2019 LUPUS CLASSIFICATION CRITERIA SCORE PREDICTS FUTURE LUPUS HOSPITAL ADMISSION

Background: There are several validated tools to quantitate lupus disease activity, end-organ damage and overall fragility. An algorithm to predict the hospitalization risk in lupus patients was proposed by Li et al . That algorithm was able to effectively screen patients at increased risk of hospitalization using EHR information only. Recently, the new 2019 Lupus classification criteria score has been noted to accurately predict 10 year mortality . Objectives: To test the above 2 algorithms with potential to predict lupus related hospital admissions. First, we attempted to validate the existing algorithm from the index study of Li et al to predict lupus hospitalization. Second, we tested the 2019 lupus clinical classification score for its ability to predict hospitalizations. Methods: A retrospective chart review was performed using EHR data collected from 2013 to 2018 at University of Kentucky (UK) Medical Center. Inclusion criteria were 18 years or older at first outpatient rheumatology appointment at UK, at least 3 outpatient rheumatology visits at UK, and ICD 9/10 code for Lupus. A total of 217 patients met inclusion criteria. Variables similar to the index study were extracted from patients’ first outpatient rheumatology visit at UK. Additionally, 2019 Lupus Classification Criteria score was calculated. Patients who were subsequently hospitalized, manual chart review was done to determine if the hospitalization was attributable to lupus or not. Results: Table 1 shows differences between the variables predicting hospitalization in patients in this study (UK) and the Ohio State University (OSU) cohort from whom the admission predicting algorithm was derived . All the risk factors that were found to predict lupus hospitalization in the index study, failed to achieve a statistical significance in our validation study. Table 1. Differences in the variables predicting hospitalization between Index and Validation Cohort Variables predicting Lupus Hospitalization Index Study (Ohio State), % of patients (n=226 ) Validation Study (University of Kentucky ) % of patients; (n=217 ) African American 33% 18% Creatinine > 1.2 17% 7% Hemoglobin < 11g/dl 79% 18% Platelets < 180 / uL 75% 22% High Risk immunosuppression 35% 9. 2% Missed appointment 27% 25% There was more success predicting lupus hospitalization using the 2019 lupus classification criteria score (CCS) ( Figure 1 ). A CCS >=19 predicted higher risk of lupus related hospitalization vs CCS < 19 over the ensuing 2 years (p-0.05). Figure 1. Kaplan- Meier Survival Analysis comparing the risk of hospitalization between the groups with 2019 Lupus classification criteria score (CCS) of less than 19 (red) and more/equal to 19 (blue). A time-dependent effect, with the admission free survival curves crossing at two years (Figure 1), indicated a 1 out of 3 chance of lupus related admission during the first 6 months for a high CCS score > =19. Conclusion: We failed to validate the EHR algorithm identifying patients at high risk for lupus hospitalization in our less severely affected cohort with fewer admission events to analyze. Nonetheless, “criteria counting” using the weightings of the 2019 lupus classification criteria was granular enough to make these case finding criteria themselves prognostic for future hospitalization risk. It is likely that existing EHRs, using protocols based upon classification criteria scores, are now capable of predicting survival, costs, and admissions automatically. REFERENCES: [1]D Li et al. Determining risk factors that increase hospitalizations in patients with systemic lupus erythematosus, Lupus (2018) 27, 1321–1328 [2]Carneiro et al. A comparison of three classification criteria sets for Systemic Lupus Erythematosus – a study looking at links to outcome and mortality; Arthritis Care Res (Hoboken). 2019 Sep 10. doi: 10.1002/acr.24061 Disclosure of Interests: Saurav Suman: None declared, Mervat Eissa: None declared, Heidi Rogers: None declared, Aleksander Lenert: None declared, Arnold Stromberg: None declared, william roberts Shareholder of: Own Stocks of Pfizer and Novartis Citation: Ann Rheum Dis, volume 79, supplement 1, year 2020, page 1902Session: Epidemiology, risk factors for disease or disease progression (Abstracts Accepted for Publication)