| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
Correspondence: Address correspondence to Gretchen E. Alkema, MSW, LCSW, Leonard Davis School of Gerontology, University of Southern California, 3715 McClintock Ave., Los Angeles, CA 90089-0191. E-mail: alkema{at}usc.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Home- and community-based services • Medicare managed care • Service utilization • Case management • Chronic care
Given the need to improve service planning and capacity, it is important for policy makers and practitioners to understand what types of services older adults actually use when they are offered a choice, and what consumer characteristics are associated with the use of specific services. Previous studies of risk factors for service utilization have focused on state plan or Medicaid-waiver services, primarily to discuss the public costs of these services and their potential to reduce nursing facility admissions (Borrayo, Salmon, Polivka, & Dunlop, 2002; Fitzner, Bennett, Weinraub, & Heckinger, 2002; Shugarman, Buttar, Fries, Moore, & Blaum, 2002). Studies, however, have not examined individual behavior in selecting and utilizing services for those consumers who have some financial responsibility for the purchase of services. To begin to address this gap, we examined individual characteristics associated with use in six areas: overall service use, four specific categories of HCBS, and referrals to insured medical services. The sample for the study was composed of high-risk Medicare managed care consumers who participated in the Care Advocate Demonstration Program—a demonstration developed to bridge medical and social care systems by offering those in Medicare managed care access to HCBS.
Background
Previous research on individual predictors of HCBS utilization has showed varied results, depending on the sample and program auspice. Some researchers have found that users of services include people with chronic diseases and need-based characteristics, such as functional decline (Borrayo et al., 2002). However, those eligible for Medicaid were higher users of HCBS regardless of their need characteristics (Borrayo et al.). This phenomenon might have occurred because Medicaid was the payer for many service programs, suggesting evidence for a woodwork effect, or because Medicaid eligibility is a proxy for unmeasured need. After controlling for social support and functional status in a Medicaid population, researchers have not shown age and gender to be significant predictors of utilization patterns (Benjamin, 2001). Others studying residents of continuing care retirement communities (CCRC) have found that social support, functional decline, and marital status have a significant impact on service use (Krout, Oggins, & Holmes, 2000). However, these results have generalizability only to CCRC residents, who tend to be upper income individuals who have made a conscious choice to live in a community saturated with age-based services.
Fundamental questions for any social service delivery system are who gets services, what kind of services do they get, who pays for the service, and who monitors the quality of the service (Branch, 2001). Each HCBS program has its own criteria for access, service parameters, costs, and quality assurance, whether it is a county adult protective services program safeguarding vulnerable adults from victimization or a hospital-sponsored bereavement support group. Dually eligible older adults can receive some state plan or waiver HCBS with a minimal or no copayment, depending on the service's geographic and programmatic accessibility. In contrast, for Medicare managed care consumers who desire HCBS support, program costs are a significant issue as the consumer is usually the responsible party for any associated fees.
Although Medicaid-eligible consumers who use HCBS must address the psychological issue of accepting help from formal care networks, Medicare managed care consumers face additional decisions about how and when to pay for services designed to support community-based living. It is in this situation that models of care built from a consumer-choice philosophy promote and enhance self-efficacy as older adults seek practical care alternatives to institutionalization (Benjamin, Matthias, & Franke, 2000; Meiners, Mahoney, Shoop, & Squillace, 2002).
There is limited research on predictors of HCBS use in general, and we are not aware of any studies that have examined characteristics associated with noninsured HCBS utilization by older Medicare managed care health plan enrollees. We address this gap by identifying individual characteristics associated with HCBS utilization for Medicare managed care consumers who participated in a social case management demonstration project: the Care Advocate Demonstration Program. For this purpose, we define HCBS as those services based in the home or community that are not covered as part of an insured benefit from a health plan. Building on the literature, we expected that higher utilization of HCBS for this specialized population would be found in (a) older participants; (b) women; (c) individuals who lived alone; (d) those without social support; (e) those with higher levels of functional impairment; and (f) those with greater levels of health service utilization. After controlling for these other factors, we believed that age and gender would not be significant predictors of utilization. We also, by way of comparison, examined characteristics of those in social care management who were referred back to medical services.
The Care Advocate Demonstration Program
The Care Advocate Demonstration Program was designed to build a collaborative relationship between the disparate worlds of medical and social service delivery. Funded by the California HealthCare Foundation through the Programs for Elders in Managed Care, the Care Advocate Demonstration Program offered a social case management intervention to frail older adults enrolled in a Medicare managed care health plan (Wilber, Allen, Shannon, & Alongi, 2003; Wilber & Shannon, 2003). Institutional review board approval was granted for all health and social service partners. In addition, the University of Southern California Institutional Review Board approved the study.
