Medicaid-Financed Institutional Services: Characteristics of Nursing Home and ICF/IID Residents and Their Patterns of Care. Notes


  1. Olmstead v. L.C. 527 U.S. 581 (1999).

  2. Consequently, references in this report to "residents" or "recipients" of institutional care indicate only those institutionalized individuals enrolled in Medicaid.

  3. Because our data were censored at the end of 2007, however, we limited the analysis to only those spells beginning in the latter half of 2006 when analyzing the length-of-spells. Details are in Appendix B.

  4. Our sample was restricted to individuals enrolled in Medicaid at some point between June 1, 2006, and December 31, 2007, and living in non-excluded states (Arizona, Indiana, Maine, Minnesota, New Hampshire, and Utah). Enrollees with new spells of care included those using Medicaid-financed nursing home services for the first time, as well as those who had previously resided in a nursing home but had more recently been hospitalized or living in the community.

  5. Approximately 67 percent of stays in Alaska and 53 percent of stays in Oregon lasted less than three months, compared with 47 percent for all states together (see Appendix Table D.1).

  6. In addition to the six states excluded from the analysis of 2006-2007 data, three other states -- Hawaii, Michigan, and Tennessee -- and the District of Columbia were excluded from the earlier study based on 2001-2002 data (Wenzlow et al. 2008).

  7. The high-HCBS states were Alaska, Arkansas, California, Colorado, Idaho, Illinois, Kansas, Massachusetts, Missouri, Montana, Nevada, New Mexico, New York, North Carolina, Oklahoma, Oregon, Texas, Vermont, Virginia, Washington, West Virginia, and Wyoming. All other states analyzed in this report were low-HCBS states.

  8. Full definitions of the cash assistance-related, medically needy, poverty-related, Section 1115 demonstration waiver-related,and "other" criteria are in Appendix A.

  9. The "other" category also includes those covered under more restrictive requirements than SSI standards in Section 209(b) states; three of the remaining four states with "other" as the plurality MAS category were Section 209(b) states.

  10. Over three-fifths of the individuals (2,737 out of 4,481) who qualified for benefits through a Section 1115 demonstration waiver were enrolled in Wisconsin's Badger Care initiative two months prior to admission, but only 57 remained enrolled under Section 1115 criteria at the beginning of the Medicaid-financed spell, with the vast majority of the others qualifying under "other" criteria such as the 300 percent rule.

  11. Because the study period ended on December 31, 2007, all spells that were active on that date were censored. To obtain more meaningful statistics regarding length-of-stay, for this portion of the analysis we followed Wenzlow et al. (2008) and considered only spells beginning in the last six months of 2006 to ensure that each spell could potentially last at least one year. (Otherwise, spells beginning, for example, in October 2007 would appear to represent short stays because they would be censored at the end of the year, yielding an incorrectly high number of shorter stays as a percentage of all stays.) We likewise limited the analysis to spells beginning in the latter half of 2006 when analyzing (below) the associations between length-of-stay and state policy variables.

  12. Data are in Appendix Table D.2.

  13. Although a detailed analysis of the relationship between use of HCBS and nursing home services by the same individual was beyond the scope of this report, the person-level and claims-level data available through MAX are well suited for pursuing such an analysis in the future.

  14. The number of elderly or disabled individuals in each state was estimated using the American Community Survey (ACS). See Wenzlow et al. (2011).

  15. These associations were statistically significant at the 5 percent level or better.

  16. Although many states characterized as "high-HCBS" states by Kaye et al. (2009) -- including Alaska, California, Colorado, Kansas, New Mexico, Vermont, and Washington -- tended to have higher values of all four re-balancing variables, and conversely for "low-HCBS" states, a small handful were clear outliers. For example, Alaska devoted nearly three-quarters of its Medicaid long-term care spending to HCBS, compared with approximately 40 percent for all states (Figure II.2). The positive associations between length-of-stay and the re-balancing variables remained positive when outlier states were excluded, but the strength of the associations was generally weaker than in Table II.8 (see Appendix Table D.3).

  17. Data are available upon request.

  18. Data are available upon request.

  19. Populations in large private ICFs/IID are similarly decreasing. Between June 1997 and June 2009, their populations declined from 28,200 people to 18,400.

  20. The population living in ICFs/IID serving 4-6 residents decreased about 7 percent from 21,500 in 1997 to 20,100 in 2009; the number of individuals in ICFs/IID serving 7-9 residents decreased by about 18 percent from 23,650 to 19,400 over the same period. (In the meantime, the number of HCBS recipients has grown from 223,200 to 562,000.)

