Using patient-reported measures to predict hospitalisation in a population-based lupus cohort

Area of Science:

• Rheumatology
• Autoimmune Diseases
• Predictive Modeling

Background:

• Systemic Lupus Erythematosus (SLE) is a multisystem autoimmune disease characterized by unpredictable disease flares.
• These flares can lead to irreversible organ damage and increased healthcare utilization, including hospitalizations.

Purpose of the Study:

• To develop a patient-centric predictive model for identifying SLE patients at higher risk of hospitalization.
• Utilizing real-world data to enhance risk stratification for SLE management.

Main Methods:

• An observational, retrospective analysis of the Georgians Organized Against Lupus (GOAL) cohort (2011-2013).
• Data linkage between GOAL surveys (sociodemographic, clinical, Systemic Lupus Activity Questionnaire [SLAQ] symptoms) and the Georgia Hospital Discharge Database.
• A two-step approach using logistic regression and Classification and Regression Tree (CART) models to predict all-cause hospitalizations.

Main Results:

• 846 participants completed 1486 surveys; hospitalized patients were younger, poorer, and reported more SLE symptoms.
• CART modeling identified unintentional weight loss, severe fatigue, and Raynaud's symptoms as key predictors of hospitalization.
• Patients in the high-risk subset (34%) had a 2.6-fold higher hospitalization rate than the overall cohort (13%).

Conclusions:

• Patient-reported SLE symptoms and disease activity, particularly components of the SLAQ, are valuable for risk management.
• These patient-reported outcomes can inform strategies aimed at reducing SLE hospitalizations.
• The study highlights the utility of real-world data and patient-reported outcomes in personalized SLE care.