The Care Advocate Demonstration Program used a brokerage model of case management, in which case managers individualized referrals within a service delivery network that, for the most part, lacked formal coordination of providers (Scharlach, Giunta, & Mills-Dick, 2001). This process tended to introduce much greater variation than slot-based waiver programs, which utilize formal service contracts and purchase of service arrangements. The underlying philosophy of the Care Advocate Demonstration Program was to offer professional support for informed consumer choice. In keeping with this philosophy, care advocates made recommendations by serving in the roles of educator, consultant, and coach rather than gatekeeper or direct implementer (Kunkel, Duffy-Durham, & Scala, 2000). Literature on the philosophy of consumer choice for older adults supports the use of professionals to help with identifying and assisting with service needs, but maintains that the consumer is responsible for decisions about when, if, and how he or she wants to use services (Coleman, 2001; Kane, 2001; Squillace & Firman, 2002).
The program used master's level social work case managers, called care advocates, located in two community-based social service agencies to help consumers identify and access relevant HCBS based on the participants' individualized needs. The social service agencies had no prior or ongoing financial relationship with the health plan or medical group beyond the scope of this externally funded demonstration. Care advocates completed an 83-item psychosocial and functional assessment with participants to identify immediate as well as more long-term service needs. They used this assessment to discuss options and offer participants HCBS information that was based on their individual needs and preferences. When participants requested additional assistance, care advocates helped make contacts to providers, complete necessary forms, and advocate for services. Care advocates called participants within 1 week of assessment, and they made monthly follow-up calls to monitor progress, offer support and coaching, and provide additional HCBS information. During this 1-year intervention, care advocates encouraged participants to contact them if they had questions, crisis issues, or needed further consultation. Upon completion of the intervention phase, participants received additional community referrals to ensure an ethical termination of the care advocate–participant relationship. Care advocates engaged in several practices to standardize the intervention process across the social service agency sites. They used a uniform assessment tool, developed standardized protocols that guided the referral processes, and participated in monthly care coordination meetings and weekly dialogue to maintain the efficacy of the standardized treatment protocol.
Core elements of the Care Advocate Demonstration Program were emphasizing individualized assessment and referral, and applying the philosophy of consumer choice in the care advocate–participant interactions. Care advocates assessed strengths and challenges of participants' daily living situation within their psychosocial, functional, health, and environmental context. Referral information and direct linkages to services were based on a variety of factors, including (a) participants' individual needs identified from the assessment; (b) their willingness to accept and often pay for services; and (c) service availability based on eligibility criteria and geographic proximity that are common to every community.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Drawing from a population of approximately 30,000 health plan members in the four medical groups, we found that over 5,800 elders met the aforementioned criteria. From this base, we randomly selected participants in the original study to participate in the care advocate intervention to reach a sample size of at least 800 (N = 823). We then randomly assigned participants to an intent-to-treat (n = 389) or control group (n = 434). One hundred eighteen people refused participation, died, or left the health plan from the intent-to-treat group, leaving 271 participants in this intervention group.
In the present study, we used data only from those in the intervention group who were assessed by care advocates, as the service data required for the analysis were only available for these participants. Care advocates assessed 271 participants; after assessment but before any referrals were made, 4 members died and 16 members left the health plan. Of the 251 members who received the information and referral intervention, we excluded 22 participants because they had Medi-Cal eligibility (given their access to services paid for by this funding source). We excluded an additional 5 participants who were screened into the original study solely on the basis of advanced age (aged 85 and older). Previous analyses indicated that these individuals had low service use. Both exclusions resulted in an analytical sample of 224. Figure 1 describes how the analytic sample was achieved.
|
Using the Katz index of independence in activities of daily living (ADLs; Katz, Ford, Moskovitz, Jackson, & Jaffe, 1963), we measured participants' functional level by using binary variables to denote difficulty for each activity of daily living (bathing, dressing, grooming, incontinence, toileting, or transferring). We used Lawton's instrumental activities of daily living (IADLs; Lawton & Brody, 1969) to measure participants' difficulty with cleaning, doing laundry, preparing meals, shopping, and using transportation. We also aggregated the number of ADLs and IADLs in which each participant had difficulty, resulting in a range from 0 to 6 and 0 to 5, respectively.
