Skip to main content
. 2020 Apr 10;59(9):2207–2216. doi: 10.1093/rheumatology/keaa113

Table 3.

Potential predictors of the specific and overall treatment effect in one-stage IPD-MA

Predictor N Specific effect
N Overall effect
n β (95% CI) n β (95% CI)
Sex (female = 1, male =0) 11 2939 4.28 (7.98,0.58) 15 1857 3.17 (5.53,0.81)
Age, years 11 2938 0.06 (−0.11, 0.22) 15 1857 −0.02 (−0.13, 0.09)
Baseline pain (0 − 100) 11 2939 0.05 (−0.06, 0.16) 15 1857 0.53 (0.59,0.47)
BMI, kg/m2 10 2863 0.03 (−0.27, 0.32) 12 1633 0.09 (−0.12, 0.29)
Inflammation (yes = 1, no = 0) 7 2339 −0.31 (−4.98, 4.36) 8 1329 −2.16 (−5.42, 1.09)
 Clinical 4 1550 −1.84 (−8.24, 4.55) 5 924 −3.03 (−7.40, 1.34)
 Biochemical 6 2116 0.90 (−1.53, 3.33) 6 1088 1.18 (−0.55, 2.90)
Duration, months 3 280 −0.04 (−0.15, 0.07) 4 181 −0.03 (−0.09, 0.03)
XR severity (0–100) 7 2576 −0.01 (−0.12, 0.14) 9 1412 0.08 (−0.00, 0.17)

β presented on a 0–100 scale. Significant interactions (specific effect) or associations (overall effect, unadjusted model) are shown in bold. A negative interaction effect for sex (specific effect) indicates a greater specific effect for topical NSAIDs in women relative to men. A negative association between sex and overall treatment effect indicates more pain reduction in women relative to men. A negative association between baseline pain and overall treatment effect indicates that the higher the baseline pain score, the more reduction in pain.

β: beta-coefficient for interaction effect (specific effect) or covariate (overall effect); n: number of participants; N: number of trials; XR: radiographic severity.