Skip to main content

Table 4 Receiver operator curve validation of multivariable models predicting the presence of any abnormal PFTs

From: Respiratory symptoms and disease characteristics as predictors of pulmonary function abnormalities in patients with rheumatoid arthritis: an observational cohort study

 

Full cohort (n = 159)

Cohort excluding reported lung disease (n = 129)

 

AUC

95% CIa

P value

AUC

95% CIa

P value

Symptoms onlyb

0.61

(0.53 to 0.69)

-

0.64

(0.55 to 0.76)

-

Symptoms + other factorsc

0.77

(0.68 to 0.86)

0.0006*

0.79

(0.67 to 0.89)

0.005*

Symptoms + factors + lung diseased

0.80

(0.71 to 0.87)

0.26**

-

-

-

  1. PFT, pulmonary function test. a95% confidence interval (CI) obtained with repeated sampling and calculation of area under the curve (AUC) for 1,000 bootstrap resamples with replacement. bSymptoms included were those from the final predictive multivariable model (chronic phlegm and breathlessness with ambulating 100 yards). cOther factors included were those from the final predictive multivariable model (current smoking, body mass index, presence of anti-cyclic citrullinated peptide antibodies, and current prednisone use). dLung disease includes patient self-report of emphysema, asthma, and rheumatoid lung disease. *P values represent the comparison of AUC functions for the symptoms + other factors model vs. the symptom-only model. **P values represent the comparison of AUC functions for the symptoms + factors + lung disease model vs. the symptoms + factors model.