We measured each participant's health status on the basis of self-reported medical diagnoses or conditions. We first selected individual diseases and conditions by identifying high frequencies for each in the sample population. We then combined these diagnoses with similar diseases into the following categories as dichotomous variables (presence = 1): heart conditions, neurological conditions, respiratory conditions, sensory impairments, and skeletal impairments. Diseases grouped under heart conditions were any heart disease, hypertension, or circulatory problems. Neurological conditions included Alzheimer's disease, dementia, Parkinson's disease, or neurological effects from a stroke. Respiratory conditions included pneumonia, chronic obstructive pulmonary disease, and shortness of breath. Sensory impairments included vision problems (glaucoma, cataracts, and vision impairment requiring corrective lenses) and hearing loss. Skeletal impairments included arthritis, osteoporosis, and previous skeletal injuries. Health service utilization indicated the participant's algorithm score, ranging from 4 to 11 points. However, as a result of small cell sizes, we collapsed scores of 7 and above into a single category.
Participant Characteristics
Table 1 shows that the average age of participants was 81 years. Sixty-six percent were female, 40% were married, and 40% reported living alone. Nearly 80% identified at least one person who provided social support. The average and modal educational attainment was a high school diploma. The three highest percentages of ADL difficulties were transferring (21.4%), bathing (19.2%), and incontinence (17.9%). Overall, participants had high levels of IADL difficulty, with cleaning (54%), shopping (46.9%), and doing laundry (38.4%) ranking as the most prevalent. The average participant had one ADL deficit and two IADL deficiencies. The majority of participants had sensory impairments (89.3%), followed by heart conditions (83.9%) and skeletal impairments (70.1%). The average health service utilization algorithm score was just over 5 points.
|
|
We also observed the referrals back to the participants' medical group through the primary care physician under the medical services category, which was used by 31.7% of participants. Although these medical services were insured by the health plan, we kept this referral category in the analysis for two complementary reasons. First, medical services referrals are a key element in the delivery of chronic care services—the direct link back to health care services—supporting the connection between medical and social services. Second, the inclusion of medical services is useful to compare with HCBS utilization, as both represent different types of service needs. Using cross-tabulation, we found that 23% of participants used both a HCBS and a medical service referral, whereas 40% used only a HCBS and 9% used only medical services (
2 = 3.86; p =.049).
Analysis
We used the Statistical Package for the Social Sciences (SPSS) to analyze cross-sectional bivariate and multivariate data for the 224 participants. We used logistic regression for examining the likelihood of utilizing any HCBS, individual categories of HCBS, and medical services with the following independent variables: sociodemographic characteristics, ADL difficulties, IADL difficulties, diseases and conditions, and health service utilization algorithm score. Marital status correlated highly with living situation (r =.581; p <.001), and therefore we removed it from the analyses because of the potential multicollinearity. Given that participants could use multiple services across categories, we evaluated the odds of HCBS use within each category (e.g., using home safety services vs not using home safety services) for the full sample. We dropped use of in-home care and nutrition services from the individual category analysis because fewer than 10% of the participants used these types of HCBS.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
2 = 23.14; p <.001). When controlling for other factors, we found that for each increasing year of education, participants were 14% less likely to use any HCBS (p =.009). However, those with incontinence were three times more likely to use any service (p =.011). In addition, participants with sensory impairments had over five times greater odds of any HCBS utilization (p <.001).
|
2 = 16.29; p <.001). When we controlled for education, we found that women were more than twice as likely as men to use home safety referrals (p =.046). In addition, participants were 20% less likely to use home safety referrals with each increasing year of education (p <.001).
Women, those who lived alone, and those without difficulty with laundry activities were more likely to use transportation resources than were their counterparts (2 = 22.13; p <.001). Controlling for other variables, we found that women had 2.6 times greater odds than men of using transportation services (p =.045), and those living alone had 2.5 times greater odds of this service utilization (p =.014). Participants with difficulty with laundry tasks were 65% less likely to use transportation services (p =.029).
Incontinence was the only variable associated with adaptive equipment use (2 = 6.55; p =.011). Participants who reported difficulty with continence were three times more likely than their counterparts to use adaptive equipment services (p =.008).
Supportive service use was significantly associated with difficulty transferring and sensory impairments; the relationship with age was nonlinear (2 = 22.02; p <.001). Those with difficulty transferring were twice as likely as their counterparts to utilize supportive services (p =.05). Participants with sensory impairments were nearly four times as likely as those without them to use supportive services (p =.022). Age first showed a 60% decrease in use with each increasing year. This trend reversed for those participants at age 92, demonstrating an increased propensity to use supportive services for those at older ages. We tested a cubed form of age was to see if the curve would resume a downward trend, but it was not shown to be significant (see Figure 2).
