This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation

The Minnesota Senior Health Options Program

An Early Effort at Integrating Care for the Dually Eligible

^a University of Minnesota School of Public Health, Minneapolis

Robert L. Kane, University of Minnesota School of Public Health, Mayo Mail Stop 197 (Room D351), 420 Delaware Street SE, Minneapolis, MN 55455 E-mail: kanex001{at}umn.edu.

Decision Editor: John E. Morley, MB, BCh

	Abstract

Top Abstract Methods Results Discussion References

 
Background. Duplication of funding and resultant inefficiencies have prompted active consideration of pooling the funding for persons covered by both Medicare and Medicaid into a single managed care program. This study reports the initial results of the first such program.

Methods. A sample of enrollees in Minnesota Senior Health Options (MSHO) and two sets of controls (within the same catchment area and outside it) as well as their families were interviewed to assess their functional status and satisfaction with their medical care. Respondents included those living in the community and those living in nursing homes.

Results. The MSHO and control samples were generally alike in terms of demographics and illness patterns. The differences that were found reflected those attributable to geographic location more than program. The groups were also similar with regard to functional status. There were few satisfaction differences among the community-dwelling samples, but the MSHO nursing residents and especially their families expressed more satisfaction with several aspects of care.

Conclusions. Whereas no causal conclusions about outcomes can be drawn from a cross-sectional sample, there is no indication that managed care for the dually eligible population has profound impacts on care. However, the system of care provided to nursing home residents is appreciated over traditional care.

THE Department of Health and Human Services has been encouraging efforts to better coordinate services provided to individuals eligible for both Medicare and Medicaid. These individuals are often frail elderly Medicare beneficiaries who have high medical costs and have spent down their income and assets to become Medicaid eligible, or they are low-income elderly persons who have aged into the Medicare program but are not necessarily frail. Approximately 17% of individuals enrolled in Medicare were also covered by the Medicaid program and represented 28% of total Medicare expenditures during 1997. Dual eligibles represent approximately 19% of the Medicaid population and account for 35% of Medicaid expenditures ⁽¹⁾. Whereas these two programs potentially provide a comprehensive package of benefits, some observers note a lack of coordination in benefits and resultant fragmented care, which is both wasteful and inefficient. Medicare and Medicaid have different payment rules and provider qualifications, offer different benefit packages, and have little incentive for integration.

In response to these perceived problems, a series of demonstration programs was launched to test the effects of merging payments for health and long-term care of these so-called dual eligibles into a single managed care program. The first of these demonstrations was the Minnesota Senior Health Options program (MSHO). MSHO expands Minnesota's extant mandatory Medicaid prepaid managed care program (PMAP) to cover more Medicaid services (specifically a portion of nursing home costs) and merges it with prepaid Medicare managed care. Under MSHO, capitated Medicare and Medicaid payments are pooled to create a single fund. The size of the payment depends on whether the enrollee resides in a nursing home or is deemed nursing home eligible but lives in the community and is not seriously disabled. Whereas this money must be used to cover at least the usual benefits covered by each program, the funds can be used to obtain services that are outside the range of defined benefits to meet the specific needs of the enrollee. This flexibility, together with mandated care coordination, is seen as a major advantage of the integrated program.

The MSHO program relies upon capitation and subcapitation arrangements to encourage cost controls and adds a case management service to coordinate care across these autonomous groups. Whereas each plan establishes arrangements with providers to meet the needs of their patient populations, it is up to the care coordinator to navigate the system and obtain these services. The plans are under no mandate to change the way in which care is provided. Because the care is provided primarily by individual provider groups for whom MSHO represents a very small proportion of their overall business, most are unwilling or unable to change their practice style to accommodate MSHO enrollees. A majority of MSHO enrollees are nursing home residents receiving EverCare services that do not differ substantially from what the program already provides through its typical Medicare HMO plan ⁽²⁾.

The demonstration program began in February 1997 and was initially offered in four counties. It has expanded to additional counties in the seven-county waiver demonstration area within the metropolitan Twin Cities area. The MSHO program contracts with three health plans (Medica, UCare, and Metropolitan Health Plan) that in turn subcontract with various programs and provider organizations to offer the full array of acute and long-term services mandated by Medicare and Medicaid. The specific arrangements vary by plan and by the nature of the enrollee, but the basic arrangement is a series of subcontracts, some of which include subcapitated payment and some of which employ other arrangements. Care coordination is provided to varying degrees to all enrollees; its extent depends on their level of disability and the complexity of their care.

