This section begins an examination of some of the component services that contribute to the total Medicaid expenditures. We begin with hospital use, often a contributor to high expenditures. Hospital use may also serve as an indicator for problems in medical care and quality of home care. Table 14 shows unadjusted mean expenditures for hospital care (among those having a hospital stay) organized by age and provider group; and by continuing and recipients enrolling in IHSS during 2005. The highest mean monthly hospital expenditures (incurred after IHSS enrollment) are among those ages 3-17. For all age groups, but especially for those age 65 or more, it is important to recognize that these figures may be biased downward relative to total “all-payer” expenditures as the Medicaid results do not include reimbursements from other payers (e.g., private insurance, Medicare, Veterans Administration (VA), out of pocket).33 Among all recipients age groups the unadjusted average monthly Medicaid hospital expenditures generally show the IHSS Plus Waiver recipients (i.e., spouses of adults, parents of minor children) to have either the lowest mean expenditures or to have expenditures approaching the lowest group. IHSS recipients entering the program in 2005 tend to have higher mean monthly expenditures than recipients continuing from 2004. This may be, in part, a function of the fewer IHSS participation days among new recipient. As seen later, new recipients have lower incidences of hospital stays. Further as shown in the “Mean Total $” rows in Table 14, there is little difference within age group in the average of hospital expenditures incurred over the observed months by hospital users in each recipient-provider group. In general, these expenses are indicative of short stays, but as evident from the standard deviations, some recipients accumulated ten’s of thousands of dollars in hospital costs.
Any Cause Hospital Stays. The next several tables refine the hospital expenditure analyses to assess whether there are differences between provider groups in the likelihood of having hospital stays. Hospital use may be indicative of differences in recipient case mix and/or of the quality of IHSS and the condition management assistance received. Table 15 shows the unadjusted probability of an “any cause” hospital stay in 2005. These incidents occurred after IHSS enrollment (or in the same month as IHSS enrollment). The unadjusted likelihood of a hospital stay is relatively comparable among the adult recipients, with rates about double those for minor children. IHSS recipients across the provider types generally have similar rates, although recipients of Spouse providers are more likely to have stays.
Table 16 extends the analysis of hospital use by adjusting for recipient case mix differences. These logistic regression models compare the difference in odds (expressed as an odds ratio) of an “any cause” hospital stay during 2005 between each of the provider groups. These comparisons are based on consideration of the main effect of provider type. (Interactions between provider type and the number of health conditions, as a group, did not statistically improve the model and were not retained in the analysis.) With the modeled adjustments the differences between recipients having IHSS Plus Waiver-permitted providers (i.e., parent and spouse respectively) and those with Non-Relative providers generally become statistically non-significant. This finding holds among all but the non-aged adults who have Spouse providers. These recipients are about 15% more likely to have hospital stays than those with non-relatives. Recipient outcome comparisons between those with Non-Relative providers and Parents (of those 18-64), as with the unadjusted results, show substantially lower odds of a hospital stay for those with Parent providers. This difference is reduced to about 25%, rather than 50% in the unadjusted results. Non-aged adults with Other Relatives as providers show about a 10% lower risk of hospital stays than those with Non-Relatives -- an advantage not evident in the unadjusted results. Such comparisons are non-significant or very minor among the aged and children recipients.
Another finding of interest in this table is that adults in non-White race/ethnic groups tend to have higher odds of hospital stays than Whites. This effect is examined further in subsequent analyses of access to physician services. Also of note is the lower likelihood of hospital use among new IHSS recipients than continuing recipients. This is consistent with the likelihood that a hospital stay in a year increases over time for IHSS recipients if the become more disabled.
Ambulatory Care Sensitive Hospital Admissions. Hospital stays for which the primary admission diagnosis is an ACSC are thought to be indicative of the quality or performance of primary health care (AHRQ, 2007a, 2007b). Better care would be suggested by low rates of these potentially “avoidable” hospital stays. The unadjusted prevalence of ACSC hospital admissions in 2005 is shown in Table 17. Comparing unadjusted “any cause” hospital stays (i.e., Table 15) with the unadjusted ACSC stays shows almost an eight-fold decrease among children and more than 3x decrease among adults using the more restricted ACSC criteria. Differences between provider groups narrow substantially when only ACSC outcomes are considered. Recipients with Spouse IHSS providers continue to have the highest unadjusted hospitalization rate.
Table 18 shows the predicted odds of ACSC hospitalization adjusting for recipient characteristics. Holding other factors constant, there were no statistically significant differences comparing the recipient outcomes of provider groups among children. This finding is consistent with the “any cause” hospital stay comparisons. Among recipients age 18-64, a similar finding also occurs when comparing Spouse and Other Relative providers to Non-Relatives. On the other hand, recipients in this age group with Parent providers have lower adjusted odds for an ACSC hospital stay than Non-Relatives. Finally, among recipients age 65 or more, there is significant difference in the spouse/non-relative comparison. Recipients of Spouse providers have reduced risk of an ACSC hospital stay. There are no statistically significant differences comparing those with Other Relatives to those with Non-Relative providers. (Interaction tests involving provider type with the number of health conditions were non-significant and are not included in the final models.)
Consistent with the “any cause” hospital stays, non-White adult age recipients tend to have increased risk for ACSC admissions. Whether this is a function of differences in access to care, or problems in culturally appropriate care is not known. Among minors, new IHSS recipients have no differences from continuing recipient in ACSC admission. For adults, new recipients have about half the risk of ACSC admission than recipients continuing from 2004.
33. For example, it is likely that Medicare will be the primary payer for health care expenditures by the aged and those non-aged disabled adults eligible for Medicare due to their disability. Similarly, the VA is the primary payer for medical care among qualified veterans if they elect to use VA facilities.