How Cash and Counseling Affects Informal Caregivers: Findings from Arkansas, Florida and New Jersey

07/01/2005

U.S. Department of Health and Human Services

How Cash and Counseling Affects Informal Caregivers: Findings from Arkansas, Florida, and New Jersey

Leslie Foster, Randall Brown, Barbara Phillips and Barbara Lepidus Carlson

Mathematica Policy Research, Inc.

July 2005

PDF Version


This report was prepared under contract #HHS-100-95-0046 between the U.S. Department of Health and Human Services (HHS), Office of Disability, Aging and Long-Term Care Policy (DALTCP) and the University of Maryland. For additional information about the study, you may visit the DALTCP home page at http://aspe.hhs.gov/_/office_specific/daltcp.cfm or contact the ASPE Project Officer, Pamela Doty, at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, SW, Washington, DC 20201. Her e-mail address is: Pamela.Doty@hhs.gov.


TABLE OF CONTENTS

ACKNOWLEDGMENTS

EXECUTIVE SUMMARY

INTRODUCTION

A NEW MODEL OF MEDICAID SUPPORTIVE SERVICES

GOALS AND FEATURES OF THE DEMONSTRATION

Covered Services

Target Populations and Eligibility

Enrollment and Random Assignment

Calculation of Program Allowances

Permitted Uses of the Allowances

Counseling and Fiscal Services

CONSUMER DIRECTION AND INFORMAL CAREGIVERS

Previous Research

Hypotheses About Caregiver Outcomes

METHODS

Data Collection and Samples

Estimation of Program Effects

Baseline Characteristics of Care Recipients and Primary Informal Caregivers

RESULTS

Receipt and Use of the Allowance in the Treatment Group

Estimated Program Effects

Primary Informal Caregivers Who Became Paid Workers

Caregiver Effects, by Care Recipients' Age Group

DISCUSSION

Summary and Interpretation

Limitations

Implementation Lessons and Policy Implications

REFERENCES

APPENDIX A: ADDITIONAL INFORMATION ON METHODS

APPENDIX B: ADDITIONAL RESULTS

NOTES

LIST OF TABLES

TABLE 1: Key Features of Cash and Counseling Programs, by State

TABLE 2: Baseline Characteristics of Care Recipients, by State

TABLE 3: Characteristics of Primary Informal Caregivers, by State

TABLE 4: Receipt and Use of the Allowance Among Care Recipients in the Treatment Group, by State

TABLE 5: Estimated Effects on Frequency, Amounts, and Timing of Assistance Provided by Primary Informal Caregivers, by State

TABLE 6: Estimated Effects on the Quality of Relationships Between Primary Informal Caregivers and Care Recipients, by State

TABLE 7: Estimated Effects on Primary Informal Caregivers' Satisfaction with Care Quality, by State

TABLE 8: Estimated Effects on Primary Informal Caregivers' Emotional Well- Being, by State

TABLE 9: Estimated Effects on Primary Informal Caregivers' Job Performance and Financial Well-Being, by State

TABLE 10: Estimated Effects on Primary Informal Caregivers' Physical Well- Being and Health, by State

TABLE 11: Estimated Effects on Primary Informal Caregivers' Satisfaction with Life, by State

TABLE A.1: Outcome Measures for Primary Informal Caregivers, by Type

TABLE A.2a: Baseline Characteristics of Care Recipients and Primary Informal Caregivers, by Evaluation Status: Arkansas

TABLE A.2b: Baseline Characteristics of Care Recipients and Primary Informal Caregivers, by Evaluation Status: Florida

TABLE A.2c: Baseline Characteristics of Care Recipients and Primary Informal Caregivers, by Evaluation Status: New Jersey

TABLE A.3: Minimum Detectable Effects

TABLE A.4: Examples of Adjustments of Hours of Care Provided

TABLE B.1. Estimated Effects on Living Arrangements for, and Assistance Provided by, Primary Informal Caregivers, by State

TABLE B.2. Estimated Effects on Primary Informal Caregivers' Knowledge and Preparedness, by State

TABLE B.3. Estimated Effects on Primary Informal Caregivers' Physical Functioning, by State

TABLE B.4. Estimated Effects of Care Recipient and Caregiver Characteristics on Whether Primary Informal Caregivers Became Paid Workers, by State

TABLE B.5. Reasons Treatment Group Caregivers Did Not Become Paid Workers

TABLE B.6. Estimated Effects of Becoming a Paid Worker on Key Caregiver Outcomes, by State

TABLE B.7a. Estimated Effects on Key Caregiver Outcomes, by Care Recipients' Age Group: Arkansas

TABLE B.7b. Estimated Effects on Key Caregiver Outcomes, by Care Recipients' Age Group: Florida

TABLE B.7c. Estimated Effects on Key Caregiver Outcomes, by Care Recipients' Age Group: New Jersey

TABLE B.8. Estimated Effects of the Arkansas Program on Caregiver Satisfaction and Well-Being for Subgroups Defined by Whether Care Recipients Participated in ElderChoices During Follow-Up

ACKNOWLEDGMENTS

This report has benefited greatly from the thoughtful comments and suggestions of several people. In particular, we appreciate input from external reviewers A.E. (Ted) Benjamin (UCLA Center for Health Policy Research), Rosalie Kane (University of Minnesota School of Public Health), and Robyn Stone (Institute for the Future of Aging Services). Several members of the Cash and Counseling Demonstration and Evaluation management team--Kevin Mahoney, Marie Squillace, and members of the staff of the Centers for Medicare & Medicaid Services (CMS)--provided useful comments. We also appreciate comments from the Florida Consumer Directed Care Plus program in the Department of Elder Affairs.

Several colleagues at Mathematica Policy Research, Inc. made the report possible. Nora Paxton and Amy Zambrowski programmed the analysis, and Valerie Cheh provided comments on an earlier draft. Patricia Ciaccio edited the report, and Jill Miller produced it.

The opinions presented here are those of the authors and do not necessarily reflect those of the funders (the Robert Wood Johnson Foundation and the U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation), the Cash and Counseling National Program Office, CMS, or the demonstration states.

EXECUTIVE SUMMARY

Cash and Counseling Is a Promising Way to Deliver Medicaid Supportive Services.

Adult Medicaid beneficiaries who have disabilities and live at home rely mostly on unpaid family members and other informal caregivers for personal care. States supplement beneficiaries' informal assistance with disability-related supportive services. These are usually delivered through a Medicaid state plan as personal care services (PCS) or through a Medicaid waiver program as home and community-based services (HCBS). If beneficiaries find their services unsatisfactory or too inflexible to meet their needs, the burden to compensate for those shortcomings often falls on informal caregivers, potentially causing them emotional, physical, and financial strain.

The Cash and Counseling Demonstration and Evaluation, implemented in Arkansas, Florida, and New Jersey, gave eligible Medicaid beneficiaries the opportunity to receive a monthly allowance to purchase supportive services of their choosing. By helping beneficiaries avoid the potential drawbacks of traditional PCS and HCBS, Cash and Counseling programs could also be expected to improve the well-being of beneficiaries' informal caregivers.

This report estimates the effects of Cash and Counseling on the caregivers who were providing the most unpaid assistance to adult beneficiaries when those beneficiaries volunteered to participate in the demonstration and completed a baseline interview. The caregivers in this analysis--identified by beneficiaries as their primary informal caregiver at baseline--were interviewed about 10 months later.

A Rigorous Design Provided Definitive Evidence of Program Effects on Caregivers.

The demonstration and evaluation used a randomized design. After a completing their baseline telephone interview, participating beneficiaries were randomly assigned to direct their own Medicaid supportive services as Cash and Counseling consumers (the treatment group) or to rely on PCS or HCBS as usual (the control group). The primary informal caregivers of treatment group consumers could be affected by Cash and Counseling if those consumers: (1) hired the primary informal caregiver as a worker or named him or her as their representative decision maker; (2) adjusted the amount, timing, and types of services used; or (3) purchased assistive devices or home modifications.

Data on caregiver outcomes were collected through telephone interviews. These were conducted between February 2000 and May 2003, depending on the state, and were completed by 1,433 caregivers in Arkansas, 1,193 in Florida, and 1,042 in New Jersey. Caregivers, who typically were related to their care recipients, were asked factual questions about the frequency, amounts, timing, and types of assistance they provided and about their labor force participation and income. They were asked for their opinions on the quality of their relationships with care recipients; their satisfaction with care recipients' care; and their own emotional, financial, and physical well-being, and satisfaction with life. Program effects were estimated by comparing outcomes for the caregivers of treatment group members with those for the caregivers of control group members. Regression models controlled for the baseline characteristics of care recipients and the demographic characteristics of caregivers.

Treatment Group Caregivers Reported Greater Well-Being in All Three States.

Despite variations in design and implementation across states, all three demonstration programs positively affected the well-being of caregivers. On average, the caregivers of treatment group members were less likely than their control group counterparts to report high levels of physical and financial strain. Treatment group caregivers worried less about insufficient care and safety and were more likely to be very satisfied with their care recipient's overall care arrangements. They were also less likely to say caregiving impinged on their privacy, social lives, and job performance. On average, treatment group caregivers were much more likely than their control group counterparts to be very satisfied with their own lives.

For some outcomes, the estimated effects were proportionally largest, relative to control group means, in Arkansas. Arkansas also had the largest proportion of treatment group members receiving the Cash and Counseling allowance when we interviewed caregivers.

As expected, some treatment group caregivers were paid for caregiving during the follow-up period (56 percent in Arkansas, 29 percent in Florida, and 42 percent in New Jersey). To explore whether Cash and Counseling affected paid and unpaid treatment group caregivers differently, we estimated program effects separately for each of these subsets. The results suggested that getting paid for caregiving was not the sole determinant of improved well-being but that it did contribute to magnitude of many treatment-control differences.

An analysis of program effects by care recipients' age group suggested that Cash and Counseling can positively affect the well-being of caregivers for elderly and nonelderly adults (as evidenced in Arkansas and Florida). The same analysis suggested that benefits to well-being may diminish if caregivers provided more hours of assistance under Cash and Counseling than they would have provided otherwise (as evidenced in New Jersey among caregivers for nonelderly care recipients).

The Implications for Medicaid Policy Are Encouraging.

The three-state findings were robust and encouraging. Viewed with earlier evaluation results, they illustrate that when Medicaid beneficiaries wish to direct their own supportive services and do so, both they and their primary informal caregivers benefit. Improvements seemed to come about because some informal caregivers became paid workers and some care recipients made service arrangements that lightened their caregiver's burden. The three demonstration states continue to operate their Cash and Counseling programs under Section 1115 authority of the Social Security Act (without random assignment). Other states have good reason to consider offering consumer-directed options as part of their Medicaid systems.

INTRODUCTION

Medicaid beneficiaries who have disabilities and live at home receive much more personal care from unpaid family members and other informal caregivers than they do from paid sources (U.S. Department of Health and Human Services 2002). Providing informal care often is rewarding, but it also can be emotionally, physically, and financially taxing. If the strain of caregiving becomes debilitating, it could imperil the health of caregivers and hasten Medicaid beneficiaries' entry into nursing homes or other institutions, thereby increasing public costs (Doty 1997).

States help beneficiaries live at home by supplementing their informal care with disability-related supportive services. These are usually delivered through a Medicaid state plan as personal care services (PCS) or through a Medicaid waiver program as home- and community-based services (HCBS). However, states often limit the amounts, types, and providers of the PCS or HCBS they cover (Doty 2004). If beneficiaries find the services unsatisfactory or too inflexible to meet their needs, the burden to compensate for those shortcomings often falls on informal caregivers.

The Cash and Counseling Demonstration and Evaluation, implemented in Arkansas, Florida, and New Jersey, gave eligible Medicaid beneficiaries the opportunity to receive a monthly allowance to purchase disability-related supportive services of their choosing. By increasing beneficiaries' autonomy and control over their care, Cash and Counseling programs are meant to help them avoid the potential drawbacks of PCS or HCBS from personal care agencies and other providers. In turn, the programs could also be expected to improve the well-being of beneficiaries' informal caregivers.

This report describes the effects of Cash and Counseling on the caregivers who were providing the most unpaid assistance to adult beneficiaries when those beneficiaries volunteered to participate in the demonstration and completed a baseline telephone interview. The caregivers in this analysis--identified by beneficiaries as their primary informal caregiver at baseline--were interviewed about 10 months later. This report follows earlier findings that beneficiaries who were randomly assigned to participate in a Cash and Counseling program (the treatment group) were more satisfied with their supportive services than were beneficiaries who were assigned to rely on PCS or HCBS as usual (the control group) (Carlson et al. 2005).

The report begins with a brief overview of the Cash and Counseling model and its implementation by the demonstration states. It sets forth research hypotheses about caregiver outcomes and briefly describes the methods used to estimate program effects. (Readers familiar with the demonstration and evaluation may wish to skip these introductory sections.) The report then presents findings for each state and concludes with a discussion of findings, limitations, implementation lessons, and policy implications. Appendices contain additional detail about analytic methods and tables of results that are noted, but not shown, in the body of the report.

A NEW MODEL OF MEDICAID SUPPORTIVE SERVICES

About 1.4 million Medicaid beneficiaries receive disability-related supportive services in their homes (Harrington and Kitchener 2003). Most receive PCS or HCBS from personal care agencies and other state-certified providers, but states are increasingly allowing beneficiaries to direct some aspects of their care, as service "consumers" (O'Brien and Elias 2004). In 1999, an estimated 139 publicly funded consumer-directed programs served adults or children with physical or developmental disabilities (Flanagan 2001).

Cash and Counseling, which is a fairly expansive model of consumer-directed care, provides a monthly allowance that consumers may use to hire workers, including relatives, and to purchase other services and goods related to their need for personal care (within state guidelines). Cash and Counseling allows consumers to designate a representative, such as a relative or friend, to help them make decisions about managing their care. It also offers counseling and fiscal services to help consumers and representatives handle their program responsibilities. These tenets of Cash and Counseling--a flexible allowance, use of representatives, and availability of counseling and fiscal services--are meant to make consumer direction adaptable to consumers of all ages and abilities.

As noted, Cash and Counseling was tested in Arkansas, Florida, and New Jersey as a three-state demonstration. The Robert Wood Johnson Foundation (RWJF) and the Office of the Assistant Secretary for Planning and Evaluation of the U.S. Department of Health and Human Services funded the demonstration. The Centers for Medicare & Medicaid Services approved the demonstration programs under Section 1115 authority of the Social Security Act. The National Program Office for the demonstration, at Boston College and the University of Maryland, coordinated the overall demonstration, provided technical assistance to the states, and oversaw the evaluation. Mathematica Policy Research, Inc. (MPR) is the demonstration evaluator.

GOALS AND FEATURES OF THE DEMONSTRATION

With the principles of consumer autonomy, choice, and control foremost in mind, each demonstration state had the practical goal of assessing the political and economic feasibility of publicly funded consumer-directed care. In addition, Arkansas (more so than Florida and New Jersey) hoped to increase access to services in parts of the state where agency workers were in short supply. All three states had to meet federal budget neutrality requirements over the life of the demonstration, but none had the goal of saving public funds.

Because the Medicaid programs and political environments of the demonstration states differed considerably, the states were not required to implement a standardized Cash and Counseling program. They did have to adhere to the model's basic tenets, however. The key features of each state's program are described below and summarized in Table 1.1

Covered Services

The demonstration programs in Arkansas and New Jersey offered an allowance instead of the supportive services, such as help with eating, bathing, housekeeping, and shopping, that beneficiaries otherwise would have received as Medicaid state plan PCS. Florida's program offered an allowance instead of the benefits usually provided through an HCBS waiver program, such as in-home nursing, professional therapies, care-related supplies and equipment, caregiver respite, and help with personal care.

Target Populations and Eligibility

Each state's target population included adult Medicaid beneficiaries with primarily physical disabilities. In addition, Florida targeted beneficiaries with primarily developmental disabilities.

In Arkansas the demonstration was open to adult Medicaid beneficiaries who were eligible for, but not necessarily receiving, state plan PCS. Beneficiaries who were participating in either of two HCBS waiver programs--ElderChoices or Alternatives--could also participate in the Arkansas demonstration. (Waiver benefits were delivered as usual during the demonstration; they were not "cashed out" as part of the Cash and Counseling allowance. ElderChoices provides nurse-supervised homemaker, chore, and respite service to nursing home-qualified elderly adults. Alternatives provides attendant care and environmental modifications for nonelderly adults and lets them choose and supervise paid caregivers.)

