The principal implications of our findings are that community based services can, in theory, have a very substantial impact on aggregate nursing home use when committed solely to this purpose and that existing systems of allocation are technically very inefficient in this, regard. On the one hand this is neither surprising nor necessarily undesirable: most use of community based services reflects individual and family resources and preferences, and arise from decisions that do not have reducing global nursing home use as a purpose in any event. Even in cases where services are managed, case managers have neither necessary information, or necessarily the incentives, to pursue such a goal--even when it is the avowed program purpose. However, our findings indicate that if we want to develop systems of community service management and financing that do more explicitly focus on overall reducing nursing home use, then the theoretical potential to do so is there.
The optimization produced substantial shifts in the aggregate pattern of service use by type. It uses considerably more home nursing services than was actually observed in use, indicating that medically more intensive long-term care services are being underutilized, perhaps in part because their clearly observable higher unit cost outweighs their less readily observable higher unit impact relative to other services. The optimization also shifted much more resources to the opposite end of the care intensity spectrum (housekeeping), while pulling resources out of the middle of the spectrum (home health aide and personal care services). We find the very dramatic shifts here to be prima facie rather implausible, and we believe that what is very likely going on is that these service categories overlapped considerably in their actual measurement as derived from client perceptions (see Appendix A).
Of greater interest is the pattern of changes in service allocations to individuals. The strong shift of resources away from minority elders, for example, make clear that a single-minded focus on reducing aggregate nursing home use may have redistributional consequences that reinforce existing social inequities: an outcome undesirable on other grounds.
On the other hand, it is clear that the optimization makes more explicit than previous studies the advantages to be gained by allocating resources on the basis of classic need-related factors (impairment and less supportive social and economic circumstances). Adhering rigorously to such targeting criteria is a potential source of substantial efficiency gains in using community services to restrain nursing home use.
Finally, we would point out that the work reported here has important weaknesses and imitations that should not be underestimated. This is particularly the case in longitudinal measurement of service use levels, and in several necessary but simplistic assumptions made in the formal models. Our detailed findings should be the impetus for further research and methodological refinement, not the basis for confident conclusions. But we hope that the advances given here in the formal framing of the issues, the methodological approaches to them, and the strong general indication of the presence of important unrealized efficiencies in community care in mitigating risk for nursing home use, will facilitate further constructive work in the area.
| TABLE 1: Baseline Descriptive Statistics for Analysis Sample
|Financial Control Site||0.527||0.499|
|TABLE 2: Maximum Likelihood Estimates of Transition Logits|
|Nursing x Wheelchair||-0.306*||0.283|
|Nursing x No Wheelchair||-0.091||0.320*|
|Home Health x Cognitive Impairment||-0.028**||0.083***|
|Personal Care x ADL||-0.007**||0.013***|
|Housekeeper x IADL||-0.033*||0.125***|
|Housekeeper x No IADL||-0.021||0.044|
|ADL (1=very or extremely severe)||0.537***||-0.146|
|IADL (1=very or extremely severe)||0.501***||-0.862***|
|Cognitive Impairment (1=severe or worse)||0.575***||-0.357**|
|Number of Surviving Children||-0.061***||0.027|
|Treatment Group/Basic Site||-0.086||0.152|
|Treatment Group/Financial Control Site||-0.025||0.427***|
|* p < 0.1
** p < 0.05
*** p < 0.005
|TABLE 3: Distributions of Nursing Home Risk and Hours per Month of Services under Observed and Optimum Service Allocations|
|Nursing Home Risk ()||Nursing Hours||Home Health Hours||Personal Care Hours||Housekeeping Hours|
|Percent Receiving Service||0.15||0.14||7||2||49||0||22||71|
|TABLE 4: Monthly Hours of Services under Observed and Optimal Assignment|
|Nurse|| Home Health
|TABLE 5: Linear Regression of Changes in Risk on Selected Individual Characteristics|
|Number of Surviving Children||0.006**|
|Adjusted R2 = 0.62**
* p < 0.05
** p < 0.0001