA, reported that their survey of state Aging
administrators found that states see consumer direction as effective and “simply the right thing to
do.”
While giving people with disabilities greater control over their own care and destiny is
appealing, consumer direction raises a number of potential problems and questions about the
safety of the beneficiary, potential fraud and abuse of the benefit, abuse of caregivers, and cost to
the state. Until recently, there has been very little evidence on these issues. Furthermore, states
wishing to adopt a consumer directed program for personal care or waiver services must make a
great many operational decisions that have crucial implications for these concerns and for the
political acceptability of the program to consumers, advocates, state government, and agencies.
In this paper I provide an overview of the options for consumer directed care that are now
available to consumers, present the available evidence on how well they work compared to the
traditional system, and indicate some of the remaining questions and challenges for states
1
Benjamin (2001) suggests that three other factors heightened states’ interest in consumer direction—hopes
that it might be less expensive than agency care, expectations that it would help address the workforce shortages that
agencies faced, and movement away from a purely medical model of providing care for people with disabilities.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 2 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
wishing to make such options available. Until very recently, there was little hard evidence on the
effects of consumer direction on consumers and caregivers, and no studies of how consumer
direction affected costs to Medicaid. Thanks to the creativity and funding of the Robert Wood
Johnson Foundation and the Office of the Assistant Secretary for Planning and Evaluation in
U.S. Department of Health and Human Services (ASPE), and the waivers granted by the Centers
for Medicare and Medicaid Services, a wealth of data and estimates are now available from the
Cash and Counseling Demonstration and Evaluation.2 Given that Cash and Counseling is an
innovative and particularly flexible form of consumer direction that is now expanding to
11 additional states, was tested with a rigorous experimental design, and covers 3 different states
with different target populations, benefit levels, and rules, I draw most of the evidence presented
from Mathematica Policy Research’s evaluation of that program.
The evaluation provides
estimates of program effects on consumers, caregivers, hired workers, and costs to Medicaid.
CONSUMER DIRECTION TODAY IN THE US
The term “consumer direction” when applied to personal care and HCBS means, at a
minimum, that consumers of these services have some control over who delivers their care, and
how and when it is delivered. However, there are many variations in how consumer directed
plans are implemented. Programs differ in the target populations to which they are offered, the
services that the consumer has some control over, whether consumers are given control over an
allowance or just some input on who their workers will be and how they do their job, how much
freedom the consumer has in deciding who they would hire and how such an allowance would be
2
The National Program Office at the University of Maryland and Boston College oversaw the development of
the program, provided extensive technical assistance to the states, and oversaw the evaluation.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 3 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
spent, the wage rate paid to hired workers, how much assistance and oversight the state provides,
and many other factors.
The features of, and constraints on, consumer directed plans are dictated in part by the
authority under which they exist. In states that offer personal care services under their state
Medicaid plan, CMS expressly permits consumer-direction, but there are limitations on what this
entails that can make it less attractive to many consumers and their prospective workers. First,
the worker must be an independent Medicaid provider, meaning that they must meet the state’s
licensure requirement and must either contract with the state or be hired by a Medicaid provider
agency. Second, the state sets the wage rate to be paid. Third, the consumer cannot hire a
legally liable relative. Fourth, consumers cannot use the benefit to manage a self-selected array
of supplies, equipment, and services as they see fit, nor can they take some of the authorized
benefit amount in cash to use for transportation or incidental expenses.
More flexibility is allowed if the state obtains a 1915(c) waiver to provide HCBS, and that
approach has just this year become even more flexible. This authorization also requires hired
workers to be Medicaid certified, and until recently, did not allow hiring of legal liable relatives
(but does now). However, it allows flexible uses of the benefit and allows the consumer to set
the wage rate for the worker they hire. CMS’s Independence Plus initiative has made it easier
for states to obtain 1915(c) waivers, previously a somewhat arduous process, by providing a
template and technical assistance. As of January 2005, 11 states had obtained waivers under this
template.3
3
This template defines, for the first time, the features that CMS requires for approval of the consumer directed
option: a person centered approach, individualized budgets, fiscal intermediary services, support brokers directed by
the consumer, a quality assurance and improvement system, and consumer protections like emergency back up and
incident management.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 4 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Consumer directed programs authorized under Section 1115 waivers offer even greater
flexibility to states, including permission to hire legally liable relatives, to give cash directly to
consumers, to loosen Medicaid eligibility requirements, and to waive the requirement that the
state can pay only providers with Medicaid contracts. However, the program has to be budget
neutral—that is, must cost no more than the state would have spent in the absence of the program
over a five year period.
Finally, the Cash and Counseling program offers a very specific type of consumer direction
that features fewer constraints than those previously offered by earlier programs. The program
was originally offered as a demonstration in three states—Arkansas, Florida, and New Jersey—
and was authorized under 1115 waivers.
The original states continue to offer Cash and
Counseling options, and programs in 11 new states have been funded by the Robert Wood
Johnson Foundation and ASPE and are expected to begin enrolling consumers in 2005 and 2006.
Cash and Counseling provides each consumer with an allowance, based on the hours of care
or other services in their care plan, which they can use to hire workers, to purchase equipment or
supplies, or to make modifications to their vehicle or home that enable them to live more
independently. In order to receive the allowance, consumers must develop a spending plan for
the allowance, and all expenditures are monitored against this plan. The consumers are allowed
to receive the full allowance as cash if they can demonstrate the ability to perform all of the
activities of an employer (write checks, withhold taxes, file social security contributions, etc.),
but a fiscal intermediary or bookkeeper is offered to consumers to perform these functions, and
virtually all consumers have chosen this option. Consumers are allowed to hire legally liable
relatives in two of the states (Florida and New Jersey), and these workers need not contract with
the state or be employed by an agency, nor meet any particular qualifications or training
requirements. Counselors help the consumers develop an acceptable spending plan, monitor
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 5 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
consumers’ spending against these plans. States vary in whether counselors make decisions on
whether requests to purchase items not in the spending plan are allowable. Consumers can
receive modest amounts of the monthly allowance (10 to 20 percent, depending on the state) in
cash to use for incidental expenses, and a substantial number of consumers do so. They also can
save up funds over several months to purchase allowable items if they so desire. Consumers can
appoint a representative to assist them with managing the allowance, making the program
available to all consumers regardless of the nature and severity of their disabilities.
Although there are no available estimates of the number of consumers currently directing
their own care, and no definitive list of the programs currently being offered, it is clear that
consumer direction is growing rapidly in the United States. Flanagan (2001) identified 139
consumer-directed programs operating in 2001, and many more have developed since then.
Most of these programs were small, serving fewer than 1,000 individuals (42 percent served
fewer than 500 consumers), but the California In-Home Supportive Services (IHSS) program
enrolled about 250,000 consumers, and the 139 programs together served nearly 500,000
consumers. Substantially more consumers are enrolled in consumer directed care programs in
2005.
EVIDENCE ON THE EFFECTS OF CONSUMER DIRECTION
As noted above, while offering consumers more choice and control over their care seems
attractive, states and policy makers have a number of concerns about consumer direction. The
key questions regarding the effects of consumer directed programs are:
• How much and in what ways does it actually improve the lives of consumers
• Does it keep consumers safe
• How does it affect the people who had been the primary informal caregivers
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 6 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
• How do the workers hired under consumer direction fare
• How do Medicaid costs compare to what they would have been
• How many and what type of beneficiaries are likely to be interested
Very little evidence is available on the effects of consumer direction in general. The only
thorough evaluations are the evaluation of California’s IHSS program (Benjamin and Matthias
2001; Benjamin 2001; Benjamin and Matthias 2004) and the Cash and Counseling evaluation of
the three Cash and Counseling states described here. The two studies examine some of the same
issues, but the programs differ substantially in a number of ways that could lead to quite different
program effects. First, IHSS assigned consumers to the consumer directed and agency-directed
models, whereas Cash and Counseling was a strictly voluntary program and the demonstration
used random assignment. Second, the maximum monthly benefit provided in the IHSS program,
equivalent to 283 hours of care (9 to 10 hours per day), far exceeded the maximum benefit
provided by any of the three Cash and Counseling states. Third, the IHSS program did not
include a counseling component. Fourth, consumers in the Cash and Counseling program had
more flexibility in the use of their allowances (e.g., they could purchase other goods and
services).
Benjamin and Matthias’s (2001) study of 511 consumers directing their own care in the IHSS
program and 584 receiving agency services showed that consumers under both models were
satisfied with their care, but those in the consumer-directed model generally reported more
positive outcomes, especially those who hired a family member. In general, those hiring their
own worker reported a greater sense of security, wider choices, and greater satisfaction
compared to those receiving care from agency providers. Benjamin and Matthias (2004) found
that, relative to agency workers, workers hired under consumer direction received 30 percent
lower wages and were less likely to receive fringe benefits. They had closer relationships with
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 7 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
the care recipient, but experienced more emotional strain. These differences were essentially
driven by those directly hired workers who were related to the client. No data on the relative
costs of the consumer directed and agency directed IHSS models were available. Benjamin
(2001) refers to Cash and Counseling as the first opportunity to obtain rigorous estimates of the
relative costs under the two systems.
The Cash and Counseling evaluation is much more extensive than the IHSS studies, and
features random assignment of interested eligible consumers into the treatment group, which had
the opportunity to manage their own allowance, or the control group, which had to get care from
agencies in the usual fashion. As noted above, it was implemented in three states, with different
target populations, benefit levels, rules about hiring legally liable relatives, methods for
calculating the allowance amount and funding the fiscal agent and counseling services, and
methods for reassessment. Table 1 displays the features of the programs in the three states.
While all of these features are important for understanding the impacts on consumers and costs
and why they differ across states, three of particular significance to bear in mind when reviewing
the results below are:4
• Arkansas allowed consumers who were eligible for but not already receiving services
under the traditional program to enroll in Cash and Counseling, whereas Florida
offered their program only to consumers already receiving waiver services, and New
Jersey offered theirs only to consumers who already had been assessed by an agency.
• Florida offered their program to children and non-elderly adults who were receiving
service under the state’s waiver program for beneficiaries with developmental
disabilities; the other states opened their programs to adults only, and targeted it
primarily toward those with physical disabilities and functioning problems rather than
cognitive problems.
4
For additional information about demonstration implementation in the three states, see Phillips and Schneider
(2002, 2003, and 2004).
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 8 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 9 of 64
HCBS waiver services, except case
management/support coordination*
Personal care
Could not hire legally responsible
relatives (such as spouses or parents) or
representative
Provider specific, ranging from 70 to
91 percent and averaging 86 percent
across all enrollees
$8 per hour in care plan multiplied by
provider-specific adjustment factor
Services included in
calculating the allowance
amount
Hiring restrictions
Care plan adjustment factor
used in setting allowance
Method for calculating
allowance
Claims history or adjustment factor
multiplied by value of care plan.
(Care plan always used for those
with developmental disabilities.
Also used care plan if claims history
was not stable or if care plan value
was at least $50 per month more
than claims history.)
89 percent for elderly adults,
83 percent for adults with physical
disabilities, 92 percent for children
and adults with developmental
disabilities
None
Elderly adults and nonelderly adults
with physical disabilities, and
children and adults with
developmental disabilities who were
receiving services under the HCBS
waiver
June 2000 to July 2002 (adults) and
June 2000 to August 2001 (children)
Florida—CDC
Adults (elderly and nonelderly) with
physical disabilities (may also have
cognitive disabilities) who were eligible
for the state plan Medicaid personal
care program
December 1998 to April 2001
Arkansas—IndependentChoices
Eligible population
Demonstration enrollment
period
Feature
Value of care plan 10 minus percent setaside for fiscal agent and counseling services
None
Could not hire representative
Personal care
Adults (elderly and nonelderly) with physical
disabilities who were already enrolled in the
state plan Medicaid personal care program
November 1999 to July 2002
New Jersey—Personal Preference Program
KEY FEATURES OF CASH AND COUNSELING PROGRAMS, BY STATE
TABLE 1
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 10 of 64
Agencies (for traditional program) and
counselors (for allowance recipients)
Demonstration enrollees could also
participate in the HCBS waiver
programs ElderChoices or Alternativesa
Entity conducting
reassessment
Participation in other
consumer-directed or home
care programs
For adults with developmental
disabilities, the demonstration
excluded some northern counties
with a state-funded consumerdirected program.