TABLE 1: Key Features of Cash and Counseling Programs, by State
  Arkansas Florida New Jersey
Program Name IndependentChoices Consumer Directed Care Personal Preference
Evaluation Intake Period December 1998-April 2001 June 2000-July 2002 (adults); June 2000-August 2001 (children) November 1999-July 2002
Eligible Medicaid Adults 18 or older with primarily physical disabilities who were eligible for Medicaid state plan PCS Adults with primarily physical disabilities, and children and adults with primarily developmental disabilities, who were receiving HCBS Adults 18 or older with primarily physical disabilities who were receiving or had been assessed for Medicaid state plan PCS
Benefits Included in the Calculation of the Program Allowance Medicaid state plan PCS Medicaid HCBS, except for case management/support coordination services Medicaid state plan PCS
Hiring Restrictions Could not hire legally responsible relatives (that is, spouses), and representative decision makers could not also be paid workers. None during the evaluation period. If a representative was also a paid worker, someone else from the care recipient's "circle of support" was asked to verify the completion of agreed upon work. Representative decision makers could not also be paid workers.
Method for Calculating the Program Allowance $8 per hour in care plan multiplied by a provider-specific adjustment factor. (Adjustment factors were used to help the state meet federal budget neutrality requirements.) Value of recent claims or care plan multiplied by an adjustment factor. (Care plans were used for beneficiaries with developmental disabilities, if claims history was not stable, or care plan value exceeded recent claims by $50 or more per month.) Value of care plan minus 10 percent set-aside for fiscal agent and counseling services.
Median Monthly Allowance at Baseline $313 $829 (adults) and $831 (children) $1,097
Participation in Other Consumer-Directed or Home Care Programs Demonstration enrollees could also participate in the HCBS waiver programs ElderChoices and Alternatives.a For adults with developmental disabilities, the demonstration excluded several northern counties with a state-funded consumer-directed program. Demonstration enrollees could not participate in HCBS waiver programs or a state-funded consumer-directed program.
SOURCE: Program records and in-person and telephone discussions with state officials and demonstration program staff members. Discussions were conducted about 18 months after each state began random assignment (see Phillips and Schneider, 2002, 2003, and 2004).
  1. 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. Sixty-one percent of elderly demonstration enrollees participated in ElderChoices at baseline, and 3 percent of the nonelderly participated in Alternatives.

In Florida the demonstration was open to Medicaid beneficiaries who were receiving HCBS under the state's Developmental Disabilities (DD) Waiver or Aged/Disabled Adult (ADA) Waiver and living in selected areas of the state.2 Together, these waivers serve adults and children with developmental disabilities, frail elderly adults, and adults with physical disabilities.3 For adults with developmental disabilities, the demonstration catchment area was the entire state except several northern counties where a state-funded consumer-directed program was being piloted. For elderly adults and those with physical disabilities, the catchment area consisted of 19 counties, including most of the state's major metropolitan areas.

In New Jersey the demonstration was open to adult Medicaid beneficiaries who: (1) were using PCS or had been assessed as eligible for it, (2) were not also participating in an HCBS waiver program or state-funded consumer-directed program, and (3) were expected to require PCS for at least 6 months. Recipients of both PCS and HCBS were excluded because authorization procedures for those services differed and beneficiaries would have received assistance from Cash and Counseling consultants and HCBS case managers, which the state program office feared would cause confusion. Beneficiaries who were expected to require PCS for less than 6 months were excluded because Cash and Counseling consumers were expected to require several months to develop and implement a plan for spending the monthly allowance.

Except for applying these eligibility criteria, the three demonstration states relied on beneficiaries and their families to decide whether they wanted to take on the responsibilities of consumer direction. Beneficiaries could disenroll from consumer direction at any time.

Enrollment and Random Assignment

The demonstration states were responsible for outreach and enrollment activities, including the collection of informed consent and basic intake data (such as contact information). In general, states used a combination of direct mailings, telephone calls, and home visits to inform eligible beneficiaries about the opportunity to participate in the demonstration. Within a week of each beneficiary's enrollment, MPR conducted a baseline telephone interview with the beneficiary or a knowledgeable proxy respondent, and then randomly assigned the beneficiary to direct their own supportive services as Cash and Counseling consumers (the treatment group) or to rely on PCS or HCBS as usual (the control group).

Enrollment and random assignment periods differed by state, depending on when each was ready to conduct outreach and enrollment activities and implement its consumer-directed program. Arkansas enrolled 2,008 beneficiaries between December 1998 and April 2001; Florida enrolled 1,816 adult beneficiaries between June 2000 and July 2002; and New Jersey enrolled 1,755 beneficiaries between November 1999 and July 2002. Half the enrollees in each state were randomly assigned to the treatment group.

Calculation of Program Allowances

Arkansas and New Jersey based program allowances on the number of hours in consumers' Medicaid PCS plans and the average rates the state would otherwise pay for agency services. Plan hours were capped at 16 per week in Arkansas and 25 per week in New Jersey absent special authorization for more hours. Florida based allowances on all the benefits in consumers' HCBS care plans or recent Medicaid waiver claims, except those for case management/support coordination.4 To meet federal budget neutrality standards, Arkansas and Florida applied adjustment factors to consumers' allowances. The states had determined that beneficiaries in the predemonstration period had not, on average, received all the services authorized in their plans (for example, because in-home services were suspended during hospitalizations or agency workers failed to arrive as scheduled). In contrast, New Jersey determined that consumers' actual and planned costs had been roughly equal historically, indicating that adjustment factors were not necessary to keep the demonstration budget neutral. Median monthly allowances calculated for adult demonstration participants varied considerably by state. They were $313 in Arkansas, $829 in Florida, and $1,097 in New Jersey.

Permitted Uses of the Allowances

Consumers (or their representatives) in all three programs were required to develop a written spending plan that specified the goods and services they wished to purchase with the allowance. States permitted purchases of services and goods that were related to consumers' care needs and ability to exercise independence. For example, they might permit microwave ovens and washing machines, but not televisions. Consumers could elect to receive small portions of their allowance (10-20 percent) as cash for incidental expenses, such as taxi fare, that were not readily purchased through an invoicing process. They could also save portions of the allowance for larger, one-time purchases, such as home modifications.

Although consumers were permitted to use their allowances to hire relatives, some restrictions applied. Arkansas did not allow consumers to hire spouses, despite a federal waiver that permitted states to let consumers pay legally responsible relatives (spouses, parents of minors, and other legal guardians). Neither Arkansas nor New Jersey allowed the same person to serve as both a representative and paid worker. Florida had no such restriction during the evaluation period, because it recognized that parents typically represent and care for children with developmental disabilities. To protect consumers in cases where the representative and the worker were the same person, however, Florida required that someone else from the consumer's "circle of support" verify that the representative/worker performed the agreed upon services.5

Counseling and Fiscal Services

In all three demonstration programs, consumers were offered the assistance of counselors (called "consultants" in Florida and New Jersey) and a fiscal agent (called a "bookkeeper" in Arkansas). Counselors interacted with consumers to: (1) develop, review, and revise written plans for spending the monthly allowance in permissible ways; (2) offer advice about recruiting, hiring, and training workers; (3) monitor consumers' well-being; and (4) monitor use of the allowance. Florida and New Jersey required that state or district-level staff review all spending plans. Arkansas required such review only if a plan included goods and services not on a preapproved list; otherwise, counselor review sufficed. Interaction between counselors and consumers occurred during telephone calls and home visits, the frequency of which varied by state. Counselor services were provided at no direct charge to consumers, but the costs of these services were included in federal budget neutrality calculations.

Consumers in the three programs were offered assistance with fiscal tasks, including the payroll functions of an employer (such as preparing and submitting payroll tax returns) and check writing. Florida and New Jersey charged consumers modest fees for fiscal services, while Arkansas did not charge them directly. Although consumers who demonstrated ability to handle fiscal tasks themselves were allowed to do so--and thus receive the allowance as cash each month--all but a handful of consumers chose to have the fiscal agent maintain a program account on their behalf.

The demonstration programs prevented misuse of the allowance by comparing check requests and worker time sheets with consumers' spending plans before they disbursed funds. In Florida and New Jersey, the fiscal agent was responsible for performing this verification; in Arkansas, counselors were responsible for performing it. Arkansas and Florida required consumers to keep receipts for purchases made with the allowance, and receipts were subject to review by counselors. New Jersey did not require that consumers keep receipts.

CONSUMER DIRECTION AND INFORMAL CAREGIVERS

Previous Research

This report provides rigorous, empirical evidence on how an innovative model of paid supportive services affects the well-being of unpaid caregivers. It helps bridge an extensive literature on informal caregiving and a more nascent one on consumer-directed care.

The emotional, physical, and financial tolls of informal caregiving are well documented (Rabow et al. 2004; Zarit 2004; Schulz and Beach 1999). According to a recent national survey, for example, one-third of caregivers for elderly adults describe caregiving as emotionally stressful, 15 percent suffer physical or mental health problems as a result of caregiving, and half report that caregiving detracts from time spent with other family members (National Alliance for Caregiving and AARP 2004).

The literature also identifies a reciprocal relationship between the health status and quality of life of informal caregivers and care recipients. Caregivers of depressed elderly people report a poorer quality of life than do caregivers of elderly people who are not depressed (Sewitch et al. 2004). Caregivers of elderly people with dementia are more likely to institutionalize their care recipient if they rate their own burden as great and their own health as poor (Gaugler et al. 2003). Whether caregivers feel they provide good care and whether care recipients had problems with household tasks and/or woke the caregiver or other family members during the night have been identified as predictors of nursing home placements for elderly adults (Schur and Whitlatch 2003).

Because of the relationships between the well-being of care recipients and informal caregivers and between caregiver burden and care recipients' institutionalization, the effects of consumer direction on caregivers merit attention. Evidence is mounting that consumer direction benefits consumers (see, for example, Carlson et al. 2005 and Benjamin et al. 2000). By contrast, this and other recent studies (see, for example, Benjamin and Matthias 2004) examine whether consumer benefits translate into caregiver benefits or into caregiver strain, and identify mechanisms leading to these outcomes.

Hypotheses About Caregiver Outcomes

Under Cash and Counseling, consumers could change their use of Medicaid supportive services in many ways. They could choose their workers and designate a representative to help them manage their care. They could adjust the amount and timing of assistance they received from paid workers and informal caregivers, and they could buy assistive devices and home modifications. They could also use the program's counseling and fiscal services. These changes, in turn, could affect:

  • The frequency, amounts, timing, and types of assistance that primary informal caregivers provided.
  • The quality of relationships between caregivers and care recipients.
  • Caregivers' satisfaction with care recipients' supportive services.
  • Caregivers' emotional, physical, and financial well-being.
  • Caregivers' satisfaction with life.

The amount, types, and timing of assistance the caregivers in this analysis provided could increase or decrease under Cash and Counseling. This would depend on whether they were hired to perform additional tasks (beyond those they performed without pay) or whether other workers were hired to relieve some of their burden.

Similarly, outcomes measuring quality of life could be favorably affected if Cash and Counseling consumers replaced unsatisfactory agency workers with workers they and their caregiver liked and trusted; began paying others to perform tasks the primary informal caregiver found especially difficult, unpleasant, time-consuming, or hard to schedule; purchased assistive devices to increase their own independence and ease the caregiver's physical strain; or began paying the primary informal caregiver, thereby acknowledging the value of his or her assistance and reducing financial strain. Cash and Counseling could also improve caregivers' financial well-being by affording them enough flexibility in meeting their caregiving responsibilities to enter the labor force or change jobs.

Conversely, Cash and Counseling could make matters worse for primary informal caregivers. Assuming the responsibilities of an agency worker or other paid provider could create emotional and physical stress. Becoming the paid employee of a relative could strain familial relationships, as could continuing to provide unpaid assistance while other family members start to receive pay. Serving as a consumer's program representative could be burdensome, especially if the program's counseling or fiscal services were inadequate or if other workers were hired and did not meet the consumer's or caregiver's expectations.

METHODS6

Data Collection and Samples

Data on the types of caregiver outcomes discussed in the previous section were collected through computer-assisted telephone interviews with caregivers. (Table A.1 lists specific outcome measures.) As noted, the caregivers in this analysis were identified by beneficiaries as their primary informal caregiver during each beneficiary's baseline interview. Specifically, beneficiaries were asked for the name and telephone number of the person (if any) who had given them the most unpaid assistance during the previous week with personal care, household and community tasks, routine health care, and transportation. Interviews with caregivers were conducted 10 months after the baseline beneficiary interview, between February 2000 and June 2003, depending on the state.7 The number of caregivers completing interviews were 1,433 caregivers in Arkansas, 1,193 in Florida, and 1,042 in New Jersey. Overall response rates were 84 percent for caregivers associated with treatment group members and 78 percent for those associated with control group members. Proxy respondents were not allowed.

To preserve the benefits of random assignment and obtain a complete picture of caregivers' experiences, caregiver interviews were conducted even with caregivers whose care recipient had disenrolled from the demonstration program (25-29 percent across states) or died (3-7 percent) (not shown). Caregivers who had not helped their care recipients shortly before the caregiver interview (6-9 percent) were included in the sample. For most measures, interviewers asked these caregivers to recall the period when they were last helping care recipients. However, if recall error seemed likely (in questions measuring hours of assistance, for example), interview questions were skipped.

Estimation of Program Effects

Program effects on caregivers were measured by comparing the outcomes of all caregivers in the treatment group with those of all caregivers in the control group, regardless of whether a particular care recipient in the treatment group actually received the monthly allowance. Because some care recipients had not received the monthly allowance by the time their caregiver was interviewed (for reasons described in the Results section of this report), this "intent-to-treat" analysis understates program impacts on caregivers whose care recipients participated fully in a consumer-directed program.8

Models. For each state, the analysis used logit models to estimate program effects on categorical outcomes, an ordered logit to estimate effects on caregivers' level of household income, and ordinary least squares models to estimate effects on the frequency and amount of assistance.

Many outcome measures were constructed by converting responses to survey questions with four-point scales (for example, degree of satisfaction) into two alternative binary measures. One measure represented the most favorable rating (very satisfied), the other an unfavorable rating (somewhat or very dissatisfied).9 We used this approach so readers could easily see the basis on which we drew inferences about the key questions for each outcome: Did consumer direction increase the proportion of highly satisfied caregivers, reduce the proportion of dissatisfied ones, or have both (or neither) effects?

Except for treatment-control differences in the amount of care provided (which were estimated with ordinary least squares regression), we measured program impacts by using the estimated coefficients from the logit models to calculate the treatment-control difference in average predicted probabilities that the binary dependent variable took a value of 1. The p-values of the estimated coefficients on the treatment status variable are reported in our tables and were used to test whether treatment-control differences were significantly different from zero.

Control Variables. The models controlled for care recipient and caregiver characteristics. Most of these characteristics were measured during baseline interviews with care recipients or their proxy respondents, and a few were drawn from baseline program records or the caregiver interview. As specified by state in Table A.2a, Table A.2b and Table A.2c, the models controlled for care recipients' demographic characteristics, health and functioning, use of unpaid and paid supportive services, satisfaction with care and life, work and community activities, unmet needs, reasons for and month of enrollment, use (if any) of a proxy respondent for all or most of the baseline interview, and whether named a representative to manage the monthly allowance by the time of the baseline interview (applicable only in Arkansas and Florida).

The models also controlled for the amount or value of benefits in the care recipient's baseline Medicaid care plan (expressed as hours of authorized PCS in the Arkansas models, and as a categorical measure of the prospective weekly allowance amount in the Florida and New Jersey models). These variables were drawn from each state's program records.

The models controlled for the familial relationship between care recipients and primary informal caregivers and whether the caregiver was employed, interested in being paid for caregiving, and living with the care recipient, as reported by care recipients at baseline. Finally, the models controlled for caregiver age, sex, race, education, marital status, and whether has dependent children. These variables were measured during the caregiver interview.

Appendix Table A.2a, Table A.2b and Table A.2c present the mean baseline characteristics of the care recipients and caregivers in this analysis, by state and treatment status. Although nearly all the characteristics were distributed similarly across the treatment and control groups (as expected with random assignment), a few treatment-control differences emerged within the subset of sample members whose primary informal caregivers responded to our survey, whether by chance or differential attrition. As noted, the regression models controlled for these differences.

Statistical Power. Given the number of caregivers in each state's sample, the analysis had 80 percent power to detect treatment-control differences of 7-8 percentage points for binary variables with means of 0.5, assuming two-tailed tests at the 0.05 significance level (Appendix Table A.3). For a key continuous outcome--the hours of assistance provided by live-in caregivers during two weeks--the analysis could detect differences of 14-17 hours (about 8-11 percent of the sample means).

Baseline Characteristics of Care Recipients and Primary Informal Caregivers

The care recipients associated with the caregivers in our analysis samples differed considerably by state. In Arkansas, care recipients typically were elderly (75 percent) and female (78 percent) (Table 2). Sixty-one percent were white, 34 percent were black, and 1 percent was Hispanic (regardless of race). About four in ten Arkansas care recipients said they lived in rural areas of the state, and about one-quarter lived alone. Nearly half of Arkansas care recipients said they were in poor health compared to other people their age, and nearly half named a representative who would help them manage Cash and Counseling responsibilities. In addition, 61 percent of elderly care recipients were enrolled in the HCBS waiver program ElderChoices at baseline, and 3 percent of the nonelderly were enrolled in Alternatives (not shown).

