Medicare and Medicaid Home Health and Medicaid ... - Oxford Journals

The Gerontologist Vol. 44, No. 6, 739–749

Copyright 2004 by The Gerontological Society of America

Medicare and Medicaid Home Health and Medicaid Waiver Services for Dually Eligible Older Adults: Risk Factors for Use and Correlates of Expenditures Richard H. Fortinsky, PhD,1 Juliane R. Fenster, MS, MPH,1 and James O. Judge, MD2 Purpose: The purpose of this work was to, among frail dually eligible older adults, determine risk factors for the likelihood of using Medicare home health and Medicaid home health services and to, among service users, determine correlates of Medicare home health, Medicaid home health, and Medicaid waiver service expenditures. Design and Methods: Dually eligible individuals enrolled in Connecticut’s Medicaid homeand community-based services (HCBS) waiver program for the aged (N = 5,232) were identified from a statewide database containing person-level linked data from Medicare claims, Medicaid claims, and uniform clinical assessment forms. Expenditures, based on claims data, were observed from the month following clinical assessment over the period August 1995 to December 1997. Results: In multivariate models controlling for medical conditions and sociodemographic variables, similar functional disability measures were strongly associated with the probability of the use of, and expenditures for, Medicare home health and Medicaid home health services; severe cognitive impairment was strongly associated with greater Medicaid waiver service expenditures. Implications: Given the similarity of

This work was supported by the Robert Wood Johnson Foundation and the Connecticut Department of Social Services under a grant to the New England Consortium, Medicare and Medicaid Integration Project. This publication does not express the views of the Connecticut Department of Social Services or the State of Connecticut. The views and opinions expressed are those of the authors. We also gratefully acknowledge data analysis assistance provided by Ramon I. Garcia, MS, and guidance and insights regarding the study database provided by Daniel Gilden of JEN Associates, Inc., Cambridge, MA. Address correspondence to Richard H. Fortinsky, PhD, Center on Aging, University of Connecticut Health Center, 263 Farmington Ave., Farmington, CT 06030-5215. E-mail: [email protected] 1 Center on Aging, University of Connecticut Health Center, Farmington. 2 Connecticut VNA, Wallingford.

Vol. 44, No. 6, 2004

739

factors associated with Medicare and Medicaid home health service use and expenditures, greater integration of Medicare and Medicaid financing, reimbursement, and delivery strategies for home health services may be feasible and warranted for dually eligible older adults enrolled in state Medicaid HCBS waiver programs. Key Words: Medicare and Medicaid integration, Dual eligibles, Linked data, Functional disability, Home- and community-based care

Current fiscal challenges facing state Medicaid budgets (Boyd, 2003) and discussions in Congress of Medicare reform shine an especially bright light on older adults eligible for both publicly funded health insurance programs (i.e., dually eligible). Dually eligible individuals are more likely than Medicareonly beneficiaries to suffer from chronic physical and mental health problems and disabilities; therefore, they cost the federal government four times what each Medicare-only beneficiary costs, and they account for nearly one-third of combined Medicare and Medicaid spending (Clark & Hulbert, 1998; Saucier et al., 1998). Because of their vulnerabilities, dually eligible beneficiaries often require a comprehensive set of acute and long-term care services. They may, however, be confronted with a lack of clinical care continuity as well as administrative and funding fragmentation between Medicare and Medicaid programs. Although innovative programs such as On Lok and Program of All-Inclusive Care for the Elderly (PACE) replication sites have been established to address these problems, they currently operate only in selected locales and serve limited numbers of clients (Bodenheimer, 1999; Eng, Pedulla, Eleazer, McCann, & Fox, 1997). Meanwhile, several states are considering or developing integrated care

systems for dually eligible beneficiaries because of the growing magnitude of their impact on state Medicaid budgets (U.S. Government Accounting Office [GAO], 2000). Better information about dually eligible beneficiaries is essential in planning these initiatives, in obtaining necessary federal approvals, and in developing creative reimbursement strategies to integrate Medicare and Medicaid funding streams and service-delivery mechanisms (Clark & Hulbert, 1998; Miller & Weissert, 2003). Such information would be especially beneficial to help integrate Medicare and Medicaid services for frail community-dwelling dually eligible older adults who are eligible for nursing facility admission, similar to the target populations in On Lok and PACE replication sites. To address these knowledge gaps, we studied a sample of dually eligible older adults enrolled in a Medicaid home- and community-based services (HCBS) waiver program in the State of Connecticut. By federal law, Medicaid HCBS waiver program clients in all states must be classified as eligible for nursing home admission; therefore, medically and functionally, they closely resemble On Lok and PACE program clients, for whom integrated Medicare and Medicaid financing strategies have already been developed. We sought to determine whether similar risk factors were associated with the use of and expenditures for both Medicare-reimbursed and Medicaid-reimbursed services to stimulate more widespread discussions among public policy makers and service providers about evidence required to develop more integrated Medicare and Medicaid initiatives for frail dually eligible older adults residing in a given geographic area. Specifically, this article addresses two major questions: In a sample of Medicaid HCBS waiver program for the aged clients, (a) What risk factors are associated with the likelihood of using Medicare home health and Medicaid home health services? (b) Among those who use services, what are the most important correlates of Medicare home health care expenditures, Medicaid home health expenditures, and Medicaid waiver service expenditures? Background Connecticut and the Medicaid HCBS Waiver Program The Medicaid HCBS waiver program was established in 1981 as part of the Omnibus Budget Reconciliation Act and is limited to individuals living at home who would otherwise require the level of care provided in nursing facilities (Harrington, Carrillo, Wellin, Miller, & LeBlanc, 2000; O’Keeffe, 1996). States may target Medicaid HCBS programs to specific groups of individuals such as older adults (aged), younger adults with physical disabilities (disabled), and individuals with mental retardation or developmental disabilities. States also have 740