Support coordinators or case
managers (for traditional program)
and counselors (for allowance
recipients)
Fiscal agent fees paid for by
schedule of fees charged to
consumers (for example, $5 per
check)
Counseling paid for through
existing Medicaid funding stream
for case management and support
coordination in traditional program.
$829 (adults) and $831 (children)
Florida—CDC
Demonstration enrollees could not
participate in HCBS waiver programs or a
state-funded consumer-directed program.
Agencies (for traditional program) and
Medicaid nurses (for allowance recipients)
Set aside 10 percent of care plan value to
cover counseling services and some fiscal
agent costs. From this pool of money, the
state paid human services agencies a lump
sum per consumer to complete a cash
management plan and an hourly fee
thereafter for consulting; state also paid
fiscal agent for some tasks, such as the
processing of employment-related forms.
Consumers paid some fiscal agent fees (such
as for cutting and stopping checks)
$1,097
New Jersey—Personal Preference Program
HCBS = home- and community-based services.
*HCBS services covered under Florida’s waiver included behavioral therapy and personal care supplies, as well as personal care.
ElderChoices provides nurse-supervised homemaker, chore, and respite services to nursing-home qualified elderly adults. Alternatives provides attendant care
and environmental modifications to nonelderly adults and lets them choose and supervise caregivers. Among demonstration enrollees, 62 percent of the elderly
participated in ElderChoices, and 9 percent of the nonelderly participated in Alternatives.
Paid for through pool of money
generated from difference between
$12.36 per hour paid to agencies and
$8.00 per hour rate at which allowance
was cashed out. Originally, agencies
were paid a per-client per-month rate for
counseling and fiscal services, which
was reduced at six-month intervals.
Later in the demonstration, agencies
were paid a fixed rate for developing a
spending plan and then paid per client
per month for counseling and fiscal
services.
Funding for fiscal agent and
counseling services
a
$313
Arkansas—IndependentChoices
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Median monthly prospective
allowance of all
demonstration enrollees
Feature
TABLE 1 (continued)
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
• Unlike the other two states, New Jersey did not apply a discount factor to the
estimated value of the care plan in setting the allowance amount. The discount factor
was intended to adjust the allowance downward to reflect the fact that consumers
receiving agency services typically get somewhat less on average than is called for in
their care plan.
Consumers enrolled in the demonstration over about a two and one-half year period in each
state, with Arkansas beginning in December 1998, New Jersey in late 1999, and Florida in June
2000. States each sought to enroll a target of about 2000 adult consumers; Florida also opened
its program to children with developmental disabilities, and set a target of 1,000 children. States
stopped enrolling beneficiaries into the evaluation when they reached their targets; unless they
had not done so by July 2002, when all enrollment ended. New Jersey and Florida each fell
about 10 percent short of their targets for adults.
The Cash and Counseling Evaluation Design
Wherever possible, program impacts were estimated by comparing outcomes for the entire
treatment group to the entire control group, regardless of whether the treatment group consumer
received the allowance or whether the control group received agency services.5 This intent-totreat approach underestimates the impacts on actual participants, so the results are conservative
estimates of the program’s true impacts on those who receive an allowance. Separate estimates
were obtained for elderly and non-elderly consumers within each of the three states, and for
children in Florida.
Data for the study came from telephone surveys of consumers, caregivers, and workers, and
from Medicaid and Medicare claims data and from program records. Consumers volunteering
5
For some measures, such as satisfaction with the paid care received, only those receiving care could be
analyzed.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 11 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
for the study were required to complete a baseline interview before being randomized. The data
on consumers’ outcomes were obtained from followup surveys conducted 9 months after they
enrolled. Data on their costs were obtained from Medicaid and Medicare claims data for the year
after enrollment for all consumers, and for two years after enrollment for those enrolling during
the first year of intake. Data on outcomes for the primary informal caregivers were obtained
from a telephone survey of these individuals, who were identified by the consumers in the
baseline interview. Data on workers were drawn from a survey of these individuals, who were
identified by the consumers at the time of their 9-month followup interview. Both the informal
caregiver and paid worker surveys were conducted about 10 months after the consumer enrolled
in the study. The proportion of sample members completing the survey ranged from 80 to
92 percent for consumers, depending on the age group, and was typically about 5 percentage
points higher for the treatment group than the control group. For the caregivers, response rates
ranged from 72 to 90 percent, with a similar treatment-control differential. Rates were similar
for the paid workers survey, but the treatment-control difference in rates were somewhat larger,
as expected. Many consumers elected to have proxies respond for them, despite interviewers’
encouraging them to complete the interview themselves and assuring them that they could do so.
Program impacts were estimated using logistic models for binary outcomes and regression
models for the few continuous measures, such as Medicaid costs. We tested hypotheses about
program effects by examining the statistical significance of the coefficient on treatment/control
status in these models; only those outcomes for which this coefficient was significantly different
from zero at the .05 level (for a two-tailed test) were considered to have been affected by the
program. However, to make the magnitudes of effects easier to grasp, we have presented
average predicted values for the outcomes over all sample members, first assuming they are in
the treatment group and then assuming they are in the control group.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 12 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Sample sizes were sufficiently large that we were confident of detecting true program
impacts of any meaningful magnitude within any state-age group category. The minimum
detectable effect at 80 percent power was about 10 percentage points for binary outcomes with a
mean of .4 to .6, for any age group in any state except the smallest one (nonelderly adults in
Arkansas) for which the detectable effect was about 12 percentage points. Detectable effects for
total Medicaid costs were roughly 10 percent of the control group mean for all but the smallest
group. We examined impacts separately for various subgroups, but sample sizes were generally
too small to detect all but very large differences. Thus, only subgroups defined by age are
presented, because of the long-standing question about the suitability of consumer direction for
elderly adults.
Enrollment and Participation in Cash and Counseling6
Relatively few (5 to 8 percent) of both younger and older eligible adult consumers in each of
the states enrolled in the program, but the program attracted a higher proportion of children with
disabilities, with 15 percent being enrolled by their parents. Eligibles receiving relatively large
benefits in the traditional program, those who had been receiving PCS/HCBS services for a
longer time, and those who were alive throughout the state’s 2½ year intake period were more
likely to enroll in the program than eligibles without these characteristics.
Enrollees were quite impaired, in all three states, as expected.
Half to two-thirds of
beneficiaries in any state-age group category needed help moving to or from a bed or chair.
Over half the elderly respondents requested that a proxy complete the baseline interview for
them due to poor health, cognitive problems, or difficulty communicating.
6
Most of the results presented in this section were drawn from Foster et al. (2005a), Schore and Phillips
(2004), and Foster et al. (2005c and 2005d). See those reports for more detailed results and discussion of
methodology.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 13 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
In each age group in each state, a substantial portion of the treatment group consumers never
received an allowance (Table 2). The proportion that did receive an allowance to manage during
their first year after enrollment ranged from only 42 percent for elderly adults in Florida to
89 percent of nonelderly adults in Arkansas. In Arkansas and New Jersey, the great majority of
those receiving their allowance had started getting it by month 6, but about one-third of children
and non-elderly adults in Florida (90 percent of whom had developmental disabilities) who
received their allowance did not get it until 7 to 12 months after enrollment. The disparity across
states was due to differences in recruitment and spending plan approval processes, as discussed
later.
Most beneficiaries (80 to 90 percent) who received an allowance used it to hire workers;
few used it to modify their cars or homes. Consumers tended to hire mainly family members,
but sizable proportions (22 to 42 percent) hired only individuals who were unrelated to them.
The second most popular use for the allowance was purchase of equipment or supplies, typically
communication devices or medical alert systems. Sizable proportions (30 to 60 percent) of each
state-age group took part of their allowance (limited by the states) in cash to cover transportation
costs and incidental expenses.
Finally, while large proportions of program participants would recommend the program to
others seeking more control over their care, a substantial proportion disenrolled from the
program. Among treatment group members at 12 months after enrollment, 15 to 25 percent of
each age group in each state had disenrolled voluntarily (except for Florida’s elderly, for which
the rate was 38 percent).7 Nonetheless, over half the younger adults in each state and 70 percent
of the children in Florida were still receiving allowances two years after enrollment.
7
This disenrollment rate is consistent with the low proportion of consumers who received an allowance.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 14 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 15 of 64
86.0
87.5
89.3
89.3
371
67
291
359
441
1,775
Among Recipients, Started Allowance
by End of (percent)b:
Month 3
Month 6
Month 9
Month 12
Allowance Amount (dollars)c
Mean
Minimum
Bottom quartile
Median
Top quartile
Maximum
453
190
902
142
509
817
1,158
5,655
19.0
35.5
38.9
40.8
31.8
41.7
60+
404
270
1,049
226
604
1,091
1,364
2,436
31.7
58.7
67.1
69.6
59.4
67.2
18 to 64
Schore and Phillips (2004), Foster et al. (2005c, 2005d).
Calculated over only treatment group members who actually received the allowance.
c
b
See Dale and Brown (2005).
a
A few sample members were missing one of the components (such as the discount rate) used to calculate the monthly prospective allowance.
456
262
1,929
93
753
1,465
2,596
28,298
13.8
42.8
54.0
57.9
54.8
57.5
18 to 59
Note:
501
356
1,212
50
418
870
1,663
5,600
21.8
53.9
65.9
71.1
68.3
71.1
3 to 17
31.3
55.0
62.7
64.0
54.6
64.1
65+
467
294
1,066
181
608
1,097
1,413
2,354
New Jersey
Medicaid claims for first row; program records from each demonstration state for all other entries.
725
585
310
29
216
301
414
2,017
79.3
81.5
82.1
83.2
60.0
81.7
65+
Florida
Source:
279
240
71.0
Percent Receiving Allowance in Month 12a
Sample Size (Treatment Group)
All
Recipients Only
88.8
Percent Receiving Allowance at Any Time in
Year 1a
18 to 64
Arkansas
MONTHS FROM RANDOM ASSIGNMENT TO START OF MONTHLY ALLOWANCE
TABLE 2
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Effects on Consumers’ Well-Being
Treatment group members in Arkansas and New Jersey were substantially more likely than
control group members in those two states to receive paid care. With one exception, treatment
group members in all three states also were much more satisfied with the care they received.
Elderly consumers in Florida were the sole exception with respect to satisfaction, as fewer than
half the treatment group members received their allowances (continuing to rely instead on
agency-supplied services).8
Treatment group members were generally more likely to receive paid personal care, but
received comparable total hours (paid and unpaid) of care. Nine months after enrollment, for six
of the seven state-age group combinations we examined, the treatment group was significantly
more likely than the control group to be receiving paid personal assistance during the two-week
reference period (Table 3). The difference was largest in Arkansas, where many beneficiaries
faced limited access to services due to worker shortages, but it also was sizeable in New Jersey
and in Florida (except in the case of elderly consumers). Ninety percent or more of both the
treatment and control group members in every state and age group were receiving some unpaid
assistance at nine months (not shown; see Carlson et al. 2005).
By contrast, the total number of hours of care received was either no different for the
treatment group than for the control group or, among younger adults in Arkansas and older adults
in Florida, was significantly lower for the treatment group. The control group had a very large
number of unpaid hours of care, accounting for 75 to 85 percent of their total hours of care. The
treatment group also reported high levels of unpaid hours, but consistently less than the control
group for consumers of all ages in all three states. The lower number of unpaid hours essentially
8
Results in this section are drawn from Foster et al. (2003), Carlson et al. (2005) and Foster et al. (2004).
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 16 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 17 of 64
473
Sample Size
1,266
101.3
115.1
–13.8*
(.036)
859
197.7
217.1
–19.4**
(.009)
39.5
29.6
9.9**
(<.001)
811
148.1
159.8
–11.7
(.130)
39.4
28.9
10.5**
(<.001)
187.5
188.7
–1.2
(.878)
736
111.8
125.6
–13.7
(.109)
28.0
32.9
–4.9
(.140)
139.9
158.4
–18.6*
(.042)
94.0
91.2
2.8
(.176)
60+
682
106.5
116.7
–10.2
(.242)
38.8
33.2
5.6*
(.023)
145.2
149.9
–4.7
(.612)
91.6
78.7
12.9**
(<.001)
18 to 64
783
94.2
111.7
–17.6*
(.034)
39.1
31.2
7.9**
(<.001)
133.2
142.9
–9.7
(.283)
93.9
81.9
12.0**
(<.001)
65+
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
Means predicted using ordinary least squares regression models. Analysis includes only those with complete data for every component of total hours (about 90
percent of the sample in each state).
b
Effects estimated by pooling the two adult age groups and including and age*treatment status interaction term in the model. Means were predicted using logit
models.
a
Sample is restricted to consumers residing in the community at the time of the 9-month interview. Hours are measured over the two-week period
preceding the interview.
