Table 3. Multivariable Logistic Model for Successful Opioid Modification at 1 Yeara.
| Outcome | Odds ratio (95% CI) | P value |
|---|---|---|
| Age per 10 y | 1.06 (0.98-1.15) | .13 |
| Sex | ||
| Male | 1 [Reference] | .10 |
| Female | 0.83 (0.67-1.04) | |
| Instrumentation | 0.86 (0.68-1.10) | .22 |
| Charlson Comorbidity Index | ||
| 0 | 1 [Reference] | <.001 |
| 1 | 0.67 (0.50-0.90) | |
| 2-3 | 0.61 (0.45-0.83) | |
| ≥4 | 0.45 (0.32-0.63) | |
| Preoperative opioid availability | ||
| No | 1 [Reference] | <.001 |
| Short-term | 0.61 (0.48-0.77) | |
| Episodic | 0.95 (0.64-1.40) | |
| Long-term | 0.49 (0.30-0.82) | |
| Depression | 0.89 (0.70-1.13) | .32 |
| Tobacco use | 0.84 (0.67-1.05) | .13 |
| Anxiety | 0.76 (0.58-0.99) | .04 |
| Discharge location | ||
| Home | 1 [Reference] | .06 |
| Not home | 0.66 (0.47-0.94) | |
| Not documented | 1.02 (0.50-2.08) |
a
Only patients living in the catchment area at 1 year after discharge were included in the analysis (n = 2148). In addition to the covariables presented in the table, to account for variability in prescribing practices across the time frame of the study, the model was adjusted for date of surgery using restricted cubic splines. Model concordance was 0.64, and there was no evidence of significant lack of fit (Hosmer Lemeshow goodness of fit test, P = .38).