flexibility to establish need criteria for Medicaid HCBS waiver program eligibility (Harrington et al., 2000). Connecticut, like most states, gives greatest weight to medical and nursing needs and functional needs when determining waiver program eligibility (O’Keeffe, 1996). Eligibility criteria for the Medicaid HCBS waiver program for the aged did not change in Connecticut during the study period. The range of Medicaid waiver services provided in Connecticut is similar to that of services provided in most states’ Medicaid HCBS waiver programs for the aged (O’Keeffe, 1996). However, the Connecticut Medicaid program does not offer a personal care services benefit in addition to the Medicaid home health benefit, and no Medicaid-managed care program is available for older adults in Connecticut. To place the Medicaid HCBS waiver program for the aged in the context of Connecticut’s older dually eligible population during the study period, in 1995– 1996, there were an estimated 27,000 Medicare beneficiaries aged 65 and older in Connecticut also covered by Medicaid (Lamphere, Brangan, Bee, & Semansky, 1998). In 1996, an estimated 18,000 of these dually eligible older adults resided in nursing facilities (Bectel & Graves, 1998); therefore, an estimated 9,000 dually eligible older adults lived at home. The average monthly caseload in the Medicaid HCBS waiver program for the aged during state fiscal year (SFY) 1996 was 4,442 (Connecticut Department of Social Services, 1997). Therefore, approximately 50% of Connecticut’s dually eligible older population living at home during the study period was enrolled in the Medicaid HCBS waiver program. Focus on Functional Disability and Cognitive Impairment as Risk Factors and Correlates Numerous studies have shown that functional disability, measured most often as dependence in personal activities of daily living (ADL) and instrumental ADL (IADL), is a major predictor of home care use (Branch et al., 1988; Coughlin, McBride, Perozek, & Liu, 1992; Fortinsky, Granger, & Seltzer, 1981; Fried, Bradley, Williams, & Tinetti, 2001; Kemper, 1992; Madigan, Tullai-McGuinness, & Neff, 2002; Solomon, Wagner, Maremberg, Acampora, Cooney, & Inouye, 1993). However, none of these studies examined Medicare and Medicaid home health care use in the same cohort of dually eligible older adults. Cognitive impairment is characterized by memory loss, compromised judgment, and impaired decisionmaking skills. Alzheimer’s disease, the most common cause of cognitive impairment, affects an estimated 4 million older Americans today; prevalence is expected to reach 14 million by the year 2040 (Costa et al., 1996; Evans et al., 1992; Small et al., 1997). Several studies based on Medicare claims data found that Medicare expenditures are considerably higher for beneficiaries The Gerontologist

with Alzheimer’s disease and other types of cognitive impairment compared with those without such a diagnosis (Newcomer, Clay, Luxenberg, & Miller, 1999; Taylor & Sloan, 2000; Weiner, Powe, Weller, Shaffer, & Anderson, 1998). From the perspective of the Medicaid program, cognitive impairment has been implicated as a risk factor for greater use of long-term care services among community-dwelling older adults, but little evidence exists to support this contention. Moreover, uniform clinical measures of cognitive impairment are not routinely incorporated into eligibility criteria for Medicaid home- and community-based care programs (Fogel, Brock, Goldscheider, & Royall, 1994; O’Keeffe, 1999).

Design and Methods Sources of Data Results in this article are based on data from the Connecticut Long Term Care Health (CLTC) database constructed for the Connecticut Department of Social Services by JEN Associates, Inc. (Cambridge, MA). The CLTC database contains data extracted from Medicare and Medicaid claims and linked at the person level and aggregated into monthly Medicare and Medicaid expenditures for a wide variety of covered health services. In this work, we focused on monthly expenditures for Medicare home health services, Medicaid home health services, and services covered by Connecticut’s Medicaid HCBS waiver program (hereafter referred to as Medicaid waiver services). Claims data from the CLTC database relevant to this study cover the time period August 1995 to December 1997. The CLTC database also contains clinical assessment data linked at the person level to claims data for dually eligible individuals served by Medicaid waiver services for the aged. Clinical assessments were done using the uniform assessment tool known as the Modified Community Care Assessment tool (O’Keeffe, 1996), which was developed for use in Connecticut by adapting measures of medical, functional, and behavioral problems from a variety of sources. This tool is used to finalize waiverprogram eligibility determination, determine service needs for eligible clients, and calculate whether a plan of care can be developed that will safely and cost-effectively meet the client’s needs in the community (O’Keeffe, 1996). The lag between clinical assessment and medical eligibility determination is usually 30 days. Clinical assessments are conducted annually after eligibility is established. Study Cohort Individuals from the CLTC database were considered eligible for the study cohort if they had Vol. 44, No. 6, 2004

741

a clinical assessment completed between July 1995 and June 1997, followed by at least 1 month of Medicaid waiver program for the aged eligibility. We chose this time period because claims data in the CLTC database were truncated at December 1997, and we chose to allow each eligible individual the opportunity to accumulate Medicare and Medicaid expenditures for at least 6 months following the clinical assessment. For clients with more than one assessment form in the database between July 1995 and June 1997, the first assessment form was used in this study; for some clients, this assessment was the initial assessment that determined waiver program eligibility, and for others, this assessment was an annual assessment conducted for continuing waiver program clients. These selection criteria yielded a sample size of 5,232 unduplicated individuals in the study cohort. Variable Measurement: Dependent Variables Our two goals in constructing dependent variables were to (a) classify each cohort member as either using or not using Medicare and Medicaid home health and Medicaid waiver services and (b) calculate, among service users, expenditure measures with policy relevance. We constructed an observation period for each cohort member over which these Medicare and Medicaid expenditures were identified and accumulated. The start month for the observation period was defined as the first month following the clinical assessment month that the client was eligible for the Medicaid waiver program (nearly always the immediately following month, except when new clients delayed seeking financial eligibility for Medicaid), and the end month was either December 1997 or the last month before December 1997 that the client was eligible for the Medicaid waiver program. Three separate types of services were measured: (a) Medicare home health services, (b) Medicaid home health services, and (c) Medicaid waiver services. Medicare home health services include skilled nursing, home health aide, physical therapy, occupational therapy, speech therapy, and medical social work. Expenditures for these six services cannot be disaggregated in the CLTC database; therefore, we report expenditures for all Medicare home health services combined. Medicaid home health services include skilled nursing and home health aide, but because they cannot be disaggregated in the CLTC database, we report expenditures for these two services combined. Medicaid waiver services include supportive inhome and community-based services covered under Connecticut’s Medicaid HCBS waiver program for the aged. Three services accounted for .80% of Medicaid waiver expenditures in SFY 1996: homemaker (44%), case management (21%), and adult day health (16%); home-delivered meals and

