Table 4.
Nonlinear and linear decomposition of home healthcare use and expenditures. Elderly Medicare beneficiaries with and without PD. Medicare 5% claims database (2006, 2007).
| % with home healthcare use (PD) % with home healthcare use (no PD) |
22.5%
9.2% |
|
|
| ||
| Average log-transformed home healthcare expenditures (PD) | 8.44 | |
| Average log-transformed home healthcare expenditures (no PD) | 8.09 | |
|
| ||
| Difference | 13.3 | 0.35 |
|
| ||
| Total explained | 3.64 | 0.063 |
|
| ||
| Nonlinear decomposition | Linear decomposition | |
|
| ||
| Predisposing characteristics | ||
| Gender, race/ethnicity, age | 5.3% | 1.8% |
| Enabling characteristic | ||
| Public assistance | −1.5% | 1.6% |
| Need characteristic | ||
| Number of comorbidities | 17.5% | 4.1% |
| Personal health choices | ||
| Any type of SUD Baseline resource use |
6.0% | 12.6% |
| External environmental factors | ||
| Census region Metro status |
0.2% | −2.1% |
|
| ||
| Explained to total difference | 27.5% | 18.0% |
| Unexplained portion | 72.5% | 82.0% |
Note: nonlinear decomposition analysis based on 10,865 and 50,009 elderly (age 65 or older) Medicare beneficiaries with and without PD.
Difference in home healthcare use = 13.3 percentage points. Percentage points in home healthcare use are explained by each independent variable. The percentage explained is derived by dividing the total explained portion by the 13.3-percentage-point difference between elderly Medicare beneficiaries with and without PD.
Linear decomposition analysis based on 2,445 and 4,601 elderly (age 65 or older) Medicare beneficiaries with and without PD and who used home healthcare.
SUD: substance use disorder; Exp.: expenditure.