ObjectiveTo establish the predictive value of the QRESEARCH risk estimator version 3 (QRISK3) algorithm in identifying Spanish patients with ankylosing spondylitis (AS) at high risk of cardiovascular (CV) events and CV mortality. We also sought to determine whether to combine QRISK3 with another CV risk algorithm: the traditional SCORE, the modified SCORE (mSCORE) EULAR 2015/2016 or the SCORE2 may increase the identification of AS patients with high-risk CV disease. MethodsInformation of 684 patients with AS from the Spanish prospective CARdiovascular in ReuMAtology (CARMA) project who at the time of the initial visit had no history of CV events and were followed in rheumatology outpatient clinics of tertiary centers for 7.5 years was reviewed. The risk chart algorithms were retrospectively tested using baseline data. ResultsAfter 4,907 years of follow-up, 33 AS patients had experienced CV events. Linearized rate=6.73 per 1000 person-years (95 % CI: 4.63, 9.44). The four CV risk scales were strongly correlated. QRISK3 correctly discriminated between people with lower and higher CV risk, although the percentage of accumulated events over 7.5 years was clearly lower than expected according to the risk established by QRISK3. Also, mSCORE EULAR 2015/2016 showed the same discrimination ability as SCORE, although the percentage of predicted events was clearly higher than the percentage of actual events. SCORE2 also had a strong discrimination capacity according to CV risk. Combining QRISK3 with any other scale improved the model. This was especially true for the combination of QRISK3 and SCORE2 which achieved the lowest AIC (406.70) and BIC (415.66), so this combination would be the best predictive model. ConclusionsIn patients from the Spanish CARMA project, the four algorithms tested accurately discriminated those AS patients with higher CV risk and those with lower CV risk. Moreover, a model that includes QRISK3 and SCORE2 combined the best discrimination ability of QRISK3 with the best calibration of SCORE2.
Read full abstract