Table 2

Multivariate models predicting new carotid plaque formation in nondiabetic patients with rheumatoid arthritis over 3.6 ± 0.2 years^a

New carotid plaque formation during follow-up (yes vs no)

Parameters

β coefficient

P value

Odds ratio (95% CI)

Model 1: traditional CV risk factors (r² = 0.0453)

Age at follow-up end (years)

0.146

0.004

5.752^b (1.770 to 18.694)

Sex

0.038

0.970

0.963^b (0.134 to 6.927)

Total years of smoking (at follow-up end)

0.051

0.023

2.628^b (1.145 to 6.029)

SBP AUC (mmHg × months)

0.001

0.346

1.495^b (0.648 to 3.449)

LDL AUC (mg/dl × months)

0.000

0.126

0.507^b (0.212 to 1.211)

BMI (at follow-up end)

0.088

0.356

1.445^b (0.661 to 3.160)

Model 3 RA treatment (months of treatment during follow-up) (r² = 0.248)

Corticosteroids

0.067

0.016

3.058 (1.231 to 7.593)

Biologic agents

-0.035

0.046

0.502 (0.255 to 0.987)

Nonbiologic DMARDs

-0.017

0.477

0.802 (0.437 to 1.473)

Final model (r² = 0.559)^c

Age at follow-up end (years)

0.136

0.006

5.098 (1.599 to 16.250)

Total years of smoking (at follow-up end)

0.055

0.009

2.875 (1.305 to 6.335)

Corticosteroids (months during follow-up)

0.086

0.016

4.170 (1.299 to 13.385)

Biologic agents (months during follow-up)

-0.033

0.118

0.520 (0.230 to 1.180)

^aAUC: area under the curve, BMI: body mass index, CV: cardiovascular, DMARDs: disease modifying antirheumatic drugs, LDL: low-density lipoprotein, RA: rheumatoid arthritis, SBP: systolic blood pressure. ^cOnly the variables from models 1 to 3 that predicted new plaque formation were used in the final model. Model 2 is described in the text only. ^bThe odds ratio for the continuous values was assessed for 1 SD increase and so is provided by exp(β*SD) (SD for age 12.0, for total years of smoking 19.15, for SBP (AUC) 629.10, for LDL (AUC) 1343.77, for BMI 4.25, for corticosteroid treatment (months during follow-up) 16.60, for biologic agents received (months during follow-up) 19.90, for nonbiologic DMARDs (months during follow-up) 13.20). Dependent variable 0 is subjects with no new plaque (n = 45), variable 1 is subjects with at least one plaque (n = 19) at the end of follow-up (binary logistic regression analysis using enter method is presented, and identical results in all models were obtained using the backward method).

Zampeli et al. Arthritis Research & Therapy 2012 14:R44   doi:10.1186/ar3757