Prediction of short term mortality in Systemic Lupus erythematosus with time dependent measures of disease activity

Objective, To identify predictors of short term mortality in systemic lupus erythematosus (SLE) in terms of time dependent clinical indicators of dise ase activity From the Systemic Lupus Erythematosus Disease Activity Index ( SLEDAI). Methods. We studied data collected on patients followed at the University o f Toronto Lupus Clinic. Clinical and laboratory indicators of disease activ ity are recorded at each clinic visit and a SLEDAI summary score is calcula ted. Survival analyses were conducted in which the prognostic value of the time dependent indicators of disease activity was examined on 6-month morta lity through a multivariate Cox regression model. Relative risks, confidenc e intervals, and significance levels were obtained for each indicator to re flect their clinical importance and statistical significance. Results. The sample consisted of 806 patients followed for a median of 6.6 years; 702 (87%) were female, 671 (83%) were Caucasian, and the mean age at first clinic visit was 36 years. Seventy-two patients died within 6 months of their last clinic visit. In a univariate regression model, a categorica l variable reflecting total SLEDAI score was highly prognostic for mortalit y (p < 0.001) and yielded increasing relative risks of 1.28 for SLEDAI 1-5 vs 0, 2.34 for SLEDAI 6-10 vs 0, 4.74 for SLEDAI 11-19 vs 0, and 14.11 for SLEDAI greater than or equal to 20 vs 0. In a separate multivariate Cox mod el examining the individual components of SLEDAI. presence of organic brain syndrome, retinal changes, cranial nerve involvement, proteinuria, pyuria, pleurisy, fever, thrombocytopenia, and leukopenia each significantly incre ased the risk of death, while new rash and anti-DNA antibodies conferred pr otective effects. Conclusion, This time dependent Cox regression analysis identified the exte nt to which SLE disease activity, revealed by SLEDAI, is prognostic for sho rt term mortality. Further, important individual components were identified and their prognostic value for death was estimated.