Skip to main content
. 2023 Feb 21;82(5):611–620. doi: 10.1136/ard-2022-222519

Table 2.

Comparison of different logistic regression models for response to TNFi, comparing grouped responders (moderate and good) and non-responders at 3-month follow-up, dichotomised based using high (75th) and low (25th) concentrations of MRP8/14 or CRP

Responses categories calculated using:
4-component DAS28 3-component DAS28 2-component DAS28
Model Cases OR (95% Cl) Model fit (pseudo R2) P value Cases OR
(95% Cl)
Model fit (pseudo R2) P value Cases OR (95% Cl) Model fit (pseudo R2) P value
Low vs high MRP8/14 (µg/mL) 218 1.22
(0.62 to 2.38)
0.0015 0.564 241 1.92
(1.04 to 3.54)
0.017 0.037 241 2.03
(1.09 to 3.78)
0.020 0.025
Low vs high MRP8/14 adjusted for pre-treatment CRP 218 1.10
(0.53 to 2.30)
0.0033 0.795 241 1.53
(0.78 to 2.98)
0.026 0.21 241 1.60
(0.82 to 3.15)
0.030 0.171
Low vs high CRP only (mg/L) 215 1.64
(0.78 to 3.4)
0.0092 0.188 234 3.79
(1.81 to 7.93)
0.062 <0.001 234 3.58
(1.74 to 7.35)
0.058 <0.001
Low vs high CRP adjusted for pre-treatment MRP8/14 215 1.79
(0.79 to 4.06)
0.011 0.163 234 4.11
(1.83 to 9.25)
0.063 0.001 234 3.73
(1.70 to 8.19)
0.059 0.001

Response categorises were calculated using 4 components, 3 components or 2 components.

CRP, C reactive protein; DAS28, disease activity scores in 28-joints; TNFi, tumour necrosis factor inhibitor.