personal emergency response systems accounted for most of the remainder of expenditures (Connecticut Department of Social Services, 1997). These services cannot be disaggregated in the CLTC database; therefore, we report expenditures for all Medicaid waiver services combined. The first set of dependent variables was dichotomous: Cohort members with any identified expenditures of each service type were distinguished from cohort members with no identified expenditures. Among service users, expenditure measures were constructed by dividing total expenditures accumulated by the total number of months in the observation period, thereby adjusting total expenditures for the specific number of months eligible for the Medicaid waiver program. Expenditures are therefore expressed as ‘‘per client per month.’’ This measurement approach was chosen to maximize policy relevance; monthly capitated Medicare rates per client are often negotiated between the Centers for Medicare and Medicaid Services and service providers caring for dually eligible older adults enrolled in the On Lok and PACE programs (GAO, 2000).

Variable Measurement: Independent Variables Functional Disability.—Four measures of functional disability were created based on data from the clinical assessment tool. First, we measured ability to conduct personal care ADL using five ADL from the clinical assessment tool: bathing, dressing, eating, toileting, and transferring. These five items are widely used in published studies to measure personal care disability (e.g., Fortinsky, Covinsky, Palmer, & Landefeld, 1999). We combined the two dependent categories for each item (partially and fully dependent) and then classified each individual according to the number of dependent ADL (zero to five). Second, we measured ability to conduct eight IADLs. Specific IADL items available in the clinical assessment tool were very similar to items used by Lawton and Brody (1969) to measure IADL disability: prepare meals, do housework, do laundry, use telephone, shop for food, travel outside the home, and take oral medications. As with ADL, we combined the two dependent categories for each item and then classified each individual according to the number of dependent IADL (zero to eight). Third, bladder and bowel incontinence items were combined into a separate measure. Cohort members designated as partially or totally incontinent in either function were considered incontinent in that function. We then classified individuals into three categories: continent in both bladder and bowel, incontinent in bowel or bladder function, or incontinent in both. Finally, because lower-body functional disability has been shown to predict a variety of adverse health outcomes in older adults (Guralnik, Ferrucci, 742

Simonsick, Salive, & Wallace, 1995), we created separate single-item measures of dependence in ambulation and in climbing stairs. For both measures, individuals were categorized as either independent or dependent. Cognitive Impairment.—The clinical assessment form includes the 10-item Short Portable Mental Status Questionnaire (MSQ) (Pfeiffer, 1975). Following recommendations by Pfeiffer, five or more errors in response to the MSQ items was considered severely cognitively impaired. Clients with zero to one errors were considered to be cognitively intact. The remainder of the cohort was equally divided between those making two errors and those making three to four errors; we considered the former group mildly cognitively impaired and the latter group moderately cognitively impaired. Additional independent variables were chosen as factors potentially associated with Medicare or Medicaid home care expenditures, as suggested in previous home care studies cited above. Sociodemographic characteristics were gender, age group, and ethnicity. Following the health services utilization model developed by Andersen (1995), these are viewed as predisposing characteristics. Clients were grouped into the following age groups, consistent with many published studies: 65–74, 75–84, and 85 years. Ethnic group categories were based on clinical assessment form data; clients self-identified their ethnic group identification. Categories included White, African American or Black, Hispanic, and other (mostly Asian American). Clients’ living arrangements at the time of the clinical assessment were the best available approximation of their social support system, often viewed as an enabling characteristic in health services research (Andersen, 1995). Clients were classified as living alone, living with spouse only, living with children only, living with spouse and children, or other living arrangements. Finally, we created independent variables for several medical conditions based on Medicare claims data. Medical conditions, defined by conventional and uniform International Classification of Diseases (9th ed.) codes or code groups, were summarized as customized variables in the CLTC database during each calendar year. For this report, medical condition variables were selected if they appeared in the CLTC database during any calendar year(s) corresponding to an individual’s observation period. Physical conditions include stroke or other cerebrovascular accident (CVA), congestive heart failure (CHF), diabetes, pneumonia, pulmonary conditions (chronic obstructive pulmonary disease, bronchitis, or asthma), osteoporosis, and urinary tract infection (UTI). Mental conditions selected because of their clinical importance in the dually eligible population included depression, schizophrenia or other psychosis, and alcohol or drug abuse. The Gerontologist

Analysis Procedures

Table 1. Sample Characteristics (N = 5,232)

The first research question was addressed by employing multivariate logistic regression techniques to estimate the probability of using each service, yielding odds ratios for each independent variable, or risk factor, in each equation. Each odds ratio was adjusted for the effects of all other risk factors; adjusted odds ratios with 95% confidence intervals that did not cross 1.0 (in either direction) were considered statistically significant. To address the second research question, we included only service users and determined the extent to which each of the home- and community-based care expenditure variables was associated with each risk factor. Multivariate ordinary least squares regression modeling techniques were employed; separate regression models were created for each dependent variable. Due to skewed distributions of expenditures, transformations (i.e., log, square root) were calculated and tested in separate regression models. Results based on transformations were comparable with those obtained with untransformed expenditure values; accordingly, we present untransformed values. Parameter estimates are unstandardized regression coefficients, indicating the net excess (or decreased) expenditures per client per month associated with each independent variable. Statistical significance levels were based on two-tailed tests; p values of .05 were considered statistically significant. In all regression models, independent variables representing measures of physical disability and cognitive impairment were constructed using dummy variables and a reference group that was omitted from the model. The reference group was consistently the least functionally disabled and cognitively impaired categories of these variables. Dummy variables were constructed for one to five ADL dependencies; coefficients associated with each dummy represent the difference in expenditures between that number of ADL dependencies and no ADL dependency. For IADL, zero to three dependencies was the reference group; additional dummy variables were constructed for bladder and bowel incontinence, dependence in ambulation, and dependence in climbing stairs. For cognitive impairment, dummy variables were constructed for mild (two errors), moderate (three to four errors), and severe (five or more errors) cognitive impairment. Dummy variables were constructed for each medical condition and entered as independent variables in all regression models.