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
We expected that age and gender would significantly impact service utilization patterns, but that the effect would disappear in multivariate models. When controlling for other factors, we found that age was a significant predictor in only one of the types of referrals, suggesting that age itself may not be an appropriate proxy for increasing disability or chronic care needs in an older sample. In a sample of participants that was selected on the basis of high utilization of insured services, younger participants had higher supportive service utilization than those in their seventies and eighties, with the trend increasing again for those at advanced ages. Although this result is somewhat puzzling, we assume that perhaps to screen into the study, these younger participants had a higher level of acuity not otherwise captured in the functional, diagnostic, or health service utilization measures. A mechanism to address this issue in future studies would be to evaluate whether health service use is for an acute event, possibly leading to an immediate change in functioning, or for treatment of a chronic illness that may not trigger increased use of supportive services.
For gender effects, we found that female participants used home safety services and referrals significantly more did than male participants, supporting previous research that women are also more likely than men to make home modifications and use safety equipment (Mathieson, Kronenfeld, & Keith, 2002). Female participants were also more likely than male participants to use transportation services. This may be due to a woman's willingness to stop driving when health conditions or functional impairments warrant (Siren, Hakamies-Blomqvist, & Lindeman, 2004), leading her to be more amenable to community transportation options. Similarly, gender and generational norms about balancing safety and independence offer a potential explanation for the greater likelihood among women in the sample to use home safety devices and transportation.
We also expected that participants who lived alone would have higher levels of HCBS utilization. Those living alone did show a higher probability of utilizing transportation services than did those living with someone, supporting the notion that older adults rely on easily accessible support networks to meet individualized care needs (Harris Interactive Inc., 2000; Kaiser Family Foundation, 2002). Older adults living alone may perceive the existence of social support from family and friends living outside a local area, but they may be less likely to receive instrumental assistance from them (e.g., getting transportation when needed).
The effect of social support on medical services utilization was an intriguing result. Individuals without social support were more likely to use medical services than those with support. One possible explanation is that older adults may defer to health care providers as treatment professionals, educators, and consultants on any particular problem, including nonmedical issues. Without friends and family to discuss the concerns of daily living, older adults may be more willing to turn to a health care provider for guidance and solutions. Previous literature shows that older adults without social support are more frequent users of multiple types of health services (Byrne et al., 2003; Daugird & Spencer, 1989; Mistry, Rasansky, McGuire, McDermott, & Jarvik, 2001). This concept lends support to the benefit of employing social service providers, such as care advocates, at primary care physician offices as a strategy to improve chronic care services while reducing overall health care utilization.
We posited that participants with higher levels of functional impairment would have higher HCBS utilization than would participants with lower levels of impairment. Our initial analyses that explored aggregate measures of ADL and IADL impairment did not show significant relationships for any of the HCBS categories. However, when we included individual measures of functioning, we found that different types of impairment were associated with different types of service use for two ADLs and one IADL. Incontinence was associated with any HCBS and adaptive equipment use. On the basis of our debriefing the care advocates, we believe that referrals to incontinence product suppliers were driving the relationship between adaptive equipment and incontinence. Difficulty transferring was associated with supportive service use, which might be related to use of physical activity and health education programs that were captured in this HCBS category. Debriefing sessions with care advocates suggested that referrals included activities such as balance classes and water aerobics sponsored by the Arthritis Foundation. For IADLs, those participants who had difficulty with laundry tasks were associated with any HCBS use. This result supports the idea that IADLs often involve complex, multistage tasks that test multiple domains of physical and cognitive functioning, leading participants with this difficulty to need a variety of services for additional support. Overall, these results suggest that specific functional difficulties are more important in predicting HCBS utilization than the total number of ADL or IADL impairments.
Finally, we focused on the role of health service utilization as a predictor of HCBS use. This factor, measured by the algorithm, showed no significance in any of the models. Given that it was used to identify those participants in Medicare managed care who were at high risk, and more than two thirds of those so identified went on to use services, we believe that the algorithm's value is more likely to be one of identifying those who need HCBS rather than discriminating different levels of need among those with scores over the threshold.
