Unweighted and inverse-probability treatment weighting (IPTW) multivariable regression effects of broad-spectrum antibiotics on secondary outcomes
| Outcome | Unweighted regression eβ (95% CI)# | p-value | IPTW-ATT eβ (95% CI)# | p-value |
| Length of stay | 1.66 (1.53–1.8) | <0.001 | 1.52 (1.41–1.63) | <0.001 |
| Cost | 1.83 (1.68–2.01) | <0.001 | 1.7 (1.57–1.84) | <0.001 |
| Clostridioides difficile infection | 3.85 (1.55–10.93) | 0.006 | 5.79 (1.86–27.51) | 0.008 |
ATT: average treatment effects on the treated. #: exponentiated β coefficients (eβ) are interpreted as multipliers for Γ-distributed variables (length of stay and cost) and as odds ratios for binary outcomes (C. difficile infection).