Table 2.

Effects of possible risk factors on self-reported incident AS, by multivariate logistic regression.

OR95% CIp
A. Main model (n = 98 patients and 33,127 controls)i
  Female sex1.380.91–2.090.13
  Ageii0.960.94–0.97< 0.001
  Previous CVD1.020.24–4.270.98
    Never smoker1Reference
    Former smoker1.150.66–2.020.62
    Current smoker1.991.28–3.110.002
B. Smoking represented as total pack-yrs (n = 100 patients and 32,573 controls)i,iii
  Female sex1.440.95–2.190.09
  Ageii0.950.93–0.97< 0.001
  Total pack-yrsii1.010.99–1.040.21
C. Smoking represented as change in smoking habits from HUNT2 to HUNT3 (n = 95 patients and 30,678 controls)i,iii
Female sex1.140.75–1.740.53
  Ageii0.950.93–0.97< 0.001
  Change in smoking habits
    Never smoker1Reference
    Quit smoking1.470.91–2.400.12
    Started smoking or continuous smoker1.921.16–3.170.012
  • i Cases with complete data. No. missing for control cases: sex and age: 0; BMI: 132 (0.4%); diabetes 33 (0.1%); HTN: 122 (0.4%); previous CV disease: 107 (0.3%); never/present/former smoker: 1876 (5.3%); pack-yrs HUNT2: 2432 (6.9%); change in smoking habits HUNT2 to HUNT3: 4334 (12.3%). No. missing for incident AS cases: sex, age, BMI, diabetes, HTN, previous CV disease: 0; never/present/former smoker: 9 (8.4%); pack-yrs HUNT2: 7 (6.5%); change in smoking habits HUNT2 to HUNT3: 12 (11.2%). No. missing for the smoking-related variables was not significantly different between cases and controls (p > 0.15).

  • ii Continuous variable, OR per 1 unit change.

  • iii The model was also adjusted for BMI, diabetes, and previous CV disease. AS: ankylosing spondylitis; BMI: body mass index; HTN: hypertension; CVD: cardiovascular disease.