73.6
96.8
–23.2**
(.008)
Unpaid Hoursb
Treatment
Control
Difference
p-Value
22.7
18.2
4.5**
(.001)
237.2
246.8
–9.5
(.228)
76.4
64.2
12.2**
(<.001)
18 to 59
Note:
23.1
23.0
0.2
(.959)
Paid Hours of Careb
Treatment
Control
Difference
p-Value
124.0
133.3
–9.4
(.185)
79.3
65.1
14.2**
(<.001)
3 to 17
New Jersey
Nine-month evaluation interview conducted by Mathematica Policy Research, Inc. between September 1999 and March 2002 for Arkansas, March
2001 and May 2003 for Florida, and August 2000 and June 2003 for New Jersey.
96.7
119.8
–23.1*
(.014)
Total Hours of Careb
Treatment
Control
Difference
p-Value
94.2
78.8
15.4**
(<.001)
65+
Florida
Source:
94.5
67.8
26.7**
(<.001)
Percent Receiving Paid Assistancea
Treatment
Control
Difference
p-Value
18 to 64
Arkansas
PAID AND UNPAID CARE RECEIVED, BY AGE GROUP
TABLE 3
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
offset the higher number of paid hours observed for treatment group members in most state-age
groups.
Treatment group members were much more likely to have their needs met, and to be very
satisfied with their care (Table 4). With the exception of elderly consumers in Florida, treatment
group members were much less likely than control group members to report unmet needs, more
likely to state that their caregivers performed reliably and appropriately, and more satisfied with
the help they received. Relative to control group members, treatment group members had fewer
remaining unmet needs for help with daily living activities, help around the house, and routine
health care, and they reported much higher satisfaction with the way paid caregivers helped with
those services. These differences reflect the treatment group’s higher reported rates of paid
caregivers’ arriving on time and completing their work, and lower rates of being neglected,
treated disrespectfully, or (in some instances) having things stolen from them.
Elderly Florida consumers’ lack of improvement in unmet needs and dissatisfaction with
their care appears to be due to the low proportion of treatment group members who actually
participated in Cash and Counseling (that is, received their allowances). Among all treatment
group recipients of paid care at nine months after enrollment, two-thirds or more of those in the
other state and age groups were purchasing that assistance with their allowances, but only 4 in
10 elderly Florida consumers were doing so (Table 4, bottom row). As a result, any favorable
program effects on elderly consumers in Florida who were receiving the allowance were not
large enough to produce a statistically significant treatment-control group difference in the full
sample of randomized consumers.
The impacts on unmet needs and satisfaction with care were large, as signified by the double
+ signs in Table 4. Table 5 provides some illustrative estimates for representative outcomes in
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 18 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 19 of 64
81
Percentage of Paid Care Recipients in
Treatment Group Who Were
Receiving Allowance at Nine Months
439
++
++
++
++
++
Very Satisfied with Assistance:
Help with daily living tasksa,f
Help around house/communityb
Routine health cared,f
Transportation aidc
Overall care arrangementsg
Sample Size (Maximum)
++
++
1,048
74
+
++
++
++
++
+
++
++
++
++
+
+
++
++
++
++
++
++
++
++
++
++
65+
Paid Caregiver Behavior
Neglected consumer
Was rude or disrespectful
Took something without askingf
Gave unwanted helpf
Consumer very satisfied with
relationshipf
Caregiver Reliability
Always completed taskse
Never arrived late
Very satisfied with schedulef
Easy to change schedule
Unmet Needs for Help with:
Daily livinga
Household activitiesb
Transportationc
Routine health cared
18 to 64
Arkansas
796
76
++
++
++
++
++
++
++
++
++
++
+
++
++
+
++
3 to 17
746
68
++
++
++
++
++
++
++
++
+
+
+
18 to 59
Florida
625
41
+
60+
SUMMARY OF RESULTS ON SATISFACTION WITH AND QUALITY OF PAID CARE
TABLE 4
637
67
++
++
++
++
++
++
++
+
++
++
++
+
++
+
18 to 64
65
++
++
++
++
++
++
+
++
++
++
++
++
++
++
65+
680
New Jersey
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 20 of 64
Nine-month evaluation interview conducted by Mathematica Policy Research, Inc. between September 1999 and March 2002 for Arkansas, March
2001 and May 2003 for Florida, and August 2000 and June 2003 for New Jersey.
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
+Signifies statistically significant treatment-control difference (p < .05) that favors the treatment group and is modest (less than 10 percentage points and less
than half the size of the control group proportion or its complement).
++Signifies statistically significant treatment-control difference (p < .05) that favors the treatment group and is large (at least 10 percentage points or at least half
the size of the control group proportion [pc] or its complement [1-pc]).
Includes arrangements for unpaid and paid help with personal care, activities around the house and community, routine health care, community services,
transportation, and for use of care-related equipment.
g
Effects were estimated by pooling the two adult age groups and including an age*treatment status interaction term in the model.
f
Routine health care includes help with medications, checking blood pressure, and doing exercises.
This measure was derived from a survey question with a five-point scale. The binary variable shown here represents the most favorable rating (always).
e
d
Transportation includes transportation to and from a physician’s office, shopping, school, work, and social and recreational activities.
c
Household activities include preparing meals, doing laundry, doing housework, and doing yard work. Help doing things around the house and community does
not include help with transportation.
b
Daily living activities include eating, dressing, toileting, transferring, and bathing.
a
Source:
TABLE 4 (continued)
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 21 of 64
25.8
41.0
–15.2**
(.001)
10.5
29.5
–18.9**
(<.001)
90.4
64.0
26.4**
(<.001)
71.0
41.9
29.2**
(<.001)
Has Unmet Needs for Help with Daily
Living Activitya
Treatment
Control
Difference
p-Value
Paid Caregiver Was Rude or
Disrespectful
Treatment
Control
Difference
p-Value
Very Satisfied with Way Caregiver
Helped Around House/Communityb
Treatment
Control
Difference
p-Value
Very Satisfied with Overall Care
Arrangementsc
Treatment
Control
Difference
p-Value
18 to 64
11.8
16.4
–4.7
(.051)
35.9
36.5
–0.7
(.823)
65+
68.3
54.0
14.3**
(<.001)
87.3
68.3
19.0**
(<.001)
Arkansas
56.4
26.8
29.7**
(<.001)
85.3
73.1
12.3**
(.001)
10.8
15.1
–4.3
(.097)
32.8
44.6
–11.8**
(<.001)
3 to 17
68.2
48.0
20.2**
(<.001)
85.4
70.9
14.5**
(.001)
16.5
18.0
–1.4
(.671)
26.7
33.8
–7.1*
(.014)
18 to 59
Florida
50.0
46.9
3.1
(.463)
70.4
66.1
4.3
(.351)
15.6
19.7
–4.1
(.228)
42.8
46.5
–3.7
(.336)
60+
UNMET NEEDS AND SATISFACTION RECEIVED WITH PAID CARE
(Percent)
TABLE 5
51.9
35.0
16.9**
(<.001)
84.4
66.0
18.4**
(<.001)
18.7
30.1
–11.4**
(.002)
46.1
54.5
–8.4*
(.028)
18 to 64
65+
56.5
36.6
19.9**
(<.001)
78.9
58.8
20.1**
(<.001)
15.4
20.0
–4.6
(.130)
44.1
57.7
–13.7**
(<.001)
New Jersey
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 22 of 64
746
63.5
50.2
13.3**
(.001)
18 to 59
625
35.9
27.9
8.0*
(.049)
60+
637
37.5
21.0
16.5***
(<.001)
18 to 64
680
47.1
25.3
21.9**
(<.001)
65+
Help doing things around the house/community does not include help with transportation.
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
Includes arrangements for unpaid and paid help with personal care, activities around the house and community, routine health care, community services,
transportation, and for use of care-related equipment.
c
b
Daily living activities include eating, dressing, toileting, transferring, and bathing.
a
Means were predicted using logit models. Sample sizes for some variables in this table were smaller because questions about paid caregiver
behavior were asked only for those with such caregivers, and because of differences in item nonresponse.
796
51.9
28.7
23.2**
(<.001)
3 to 17
New Jersey
Note:
1,048
55.5
37.0
18.5**
(<.001)
65+
Florida
Nine-month evaluation interview conducted by Mathematica Policy Research, Inc. between September 1999 and March 2002 for Arkansas, March
2001 and May 2003 for Florida, and August 2000 and June 2003 for New Jersey.
439
43.4
22.9
20.5**
(<.001)
18 to 64
Arkansas
Source:
Sample Size
Very Satisfied with Life
Treatment
Control
Difference
p-Value
TABLE 5 (continued)
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
each of the four categories of indicators of consumers’ satisfaction with services received.
Despite the services and sizeable amounts of unpaid care received, one-third to one-half or more
of enrollees reported unmet needs for help with personal care, help around the house, help with
routine health care, and help with transportation. For most measures, Cash and Counseling
enabled the treatment group to reduce those unmet needs by 10 to 40 percent below the incidence
for the control group. The treatment-control differences in the proportion reporting that their
caregivers were rude or disrespectful were less dramatic, but still significantly lower for the
treatment group among younger adults in Arkansas and New Jersey, and of the same sizeable
magnitude (as a proportion of the control group mean) across all three states for elderly
consumers. The proportion reporting that they were very satisfied with the different types of
care received, such as help around the house or help traveling around the community, and with
their care overall was much higher for the treatment group in every state and age group with the
exception of older adults in Florida.
Many of the treatment-control differences were significant for all three age groups, but
differences generally were larger for nonelderly adults and children than for older adults. As
Table 5 illustrates, the treatment-control difference in the proportion of nonelderly adults in
Arkansas and in Florida that was very satisfied with their care overall was more than
20 percentage points, but this difference was half that size or less among elderly adults in those
states. (The estimated differences for the two age groups in New Jersey were comparable to
each other.) Differences in the proportion who were dissatisfied with their care also favored
treatment group members; thus, the explanation for the differences is not simply that treatment
group members received better care, but that, compared with control group members, treatment
group members were far less likely to consider the quality of their care to be unacceptable.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 23 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Importantly, treatment group members were no more likely to suffer care-related health
problems. None of the 11 measures of health problems or adverse events examined showed
worse outcomes for the treatment group than the control group, for any state or age group.9
Furthermore, for nearly one-third of the 77 comparisons, the treatment group was significantly
less likely to experience health problems. The significant differences were scattered across
measures, age groups, and states, revealing no consistent pattern. For example, among the four
representative measures presented in Table 6, we find the treatment group to be significantly less
likely than the control group to have fallen (in New Jersey, for both age groups), to have
contractures develop or worsen (for older beneficiaries in Arkansas and Florida), to have urinary
tract infections (nonelderly in Florida), or to have bedsores develop or worsen (younger adults in
Arkansas and New Jersey).
The significant differences are sizeable, ranging from 20 to
50 percent of the control group means. Thus, concerns that consumer direction would place care
recipients at greater risk of injury or illness related to the quality of their care are unwarranted in
the Cash and Counseling model as implemented by the three demonstration states; in some
instances, consumer direction may actually have reduced consumer’s’ risk of such problems.
Finally, on what is perhaps the ultimate measure of the value of Cash and Counseling, we
see that treatment group members were 25 to 90 percent (8 to 23 percentage points) more likely
than control group members to report that they were very satisfied with how they were leading
their lives, and generally half as likely to report that they were dissatisfied with their lives. The
smallest of these overwhelmingly positive effects on consumers’ self-reported quality of life was
for older adults in Florida, the group for which only 40 percent of treatment group members
9
Measures examined, in addition to the four shown in Table 6, included whether saw a doctor due to a fall;
whether saw a doctor because of a cut, burn, or scald; whether injured while receiving paid help; whether shortness
of breath developed or worsened; whether had a respiratory infection; whether current health was poor; and whether
hospitalized or in a nursing home during the past two months (see Carlson et al. 2005 for all results).
