Background/Purpose:   NSAIDs are the first line pharmacologic therapy in ankylosing spondylitis (AS). Several NSAIDs are efficacious in AS without superiority demonstrated by any particular drug.  The literature frames NSAID use in terms of type, dose and duration through the NSAID index. This study aims to identify the effect of NSAID use in AS as defined by the percent of the full inflammatory dose taken over time (the NSAID Index) on disease activity over time.

Methods: A prospective cohort study of 755 AS patients who met the modified New York criteria were followed for at least 2 visits from multiple sites. Patients underwent a comprehensive clinical evaluation including assessing their disease activity by the Bath AS Disease Activity Index (BASDAI) in addition other demographic, social, and psychological variables collected. . Medications taken concurrently, inflammatory markers and radiographs were taken at each visit.  NSAID use was captured by self-report of numbers of pills taken in the last week, month, and 6 month interval. Longitudinal multivariable analyses using mixed effect Poisson regression models that account for the correlation of repeated measures over time were conducted to assess associations between NSAID use (NSAID index of 0%, >0 & ≤50%, and >50%) and BASDAI in tandem with other covariables..

Results: Of the 5101 visits included in this analysis, 41%, 29% and 26% had an NSAID index of 0, >0 & ≤50, and >50 respectively. Table 1. show the longitudinal associations between NSAID use and other covariables over a 6 months period on BASDAI scores.  We found that NSAID use was significantly related to higher BASDAI (overall p<0.0001); the patients who had high NSAID index (>50%) and low NSAID index (>0 and ≤50%) were more likely to have high BASDAI compared to the ones with no NSAID use (ARR=1.21; p<0.0001 and ARR=1.15; p<0.0001, respectively).  This effect was attenuated when we compare patients with high versus low NSAID index (>50% vs. >0 and ≤50%) (ARR= 1.05; p=0.0809). Smoking, C-reactive protein (CRP), baseline radiographic disease, depression (defined by Center for Epidemiologic Studies Depression Scale (CESD) score>16) and opioid use were positively associated with disease activity.  Negative associations were seen with TNF inhibitors (TNFi), male gender, higher education. associated with disease activity (ARR=0.9; p=0.007). Nearly identical results were found if only the NSAID index of the previous month was considered.

Conclusion: We found a positive association between any NSAID use and higher disease activity, while high vs. low NSAID use was not associated. This may be due to AS patients with higher disease activity taking more NSAIDs.  Our findings also hint at a modest treatment effect of NSAIDs in AS as high vs. low use was not associated with change in disease activity.     Table 1. NSAID Use and Covariables Associated with Longitudinal Disease Activity in Multivariable Regression Modeling  

a.        modified Stoke Ankylosing Spondylitis Spinal Scores (mSASSS).

b.        Center for Epidemiologic Studies Depression Scale (CESD)