Skip to main content
. 2010 May 27;12(3):R104. doi: 10.1186/ar3037

Table 3.

Multivariate associations of pulmonary symptoms, cardiac findings, and patient characteristics with any PFT abnormality

Characteristic Model 1 (null) Model 2 (complex) Model 3a (simplified) P value



OR 95% CI OR 95% CI OR 95% CI
Frequent cough 1.61 (0.49 to 5.25) 2.05 (0.41 to 10.3)
Chronic phlegm 2.64 (0.87 to 8.02) 3.01 (0.56 to 16.3) 4.16 (1.21 to 14.4) 0.02
Frequent wheezing 2.84 (0.74 to 10.9) 2.12 (0.37 to 12.1)
Breathlessness at 100 yards 2.24 (0.60 to 8.32) 11.2 (1.54 to 81.6) 5.96 (1.25 to 28.3) 0.03
Reported diagnosis of pulmonary disease 4.57 (1.35 to 15.4) 3.77 (1.42 to 10.0) 0.01
CAC >100 1.25 (0.38 to 4.11)
Age, per year 1.04 (0.97 to 1.12)
Male vs. female 0.87 (0.27 to 2.84)
White vs. other race 0.69 (0.20 to 2.31)
Reported exercise, per quartile 0.95 (0.57 to 1.57)
BMI, per kg/m2 0.91 (0.82 to 1.00) 0.93 (0.85 to 1.01) 0.08
Current smoking 5.34 (1.12 to 25.4) 4.01 (1.22 to 13.3) 0.02
Ever smoking vs. never smoking 0.73 (0.23 to 2.31)
RA duration, per year 1.01 (0.95 to 1.07)
RF seropositivity 1.53 (0.44 to 5.29)
Anti-CCP seropositivity 2.99 (0.69 to 12.9) 2.82 (0.96 to 8.29) 0.06
DAS28, per unit 0.78 (0.45 to 1.33)
HAQ, per unit 0.86 (0.35 to 2.10)
Log-modified Sharp score, per log unit 1.16 (0.71 to 1.91)
Current prednisone use 2.54 (0.95 to 6.85) 2.65 (1.12 to 6.25) 0.03
Current nonbiologic use 0.38 (0.08 to 1.88)
Current biologic use 0.82 (0.27 to 2.52)

Model 1 (null), pulmonary symptoms only; Model 2 (complex), multivariable model with extended covariate list; Model 3, multivariate model with covariates reduced based on hierarchical modeling. BMI, body mass index; CAC, coronary artery calcium; CCP, cyclic citrullinated peptide; CI, confidence interval; DAS28, RA Disease Activity Score in 28 joints; HAQ, Health Assessment Questionnaire; OR, odds ratio; PFT, pulmonary function test; RA, rheumatoid arthritis; RF, rheumatoid factor. aP value for the likelihood ratio test comparing Models 2 and 3 = 0.801, indicating that the simpler model was not statistically different in predicting the variance in the outcome than the more complex model.