Skip to main content

Table 2 Hazard ratios (HR) of predictor variables used in CVD risk models (training dataset, N = 20,476)

From: Derivation and internal validation of a multi-biomarker-based cardiovascular disease risk prediction score for rheumatoid arthritis patients

Predictors

Univariable models

Multivariable models

HR (CI)

p value

Age + Sex

HR (CI)

Age + Sex + CRP

HR (CI)

Clinical

HR (CI)

Clinical + CRP

HR (CI)

Final MBDA-based model

HR (CI)

p value

Age

1.05 (1.04–1.06)

3.45 × 10−24

1.05 (1.04–1.06)

1.05 (1.04–1.06)

1.04 (1.03–1.05)

1.04 (1.03–1.05)

1.03 (1.02–1.04)

6.68 × 10−11

Male

1.39 (1.15–1.68)

9.29 × 10−4

1.43 (1.19–1.73)

1.43 (1.19–1.73)

1.31 (1.08–1.59)

1.32 (1.09–1.60)

Comorbidities*

 Diabetes

1.49 (1.27–1.74)

1.08 × 10−6

1.31 (1.11–1.54)

1.29 (1.10–1.52)

1.31 (1.11–1.55)

0.0012

 History of CVD

2.01 (1.71–2.35)

1.08 × 10−17

1.47 (1.24–1.74)

1.44 (1.21–1.71)

1.40 (1.18–1.66)

9.85 × 10−5

 Hyperlipidemia

1.36 (1.11–1.66)

0.0023

 Hypertension

2.52 (1.84–3.46)

5.49 × 10−11

1.35 (1.04–1.75)

1.30 (1.00–1.69)

1.31 (1.01–1.71)

0.0405

 Tobacco use

1.38 (1.16–1.65)

4.55 × 10−4

1.42 (1.19–1.70)

1.35 (1.12–1.61)

1.31 (1.09–1.57)

0.0044

Molecular

 MBDA score

1.02 (1.02–1.03)

4.98 × 10−14

 Tanh-MBDA

15.35 (7.17–32.87)

3.70 × 10− 14

4.99 (2.24–11.13)

4.22 × 10−5

 CRP, mg/L

1.20 (1.13–1.26)

7.04 × 10−11

1.22 (1.16–1.29)

1.20 (1.13–1.26)

 EGF, ng/mL

0.92 (0.85–1.00)

0.0650

 IL-6, ng/mL

1.29 (1.20–1.38)

1.92 × 10−11

 Leptin, ng/mL

0.87 (0.81–0.93)

7.93 × 10−5

0.84 (0.79–0.90)

2.99 × 10−6

 MMP-1, ng/mL

1.07 (0.96–1.19)

0.2497

 MMP-3, ng/mL

1.49 (1.35–1.65)

4.43 × 10−14

1.16 (1.03–1.30)

0.0139

 Resistin, ng/mL

1.58 (1.32–1.90)

7.32 × 10−7

 SAA, ng/mL

1.19 (1.13–1.26)

9.38 × 10−10

 TNF-R1, ng/mL

2.83 (2.39–3.35)

8.32 × 10−29

1.77 (1.43–2.19)

3.33 × 10−11

 VCAM-1, ng/mL

3.04 (2.36–3.91)

1.13 × 10−16

 VEGF, ng/mL

1.35 (1.19–1.54)

5.20 × 10−6

 YKL-40, ng/mL

1.60 (1.46–1.76)

1.30 × 10−21

  1. These predictors comprise the complete list of predictors tested by backwards elimination to build the final MBDA-based model; also shown is the adjusted MBDA score, untransformed. P values by the likelihood ratio test
  2. *Based on diagnostic codes during the baseline period (see the “Methods” section)
  3. Natural log transformed
  4. Hyperbolic tangent transformed (tanh [a × MBDA Score [adjusted]], where a = 1/33.0807)
  5. Clinical model includes age + sex + tobacco use + diabetes + hypertension + history of CVD (excluding MI and stroke)
  6. CI 95% confidence interval, CVD cardiovascular disease, MBDA multi-biomarker disease activity