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 24 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 25 of 64
5.9
12.6
–6.7*
(.012)
19.4
21.6
–2.2
(.560)
Bedsores Developed/Worseneda,b
Treatment
Control
Difference
p-Value
Had a Urinary Tract Infectionb
Treatment
Control
Difference
p-Value
696
19.5
21.5
–2.0
(.516)
7.9
9.3
–1.4
(.511)
20.0
21.9
–2.0
(.534)
17.5
19.7
–2.2
(.468)
60+
668
16.6
19.4
–2.8
(.329)
9.0
13.0
–4.1
(.094)
24.5
28.1
–3.7
(.269)
18.7
28.0
–9.3**
(.004)
18 to 64
742
15.7
15.8
–0.1
(.966)
7.2
7.1
0.1
(.970)
17.5
27.1
–9.6**
(.002)
13.2
20.4
–7.2**
(.009)
65+
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
For Florida children, impact could not be estimated from the logit model. Results presented are the unadjusted means and treatment-control differences.
Effects were estimated by pooling the two adult age groups and including an age*treatment status interaction term in the model.
b
a
Means were predicted using logit models. Sample sizes for some variables in this table were smaller because of differences in item nonresponse and skip patterns.
808
7.7
11.7
–4.0*
(.043)
4.1
5.9
–1.8
(.252)
9.0
14.0
–5.0*
(.021)
14.5
17.5
–3.0
(.235)
18 to 59
Note:
857
2.5
6.0
–3.5*
(.011)
3.0
6.0
–3.0*
(.033)
9.4
13.4
–4.0*
(.049)
27.3
36.2
–8.9**
(.004)
3 to 17
New Jersey
Nine-month evaluation interview, conducted by Mathematica Policy Research, Inc. between September 1999 and March 2002 for Arkansas, March 2001 and May
2003 for Florida, and August 2000 and June 2003 for New Jersey.
1,164
18.2
21.0
–2.8
(.230)
7.5
6.8
0.7
(.640)
15.9
19.7
–3.9
(.089)
19.0
18.6
0.4
(.869)
65+
Florida
Source:
462
26.0
25.2
0.8
(.826)
Contractures Developed/Worsened
Treatment
Control
Difference
p-Value
Sample Size
28.4
28.7
–0.4
(.931)
Had a Fall
Treatment
Control
Difference
p-Value
18 to 64
Arkansas
(Percent)
CARE-RELATED HEALTH PROBLEMS AND EVENTS
TABLE 6
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
received their Cash and Counseling allowance. Even among this group, however, the treatment
group was significantly more likely (by nearly 30 percent of the control group mean) to report
that it was “very satisfied.” These estimates are buttressed by the findings that half to two-thirds
of allowance recipients in every age group in every state reported that the program “improved
[their] lives a great deal,” and that more than 85 percent in any state or age group would
recommend the program to others wanting more control over their personal care services (Schore
and Phillips 2004; Foster et al. 2005c and 2005d). Thus, the message from the consumers’
perspective is clear—Cash and Counseling led to a major improvement in their care and overall
well-being, in every state and age group.
Effects on Unpaid Caregivers
Under Cash and Counseling, many of the treatment group’s primary informal caregivers at
baseline (29 percent for adults in Florida, 42 percent in New Jersey, and 56 percent in Arkansas)
began receiving pay from consumers during the demonstration. This change in the consumercaregiver relationship affects how we interpret findings for both the paid workers and the unpaid
workers. The findings are quite consistent across adult age groups for nearly all of the outcomes
examined. Thus, each state’s results are for the primary informal caregivers of younger and
older adults combined; results for children are presented separately.10
As expected, nearly all sample members’ primary informal caregivers were relatives (not
shown). Children’s primary unpaid caregivers usually were their mothers. For younger adults,
caregivers usually were parents. For older adults, the caregivers usually were daughters.
In Arkansas and Florida, unpaid treatment group caregivers provided fewer total hours of
care during the two weeks before the interview than did control group caregivers; in New Jersey,
10
Results in this section were drawn from reports on unpaid caregivers for adults (Foster et al. 2005e), unpaid
caregivers for children (Foster et al. 2005b), and paid workers for all age groups (Dale et al. 2005).
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 26 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
this pattern was reversed (Table 7).
In all three states, both treatment and control group
caregivers reported providing more than 100 hours of care during the two most recent weeks at
home before the survey, with live-in caregivers generally reporting twice as many hours as
visiting caregivers. Compared with their respective control group caregivers, treatment group
members in Arkansas reported nine percent fewer total hours of care, and those in Florida
reported about seven percent fewer hours. In New Jersey, the treatment group’s average reported
care hours exceeded the control group’s, by about nine percent.
Although none of these estimates is significantly different from zero at the .05 level, all have
p-values between .05 and .11, suggesting that the differences may be effects of the program,
rather than chance. Separate analysis of this measure by age subgroups showed that this result
for New Jersey was confined entirely to the caregivers (both visiting and live-in) of younger
adults, who provided 20 hours more care than did corresponding control group caregivers during
the reference period. Thus, although Cash and Counseling provided some relief for the primary
unpaid caregivers in two states, the caregiving burden (as measured in hours) of unpaid
caregivers for nonelderly adults in New Jersey increased, on average. In Florida, the total
numbers of hours of care provided by the primary informal caregivers of children (usually the
mothers) was similar for the treatment and control groups, as might be expected.
In all three states, unpaid treatment group caregivers were much more satisfied than control
group caregivers with the care that consumers received, and they worried less about it.
Treatment group caregivers were 18 to 20 percentage points more likely than control group
caregivers to say that they were “very satisfied” with their care recipients’ overall (paid and
unpaid) care arrangements (Table 8). Furthermore, they were only half as likely as the control
group caregivers to report being dissatisfied. (One-fifth to one-third of control group caregivers
were dissatisfied with the beneficiary’s care.) Treatment group caregivers were consistently and
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 27 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
TABLE 7
HOURS OF CARE PROVIDED BY PRIMARY UNPAID CAREGIVERS
Adults
Outcome
Children
Arkansas
Florida
New Jersey
Florida
All Primary Caregivers
Treatment
Control
Difference
p-Value
106.6
117.0
–10.4
.089
123.7
132.7
–9.0
.111
123.2
113.3
9.9
.057
150.1
155.0
–4.9
.353
Among Live-in Caregivers
Treatment
Control
Difference
p-Value
140.1
153.0
–12.9*
.035
139.5
149.3
–9.8
.069
148.4
140.1
8.3
.279
154.1
159.9
–5.8
.227
Among Visiting Caregivers
Treatment
Control
Difference
p-Value
61.9
68.9
–7.0
.164
48.9
54.0
–5.1
.676
84.8
72.5
12.3
.090
—a
—
—
—
1,433
1,193
1,042
Number of Respondents
Source:
a
829
Survey of primary unpaid caregivers conducted by Mathematica Policy Research, Inc.
See Foster et al. (2005b and 2005e).
Only 30 of the primary unpaid caregivers for Florida children were visiting caregivers; hence,
results are not presented for this small group.
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 28 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
TABLE 8
PRIMARY UNPAID CAREGIVERS’ SATISFACTION WITH RECIPIENTS’ CARE
(Percent)
Adults
Outcome
Children
Arkansas
Florida
New Jersey
Florida
Very Satisfied
Treatment
Control
Difference
p-Value
60.8
42.7
18.1 **
<.001
47.9
29.8
18.1 **
<.001
51.6
31.7
19.9 **
<.001
42.3
22.0
20.3**
<.001
Dissatisfied
Treatment
Control
Difference
p-Value
9.1
22.8
-13.7 **
<.001
15.7
27.5
-11.8 **
<.001
13.3
32.2
-18.8 **
<.001
14.6
36.9
-22.4 **
<.001
Recipient Does Not Have Enough Help
Treatment
Control
Difference
p-Value
35.8
53.5
-17.6 **
<.001
47.8
60.5
-12.7 **
<.001
52.2
70.2
-18.0 **
<.001
47.2
64.7
-17.6 **
<.001
Recipient’s Safety Is at Risk
Treatment
Control
Difference
p-Value
39.3
53.4
-14.1 **
<.001
43.1
52.3
-9.2 **
.001
53.5
64.8
-11.3 **
<.001
43.5
57.3
-13.8 **
<.001
14.0
20.3
-6.3 **
.001
22.2
29.2
-7.0 **
.005
24.9
30.3
-5.5 **
.041
25.0
34.7
-9.7 **
.002
Level of Satisfaction with Care Recipient’s
Overall Care Arrangements
When Not with Care Recipient, Worries
Quite a Lot that:
Someone Will Take Recipient’s Belongings
Treatment
Control
Difference
p-Value
Number of Respondents
Source:
1,433
1,193
1,042
829
Survey of primary unpaid caregivers conducted by Mathematica Policy Research, Inc. See Foster
et al. (2005b and 2005e).
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 29 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
significantly less likely than control group caregivers to report worrying that (in their absence)
consumers had insufficient care, were not safe, or would have things stolen from them.
The observed differences were large and statistically significant in each state, ranging from
10 to 100 percent of the control group mean, for caregivers of adults and caregivers of children.
Both ends of the spectrum of satisfaction were affected, with treatment group caregivers less
likely to report that they “worried quite a lot” about these issues, and much more likely to report
that they worried “rarely or not at all.”
In addition to being more satisfied than control group caregivers about the care that
consumers were receiving, treatment group caregivers fared better personally.
They were
significantly less likely than control group caregivers to say that caregiving limited their privacy
or impeded their social lives, and (except in Florida) significantly less likely to say that
caregiving caused severe emotional strain (Table 9). Again, these differences generally were
large. The somewhat smaller differences in Florida may reflect the fact that almost all of
Florida’s children and younger adults had developmental disabilities (potentially a more
emotionally taxing group to care for), and that a lower proportion of treatment group adults in
Florida than in either Arkansas or New Jersey ever received their cash allowances.
Cash and Counseling also appeared to cause fewer work-related and financial problems for
caregivers. About half of both treatment and control group caregivers in each state had jobs
(other than caregiving), and a remarkable one-third of each group (one-half, for children’s
caregivers) reported that caregiving forced them to quit their jobs or reduce their hours (not
shown). Although the program had no effect on caregivers’ hours worked, in all three states,
treatment group caregivers for adults were significantly less likely to report that they could not
look for a job (or different job) because of caregiving responsibilities.
They also were
significantly less likely to say that caregiving caused them to miss or arrive late for work.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 30 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
TABLE 9
EMOTIONAL, PHYSICAL, AND FINANCIAL STRESS ON
PRIMARY INFORMAL CAREGIVERS
(Percent)
Adults
Outcome
Children
Arkansas
Florida
New Jersey
Florida
Caregiving Limits Privacy
Treatment
Control
Difference
p-Value
38.7
52.7
-14.1 **
<.001
52.3
57.1
-4.8
.084
41.1
50.5
-9.4 **
.001
61.0
65.9
-4.9
.125
Limits Free Time/Social Life
Treatment
Control
Difference
p-Value
52.5
63.8
-11.3 **
<.001
66.9
73.3
-6.5 **
.008
54.8
60.1
-5.3
.061
80.9
81.6
-0.7
.778
35.7
38.6
-2.9
.286
42.3
49.4
-7.1 *
.017
39.4
41.6
-2.2
.495
Emotional Indicators
Suffered Great Deal of Emotional
Strain Due to Caregiving
Treatment
Control
Difference
p-Value
26.8
34.3
-7.5 **
.002
Financial Indicators
Wanted to Look for a Job But Didn’t
Treatment
Control
Difference
p-Value
23.5
38.6
-15.1 **
<.001
35.1
41.8
-6.7 *
.011
33.9
44.1
-10.3 **
<.001
52.7
57.0
-4.3
.192
Missed Work or Arrived Late
Treatment
Control
Difference
p-Value
48.6
60.6
-12.0 **
.001
60.9
67.1
-6.2
.095
53.6
65.8
-12.2 **
.002
84.0
82.6
1.4
.657
Experienced Great Deal of Financial
Strain
Treatment
Control
Difference
p-Value
22.4
35.7
-13.3 **
<.001
29.9
38.9
-9.0 **
.001
DRAFT: DO NOT CITE OR DISTRIBUTE
30.0
38.6
-8.6 **
.001
43.7
55.6
-11.9 ***
<.001
Page 31 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
TABLE 9 (continued)
Adults
Outcome
Children
Arkansas
Florida
New Jersey
Florida
Experienced Great Deal of Physical
Strain
Treatment
Control
Difference
p-Value
23.0
32.0
-9.0 **
<.001
28.4
38.8
-10.4 **
<.001
31.7
41.8
-10.1 **
<.001
34.5
42.1
-7.6 *
.020
Physical Health Has Suffered as
Result of Caregiving
Treatment
Control
Difference
p-Value
23.6
34.3
-10.7 **
<.001
32.7
44.9
-12.2 **
<.001
30.7
40.3
-9.6 **
.001
41.8
55.4
-13.6 **
<.001
Current Health Is Fair/Poor Relative to
Peers
Treatment
Control
Difference
p-Value
35.5
46.7
-11.2 **
<.001
31.8
39.6
-7.8 **
.004
30.3
42.3
-12.0 **
<.001
27.4
36.8
-9.4 **
.003
Very satisfied
Treatment
Control
Difference
p-Value
51.3
39.9
11.4 **
<.001
47.0
35.2
11.8 **
<.001
51.6
37.5
14.1 **
<.001
36.9
23.8
13.2 **
<.001
Dissatisfied
Treatment
Control
Difference
p-Value
13.1
23.2
-10.1 **
<.001
16.7
22.8
-6.1 **
.008
15.2
27.3
-12.2 **
<.001
16.7
31.1
-14.4 **
<.001
Number of Respondents
1,433
Physical Well-Being Indicators
Overall Satisfaction with Life
Source:
1,193
1,042
829
Survey of primary unpaid caregivers conducted by Mathematica Policy Research, Inc. Foster et al.
