Table 2 Association between CSS phenotype and ANCA status (dependent variable)
ANCA positive (n = 26)ANCA negative (n = 22)ANCA not recordedp Value (univariate analysis)OR (logistic regression analysis, 95% CI)p Value (multivariate logistic regression)OR (multivariate logistic regression, 95% CI)
LTA type
    Pranlukast12 (80%)3 (20%)10.04†6.7 (1.5 to 30)*0.334.6 (0.4 to 48)*
    Zafirlukast5 (56%)4 (44%)82.1 (0.4 to 9.8)*0.204.5 (0.1 to 194)*
    Montelukast9 (38%)15 (62%)51.00.431.0
Organ
    Lung17 (46%)20 (54%)110.050.2 (0.04 to 0.99)0.040.12 (0.01 to 0.98)
    Heart3 (38%)5 (62%)80.30.4 (0.09 to 2.1)0.290.12 (0.003 to 5.8)
    Nerves19 (68%)9 (32%)90.023.9 (1.1 to 13.2)0.198.6 (0.3 to 225)
    Skin13 (72%)5 (28%)60.053.4 (1.05 to 12.0)0.0114.3 (1.8 to 110)
Interval between LTA use and CSS (mean (SEM))6.7 (2.3)5.9 (1.0)0.8
No of organs affected (mean (SEM))2.2 (0.2)2.2 (0.1)0.950.11
  • Univariate analysis was performed using the χ2 test with logistic regression (with a single predictor variable) used to generate ORs.

  • Using multivariate logistic regression analysis, lung involvement was independently associated with a negative ANCA test and skin involvement with a positive ANCA, but the association with LTA type and neurological involvement was lost.

  • *Relative to montelukast. †For all three drugs.

  • ANCA, antineutrophil cytoplasmic antibodies status; LTA, leukotriene antagonist; CSS, Churg–Strauss Syndrome.