TABLE 2: Baseline Characteristics of Care Recipients, by State(Percentages)
Characteristic Arkansas Florida New Jersey
AGE IN YEARS
   18 to 39 8.3 47.8 18.4
   40 to 64 17.2 17.1 26.7
   65 to 79 35.5 15.3 28.2
   80 or older 39.0 19.8 26.7
ELIGIBLE FOR DEMONSTRATION BECAUSE OF DEVELOPMENTAL DISABILITY n.a. 61.5 n.a.
HISPANIC 1.2 25.7 35.2
RACE
   White 60.6 73.5 55.1
   Black 34.3 22.8 35.2
   Other 5.1 3.7 9.7
FEMALE 77.5 58.8 72.6
GRADUATED FROM HIGH SCHOOLa 23.8 76.0 38.4
LIVES ALONE 23.3 9.0 21.0
AREA OF RESIDENCE
   Rural 40.6 15.3 12.2
   Nonrural but high-crime or with poor public transportation 27.6 39.9 40.8
IN POOR HEALTH RELATIVE TO PEERS 48.1 23.3 41.6
NEEDED HELP USING TOILET IN PAST WEEK 70.2 68.7 75.9
NUMBER OF INFORMAL CAREGIVERS IN PAST WEEK
   1 32.1 26.5 29.7
   2 31.0 26.7 26.7
   3 or more 36.9 46.8 43.7
RECEIVING PUBLICLY FUNDED HOME CARE (at All or for at Least 6 Months)b 73.4 66.8 44.6
DISSATISFIED WITH OVERALL CARE ARRANGEMENTSc 18.6 18.0 27.7
APPOINTED A REPRESENTATIVE 47.0 85.8 NA
NUMBER OF RESPONDENTS 1,433 1,193 1,042
SOURCE: Program records from each of the demonstration states and MPR's baseline interview. Interviews were conducted in Arkansas between December 1998 and April 2001; in Florida between June 2000 and July 2002; and in New Jersey between November 1999 and July 2002.
  1. For Arkansas and New Jersey, the percentages reflect the care recipient's education. For Florida, the percentage reflects the education of the person (that is, the care recipient or a representative if the representative responded to the baseline interview) who would make care-related decisions in the demonstration program.
  2. For Arkansas, the percentage represents care recipients who were receiving publicly funded home care at baseline, regardless of how long they had been receiving it. For Florida and New Jersey, the percentages represent care recipients who had been receiving publicly funded home care for 6 months or longer at baseline. The measures differ because care recipients in Florida and New Jersey typically received traditional home care services before they enrolled in the demonstration; in Arkansas, this was not necessarily the case.
  3. Calculations of the percent dissatisfied excludes beneficiaries who said they used no paid services or goods during the week before the baseline interview (see Appendix Table A.2a, Table A.2b and Table A.2c).

n.a. = not applicable; NA = data not available.

In Florida, where the demonstration was targeted in part to adults with developmental disabilities, a larger proportion of care recipients were nonelderly than elderly (65 percent versus 35 percent) (Table 2). Most care recipients were White (74 percent) or Black (23 percent), and one-quarter were Hispanic (regardless of race). Relatively few Florida care recipients (15 percent) said they lived in a rural area, but 40 percent said their neighborhood had high-crime or poor public transportation. Because a large proportion of Florida care recipients had developmental disabilities, more than eight in ten said they would use a representative in the consumer-directed program, and few (9 percent) lived alone. About one-quarter said their health was poor.

Care recipients in New Jersey were, as a group, more demographically diverse than those in Arkansas and Florida (Table 2). About 55 percent were elderly. Fifty-five percent were White, 35 percent were Black, 10 percent reported another race, and 35 percent were Hispanic. In addition, 21 percent of New Jersey care recipients lived alone, and 41 percent said their neighborhood had high-crime or poor public transportation. Slightly more than four in ten care recipients said their health was poor.10 New Jersey care recipients were somewhat more likely than their counterparts in the other states to be dissatisfied with their overall care arrangements at baseline (28 percent in New Jersey, 19 percent in Arkansas, and 18 percent in Florida).

The caregivers in our analysis samples had many characteristics in common, regardless of state. Roughly eight in ten caregivers were female, a majority had graduated from high school, and between 52 percent and 58 percent were married (Table 3). Many caregivers were employed at baseline (38-49 percent), and many were interested in getting paid for caregiving (31-38 percent). The most notable difference across states was that Florida caregivers were more likely to be the care recipient's parent (48 percent versus 7 percent in Arkansas, and 18 percent in New Jersey). Thus, Florida caregivers were also somewhat older, on average, than their counterparts in the other states. (Twenty-two percent of Florida caregivers were 65 or older, compared with 14 percent of Arkansas caregivers and 17 percent of New Jersey caregivers.)

TABLE 3: Characteristics of Primary Informal Caregivers, by State(Percentages)
Characteristic Arkansas Florida New Jersey
AGE IN YEARS
   39 or younger 22.4 8.1 19.7
   40 to 64 64.1 70.4 63.5
   65 or older 13.5 21.5 16.8
HISPANIC 1.2 24.8 35.7
WHITE 61.0 70.7 45.4
FEMALE 86.0 83.9 79.4
GRADUATED FROM HIGH SCHOOL 69.0 81.1 70.7
MARRIED 55.3 57.6 51.7
HAS CHILD(REN) YOUNGER THAN 18 29.3 18.6 29.9
RELATIONSHIP TO CARE RECIPIENT
   Spouse 5.5 6.1 8.3
   Parent 7.0 48.3 17.9
   Daughter or son 62.3 26.3 49.9
   Other relative 17.7 12.6 15.4
   Nonrelative 7.4 6.7 8.4
LIVED WITH CARE RECIPIENT AT BASELINE 61.7 83.1 65.0
EMPLOYED AT BASELINEa 37.7 45.9 49.0
INTERESTED IN BEING PAID FOR CAREGIVINGa 35.8 31.1 38.2
NUMBER OF RESPONDENTS 1,433 1,193 1,042
SOURCE: MPR's caregiver and baseline interviews. Caregiver interviews were conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. Baseline interviews were conducted in Arkansas between December 1998 and April 2001; in Florida between June 2000 and July 2002; and in New Jersey between November 1999 and July 2002. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline.
  1. As reported by care recipients during the baseline interview.

RESULTS

Receipt and Use of the Allowance in the Treatment Group

As noted earlier, all beneficiaries randomly assigned to the Cash and Counseling treatment group could receive a monthly allowance if they developed an acceptable spending plan. In fact, the proportions of treatment group members who had received the program allowance for at least 1 month by the time caregivers were interviewed for this analysis (about 10 months after beneficiaries' random assignment) varied considerably by state. The proportions were 88 percent in Arkansas, 55 percent in Florida, and 75 percent in New Jersey (Table 4). The variations stemmed, in part, from the complexity of states' allowance planning procedures, the number of people involved in allowance planning, and the degree to which program counselors were responsible for getting consumers started on the allowance.

Furthermore, although this analysis is of people who were providing unpaid care at baseline, some treatment group caregivers were paid for caregiving during the follow-up period. This was true for 56 percent of all treatment group caregivers in Arkansas, 29 percent of treatment group caregivers in Florida, and 42 percent of treatment group caregivers in New Jersey (Table 4). Among caregivers who became paid, those in Arkansas were paid for about 13 hours of care per week and earned $6 an hour; those in Florida and New Jersey were paid for about 20 hours of care per week and earned about $10 an hour.

TABLE 4: Receipt and Use of the Allowance Among Care Recipients in the Treatment Group, by State
  Arkansas Florida New Jersey
PERCENTAGE OF CARE RECIPIENTS WHO RECEIVED A PROGRAM ALLOWANCEa 87.7 55.4 74.7
PERCENTAGE OF CAREGIVERS PAID FOR HELPING CARE RECIPIENTb 56.4 28.5 41.8
PERCENTAGE OF CAREGIVERS PAID FOR HELPING CARE RECIPIENT, IF CARE RECIPIENT RECEIVED A PROGRAM ALLOWANCE 62.6 45.9 55.6
   Number of paid hours (if any) provided during 2-week reference period (mean)c 12.5 19.6 20.6
   Hourly wage (mean)c $6.07 $10.26 $9.84
NUMBER OF CARE RECIPIENTS/CAREGIVERS IN THE TREATMENT GROUP 721 617 546
SOURCE: Program records from each of the demonstration states and MPR's caregiver interview. Interviews were conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003.
  1. Measured 10 months after care recipients were randomly assigned to the treatment group and, thus, about the same time as caregivers were interviewed.
  2. The percentage of caregivers who were paid for any of the help they provided since care recipients' random assignment, according to caregiver self-reports.
  3. Paid hours and hourly wages were measured for a subset of caregivers who were care recipients' primary paid workers during the follow-up period. The mean paid hours provided by all caregivers is probably somewhat lower.

Estimated Program Effects

As noted, program effects were estimated over all responding caregivers of treatment and control group members, whether or not treatment group members received the allowance or hired caregivers. Because payment of caregivers was a program effect, however, it and its influence on other outcomes are important to measure. Thus, following the presentation of overall program effects, this report discusses the characteristics associated with becoming paid, describes caregivers' self-reported reasons for remaining unpaid, and examines the extent to which outcomes differed by whether caregivers became paid workers.

Frequency, Amounts, and Timing of Assistance.11 Regardless of treatment status, the caregivers in this analysis devoted substantial amounts of time to activities that benefited care recipients exclusively (such as help eating and bathing) and to activities that also may have benefited others in the household (such as preparing meals and doing laundry). Program effects on the amount of assistance provided by caregivers varied by state. The frequency of assistance was affected only in New Jersey, and the timing of assistance was affected only in Arkansas.

Caregivers in Arkansas provided assistance on 12 days during the two-week period they were asked about, regardless of treatment status (Table 5). During that time, treatment group caregivers provided about 107 hours of assistance and control group caregivers provided about 117 hours. The mean difference equaled one less hour of assistance per day (-10.4 hours/14.0 days = -0.7 hours) and was significant at the 0.10 level. This overall impact was driven by a 13-hour treatment-control difference in the amount of care provided by live-in caregivers (57 percent of the Arkansas sample), including a 9 hour difference in the time live-in caregivers spent on activities that benefited the entire household. In addition, Arkansas treatment group caregivers were slightly less likely than their control group counterparts to provide care after 6:00 P.M. on weekdays.

Florida caregivers provided assistance on nearly 13 days during the 2-week period they were asked about, regardless of treatment status (Table 5). The treatment-control difference in the hours reported by all caregivers was not statistically significant. However, among caregivers who lived with their care recipient (83 percent of the Florida sample), those in the treatment group provided 140 hours of assistance during the 2-week reference period, compared with 149 hours of assistance provided by control group caregivers. The mean difference of -0.7 hours per day was significant at the 0.10 level. The treatment-control difference was not significant for visiting caregivers.

TABLE 5: Estimated Effects on Frequency, Amounts, and Timing of Assistance Provided by Primary Informal Caregivers, by State
Outcome Arkansas Florida New Jersey
Predicted Means EstimatedEffect(p-Value) Predicted Means EstimatedEffect(p-Value) Predicted Means EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
Frequency of Assistance Provided in Recent Two Weeksa
NUMBER OF DAYS HELPED CARE RECIPIENT 12.4 12.2 0.2(.371) 12.7 12.7 0.0(.844) 12.2 11.8 0.5**(.036)
Time Spent Meeting the Needs of Care Recipient and Other Household Members in Recent Two Weeksa
TOTAL HOURS OF ASSISTANCEb 106.6 117.0 -10.4*(.089) 123.7 132.7 -9.0(.111) 123.2 113.3 9.9*(.057)
   Among live-in caregivers 140.1 153.0 -12.9**(.035) 139.5 149.3 -9.8*(.069) 148.4 140.1 8.3(.279)
   --   Hours that benefited care recipient onlyc 64.8 68.2 -3.5(.364) 72.0 77.7 -5.7(.126) 77.0 74.7 2.3(.652)
   --   Hours that benefited entire householdd 75.4 84.8 -9.4**(.012) 67.6 71.6 -4.1(.190) 71.4 65.4 6.0(.195)
   Among visiting caregiverse 61.9 68.9 -7.0(.164) 48.9 54.0 -5.1(.676) 84.8 72.5 12.3*(.090)
Timing of Assistance Provided in Recent Two Weeksa (Percents)
ASSISTED CARE RECIPIENT
   Before 8:00 A.M. weekdays 59.6 62.5 -3.0(.264) 70.0 73.7 -3.8(.121) 64.5 63.9 0.6(.830)
   After 6:00 P.M. weekdays 84.0 87.5 -3.5*(.081) 91.3 92.3 -0.9(.533) 89.3 90.2 -0.1(.650)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with ordinary-least-squares or logit models.
  1. The most recent 2 weeks during the two months before the interview in which the care recipient lived in the home or community.
  2. The estimates for total hours are constructed as weighted averages of the estimates for caregivers who are live-in and those who are visiting, with the weights being the proportion of all caregivers (treatment and control groups combined) who were live-in or visiting at follow-up. In Arkansas, these proportions were 0.571 (live-in) and 0.429 (visiting); in Florida, they were 0.826 and 0.174; and in New Jersey, they were 0.603 and 0.397. The variance of the estimated effect used to construct the t-statistic and corresponding p-value is are the variances of the estimated coefficients on the treatment status indicator from the regressions on live-in and visiting caregivers, respectively. This approach ensures that the impact on total hours is a weighted average of the impacts on live-in and visiting caregivers.
  3. Includes routine health care, personal care, and transportation.
  4. Includes preparing meals and doing housework, laundry, shopping, and yard work to meet the needs of all household members.
  5. Includes time spent helping away from the care recipient's home, such as shopping for him or her or preparing meals to take to him or her. Excludes purely social visits and time traveling, doing paperwork, or not actively helping.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

In New Jersey, treatment group caregivers reported providing assistance on 12.2 days, somewhat more than the 11.8 days reported by control group caregivers (Table 5). During the two-week reference period, treatment group caregivers reported providing 10 more hours of assistance overall. Visiting caregivers (40 percent of the New Jersey sample) drove this difference. Visiting caregivers in the treatment group reported providing about 85 hours of assistance during the 2-week reference period, compared with 73 hours reported by the control group (a difference of 0.9 hours per day).

Quality of Relationships with Care Recipients. All three Cash and Counseling programs seemed to positively affect some aspects of the relationships between caregivers and care recipients (Table 6). Caregivers were asked how well they got along with care recipients, whether the relationship had improved since baseline, whether they consulted care recipients with questions about their personal care, and whether care recipients refused to cooperate when the caregiver offered help. Statistically significant effects were seen for three of the four measures in Florida, and for two of the four measures in Arkansas and New Jersey. All the treatment-control differences were modest (ranging from 4 to 8 percentage points). In no instances did treatment group caregivers report worse outcomes than control group caregivers.

TABLE 6: Estimated Effects on the Quality of Relationships Between Primary Informal Caregivers and Care Recipients, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
Caregiver and Care Recipient Get Along Very Wella 87.5 86.1 1.4(.467) 88.7 84.9 3.8**(.047) 84.1 79.5 4.6*(.057)
Relationship Is Better Now than at Enrollment 27.9 26.7 1.2(.619) 24.7 20.2 4.5*(.062) 30.8 25.6 5.2*(.064)
Caregiver Consults Care Recipient About Personal Care Questionsa,b 21.9 13.7 8.2***(.000) 9.7 12.3 -2.6(.224) 23.0 23.0 0.0(.993)
Care Recipient Refuses to Cooperate When Caregiver Tries to Helpa 28.6 34.5 -5.9**(.019) 28.5 34.0 -5.5**(.039) 28.1 29.6 -1.6(.574)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit models.
  1. Measured "at present" or "when you were last helping."
  2. Measured only for caregivers who provided assistance with personal care, such as eating, bathing, and using the toilet.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

Satisfaction with Quality of Care.12 In all three states, treatment group caregivers were substantially more satisfied with care recipients' overall care arrangements and were less worried about insufficient care, safety, and theft when they were not with the care recipient (although many still worried) (Table 7).

Program effects on the proportions of caregivers that were very satisfied with care recipients' overall care arrangements were large in all states (Table 7). Specifically, the proportions of very satisfied treatment group caregivers were about 60 percent larger than the control group proportions of 30 percent in Florida and 32 percent in New Jersey, and 42 percent larger than the control group proportion of 43 percent in Arkansas. The proportions of dissatisfied treatment group caregivers were smaller than the control group proportions by similarly large amounts.

In all states, smaller proportions of treatment group caregivers than of control group caregivers said they worried quite a lot that care recipients would not get enough care in their absence, that care recipients' safety was at risk, or that someone would take care recipients' money or other belongings (Table 7). As a percentage of the control group means, differences for these measures were generally somewhat larger in Arkansas than they were in the other states. For example, in Arkansas, the proportion of treatment group caregivers who said they worried quite a lot about their care recipient's safety was 26 percent smaller than the control group proportion of 53 percent, while the treatment group proportions in Florida and New Jersey were about 17 percent smaller than the control group proportions of 52 percent and 65 percent, respectively.

TABLE 7: Estimated Effects on Primary Informal Caregivers' Satisfaction with Care Quality, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
HOW SATISFIED WITH CARE RECIPIENT'S OVERALL CARE ARRANGEMENTS
   Very satisfied 60.8 42.7 18.1***(.000) 47.9 29.8 18.1***(.000) 51.6 31.7 19.9***(.000)
   Dissatisfied 9.1 22.8 -13.7***(.000) 15.7 27.5 -11.8***(.000) 13.3 32.2 -18.8***(.000)
WHEN NOT WITH CARE RECIPIENT, HOW OFTEN WORRIES THAT:a
Care Recipient Does Not Have Enough Help
   Quite a lot 35.8 53.5 -17.6***(.000) 47.8 60.5 -12.7***(.000) 52.2 70.2 -18.0***(.000)
   Rarely or not at all 35.6 20.2 15.4***(.000) 27.3 13.4 13.8***(.000) 23.9 12.7 11.2***(.000)
Care Recipient's Safety Is at Risk
   Quite a lot 39.3 53.4 -14.1***(.000) 43.1 52.3 -9.2***(.001) 53.5 64.8 -11.3***(.000)
   Rarely or not at all 31.8 21.9 9.9***(.000) 30.6 18.7 11.9***(.000) 23.4 13.6 9.8***(.000)
Someone Will Take Care Recipient's Money or Other Belongings
   Quite a lot 14.0 20.3 -6.3***(.001) 22.2 29.2 -7.0***(.005) 24.9 30.3 -5.5**(.041)
   Rarely or not at all 69.4 62.7 6.8***(.007) 61.5 49.9 11.6***(.000) 62.0 53.4 8.7***(.004)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit models.
  1. Measured "at present" or "when you were last helping."