Because MSHO involves managed Medicare, enrollment is voluntary. Those opting not to enroll can continue to receive Medicare-covered services in a fee-for-service mode, whereas they obtain Medicaid under a mandatory managed care program.

This paper describes the initial findings from the federally mandated evaluation of the MSHO program, comparing the nature of the enrollees in MSHO with that of two control groups chosen to address issues of potential selection bias.

	Methods

Top Abstract Methods Results Discussion References

 
Sample
The respondents were all dual-eligible individuals (those enrolled in PMAP and eligible for Medicare) who lived in the defined study areas. Family members of dual eligibles were also asked to participate, when contact with a knowledgeable family member could be established. At the start of the sampling process in October 1998, over 21,665 people were enrolled in PMAP in the defined study areas, and 4,127 individuals were enrolled in MSHO.

To address the issue of selection bias, the evaluation used two separate comparison groups: dually eligible individuals who chose not to enroll and a similar group who did not have the option to enroll because they lived in an area where the program was not offered. By comparing the MSHO enrollees with a group that was comparable but did not have the option to enroll, we can control for the effect of selection. Using a group that was exposed to the same group of providers but chose not to enroll in MSHO (the in-area control group) addresses the effect of enrollment. Both control groups consisted of individuals enrolled in PMAP who were also enrolled in Medicare.

The two control groups were each matched to their corresponding experimental groups (both institutional and community) on the basis of age, gender, race, previous nursing home residence since the start of MSHO, and their Medicare managed care enrollment status prior to MSHO enrollment. An additional criterion, county residency, was used to match the in-area controls to the MSHO group. Individuals were selected for interview at random from these matched control groups. Parallel sampling strategies were implemented for both community-dwelling and nursing home resident enrollees. On the basis of information provided from state data files, individuals were identified as living in the community or in a skilled nursing facility and were further separated into experimental, in-area control, or out-of-area control groups on the basis of their enrollment status and area of residence. Approximately one half of the PMAP population and 75% of the MSHO population lived in nursing homes.

Data Collection
The survey data collection began in October 1998. MSHO and control subjects were interviewed in person by trained interviewers. When the respondent could not provide coherent information, a proxy was identified. The first choice for the proxy respondent was a family member who had regular contact with the enrollee, but in some instances (especially among nursing home residents) no family member could be identified. In these cases a staff proxy was used. In addition to the family proxies, for each enrollee or control beneficiary or staff proxy completing a survey, a family member was identified, when possible, to complete a telephone interview to obtain their perspective and to gather data on caregiving burden.

Survey Instruments
Areas covered by the survey instruments included a description of the person, information regarding the use of formal and informal care (for community-dwelling respondents), satisfaction (including issues of access to services, advance medical directives, general health, functional status [activities of daily living (ADLs) and instrumental activities of daily living (IADLs)]), and informal caregiver burden (for family of community respondents). Two potential sources of information were used: the client (a family proxy, or a staff proxy when the client was unable to complete the interview) and a family member. In the majority of cases a beneficiary and their family member were interviewed. Questions regarding satisfaction with MSHO specifically and health care in general were asked of clients and their families.

Wherever possible, previously tested measures were utilized. Items regarding ADLs were derived from instruments developed by Finch and colleagues ⁽³⁾. Questions focus on beneficiaries' ability to engage in basic daily activities (e.g., bathing, dressing, toileting, transferring, feeding, and walking). IADL items focus on the beneficiaries' ability to engage in a range of higher-order daily activities from shopping to the management of finances.

In addition to questions about ability to perform ADLs and IADLs, questions adapted from the work of Allen and Mor were also posed to those who reported functional difficulties to look for evidence of adverse events as a result of unmet need ⁽⁴⁾. For example, someone who reported difficulties using the toilet was asked if they had needed help and did not receive it, if they were wet or soiled because they did not receive this help, and if they had to wait more than 20 minutes before receiving help. Unmet need was also measured with regard to transferring, where patients were asked if they needed additional help getting in or out of a bed or a chair and did not receive it, and if they had fallen as a result of this lack of assistance.

Satisfaction questions were adapted from an approach that had been used previously in a study of assisted living. In this method, respondents were first asked to agree or disagree with a statement regarding receipt of services or about the nature of the services they received (e.g., were services provided when needed, did personnel communicate effectively, did patients [or families] participate in decision making). For each item, they were then asked to indicate how important such a service or style of practice was to them. The latter weight was used to create an individual preference-weighted satisfaction score. Parallel versions of the satisfaction items were used with both enrollees and family respondents.

Sample Size
The sample was stratified by community and nursing home residence and included a matched number of individuals in each of the two control groups. A target sample of 600 was chosen to ensure an end count of approximately 400 surveys without replacement.

