Use of risk chart algorithms for the identification of psoriatic arthritis patients at high risk for cardiovascular disease: findings derived from the project CARMA cohort after a 7.5-year follow-up period
Polo y La Borda,J.,Castaneda,S.,Heras-Recuero,E.,Sanchez-Alonso,F.,Plaza,Z.,Garcia Gomez,C.,Ferraz-Amaro,I.,Sanchez-Costa,J. T.,Sanchez-Gonzalez,O. C.,Turrion-Nieves,A. I.,Perez-Alcala,A.,Perez-Garcia,C.,Gonzalez-Juanatey,C.,Llorca,J.,Gonzalez-Gay,M. A.,On behalf of the CARMA Project Collaborative Group
DOI: https://doi.org/10.1136/rmdopen-2024-004207
2024-04-18
RMD Open
Abstract:Objective To assess the predictive value of four cardiovascular (CV) risk algorithms for identifying high-risk psoriatic arthritis (PsA) patients. Methods Evaluation of patients with PsA enrolled in the Spanish prospective project CARdiovascular in RheuMAtology. Baseline data of 669 PsA patients with no history of CV events at the baseline visit, who were followed in rheumatology outpatient clinics at tertiary centres for 7.5 years, were retrospectively analysed to test the performance of the Systematic Coronary Risk Assessment (SCORE), the modified version (mSCORE) European Alliance of Rheumatology Associations (EULAR) 2015/2016, the SCORE2 algorithm (the updated and improved version of SCORE) and the QRESEARCH risk estimator version 3 (QRISK3). Results Over 4790 years of follow-up, there were 34 CV events, resulting in a linearised rate of 7.10 per 1000 person-years (95% CI 4.92 to 9.92). The four CV risk scales showed strong correlations and all showed significant associations with CV events (p<0.001). SCORE, mSCORE EULAR 2015/2016 and QRISK3 effectively differentiated between low and high CV risk patients, although the cumulative rate of CV events observed over 7.5 years was lower than expected based on the frequency predicted by these risk scales. Additionally, model improvement was observed when combining QRISK3 with any other scale, particularly the combination of QRISK3 and SCORE2, which yielded the lowest Akaike information criterion (411.15) and Bayesian information criterion (420.10), making it the best predictive model. Conclusions Risk chart algorithms are very useful for discriminating PsA at low and high CV risk. An integrated model featuring QRISK3 and SCORE2 yielded the optimal synergy of QRISK3's discrimination ability and SCORE2's calibration accuracy.
rheumatology