This section shifts the analysis from a description of IHSS program recipients to the consideration of the quality of care and other outcomes given the “choice” of provider types. The data sources used for this comparison are the IHSS assessments and Medicaid claims data. In combination these data sources enable us to investigate the following question:
Adjusting for disability and other attributes, what are the Medicaid program expenditures and health care events incurred by IHSS Plus Waiver program and non-waiver recipients? Are there differences by age group?
Included in these comparisons are all IHSS services, as well as personal care from Medicaid HCBS waiver programs. These services are available to eligible Medicaid recipients, and are unaffected by whether the recipient is enrolled in Medicaid managed care. Additionally, we examine Medicaid-reimbursed hospital, ER, nursing home, home health, and medical provider claims. These services generally do not generate a claims-record for persons in Medicaid managed care, so the sample size for analyses involving these services reduce to beneficiaries receiving health care reimbursed through fee for service claims. For hospital, nursing home, and ER use, the compilation of claims starts with encounters occurring within or subsequent to the first month of IHSS eligibility in 2005. Expenditures and utilization for all remaining months in 2005 are compiled as the basis for calculating mean monthly expenditures for these services. The compilation of chronic health conditions from Medicaid claims, includes all claims in 2005, regardless of months of IHSS participation. This was done under the assumption that chronic conditions are pre-existing in 2005, and with recognition that the inclusion of all claims reduced some of the under reporting of conditions that occurs if only prior year claims are used in identifying diagnoses.
Both unadjusted descriptive and multivariate analyses of expenditures and health care events are reported. The measures of primary interest in the multivariate models are the coefficients for IHSS provider types. All models adjust for recipient gender, race/ethnicity; household size; cognitive, ADL, and breathing limitations; the number of chronic health conditions. Household size and recipient limitations are the basis of IHSS benefit eligibility. Total Authorized IHSS Hours are also included as potentially reflecting changes in functional limitations or living arrangement that may not be reflected on the baseline IHSS assessments. Authorized hours are reduced as the availability of informal care increases, so that higher hours (up to the cap of 283 hours) corresponds to an increasing reliance on paid IHSS assistance. Complementing the recipient characteristics is one county indicator: per capita income, an adjustment for prevailing cost of living. The models also include a dummy variable representing whether the individual was a new IHSS recipient in 2005 or continuing from 2004. This tests whether new recipients had different expenses and utilization than continuing recipients, after adjusting for recipient characteristics.30
30. A series of equations that included interactions between provider type and the number of chronic conditions were evaluated. These items did not sufficiently improve the fit of the model to be retained in the analyses presented. Additionally, the ordinary least squares analyses were replicated using logarithm transformations of the expenditures measures instead of raw expenditures data. These models generally had higher R2 values, but as the results testing whether the coefficients on the Parent and Spouse provider measures were significantly different from Non-Relative providers were consistent (in terms of the direction of the sign) with those in the non-transformed models we have elected to report only models with the non-transformed data. These models have the advantage of being in dollar units, and more readily understandable than the percentage comparisons possible using the logarithm transformations.