There were many more instances in the community-dwelling control group where sampled individuals did not speak English or did not speak it well enough to understand the survey. Replacements were made to the study sample only for the in-area control group. Shortly into data collection our tracking records revealed a large number of cases falling out of the sample due to language barriers. It was also found that a large number of proxies were being used to complete many of these interviews for the same reason. Whereas we had anticipated a 20% proxy rate for the community sample, the proxy rate for the community control group exceeded our expectations. It was important to maintain a control group that reflected our experimental group, and for this reason, replacements were made to maintain the balance. Therefore, if the individual was identified as non-English speaking, we replaced them without attempting to identify a proxy.

Univariate Analysis
Because of wide differences in the composition of the groups, the findings are presented separately for the nursing home and community-dwelling samples with no efforts made to create overall rates. Chi-square independence tests were used for categorical data, and independent sample t tests were used for continuous variables. Each control group was compared with the MSHO sample separately. When appropriate (e.g., demographics, disability), proxy and respondent data were combined, but usually they were analyzed separately. However, the family-specific responses were separated by only the location of the enrollees (i.e., nursing home or community).

Multivariate Analysis of Satisfaction
To determine the effect of MSHO enrollment on the individual satisfaction items, risk-adjusted logistic regression models were used to account for possible effects of exogenous variables on satisfaction ratings. Characteristics that might affect an individual's response were included in a stepwise regression model. The final variables included in the regression models are indicated by superscripts in the following list. "C" indicates that the variable was significant for community cases, and "N" indicates that the variable was significant for nursing home cases. For beneficiaries, the characteristics included age^C,N, gender^C,N, education^C,N, race^C,N, ADL score^N, IADL score^C, number of major diagnoses^N, number of items missed on the mental status questionnaire (MSQ)^C,N, and study group^C,N. For family members, the initial set of control variables included beneficiary age^C,N, ADL score^N, IADL score^C, number of major diagnoses, number of items missed on MSQ; family member age^C, gender^N, education^N and race^C, and study group^C,N.

Stepwise regressions were used to determine which of these variables were pertinent to include in each risk-adjusted model. The results of these analyses revealed that only a few variables were necessary. Whereas depression was considered a characteristic of the beneficiary and significant in predicting satisfaction, it was not included in this analysis because it is a treatable problem and the beneficiary's satisfaction may be a result of the quality of his or her treatment.

Adjusted odds ratios were calculated for each of the satisfaction questions. The respondents' answers to each satisfaction item were coded zero if they agreed with the item and one if they disagreed. We emphasized dissatisfaction rather than satisfaction because agreement was the norm. Individual satisfaction items were regressed on the variables identified from the stepwise analysis, including an indicator variable for study group membership. The comparison group for the indicator variable was the MSHO group.

	Results

Top Abstract Methods Results Discussion References

 
Although there was a need to replace subjects for the in-area controls because of language problems, the actual response rates for the rest of the samples ranged from 78% to 90%. Of those approached, only 2% to 3% of nursing home residents and 7% to 13% of community-dwelling residents refused to participate.

As Table 1 shows, the demographics for the patients were fairly similar across the three study groups. Among the community sample, the out-of-area controls tended to have more white respondents with less education. This group also reported less hypertension. Among the nursing home samples, the out-of-area differences were more substantial. In addition to being more white and less educated, they were less cognitively impaired than the MSHO sample. Both control groups were more likely to rate their health status as fair or poor than the MSHO sample. There were also differences in disease reports. Consistent with the cognitive performance, both control groups had less dementia. The in-area control had more myocardial infraction than the MSHO sample. The out-of-area control had less "other mental health" and more stroke than the MSHO sample.

When the extent of unmet needs was explored among those who reported at least some dependency, only a small number of differences were found. Table 3 presents the data in the form of a ratio to emphasize the small number of cases eligible for each item. In the community sample, the in-area control respondents were more likely to report that they needed help transferring and did not receive it (25% compared with 9.4% for the MSHO respondents). Among the nursing home sample, the out-of-area controls had significantly less unmet need with regard to putting on clean clothes (7.7% compared with 20% for MSHO) and being wet or soiled because no help was available (20% vs 44.4% of the MSHO respondents). Despite the small numbers of persons eligible to respond to each item, there is no consistent pattern favoring any group in either the community or the nursing home.

	Discussion

Top Abstract Methods Results Discussion References

 
Overall, the study groups seem to be generally comparable. The differences that do exist reveal more of a geographic effect than a program effect, suggesting that the differences are more a result of resource availability in the nonmetropolitan areas than a program effect due to enrollment in MSHO.

