To investigate the effects of glucocorticoids (GC) on osteoporosis (OP) risk and changes in bone mineral density (BMD) in female patients with rheumatoid arthritis (RA) across different levels of disease activity, and to develop individualized risk prediction models.

A retrospective analysis was conducted in 186 female RA patients, who were divided into remission/low disease activity group (DAS28 ≤ 3.2, n = 59) and moderate/high disease activity group (DAS28 > 3.2, n = 127) based on baseline DAS28 score. Univariate analysis and LASSO regression were used for variable selection. Predictive models for 1-year OP risk and 1-year change in BMD were constructed in the overall population and in the two subgroups, with internal validation performed using bootstrap resampling.

Multivariate logistic regression showed that in the overall population, age (OR = 1.09), baseline DAS28-ESR (OR = 1.56), and cumulative days of GC usage (OR = 1.01) were independent risk factors (all P < 0.01), while BMI (OR = 0.79), TNF-α inhibitor use ≥ 3 months (OR = 0.36), cumulative calcium dose (OR = 0.99), and use of denosumab or zoledronic acid (OR = 0.41) were independent protective factors (all P < 0.05). Subgroup analysis revealed that in the remission/low disease activity group, baseline DAS28-ESR (OR = 4.87) and cumulative days of GC usage (OR = 1.04) were risk factors, while TNF-α inhibitor use ≥ 3 months (OR = 0.03) was a protective factor. In the moderate/high disease activity group, age (OR = 1.15) and baseline DAS28-ESR (OR = 11.64) were risk factors, while cumulative days of calcium usage (OR = 1.00) and use of denosumab or zoledronic acid (OR = 0.17) were protective factors. The three models demonstrated good discrimination, with area under the curve (AUC) values ranging from 0.885 to 0.942. Multiple linear regression showed that in the overall population, age was negatively correlated with change in BMD (β = –0.02), while BMI was positively correlated (β = 0.08) (both P < 0.001). In subgroup analysis, age, baseline DAS28-ESR, and cumulative days of GC usage were negatively correlated with change in BMD in the remission/low disease activity group (all P < 0.01). In the moderate/high disease activity group, age and baseline DAS28-ESR were negatively correlated, while BMI, cumulative days of GC usage, and use of denosumab or zoledronic acid were positively correlated (all P < 0.05). The three models explained 33.8%, 75.5%, and 80.1% of the variance in BMD change, respectively.

In female RA patients, cumulative GC use is associated with an increased risk of OP, and this association is modulated by disease activity. GC may attenuate BMD loss in patients with moderate/high disease activity by suppressing inflammatory status. This study was supported by the National Natural Science Foundation of China (No. 82574760).