Characteristic Female Age group 65–74 75–84 85

% 79.1 20.5 41.9 37.6

Ethnic background White African American Hispanic Other

81.0 14.2 3.1 1.7

Living arrangements Alone Spouse only Child(ren) only Spouse and child(ren) Other

55.9 12.1 21.2 1.9 8.9

Number of dependent personal care ADLa 0 1 2 3 4 5

9.8 17.5 43.6 12.7 8.8 7.6

Number of dependent instrumental ADLb 0 1 2 3 4 5 6 7 8

0 0.1 0.6 1.8 5.7 14.4 22.6 32.4 22.6

Dependent in walking

19.4

Dependent in climbing stairs

77.3

Cognitive impairment (no. MSQ errors)c 0–1 2 3–4 5

61.4 11.1 10.6 16.8

Continence Continent of bladder and bowel Incontinent of bladder or bowel Incontinent of bladder and bowel

66.8 20.4 12.9

Results

Notes: ADL = activities of daily living; MSQ = Mental Status Questionnaire. a Personal care ADL: bathing, dressing, eating, toileting, and transferring. b Instrumental ADL: preparing meals, doing housework, doing laundry, using telephone, managing money, shopping for food, traveling outside the home, and taking oral medications. c Mental status questions: today’s date, current month and year, street address, town, age, month born, year born, presi-

Sociodemographic, functional disability, and cognitive impairment characteristics of sample members (N = 5,232) are summarized in Table 1. Nearly 80% were female, 14% were African American, and more than one-third were 85 years or older (mean 6 SD age = 82 6 8 years). Functional disability was wide-

spread; nearly 30% were dependent in three or more ADL, 56% were dependent in seven or eight IADL, and 77% were dependent in climbing stairs. Cognitive impairment was evident in nearly 40% of the sample, including 17% with severe cognitive impairment.

Vol. 44, No. 6, 2004

743

Table 2. Prevalence of Physical and Mental Health Diagnoses in Study Samplea (N = 5,232) Diagnoses

Table 3. Home- and Community-Based Care Expenditures per Client per Month (N = 5,232) Type of Expenditure

%

Physical health diagnoses CVA, stroke Congestive heart failure Urinary tract infection Pulmonary conditions Pneumonia Diabetes Osteoporosis

66.6 52.8 52.4 44.5 44.3 44.2 25.7

Mental health diagnoses Depression Schizophrenia, other psychosis Alcohol drug abuse

35.6 20.2 5.0

Medicare home health Medicaid home health Medicaid waiver

Notes: CVA = cerebrovascular accident. a Diagnoses appearing on Medicare Part A hospital or Part B physician claims during calendar years associated with each sample member’s observation period.

Many medical conditions were common (Table 2); two-thirds of sample members had a diagnosis of stroke or CVA, about one-half of sample members had a diagnosis of CHF or UTI, .40% had diagnoses of pulmonary disease, pneumonia, or diabetes, and 26% had osteoporosis. Mental health conditions were prevalent in a considerable minority of the cohort; 36% had diagnosed depression, 20% had a diagnosed psychosis, and 5% had a substance abuse diagnosis. Patterns of Medicare and Medicaid Home- and Community-Based Care Expenditures Table 3 shows the proportion of the sample that generated expenditures for each of the home- and community-based services as well as the distribution of expenditures among users of each service. The mean length of the observation period was 14.9 months (SD = 7.7 months, median = 16 months). More than 70% of sample members used Medicare home health care during their observation periods, 76% used Medicaid home health care, and nearly all clients used Medicaid waiver services. Among service users, mean expenditures per client per month were $552 for Medicare home health, $364 for Medicaid home health, and $666 for Medicaid waiver services. Regression Results Featuring Medicare Home Health Use and Expenditures Table 4 summarizes results of the logistic regression analysis conducted to determine factors associated with the probability of using any Medicare home health services. Results show that increased dependence in personal care ADL, dependence in walking, and dependence in climbing stairs were all risk factors for using Medicare home health care in the study cohort. Compared with those with no ADL dependency, those with four or 744

Mean $

SD $

Median $

% with Expenditures

552 364 666

799 432 442

208 220 570

70.6 76.1 99.0

five ADL dependencies were more than twice as likely to have used Medicare home health care; even a single ADL dependency increased the probability of using Medicare home health care by 68%. Individuals with severe cognitive impairment were 78% as likely as those with no cognitive impairment to use Medicare home health care. Several medical conditions were associated with the probability of using Medicare home health care, including (in descending order by size of adjusted odds ratio) alcohol or substance abuse, CHF, diabetes, pulmonary disease, pneumonia, UTI, osteoporosis, and stroke. Among sociodemographic factors, sample members living with relatives other than spouses or adult children were less likely than those living alone to use Medicare home health services, and increased age was associated with a greater probability of using Medicare home health services. Table 5 shows factors associated with the amount of Medicare home health expenditures per month among those who used Medicare home health care during their observation periods (71% of the study cohort). The number of personal care ADL dependencies showed a statistically significant, linear, and positive association with Medicare home health expenditures. Dependence in walking and climbing stairs also led to statistically significantly greater Medicare home health expenditures, as did bladder and bowel incontinence compared with continent clients. Among medical conditions, only UTI, CHF, and diabetes were associated with greater Medicare home health expenditures per month. Cognitive impairment severity was not statistically significantly associated with Medicare home health expenditures. Among sociodemographic characteristics, living with others was associated with lower Medicare home health expenditures than living alone (the reference group not in the table), as indicated by negative parameter estimates for living arrangements. All independent variables combined explained 15% of the variance in Medicare home health expenditures. Regression Results Featuring Medicaid Home Health Use and Expenditures Table 6 summarizes results of the logistic regression analysis conducted to determine factors associated with the probability of using any Medicaid home health services. As with Medicare home health services, the number of personal care ADL The Gerontologist