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

A somewhat different pattern emerged for estimated effects on the proportion of caregivers that worried only rarely or not at all about sufficient care, safety, and theft. As a percentage of the control group means, impacts were somewhat smaller in Arkansas than they were in Florida or New Jersey, but they were still sizable and significant (Table 7).

Emotional Well-Being. Cash and Counseling seemed to favorably affect caregivers' emotional well-being by all of five measures in Arkansas and New Jersey, and by three of five measures in Florida (Table 8). First, although many caregivers in each state said care recipients required their almost constant attention, the treatment group proportions were 8-9 percent smaller than the control group proportions, which ranged from 57 percent in Arkansas to roughly 70 percent in Florida and New Jersey. Second, the proportions of treatment group caregivers who said caregiving impinged on their privacy were 27 percent smaller than the control group proportions in Arkansas, 19 percent smaller than in New Jersey, and only 8 percent smaller than in Florida, with slightly more than half the control group caregivers in each state reporting that caregiving affected their privacy. Third, treatment group caregivers were less likely to say caregiving limited their free time or social lives in each state, though as a percentage of the control group mean the effects were twice as large in Arkansas as they were elsewhere.

Finally, the proportions of treatment group caregivers saying caregiving caused them a great deal of emotional strain were substantially smaller than the control group proportions in Arkansas and New Jersey (Table 8). Likewise, the treatment group proportions reporting little or no emotional strain were substantially larger than the control group proportions in both states. In Florida, treatment and control group caregivers reported similar levels of emotional strain.

TABLE 8: Estimated Effects on Primary Informal Caregivers' Emotional Well-Being, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
CARE RECIPIENT REQUIRES ALMOST CONSTANT ATTENTION FROM CAREGIVER 52.6 57.2 -4.6*(0.54) 63.7 68.9 -5.2**(.028) 64.0 70.5 -6.5**(.013)
CAREGIVING LIMITS:
   Privacy 38.7 52.7 -14.1***(.000) 52.3 57.1 -4.8*(.084) 41.1 50.5 -9.4***(.001)
   Free time or social life 52.5 63.8 -11.3***(.000) 66.9 73.3 -6.5***(.008) 54.8 60.1 -5.3*(.061)
LEVEL OF EMOTIONAL STRAIN AS A RESULT OF CAREGIVING
   A great deal 26.8 34.3 -7.5***(.002) 35.7 38.6 -2.9(.286) 42.3 49.4 -7.1**(.017)
   Little or none 41.0 35.0 6.0**(.015) 29.4 28.1 1.3(.600) 28.3 23.1 5.2**(.049)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Outcomes were measured "at present" or "when you were last helping." Means were predicted with logit models. *Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

Job Performance and Financial Well-Being. In all three states, substantial proportions of caregivers said that caregiving adversely affected their choice and performance of jobs, but some problems were less prevalent in the treatment group (Table 9). For example, while at least 38 percent of control group caregivers in each state said they had not looked for a job, or another job, although they wanted to, since their care recipients' demonstration enrollment, the proportions of treatment group caregivers reporting this problem were 39 percent smaller than the control group proportions in Arkansas, 24 percent smaller than in New Jersey, and 16 percent smaller than in Florida. Among caregivers who were working for pay (other than for their care recipient) when interviewed, those in the treatment group were less likely than those in the control group to say they missed work or arrived late because of caregiving. Again, the estimated program effects were proportionally largest in Arkansas and smallest in Florida. Also in Arkansas, the proportion of treatment group caregivers that declined a better job or a promotion because of caregiving was significantly smaller than the control group proportion of 28 percent.

Although Cash and Counseling did not discernibly affect caregivers' household income, it seemed to alleviate their perception of financial strain in all three states (Table 9). In Arkansas, the proportion of treatment group caregivers reporting a great deal of financial strain as a result of caregiving was nearly 40 percent smaller than the control group proportion, while in Florida and New Jersey the treatment group proportions were about one-fourth smaller than the control group proportions. Similarly, treatment group caregivers were substantially more likely than control group caregivers to report little or no financial strain in each state.

TABLE 9: Estimated Effects on Primary Informal Caregivers' Job Performance and Financing Well-Being, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
Job Performance Since Care Recipient's Enrollment
WORKED FOR PAY, OTHER THAN FOR CARE RECIPIENT 48.4 49.2 -0.8(.683) 50.6 49.7 0.9(.613) 54.2 57.0 -2.9(.171)
DID NOT LOOK FOR A JOB OR ANOTHER JOB BUT WANTED TO 23.5 38.6 -15.1***(.000) 35.1 41.8 -6.7**(.011) 33.9 44.1 -10.3***(.000)
AMONG THOSE WHO WORKED FOR PAY, OTHER THAN FOR CARE RECIPIENT, CAREGIVING CAUSED THEM TO:
   Miss work or arrive late 48.6 60.6 -12.0***(.001) 60.9 67.1 -6.2*(.095) 53.6 65.8 -12.2***(.002)
   Decline a better job or promotion 21.5 27.8 -6.3**(.046) 35.3 35.4 -0.1(.975) 31.8 32.3 -0.4(.906)
   Quit job or reduce hours 29.2 31.5 -2.2(.511) 29.3 34.0 -4.7(.179) 34.3 37.1 -2.8(.464)
Financial Well-Being
LEVEL OF FINANCIAL STRAIN AS A RESULT OF CAREGIVINGa
   A great deal 22.4 35.7 -13.3***(.000) 29.9 38.9 -9.0***(.001) 30.0 38.6 -8.6***(.001)
   Little or None 48.3 34.9 13.4***(.000) 36.2 25.3 10.9***(.000) 39.8 31.5 8.4***(.005)
HOUSEHOLD INCOME LAST MONTH
   $1,000 or less 40.7 42.1 -1.4(.504) 24.8 25.0 -0.2(.903) 37.4 40.4 -3.0(.175)
   $1,001 to $2,000 41.5 41.1 0.4(.504) 36.5 36.5 0.0(.903) 31.0 30.7 0.3(.175)
   $2,001 to $3,000 10.6 10.2 0.5(.504) 20.1 20.0 0.1(.903) 15.6 14.7 0.9(.175)
   $3,001 or more 7.1 6.6 0.4(.504) 18.6 18.4 0.2(.903) 16.0 14.1 1.8(.175)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit or ordered-logit models.
  1. Measured "at present" or "when you were last helping."

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

Physical Well-Being and Health.13 The physical burden so often associated with informal caregiving was markedly less prevalent under Cash and Counseling (Table 10). Across states, the proportions of treatment group caregivers reporting that caregiving caused them a great deal of physical strain were smaller than those of control group caregivers by about 24-28 percent, with control group proportions ranging from 32-42 percent. Smaller proportions of treatment group caregivers reported that their physical health suffered as a result of caregiving, and smaller proportions said their own health was fair or poor (as opposed to good or excellent) compared to the health of other people their age. For example, while 34-45 percent of control group caregivers in each state said their health had suffered, the proportions of treatment group caregivers saying the same were 23-31 percent smaller.

TABLE 10: Estimated Effects on Primary Informal Caregivers' Physical Well-Being and Health, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
LEVEL OF PHYSICAL STRAIN AS A RESULT OF CAREGIVINGa
   A great deal 23.0 32.0 -9.0***(.000) 28.4 38.8 -10.4***(.000) 31.7 41.8 -10.1***(.000)
   Little or none 37.6 30.6 7.0***(.004) 30.3 25.1 5.3**(.034) 33.3 24.3 9.0***(.001)
PHYSICAL HEALTH HAS SUFFERED AS A RESULT OF CAREGIVINGa 23.6 34.3 -10.7***(.000) 32.7 44.9 -12.2***(.000) 30.7 40.3 -9.6***(.001)
CURRENT HEALTH IS FAIR OR POOR RELATIVE TO THAT OF PEERS 35.5 46.7 -11.2***(.000) 31.8 39.6 -7.8***(.004) 30.3 42.3 -12.0***(.000)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit models.
  1. Measured "at present" or "when you were last helping."

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

TABLE 11: Estimated Effects on Primary Informal Caregivers' Satisfaction with Life, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
CURRENT SATISFACTION WITH LIFE
   Very satisfied 51.3 39.9 11.4***(.000) 47.0 35.2 11.8***(.000) 51.6 37.5 4.1***(.000)
   Dissatisfied 13.1 23.2 -10.1***(.000) 16.7 22.8 -6.1***(.008) 15.2 27.3 -12.2***(.000)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit models. *Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

Satisfaction with Life. In all three states, treatment group caregivers were much more likely than control group caregivers to be very satisfied with their own lives and much less likely to be dissatisfied (Table 11). While 35-39 percent of control group caregivers in each state were very satisfied with their lives, the proportions of very satisfied treatment group caregivers were roughly one-third larger in each state. Moreover, while about a quarter of control group caregivers in each state was dissatisfied with life, the proportions of dissatisfied treatment group caregivers were smaller by nearly half in Arkansas and New Jersey, and by one-fourth in Florida.

Primary Informal Caregivers Who Became Paid Workers

Explanatory Variables. Under Cash and Counseling, the treatment group's caregivers continued to provide many hours of assistance to care recipients, but only some were paid. We used logistic regression to estimate the odds that caregivers became paid workers as a function of their observable characteristics and those of care recipients.14 (The dependent variable was a binary self-reported measure of whether the caregiver was paid for helping the care recipient at any time since enrollment.)

Across states, a few explanatory variables were consistently associated with whether primary informal caregivers became paid workers. These were: (1) the sex of caregivers, (2) whether care recipients used a proxy respondent for the baseline interview, and (3) whether care recipients considered it very important to be able to pay family or friends for caregiving when they enrolled in the demonstration (Appendix Table B.4). All else equal, female caregivers were more than twice as likely as male caregivers to become paid in all three states. Care recipients who--because of physical or cognitive impairment--used a proxy respondent for the baseline interview were less likely than others to pay their primary informal caregiver (odds ratios ranged from 0.49 to 0.84). These caregivers may have preferred that the allowance be used to pay others to perform some stressful tasks, rather than be paid themselves.

No other strong patterns of association emerged across states. In Arkansas, caregivers' income level was most strongly associated with becoming paid (after being female). Primary informal caregivers who had monthly household incomes greater than $1,000 were less likely than other caregivers to become paid workers, presumably because they had less need of additional income. In Florida, if a representative managed or helped manage the care recipient's program responsibilities, the primary informal caregiver was more than twice as likely to become paid, relative to there being no representative. Representatives may have advocated paying the primary informal caregiver when developing the allowance spending plan. Indeed, the representative may have been the primary informal caregiver and advocated self-payment. In New Jersey, care recipients who considered it very important, at baseline, to be able to pay family members or friends for caregiving were more than three times as likely as other care recipients to hire their primary informal caregiver. (Appendix Table B.4 gives the estimated odds ratios for all variables in the models.)

Reasons for Remaining Unpaid. Treatment group caregivers who were not paid for caregiving at the time of their interview were asked why not. (The survey question was open-ended, but interviewers recorded responses with the aid of a precoded list.) No single reason predominated in any state; however, caregivers in all three states most commonly said they helped out of love, devotion, or family tradition (cited by 26-28 percent of caregivers) (Appendix Table B.5). Across states, 11-16 percent of caregivers said they were not able to perform all the tasks care recipients required, lived too far from them, or had other obligations. Others (8-11 percent) said the allowance was not large enough to pay them and meet the other needs of the care recipient. Less than 5 percent of caregivers in each state said they were not paid because their care recipient had not yet received the program allowance. Finally, substantial proportions of caregivers said they did not know they could be paid--9 percent in Arkansas, 14 percent in Florida, and 15 percent New Jersey.

Estimated Effects of Payment. To explore whether Cash and Counseling affected paid and unpaid treatment group caregivers differently, we estimated separate programs effects for each of these subsets. For 15 key measures of caregiver well-being and satisfaction, similar patterns emerged in all three states: (1) paid and unpaid treatment group caregivers both fared significantly better than control group caregivers for at least half the outcomes; and (2) in most such instances, the estimated effects were substantially larger for paid caregivers than they were for unpaid caregivers. Thus, while the results suggest that getting paid for caregiving was not the sole factor that led treatment group caregivers to report better outcomes than control group caregivers, getting paid did seem to increase the magnitude of most treatment-control differences. We discuss specific findings below; however, differences between paid (or unpaid) caregivers in the treatment group and all caregivers in the control group must be interpreted with caution. Estimated effects may be driven more by unobserved differences between paid and unpaid caregivers in the treatment group than they were by payment. Such "self-selection bias" could arise, for example, if caregivers who became paid had had more responsibility, on average, for arranging care recipients' personal care than caregivers who remained unpaid. Caregivers accustomed to responsibility and control may have benefited most from Cash and Counseling.

In all three states, both subsets of treatment group caregivers (paid and unpaid) fared better than control group caregivers with respect to satisfaction with overall care arrangements, worries about insufficient care, pursuit of desired jobs, and level of financial strain (Appendix Table B.6). In Arkansas, both paid and unpaid treatment group caregivers were also less likely than control group caregivers to report limitations on privacy and free time. In Florida and New Jersey, both paid and unpaid treatment group caregivers had greater satisfaction with life than control group caregivers. In all three states, the estimated program effects on whether caregivers did not pursue desired jobs were similar for paid and unpaid treatment group caregivers, relative to control group caregivers.

For other key satisfaction and well-being outcomes, paid treatment group caregivers had significantly better outcomes than control group caregivers, but unpaid treatment group caregivers and control group caregivers had similar ones (Appendix Table B.6). For example, in all three states, only paid treatment group caregivers fared better than control group caregivers with respect to the important measures of emotional strain and physical health. In addition, in Arkansas, only paid treatment group caregivers fared better than control group caregivers with respect to a high level of satisfaction with life and whether care recipients cooperated when caregivers tried to help. In Florida and New Jersey, only paid treatment group caregivers were less likely than control group caregivers to report limitations on privacy and free time.

We also examined whether estimated program effects on the amount of care provided may have stemmed from caregiver payment. The results differed by state. In Arkansas, live-in treatment group caregivers provided fewer hours of care than their control group counterparts, whether or not they became paid workers. Among visiting caregivers, however, those in the treatment group provided fewer hours of assistance than did those in the control group only if they remained unpaid. In Florida, unpaid treatment group caregivers provided substantially less care (about 17 fewer hours per 2 weeks) than control group caregivers. The mean number of hours for paid treatment group caregivers exceeded that of the control group by about 10 hours, but the difference was not statistically significant. In contrast, in New Jersey, no program effects were seen on the amount of care provided by unpaid treatment group caregivers, but those who became paid provided about 21 more hours of care per 2 weeks, relative to the control group, and they helped more often. This was true for both live-in and visiting caregivers.

Caregiver Effects, by Care Recipients' Age Group

Because providing personal assistance to an elderly care recipient may be very different from caring for a nonelderly one, we conducted a subgroup analysis to assess whether key caregiver outcomes differed by care recipients' age group. Because some of the subgroups were quite small, only very large differences within and between groups were statistically significant; however, the size and direction of the differences revealed interesting patterns.

First, regardless of care recipients' age group, treatment group caregivers in all three states were much more likely than their control group counterparts to be very satisfied with their care recipient's overall care arrangements (Appendix Table B.7a, Table B.7b and Table B.7c). For key measures of caregivers' emotional, physical, and financial well-being, however, the results differed by state. In Arkansas, treatment group caregivers fared better than their control group counterparts regardless of care recipients' age group. In New Jersey, program effects were large and positive for treatment group caregivers who assisted elderly care recipients, but they were smaller and not significant for treatment group caregivers who assisted nonelderly adults. In Florida, treatment group caregivers who assisted elderly adults reported less physical strain than their control group counterparts, while treatment group caregivers who assisted nonelderly adults reported less physical and financial strain. Cash and Counseling did not affect emotional well-being in either Florida subgroup.

The subgroup results for Arkansas and New Jersey suggest that the amount of assistance caregivers provided may have been related to their levels of emotional, physical, and financial strain. In Arkansas, treatment group caregivers in both subgroups provided fewer (or similar) hours of assistance to care recipients as control group caregivers provided (Appendix Table B.7a). As noted, treatment group caregivers in both subgroups were less likely than their control group counterparts to report a great deal of emotional, physical, and financial strain, on average. In contrast, in New Jersey, treatment group caregivers for nonelderly adults provided substantially more hours of assistance than their control group counterparts, and their levels of emotional, physical, and financial strain were similar to those of control group caregivers (Appendix Table B.7c). (Although the estimated effects favored the treatment group, they were not statistically significant.) Within the subgroup of caregivers for elderly adults in New Jersey, treatment and control group caregivers provided about the same hours of assistance, and treatment group caregivers reported less strain.

