Table 3. Adjusted Treatment Persistence and Risk of Discontinuation Among All Patients (N=265)a.
| Persistence at 12 Months: OR (95% CI) | P-Value | Risk of All-Cause Discontinuation: HR (95% CI) | P-Value | Risk of Discontinuation for Disease Progression | P-Value | |
| Clinic | ||||||
| 1 vs. 6 | 3.208 (0.839 - 12.275) | 0.089 | 0.459 (0.214 - 0.985) | 0.046 | 0.472 (0.144 - 1.547) | 0.215 |
| 2 vs. 6 | 1.679 (0.522 - 5.396) | 0.385 | 0.481 (0.242 - 0.957) | 0.037 | 0.484 (0.163 - 1.436) | 0.191 |
| 3 vs. 6 | 14.754 (3.440 - 63.274) | <0.001 | 0.286 (0.144 - 0.569) | <0.001 | 0.192 (0.060 - 0.615) | 0.006 |
| 4 vs. 6 | 3.656 (1.076 - 12.425) | 0.038 | 0.299 (0.146 - 0.614) | 0.001 | 0.016 (0.002 - 0.150) | <0.001 |
| 5 vs. 6 | 1.762 (0.397 - 7.809) | 0.456 | 0.562 (0.236 - 1.339) | 0.194 | 0.930 (0.273 - 3.167) | 0.908 |
| Male vs. Female | 0.508 (0.255 - 1.010) | 0.053 | 1.216 (0.826 - 1.791) | 0.321 | 1.280 (0.688 - 2.380) | 0.435 |
| Age, Years | 1.013 (0.988 - 1.039) | 0.319 | 1.002 (0.988 - 1.016) | 0.803 | 0.996 (0.973 - 1.019) | 0.711 |
| CCI | 1.003 (0.731 - 1.377) | 0.985 | 1.012 (0.842 - 1.216) | 0.898 | 0.982 (0.730 - 1.321) | 0.903 |
| Years from RA Diagnosis | 0.988 (0.951 - 1.027) | 0.540 | 1.010 (0.987 - 1.033) | 0.408 | 1.018 (0.981 - 1.056) | 0.341 |
| No. Physician Office Visits (1-year pre-index) | Not significant | Not significant | Not significant | |||
| No. Hospitalizations (1-year pre-index) | Not significant | Not significant | Not significant | |||
| Abatacept vs TNFi | 1.980 (0.943 - 4.167) | 0.071 | 0.584 (0.384 - 0.887) | 0.012 | 0.293 (0.138 - 0.620) | 0.001 |
Abbreviations: CCI, Charlson comorbidity index; CI, confidence interval; HR, hazard ratio; OR, odds ratio; RA, rheumatoid arthritis; TNFi, tumor necrosis factor inhibitor.
a The initial models included age, sex, CCI, years from RA diagnosis to index, and cohort as independent variables. We then used a forward selection method to include additional significant covariates (P<0.05) for the final models. The following covariates were considered: number of physician office visits (1-year pre-index), number of hospitalizations (1-year pre-index), and site.