Table 6.
Multivariate analysis∗ of predictors of veterans' utilization of inpatient hospitalization (n = 195).
| Logistic regression (n = 296) | Negative binomial (n = 104) | |||||
|---|---|---|---|---|---|---|
| Estimate | P value | Confidence interval | Estimate | P value | Confidence interval | |
| Patient | ||||||
| Marital status | 0.493 | 0.029 | 0.261–0.930 | — | — | — |
| Personal care dependency | 1.048 | 0.007 | 1.029–1.197 | 0.040 | 0.035 | 0.003–0.085 |
| Behavior problems | — | — | — | 0.044 | 0.157 | −0.015–0.095 |
| VA site | ||||||
| Houston versus Providence (1 versus 5) | 1.048 | 0.4556 | 0.535–2.054 | −0.152 | 0.487 | −0.579–0.276 |
| Boston versus Providence (2 versus 5) | 0.542 | 0.2590 | 0.194–1.509 | 0.554 | 0.034 | 0.041–1.067 |
| Oklahoma City versus Providence (3 versus 5) | 0.519 | 0.2541 | 0.171–1.575 | −0.445 | 0.309 | −1.302–0.412 |
| Beaumont versus Providence (4 versus 5) | 1.575 | 0.0163 | 0.849–2.921 | 0.134 | 0.458 | −0.220–0.487 |
*Controlled for site.