As expected, the community-dwelling group was much healthier than the nursing home residents. The community-dwelling population on the whole is a very functional group, with few differences across study groups. Less than 10% of each group has limitations with toileting, transferring, feeding, or walking. Because the survey relied on a cross-sectional design, no causal inferences can be made about this lack of differences.

The greater extent of advanced directives among MSHO enrollees may reflect a deliberate effort to obtain such information. Managed care has an incentive to avoid futile care. However, neither of the MSHO samples reported a higher rate of feeling pressure to establish advance directives.

MSHO families seem more satisfied with their relatives' care than families of the controls. This is especially true for the families of nursing home residents. One possible explanation for this effect might be the greater exposure to the EverCare model ⁽²⁾. Much of the primary care for nursing home residents under MSHO is delivered by EverCare or other organizations that use nurse practitioners as active members of the primary care team.

The findings here suggest that those who opt for MSHO appreciate some of the benefits it is striving to deliver. The families are more satisfied with the care being given. Enrollees are more likely to have advanced directives. There is no evidence that the informal care burden has been reduced, and the cross-sectional design does not allow any inferences about the implications of the similarities in functional and medical status between MSHO and control groups.

The inability to detect a difference in unmet needs may reflect the small sample sizes available. Few people in the community had substantial levels of disability, and there is no immediate reason to believe that the nursing home care itself should be different under MSHO.

Some of our observations have serious implications for the next phase of the MSHO evaluation. The majority of the community-dwelling clients do not appear to be severely disabled. It is not clear how much they would profit from a program that offers more coordinated care. The MSHO effect is seen more in the nursing home. Whereas care coordination should affect community residents more, many of these enrollees are too functional to trigger such attention. Care for the nursing home residents, by contrast, relies heavily on nurse practitioners to provide more intensive primary care. However, the use of an EverCare approach is not unique to MSHO. Indeed, EverCare was designed as a Medicare HMO.

The MSHO model itself is limited. As noted earlier, much of the primary care in the community is provided by physicians for whom such work constitutes only a small part of their practice. Getting them to change practice behaviors will not be easy under these circumstances. The task of creating a seamless system out of the currently fragmented care delivery programs falls to the care coordinators, but many of the community-dwelling enrollees are too functional to trigger extensive care coordination. For those in nursing homes the care coordinators are the same nurse practitioners who provide the primary care. For those not receiving NP care, it is unclear just what the care coordinators can do beyond maintaining closer communication with families. Such attention may be the reason for the higher satisfaction scores.

Subsequent analyses of the MSHO demonstration will include an examination of longitudinal utilization data beginning prior to the implementation of the MSHO program to determine patterns of change in the use of services among both MSHO enrollees and controls in the community and nursing home settings. Evidence of changes in participants' functional status will come from data collected in a follow-up survey of the community-dwelling population and administrative data on case mix for the nursing home sample. This evaluation should provide useful insights on the effectiveness of this effort to coordinate Medicare and Medicaid services for older people.

	Acknowledgments

 
This work was performed as part of a HCFA contract (HCFA 500-96-0008, Task Order 3). However, the opinions expressed are solely those of the authors and do not reflect those of HCFA, the federal government, or the University of Minnesota.

Received August 9, 2000

Accepted August 17, 2000

	References

Top Abstract Methods Results Discussion References

 

1. Clark WD, Hulbert MM, 1998. Research issues: dually eligible Medicare and Medicaid beneficiaries, challenges and opportunities. Health Care Financing Rev 20:1-10.
2. Kane RL, Huck S, 2000. The implementation of the EverCare demonstration project. J Am Geriatr Soc 44:218-228.
3. Finch M, Kane RL, Philp I, 1995. Developing a new metric for ADLs. J Am Geriatr Soc 43:877-884. [Medline]
4. Allen S, Mor V, 1997. The prevalence and consequences of unmet need. Med Care 35:1132-1148. [Medline]

This article has been cited by other articles:

				R. L. Kane, P. Homyak, B. Bershadsky, T. Lum, S. Flood, and H. Zhang The Quality of Care Under a Managed-Care Program for Dual Eligibles Gerontologist, August 1, 2005; 45(4): 496 - 504. [Abstract] [Full Text] [PDF]

				R. L. Kane, P. Homyak, and B. Bershadsky Consumer Reactions to the Wisconsin Partnership Program and Its Parent, the Program for All-Inclusive Care of the Elderly (PACE) Gerontologist, June 1, 2002; 42(3): 314 - 320. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation