Table 3

Univariate HRs and results of multiple Cox regression

UnivariateMultiple
Cox regression 1
Multiple
Cox regression 2
HR95% CIHR95% CIHR95% CI
Age at event (by 5 years)1.48(1.25 to 1.75)1.55(1.30 to 1.84)1.57(1.32 to 1.87)
Male1.68(0.84 to 3.34)1.58(0.79 to 3.20)1.45(0.72 to 2.90)
BMI1.00(0.94 to 1.06)
bDMARD failures (reference: 0)
 1 bDMARD1.54(0.67 to 3.52)
 ≥2 bDMARDs0.71(0.17 to 3.00)
DAS28 (current)1.14(0.92 to 1.42)
DAS28 (average last 12 months)1.16(0.91 to 1.48)
DMARD (reference: csDMARDs)
 TNFi0.84(0.39 to 1.80)1.00(0.46 to 2.20)1.04(0.48 to 2.26)
 Other bDMARDs0.40(0.09 to 1.78)0.41(0.09 to 1.84)0.33(0.08 to 1.44)
 Tocilizumab4.17(1.87 to 9.27)5.11(2.31 to 11.3)4.48(2.01 to 9.99)
GCs
 Current GC (by 5 mg)1.22(1.13 to 1.31)1.28(1.18 to 1.38)
 Cumulative GCs*1.81(1.47 to 2.22)1.87(1.50 to 2.33)
NSAIDs
 Current NSAID1.80(0.92 to 3.53)2.18(1.11 to 4.31)
 Cumulative NSAIDs†2.71(1.20 to 6.12)3.00(1.33 to 6.82)
  • In Cox regression 1, we adjusted for current doses of GCs and NSAIDs, whereas in Cox regression 2 for cumulative doses.

  • *Concomitant GC use (range: 0 to 1) was calculated for each patient as the area under curve of follow-up month with medium doses (>5 to 10 mg/day, weight of 0.5) plus follow-up month with high doses (>10 mg/day, weight of 1) and then divided by total no. of follow-up months.

  • †Cumulative treatment with NSAIDs (range: 0 to 1) was calculated for each patient as: no. of follow-ups with concomitant NSAIDs use divided by the total no. of follow-ups.

  • bDMARD, biologic disease-modifying anti-rheumatic drug; BMI, body mass index; csDMARDs, conventional synthetic disease-modifying anti-rheumatic drugs; DAS28, disease activity score based on 28 joints; GC, glucocorticoid; NSAIDs, non-steroidal anti-inflammatory drugs; other bDMARDs, abatacept and rituximab; TNFi, tumour necrosis factor-α inhibitor.