Skip to main content
. 2015 Apr 27;2015:279890. doi: 10.1155/2015/279890

Table 3.

Logistic regression and propensity score-based analyses to predict treatment response with tocilizumab versus TNFi.

N Remission Low disease activity
Odds ratio 95% CI P value Odds ratio 95% CI P value
DAS28 < 2.6 DAS28 < 3.2
Multivariate LR 489 13.3 6.9–25.4 <0.001 7.3 3.9–13.6 <0.001
Univariate LR + PS 489 8.3 4.4–15.4 <0.001 5.5 2.9–10.5 <0.001
Multivariate LR + PS 489 11.0 5.6–21.6 <0.001 6.2 3.2–12.0 <0.001
PS matching 259 7.9 4.3–14.6 <0.001 5.0 2.7–9.2 <0.001
PS matching + multivariate LR 259 12.3 6.0–25.4 <0.001 7.5 3.7–15.1 <0.001

CDAI ≤ 2.8 CDAI ≤ 10
Multivariate LR 308 2.9 1.3–6.6 0.009 3.0 1.5–6.1 0.002
Univariate LR + PS 308 2.5 1.1–5.6 0.022 2.3 1.1–4.6 0.019
Multivariate LR + PS 308 2.8 1.2–6.5 0.016 2.6 1.3–5.5 0.010
PS matching 179 2.1 0.92–5.0 0.078 2.0 1.00–4.1 0.048
PS matching + multivariate LR 179 3.3 1.3–8.4 0.015 2.6 1.2–5.6 0.017

SDAI ≤ 3.3 SDAI ≤ 11
Multivariate LR 282 4.1 1.7–9.5 0.001 2.9 1.4–6.3 0.005
Univariate LR + PS 282 3.1 1.3–7.0 0.008 2.2 1.04–4.8 0.038
Multivariate LR + PS 282 3.6 1.5–8.7 0.005 2.5 1.1–5.5 0.024
PS matching 158 2.6 1.1–6.4 0.029 1.6 0.8–3.5 0.209
PS matching + multivariate LR 158 4.0 1.5–10.8 0.007 2.2 0.95–5.0 0.065

Boolean
Multivariate LR 442 2.1 0.91–4.8 0.083
Univariate LR + PS 442 1.8 0.76–4.0 0.184
Multivariate LR + PS 442 1.9 0.77–4.8 0.159
PS matching 216 1.2 0.54–2.9 0.629
PS matching + multivariate LR 216 1.4 0.54–3.9 0.463

Good EULAR Good/moderate EULAR
Multivariate LR 489 6.8 3.8–12.3 <0.001 2.5 1.1–5.9 0.035
Univariate LR + PS 489 6.3 3.4–11.8 <0.001 1.8 0.8–4.1 0.143
Multivariate LR + PS 489 6.4 3.4–12.0 <0.001 1.8 0.8–4.5 0.182
PS matching 259 6.2 3.3–11.6 <0.001 2.2 0.93–5.2 0.074
PS matching + multivariate LR 259 7.8 4.0–15.4 <0.001 2.4 0.98–6.1 0.056

The odds ratio and 95% confidence interval (95% CI) for the effect of tocilizumab versus TNF inhibitors (TNFi) in the considered outcomes are represented according to the statistical methodology used. Multivariate logistic regression (LR) adjusted for other significant confounders as described in Section 2, namely, age, disease duration, number of previous biologics, and baseline disease activity (DAS28 for DAS28/Boolean remission, DAS28 low disease activity (LDA), and EULAR response; CDAI for CDAI remission/LDA; SDAI for SDAI remission/LDA). Propensity scores (PS) predicting biologic class were calculated and incorporated in the analysis, via LR and/or matching (caliper 1 : 5 with replacement). P value significant at <0.05; significant differences highlighted in bold.