Table 3.
Interactions Between Age and HOUSES and outcomes using multivariable Cox, linear, and logistic regression
| Age <50 N=740 |
Age ≥ 50 N=2683 |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| Outcome | N | Outcome | estimate † | p- value † |
N | Outcome | estimate † | p- value † |
Inter actio n p- valu e ‡ |
| Time to With-in Hospital Mortality | N | N Event (%) | Hazard Ratio | 0.07 | N | N Event (%) | Hazard Ratio | 0.01 | 0.01 |
| HOUSES Q1 | 243 | 4 (1.7%) | 1.00 (ref) | 595 | 83 (14.0%) | 1.00 (ref) | |||
| HOUSES Q2-Q4 | 497 | 24 (4.8%) | 2.70 (0.93, 7.87) | 2043 | 210 (10.3%) | 0.72 (0.56, 0.93) * | |||
| ICU Free Days | N | Mean (SD) | Mean Estimate | 0.005 | N | Mean (SD) | Mean Estimate | 0.004 | <.01 |
| HOUSES Q1 | 243 | 26.0 (4.1) | 0.00 (ref) | 595 | 22.3 (9.1) | 0.00 (ref) | |||
| HOUSES Q2-Q4 | 497 | 25.0 (6.1) | −1.16 (−1.96, −0.36) ** | 2043 | 23.1 (8.2) | 1.08 (0.34, 1.82) ** | |||
| Hospital Free Days | N | Mean (SD) | Mean Estimate | 0.13 | N | Mean (SD) | Mean Estimate | 0.002 | <.01 |
| HOUSES Q1 | 243 | 23.1 (6.1) | 0.00 (ref) | 595 | 18.4 (9.3) | 0.00 (ref) | |||
| HOUSES Q2-Q4 | 497 | 22.6 (6.9) | −0.73 (−1.68, 0.21) | 2043 | 19.2 (8.5) | 1.20 (0.45, 1.96) ** | |||
| ICU Readmission | N | N Event (%) | Odds Ratio | 0.23 | N | N Event (%) | Odds Ratio | 0.10 | 0.50 |
| HOUSES Q1 | 243 | 10 (4.1%) | 1.00 (ref) | 595 | 59 (9.9%) | 1.00 (ref) | |||
| HOUSES Q2-Q4 | 497 | 12 (2.4%) | 0.59 (0.24, 1.41) | 2043 | 170 (8.3%) | 0.77 (0.56, 1.06) | |||
*
P <.05
**
P <.01
†
Estimate and P-value are adjusted for all covariates listed in the table
‡
P-value is the interaction p-value between HOUSES and AGE. If P<.05, it means the estimate for HOUSES is significantly modified by age <50 vs. >=50. P-values comes from a model on the full sepsis cohort, adjusted for the effects of HOUSES, age, sex, race, APACHE, and the HOUSES*age interaction.