Background/Purpose:   Spondyloarthropathies (SpA) have been associated with increased risk of myocardial infarction (MI), however it remains unclear if the risk is related solely to the underlying inflammatory disease, or also due to use of medications that increase MI risk, such as certain non-steroidal anti-inflammatory drugs (NSAIDs).   Alternatively, NSAIDs may reduce MI risk through reduction in systemic inflammation.  Although diclofenac is commonly prescribed for treatment of SpA symptoms, its effect on MI in the SpA population has not been well quantified.    By comparing risk of MI with diclofenac use in the SpA population with that in non-inflammatory back pain, we aimed to determine if the increased MI risk is solely due to the inflammatory disease or also the result of NSAID use.

Methods:   We performed a cohort study assessing the effect of diclofenac on MI risk using 1995-2013 data from The Health Improvement Network, a medical record database from the United Kingdom, comprising records of over 11 million patients from over 600 general practitioners.  We included adults with at least 1 year enrollment in the database, followed by a diagnosis of ankylosing spondylitis or psoriatic arthritis (combined SpA population) or who had a low back pain (LBP) diagnosis without any prior SpA history.  From the SpA and LBP populations, we identified subjects who used either diclofenac or naproxen after SpA or LBP diagnosis and followed them until diagnosis of MI, death or end of the study period.  We divided subjects into four categories:  SpA/diclofenac, SpA/naproxen, LBP/diclofenac, and LBP/naproxen.  We used Cox proportional hazards models with adjustment for potential confounders.  The risk of MI in each cohort was compared to that in the LBP/naproxen cohort to estimate the effects of diclofenac and SpA on MI independently.

Results:   We identified 2590 naproxen users and 3573 diclofenac users with SpA, as well as 147510 naproxen users and 280442 diclofenac users with LBP.   SpA subjects tended to be younger, and less commonly female than LBP subjects.   In all cohorts, baseline smoking, obesity, hypertension and use of cardioprotective medications was common.  After adjustment for potential confounders, relative to the LBP/naproxen cohort, hazard ratios were 1.02 (985% CI 0.91-1.14) for LBP/diclofenac, 1.26 (0.71-2.23) for SpA/naproxen, and 1.50 (1.02-2.22) for SpA/diclofenac (Table).

Conclusion:   Diclofenac use among SpA patients was associated with 50% increased risk of MI, while naproxen use did not significantly increase risk relative to its use in LBP patients.  Notably in LBP patients, diclofenac use did not increase MI risk, potentially reflecting different patterns of use among LBP and SpA populations (eg- dose, frequency).  These findings suggest that diclofenac use in SpA increases MI risk beyond any potential risk conferred by the underlying inflammatory disease.