Cardiovascular risk with NSAIDs in rheumatoid arthritis: an analysis using routinely collected data

Rheumatology key message Contemporary data from routine care show no association between cardiovascular risk and NSAIDs in RA. Sir, It is well known that NSAIDs are associated with increased cardiovascular (CV) risk in the general population [1]. Rheumatoid arthritis (RA) is also associated with increased CV morbidity and mortality [2]. Numerous guidelines recommend that RA patients should be assessed annually to assess CV risk factors using tools such as QRISK2 [3, 4]. Therefore, it would be logical to assume that NSAID use might be particularly hazardous in the setting of RA, where background CV risk is greater. However, data from a large RA cohort study found no association between NSAID use and CV risk [5]. Two possible explanations exist for the lack of any association: NSAIDs truly have a different safety profile in the setting of systemic inflammation, with the deleterious vascular effects of cyclooxygenase-2 (COX-2) inhibition offset by the beneficial impact of lower systemic inflammatory burden [6]; or selection bias exists, whereby clinicians preferentially avoid prescribing NSAIDs in high-risk subjects. Despite the conflicting evidence base, NSAIDs remain a powerful therapeutic adjunct to alleviate symptoms in RA patients and understanding their safety profile in contemporary practice is important. We therefore aimed first to investigate the prescribing habit of NSAIDs in our cohort in RA in a single tertiary referral unit in London and also to evaluate the relationship between NSAIDs use and hospitalization due to CV events. Data were captured using the electronic medical record (EMR) from King’s College Hospital (KCH) in London. KCH provides care using a paperless record with electronic coding of all hospital admissions. In addition, a rheumatology-specific database (RSD) contains detailed information on disease as well as drug exposure information. Patients with RA were classified into regular NSAIDs users or non-users according to whether they had NSAIDs recorded on their regular medication at their earliest documented visit in the EMR. Cox-2 inhibitors accounted for a small proportion of prescriptions and were combined with all NSAIDs. Low-dose aspirin was not considered an NSAID in this analysis. Incident CV events were identified from the International Classification of Diseases, Tenth Revision coded hospital episode statistics submissions and then verified manually. Data on potential confounders were collected from the RSD. In order to adjust for disease burden, a cumulative estimate of disease activity during follow-up was calculated using an area under the curve approach. The follow-up window included time from either March 2008 (when the current EMR coding system was established) or date of diagnosis, whichever came later. Follow-up was censored at the first CV event (including stroke) or October 2013, whichever came first. Cox proportional hazards modelling was used to compare the risk of cardiac events between NSAID users and non-users. In total, 1089 RA patients were under follow-up during the time period available for analysis. The characteristics are typical of a UK cohort with RA, with a mean age 55 years, a female predominance (78%), mean baseline 28-joint DAS score of 4 and a moderate level of disability as measured by the HAQ (median 1.625). Twenty-eight percent of the cohort were treated for hypertension. The median follow-up time was 3.7 years. The annual incidence of CV events was 1.2%. The absolute event rate was lower in the NSAID users, with a hazard ratio (HR) of 0.77 (95% CI 0.39, 1.51). Adjusting for confounders had minimal impact on the HR, which remained non-significantly lower in the NSAID user cohort. Sensitivity analyses including as-required NSAID users did not alter estimates, however, excluding naproxen users (who accounted for 24% of NSAID users) increased the HR estimate to 1.2 (95% CI 0.54, 2.45). Only 10% of NSAID users were receiving COX-2 inhibitors, making the sample size too small for meaningful analyses. Fig. 1 shows the Nelson–Aalen plot comparing the unadjusted risk of CV events in NSAID users and non-users (this can be considered similar to an upside Kaplan–Meier method, plotting cumulative hazard on the y-axis against time on the x-axis). Comparison of unadjusted CV events in NSAID users and non-users These data are reassuring about the use of NSAIDs in RA in UK practice. In our unit, the use of NSAIDs overall does not appear to be associated with an increased CV risk. The findings should not be interpreted as demonstrating no CV risk with NSAID use. Rather, these data support the premise that the manner in which we use NSAIDs appears not to be associated with an increased CV risk. There are significant limitations to acknowledge, namely the small sample size, confounding effect of co-morbidities, the potential for misclassification bias as well as left and right censorship. Regarding right censorship, as we rely upon capturing hospital events from our institution alone, we are likely to have underestimated the true CV event rate (although there is no reason to suspect that this latter bias would impact either group differentially). We were also unable to specifically explore the impact of COX-2 selective agents, as these were prescribed infrequently. This approach to analysis using the EMR opens doors to future research, highlighting the power of routinely captured data. As routinely captured data are more frequently utilized to address clinical questions in everyday practice, the strengths and limitations of such datasets will become more apparent. Funding: No specific funding was received from any funding bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article. Disclosure statement: The authors have declared no conflicts of interest.