Across all age groups participating in IHSS, mean unadjusted Medicaid expenditures (excluding pharmacy payments) range from $1,400 to $1,700 per IHSS participation month. This is a cost inclusive of Medicaid-reimbursed personal assistance-related expenses, which averaged about $825 in 2005. The highest average total expenditures are among those ages 18-64, the lowest among those ages 65 or more. Lower expenditures among this latter group are explained, in part, by more of these recipients having access to Medicare -- their primary payer for hospital, physician, and other health care use.
As shown in Table 11, mean unadjusted monthly expenditures for all age groups vary substantially by the number of an IHSS recipient’s exposure months in 2005. Expenditures shown in this table are accumulated for every month after IHSS eligibility in 2005. For recipients continuing from 2004, all would be eligible in January 2005. The new recipients could have entered in any month starting with January. Approximately 4,000 recipients entered the program each month, with an approximately equal number leaving. Persons with fewer than three months in the year tend to have average monthly expenditures that are about three times higher than the average monthly expenditures for those in the IHSS program for a full year. The causes for the difference across exposure months are beyond the scope of this analysis, but they likely are associated with changes in health status immediately preceding program entry or that contribute to leaving the program.
Provider Type and Medicaid Expenditures. Table 12 arrays the mean expenditures data by age and IHSS provider groups. These unadjusted results show a tendency for recipients of Spouse providers to have lower mean monthly expenditures than those receiving care from other providers. There are relatively few unadjusted differences in mean expenditures comparing recipients of Other Relatives and Non-Relatives providers. Expenditures among new recipients tend to be lower than for those of continuing IHSS recipients.
A set of ordinary least squares regression models, Table 13, were used to provide a comparison of adjusted provider effects on expenditures. Each column presents a model for a particular IHSS recipient age group. The comparisons of interest in the analysis are those of provider type. The reference category for the provider types is Non-Relatives. The coefficients, multiplied by 1,000, convert the effect into the metric of dollar units and facilitate interpretation of the differences among the recipient-provider groups in terms of average monthly dollar expenditures.31 While the models do not fit the data particularly well, the purpose is to test the adjusted predicted expenditure differences between providers. The individual covariates for these comparisons tend to have high levels of statistical significance, even for small difference in the predicted mean monthly expenditures. This is due, in part, to the large sample size.32
Among these age 3-17, Parent providers tend to have about $920 lower adjusted Medicaid expenditures than Non-Relatives. Other Relatives seem to have slightly lower adjusted expenditures than non-relatives, but this difference is not statistically or practically significant. Recipients age 18-64 with Spouse IHSS providers have predicted mean monthly Medicaid expenditures (holding everything else constant) about $1,000 lower than do those with Non-Relative providers. This estimate is somewhat larger than the difference in the unadjusted comparisons. Recipients with Other Relative providers have mean expenditures about $170 lower than Non-Relatives. There was no statistically significant difference between those with Parent providers and Non-Relatives. Among recipients age 65 or more, those with Non-Relative providers have predicted average monthly month expenditures that are higher than those for either recipients with Spouse providers ($780), or Other Relatives ($110).
Expenditures comparing new with continuing IHSS recipients showed only minor differences in adjusted mean monthly expenditures: non-significant among children recipients, slightly higher among those 18-64, slightly lower among the aged.
The association of provider type with expenditures was evaluated as both a main effect, and as the interaction of provider type and the number of the recipient’s health conditions. The interaction models did not improve the model and were not retained.
Additional models were estimated to test the stability of the provider findings. These included models limited to those with 12-month participation, and those with fewer than 12 months. The former had higher R2 values, the latter, lower R2 values. This is consistent with the higher variability in this latter group. In spite of these differences in model fit, the effect of provider type remained relatively constant. There were no changes in statistical significance or direction of effect, nor in substantively meaningful magnitude. Analyses were also conducted using the logarithm of expenditures. These models produce findings consistent with the non-transformed models. They are available on request.