The Florida results for physical and financial strain follow a similar pattern. In both age groups, live-in treatment group caregivers (who comprised 83 percent of the analysis sample) provided similar or fewer hours of assistance and were less likely to report a great deal of at least one type of strain (Appendix Table B.7b). The relationships were not entirely consistent across subgroups, however. For example, while the emotional strain for caregivers of nonelderly care recipients was not significantly different from zero, it was close in magnitude to, and not significantly different from, the estimate for caregivers for elders, despite the large difference in the effects on hours of care.

Overall, the estimates did follow a general pattern that is consistent with expectations, given the burden that caregivers face. If caregivers provide more hours of assistance to care recipients under consumer direction than they would provide otherwise, their emotional, physical, and financial well-being may not improve. In contrast, if caregivers provide approximately the same amount of care or less than they would otherwise, they may be more likely to benefit emotionally, physically, and financially.

DISCUSSION

Summary and Interpretation

The demonstration programs implemented in Arkansas, Florida, and New Jersey all tested the Cash and Counseling model of delivering Medicaid supportive services. The programs had the goal of improving consumer satisfaction and quality of life and were expected, in turn, to improve the well-being of consumers' primary informal caregivers.

Regardless of treatment status, the caregivers who responded to our interview devoted a great deal of time to caregiving and reported considerable emotional, physical, and financial strain, on average. Cash and Counseling did not eradicate caregiver strain, but it appeared to alleviate it. Compared with the informal caregivers in the randomly assigned control group, those in the treatment group had better outcomes with respect to:

  • Feeling very satisfied with the care recipient's overall care arrangements.
  • Worrying about insufficient care, safety, and theft.
  • Having privacy and free time.
  • Getting along with the care recipient.
  • Experiencing financial strain as a result of caregiving.
  • Being able to pursue desired jobs and getting to work on time.
  • Experiencing physical strain, or physical health problems, as a result of caregiving.
  • Feeling very satisfied with their own lives.

Of course, results were not entirely uniform. Although positive program effects were seen for these key outcomes in all three states, the size of the effects relative to the control group means often differed. For some outcomes, the estimated effects were proportionally largest in Arkansas, where care recipients were more likely to receive an allowance (87 percent) and the proportion of primary informal caregivers hired by care recipients was largest (56 percent). In addition, the direction of estimated program effects on the amounts of assistance provided by caregivers differed by state. Although Cash and Counseling seemed to somewhat reduce the amount of assistance caregivers provided in Arkansas and Florida, it seemed to increase the amount provided by the caregivers of nonelderly care recipients in New Jersey, perhaps diminishing gains for some other outcomes. Finally, the Arkansas and New Jersey programs seemed to have large, positive effects on the level of emotional strain reported by caregivers overall. In contrast, there were no apparent effects on emotional strain in Florida, where slightly more than half the care recipients were receiving the allowance, and only about one-fourth hired their primary informal caregiver.

Several mechanisms probably contributed to the positive effects we observed. For example, reductions in caregivers' financial strain may have stemmed from being hired as workers, from having more flexibility to do other paid work, or from consumers using their program allowance to buy care-related items that caregivers might have paid for otherwise. Caregivers may have been more likely to get to work on time because consumers hired workers who were more reliable than those sent by agencies. Likewise, caregivers may have worried less about safety and theft because consumers hired workers whom they, and their caregivers, knew personally and trusted. Apparent reductions in caregivers' physical strain may have resulted from consumers hiring other workers to perform the physically demanding tasks (especially if caregivers themselves were frail) or from consumers' using the program allowance to buy assistive devices to help them transfer or increase mobility. For example, a companion report for the evaluation found that nonelderly treatment group members in Arkansas were more likely than control group members to have modified their homes or purchased assistive equipment (Carlson et al. 2005).

Our assessment of the effects of payment on caregiver outcomes suggests that primary informal caregivers who became paid workers derived substantial benefit from their change in status, even though they were paid for only a fraction (12-17 percent) of the hours they worked. Most notably, paid treatment group caregivers in all three states were about 20 percentage points more likely than control group caregivers to be very satisfied with their own lives at the time they were interviewed. Although estimated program effects on caregiver satisfaction and well-being often were smaller for unpaid caregivers than they were for paid caregivers, unpaid caregivers fared better than control group caregivers on many measures of satisfaction and well-being. In addition, compared with control group caregivers in Arkansas and Florida, unpaid treatment group caregivers in those states provided substantially fewer hours of assistance to care recipients, which could be seen as an important caregiver benefit.

Across states, some of the differences in the relative magnitude of overall program effects may have stemmed from differences in program implementation and design. As noted, it is not surprising to see some of the largest program effects in Arkansas, because its treatment group had the largest proportion of active care recipients--those who developed an allowance spending plan and used the allowance to meet their needs for supportive services. Moreover, the demonstration states enrolled markedly different consumer populations in terms of age and disability. If Florida's demonstration program did not affect caregivers' emotional strain, it may have been because the emotional strain experienced by the caregivers of younger adults with primarily developmental disabilities (such as Florida enrolled) was different from that experienced by the caregivers of elderly adults with primarily physical disabilities (such as Arkansas and New Jersey enrolled) in ways consumer direction could not address. For example, compared with caregivers in Arkansas and New Jersey, those in Florida may have worried more about consumers' future care arrangements if the consumers outlived them. Consumer direction, despite being attractive in the near term, might not allay the emotional strain experienced by aging parents of adult children with developmental disabilities.

Limitations

Each state's analysis was based on a randomized design and yielded estimated program effects that were quite large and consistent across many types of measures. Despite the robustness of our findings, a few caveats are warranted about study duration, possible reporting bias, and the desirability of additional data. In Arkansas, our analysis also may have been affected by consumers' participation in an HCBS waiver program.

First, given the relatively short duration of our follow-up period, we do not know whether the positive effects observed for treatment group caregivers would persist. Improvements in the satisfaction and well-being of caregivers might not last, for example, if changes in the amount of assistance caregivers provided were not sustainable, the gratification derived from getting paid for caregiving were to diminish, or consumers made short-term or unstable care arrangements (such as by hiring young relatives who later went away to school).

Second, we cannot rule out the possibility that some caregivers for treatment group members might have inflated reports on some outcomes, such as their own health status, because Cash and Counseling brought them other benefits and they wanted the program to continue. Conversely, some caregivers for control group members may have overstated their dissatisfaction or burden because they were disappointed that their care recipient was not randomly assigned to the treatment group.

Third, having data on additional caregiver characteristics would have enriched our analysis in two ways. First, having data on caregivers' baseline health status and levels of strain would have enabled us to determine how Cash and Counseling affected subgroups of caregivers defined by those characteristics. Second, if the models used to estimate payment effects had controlled for such variables, we would be more confident in concluding that observed differences in the outcomes between paid and unpaid caregivers actually resulted from their payment status and not from unobserved differences between the groups.

As noted, the estimated program effects for Arkansas must be considered in light of the fact that 45 percent of the caregivers in the analysis were helping care recipients who participated in ElderChoices, an HCBS waiver program, during the evaluation followup. Because ElderChoices provides nurse supervision, the treatment group caregivers in this analysis may have felt more secure about their care recipients' foray into consumer direction than they would have otherwise. If this were true, our analysis would overstate the favorable effects of Cash and Counseling on caregivers. To the contrary, our tests of this hypothesis showed that, for all but three key outcomes, the estimated effects of Cash and Counseling were significantly larger for treatment group caregivers whose care recipients did not participate in ElderChoices than they were for treatment group caregivers whose care recipients did participate (Appendix Table B.8). Perhaps as long as workers from the ElderChoices program--as opposed to workers consumers had chosen and hired themselves--were still visiting the homes of treatment group consumers, their caregivers were more likely to worry about safety and theft.15 All else equal, if Arkansas's Cash and Counseling program were replicated in states without programs like ElderChoices, even larger caregiver impacts might be expected.

Implementation Lessons and Policy Implications

Implementation Lessons. The findings from this three-state analysis suggest two lessons for state program administrators. Clearly, the sooner interested consumers begin receiving the allowance and implementing decisions about their care, the sooner their caregivers benefit. Thus, one lesson is that programs should give consumers whatever assistance they might need to develop their allowance spending plans, possibly holding program counselors responsible for starting most consumers on the allowance within a certain time frame. Arkansas contractually obligated its fiscal/counseling agency to start consumers on the allowance within 45 days of random assignment and developed a database program that reminded counselors about consumers who were not yet allowance recipients. Neither Florida nor New Jersey held counselors responsible in this way. Second, because some caregivers benefited from becoming paid workers, programs should give consumers and caregivers clear and accurate information about their hiring policies. Presumably, at least some of the caregivers in Arkansas, Florida, and New Jersey who remained unpaid during the evaluation period because they did not know they could be paid would have benefited from payment.

Policy Implications. Findings from this study may enlighten policy debates about whether consumer direction threatens care quality and whether to allow consumers to pay family members for caregiving. The demonstration states addressed concerns about care quality procedurally (for example, by mandating contact between consumers and program counselors). Although not in any official capacity, the caregivers in our sample also monitored care quality; their survey responses may be viewed as testimony to the programs' success. The caregivers in this study were less likely to report worrying about insufficient care, safety, and theft by paid workers. Moreover, treatment group caregivers, like treatment group consumers (see Carlson et al. 2005), were significantly more likely than their control group counterparts to be very satisfied with consumers' overall care arrangements.

Policymakers who are concerned about paying family caregivers may worry that the practice will erode traditional values about familial responsibility and induce demand for caregiver payment. The programs in this study had positive effects, or no effects, on the quality of relationships between caregivers and care recipients and, thus, do not add to concerns about the erosion of family values. Moreover, Cash and Counseling did not appear to induce widespread demand for caregiver payment. Although many caregivers in our samples became paid workers under Cash and Counseling, 44 percent or more in each state remained unpaid and even those who were paid continued to provide many hours of unpaid assistance.

As states and the Federal Government seek to increase Medicaid beneficiaries' ability to live at home, instead of in institutions, attention must be paid to the burden society places on informal caregivers. The expanding availability of Medicaid supportive services, made possible in recent years through the federal New Freedom Initiative and Systems Change grants and through state-funded efforts, helps the nation's growing population of informal caregivers. The Cash and Counseling Demonstration and Evaluation has shown that expanding the availability of consumer-directed supportive services could help these caregivers even more.

REFERENCES

Benjamin, A.E., and Ruth Matthias. "Work-Life Differences and Outcomes for Agency and Consumer-Directed Home-Care Workers." The Gerontologist, vol. 44, no. 4, August 2004, pp. 479-488.

Benjamin, A.E., Ruth Matthias, and Todd Franke. "Comparing Consumer-Directed and Agency Models for Providing Supportive Services at Home." Health Services Research, vol. 35, no. 1, part 2, April 2000, pp. 351-366.

Carlson, Barbara Lepidus, Stacy Dale, Leslie Foster, Barbara Phillips, Randall Brown, 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., May 2005. Available at [http://aspe.hhs.gov/daltcp/reports/adultpcw.htm].

Doty, Pamela. "Consumer-Directed Home Care: Effects on Family Caregivers." Policy Brief. San Francisco, CA: Family Caregiver Alliance, October 2004. Available at [http://www.caregiver.org]. Accessed July 1, 2005.

Doty, Pamela. "Informal Caregiver 'Burnout': Predictors and Prevention." ASPE Research Notes. Washington, DC: U.S. Department of Health and Human Services, April 22, 1997. Available at [http://aspe.hhs.gov/daltcp/reports/rn05.htm].

Flanagan, Susan. "An Inventory of Consumer-Directed Support Service Programs: Overview of Key Program Characteristics." Presented at the Annual Meeting of the Cash and Counseling Demonstration and Evaluation, Arlington, VA, October 24, 2001.

Foster, Leslie, Stacy Dale, Randall Brown, Barbara Phillips, Jennifer Schore, and Barbara Lepidus Carlson. "Do Consumer-Directed Medicaid Supportive Services Work for Children with Developmental Disabilities?" Princeton, NJ: Mathematica Policy Research, Inc., September 2004. Available at [http://aspe.hhs.gov/daltcp/reports/ddkidsMss.htm].

Foster, Leslie, Randall Brown, Barbara Phillips, and Barbara Lepidus Carlson. "The Effects of Cash and Counseling on the Primary Informal Caregivers of Children with Developmental Disabilities." Princeton, NJ: Mathematica Policy Research, Inc., April 2005. Available at [http://aspe.hhs.gov/daltcp/reports/ddkidpic.htm].

Gaugler, Joseph, Robert Kane, Rosalie Kane, Ted Clay, and Robert Newcomer. "Caregiving and Institutionalization of Cognitively Impaired Older People." The Gerontologist, vol. 43, no. 2, April 2003, pp. 219-229.

Harrington, Charlene, and Martin Kitchener. "Medicaid Long-Term Care: Changes, Innovations, and Cost Containment." San Francisco, CA: University of California, San Francisco, July 2003. Available at [http://www.ncsl.org/programs/health/harrington]. Accessed March 18, 2004.

National Alliance for Caregiving and AARP. Caregiving in the U.S. Bethesda, MD: NAC and AARP, 2004. Available at: [http://www.caregiving.org/04finalreport.pdf]. Accessed July 27, 2004.

O'Brien, Ellen, and Risa Elias. "Medicaid and Long-Term Care." Menlo Park, CA: Henry J. Kaiser Family Foundation, May 2004.

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., July 2004. Available at [http://aspe.hhs.gov/daltcp/reports/FLchange.htm].

Phillips, Barbara, and Barbara Schneider. "Enabling Personal Preference: The Implementation of the Cash and Counseling Demonstration in New Jersey." Princeton, NJ: Mathematica Policy Research, Inc., March 2003. Available at [http://aspe.hhs.gov/daltcp/reports/enablepp.htm].

Phillips, Barbara, and Barbara Schneider. "Moving to IndependentChoices: The Implementation of the Cash and Counseling Demonstration in Arkansas." Princeton, NJ: Mathematica Policy Research, Inc., June 2002. Available at [http://aspe.hhs.gov/daltcp/reports/movingic.htm].

Rabow, Michael, Joshua Hauser, and Jocelia Adams. "Supporting Family Caregivers at the End of Life." Journal of the American Medical Association, vol. 291, no. 4, January 28, 2004, pp. 483-491.

Schulz, Richard, and Scott Beach. "Caregiving as a Risk Factor for Mortality: The Caregiver Health Effects Study." Journal of the American Medical Association, vol. 282, no. 23, December 15, 1999, pp. 2215-2219.

Schur, Dorothy, and Carol Whitlatch. "Circumstances Leading to Placement: A Difficult Caregiving Decision." Lippincott's Case Management, vol. 8, no. 5, September/October 2003, pp. 187-195.

Sewitch, M.J., J. McCusker, N. Dendukuri, and M.J. Yaffe. "Depression in Frail Elders: Impact on Family Caregivers." International Journal of Geriatric Psychiatry, vol. 19, no. 7, 2004, pp. 655-665.

U.S. Department of Health and Human Services. "Delivering on the Promise: Preliminary Report on the New Freedom Initiative." Washington, DC: U.S. Department of Health and Human Services, April 2002. Available at [http://www.hhs.gov/newfreedom/prelim/cargive.html]. Accessed October 22, 2002.

Zarit, Steven H. "Family Care and Burden at the End of Life." Commentary, E-Journal of the Canadian Medical Association, vol. 170, no. 12, June 8, 2004. Available at [http://www.cmaj.ca/cgi/content/full/170/12/1811]. Accessed July 7, 2005.

APPENDIX A: ADDITIONAL INFORMATION ON METHODS

This appendix provides additional information on analytic methods. Table A.1 presents a complete list of the outcomes measured. Table A.2a, Table A.2b and Table A.2c show the distribution of control variables for the treatment and control groups by state, and Table A.3 shows the minimum program effects we could confidently detect. The approaches we used to estimate logit models, measure care hours, and impute values for some control variables are described in detail below.

Methods

Use of Logit Models

As noted in the body of the report, we measured most program impacts by using the estimated coefficients from logit models to calculate the treatment-control difference in average predicted probabilities that the binary dependent variables took a value of 1. That is, we calculated two predicted probabilities that Y = 1 (for example, whether very satisfied with life) for each primary informal caregiver in the sample--first assuming the case was in the treatment group, then assuming it was in the control group--then calculated the mean probability for these two series to get predicted treatment and control values, and the difference between these means. This approach provides a more intuitive measure of the size and importance of the impact than would the traditional odds ratio, which is obtained by exponentiating the logit coefficient on the treatment status variable.

Measuring Outcomes Derived from Scales

As noted in the body of the report, many of our outcome measures were derived from survey questions with four or five-point scales. We converted responses into two binary measures--one for the most favorable rating and one for an unfavorable rating. We then estimated impacts on each of these two measures. Although we could have measured both impacts with one multinomial logit model, such estimates would be less precise because of the relatively large number of parameters estimated. Ordered logit models are designed for such outcome measures but may mask important nonlinear patterns of impacts. Thus, after examining simple frequencies and determining that using two binary measures would not obscure important results, we proceeded with this approach.

Statistical Power

We had 80 percent power to detect impact of the sizes listed in Table A.3, assuming two-tailed tests at the .05 significance level.

Measuring Hours of Assistance Provided

Data on hours of assistance were collected for live-in and visiting caregivers who provided any help with routine health care, personal care, or household or community chores during a 2-week reference period (the last 2 weeks in the month before the caregiver interview during which the care recipient lived at home, and not in a hospital or nursing home).

Visiting caregivers were asked for the number of hours spent actively helping care recipients during the reference period, excluding purely social visits, or time spent on paperwork or travel to and from care recipients' homes.

Live-in caregivers were separately asked for: (1) the number of hours spent helping care recipients with personal care, routine health care, or transportation; and (2) the number spent on other things around the house and community. This approach was used to distinguish hours that benefited care recipients exclusively from those that may also have benefited others in the household.

For live-in caregivers, the total hours of assistance were calculated by adding these "care recipient hours" and "household hours." When total reported hours exceeded 336, the total number of hours in the 2-week reference period, the following adjustments were made, so that no case had more than 336 total:

  • If the caregiver reported the same number of hours for time spent helping the care recipient as for time spent on other household or community chores, we divided both types of hours in half and summed the halved amounts (for example, see case 1, Table A.4).
  • Otherwise, if the caregiver reported that care recipient hours equaled 336 and household hours were less than 336, we kept the household value, decreased the care recipient value by that amount, and summed those values (case 2).
  • Otherwise, if the caregiver reported different totals for care recipient hours and household hours, we kept the lesser value, decreased the greater value by that amount, and summed those values (case 3).

Imputation Procedures for Control Variables

Prior Education and Work Experience. We used an imputation procedure for a handful of measures of education and prior experience with work, hiring, and supervising others that were collected during baseline interviews because of their supposed relevance to managing a budget and hiring workers. If the interview respondent was a demonstration enrollee, the questions about these variables were asked in reference to the enrollee. If a representative was a proxy respondent for the enrollee, however, the questions were asked in reference to the representative, under the assumption that the representative would be the primary decision maker for the enrollee under Cash and Counseling. If someone other than a representative decision maker was a proxy respondent for the enrollee, the questions were asked in reference to the enrollee. To provide consistently defined variables within each of the three states, we imputed values as follows:

  • In Arkansas, the intention was to make all responses refer to the characteristics of enrollees. Thus, the responses of representative proxies were replaced with imputed values from cases in the most comparable group: that of nonrepresentative proxies for enrollees who also had a representative. In both types of cases, the enrollee had a representative and a proxy respondent. For cases requiring imputation (those in which the proxy was the representative), a random case was selected from the "donor" group who fell into the same demographic cell defined by age group, race, and sex. That donor's values for the set of variables were imputed to the case requiring imputation.
  • In Florida, the intention was to make all responses refer to the person who would be the primary decision maker under Cash and Counseling. (There were too few interviews conducted with nonrepresentative proxies for enrollees who also had representatives, and thus too few donor cases, to replicate the procedure used in Arkansas.) Thus, the responses of enrollees or nonrepresentative proxies--if the enrollee also had a representative--were replaced with imputed values from the most comparable group: that of representative proxies. In all these cases, the enrollee had a representative, and donor cases were randomly selected in the manner described above, except that the donor group was further defined by the sample member's feeder program (Adult Services, Developmental Services, or Aged and Disabled).
  • In New Jersey, no imputation was needed. New Jersey demonstration enrollees were not asked to name representatives until they had completed a baseline interview and then been randomly assigned. Thus, all responses to questions about education, work, and supervisory experience were asked in reference to the enrollee.

Missing Values for Control Variables. When care recipients or caregivers were missing data on one or two control variables, the sample mean of the missing variable(s) was imputed to keep the case in the analysis. Overall, few survey respondents refused to answer any given question or did not know the answer to it.

TABLE A.1: Outcome Measures for Primary Informal Caregivers, by Type
Living Arrangements and Types of Assistance Provided in Recent Two Weeks Whether lived with care recipient Whether lived within 10 minutes' travel time of care recipient Whether provided assistance Among those providing assistance:
  • Helped with personal care
  • Helped with household tasks
  • Helped with routine health care
  • Kept care recipient company
Knowledge, Preparedness, and Consulting Others Whether feels well informed about care recipient's condition and services Whether feels fully prepared to meet expectations in helping care recipient Whether consults care recipient with personal care questions Caregiver's Quality of Life Whether caregiving limits:
  • Privacy
  • Free time or social life

Whether care recipient requires almost constant attention from caregiver Level of emotional strain as a result of caregiving How satisfied with life in general

Financial Well-Being Level of financial strain as a result of caregiving Household income last month
Frequency, Amount, and Timing of Care Provided in Recent Two Weeks Mean number of days provided care Among live-in caregivers:
  • Mean hours of care that benefited care recipient only
  • Mean hours of care that benefited entire household

Among visiting caregivers:

  • Mean hours of care per day

Whether provided care:

  • Before 8:00 A.M. weekdays
  • After 6:00 P.M. weekdays
Caregiver-Care Recipient Relationship How well caregiver and care recipient get along Whether relationship is better or worse than, or about the same as, it was at enrollment Whether care recipient refuses to cooperate when caregiver tries to help Job Choice and Performance Since Enrollment Whether worked for pay, other than for care recipient Among those who did, whether caregiving caused them to:
  • Miss work or arrive late
  • Turn down a better job or promotion
  • Quit job or reduce hours

Whether did not look for a job, or another job, although wanted to

Physical Well-Being Whether physical health suffered as a result of caregiving Level of physical strain as a result of caregiving
Perception of Care Quality How satisfied with care recipient's overall care arrangements When caregiver is not with care recipient, how often worries about:
  • Care recipient not getting enough care
  • Care recipient's safety
  • Someone taking care recipient's money or other belongings
Health and Functioning Current health status relative to that of peers Whether illness or disability caused problems with:
  • Preparing meals, doing housework, laundry, shopping, taking medicine, or managing money
  • Eating, getting out of bed or a chair, dressing, bathing, or using the toilet
NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Outcomes were measured about 10 months after baseline.
TABLE A.2a: Baseline Characteristics of Care Recipients and Primary Informal Caregivers, by Evaluation Status: Arkansas(Percentages)
Characteristic Treatment Group Control Group
Care Recipients' Demographics
AGE IN YEARS
   18 to 39 8.9 7.8
   40 to 64 15.7 18.6
   65 to 79 36.1 35.0
   80 or older 39.3 38.7
FEMALE 77.0 78.1
HISPANICa 1.4 1.0
RACE
   White 60.1 61.0
   Black 34.8 33.9
   Other 5.1 5.2
LIVE ALONE 23.4 23.2
GRADUATED FROM HIGH SCHOOL 22.8 24.9
DESCRIBED AREA OF RESIDENCE AS:
   Rural 40.3 41.0
   Not rural but high-crime or lacking adequate public transportation 26.5 28.6
   Not rural, not high-crime, having adequate public transportation 33.2 30.4
Care Recipients' Health and Functioning
RELATIVE HEALTH STATUS
   Excellent or good 19.7 23.5
   Fair 30.3 30.3
   Poor 50.0 46.2
NOT INDEPENDENT IN PAST WEEK IN:b
   Getting in or out of bed 69.4 70.8
   Bathing 93.3 92.8
   Using toilet/diapers 71.7 68.7
Care Recipients' Use of Personal Assistance
RECEIVED ANY HELP IN PAST WEEK WITH:
   Household activitiesc 98.3 97.8
   Personal cared 91.1 91.3
   Transportatione 65.2 67.5
   Routine health caref 79.9 78.5
NUMBER OF UNPAID CAREGIVERS WHO PROVIDED HELP IN PAST WEEK
   1 33.8 30.4
   2 30.4 31.6
   3 or more 35.9 38.0
LENGTH OF TIME WITH PUBLICLY FUNDED HOME CARE
    **
   Less than 1 year 21.2 20.2
   1 to 3 years 19.6 23.7
   More than 3 years 21.1 19.9
   Respondent said no care last week, but program says current user 12.8 8.5
   Not a current recipient 25.3 27.7
NUMBER OF PAID CAREGIVERS IN PAST WEEK
   0 34.1 33.1
   1 38.2 39.2
   2 or more 27.7 27.8
NUMBER OF HOURS PER WEEK IN MEDICAID CARE PLAN
   1 to 6 24.7 22.5
   7 to 11 34.8 37.6
   12 or more 40.5 39.9
Care Recipients' Satisfaction with Care and Unmet Needs
HOW SATISFIED WITH OVERALL CARE ARRANGEMENTS
    **
   Very satisfied 42.7 40.3
   Satisfied 30.9 32.6
   Dissatisfied 14.7 18.8
   No paid services or goods in past week 11.7 8.3
NOT GETTING ENOUGH HELP WITH:
   Household activitiesc 65.0 65.4
   Personal cared 61.0 65.7*
   Transportatione 43.4 46.9
Care Recipients' Preferences About Consumer Direction
BEING ALLOWED TO PAY FAMILY MEMBERS OR FRIENDS WAS VERY IMPORTANT 88.1 86.7
HAVING A CHOICE ABOUT PAID WORKERS' SCHEDULE WAS VERY IMPORTANT 81.7 82.1
HAVING A CHOICE ABOUT TYPES OF SERVICES RECEIVED WAS VERY IMPORTANT 86.8 87.4
Care Recipients' Work Experience and Other Characteristics
EVER SUPERVISED SOMEONE 27.8 27.4
EVER HIRED SOMEONE PRIVATELY 31.2 29.8
EVER WORKED FOR PAY 83.5 83.3
PROXY COMPLETED ALL OR MOST OF BASELINE SURVEY 52.7 53.0
APPOINTED A REPRESENTATIVE AT ENROLLMENT 45.2 48.7
ENROLLMENT MONTH WAS BETWEEN:
   December 1998 and December 1999 51.0 51.7
   January 2000 and April 2001 49.0 48.3
Primary Informal Caregivers' Characteristics
AGE IN YEARS
   39 or younger 21.9 22.9
   40 to 64 66.0 62.3
   65 or older 12.1 14.8
FEMALE 85.1 87.7
RELATIONSHIP TO CARE RECIPIENT
   Spouse 5.1 6.0
   Parent 7.6 6.5
   Daughter or son 64.1 60.6
   Other relative 17.1 18.3
   Nonrelative 6.2 8.6
HISPANICa 1.4 1.0
WHITE 61.3 60.7
MARRIED 53.5 57.2
HAS CHILD(REN) YOUNGER THAN AGE 18 28.4 30.2
HIGHEST LEVEL OF EDUCATION
   8 years or less 11.1 12.4
   9 to 12 years, but no high school diploma or GED 18.1 20.5
   High school diploma or GED 39.8 40.5
   At least some college 31.0 26.7
EMPLOYEDg 36.5 38.9
EXPRESSED INTEREST IN BEING PAID FOR CAREGIVINGg 32.1 39.5***
SAMPLE SIZE 721 712
SOURCE: MPR's baseline evaluation interview, conducted between December 1998 and April 2001; caregiver interview, conducted between February 2000 and April 2002; and program records.
  1. Because this characteristic was rare, it was not controlled for in regression models.
  2. Needed hands-on or standby help or did not perform activity at all.
  3. Household activities include meal preparation, laundry, housework, and yard work.
  4. Personal care includes eating, dressing, and bathing.
  5. Transportation includes trips for medical and nonmedical reasons.
  6. Routine health care includes checking vital signs and help taking medicine or doing exercises.
  7. As reported by care recipients during the baseline interview.

*Difference between treatment and control groups significantly different from 0 at the .10 level, two-tailed test. **Difference between treatment and control groups significantly different from 0 at the .05 level, two-tailed test. ***Difference between treatment and control groups significantly different from 0 at the .01 level, two-tailed test.

TABLE A.2b: Baseline Characteristics of Care Recipients and Primary Informal Caregivers, by Evaluation Status: Florida(Percentages)
Characteristic Treatment Group Control Group
Care Recipients' Demographics
AGE IN YEARS
   18 to 39 48.1 47.4
   40 to 64 17.2 17.0
   65 to 79 15.4 15.3
   80 or older 19.3 20.3
FEMALE 57.2 60.6
HISPANIC 24.0 27.5
RACE
   White 72.8 74.2
   Black 23.9 21.6
   Other 3.2 4.2
LIVE ALONE 8.3 9.7
DESCRIBED AREA OF RESIDENCE AS:
   Rural 15.5 15.0
   Not rural but high-crime or lacking adequate public transportation 38.0 42.0
   Not rural, not high-crime, having adequate public transportation 46.5 43.0
Care Recipients' Health and Functioning
RELATIVE HEALTH STATUS
   Excellent or good 49.4 48.8
   Fair 27.1 28.2
   Poor 23.5 23.1
NOT INDEPENDENT IN PAST WEEK IN:a
   Getting in or out of bed 58.8 62.5
   Bathing 84.6 84.2
   Using toilet/diapers 68.9 68.4
Care Recipients' Use of Personal Assistance
RECEIVED ANY HELP IN PAST WEEK WITH:
   Household activitiesb 97.9 98.1
   Personal carec 85.3 84.9
   Transportationd 79.7 78.1
   Routine health caree 83.8 81.3
NUMBER OF UNPAID CAREGIVERS WHO PROVIDED HELP IN PAST WEEK
   1 26.1 27.1
   2 28.9 24.3
   3 or more 45.1 48.6
RECEIVING WAIVER SERVICES FOR 6 MONTHS OR LONGER 66.1 67.5
NUMBER OF PAID CAREGIVERS IN PAST WEEK
   0 28.3 29.5
   1 36.0 34.9
   2 or more 35.7 35.6
PROPOSED WEEKLY ALLOWANCE
   Less than $150 33.2 36.6
   $150 to $299 32.9 31.8
   $300 to $499 17.0 15.6
   $500 or more 16.9 16.0
DEMONSTRATION FEEDER PROGRAM
   Department of Elder Affairs 38.7 38.3
   Developmental Services 56.6 57.5
   Adult Services 4.7 4.2
Care Recipients' Satisfaction with Care and Unmet Needs
HOW SATISFIED WITH OVERALL CARE ARRANGEMENTS
   Very satisfied 40.5 47.1
   Satisfied 37.5 31.3
   Dissatisfied 17.5 16.8
   No paid services or goods in past week 4.5 4.8
NOT GETTING ENOUGH HELP WITH:
   Household activitiesb 73.2 72.0
   Personal carec 59.6 56.1
   Transportationd 54.5 55.5
Care Recipients' Preferences About Consumer Direction
BEING ALLOWED TO PAY FAMILY MEMBERS OR FRIENDS WAS VERY IMPORTANT 75.0 75.2
HAVING A CHOICE ABOUT PAID WORKERS' SCHEDULE WAS VERY IMPORTANT 84.3 85.6
HAVING A CHOICE ABOUT TYPES OF SERVICES RECEIVED WAS VERY IMPORTANT 92.7 91.7
Education and Work Experience of Decision Makerf
GRADUATED FROM HIGH SCHOOL 76.0 76.1
EVER SUPERVISED SOMEONE 66.1 64.6
EVER HIRED SOMEONE PRIVATELY 68.1 67.5
EVER WORKED FOR PAY 96.0 95.0
Other
PROXY COMPLETED ALL OR MOST OF BASELINE SURVEY 78.4 77.4
APPOINTED A REPRESENTATIVE AT ENROLLMENT 86.1 85.4
ENROLLMENT MONTH WAS BETWEEN:
   June 2000 and May 2001 50.7 51.0
   June 2001 and July 2002 49.3 49.0
Primary Informal Caregivers' Characteristics
AGE IN YEARS
   39 or younger 7.9 8.3
   40 to 64 70.7 70.1
   65 or older 21.4 21.5
FEMALE 83.8 84.2
RELATIONSHIP TO CARE RECIPIENT
    **
   Spouse 5.7 6.6
   Parent 50.2 46.2
   Daughter or son 28.0 24.5
   Other relative 10.9 14.4
   Nonrelative 5.2 8.3
HISPANIC 23.0 26.7
WHITE 70.2 71.2
MARRIED 57.8 57.4
HAS CHILD(REN) YOUNGER THAN AGE 18 17.7 19.7
HIGHEST LEVEL OF EDUCATION
   8 years or less 6.7 6.6
   9 to 12 years, but no high school diploma or GED 11.8 12.7
   High school diploma or GED 30.8 32.9
   At least some college 50.7 47.7
EMPLOYEDg 45.9 45.9
EXPRESSED INTEREST IN BEING PAID FOR CAREGIVINGg 34.1 27.9**
SAMPLE SIZE 617 576
SOURCE: MPR's baseline evaluation interview, conducted between June 2000 and July 2002; caregiver interview, conducted between May 2001 and May 2003; and program records.
  1. Needed hands-on or standby help or did not perform activity at all.
  2. Household activities include meal preparation, laundry, housework, and yard work.
  3. Personal care includes eating, dressing, and bathing.
  4. Transportation includes trips for medical and nonmedical reasons.
  5. Routine health care includes checking vital signs and help taking medicine or doing exercises.
  6. Reflects the characteristics of the person (the care recipient or a representative if the representative responded to the baseline interview) who would make care-related decisions in the demonstration program. See text for description of imputation procedures used when the characteristics of the decision maker were not observed.
  7. As reported by care receipitnes during the baseline interview.

*Difference between treatment and control groups significantly different from 0 at the .10 level, two-tailed test. **Difference between treatment and control groups significantly different from 0 at the .05 level, two-tailed test. ***Difference between treatment and control groups significantly different from 0 at the .01 level, two-tailed test.

