Fall risk in rheumatoid arthritis revisited: A follow-up study focusing on sarcopenia and glucocorticoid use Authors: Naoki Nakagawa, Keita Nagira, Masako Hayashibara, Takashi Wada, Yuki Kitsuda, Yuta Takasu, Chikako Takeda, Mari Osaki, Hiroshi Hagino, Hideki Nagashima Published: 2026 ABSTRACT AbstractObjectivesTo re-evaluate fall incidence and its determinants in a contemporary cohort of ambulatory women with rheumatoid arthritis (RA) in a follow-up study based on our previously published cohort data.MethodsWe prospectively evaluated 124 ambulatory female patients aged ≥50 years with RA at a single institution in 2022–2023, monitoring falls for 1 year. Outcomes included patient characteristics, sarcopenia, disease activity, and physical function. Multivariable logistic regression and Firth-penalized logistic regression were used in sensitivity analysis to assess the interaction between sarcopenia and glucocorticoid (GC) use.ResultsThe annual fall rate in the 2022–2023 cohort was lower than in our previous cohort (27.4% vs. 50.0%). History of falls (odds ratio [OR] 2.76, 95% confidence interval [CI] 1.02–7.45, P = 0.045) and hypnotics/anxiolytics use (OR 3.92, 95% CI 1.01–15.20, P = 0.048) were independent risk factors for falls. A significant interaction between sarcopenia and GC use was observed in the initial analysis but was no longer statistically significant in sensitivity analyses.ConclusionsCompared with our previously reported cohort, the contemporary RA cohort showed a lower fall rate and a different pattern of associated risk factors. Further research is needed to explore the potential interaction between sarcopenia and glucocorticoid use. The observed interaction should be considered exploratory.Keywords: Rheumatoid arthritis, Falls, Sarcopenia, Frailty, Aging Introduction Rheumatoid arthritis (RA) is a chronic inflammatory disease that erodes joint cartilage and bone, leading to functional impairment [1]. Patients with RA are considered to be at increased risk of falls [2], although some contemporary cohorts have reported fall rates comparable to those of controls [3]. Recent studies have shown that sarcopenia is common in patients with RA, although its association with falls remains uncertain [[4], [5], [6]]. RA management has changed substantially following the introduction of the 2010 American College of Rheumatology/European Alliance of Associations for Rheumatology (ACR/EULAR) classification criteria, the treat-to-target strategy, and the development of biological or targeted synthetic disease-modifying antirheumatic drugs (b/tsDMARDs) [[7], [8], [9], [10]]. These advances have improved disease control, functional outcomes, and life expectancy in patients with RA [[9], [10], [11]]. Accordingly, the clinical characteristics of patients with RA have changed considerably. At our institution, knee arthroplasty among patients with RA <60 years has significantly decreased [12]. Recent studies have also reported increasing osteoarthritis-like features and a shift toward older age at RA onset [[13], [14], [15]], along with atypical clinical presentations, such as concomitant carpal tunnel syndrome and shoulder monoarthritis [16,17]. However, few studies have investigated fall frequency and risk factors in patients with RA. Reported 1-year fall rates among these individuals range from 20.2% to 50.0%, with risk factors including age, history of falls, swollen or painful joints, HAQ score, and antihypertensive medication [4,[18], [19], [20]]. Our previous prospective study (2006) was conducted before biologic agents became widely available [20]. Since then, treatment strategies and patient characteristics have changed significantly, warranting a re-evaluation of fall risk in a modern cohort. This study therefore serves as a second report building on our previous cohort study using the same methodology. It also includes an exploratory assessment of the interaction between sarcopenia and glucocorticoid (GC) use. Methods Patients From January 2022 to July 2022, 136 female patients aged ≥50 years with RA who were able to walk independently and visited the Orthopedic Surgery Outpatient Department at Tottori University Hospital, met the eligibility criteria for this study. Eligibility required meeting the 2010 ACR/EULAR classification criteria for RA [7] or having received a diagnosis of RA from a specialist ≥1 year before and having been continuously followed in our department for ≥1 year. Patients with connective tissue diseases other than RA, or those unable to walk independently due to conditions other than RA, were excluded. Ultimately, 126 patients met all selection criteria and provided written consent for participation in the study (Fig. 1). This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Tottori University Hospital (Approval Number: 21A050). Flow diagram of participant selection and follow-up in the 2022–2023 cohort. In 2022–2023, 136 patients were screened, of which 126 were included. Patients were followed prospectively for 1 year and classified as fallers or non-fallers according to history of falls during the observation period. RA, rheumatoid arthritis. In addition, data from our previously published 2006–2007 prospective cohort [20] were used as a historical reference to contextualize the findings of the present study. Sample size Sample size was estimated using Easy R (EZR), assuming an annual fall rate of ∼30% in the contemporary cohort based on our previous study [20]. We predicted a lower fall rate in the contemporary cohort than that previously reported and therefore adopted a one-sided test. Assuming 80% power and a 5% significance level (one-sided), the required sample size was estimated to be 104 participants. Patient characteristics At study initiation, detailed medical histories regarding RA, including age, disease duration, and history of falls within the past year, were collected through physical examination and interviews. Walking ability, including cane use, was assessed as well. RA disease stage and functional status were evaluated based on the Steinbrocker classification. The stage was determined based on the condition of the most severely affected joint. Current RA medications, including GCs, methotrexates, and b/tsDMARDs, as well as regular medications such as antihypertensives, diuretics, hypnotics, and anxiolytics, were recorded. Fall assessment A fall was defined as an unintentional event that resulted in a person coming to rest on the ground, floor, or another lower level [21]. On the first day of the study, the patients underwent motor function testing, physical examination, and balance function testing. All patients were provided with a 1-year ‘fall calendar’ and instructed to record each fall as it occurred. The calendars were collected during outpatient visits every 3 months. In cases of incomplete records or non-collection, the patients were contacted by phone. After the 1-year observation period, the participants were classified into fall (those with ≥1 fall) and non-fall groups (those without) based on calendar records. Symptoms, disease activity, and foot deformity RA disease activity was assessed based on the number of swollen and tender joints in 66 joints, C-reactive protein (mg/dL) levels, erythrocyte sedimentation rate (ESR, mm/h), anti-cyclic citrullinated peptide antibody (IU/mL) levels, rheumatoid factor (U/mL) levels, matrix metalloproteinase-3 (MMP-3, ng/mL) levels, disease activity score in 28 joints for RA using erythrocyte sedimentation rate (DAS28-ESR), and the Modified Health Assessment Questionnaire (mHAQ). For foot deformities, left foot radiographs were used to measure and evaluate the hallux valgus (HV) angle, first/second metatarsal (M1/2) angle, first/fifth metatarsal (M1/5) angle, and calcaneal pitch (CP) angle. Physical function On the first day of the study, a physical therapist from the Rehabilitation Department conducted the following physical function tests: •Grip strength: With the subject standing and arms hanging down, each hand was measured twice, and the higher value was recorded. A Smedley-type digital grip strength tester (Takai Kiki Co., Ltd.) was used for this purpose. If measurement was impossible due to joint deformity, it was recorded as 0 kg. Grip strength: With the subject standing and arms hanging down, each hand was measured twice, and the higher value was recorded. A Smedley-type digital grip strength tester (Takai Kiki Co., Ltd.) was used for this purpose. If measurement was impossible due to joint deformity, it was recorded as 0 kg. •Walking speed: A 9-m course, consisting of a 5-m measurement section flanked by 2-m approach sections on each side, was used. The time required to traverse the 5-m section was measured, and the fastest recorded time was used to calculate walking speed. This protocol was modified from Tiedemann's method [22]. Walking speed: A 9-m course, consisting of a 5-m measurement section flanked by 2-m approach sections on each side, was used. The time required to traverse the 5-m section was measured, and the fastest recorded time was used to calculate walking speed. This protocol was modified from Tiedemann's method [22]. •Single-leg standing time: The subject stood on 1 leg at a time in a natural posture with arms at their sides. The total time standing on each leg was then recorded [23]. Single-leg standing time: The subject stood on 1 leg at a time in a natural posture with arms at their sides. The total time standing on each leg was then recorded [23]. Body composition and balance measurements Weight, height, body mass index, limb muscle mass, and body fat percentage were measured using a multi-frequency body composition analyzer (MC-780A-N; Tanita Corp., Tokyo, Japan). Postural balance was assessed on a specialized testing platform (Gravicorder GW-5000; Anima Corp., Tokyo, Japan). Subjects maintained an open-eye standing position for several minutes while center of pressure (COP), total path length (cm), and sway area (cm2) were measured. Sarcopenia assessment Skeletal muscle mass index (SMI, kg/m2) was calculated as the limb muscle mass (kg) obtained via bioelectrical impedance analysis (BIA) divided by height squared (m2). Based on the 2019 Asian Working Group for Sarcopenia diagnostic criteria, “sarcopenia” was defined as low muscle mass combined with either low grip strength or low walking speed, and “severe sarcopenia” as low muscle mass combined with both low grip strength and