Table 4. Logistic Regression Results: Likelihood of Medicare Home Health Use (n = 5,052)

Independent Variable Intercept Mild cognitive impairment Moderate cognitive impairment Severe cognitive impairment Dependent in walking Dependent in climbing 1 ADL dependency 2 ADL dependencies 3 ADL dependencies 4 ADL dependencies 5 ADL dependencies Incontinent bowel or bladder Incontinent bowel and bladder 4 IADL dependencies 5 IADL dependencies 6 IADL dependencies 7 IADL dependencies 8 IADL dependencies Diabetes Alcohol/substance abuse Schizophrenia/psychosis Depression CHF Pulmonary disease CVA, stroke Pneumonia UTI Osteoporosis Lives with spouse Lives with child(ren) Lives with spouse and child(ren) Lives with other relatives Male African American Hispanic or other Age group 75 and ,85 Age group 85

Table 5. Multivariate Regression Results: Correlates of Medicare Home Health Expendituresa (n = 3,559)

95% Adjusted Confidence Odds Interval Significance Ratio 0.28

—

1.15

0.92–1.44

0.21

0.81

0.65–1.01

0.06

0.78 1.58 1.26 1.68 1.94 1.79 2.23 2.04

0.62–0.98 1.23–2.03 1.08–1.47 1.31–2.15 1.55–2.43 1.34–2.38 1.53–3.24 1.29–3.22

0.03 ,0.01 ,0.01 ,0.01 ,0.01 ,0.01 ,0.01 ,0.01

1.00

0.84–1.18

0.96

0.96 0.91 1.07 1.08 0.90 0.71 1.63

0.74–1.23 0.56–1.47 0.69–1.67 0.70–1.67 0.58–1.4 0.44–1.14 1.41–1.87

0.73 0.7 0.77 0.74 0.65 0.16 ,0.01

2.08 1.03 1.15 1.67 1.55 1.32 1.47 1.45 1.41 1.15 0.84

1.45–2.98 0.86–1.22 1.00–1.33 1.45–1.91 1.35-1.79 1.15–1.52 1.27–1.69 1.27–1.66 1.20–1.66 0.91–1.45 0.69–1.01

,0.01 0.76 0.06 ,0.01 ,0.01 ,0.01 ,0.01 ,0.01 ,0.01 0.23 0.06

1.10 0.74 0.94 1.04 1.07 1.28 1.38

0.65–1.85 0.58–0.94 0.78–1.12 0.86–1.25 0.78–1.46 1.07–1.53 1.14–1.67

0.72 0.01 0.46 0.70 0.67 0.01 ,0.01

,0.01

Notes: Cox and Snell R2 = 0.12; Nagelkerke R2 = 0.18. ADL = activities of daily living; IADL = instrumental activities of daily living; CHF = congestive heart failure; CVA = cerebrovascular accident; UTI = urinary tract infection.

dependencies was statistically significantly associated with the probability of using Medicaid home health services. The pattern of results reveals that those with three ADL dependencies had the greatest likelihood of using Medicaid home health services, but that even one ADL dependency increased the odds by 66%. Cognitive impairment severity was found to have no statistically significant association with Medicaid home health expenditures. Stroke was the only medical condition associated with a greater Vol. 44, No. 6, 2004

745


Parameter Estimate $

Standard Error

Significance

89

102

0.38

101

41

0.02

64

46

0.17

17 197 143 98 250 277 498 713

47 45 34 56 52 63 76 91

0.72 ,0.01 ,0.01 0.08 ,0.01 ,0.01 ,0.01 ,0.01

1

33

0.97

249 37 89 7 49 4 97 23 5 45 100 48 59 46 155 19 76 147

49 104 95 94 94 102 27 56 34 28 28 28 29 28 27 29 44 37

,0.01 0.72 0.35 0.94 0.6 0.97 ,0.01 0.68 0.89 0.10 ,0.01 0.08 0.04 0.10 ,0.01 0.53 0.08 ,0.01

234 69 9 17 71 6 33

101 51 35 39 61 35 39

0.02 0.17 0.81 0.67 0.25 0.87 0.39

Notes: Adjusted R2 = 0.15. ADL = activities of daily living; IADL = instrumental activities of daily living; CHF = congestive heart failure; CVA = cerebrovascular accident; UTI = urinary tract infection. a Expenditures expressed as per client per month.

probability of using Medicaid home health services, whereas depression, pulmonary disease, and UTI were associated with a lower probability of using Medicaid home health services. Among sociodemographic variables, sample members living with spouses and children were more than twice as likely as those living alone to use Medicaid home health services, Blacks were 63% more likely than Whites, and those aged 85 and older were 32% more likely than those aged 65–74.

Table 6. Logistic Regression Results: Likelihood of Medicaid Home Health Use (n = 5,054)

Independent Variable

Table 7. Multivariate Regression Results: Correlates of Medicaid Home Health Expendituresa (n = 3,863)

95% Adjusted Confidence Odds Interval Significance Ratio

Intercept Mild cognitive impairment Moderate cognitive impairment Severe cognitive impairment Dependent in walking Dependent in climbing 1 ADL dependency 2 ADL dependencies 3 ADL dependencies 4 ADL dependencies 5 ADL dependencies Incontinent bowel or bladder Incontinent bowel and bladder 4 IADL dependencies 5 IADL dependencies 6 IADL dependencies 7 IADL dependencies 8 IADL dependencies Diabetes Alcohol/substance abuse Schizophrenia/psychosis Depression CHF Pulmonary disease CVA, stroke Pneumonia UTI Osteoporosis Lives with spouse Lives with child(ren) Lives with spouse and child(ren) Lives with other relatives Male African American Hispanic or other Age group 75 and ,85 Age group 85

