Table 5.
Baseline predictors of anti-TNF-α treatment discontinuation
| Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | ||
| Age (yr)†a | 0.994 (0.975 to 1.014) | 0.561 | ** | ||
| Gender | Female | 1 | - | - | |
| Male | 0.503 (0.321 to 0.787) | 0.003 | 0.406 (0.251 to 0.657) | 0.000 | |
| Duration of symptoms (yr)† | 0.981 (0.959 to 1.002) | 0.082 | ** | ||
| HLA-B27 | Negative | 1 | - | - | |
| Positive | 0.823 (0.468 to 1.448) | 0.500 | *** | ||
| Peripheral arthritis | Absent | 1 | - | - | |
| Present | 0.382 (0.176 to 0.830) | 0.015 | 0.320 (0.144 to 0.712) | 0.005 | |
| BASDAI (range 0 to 10)‡b | 1.162 (1.016 to 1.329) | 0.028 | 1.225 (1.053 to 1.424) | 0.008 | |
| ASDAS‡b | 1.005 (0.759 to 1.330) | 0.974 | ** | ||
| Physician's GDA (range 0 to 10)‡b | 0.907 (0.816 to 1.008) | 0.070 | ** | ||
| Patient's GDA (range 0 to 10)‡a | 1.075 (0.958 to 1.208) | 0.219 | ** | ||
| ESR (mm/h)‡a | 0.987 (0.974 to 0.999) | 0.039 | 0.983 (0.969 to 0.997) | 0.018 | |
| CRP (mg/l)‡b | 0.986 (0.972 to 1.000) | 0.049 | * | ||
| BASFI (range 0 to 10)‡ | 1.045 (0.935 to 1.168) | 0.438 | *** | ||
| Chest expansion (cm)‡a | 0.986 (0.903 to 1.076) | 0.753 | ** | ||
| Modified Schober test (cm)‡ | 1.189 (1.036 to 1.365) | 0.014 | ** | ||
| Occiput to wall distance (cm)‡a | 0.971 (0.938 to 1.006) | 0.971 | ** | ||
| Lateral lumbar flexion L (cm)‡ | 1.018 (0.973 to 1.066) | 0.434 | *** | ||
| Lateral lumbar flexion R (cm)‡ | 1.016 (0.971 to 1.062) | 0.498 | *** | ||
| TNF-α blocking agent | ETA | 1 | - | - | |
| IFX | 0.847 (0.441 to 1.627) | 0.618 | *** | ||
| ADA | 1.334 (0.769 to 2.314) | 0.305 | *** | ||
See Table 1 for definitions.
HR refers to the risk of anti-TNF-α treatment discontinuation: † per year; ‡ per 1 grade or 1 point.
a Significant difference (P < 0.05) between men and women at baseline.
b Significant difference (P < 0.05) between patients with peripheral arthritis (defined as at least one swollen joint) and only axial disease at baseline.
* CRP was not selected during forward conditional logistic regression due to the strong correlation with ESR (ρ = 0.669, P = 0.000) and the significant difference in CRP level between patients with and without peripheral arthritis (median 17 vs. 12, P = 0.014). Although, lower CRP level (HR: 0.984, 95% CI: 0.969 to 0.999) was also a significant predictor of treatment discontinuation in the presence of gender and BASDAI.
** The variable was not selected during multivariate regression analysis (P ≥0.05).
*** The variable was not tested in multivariate regression analysis because of a P-value > 0.3 in univariate regression analysis and no significant difference between men and women at baseline.