(2005b and 2005e).
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 32 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Furthermore, treatment group caregivers for consumers in every state and every age group
were significantly less likely than control group caregivers to report that they experienced a great
deal of financial strain as a result of caregiving.
Treatment group caregivers were substantially less likely than control group caregivers to
report experiencing a high level of physical strain, and to have suffered physical health problems
as a result of caregiving. They also were much less likely to rate their health as only “fair” or
“poor.” All of these differences in physical well-being measures were large (20 to 30 percent of
the control group mean, not shown), and highly consistent across states and age groups.
These various differences favoring the Cash and Counseling group are reflected in the
significantly greater proportion of treatment group caregivers reporting that they were “very
satisfied” with their lives, and the significantly lower proportions reporting that they were
dissatisfied. Whereas roughly one-fourth of all control group caregivers in each state were very
or somewhat dissatisfied with their lives, the corresponding proportions for the treatment group
caregivers ranged from 13 to 17 percent (not shown).
Treatment group caregivers fared far better than control group caregivers for beneficiaries of
all age groups and in all states, with one exception—treatment group caregivers for nonelderly
adults in New Jersey (not shown; see Foster et al. 2005e). This subgroup was the only one in
which the treatment group caregivers provided more total hours of care than did control group
caregivers, and the only one in which there were no favorable effects on caregivers’ levels of
physical, emotional, or financial strain. Thus, it appears that, if beneficiaries’ participation in
Cash and Counseling leads to an increase in the total hours of care provided by the person who
previously had provided the most unpaid care, the added burden of additional hours may offset
some of the advantages that generally accrue to the primary unpaid caregivers of care recipients
who manage their own care through Cash and Counseling.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 33 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Effects on Paid Workers
The great majority of directly hired workers had provided unpaid care to the consumers
before the consumers had enrolled in Cash and Counseling, with more than one-third having
been the primary unpaid caregiver prior to enrollment, and 40 to 50 percent living with the
consumer (Table 10). About 30 to 40 percent of directly hired workers in each state had
children, and 40 percent had jobs other than caregiving, suggesting that directly hired workers
are vulnerable to stress caused by the many competing demands on their time. Hired workers
were generally in the 40- to 64-year age range, like agency workers, but they were much more
likely to be of the same race as their care recipients. Most of the workers were related to their
care recipients; few (5 to 15 percent) did not know the care recipients prior to the
demonstration.11
Directly hired workers received roughly similar wages as did agency workers, but they were
much more satisfied with their pay. In Florida and New Jersey, treatment group consumers paid
their directly hired workers 10 to 15 percent (about $1) more per hour on average than agency
workers serving the control group were paid, whereas consumers in Arkansas paid about four
percent less per hour than agency wages (Table 11). This result is the opposite of that observed
in the IHSS program, where directly hired workers received substantially lower wages than
agency workers. Almost no directly hired workers received fringe benefits in any of the Cash
and Counseling states, but few agency workers did either.
Directly hired workers were
occasionally paid late, but rarely paid less than they were owed.
In all three states, directly hired workers were twice as likely as agency workers to report
that they were very satisfied with their compensation (Table 12) despite the similarity in wages.
11
See Dale et al. (2005) for complete results on the experiences of paid workers in Cash and Counseling.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 34 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 35 of 64
Relationship to Consumer
Related
Not related; knew consumer
before demonstration
Did not know consumer before
demonstration
n.a.
n.a.
16.4
5.4
n.a.
20.1**
39.7
78.3
51.3
47.0**
79.6**
34.2
57.7
7.5
0.7
51.7
35.6
90.7
Same Race as Consumer
Family and Work Situation
Married
Has children
Currently has job other than
caregiving
37.1
54.5
8.2
0.3
Age
18 to 39
40 to 64
65 to 79
≥80
Directly
Hired
Agency
Workers Workers
Arkansas Adults
15.8
25.8
58.4
40.3
50.2
29.9
83.6
25.2
62.4
11.7
0.7
Directly
Hired
Workers
n.a.
n.a.
n.a.
21.2**
52.6
41.6**
60.2**
**
33.3
62.0
4.3
0.4
Agency
Workers
Florida Adults
10.0
19.1
70.9
42.1
51.4
40.4
84.4
35.9
58.9
5.0
0.3
Directly
Hired
Workers
n.a.
n.a.
n.a.
22.6**
49.3
47.5
54.5**
34.7
61.7
2.3
1.3
Agency
Workers
New Jersey Adults
CHARACTERISTICS OF PRIMARY PAID WORKERS
(Percent)
20.7
25.7
53.6
50.7
50.9
54.5
86.7
44.1
47.8
7.2
0.9
Directly
Hired
Workers
n.a.
n.a.
n.a.
39.5**
51.2
42.7**
72.1**
40.9
55.5
3.1
0.6
Agency
Workers
Florida Children
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
TABLE 10
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 36 of 64
n.a.
84.4
391
Sample Size
308
n.a.
n.a.
n.a.
Agency
Workers
222
66.7
39.2
26.6
Directly
Hired
Workers
164
n.a.
n.a.
n.a.
Agency
Workers
*Mean for directly hired workers different from that of agency workers at .10 level.
**Mean for directly hired workers different from that of agency workers at .05 level.
***Mean for directly hired workers different from that of agency workers at .01 level.
n.a. = not applicable for agency workers. A small proportion of agency workers (less than five percent) were related to consumers,
lived with them, or provided unpaid care to them before the demonstration.
Sample sizes vary slightly for each measure due to item nonresponse.
382
79.6
40.1
46.1
Directly
Hired
Workers
Florida Children
Note:
255
n.a.
n.a.
n.a.
Agency
Workers
New Jersey Adults
Survey of primary paid workers conducted by Mathematica Policy Research, Inc. between September 2000 and May 2003.
See Dale et al. (2005).
298
69.7
46.3
35.2
Directly
Hired
Workers
Florida Adults
Source:
281
n.a.
n.a.
39.4
44.5
Directly
Hired
Agency
Workers Workers
Arkansas Adults
Living/Caregiving Arrangements
Lives with consumer
Is primary informal caregiver
Provided consumer with informal
care before demonstration
TABLE 10 (continued)
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 37 of 64
391
Sample Size
298
10.26
3.5
7.3
29.2
5.1
19.9
26.5
46.4
255
9.03**
16.5**
21.2**
n.a.
n.a.
16.2**
n.a.
n.a.
Agency
Workersa
382
9.84
4.6
6.9
30.8
6.8
20.3
26.5
46.8
Directly
Hired
Workers
308
8.53**
24.2**
15.1**
n.a.
n.a.
18.9
n.a.
n.a.
Agency
Workersa
New Jersey Adults
222
11.81
1.0
7.5
29.8
2.8
18.6
21.2
39.7
Directly
Hired
Workers
*Mean for directly hired workers different from that of agency workers at .05 level.
**Mean for directly hired workers different from that of agency workers at .01 level.
n.a. = not applicable. Only the handful of agency workers who were related to the consumer provided unpaid care to the consumer.
Questions about being paid late or less than owed were not asked of agency workers.
Among those living apart from consumer.
164
11.33
8.6**
7.1
n.a.
n.a.
22.8*
n.a.
n.a.
Agency
Workersa
Florida Children
Responses for agency workers pertain only to the care the workers provided to the consumers who identified them during the ninemonth follow-up survey as their primary worker.
b
a
281
6.30**
20.6**
57.8**
n.a.
n.a.
11.7
n.a.
n.a.
Agency
Workera
Directly
Hired
Workers
Florida Adults
Survey of primary paidworkers Mathematica Policy Research, Inc. between September 2000 and May 2003. Dale et al.
(2005).
6.07
1.6
5.8
35.1
6.7
Compensation
Hourly wage (dollars)
Received fringe benefits (percent)
Paid for travel timeb (percent)
Ever paid late (percent)
Ever paid less than owed (percent)
Source:
12.5
25.7
38.2
Hours of Care Provided per Week
(Percent)
Average paid hours
Average unpaid hours
Total hours
Directly
Hired
Workers
Arkansas Adults
PRIMARY PAID WORKERS’ HOURS OF CARE PROVIDED AND COMPENSATION RECEIVED
TABLE 11
Conumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 38 of 64
391
Sample Size
298
21.4
47.5
5.7
29.6
85.3
1.0
50.7
16.3
Directly
Hired
Workers
255
16.5
59.1**
4.3
33.9
82.6
2.4
22.9**
38.0**
Agency
Workers
Florida Adults
382
29.2
40.2
3.7
28.4
79.3
1.3
41.4
16.9
Directly
Hired
Workers
308
18.9**
42.0
6.5
42..0**
69.9**
3.3
19.1**
41.8**
Agency
Workers
New Jersey Adults
222
21.6
44.0
8.6
21.2
88.7
1.4
61.6
8.4
Directly
Hired
Workers
164
12.4*
51.2
*Mean for directly hired workers different from that of agency workers at .05 level.
**Mean for directly hired workers different from that of agency workers at .01 level.
4.9
32.5*
84.7
2.5
23.0**
41.0**
Agency
Workers
Florida Children
Survey of primary paid workers conducted by Mathematica Policy Research, Inc. between September 2000 and May 2003. Dale et al. (2005).
n.a. = not applicable for agency workers.
Source:
22.4**
37.1
281
70.1**
1.4
3.6
60.2
15.1
Little or No Emotional Strain
Consumer’s Family and Friends Needed to Be
More Respectful
Emotional Well-Being
Great Deal of Physician Strain
Suffered Any Injury
81.8
2.1
22.2**
37.5**
Agency
Workers
16.9
83.4
1.0
Working Conditions Overall
Very satisfied
Not satisfied
Physical Well-Being
44.6
15.6
Wages and Fringe Benefits
Very satisfied
Not satisfied
Satisfaction with:
Directly
Hired
Workers
Arkansas Adults
PAID WORKERS’ STRESS AND SATISFACTION WITH WORKING CONDITIONS
(Percent)
TABLE 12
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
On other dimensions of their jobs, the two groups of paid workers reported similar, high
rates of satisfaction. Only 50 to 70 percent of directly hired workers said that they received
training on how to perform their jobs, compared with 95 percent or more of agency workers (not
shown); however, the groups were equally likely to feel prepared to handle their responsibilities.
Directly hired workers and agency workers experienced similar levels of physical strain and jobrelated injuries. Between 17 and 30 percent of directly hired workers reported “a great deal” of
physical strain, although few (3 to 6 percent) reported being injured while providing care (Table
12). These rates are generally similar to rates reported by agency workers, although hired
workers in New Jersey were significantly less likely than agency workers in that state to report
high levels of physical strain. Hired workers in Arkansas were somewhat more likely than
agency workers to report being injured, although the proportion was small, and the difference
was reversed from the pattern for New Jersey. Furthermore, the difference in caregiving-related
injuries in Arkansas disappeared when we used a regression model to control for the much larger
total number of hours of care provided by the directly hired workers. Thus, we find no evidence
that hired workers suffered more physical problems than is normal for the tasks they were
performing and the hours of care provided, even though hired workers were much less likely to
have received formal training than were agency workers.