3.54

Intercept Mild cognitive impairment Moderate cognitive impairment Severe cognitive impairment Dependent in walking Dependent in climbing 1 ADL dependency 2 ADL dependencies 3 ADL dependencies 4 ADL dependencies 5 ADL dependencies Incontinent bowel or bladder Incontinent bowel and bladder 4 IADL dependencies 5 IADL dependencies 6 IADL dependencies 7 IADL dependencies 8 IADL dependencies Diabetes Alcohol/substance abuse Schizophrenia/psychosis Depression CHF Pulmonary disease CVA, stroke Pneumonia UTI Osteoporosis Lives with spouse Lives with child(ren) Lives with spouse and child(ren) Lives with other relatives Male African American Hispanic or other Age group 75 and ,85 Age group 85

,0.01

0.94

0.76–1.17

0.60

1.02

0.81–1.29

0.85

0.89 0.97 0.90 1.66 1.73 1.87 1.46 1.62

0.70–1.11 0.76–1.23 0.76–1.07 1.28–2.15 1.37–2.19 1.38–2.52 1.02–2.11 1.04–2.53

0.30 0.78 0.23 ,0.01 ,0.01 ,0.01 0.04 0.03

1.16

0.97–1.38

0.10

1.18 0.78 0.75 0.74 0.52 0.45 1.08 0.89 0.98 0.75 0.99 0.82 1.22 1.05 0.85 0.95 0.96 0.96

0.92–1.51 0.46–1.34 0.46–1.23 0.45–1.21 0.32–0.85 0.27–0.76 0.94–1.25 0.67–1.2 0.82–1.16 0.65–0.87 0.85–1.14 0.71–0.95 1.06–1.41 0.91–1.22 0.74–0.98 0.82–1.11 0.76–1.2 0.79–1.16

0.20 0.36 0.26 0.23 0.01 ,0.01 0.27 0.46 0.79 ,0.01 0.84 0.01 0.01 0.47 0.02 0.54 0.70 0.67

2.32 0.74 0.95 1.63 1.36 1.04 1.32

1.24–4.37 0.58–0.94 0.79–1.13 1.31–2.02 0.98–1.89 0.87–1.24 1.08–1.6

0.01 0.01 0.55 ,0.01 0.06 0.68 0.01

Independent Variable

Standard Error

Significance

164

46

,0.01

15

21

0.46

19

22

0.39

21 60 26 78 213 279 360 562

22 23 16 26 24 29 37 45

0.35 0.01 0.09 ,0.01 ,0.01 ,0.01 ,0.01 ,0.01

48

16

,0.01

45 27 82 70 132 114 6 8 7 11 28 42 14 32 39 0 122 29

24 47 43 43 43 47 13 30 17 14 13 14 14 14 13 15 22 18

0.07 0.56 0.06 0.10 ,0.01 0.02 0.67 0.79 0.66 0.41 0.04 ,0.01 0.30 0.02 ,0.01 0.99 ,0.01 0.11

45 60 9 116 79 14 30

47 25 17 18 30 17 19

0.33 0.01 0.60 ,0.01 0.01 0.42 0.10

Parameter Estimatea

Notes: Cox and Snell R = 0.03; Nagelkerke R = 0.05. ADL = activities of daily living; IADL = instrumental activities of daily living; CHF = congestive heart failure; CVA = cerebrovascular accident; UTI = urinary tract infection.

Notes: Adjusted R2 = 0.18. ADL = activities of daily living; IADL = instrumental activities of daily living; CHF = congestive heart failure; CVA = cerebrovascular accident; UTI = urinary tract infection. a Expenditures expressed as per client per month.

Table 7 shows factors associated with the amount of Medicaid home health expenditures per month among those who used Medicaid home health care during their observation periods (76% of the study cohort). The number of personal care ADL dependencies showed a statistically significant, linear, and positive association with Medicaid home health expenditures. Those with seven or eight IADL dependencies were more costly than those with no IADL dependencies, but the magnitude was not as

great as with ADL dependencies. Level of cognitive impairment was not associated with Medicaid home health expenditures. Among sociodemographic characteristics, living with a spouse or with relatives other than a spouse or adult children was associated with lower Medicaid home health expenditures compared with living alone. Black and Hispanic sample members generated greater Medicaid home health expenditures than their White counterparts. All independent variables combined explained 11% of the variance in Medicaid home health expenditures.

2

2

746

The Gerontologist

Regression Results Featuring Medicaid Waiver Service Expenditures

Table 8. Multivariate Regression Results: Correlates of Medicaid Waiver Expenditurea (n = 5,054)

Nearly all sample members used Medicaid waiver services (Table 3); therefore, results in this section are restricted to factors associated with greater or fewer Medicaid waiver service expenditures per month. As Table 8 shows, cognitive impairment severity was statistically significantly associated with Medicaid waiver services in a positive and linear fashion. Functional disability measures most associated with Medicaid waiver service expenditures were IADL dependencies, with expenditures per month showing a linear increase with each additional IADL dependency. Conversely, the most disabled sample members as measured by personal care ADL generated less Medicaid waiver service expenditures than those with no ADL dependencies. Among medical conditions, only stroke was associated with greater Medicaid waiver service expenditures. Among sociodemographic variables, negative parameter estimates for living arrangement variables indicate that clients who lived alone were more costly than clients in any of these living arrangements. Males were less costly than females by $47 per month. All independent variables combined explained 5% of the variance in Medicaid waiver expenditures.