  21. For example, Indiana and Louisiana had comparable numbers of ICFs/IID in 2007 (533 and 513, respectively), but Louisiana had 18 large facilities (16 beds or more) whereas Indiana had only six (Lakin et al. 2008).

  22. This was partly due to Oregon's extensive use of managed care, which limited the number of enrollees observed in our sample of FFS data.

  23. It is not clear why the percentage of spells that were new was higher in these states, although the relatively small base of continuing spells was a contributing factor. (A new spell in a state with few previously existing spells will constitute a higher percentage of the total.) None of these states opened new facilities in either 2006 or 2007 (Lakin et al. 2009; Lakin et al. 2008; Lakin et al. 2007). However, these states' state institutions tend to serve emergency placements or placements of last resort, which could explain the higher percentage of new spells.

  24. "Other" criteria include qualifying under the 300 percent rule, which allows individuals with income up to 300 percent of the SSI threshold to qualify for Medicaid assistance for institutional care. Medically needy provisions (a state option) allow individuals with higher incomes to qualify for Medicaid by deducting incurred medical expenses from their income and/or assets to determine financial eligibility. Poverty-related expansions enacted since 1998 allow states to, for example, elect to extend full Medicaid benefits to otherwise ineligible aged and disabled people up to 100 percent of the federal poverty level. More detailed definitions of the cash assistance-related, medically needy, poverty-related, Section 1115 demonstration-related,and "other" criteria are in Appendix A.

  25. ICFs/IID can be used to provide respite care, which could also explain some shorter stays, but respite care is covered through HCBS waivers and does not include room and board, which the individual must finance with non-Medicaid sources of payment (Prouty et al. 2008). We believe it is more likely that some of the implausibly short ICF/IID stays reflected coding errors in the data.(For example, the relatively high percentage of ICF/IID stays lasting three months or less in Connecticut was the result of 20 one-day spells at a single facility.)

  26. For individuals with ID/DD who have more limited medical needs, nursing services are available through Medicaid in ICFs/IID (both larger and community facilities) and community settings.

  27. There is a larger literature on the similar question of to what extent persons with chronic mental illness have been treated in nursing homes and whether nursing homes are adequately equipped to provide appropriate treatment for people with mental illness (Molinari et al. 2009; Li 2010).

  28. It is possible that the ICF/IID populations in some of these states might be older than those in other states and therefore at higher risk of requiring skilled nursing care on a regular basis.

  29. Because most ICF/IID stays are very long, the population of individuals with new spells of ICF/IID care followed by HCBS use in our 18-month study period is unlikely to be representative of the population of all ICF/IID residents unless these shorter stays can be attributed primarily to ICF/IID closings and would have otherwise been long-stays had the facility remained open.

  30. See Appendix Table D.6.

  31. This relationship was largely driven by the data from Wisconsin, which closed eight large facilities during this time period (Lakin et al. 2008; Lakin et al. 2007); if Wisconsin is removed from the data, the association becomes insignificantly different from zero. Although the data are presented as a scatter plot with a regression line to facilitate comparisons to other figures, the large number of states with no change in the number of ICFs/IID from 2006 to 2007 suggests that a categorical data analysis is more appropriate than linear regression; applying this alternative approach resulted in similar findings.

  32. Data are available upon request.

  33. The measured median length-of-stay for ICF/IID spells in this analysis was 425 days, which is longer than the 12-month look-ahead period that we preserved when limiting the analysis to spells beginning in the second half of 2006; hence, the true median length-of-stay is some number greater than 425 days.

  34. There are almost twice as many people living in agency-managed residential settings financed by HCBS than are living in ICFs/IID.

  35. Now known as ICFs/IID.

  36. Within a single benefit period, Medicare provides up to 100 days of such coverage. A benefit period begins upon admission to an inpatient hospital or skilled nursing facility, and ends 60 days after the date on which inpatient hospital or skilled nursing care was last used. In practice, Medicare usually reimburses less than 100 days of skilled nursing care for qualifying individuals.

  37. Services covered under managed care (including any for LTC) generally cannot be identified in MAX as they are reported in "encounter records," which are known to be incomplete in MSIS and MAX.

  38. Although admission dates provide additional information about institutional services that might not be captured in other MAX variables, not all states and facilities include admission dates on their claims records, and this information only reflects services used in the facility in which the person resided at the time the claim was paid.

View full report


"MFIS.pdf" (pdf, 1.1Mb)

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