Table 3. Multivariable proportional hazards regression models of 30-d readmission.
| Variable | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|
| aHR | 95% CI | p-Value* | aHR | 95% CI | p-Value* | |
| Outpatient IV support | 0.002 | 0.003 | ||||
| H-OPAT | 1.00 (ref) | 1.00 (ref) | ||||
| S-OPAT | 0.59 ‡ | 0.42–0.82 | 0.002 | 0.53 † ‡ | 0.35–0.81 | 0.003 |
| Healthcare funding source | 0.001 | 0.001 | ||||
| Medicare, private insurance, charity | 1.00 (ref) | 1.00 (ref) | ||||
| Self-pay | 1.75 | 1.25–2.47 | 0.005 | 1.64 | 1.15–2.32 | 0.006 |
| Medicaid | 1.62 | 1.15–2.28 | 0.001 | 1.74 | 1.21–2.49 | 0.003 |
| Type of infection requiring IV antimicrobials | 0.001 | 0.001 | ||||
| Bone/joint, skin/soft tissue, intra-abdominal, genitourinary | 1.00 (ref) | 1.00 (ref) | ||||
| Bacteremia | 1.43 | 1.03–1.99 | 0.05 | 1.43 | 1.03–1.99 | 0.03 |
| Central nervous system | 0.32 | 0.10–0.99 | 0.04 | 0.31 | 0.10–0.97 | 0.04 |
| Pulmonary/ENT | 0.43 | 0.19–0.98 | 0.05 | 0.44 | 0.19–1.02 | 0.06 |
| Other type/site | 0.55 | 0.30–0.99 | 0.003 | 0.54 | 0.29–0.98 | 0.04 |
| Chronic renal insufficiency | 0.003 | 0.003 | ||||
| No | 1.00 (ref) | 1.00 (ref) | ||||
| Yes | 1.72 | 1.21–2.46 | 0.003 | 1.74 | 1.21–2.51 | 0.003 |
| Propensity score (quintiles) | 0.12 | |||||
| 1 | 1.00 (ref) | |||||
| 2 | 1.55 | 0.40–1.05 | 0.08 | |||
| 3 | 1.09 | 0.53–1.57 | 0.75 | |||
| 4 | 1.28 | 0.44–1.39 | 0.40 | |||
| 5 | 0.89 | 0.60–2.08 | 0.72 | |||
Model 1 controls for confounding with covariates; model 2 controls for confounding with covariates and for selection bias with the propensity score.
*The p-values for the main category terms are the effects from the type 3 tests, and those for the individual category terms are the maximum likelihood estimates, all based on a sample size of 1,168.
‡Replication of the two models with multiple logistic regression analysis gave similar results for all estimates; specifically, the odds ratio for S-OPAT was 0.59 (95% CI 0.40–0.86) in model 1 and 0.55 (95% CI 0.34–0.89) in model 2.
†Reanalysis after excluding patients in quintiles 4 and 5 of the categorical propensity score gave an aHR for S-OPAT of 0.51 (95% CI 0.33–0.79; p = 0.003). When the continuous propensity score was used in the model as a quadratic effect, the aHR for S-OPAT was 0.52 (95% CI 0.34–0.80; p = 0.003).
ENT, ear/nose/throat; ref, referent category.