Discussion This article addressed two major questions in a dually eligible sample enrolled in Connecticut’s Medicaid HCBS waiver program for the aged: (a) What risk factors are associated with the likelihood of using Medicare home health and Medicaid home health services? (b) Among those who use services, what are the most important correlates of Medicare home health care expenditures, Medicaid home health expenditures, and Medicaid waiver service expenditures? Several findings are emphasized. First, personal care ADL dependencies were the most consistent factors associated with the probability of using, and with the amount of expenditures per month for, both Medicare and Medicaid home health services. The greater the number of ADL dependencies, the greater the probability of using service, and the greater the expenditures per month among service users. Second, cognitive impairment severity was positively associated with Medicaid waiver service expenditures but was unrelated to home health service use funded by either Medicare or Medicaid. Third, most medical conditions increased the probability of using Medicare home health care; only a few (UTI, CHF, diabetes) were associated with greater Medicare home health expenditures per month, and only stroke was associated with a greater probability of using Medicaid home health services and with greater Medicaid waiver service expenditures. Vol. 44, No. 6, 2004

747


Parameter Estimate $

Standard Error

Significance

603

44

,0.01

42

20

0.03

49

21

0.02

113 11 54 25 6 23 38 104

22 22 15 24 22 28 35 42

,0.01 0.60 ,0.01 0.31 0.79 0.41 0.28 0.01

5

16

0.73

6 63 108 136 194 240 5 36 22 5 48 22 29 28 18 3 137 34

23 46 43 42 42 46 13 28 16 13 13 13 13 13 13 14 21 17

0.81 0.18 0.01 ,0.01 ,0.01 ,0.01 0.70 0.20 0.16 0.73 ,0.01 0.09 0.03 0.03 0.17 0.85 ,0.01 0.05

73 80 47 10 27 6 5

48 23 16 18 29 16 18

0.12 ,0.01 ,0.01 0.58 0.35 0.74 0.79

Notes: R2 = .05. ADL = activities of daily living; IADL = instrumental activities of daily living; CHF = congestive heart failure; CVA = cerebrovascular accident; UTI = urinary tract infection. a Expenditures expressed as per client per month.

Finally, clients who lived alone showed a fairly consistent pattern of generating greater expenditures for all types of services studied compared with clients living with family members. Several important limitations of this study are acknowledged. First, it was not possible to disaggregate specific types of service expenditures within the Medicare home health, Medicaid home health, or Medicaid waiver service groupings. This restriction in the study database prevented a more penetrating analysis into whether, for example, Medicare home

health expenditures were more likely to be for skilled nursing services or whether Medicaid home health expenditures were more likely to be for home health aide services. Second, it was not possible to assign temporal interpretation to results involving medical conditions because these conditions were based on claims data extracted on an annual basis only. Consequently, results related to medical conditions should be interpreted as associations, not as cause and effect, in relation to service use and expenditures. Third, the reliability of clinical assessment data used in this study could not be formally evaluated. Fourth, study data are from only one state and from a time period when Medicare home health service reimbursement was cost based (until October 1997) and then subject to a more restrictive interim payment system under the Balanced Budget Act of 1997. Today, Medicare home health services are reimbursed under a prospective payment system; therefore, a more current linked database constructed to study similar questions might yield different results. Finally, study results cannot be generalized to other community-dwelling dually eligible individuals such as individuals younger than age 65 and adults aged 65 and older not enrolled in Medicaid HCBS waiver programs. Despite these limitations, study results yield several implications that could inform recent discussions about how to design more integrated Medicare and Medicaid initiatives for frail dually eligible older adults (Capitman, 2003; Miller & Weissert, 2003). First, efforts by policy makers to design Medicare and Medicaid service-integration initiatives might begin with dually eligible individuals enrolled in Medicaid HCBS waiver programs for the aged, because this study found that .70% of these clients used Medicare home health and Medicaid home health services in addition to Medicaid waiver services. Results from other states could help refine insights about how to improve efficiencies between Medicare- and Medicaid-reimbursed providers serving the older dually eligible population enrolled in Medicaid waiver programs. This would be a first step toward considering how to improve Medicare and Medicaid home health care integration for all dually eligible older adults living in the community. Second, state reimbursement strategies for Medicaid-funded services for older Medicaid HCBS waiver clients could be informed by study findings. For example, measures of personal ADL dependence might be factored into a case mix payment system for Medicaid home health services in a state Medicaid HCBS waiver program for the aged. It is worth noting that ADL dependence is incorporated into the Medicare prospective payment system for home health care, implemented in October 2000. Third, based on the low amount of variance explained in the Medicaid waiver model (5%), few risk factors are associated with expenditures for the 748

bundle of services covered in Connecticut’s Medicaid waiver package. A likely explanation for the low amount of variance explained is that all waiver services are bundled into a single dependent variable; each specific service might indeed have a different set of risk factors. Nevertheless, moderate to severe cognitive impairment and extensive dependence in IADL are risk factors for greater total waiver service expenditures. Thus, these types of services may be used more often to supplement the daily care of cognitively impaired clients already provided by family members or to sustain at home cognitively impaired clients with little or no family support system. Further evidence in other states might also show that expenditures for supportive services offered in their Medicaid waiver packages are greatest for clients with more progressive cognitive impairment and extensive IADL dependence. Finally, study sample members living alone are at greatest risk for high expenditures of all types of home- and community-based services examined in this study, even after controlling for many other important covariates. Medicare and Medicaid service integration would be especially useful for dually eligible individuals living alone because they are less likely to have personal advocates to help navigate both public programs. In conclusion, this study provides evidence-based insights into the important policy issue of Medicare and Medicaid integration. Similar empirical research from other states will help policy makers design strategies for integrating Medicare and Medicaid services for the growing numbers of dually eligible older adults who are projected to use both Medicare and Medicaid home- and community-based services in the next several decades (Clark & Hulbert, 1998; Kemper, 2003; Knickman & Snell, 2002; Miller & Weissert, 2003).

References Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care: Does it matter? Journal of Health and Social Behavior, 36, 1–10. Bectel, R. W., & Graves, N. G. (1998). Across the states 1998: Profiles of long-term care systems. Washington, DC: AARP Public Policy Institute. Bodenheimer, T. (1999). Long-term care for frail elderly people—the On Lok model. New England Journal of Medicine, 341, 1324–1328. Boyd, D. J. (2003). The bursting state fiscal bubble and state Medicaid budgets. Health Affairs, 22, 46–61. Branch, L., Wetle, T. T., Scherr, P. A., Cook, N. R., Evans, D. A., Hebert, L. E., et al. (1988). A prospective study of incident comprehensive medical home care use among the elderly. American Journal of Public Health, 78, 255–259. Capitman, J. (2003). Effective coordination of medical and supportive services. Journal of Aging and Health, 15, 124–164. Clark, W. D., & Hulbert, M. M. (1998). Research issues: Dually eligible Medicare and Medicaid beneficiaries, challenges, and opportunities. Health Care Financing Review, 20, 1–10. Connecticut Department of Social Services. (1997). Annual report on the Connecticut Home Care Program for Elders, July 1, 1995, through June 30, 1996. Hartford, CT: Office of the Commissioner. Costa, P. T., Williams, T. F., Albert, M. S., Butters, N. M., Folstein, M. F., Gilman, S., et al. (1996). Recognition and initial assessment of Alzheimer’s disease & related dementias. AHCPR publication

