In today's Medicaid delivery system, providing long-term services and supports (LTSS) in the least restrictive setting compatible with appropriate care and available resources has become the shared goal of payors, providers, and recipients of services. Known as “rebalancing” the LTSS population, this concept is supported by new government initiatives at both federal and state levels.
The interesting question now is: How far can rebalancing go and at what point is it not economically rational? Consumer choice has historically been the driver of rebalancing initiatives and will continue to play a major role. The Supreme Court decision of 1999 in the case of Olmstead v. L.C. supports this view, holding that public entities must provide community-based services to individuals with disabilities, consistent with appropriate services and the wishes of the individuals.1 With the recent proliferation of managed long-term care programs additional focus will inevitably be placed on what is most cost-effective. In order to do this, managed care organizations, accountable care organizations (ACOs), and individual states will need to look toward innovative uses of data that accurately capture and convey physical and cognitive limitations. There are considerable operational challenges to rebalancing the LTSS population by transitioning nursing facility residents into the community; it is a difficult and labor-intensive process. Therefore, focusing the efforts of LTSS rebalancing on those individuals who are most capable of making the transition from institution to community would likely add value.
Using the Minimum Data Set to support rebalancing
An important source of information about the institutionalized long-term care population is the National Resident Assessment Instrument (RAI) Minimum Data Set (MDS). The MDS records a mandated assessment conducted for all residents of nursing facilities on admission and every three months thereafter. In addition to these “routine” assessments, MDS assessments are also conducted whenever there is a material change in a resident’s status. The MDS contains information on clinical status and services, functional status and assistance, plus psychosocial factors. The results of MDS assessments are used to categorize residents into resource utilization groups (RUG-III), a case-mix classification system used to determine reimbursement for nursing home and skilled nursing facility (SNF) care.
Section Q of the MDS records whether or not the resident is interested in returning to the community, and also records whether family members are supportive of this preference. An analysis of first-time admissions to Minnesota nursing homes from July 2005 to July 2006 found that 64% of residents showed a preference or support for community discharge; that 40% had health and functional conditions predictive of community discharge; and that 20% had low-care requirements. Overall, 6% of residents met all of these targeting criteria at 90 days after admission, and the authors concluded that a community discharge intervention could be offered to new nursing home residents at 90 days after their admission.2
In another study, MDS assessment data was used to define a “low care” category for nursing home residents. The broadly defined “low care” category comprised those residents who did not require assistance in any of the late-loss activities of daily living (ADL)—bed mobility, toileting, transferring, and eating—and who were also not classified in the RUG-III “clinically complex” or “special rehab” categories. The narrowly defined “low care” category added the qualification that the resident had to be classified in either of the two lowest groups in the RUG-III scheme.
Under this classification, 5.1% of residents nationally met the narrow definition of “low care” and 11.8% met the broad definition. For the narrow definition, the percent in each state that were considered to be “low care” ranged from 1.0% to 11.4%; for the broad definition, the range was 1.8% to 19.0%. States that spent a greater proportion of total Medicaid LTSS expenditures on HCBS or had more residential and assisted living beds had significantly fewer nursing home residents with narrowly defined “low care” needs. The conclusion was that there were a significant number of nursing home residents who might be able to avoid permanent institutionalization.3
Fries and James used data from nursing facility transition programs in three states to predict which nursing home residents might successfully transition from nursing home to facility care and should be targeted as individuals for outreach in the Money Follows the Person (MFP) initiative. They developed a “Q+ index” that could be computed from the values of variables found on the MDS assessment. In addition to the preference/support variables from Section Q of the MDS, variables included length of stay, age, and RUG-III group, as well as functional and clinical considerations. The index had 86.5% sensitivity and 78.7% specificity for identifying candidates for transition. Statistically, these results indicate that the model is very powerful. While it was developed using MDS 2.0, the index could be applied to MDS 3.0 assessments.4
Managed care plans could use MDS-based algorithms to identify members who would be good candidates for transition from nursing home to community environments. When paired with historical claims data, such algorithms can be quite powerful by enabling managed care plans to understand and profile potentially successful and unsuccessful candidates for community services, along with their expected costs.
States have historically used cost-based reimbursement for nursing facilities; however, with the implementation of managed care delivery systems, the focus of reimbursement will shift away from cost and move toward quality and value. Managed care plans with access to and expertise with MDS and other assessment data will also be in a preferred position to differentiate rates to their facilities to better align incentives and tie payments to outcomes. Effective utilization of MDS data can also aid State Medicaid agencies in developing estimates of how much rebalancing can likely be achieved, risk adjusting Medicaid payment rates, and perhaps even setting the baseline for Medicaid managed care rates.
In conclusion, MDS data has been shown to be effective in identifying populations that can successfully transfer from nursing facilities to home- and community-based settings. When paired with historical claims data, powerful predictive models can be built to identify successful candidates and also to predict their costs, thereby enabling managed care entities and states to rebalance in a clinically and economically rational manner.
1. Olmstead: community integration for everyone. Retrieved November 8, 2013 from http://www.ada.gov/olmstead/.
2. Arling G., Kane R.L., Cooke V., Lewis T. Targeting residents for transitions from nursing home to community. Health services research. (June 2010). 45(3):691-711.
3. Mor V., Zinn J., Gozalo P., Feng Z., Intrator O., Grabowski D.C. Prospects for transferring nursing home residents to the community. Health affairs. (Nov-Dec 2007). 26(6):1762-1771.
4. Fries B.E., James M.L. Beyond section Q: Prioritizing nursing home residents for transition to the community. BMC health services research. (2012). 12:186.