TABLE A.2c: Baseline Characteristics of Care Recipients and Primary Informal Caregivers, by Evaluation Status: New Jersey(Percentages)
Characteristic Treatment Group Control Group
Care Recipients' Demographics
AGE IN YEARS
   18 to 39 19.4 17.6
   40 to 64 25.8 27.5
   65 to 79 26.8 29.5
   80 or older 28.0 25.5
FEMALE 73.3 71.8
HISPANIC 34.0 36.6
RACE
   White 57.2 53.3
   Black 34.5 35.8
   Other 8.3 10.9
LIVE ALONE 21.8 20.2
GRADUATED FROM HIGH SCHOOL 39.7 37.0
DESCRIBED AREA OF RESIDENCE AS:
   Rural 11.3 13.1
   Not rural but high-crime or lacking adequate public transportation 40.3 41.3
   Not rural, not high-crime, having adequate public transportation 48.4 45.6
Care Recipients' Health and Functioning
RELATIVE HEALTH STATUS
   Excellent or good 22.1 21.7
   Fair 39.4 33.9
   Poor 38.5 44.4
NOT INDEPENDENT IN PAST WEEK IN:a
   Getting in or out of bed 72.8 74.2
   Bathing 90.8 92.1
   Using toilet/diapers 74.9 77.0
Care Recipients' Use of Personal Assistance
RECEIVED ANY HELP IN PAST WEEK WITH:
   Household activitiesb 98.4 99.0
   Personal carec 91.8 92.7
   Transportationd 68.8 68.6
   Routine health caree 85.9 88.1
NUMBER OF UNPAID CAREGIVERS WHO PROVIDED HELP IN PAST WEEK
   1 29.2 30.0
   2 26.2 27.1
   3 or more 44.6 42.9
RECEIVING PERSONAL CARE SERVICES FOR 6 MONTHS OR LONGER 43.4 46.0
NUMBER OF PAID CAREGIVERS IN PAST WEEK
    *
   0 16.3 17.4
   1 53.8 58.8
   2 or more 29.8 23.8
PROPOSED WEEKLY ALLOWANCE
   Less than $150 24.2 19.8
   $150 to $299 43.0 43.0
   $300 to $499 27.7 30.9
   $500 or more 5.1 6.3
Care Recipients' Satisfaction with Care and Unmet Needs
HOW SATISFIED WITH OVERALL CARE ARRANGEMENTS
   Very satisfied 34.5 31.3
   Satisfied 35.6 36.0
   Dissatisfied 24.1 28.3
   No paid services or goods in past week 5.9 4.5
NOT GETTING ENOUGH HELP WITH:
   Household activitiesb 78.5 78.1
   Personal carec 75.5 75.9
   Transportationd 69.0 65.2
Care Recipients' Preferences About Consumer Direction
BEING ALLOWED TO PAY FAMILY MEMBERS OR FRIENDS WAS VERY IMPORTANT 88.4 86.1
HAVING A CHOICE ABOUT PAID WORKERS' SCHEDULE WAS VERY IMPORTANT 88.1 88.7
HAVING A CHOICE ABOUT TYPES OF SERVICES RECEIVED WAS VERY IMPORTANT 92.5 92.5
Care Recipients' Work Experience and Other Characteristics
EVER SUPERVISED SOMEONE 31.7 27.2
EVER HIRED SOMEONE PRIVATELY 28.8 24.4
EVER WORKED FOR PAY 73.7 69.8
PROXY COMPLETED ALL OR MOST OF BASELINE SURVEY 51.5 56.1
APPOINTED A REPRESENTATIVE AT ENROLLMENT NA NA
ENROLLMENT MONTH WAS BETWEEN:
   November 1999 and December 2000 47.3 48.6
   January 2001 and July 2002 52.7 51.4
Primary Informal Caregivers' Characteristics
AGE IN YEARS
    **
   39 or younger 21.3 18.1
   40 to 64 64.8 61.9
   65 or older 13.9 20.0
FEMALE 78.7 80.2
RELATIONSHIP TO CARE RECIPIENT
   Spouse 7.7 9.1
   Parent 15.8 20.4
   Daughter or son 50.7 49.0
   Other relative 16.5 14.1
   Nonrelative 9.3 7.5
HISPANIC 33.9 37.8
WHITE 44.8 46.2
MARRIED 52.6 50.6
HAS CHILD(REN) YOUNGER THAN AGE 18 30.4 29.2
HIGHEST LEVEL OF EDUCATION
   8 years or less 12.4 14.4
   9 to 12 years, but no high school diploma or GED 13.9 18.3
   High school diploma or GED 26.5 26.1
   At least some college 47.1 41.3
EMPLOYEDf 50.8 47.0
EXPRESSED INTEREST IN BEING PAID FOR CAREGIVINGf 37.5 39.1
SAMPLE SIZE 546 496
SOURCE: MPR's baseline evaluation interview, conducted between December 1998 and July 2002; caregiver interview, conducted between September 2000 and June 2003; and program records.
  1. Needed hands-on or standby help or did not perform activity at all.
  2. Household activities include meal preparation, laundry, housework, and yard work.
  3. Personal care includes eating, dressing, and bathing.
  4. Transportation includes trips for medical and nonmedical reasons.
  5. Routine health care includes checking vital signs and help taking medicine or doing exercises.
  6. As reported by care recipients during the baseline interview.

*Difference between treatment and control groups significantly different from 0 at the .10 level, two-tailed test. **Difference between treatment and control groups significantly different from 0 at the .05 level, two-tailed test. ***Difference between treatment and control groups significantly different from 0 at the .01 level, two-tailed test.NA = data not available.

TABLE A.3: Minimum Detectable Effects
Characteristic Arkansas Florida New Jersey
Detectable Effect (Percentage Points)
BINARY VARIABLE MEAN
   .50 6.57 7.20 7.71
   .30 or .70 6.02 6.60 7.07
   .10 or .90 3.94 4.32 4.63
Detectable Effect (Hours)
CONTINUOUS VARIABLE
   Hours of Care Provided by Live-In Caregivers 13.67 12.47 16.63
TABLE A.4: Examples of Adjustments to Hours of Care Provided
Case Reported Adjusted
CareRecipientHours HouseholdHours TotalHours CareRecipientHours HouseholdHours TotalHours
1 252 252 504 126 126 252
2 336 84 420 252 84 336
3 168 224 392 168 56 224

APPENDIX B: ADDITIONAL RESULTS

TABLE B.1: Estimated Effects on Living Arrangements for, and Assistance Provided by, Primary Informal Caregivers, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
Living Arrangements in Recent Two Weeksa
LIVED WITH CARE RECIPIENT 57.3 56.8 0.6(.719) 83.1 82.1 1.0(.417) 61.6 58.7 2.9*(.089)
LIVED WITHIN 10 MINUTES' TRAVEL TIME OF CARE RECIPIENT 32.0 31.9 0.1(.969) 10.5 11.5 -1.0(.488) 28.1 29.0 -0.9(.664)
Assistance Provided in Recent Two Weeksa
PROVIDED ANY ASSISTANCE 93.1 91.2 1.9(.197) 94.7 94.2 0.5(.698) 92.3 90.9 1.4(.421)
AMONG THOSE PROVIDING ASSISTANCE:
   Helped with personal careb 92.5 89.7 2.8*(.077) 85.7 86.7 -1.0(.548) 90.0 87.6 2.4(.206)
   Helped with household activitiesc,d 98.1 99.1 -1.0(.144) 99.5 99.0 0.5(.414) 97.6 98.8 -1.2(.199)
   Helped with routine health caree 90.7 90.4 0.2(.882) 88.6 88.4 0.1(.942) 92.2 86.3 5.9***(.003)
   Socialized with or kept care recipient company 96.9 97.9 -1.0(.309) 97.0 97.3 -0.3(.743) 96.7 97.5 -0.8(.479)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit models, unless noted.
  1. The most recent 2 weeks the care recipient lived in the home or community during the 2 months before the interview.
  2. Personal care included eating, bathing, transferring, and using the toilet.
  3. Household activities include meal preparation, laundry, housework, and yard work.
  4. Impacts could not be estimated with logit models. Results presented are the unadjusted means and treatment-control differences.
  5. Routine health care includes checking vital signs and help taking medicine or doing exercises.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

TABLE B.2: Estimated Effects on Pirmary Informal Caregivers' Knowledge and Preparedness, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
FEELS WELL INFORMED ABOUT CARE RECIPIENT'S CONDITION AND SERVICES
   Strongly agrees 84.9 76.5 8.4***(.000) 74.3 70.2 4.1(.117) 81.3 67.0 14.3***(.001)
   Disagrees 3.4 7.2 -3.8***(.000) 5.4 6.2 -0.8(.559) 3.2 7.2 -4.0***(.006)
FEELS FULLY PREPARED TO MEET EXPECTATIONS IN HELPING CARE RECIPIENT
   Strongly agrees 90.2 84.3 5.9***(.001) 87.1 85.3 1.8(.375) 86.3 79.9 6.4***(.006)
   Disagreesa 1.8 3.4 -1.6*(.061) 1.8 2.8 -1.0(.244) 3.2 5.4 -2.2*(.082)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Outcomes were measured "at present" or "when you were last helping." Means were predicted with logit models, unless noted.
  1. Impacts could not be estimated with the logit model. Results presented are the unadjusted means and treatment-control differences.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

TABLE B.3: Estimated Effects on Primary Informal Caregivers' Physical Functioning, by State
Outcome Arkansas Florida New Jersey
Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value) Predicted Means(Percents) EstimatedEffect(p-Value)
TreatmentGroup ControlGroup TreatmentGroup ControlGroup TreatmentGroup ControlGroup
BECAUSE OF ILLNESS OR DISABILITY, HAS PROBLEMS WITH AT LEAST ONE:
   Instrumental activity of daily livinga 11.4 18.3 -6.9***(.000) 15.3 20.1 -4.9**(.022) 12.9 16.8 -3.9*(.058)
   Activity of daily livingb 4.9 7.0 -2.2*(.091) 5.9 8.3 -2.4*(.100) 5.5 7.1 -1.6(.279)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. Includes those who became paid workers for care recipients randomly assigned to the treatment group. Means were predicted with logit models.
  1. Includes meal preparation, housework, shopping, taking medicine, and managing money.
  2. Includes eating, getting in or out of bed or chairs, dressing, bathing, or using the toilet.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

TABLE B.4: Estimated Effects of Care Recipient and Caregiver Characteristics on Whether Primary Informal Caregivers Became Paid Workers, by State(Omitted Categories in Parentheses)
Characteristic Arkansas (n = 671) Florida (n = 617) New Jersey (n = 546)
EstimatedOdds Ratio p-Value EstimatedOdds Ratio p-Value EstimatedOdds Ratio p-Value
Care Recipient Characteristics
DEMOGRAPHICS
65 (or 60) or oldera 0.69 .183 0.75 .825 1.06 .853
Female 1.47 .109 0.84 .441 1.00 .990
Hispanic n.a. n.a. 1.16 .767 2.22 .249
Racial minority 0.90 .758 0.79 .588 1.04 .902
Did not graduate from high schoolb 0.63* .057 0.86 .577 1.10 .700
LIVING ARRANEGMENTS
Lived alone 0.53** .046 0.86 .789 0.86 .661
Described area of residence as rural or high-crime or lacking public transportation 1.19 .388 1.24 .295 1.31 .212
HEALTH AND FUNCTIONING
In poor health relative to peers 0.91 .641 1.14 .599 1.19 .420
Not independent in past week in:            
   Getting in or uot of bed 0.92 .767 1.23 .464 1.62 .097
   Bathing 0.77 .532 1.76 .159 1.58 .272
   Using toilet/diapers 1.18 .584 0.85 .621 0.73 .296
UNPAID AND PAID ASSISTANCE
Number of informal caregivers who helped last week:            
   (One)            
   Two 0.97 .891 1.20 .507 2.12*** .005
   Three or more 1.03 .915 1.01 .960 0.93 .770
Receiving publicly funded home care (at all or for at least 6 months)c 0.81 .297 1.01 .944 0.88 .534
Satisfied with overall care arrangements 0.99 .969 0.69 .130 1.11 .640
Demonstration Feeder Program            
   (Department of Elder Affairs)            
   Developmental Services or Adult Services n.a. n.a. 0.78 .846 n.a. n.a.
ASSISTANCE NEEDS AND PREFERENCES
Medicaid personal care plan indicates high level of need (in hours or dollars)d 1.61** .033 1.25 .338 1.00 .994
Not getting enough help with:            
   Personal care 0.99 .995 1.19 .515 0.98 .945
   Transportation 0.79 .246 0.90 .657 0.68* .099
   Household activities 0.69 .101 1.38 .275 1.50 .176
Proxy completed most or all of baseline evaluation survey 0.56** .018 0.84 .608 0.49*** .005
Representative would manage monthly allowance 0.61** .030 2.28* .067 n.a. n.a.
Ability to pay family members or friends was very important 1.33 .326 1.75** .037 3.46*** .001
Setting paid workers' schedule was very important 0.52** .013 1.40 .307 1.33 .414
Choosing types of paid services was very important 1.63* .085 0.62 .266 0.71 .410
WORK AND SUPERVISORY EXPERIENCE
Ever supervised someone 0.85 .476 1.06 .816 0.96 .860
Ever hired someone privately 0.90 .636 1.31 .256 0.94 .804
Ever worked for pay 0.90 .685 0.97 .952 1.04 .870
Caregiver Characteristics
DEMOGRAPHICS
Age (years)            
   (39 or younger)            
   40 to 64 0.60* .055 0.97 .945 0.62 .103
   65 or older 0.48** .047 0.82 .675 0.29*** .004
Female 2.36*** .001 2.24** .012 1.97*** .010
Hispanic n.a. n.a. 0.71 .504 0.42 .211
Racial minority 1.46 .285 0.53 .130 0.67 .206
Did not graduate from high school 1.03 .903 1.22 .517 0.55** .027
Married 1.66** .011 0.75 .198 1.24 .357
Has child(ren) 18 or younger 0.88 .589 1.49 .165 1.32 .263
Employed, other than by care recipient 0.99 .981 1.64** .029 0.65* .053
Household income in month before caregiver interview            
   ($1,000 or less)            
   $1,001-3,000 0.66** .043 0.79 .366 1.09 .740
   $3,001 or more 0.17*** .000 0.74 .412 0.81 .560
   Missing income data 0.15*** .000 0.86 .716 0.22*** .001
RELATIONSHIP TO CARE RECIPIENT AND LIVING ARRANGEMENTS
Relationship to care recipient:            
   (Not related)            
   Daughter or son 1.32 .407 1.15 .802 1.31 .524
   Parent 0.94 .903 0.26** .025 1.27 .609
   Spouse n.a. n.a. 1.17 .821 0.48 .213
   Other relative 1.29 .501 0.63 .437 1.23 .788
Lived with care recipient at baseline 0.90 .709 1.43 .374 2.09** .022
SOURCE: Program records from each of the demonstration states and MPR's baseline and caregiver interviews. Baseline interviews were conducted in Arkansas between December 1998 and April 2001; in Florida between June 2000 and July 2002; and in New Jersey between November 1999 and July 2002. Caregiver interviews were conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Primary informal caregivers are those providing the most unpaid care to care recipients at baseline. The Arkansas analysis sample excludes 50 caregivers who could not become paid workers because they were married to their care recipient. Odds ratios were estimated with logit models.
  1. The Florida model controlled for whether care recipients were age 60 or older (instead of 65 or older) to parallel the age groups served by the HCBS waiver programs that fed into the demonstration.
  2. The Florida model controlled for the education of the person (that is, the care recipient or a representative if the representative responded to the baseline interview) who would make care-related decisions in the demonstration program. The Arkansas and New Jersey models controlled for the education of the care recipient, whether or not they would use a representative in the program.
  3. The Arkansas model controlled for whether recipients were receiving publicly funded home care at baseline, regardless of how long they had been receiving it. The Florida and New Jersey models controlled for whether care recipients had been receiving publicly funded home care for 6 months or longer at baseline. The measures differ because care recipients in Florida and New Jersey typically received traditional home care services before they enrolled in the demonstration; in Arkansas, this was not necessarily the case.
  4. The Arkansas model controlled for whether the care recipient's Medicaid personal care plan included 7 or more hours of care per week. The Florida and New Jersey models controlled for whether the plan included benefits valued at $150 or more per week.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.n.a. = not applicable.

TABLE B.5 : Reasons Treatment Group Caregivers Did Not Become Paid Workers(Percentages)
Reason Arkansas Florida New Jersey
Helped out of love, devotion, or family tradition 28.4 26.3 27.8
Not allowed to be paid under the programa 22.2 15.9 13.9
Not able to perform all tasks, or lived far away/had other obligations, or care recipient had someone else in mind 15.0 10.7 16.2
Benefit not enough to pay me and others or was needed for other things 10.5 9.1 8.3
Did not know could get paid or program made an error 8.8 14.4 15.2
Care recipient disenrolled from the program 8.2 10.5 8.9
Intends to be paid after care recipient receives program allowance 0.0 4.7 2.3
Too much hassle or caregiver disagrees with program philosophy 0.0 4.4 2.0
Did not need or want the money 3.9 2.1 3.3
Other 2.9 1.9 2.0
NUMBER OF RESPONDENTS 306 429 302
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: Treatment group caregivers are those identified at baseline as the primary informal caregivers of care recipients who were randomly assigned to participate in a Cash and Counseling program.
  1. In Arkansas, caregivers were not allowed to be paid if they were the care recipient's spouse or representative. In New Jersey, caregivers were not allowed to be paid if they were the care recipient's representative. Florida did not impose hiring restrictions during its evaluation period. However, if a representative was also a paid worker, someone else from the care recipient's "circle of support" was asked to verify the completion of agreed upon work.
TABLE B.6: Estimated Effects of Becoming a Paid Worker on Key Caregiver Outcomes, by State
Outcome Arkansas Florida New Jersey
Estimated Effect of Being a Treatment Group Caregiver and:
BecomingPaid(p-Value) RemainingUnpaid(p-Value) BecomingPaid(p-Value) RemainingUnpaid(p-Value) BecomingPaid(p-Value) RemainingUnpaid(p-Value)
Well-Being and Satisfaction
CARE RECIPIENT REFUSES TO COOPERATE WHEN CAREGIVER TRIES TO HELP -11.0***(.000) 1.4(.668) -12.3***(.002) -2.7(.343) -3.4(.348) -0.2(.945)
HOW SATISFIED WITH CARE RECIPIENT'S OVERALL CARE ARRANGEMENTS
   Very satisfied 23.5***(.000) 9.8***(.003) 32.4***(.000) 12.5***(.000) 28.6***(.000) 13.5***(.000)
   Dissatisfied -14.5***(.000) -9.0***(.000) -15.8***(.000) -8.4***(.001) -21.7***(.000) -12.0***(.000)
HOW OFTEN WORRIES THAT CARE RECIPIENT DOES NOT HAVE ENOUGH HELP IN CAREGIVER'S ABSENCE
   Rarely or not at all 18.6***(.000) 12.6***(.000) 25.1***(.000) 11.5***(.000) 18.7***(.000) 8.2***(.006)
   Quite a lot -20.4***(.000) -12.2***(.000) -21.2***(.000) -9.2***(.002) -24.3***(.000) -13.5***(.000)
CAREGIVING LIMITS:
   Privacy -19.4***(.000) -6.5**(.040) -15.0***(.000) -0.7(.806) -19.7***(.000) -1.7(.610)
   Free time or social life -15.4***(.000) -5.4*(.090) -19.5***(.000) -2.1(.441) -14.2***(.000) 1.2(.716)
LEVEL OF EMOTIONAL STRAIN AS A RESULT OF CAREGIVING
   Little or none 11.7***(.000) -2.0(.555) 11.3***(.005) -2.2(.415) 12.5***(.000) 0.0(.987)
   A great deal -12.7***(.000) -0.7(.827) -9.3**(.017) -0.0(.994) -13.6***(.000) -2.3(.498)
HOW SATISFIED WITH LIFE
   Very satisfied 19.2***(.000) 0.2(.948) 24.7***(.000) 6.7**(.030) 20.0***(.000) 9.7***(.004)
   Dissatisfied -13.2***(.000) -4.4*(.063) -8.1***(.000) -4.6*(.057) -20.1***(.000) -4.6*(.073)
DID NOT LOOK FOR A JOB, OR ANOTHER JOB, BUT WANTED TO -14.3***(.000) -14.0***(.000) -9.0**(.019) -5.5*(.055) -10.2***(.005) -14.0***(.003)
LEVEL OF FINANCIAL STRAIN AS A RESULT OF CAREGIVING
   Little or none 15.2***(.000) 11.4***(.001) 20.6***(.000) 7.7***(.006) 14.7***(.000) 4.0(.254)
   A great deal -16.8***(.000) -7.4***(.009) -18.3***(.000) -4.4(.121) -12.6***(.001) -6.6**(.047)
PHYSICAL HEALTH HAS SUFFERED AS A RESULT OF CAREGIVING -19.7***(.000) 0.3(.918) -22.7***(.000) -6.7**(.016) -15.9***(.001) -4.8(.120)
Amount of Assistance Provided in Recent Two Weeks
NUMBER OF DAYS PROVIDED CARE 0.7***(.001) -0.7***(.003) 0.4*(.089) -0.2(.211) 1.0***(.000) -0.2(.933)
HOURS OF ASSISTANCEa -7.5(.152) -15.8*(.075) 10.3(.169) -17.4*(.061) 21.1*(.067) -0.5(.464)
   Among live-in caregivers -12.9*(.069) -12.7(.127) 8.5(.289) -17.5***(.003) 20.8**(.028) -3.2(.724)
   --   Hours that benefited care recipient only -5.8(.196) 0.1(.986) 7.0(.211) -11.0***(.007) 8.4(.178) -3.3(.581)
   --   Hours that benefited entire household -7.1(.103) -12.8**(.013) 1.5(.737) -6.5*(.061) 12.4**(.031) 0.1(.984)
   Among visiting caregivers -0.2(.977) -19.9***(.004) 18.6(.313) -17.0(.225) 21.7**(.020) 3.6(.690)
SOURCE: MPR's caregiver interview, conducted in Arkansas between February 2000 and April 2002; in Florida between May 2001 and May 2003; and in New Jersey between September 2000 and June 2003. NOTE: The estimated effects of becoming paid (remaining unpaid) are the differences between the predicted means for treatment group caregivers who became paid workers (remained unpaid) and those for control group caregivers. Means were predicted with logit or ordinary-least-squares models. Sample sizes varied from measure to measure. In Arkansas, the largest sample included 712 control group caregivers, 403 paid treatment group caregivers, and 310 unpaid treatment group caregivers. In Florida, the largest sample included 576 control group caregivers, 176 paid treatment group caregivers, and 441 unpaid treatment caregivers. In New Jersey, the largest sample included 508 control group caregivers, 168 paid treatment group caregivers, and 380 unpaid treatment group caregivers.
  1. The estimates for total hours are constructed as weighted averages of the estimates for caregivers who are live-in and those who are visiting, with the weights being the proportion of all caregivers (treatment and control groups combined) who were live-in or visiting at followup. In Arkansas, these proportions were 0.571 (live-in) and 0.429 (visiting); in Florida, they were 0.826 and 0.174; and in New Jersey, they were 0.603 and 0.397. The variance of the estimated effect used to construct the t-statistic and corresponding p-value is are the variances of the estimated coefficients on the treatment status indicator from the regressions on live-in and visiting caregivers, respectively. This approach ensures that the impact on total hours is a weighted average of the impacts on live-in and visiting caregivers.

*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.

TABLE B.7a: Estimated Effects on Key Caregiver Outcomes, by Care Recipients' Age Group: Arkansas
Outcome Care Recipients' Age Group
18 to 64 (n = 365) 65 or Older (n = 1,068)
PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value) PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value)
Satisfaction
VERY SATISFACTION WITH CARE RECIPIENT'S OVERALL CARE ARRANGEMENTS (Percent) 62.2 41.9 20.2***(.000) 60.4 43.0 17.4***(.000)
Amount of Assistance Provided During Recent Two Weeks
LIVE-IN CAREGIVERS (Hours) 141.3 155.4 -14.1(.243) 140.6 152.8 -12.2*(.081)
VISITING CAREGIVERS (Hours) 49.5 49.6 -0.1(.990) 49.4 58.1 -8.7(.154)
Well-Being
CAREGIVING CAUSED A GREAT DEAL OF: (Percent)
   Emotional strain 24.8 36.7 -11.9**(.016) 27.5 33.6 -6.1**(.026)
   Financial strain 22.4 38.6 -16.1***(.001) 22.7 34.8 -12.1***(.000)
   Physical strain 19.3 30.0 -10.7**(.020) 24.4 32.7 -8.4***(.002)
SOURCE: MPR's caregiver interview, conducted between February 2000 and April 2002. NOTE: Means were predicted with ordinary least squares or logit models. Subgroup effects were estimated by pooling the two age groups and including an age*treatment status interaction term in the model. Sample sizes vary from measure to measure, because certain survey questions were asked only of sample members who met certain criteria and because of item nonresponse.*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.None of the subgroup effects were significantly different from each other.
TABLE B.7b: Estimated Effects on Key Caregiver Outcomes, by Care Recipients' Age Group: Florida
Outcome Care Recipients' Age Group
18 to 64 (n = 720) 65 or Older (n = 473)
PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value) PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value)
Satisfaction
VERY SATISFACTION WITH CARE RECIPIENT'S OVERALL CARE ARRANGEMENTS (Percent) 45.1 26.2 18.9***(.000) 51.9 35.3 16.5***(.000)
Amount of Assistance Provided During Recent Two Weeks
LIVE-IN CAREGIVERS (Hours) 130.9 137.7 -6.9(.285) 153.0 168.8 -15.8*(.100)
VISITING CAREGIVERS (Hours) 27.8 14.9 12.9(.505) 79.1 98.7 -19.6(.230)
Well-Being
CAREGIVING CAUSED A GREAT DEAL OF: (Percent)
   Emotional strain 29.5 32.4 -2.9(.381) 45.9 48.5 -2.6(.557)
   Financial strain 33.1 44.6 -11.5***(.001) 25.8 30.9 -5.1(.173)
   Physical strain 30.6 39.0 -8.4**(.016) 25.5 38.5 -12.9***(.001)
SOURCE: MPR's caregiver interview, conducted between May 2001 and May 2003. NOTE: Means were predicted with ordinary least squares or logit models. Subgroup effects were estimated by pooling the two age groups and including an age*treatment status interaction term in the model. Sample sizes vary from measure to measure, because certain survey questions were asked only of sample members who met certain criteria and because of item nonresponse.*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.None of the subgroup effects were significantly different from each other.
TABLE B.7c: Estimated Effects on Key Caregiver Outcomes, by Care Recipients' Age Group: New Jersey
Outcome Care Recipients' Age Group
18 to 64 (n = 365) 65 or Older (n = 1,068)
PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value) PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value)
Satisfaction
VERY SATISFACTION WITH CARE RECIPIENT'S OVERALL CARE ARRANGEMENTS (Percent) 46.4 31.6 14.7***(.001) 55.6 31.9 23.7***(.000)
Amount of Assistance Provided During Recent Two Weeks
LIVE-IN CAREGIVERS (Hours)† 154.7 133.9 20.8*(.058) 143.7 147.9 -4.2(.684)
VISITING CAREGIVERS (Hours) 75.1 48.6 26.5**(.020) 61.5 58.5 3.0(.753)
Well-Being
CAREGIVING CAUSED A GREAT DEAL OF: (Percent)
   Emotional strain 43.4 49.7 -6.3(.157) 41.6 49.2 -7.7*(.051)
   Financial strain†† 36.4 40.1 -3.7(.406) 23.5 37.8 -14.3***(.000)
   Physical strain 30.7 36.7 -6.0(.157) 32.7 45.7 -13.1***(.001)
SOURCE: MPR's caregiver interview, conducted between September 2000 and June 2003. NOTE: Means were predicted with ordinary least squares or logit models. Subgroup effects were estimated by pooling the two age groups and including an age*treatment status interaction term in the model. Sample sizes vary from measure to measure, because certain survey questions were asked only of sample members who met certain criteria and because of item nonresponse.*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.†Estimated effects for the two subgroups were significantly different from each other at the .10 level, two-tailed test. ††Estimated effects for the two subgroups were significantly different from each other at the .05 level, two-tailed test.
TABLE B.8: Estimated Effects of the Arkansas Program on Caregiver Satisfaction and Well-Being for Subgroups Defined by Whether Care Recipients Participated in ElderChoices During Follow-up(Elderly Sample Members Only)
Caregiver Outcomes Care Recipient Participated in ElderChoices Care Recipient Did Not Participate in ElderChoices
PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value) PredictedTreatmentGroup Mean PredictedControlGroup Mean EstimatedEffect(p-Value)
VERY SATISFIED WITH OVERALL CARE ARRANGEMENTS††† 58.1 45.7 12.4***(.001) 67.7 39.5 28.2***(.000)
WORRIES QUITE A LOT THAT:
   Care recipient does not have enough help† 36.6 50.2 -13.6***(.000) 30.3 54.1 -23.8***(.000)
   Care recipient's safety is at risk†† 41.3 50.0 -8.8***(.021) 37.6 59.1 -21.5***(.000)
   Someone will take care recipient's belongings† 13.5 15.3 -1.8(.517) 11.3 21.9 -10.6***(.005)
CAREGIVING LIMITS PRIVACY†† 47.1 52.6 -5.5(.137) 35.8 55.4 -19.7***(.001)
CAREGIVING CAUSES A GREAT DEAL OF:
   Emotional strain 29.1 33.7 -4.6(.182) 27.6 35.2 -7.7(.111)
   Financial strain 22.1 36.6 -14.4***(.000) 25.6 34.9 -9.3**(.048)
   Physical strain 27.3 34.7 -7.3**(.032) 20.9 32.5 -11.6**(.010)
VERY SATISFIED WITH LIFE 48.6 37.9 10.7***(.005) 56.3 40.4 15.9***(.002)
SOURCE: MPR's caregiver interview, conducted between February 2000 and April 2002, and Arkansas Medicaid data. NOTE: Means were predicted with logit models. Subgroups effects were estimated by interacting the treatment status indicator and an ElderChoices enrollment status indicator. Sample sizes vary from measure to measure (from 1,045 to 1,065) because of item nonresponse. In the largest sample used, 677 primary informal caregivers were assisting care recipients who participated in ElderChoices, and 388 were assisting care recipients who were not participating.*Significantly different from zero at the .10 level, two-tailed test. **Significantly different from zero at the .05 level, two-tailed test. ***Significantly different from zero at the .01 level, two-tailed test.†Estimated effects for the two subgroups were significantly different from each other at the .10 level, two-tailed test. ††Estimated effects for the two subgroups were significantly different from each other at the .05 level, two-tailed test. †††Estimated effects for the two subgroups were significantly different from each other at the .01 level, two-tailed test.

NOTES

  1. For more information about demonstration implementation in the three states, see Phillips and Schneider (2002, 2003, and 2004).

  2. Florida's initial demonstration design included beneficiaries in the state's Brain and Spinal Cord Injury Program (BSCIP). The participation of BSCIP was delayed, however, so BSCIP beneficiaries were excluded from the MPR evaluation.

  3. The experiences of Florida children and their primary informal caregivers are described in companion reports (Foster et al. 2004 and 2005).

  4. Claims were used to calculate the allowance if they were historically stable and consistent with the consumer's care plan. In practice, claims were not used to calculate the allowances of consumers with developmental disabilities, because their care plans were being systematically revised when the demonstration began. (The revisions resulted from a substantial increase in state funding for the HCBS waiver programs serving people with developmental disabilities.) Claims were used to calculate the allowances of consumers with primarily physical disabilities, however.

  5. After the evaluation period, Florida modified its operational protocol so that that no one could serve as both a representative and paid worker. This restriction is currently enforced in Florida's CDC+ program, which operates under a Section 1115 waiver.

  6. Appendix A includes additional detail on analytic methods.

  7. Unlike beneficiaries who participated in the demonstration, caregivers were interviewed only once; thus, we do not have analogous baseline measures of the caregiver outcomes measured at 10 months.

  8. Assuming the demonstration affected only caregivers of allowance recipients, the effects on those caregivers could be estimated by dividing the overall program effects reported in our tables by the proportion of treatment group members receiving the allowance in each state (0.88 in Arkansas, 0.55 in Florida, and 0.75 in New Jersey). For example, whereas the overall effects on whether caregiving limited privacy were -14.0, -4.8, and -9.4 percentage points in Arkansas, Florida, and New Jersey (Table 8), the respective effects on caregivers of allowance recipients would be -16.0, -8.7, and -12.5.

  9. The caregiver survey also included several questions with five-point scales. In these cases, respondents rated the level of strain they experienced, with 1 representing little or no strain and 5 representing a great deal of strain. We converted each scale into two binary measures. The first was set equal to 1 if the respondent gave a rating of 1; the other was set equal to 1 for ratings of 4 or 5.

  10. In New Jersey, data on representatives were not collected during baseline interviews, because care recipients were not asked to name them until after they were randomly assigned to the treatment group. Program staff estimated that up to two-thirds of elderly New Jersey treatment group members, and a somewhat smaller proportion of nonelderly ones, named representatives.

  11. Appendix Table B.1 shows additional results on caregivers' living arrangements, provision of any assistance, and types of assistance provided.

  12. Appendix Table B.2 shows additional results on whether caregivers considered themselves knowledgeable about, and prepared for, caregiving.

  13. Appendix Table B.3 shows additional results on caregivers' physical functioning.

  14. In each state, we estimated the model for all treatment group caregivers (except spouses in Arkansas), and then for the subset of treatment group caregivers whose care recipients had received the program allowance within 10 months of random assignment. Results were similar for both samples; thus, we report results for the full samples. Results for the restricted samples could be confounded by the interdependence of allowance receipt and payment of caregivers. That is, although caregiver payment is conditional on allowance receipt, caregivers may also affect the likelihood of allowance receipt--by being willing or unwilling to be hired.

  15. The subgroup analysis controlled for the care recipient and caregiver characteristics listed in Table A.2a; thus, it accounted for some important differences between ElderChoices participants and nonparticipants, such as self-reported health status and the number of hours authorized in consumers' Medicaid personal care plans.

Preview
Download

"ICaffect.pdf" (pdf, 37.11Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®