Directly hired workers in Arkansas and Florida were significantly more likely than agency
workers in those states to report emotional strain (Table 12). They also were more likely to say
that they received too little respect from the care recipients’ families and friends. Across states,
40 to 60 percent of the directly hired workers reported “some” or “a great deal” of emotional
strain. Agency workers in both Arkansas and Florida reported lower levels of such stress than
the directly hired workers. More than three-fourths of the hired workers in all three states felt
very emotionally close to their care recipients (not shown), and they were no more likely than
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 39 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
agency workers to say that the care recipient should have shown more respect. However, they
were much more likely to say that the care recipients’ families and friends needed to show them
more respect.
These differences appear to be due entirely to the familial relationship that most of the
directly hired workers had with the beneficiaries whom they cared for. After pooling the data
across states to yield adequate sample sizes, we see that directly hired workers who were not
related to their care recipients reported rates of emotional strain and lack of respect from the care
recipients’ families that are very similar to the rates reported by agency workers (Table 13).
Their observed rates of these problems were significantly lower than those of directly hired
workers who are related to the care recipients. Unrelated hired workers also received higher
wages, were more likely to receive training, and provided far fewer unpaid hours of care than did
related workers. Not surprisingly, it is the family dynamics and overall burden of care that
appear to explain why hired workers had more emotional problems, rather than the fact that the
worker was hired by a consumer instead of being employed by an agency.
Effects of Cash and Counseling on Medicaid Costs
The Cash and Counseling program was not designed to save money, but rather, to give
consumers much greater control and flexibility over their care without costing Medicaid any
more per month of benefits received than that care would have cost under the traditional agencybased model. In addition, states are likely to want to know how introduction of Cash and
Counseling is likely to affect their total Medicaid costs for cashed out services, and whether the
program leads to higher or lower costs for other Medicaid services. Finally, the sources of cost
increases or savings are important, as higher treatment group costs resulting from the failure of
the traditional program to serve the control group adequately may have different policy
implications than higher costs resulting from program design issues.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 40 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 41 of 64
41
279
18
2
17.0
59.1
12.0
9.29
41.5
83.0
320
19
7
18.9
56.8
13.0
9.11
42.5
83.3
`
844
16
2
19.3
56.7
14.4
7.93
21.3
82.4
*Related workers different from unrelated workers at the .05 level, two-tailed test.
**Related workers different from unrelated workers at the .01 level, two-tailed test.
Data from all three states are combined here to provide adequate sample sizes for comparisons across subgroups of directly hired workers defined
by their relationship to the consumer.
a
Sample sizes vary slightly for each measure due to item nonresponse.
751
22
48
31.7
41.5
19.5
7.63
48.8
85.4
Agency
Workers
Note:
347
17**
34**
34.6*
46.5**
19.0**
8.34**
46.3
82.1
All
Unrelated
Survey of primary paid workers conducted by Mathematica Policy Research, Inc. between September 2000 and May 2003. Dale et al.
(2005).
404
14
12
27.4
40.9
19
53
53.0
14.5
7.98
47.1
84.4
40.8
22.9
8.64
45.5
80.0
Live-In
Not
Live-In
All Related
Not
Live-In
Source:
Sample Sizea
Hours of Care Provided (Number)
Paid hours
Unpaid hours
Worker Well-Being (Percent)
Little or no emotional strain
Consumer needed to be more respectful
Consumer’s family and friends needed to be
more respectful
Working Conditions
Hourly wage (dollars)
Satisfied with wages (percent)
Satisfied with working conditions (percent)
Live-In
Unrelated
Related
Directly Hired Workers
SELECTED PAID WORKER OUTCOMES, BY CONSUMER-WORKER RELATIONSHIP
TABLE 13
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
We found that Medicaid costs for the costed-out services (personal care in Arkansas and
New Jersey, and waiver services in Florida) were significantly and substantially higher for the
treatment group than for the control group in each state for each age group (with the exception of
elderly consumers in Florida), but that other Medicaid costs typically were at least somewhat
lower for the treatment group. The treatment group’s personal care/waiver costs remained higher
into the second year after enrollment, but the effects on other Medicaid costs were less
consistent. Furthermore, the reasons why the treatment group’s personal care/waiver costs were
higher differed among the three states. Each of the states have instituted important changes in
their ongoing programs that are expected to reduce or eliminate the cost disparities between Cash
and Counseling and agency-provided care.12
On average, Medicaid personal care/waiver costs were substantially higher for the treatment
group than for the control group, in six of the seven state-age group combinations, and for both
the first and second years after enrollment (Table 14). However, the magnitudes of the cost
differences varied widely across the three states. In Arkansas, average personal care/waiver
costs per treatment group member for all adults were double the average care costs per control
group member in both years, compared with a difference of only about 15 percent in Florida in
both years (and limited to nonelderly consumers). In New Jersey, average treatment-control
personal care/waiver costs differed by 16 percent in year 1, but by 29 percent in year 2. The
treatment-control cost differences were somewhat smaller for older adults than for both younger
adults and children in each state in year 1.
The absolute amounts of average personal
care/waiver expenditures varied widely across the states, ranging during the first year from about
12
Results in this section are drawn from Dale and Brown (2005) for adults, and from Dale et al. (2004) for
children in Florida. See also Dale et al. (2003).
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 42 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 43 of 64
—
Year 2
2,349
All adults
Arkansas
Florida
14,046
12,647
15,978
14,193
10,063
18,321
34.3**
26.2**
+14.9**
+14.9**
+4.3
+20.2**
—
—
–17.3**
–8.7*
–4.2
–16.7*
—
—
8,792
9,970
10,650
9,220
—
—
8,743
8,339
7,530
10,432
Control Group Means (Dollars)
—
—
28.9**
+15.9**
+11.7**
+21.1**
16,830
16,448
5,698
5,780
5,771
5,785
–13.3**
–14.8**
6.0
–5.8
–5.1
–6.4
—
—
10,861
12,540
8,757
16,829
—
—
–1.5
–5.8
–4.7
–6.7
New
Jersey
—
—
10,582
10,688
9,822
12,862
—
—
4.7
+14.3**
+17.3**
+9.8
Arkansas
30,877
29,095
21,676
19,973
15,833
24,106
8.4
3.0
12.5**
+8.9**
+0.9
+13.8**
Florida
—
—
19,653
22,509
19,407
26,049
—
—
12.1**
+3.8
+4.3
+3.1
New
Jersey
Total Medicaid Costs
**Treatment-control difference is significantly different from zero at the .05 level.
Year 2 results were calculated for only those early enrollees for whom complete Medicaid claims data for their second year were available at the time the claims data
were provided by the state. Early enrollees were those who enrolled in the demonstration before May 2000 in Arkansas, January 2002 in New Jersey, and October 2001
in Florida. See Table 15 for sample size.
a
New Jersey
Other Medicaid Costs
Effects as Percent of Control Group Mean
Florida
Medicaid claims data. See Dale and Brown (2005).
—
Year 2
Source:
—
Year 1
Children
All adults (early cohort only)a
1,839
2,292
Ages 65+
Year 2
2,430
Ages 18 to 64
Year 1
Adults
—
Year 1
Children
All adults (early cohort only)a
+109.5**
+96.0**
All adults
Year 2
+88.2**
+123.7**
Arkansas
Personal Care Costs
Ages 65+
Ages 18 to 64
Year 1
Adults
Treatment-Control Differences
in Cost as a Percentage of
Control Group Costs
MEDICAID COSTS FOR PERSONAL CARE AND OTHER SERVICES
TABLE 14
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
$4,600 for all adult treatment group members in Arkansas, to $11,600 in New Jersey, and to
$16,000 in Florida. Personal care/waiver costs were especially high in Florida for children
($16,000) and for nonelderly adults, 90 percent of whom had developmental disabilities
($22,000).13
The treatment group’s personal care/waiver costs were higher both because treatment group
members were more likely than control group members to receive any paid care (in Arkansas
and New Jersey) and because average Medicaid payments per month of benefits received were
higher for the treatment group (in some cases). In all three states, nonelderly treatment group
adults (and children in Florida) had significantly higher costs per recipient month than did
nonelderly control group consumers; for elderly consumers, by contrast, cost per recipient month
was significantly higher for the treatment group only in Arkansas. The significant differences
ranged from 4 to 25 percent of the control group mean in year 1(Table 15).
For policymakers, the important factor is perhaps how actual costs compare with the costs
that would be expected, had consumers received the services to which they were entitled through
the traditional system.14 To assess this issue, we calculated the ratio of the actual average
Medicaid cost for the allowance (plus consultant costs) for treatment group members who
received allowances to their average expected cost, computed from the number of hours or
amounts in their care plans at enrollment. (In Arkansas and Florida, this calculation included
multiplying the dollar value of the services authorized in the care plan services by the discount
13
Year 2 results could be estimated only on the earlier enrollees for whom claims data were available at the
time the analysis was conducted. Sensitivity tests suggested that the year 1 results for the early cohort were not
markedly different from those for the full sample. Thus, differences in results between year 1 and year 2 after
enrollment are not due to the differences in the samples for the two periods (see Dale and Brown 2005).
14
Although the waiver’s budget neutrality provisions require that the Medicaid cost per recipient month for a
defined set of core services (see Table 1) be no higher for the treatment than the control group over the life of the
demonstration, that requirement need not be met during the first year.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 44 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
TABLE 15
COST PER RECIPIENT PER MONTH FOR PCS/WAIVER SERVICES
(Dollars)
Year 1
Year 2
Nonelderly
Elderly
All
Adults
Children
All
Adultsa
Children
Arkansas
Treatment
Control
Difference
p-Value
513
422
91**
<.001
420
336
84**
<.001
445
359
86**
<.001
—
—
—
—
467
369
98**
<.001
—
—
—
—
Florida
Treatment
Control
Difference
p-Value
1,884
1,593
291**
<.001
983
967
16
.509
1,460
1,292
168**
<.001
1,378
1,099
279**
<.001
1,814
1,630
184**
<.001
1,660
1,251
409**
<.001
New Jersey
Treatment
Control
Difference
p-Value
1,153
1,106
47*
.043
1,170
1,172
-2
.926
1,164
1,140
25
.112
—
—
—
—
1,264
1,219
45
.051
—
—
—
—
Sample Sizes
Arkansas
Florida
New Jersey
556
913
813
1,452
904
917
2,008
1,817
1,730
—
1,002
—
1,312
1,424
1,447
—
1,002
—
Source:
Medicaid claims data. See Dale and Brown (2005) for adults and Dale et al. (2004) for
children.
a
Year 2 results were calculated for only those early enrollees for whom complete Medicaid claims data for
their second year were available at the time the claims data were provided by the state. Early enrollees
were those who enrolled in the demonstration before May 2000 in Arkansas, January 2002 in New
Jersey, and October 2001 in Florida. Year 2 data were available for all children in the study.
PCS = personal care services.
*Significantly different from zero at the .05 level, two-tailed test.
**Significantly different from zero at the .01 level, two-tailed test.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 45 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
factor applicable to that individual. New Jersey did not discount the allowance.) We performed
the same calculations for the control group and plotted the ratios of actual to expected costs for
each of the first 24 months after enrollment (Figures 1a through 3c).
In Arkansas and New Jersey, it was a failure of the agency system to deliver the authorized
amounts of care to control group members that led to higher costs per recipient month for the
treatment group compared with for the control group, whereas in Florida, the treatment group’s
allowances were excessive for the consumers (adults and children) receiving services under the
developmental disability waiver. For both adult age groups in both Arkansas and New Jersey,
the plotted ratios show that allowance recipients in the treatment group received about what they
were expected to get (ratios of about 1.0), but, with the exception of elderly beneficiaries in New
Jersey, control group personal care/waiver recipients received less than the expected amounts.
This finding was unexpected, because the care plans had been adjusted when determining the
allowance amounts to account for historic gaps between care plan recommendations and the
amount of care actually received. The shortfall was particularly sizeable in Arkansas, where
control group recipients received only about 80 percent of the (already discounted) expected
amounts.
The pattern in Florida was quite different from the ones in both the other states, and it
differed across age group as well. In each age group, the treatment group members who received
allowances received more than had been expected based on their care plans (30 percent more for
children, 20 percent more for adults younger than age 60, and 10 percent more for adults aged
60 or older). Among control group members, nonelderly waiver recipients received about what
was expected, but older ones received more than expected. Thus, cost per recipient month was
actually about the same for treatment group and control group members among Florida’s elderly
care recipients, but substantially higher for the treatment group among younger care recipients.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 46 of 64
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 47 of 64
Mean Actual / Mean Expected Costs
0
0.2
0.4
0.6
0.8
1
1.2
1.4
2
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Month
Control Group Recipeints
Allowance Recipients (including allowances and fiscal agent/counseling costs)
Allowance Recipients (including allowances only)
3
Notes: The first month is excluded for allowance recipients, since the first allowance often was prorated. The care plan is
discounted in calculating expected costs.
