The occurrence of nursing home use is derived from Medicaid-reimbursement claims. We have limited the use of claims to those occurring during or following the period in which the individual was a recipient in the IHSS program. Only nursing home stays occurring in 2005 are counted.40 The claims (both payments and stays) available do not include skilled care placements or days covered entirely by payers such as Medicare, the VA, or private funds. A consequence of these limitations is that these data may under report short-term, skilled care days/stays; and under count total expenditures if service use was paid by these sources. Medicaid-paid co-payments are included in tabulations of Medicaid-paid nursing home stays and days. Within these biases the preponderance of nursing home claims are those involving IHSS recipients age 65 or more. The incidence of Medicaid-paid nursing home placement among IHSS recipients is low: about 0.26% among children, 2.25% among non-aged adult recipients, and 5.9% among those age 65+. As shown in Table 29 there are some differences in the unadjusted probabilities of nursing home use by IHSS recipient age and provider group. As a group, those with Other Relative providers tend to have among the lowest likelihood of placements. Adults with Spouse providers tend to be among those with a higher likelihood of placement.
These patterns are somewhat effected after adjusting for recipient characteristics, as shown in logistic regression equations in Table 30.41Among recipients age 18-64, there is a persistent adjusted effect: IHSS recipients related to their providers have a lower adjusted odds of nursing home use than persons with Non-Relative providers. Further, recipients with paid Parent providers tend to have a lower adjusted risk than recipients with either Spouse or Other Relative providers. Among recipients age 65 or more, the protective effect of relatives as providers is present only comparing Other Relatives to Non-Relatives. Spouses have a modest tendency toward a lower placement rate, but this does not reach statistical significance. In short, the IHSS program, including its waiver-permitted providers, is at least as successful in aiding families and individuals remain in the community as are recipients with Non-Relative providers. New IHSS recipients age 65+ are also less likely than continuing recipients to have a nursing home placement.
Nursing home use has a direct effect on Medicaid costs. One consequence of this is seen in the nursing home expenditures for 2005 shown inTable 31. These expenditures reflect the accumulated costs for any nursing home stay in 2005, restricted to stays occurring after entry into (and, if applicable, exit from) IHSS. Most of these expenses seem to be for non-skilled care as the average monthly rate among nursing home users approximates the 2005 Medicaid daily nursing reimbursement rate of $115. Comparisons between continuing and new recipients show generally similar average daily expenses. In both age groups, IHSS recipients tend to have relatively similar average monthly expenditures across provider types. Recipients (non-aged adult only) with Parent providers, the one exception, have the highest average daily expenditures, but this may be an artifact of sample size. Parents are the smallest subgroup and their mean values are perhaps affected upward by the wide standard deviation in these data.
Analysis of expenses associated with the transition from IHSS to nursing home care, the transitions from nursing homes into IHSS, and the total Medicaid expenditures incurred by nursing home recipients are beyond the scope of this analysis; as is an analysis of the duration of nursing home placements.
Persons in managed care have been included in these analyses, as Medicaid claims for non-skilled nursing home care are available. Custodial nursing home care is not included under managed care capitation payments. Tables showing the likelihood of nursing home place among IHSS recipients, excluding those in managed care are in Appendix F, Table F-1. The exclusion of the managed care recipients, results in minor changes in the percentages, approximately 0.05% among minor children, and 0.2% among adult recipients.
Provider by health condition interaction terms were tested in earlier models, but were not statistically significant. Estimates involving recipients age 3-17 are omitted from Table 30, as there were too few cases to estimate reliable models. Appendix F, Table F-2 shows the logistic regression results for models excluding managed care recipients. These results are very similar to those shown in Table 30.