Lacking guidance from previous research, we did not initially anticipate educational attainment and specific health conditions to have an effect on HCBS utilization. The effect of education on any HCBS and home safety utilization specifically suggests that individuals with lower levels of education had a greater likelihood than those with higher levels of accessing these service categories. Given that sociological research indicates that education may be a proxy for socioeconomic status (Seeman et al., 2004), we think that greater use of any HCBS and home safety resources for individuals with lower levels of education may suggest that these participants, because of economic constraints, may not have previously considered available resources and specifically upgrading their home with safety devices. However, when informed about the benefit and potential affordability of HCBS, such as low-tech equipment that could dramatically improve their safety, these participants were open to referrals. Regarding the effect of health status on HCBS, it was not surprising to us that participants with a heart condition used referrals back to the medical group, given the nature of the condition. In addition, those with sensory impairments were more likely than those without them to use any type of HCBS as well as supportive services specifically. This finding may suggest that, like those participants who have difficulties with functional impairment, those with sensory impairments can benefit from a variety of HCBS, depending on their particular needs.
Overall, when we evaluated only at the aggregate-use category—that is, use of any HCBS—a very different picture emerged than when we observed service categories separately. Only three individual characteristics (education, incontinence, and sensory impairments) were significant factors in the aggregate and separate HCBS and medical services models. This finding suggests that lumping services together provides a very different picture of need than that which emerges when they are separated into service categories. Future studies should drill down to more discrete service categories and patterns of use to improve our understanding of the relationships among individual characteristics and utilization.
Strengths and Limitations
The overarching strength of this study was its design, which allowed us to examine multiple categories of services selected by participants who were guided by professional support. The care advocates' approach toward consumer choice tapped each participant's individual agency through an information–referral intervention geared toward improving knowledge and decision making about options for community support. Given that participants had the choice to select from a number of individual services rather than prepackaged services, we could examine a variety of options.
Although the study offers important new information on the associations among characteristics and service choice, several limitations must be noted. First, we lacked specific data about why participants made the choices they did. We assume that, in addition to recognizing the benefits of specific services, participants considered other factors such as service availability and geographic proximity. Cost was another issue, which we were unable to measure but that probably accounted for some of the decisions to use or not to use certain services. Given that we excluded those participants eligible for Medicaid services, we assume that most of the selected services had out-of-pocket costs associated with them, but we were not able to determine how much they cost, how much cost was borne by the participant, or the impact of cost on service selection. In addition, it would have been valuable to know if any participants had long-term-care insurance or other similar means to offset costs. Sample size was another limitation that affected our ability to include several service categories that had small proportion of use. For example, fewer than 10% of participants used either in-home care or nutrition services, and five of the models had less than 20% service utilization.
Data on race or ethnicity were not available for this study, restricting our ability to explore how cultural diversity affects HCBS use. Although more than 90% of the health plan's members were Caucasian, it is unknown if the participants reflected the larger health plan membership on racial or ethnic status. Given the growing literature on health disparities among older adults (Hayward, Crimmins, Miles, & Yang, 2000) as well as racial or ethnic variations in long-term-care use (Sciegaj, Capitman, & Kyriacou, 2004), this is clearly an area that should be studied further.
Finally, generalizability may be a problem for several reasons. Although the sample was drawn from a single health plan in southern California, the use of four large medical groups within the largest Medicare managed care provider in the United States improved the generalizability of study results. Second, there is considerable community variation in what services are available. A large multicommunity study would be needed to compare and confirm results across geographic regions.
The present study should be considered exploratory. Our results suggest that researchers should conduct more studies to understand factors associated with consumer choice to use services. Future studies should look at HCBS use and participants' decision-making process as primary research questions. Although specific findings are supported by earlier studies (e.g., gender is associated with use of transportation and home modification), other findings, such as the curvilinear relationship of age and supportive services and the lack of relationship among the service categories and diagnoses, are surprising and require further examination.
Policy Implications
This study represents a positive initial step toward understanding the relationship between individual characteristics and HCBS utilization patterns for high-risk older adults in a Medicare managed care setting. This article suggests that a "one-size-fits-all" approach to service referrals will not meet the varied needs of diverse consumers. As policy makers and practitioners work to operationalize the chronic care model (Wagner et al., 2001) toward integrated medical and social care, they must avoid the overstructuring of chronic care. Rather, they must recognize the need for myriad service options and unique configurations of these services to address the multiple and varied needs of older adults with chronic health conditions.
|
Footnotes |
|---|
1 Leonard Davis School of Gerontology, University of Southern California, Los Angeles.
2 Andrus Gerontology Center, University of Southern California, Los Angeles.
Decision Editor: Linda S. Noelker, PhD
Received for publication May 4, 2005. Accepted for publication November 4, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
