Although previous studies have demonstrated that high-dose glucocorticoids (GCs) significantly increase the risk of infection in patients with autoimmune inflammatory rheumatic diseases (AIRDs), the actual risk for individual patients varies depending on other clinical factors. This study aimed to develop and validate a scoring system for identifying patients at high risk of serious infection following high-dose GC treatment for AIRDs. Patients with AIRDs treated with prolonged (≥ 4 weeks) high-dose GCs (≥ 30 mg/day prednisone) were included from two referral hospitals. Primary outcome was 1-year incidence of serious infection. Cox regression with LASSO regularization was applied to select covariates for the final model from 19 prespecified factors. A scoring system was developed based on β-coefficients of covariates in the final model. In the derivation cohort (n = 1635), serious infection occurred in 153 patients, with an incidence rate (per 100 person-years) of 10.5 (95% CI 9.0–12.3). The LASSO regularization identified older age, interstitial lung disease, decreased renal function, anemia, low serum albumin, ANCA-associated vasculitis, and concomitant rituximab and cyclophosphamide as key risk factors. The resulting scoring system, termed the CORAL score (0–20), demonstrated good predictive performance (AUROC 0.726 95% CI 0.681–0.771) for serious infections. Based on the score, patients were stratified into low- (score ≤ 6) or high-risk group (score > 6). Patients in the high-risk group (n = 343) had significantly higher rates of serious infection (HR 4.85 3.53–6.67) and all-cause mortality (HR 4.55 2.91–7.11) than those in the low-risk group. The predictive accuracy of the CORAL score was consistent in the external validation cohort (n = 672). In patients with AIRDs receiving prolonged high-dose glucocorticoids, the CORAL score successfully identified those at high risk of serious infection.
Choi et al. (Thu,) studied this question.