Introduction:

Systemic lupus erythematosus is a chronic autoimmune disease with variable severity, neuropsychiatric manifestations comprise a wide spectrum of syndromes defined according to its central and peripheral involvement. Many pathophysiological mechanisms have been involved in the development of these manifestations, such as the disrup tion of the blood-brain barrier and the cellular toxicity related to autoantibodies such as anti-aquaporin, anti-ribosomal P and anti-phospholipids. Predictive elements to identify neuropsychiatric compromise will be explored by predictive models in this article.

Methodology and methods:

Retrospective cross-sectional study, predictive models such as decision trees, naive Bayesian, K-nearest neighbours and generalized linear models were used for the analysis, the best model was selected.

Results:

122 patients were included, of whom 100 (81.9%) were women and 22 (18.1%) men, the prevalence of neuropsychiatric compromise was 59%, anxiety disorders, headache and mood disorders being the most frequent. In the bivariate analysis, oral ulcers (OR 2.52, 95% CI 1.09-6.08 p = 0.02), non-scarring alopecia (OR 4.06, 95% CI 1.76-9.68 p = <0.01), arthritis (OR 2.5, 95% CI 0.99-6.94 p = 0.04), leukopenia (OR 2.81, 95% CI 1.18-7.09 p = 0.04) and patients without accumulated damage measured by the SLICC/ACR-DI damage score (2.1, 95% CI 0.98-4.93 p = 0.01), were associated with greater neuropsychiatric compromise. The best prediction model was the decision tree model, in which non-scarring alopecia, oral ulcers and anti-Ro/SSA antibodies behaved as independent variables related to neuropsychiatric compromise with a precision of 75.6%.

Conclusions:

The prevalence of neuropsychiatric involvement in lupus was 59%, the inde pendent variables that best predict these manifestations were anti-Ro/SSA seropositivity, non-scarring alopecia, and oral ulcers through decision trees.

        · resumen en Español     · texto en Español | Inglés     · Español ( pdf ) | Inglés ( pdf )