Short-term prognostic models for functional improvement in rheumatic diseases: an exploratory prospective study

To identify baseline factors associated with clinically meaningful short-term improvement in patient-reported physical function and to develop exploratory prognostic models for LEFS and DASH response in patients with rheumatic diseases.

This prospective observational study included 64 adults with heterogeneous rheumatic and musculoskeletal diagnoses, assessed at baseline and after 4 weeks of individualized routine pharmacological treatment. Clinically meaningful improvement was defined as a ≥ 9-point increase in the Lower Extremity Functional Scale (LEFS) and a ≥ 12-point decrease in the Disabilities of the Arm, Shoulder and Hand (DASH) score. Separate univariable and multivariable logistic regression analyses were performed for each outcome. Apparent model performance was assessed using AUROC and Brier scores, while regression-coefficient uncertainty was examined using 1000 bootstrap resamples.

LEFS response occurred in 43/64 patients (67.2%) and DASH response in 35/64 (54.7%). Shorter disease duration, lower GAD-7 score, and lower BMI were independently associated with response in both models. Higher total protein and lower baseline LEFS were additionally associated with LEFS response, while higher hemoglobin was associated with DASH response. The models demonstrated high apparent discrimination in the development cohort, without optimism correction, with AUROCs of 0.928 (95% CI 0.866-0.990) for LEFS and 0.910 (95% CI 0.837-0.984) for DASH. Corresponding apparent Brier scores were 0.100 and 0.111.

Simple routinely available baseline characteristics showed promising value for short-term prognostic stratification of functional improvement. The models remain exploratory and their predictive performance requires internal and external validation before clinical application.