1
Figure 1a
Arkansas's Ratios of Mean Actual to Mean Expected Cost Ratios for the Nonelderly
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 48 of 64
Mean Actual / Mean Expected Costs
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Notes:
1
3
5
6
7
8
9
10
11
12
13
M onth
14
15
16
17
18
19
20
21
22
Control Group Recipients
Allowance Recipients (including allowances and fiscal agent/counseling costs)
Allowance Recipients (including allowances only)
4
23
24
The first month is excluded for allowance recipients, since the first allowance often was prorated. The care plan is
discounted in calculating expected costs.
2
Figure 1b
Arkansas's Ratios of Mean Actual to Mean Expected Cost Ratios for the Elderly
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 49 of 64
Mean Actual Costs / Mean Expected Costs
0
0.2
0.4
0.6
0.8
1
1.2
2
3
4
5
6
8
9
10
11
12 13
M onth
C ontrol G roup R ecipients
7
14
16
17
18
19
A llowance R ecipients
15
20
21
22
23
Notes: Early months are excluded for allowance recipients since few (less than 100) individuals had received full
allowances.
1
Figure 2a
New Jersey's Ratios of Mean Actual to Mean Expected Costs for the Nonelderly
24
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 50 of 64
Mean Actual Costs / Mean Expected Costs
0
0.2
0.4
0.6
0.8
1
1.2
2
3
4
5
6
7
9
10
11
13
M on th
12
C ontrol G roup R ecipients
8
14
16
17
18
19
A llow ance R ecipients
15
20
21
22
23
Notes: Early months are excluded for allowance recipients since few (less than 100) individuals had received full
allowances.
1
Figure 2b
New Jersey's Ratios of Mean Actual to Mean Expected Costs for the Elderly
24
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 51 of 64
Mean Actual Costs / Mean Expected Costs
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1
3
4
5
6
7
8
10
11
12 13
M onth
Control G roup R ecipients
9
14
16
17
18
19
A llowance R ecipients
15
20
21
22
Notes: Early months are excluded for allowance recipients since few (less than 100) individuals had received full
allowances.
2
Figure 3a
Florida's Ratios of Mean Actual to Mean Expected Costs for the Nonelderly
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
23
24
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 52 of 64
Mean Actual Costs / Mean Expected Costs
1
2
3
4
5
6
7
8
10
11
13
Month
12
Control Group Recipients
9
14
16
17
18
Allowance Recipients
15
19
20
21
22
Notes: Data for early months are excluded for allowance recipients since few (less than 50) individuals received
full allowances. Baseline costs include the discounted care plan value plus expected fiscal agent and
counseling costs.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Figure 3b
Florida's Ratios of Mean Actual to Mean Expected Costs for the Elderly
23
24
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 53 of 64
Mean Spending / Monthly Care Plan
4
5
6
8
9
10
11
13
M on th
12
C ontrol G roup R ecipients
7
14
16
17
18
19
20
A llow ance R ecipients
15
21
22
D ata for early m onths are excluded for allow ance recip ients, as few er than 1 1 5 (or o ne-third o f allow ance
recipients) received full allow ances. T he care plan is discounted in calculating expected costs.
3
N otes:
2
M ed icaid claim s d ata and care plan d ata. S ee D ale et al. (2004 ).
1
S ou rce:
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
F igu re 3c
F lorid a's R atios of M ean A ctu al to M ean E xp ected C osts for C h ild ren
23
24
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
The treatment-control difference was even more striking for children.
Whereas the
treatment group’s average waiver costs were about 40 percent more than the average discounted
amount in the baseline care plan, the control group had average waiver costs per month of
benefits that were about 10 to 20 percent less than the average discounted care plan amount.
This differential pattern across states and age groups in the ratio of actual to expected costs
per month of benefits received for both treatment and control groups appears to be due to a
number of unique factors. The failure of the traditional system to provide the benefit recipients
with even the discounted level of services in Arkansas and, to a lesser degree, in the two other
states for some age groups, appears likely to be due to worker shortages faced by agencies.
Agencies seeking to maximize either profits or consumer satisfaction would surely have the
incentive to provide all of the care for which a consumers has been authorized, if they could do
so.
The very high ratios of actual to expected costs for Florida’s Cash and Counseling
participants were are under the developmental disabilities waiver (children and nonelderly
adults) appear to be due to a coincidental increase in the availability of funds for this population.
As we noted above, Florida intended to set the allowance amount for each consumer at the
amount that Medicaid had been spending on waiver services for that specific consumer before he
or she had enrolled, discounted by eight percent to reflect historic differences between actual
costs and care plan amounts. However, that practice was not systematically followed for those
under the developmental disability waiver in Florida; as a result of a substantial increase in state
funding for the HCBS waiver programs serving people with developmental disabilities, care
plans were being systematically revised upward during much of the demonstration period. Thus,
when representatives for treatment group members in the developmental disabilities population
met with state counselors to discuss the allowance amounts consumers would receive, those who
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 54 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
sought allowances higher than the amount that Medicaid had been spending on them were likely
to have had a favorable reception. Counselors, following the mandate to increase spending for
this population’s waiver services, may have suggested sizeable increases themselves. Control
group members, on the other hand, would have had no reason to have their care plans
reexamined during that period.
Costs for Medicaid services other than the personal care or waiver services that the
allowance was intended to replace were lower for the treatment group than for the control group
in every state-age group category during the first year after enrollment. However, the differences
were large and statistically significant only for younger adults in Arkansas and for children in
Florida. For those two groups, other Medicaid costs were about 15 and 17 percent lower,
respectively, for the treatment group. For all other age groups, the treatment groups’ costs were
only about four to seven percent below those of the corresponding control group costs. The
particular types of services for which costs were lower were primarily costs related to long-term
care (not shown; see Dale and Brown 2005, Tables 4a to 7c) although this differed somewhat
across states and age groups. The main cost reductions in Arkansas were for nursing facility,
hospital, and home health care, and for ElderChoices (the supplementary waiver program
covering additional personal care hours for older Medicaid beneficiaries beyond what was
offered under the state’s personal care benefit).
Medicaid costs for several other services
(laboratory services, physician visits, and durable medical equipment) also were somewhat lower
for the treatment group (for the nonelderly).
In New Jersey, the treatment group had
significantly lower nursing home expenditures and home health expenditures. For adults in
Florida, the treatment group had slightly lower costs than did the control group for nursing home
and inpatient care, but these differences were not statistically significant. For children, the major
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 55 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
source of the difference in non-waiver Medicaid costs was the treatment group’s nearly
30 percent lower cost for private duty nursing ($4,773, versus $6,639 for the control group).
The second postenrollment year’s results followed a pattern similar to those of the first year;
treatment-control differences in other Medicaid costs were statistically significant only for
children in Florida and for adults in Arkansas.15 However, the magnitudes of some of the
differences changed substantially; in Arkansas in year 2, other Medicaid costs for the treatment
group were 17 percent below those for the control group, twice the difference observed during
year 1. The treatment-control difference for adults in New Jersey changed in the opposite
direction. White it was statistically insignificant in both years, the treatment group’s mean was
six percent lower than the control group’s in year 1, whereas in year 2, it was six percent higher
than the control group’s. For New Jersey adults and for children in Florida, the treatment-control
differences in year 2 were similar to those observed in year 1.
The treatment group’s lower cost for long-term care and other services partially offset its
higher personal care/waiver costs, resulting in differences in total Medicaid costs that were
statistically significant only for elderly consumers in Arkansas (17 percent) and for younger
adults in Florida (14 percent) in the first year after enrollment. For four of the five other stateage group categories, treatment group costs were four percent or less above those of the control
group (not shown).
The results for most groups for the second postenrollment year are less encouraging. In both
Florida and New Jersey, total Medicaid costs for all adults were significantly higher for
treatment group members than for control group members, by about 12 percent. For children,
15
In Year 2, the treatment-control differences in other Medicaid costs were statistically significant for both the
elderly group and the non-elderly group (not shown; see Dale and Brown 2005, Table A.2a). In year 1, the
treatment-control difference for elderly consumers was not statistically significant.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 56 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
the treatment-control difference grew from only three percent of the control group mean in year
1 to eight percent (p = .082 in year 2). Only in Arkansas does the trend suggest that the
treatment-control difference in total cost was shrinking over time—the significant 14 percent
year 1 difference decreased to an insignificant 4.7 percent in year 2.
The reason for the change in Cash and Counseling’s cost impacts from year 1 to year 2 is
different for each state. In Arkansas, the treatment-control gap in total Medicaid costs narrowed
because the unfavorable treatment-control difference in personal care costs decreased by about
$500, while the favorable difference in other Medicaid costs (mostly for nursing home care)
increased by $500. In New Jersey, the trend was exactly the opposite; the treatment group’s
16 percent higher personal care costs nearly doubled in year 2, to 29 percent, while the modest
“savings” in other Medicaid services of about 6 percent in year 1 essentially disappeared in year
2. The somewhat less favorable year 2 results for children in Florida are due to the increase in
the treatment-control difference in costs for waiver services.
Finally, examination of Medicare costs and services showed no statistically significant
treatment-control differences, for any state, in either year.
This result was not surprising.
Neither the states nor the national program office expected that offering consumers more
flexibility in managing their personal care would lead to fewer hospitalizations or to fewer uses
of the other acute care services covered under Medicare.
One question raised by the cost results is whether the favorable effects of consumer
direction on satisfaction and unmet needs are due largely to the states’ spending more per
recipient per month for those in consumer direction? Sensitivity analysis in which we estimated
effects on satisfaction and unmet needs controlling for the ratio of each consumer’s actual
benefits received to the allowance amount that was offered to them when they applied to the
program showed that program impacts were not due to this effect. Satisfaction was only weakly
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 57 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
related to this ratio, and program effects were very large, so despite the higher ratios for the
treatment than the control group in most of the state-age group subgroups examined, program
impacts were largely unaffected by controlling for this ratio.
KEY POLICY AND PRACTICE ISSUES FOR THE FUTURE
Consumer direction appears to be here to stay and growing. It is a program that appeals to
both political conservatives and liberals, and has strong advocacy groups that enthusiastically
support it. Favorable early findings from Cash and Counseling on the impacts on consumers led
ASPE and RWJF to offer grants and technical assistance for 11 additional states to develop their
own Cash and Counseling programs, and led the three demonstration states to continue offering
the programs they developed for the demonstration. Other states are pursuing their own versions
of consumer direction, with the assistance of CMS, which has relaxed some of the restrictions on
consumer directed care options offered under 1915(c) waivers and is providing guidance to states
on how to design their waivers to ensure their acceptability.
Despite states’ march toward consumer direction, however, and the many lessons learned
about consumer direction from the Cash and Counseling demonstration and its predecessors,
there remain a number of concerns about the programs that will require ongoing monitoring and
weighing of tradeoffs.16
• Cost. States’ Medicaid budgets are extremely tight, and cuts to Medicaid funding are
being considered by Congress. This scenario makes it tempting to look toward
consumer direction as a way to save money, given that states do not need to pay
agency overhead if they provide allowances directly to consumers instead of
providing agency care. However, so far Cash and Counseling has cost states more,
not less, than the traditional system per month of service. States may be leery of
adopting an option that increases their costs relative to what they would have been,
regardless of the fact that the higher cost appear to be due to the failure of the
16
See Phillips et al. (2003) for detailed lessons learned about implementing a Cash and Counseling program.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 58 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
traditional system to provide authorized services to eligible beneficiaries. Many
valuable lessons have been learned by the demonstration states about ways to control
these costs (see Phillips et al. 2003; Dale and Brown 2005); states wishing to adopt a
Cash and Counseling program but concerned about costs should take these lessons
into account in designing and implementing their programs. Most importantly,
perhaps, states should monitor the costs of personal care or HCBS relative to care
plans on a routine basis, for both the consumer directed and traditional provider
options.