The Gerontologist

97-0702. Rockville, MD: Agency for Health Care Policy and Research, U.S. Department of Health & Human Services. Coughlin, T. A., McBride, T. D., Perozek, M., & Liu, K. (1992). Home care for the disabled elderly: Predictors and expected expenditures. Health Services Research, 27, 453–479. Eng, C., Pedulla, J., Eleazer, G. P., McCann, R., & Fox, N. (1997). Program of All-Inclusive Care for the Elderly (PACE): An innovative model of integrated geriatric care and financing. Journal of the American Geriatrics Society, 45, 223–232. Evans, D. A., Scherr, P. A., Cook, N. R., Albert, M. S., Funkenstein, H. H., Beckett, L. A., et al. (1992). The impact of Alzheimer’s disease in the United States population. In R. M. Suzman, D. P. Willis, & K. G. Manton (Eds.), The oldest old (pp. 283–302). New York: Oxford University Press. Fogel, B. S., Brock, D., Goldscheider, F., & Royall, D. (1994). Cognitive dysfunction and the need for long-term care: Implications for public policy. Public Policy Institute (report 9309). Washington, DC: American Association of Retired Persons. Fortinsky, R. H., Covinsky, K. E., Palmer, R. M., & Landefeld, C. S. (1999). Effect of functional status changes before and during hospitalization on nursing home admission of older adults. Journal of Gerontology: Medical Sciences, 54A, M521–M526. Fortinsky, R. H., Granger, C. V., & Seltzer, G. B. (1981). The use of functional assessment in understanding home care needs. Medical Care, 19, 489–497. Fried, T. R., Bradley, E. H., Williams, S., & Tinetti, M. E. (2001). Functional disability and health care expenditures for older persons. Archives of Internal Medicine, 161, 2602–2607. Guralnik, J. M., Ferrucci, L., Simonsick, E. M., Salive, M. E., & Wallace, R. B. (1995). Lower extremity function in persons over the age of 70 as a predictor of subsequent disability. New England Journal of Medicine, 332, 556–561. Harrington, C., Carrillo, H., Wellin, V., Miller, N., & LeBlanc, A. (2000). Predicting state Medicaid home and community-based waiver participants and expenditures, 1992–1997. The Gerontologist, 40, 673–686. Kemper, P. (1992). The use of formal and informal care by the disabled elderly. Health Services Research, 27, 421–451. Kemper, P. (2003). Long-term care research and policy. The Gerontologist, 43, 436–446. Knickman, J. R., & Snell, E. K. (2002). The 2030 problem: Caring for aging baby boomers. Health Services Research, 37, 849–884. Lamphere, J., Brangan, N., Bee, S., & Semansky, R. (1998). Reforming the health care system: State profiles 1998. Washington, DC: AARP Public Policy Institute. Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Selfmaintaining and instrumental activities of daily living. The Gerontologist, 9, 179–186. Madigan, E. A., Tullai-McGuinness, S., & Neff, D. F. (2002). Home health services research. Annual Review of Nursing Research, 20, 267–291.

Vol. 44, No. 6, 2004

749

Miller, E. A., & Weissert, W. G. (2003). Strategies for integrating Medicare and Medicaid: Design features and incentives. Medical Care Research & Review, 60, 123–157. Newcomer, R., Clay, T., Luxenberg, R. S., & Miller, R. H. (1999). Misclassification and selection bias when identifying Alzheimer’s disease solely from Medicare claims records. Journal of the American Geriatrics Society, 47, 215–219. O’Keeffe, J. (1996). Determining the need for long-term care services: An analysis of health and functional eligibility criteria in Medicaid home and community-based waiver programs. Public Policy Institute (report 9617). Washington, DC: American Association of Retired Persons. O’Keeffe, J. (1999). People with dementia: Can they meet Medicaid levelof-care criteria for admission to nursing homes and home and community-based waiver programs? Public Policy Institute (report 9912), Washington, DC: American Association of Retired Persons. Pfeiffer, E. (1975). A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. Journal of the American Geriatrics Society, 23, 433–441. Saucier, P., Bezanson, L., Booth, M., Bratesman, S., Fralich, J. T., Gilden, D., et al. (1998). Linked data analysis of dually eligible beneficiaries in New England. Health Care Financing Review, 20, 91–108. Small, G. W., Rabins, P. V., Barry, P. P., Buckholtz, N. S., DeKosky, S. T., Ferris, S. H., et al. (1997). Diagnosis and treatment of Alzheimer disease and related disorders: Consensus statement of the American Association for Geriatric Psychiatry, the Alzheimer’s Association, and the American Geriatrics Society. Journal of the American Medical Association, 278, 1363–1371. Solomon, D. H., Wagner, R., Maremberg, M. E., Acampora, D., Cooney, L. M., & Inouye, S. K. (1993). Predictors of formal home health care use in elderly patients after hospitalization. Journal of the American Geriatrics Society, 41, 961–966. Taylor, D. H., & Sloan, F. A. (2000). How much do persons with Alzheimer’s disease cost Medicare? Journal of the American Geriatrics Society, 48, 639–646. U.S. General Accounting Office (2000). Medicare and Medicaid: Implementing state demonstrations for dual eligibles has proven challenging. GAO/HEHS-00-94. Washington, DC: Author. Weiner, M., Powe, N. R., Weller, W. E., Shaffer, T. J., & Anderson, G. F. (1998). Alzheimer’s disease under managed care: Implications from Medicare utilization and expenditure patterns. Journal of the American Geriatrics Society, 46, 762–770.

Received February 12, 2003 Accepted January 27, 2004 Decision Editor: Linda S. Noelker, PhD