• Effects on Consumer Demand. Does offering a flexible spending allowance lead
some eligible individuals to apply for PCS or waiver benefits who would not
otherwise have done so (thereby increasing overall cost to the state)? In Arkansas,
the fact that only one-third of control group members not already receiving Medicaid
PCS at the time of enrollment received it in the year after enrollment suggests that
Cash and Counseling may have had such an effect. However, Arkansas had expected,
and hoped, that some of the Arkansans who enrolled would be beneficiaries who were
unable to get agency care, due to their remote location or worker shortages. Thus,
some increase in the number of people receiving PCS benefits was desired. Florida
and New Jersey minimized the potential for such increases by limiting their programs
to people who were already receiving agency services or had been assessed by an
agency as eligible for them. Such restrictions can limit the amount of “induced
demand” (at least until consumers learn about the consumer directed option), but are
likely to defeat the laudable goal some states may have of increasing access to needed
care. Ultimately, however, such concerns may be much ado about (virtually)
nothing—the proportion of eligible beneficiaries that enrolled was low, despite
aggressive efforts by the states to promote the program. Furthermore, the proportion
of Arkansas’s program enrollees who were not already receiving Medicaid PCS
services from agencies at the time they enrolled only briefly reached the state’s
historic average of enrollees in a given year who are “new” recipients. Thus, it seems
likely that offering an allowance instead of services will have negligible effects on the
number of beneficiaries receiving PCS or HCBS services.
• Hiring legally responsible relatives. Despite the finding that there were very few
instances of fraud or abuse of the benefit, some states may still be opposed to
allowing consumers to hire relatives who are legally responsible for the consumer
(such as spouses, parents of minors, or other legal guardians). While paying people
who are legally responsible for a beneficiary to provide care for them will strike some
policy makers as inappropriate, the Cash and Counseling experience suggests that
these individuals continue to provide large amounts of unpaid care. Eliminating them
as potential paid workers is likely to mean that some interested consumers will not be
able to find a worker who is acceptable to them, and therefore will not be able to
participate. Furthermore, the lower levels of emotional, physical, and financial stress
for primary informal caregivers (often legally responsible relatives) who become paid
workers under the program may enable them to continue providing the large amounts
of paid and unpaid care needed for the consumer to be able to remain in the
community.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 59 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
• Allowing representatives to be hired workers. One issue states struggled with is
whether an individual should be permitted to be both the representative (who may
oversee use of the allowance) and a hired worker for a consumer. Ultimately, all
three states decided against allowing this, except in unusual cases (such as a single
parent of a young child). While offering some protection against fraud and abuse,
this too severely constrains some consumers’ choices.
• Hired relatives’ emotional stress. Cash and Counseling states provided relatively
little support for the workers hired by consumers, and only about half of these
workers received any training in how to carry out their tasks. Some modest amount
of support, such as educational materials, videos, and referrals to support groups may
help reduce the high levels of emotional stress many hired relatives experience.
Counselors’ attention to the hired workers’ well-being as well as the consumers may
also help reduce stress levels and address the problem of hired relatives feeling that
they do not receive enough respect from other family members of the consumer.
Finally, in addition to these and other particular concerns, a great many critical program
design issues with important implications for consumers, caregivers, and policy makers must be
worked out by states contemplating the adoption or expansion of consumer direction. These
design issues include determining to whom the program should be offered, what services are
cashed out, how the allowance amount is established initially and revised over time, what the
allowance can be used for, who can be hired, how the allowance is accessed by consumers, the
counselors’ role, how to select the fiscal intermediary and how much to pay for this service,
whether and when to recoup allowance funds not spent by consumers, and monitoring of
spending plans. For example, states must decide whether to offer consumer direction to: only
participants in certain waiver programs or the state plan PCS, only current recipients of the
cashed out services, elderly adults, children, those expected to die within six months, only people
able to make decisions for themselves, and only consumers not participating in other waiver
programs. In setting the allowance amounts, states must decide whether to base it on care
received in the past, the care plan, a new assessment using a common assessment tool that is
administered without the assessor knowing whether the consumer intends to choose agency or
consumer directed care, how much to set aside to cover the cost of counseling and fiscal
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 60 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
intermediary activities, rules regarding unplanned reassessments, and whether to apply a
discount factor to the care plan amount (and what to set it at). Rules regarding how the cash can
be used and how much, if any, can be taken as cash must be established.
AREAS FOR FUTURE RESEARCH
While the Cash and Counseling demonstration has yielded important insight into many of
the above issues concerning consumer direction, a number of key questions remain to be studied.
These questions include both questions about the effects of consumer direction on consumers,
caregivers, and costs, and questions about implementation. Among the questions about program
effects are the following:
• Does consumer direction improve quality only if the consumer hires a family
member?
• Would consumer direction work as well for Medicare home health care recipients?
For individuals with substance abuse or mental health problems? What are the key
factors that will influence whether the success of the program for consumers of PCS
and HCBS can be replicated in these other settings?
• How long do the effects of consumer direction last? Do they increase or decrease
over time?
• Are sizable reductions in other Medicaid costs for adults, especially long term care
costs, realized only if a consumer can purchase more services? That is, is consumer
control over how the funds are spent enough to reduce other Medicaid long term care
costs, or does the average benefit level have to be higher than it would have been
under the traditional model?
• Why do some eligible beneficiaries never receive services through traditional PCS or
HCBS waivers? That is, how much is participation in the PCS or HCBS program
likely to increase solely because a consumer direction option is offered?
• Does competition from consumer direction force agencies to do a better job than they
are currently doing in order to retain consumers? What level of penetration into the
market does consumer direction have to have before agencies make such a response?
• What are the mechanisms that lead to fewer reported unmet needs and greater
satisfaction despite total hours of care being lower under consumer direction? How
much is due to greater efficiency of the human assistance versus better quality versus
substitution of equipment or supplies for human assistance?
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 61 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
The operational issues about which more information is needed include the following:
• How much and what types of contact between consumers and counselors is needed to
ensure consumers’ safety from exploitation and inadequate care?
• How important is the counselor’s role in hired workers’ satisfaction? (For example,
is the finding that consumers in Cash and Counseling paid their workers slightly
higher wages than agency workers in two states, whereas in IHSS consumers paid
directly hired workers markedly less than agency workers, due to counselors’
influence?)
• How can states ensure that consumers who want to self-direct get to do so? Arkansas
was able to get a high proportion of program enrollees started with a cash allowance
by requiring counselors to get a spending plan established within 45 days after a
consumer enrolled. However, what, if anything, did the state sacrifice in terms of
control over the acceptability of consumers’ spending plans?
• What effect would worker registries have on the number of consumers likely to
participate in a consumer directed option, and would they be worth the cost? That is,
is the real value of consumer direction to consumers solely in being able to hire
someone they already know and trust?
• Will unions representing personal care workers insist on a role in consumer directed
initiatives?
Consumer direction is growing rapidly, providing a potential laboratory over the next few
years for addressing these and other important research, policy, and operational issues. The
challenge now that the Cash and Counseling demonstration has ended will be to find ways to
develop and exploit the potential knowledge base from the new programs so that consumers,
caregivers, and states can reap the sizeable benefits from consumer direction.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 62 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
REFERENCES
Benjamin, A.E. “Consumer-Directed Services at Home: A New Model for Persons with
Disabilities.” Health Affairs, vol. 20. no. 6, 2001, pp. 80-95.
Benjamin, A.E., and Ruth E. Matthias. Age, Consumer Direction, and Outcomes of Supportive
Services at Home." The Gerontologist, vol. 41, no. 5, 2001, pp. 632-642.
Benjamin, A.E., and Ruth E. Matthias. "Work-Life Differences and Outcomes for Agency and
Consumer-Directed Home Care Workers." The Gerontologist, vol. 44, 2004, pp. 479-488.
Carlson, Barbara, Stacy Dale, Leslie Foster, Randall Brown, Barbara Phillips, and Jennifer
Schore. “Effect of Consumer Direction on Adults’ Personal Care and Well-Being in
Arkansas, New Jersey, and Florida.” Princeton, NJ: Mathematica Policy Research, Inc.,
2005.
Dale, Stacy, Randall Brown, Barbara Phillips, Jennifer Schore, and Barbara Lepidus Carlson.
“The Effects Of Cash And Counseling On Personal Care Services And Medicaid Costs In
Arkansas.” Health Affairs Web Exclusive, 10.1377/hlthaff.w3.566, November 19, 2003.
Dale, Stacy, Randall Brown, and Barbara Phillips. “Medicaid Costs Under Consumer Direction
for Florida Children with Developmental Disabilities.” Princeton, NJ: Mathematica Policy
Research, Inc., 2004.
Dale, Stacy, and Randall Brown. “The Effect of Cash and Counseling on Medicaid and
Medicare Costs: Findings for Adults in Three States.” Princeton, NJ: Mathematica Policy
Research, Inc., 2005.
Dale, Stacy, Randall Brown, Barbara Phillips, and Barbara Carlson. “Experiences of Workers
Hired Under Cash and Counseling: Findings from Arkansas, Florida, and New Jersey.”
Princeton, NJ: Mathematica Policy Research, Inc., 2005.
Flannagan, Susan. “An Inventory of Consumer-Directed Support Service Programs.” Overview
of Key Program Characteristics. Presentation at the Cash and Counseling annual meeting,
Arlington, VA, 2001.
Foster, Leslie, Randall Brown, Barbara Phillips, Jennifer Schore, and Barbara Lepidus Carlson.
“Improving The Quality Of Medicaid Personal Assistance Through Consumer Direction”
Health Affairs Web Exclusive, 10.1377/hlthaff.w3.162, March 26, 2003.
Foster, Leslie, Stacy Dale, Randall Brown, Barbara Phillips, Jennifer Schore, and Barbara
Carlson. “Do Consumer-Directed Medicaid Support Services Work for Children with
Developmental Disabilities.” Princeton, NJ: Mathematica Policy Research, Inc., 2004.
Foster, Leslie, Randall Brown, and Rachel Shapiro. “Assessing the Appeal of the Cash and
Counseling Demonstration in Arkansas, Florida, and New Jersey.” Princeton, NJ:
Mathematica Policy Research, Inc., 2005a.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 63 of 64
Consumer-Directed Care and Its Implications for State and Federal Long-Term Care Policy
Randall Brown, Ph.D.
Foster, Leslie, Randall Brown, Barbara Phillips, and Barbara Carlson. “The Effect of Cash and
Counseling on the Primary Informal Caregivers of Children with Developmental
Disabilities.” Princeton, NJ: Mathematica Policy Research, Inc., 2005b.
Foster, Leslie, Barbara Phillips, and Jennifer Schore. “Consumer and Consultant Experiences in
the Florida Consumer Directed Care Program.” Princeton, NJ: Mathematica Policy
Research, Inc., 2005c.
Foster, Leslie, Barbara Phillips, and Jennifer Schore. “Consumer and Consultant Experiences in
the New Jersey Personal Preference Program.” Draft report, Princeton, NJ: Mathematica
Policy Research, Inc., 2005d.
Foster, Leslie, Randall Brown, Barbara Phillips, and Barbara Lepidus Carlson. “How Cash and
Counseling Affects Informal Caregivers: Findings from Three States.” Draft report,
Princeton, NJ: Mathematica Policy Research, Inc., 2005e.
Infield, Donna Lind. “States’ Experiencing Implementing Consumer-Directed Home and
Community Services: Results of the 2004 Survey of State Administrators, Opinion Survey,
and Telephone Interviews.” National Council on Aging and NASUA, 2004.
Phillips, Barbara, Kevin Mahoney, Lori Simon-Rusinowitz, Sandra Barrett, William Ditto,
Thomas Reimers, and Pamela Doty. Lessons from the Implementation of Cash and
Counseling in Arkansas, Florida and New Jersey. Princeton, NJ: Mathematica Policy
Research, Inc., 2003.
Phillips, Barbara, and Barbara Schneider.
“Moving to IndependentChoices:
The
Implementation of the Cash & Counseling Demonstration in Arkansas.” Princeton, NJ:
Mathematica Policy Research, Inc., 2002.
Phillips, Barbara, and Barbara Schneider. “Enabling Personal Preference: The Implementation
of the Cash & Counseling Demonstration in New Jersey.” Princeton, NJ: Mathematica
Policy Research, Inc., 2003.
Phillips, Barbara, and Barbara Schneider. “Changing to Consumer-Directed Care: The
Implementation of the Cash and Counseling Demonstration in Florida.” Princeton, NJ:
Mathematica Policy Research, Inc., 2004.
Schore, Jennifer, and Barbara Phillips. “Consumer and Counselor Experiences in the Arkansas
IndependentChoices Program.” Princeton, NJ: Mathematica Policy Research, Inc., 2004.
DRAFT: DO NOT CITE OR DISTRIBUTE
Page 64 of 64