low walking speed. The cutoff values were SMI <5.7 kg/m2, grip strength <18 kg, and walking speed <1.0 m/s [24]. Statistical analysis Statistical analysis was performed using EZR (version 1.68, Jichi Medical University Saitama Medical Center, Saitama, Japan), a statistical analysis software based on R (The R Foundation for Statistical Computing, Vienna, Austria) [25]. The significance level was P < 0.05 in all tests. Normality of continuous variables was assessed using the Shapiro–Wilk test. For variables showing normal distribution (age and SMI), a Student's t-test was used. For non-normally distributed continuous variables and ordinal variables, the Mann–Whitney U test was used. Fisher's exact test was used to compare nominal variables. Risk factors for falls were examined using multivariable logistic regression analysis. Covariates in the primary model were selected based on our previous study [20] and their clinical relevance to fall risk in contemporary patients with RA. Antihypertensive/diuretic use and COP sway area were retained from the previous model, and DAS28-ESR was used to represent current RA disease activity. Sarcopenia, GC use, history of falls, and hypnotic or anxiolytic use were included as clinically relevant contemporary factors. Additionally, a predefined subgroup analysis was conducted to examine fall risk by sarcopenia status and GC use. Multiple subgroup comparisons were performed using Fisher's exact test. After identification of significant associations, post hoc residual analyses were performed to pinpoint contributing groups using a previously described method [26]. Adjusted predicted probabilities for the 4 subgroups were estimated using Firth's penalized logistic regression. The sensitivity model used Firth's penalized logistic regression to account for sparse data and included clinically relevant covariates (age, DAS28-ESR, and mHAQ), variables related to the interaction analysis (sarcopenia, GC use, and the interaction between sarcopenia and GC use), and independent fall predictors identified in the primary analysis (history of falls and use of hypnotics or anxiolytics). Age was retained as a clinically important potential confounder regardless of its statistical significance. Given the limited number of fall events, these results were interpreted with potential overfitting in mind. Results A total of 136 patients were screened in 2022. Ten patients who did not provide consent were excluded, and the remaining 126 patients underwent 1 year of prospective observation. Two patients were excluded: 1 for sudden death unrelated to RA and another for transfer to another physician. Ultimately, 124 patients were included in the analysis. Participants were classified based on fall occurrence during the study period: 34 and 90 in the fall and non-fall groups, respectively (Fig. 1). The mean age of the participants was 68.8 ± 9.7 years, and the 1-year fall rate was 27.4%. The fall group exhibited significantly higher rate of falls in the past year, oral GC use, and hypnotics or anxiolytics use, as well as higher prednisolone (PSL)-equivalent dose, than the non-fall group (P < 0.05; Table 1). Regarding foot deformities, we evaluated HV, M1/2, M1/5, and CP angles. No significant differences were observed between the fall and non-fall groups within the 2022–2023 cohort (Table S1) [Supplementary Materials]. Similarly, no significant differences in sarcopenia rates or motor and balance functions were observed between groups (Table 1). Fall rates did not differ significantly according to sarcopenia status or the type of hypnotics or anxiolytics used (Table S2) [Supplementary Materials]. Table 1Baseline characteristics of patients with rheumatoid arthritis according to fall status.Table 1Total (N = 124)Fall group (n = 34)Non-fall group (n = 90)P-valueAge, years68.0 (63.0, 76.0)70.4 ± 9.168.2 ± 9.90.255BMI, kg/m221.8 (19.5, 24.3)22.0 (19.8, 24.5)21.7 (19.5, 24.3)0.743Duration of RA, years13.0 (6.0, 21.5)12.0 (4.0, 22.0)13.0 (7.0, 20.5)0.583Class (Ⅰ/Ⅱ/Ⅲ/Ⅳ), n53/67/4/015/18/1/038/49/3/01.000Stage (Ⅰ/Ⅱ/Ⅲ/Ⅳ), n20/40/20/443/12/4/1517/28/16/290.390History of falls, n (%)31 (25.0)14 (41.2)17 (18.9)0.019*With aids, n (%)11 (8.9)3 (8.8)8 (8.9)1.000MTX, n (%)91 (73.4)25 (73.5)66 (73.3)1.000MTX dose, mg/week6.0 (0, 8.0)6 (0.5, 8.0)6.0 (0, 8.0)0.751b/tsDMARD, n (%)48 (38.7)11 (32.4)37 (41.1)0.533GC, n (%)16 (12.9)8 (23.5)8 (8.9%)0.039*PSL-equivalent dose, mg/dayc0 (0, 0)0 (0, 0)0 (0, 0)0.032*Antihypertensives/diuretics, n (%)41 (33.1)16 (47.1)25 (27.8)0.054Hypnotics/anxiolytics, n (%)13 (10.5)7 (20.6)6 (6.7)0.043*Swollen joint count, n1.0 (0, 3.0)1.0 (0, 3.0)1.0 (0, 3.0)0.883Tender joint count, n0 (0, 1.0)0 (0, 1.0)0 (0, 1.0)0.663ACPA (negative/low/high), n (%)a30/11/69 (24.2/8.9/55.6)12/1/15 (35.3/2.9/44.1)18/10/54 (20.0/11.1/60.0)0.088RF (negative/low/high), n (%)b13/19/34 (10.5/15.3/27.4)3/7/7 (8.9/20.6/20.6)10/12/27 (11.1/13.3/30.0)0.396CRP, mg/dL0.1 (0, 0.2)0.1 (0, 0.2)0.1 (0, 0.2)0.633ESR, mm/h14.0 (9.0, 27.0)13.0 (8.0, 23.0)14.0 (9.0, 27.3)0.418MMP-3, ng/mL54.1 (42.1, 80.4)56.9 (39.4, 69.0)53.8 (43.2, 82.9)0.403DAS28-ESR2.6 (2.1, 3.2)2.6 (2.2, 3.3)2.6 (2.1, 3.2)0.828mHAQ0 (0, 0.5)0.3 (0, 0.7)0.1 (0, 0.4)0.117Sarcopenia, n (%)20 (16.1)5 (14.7)15 (16.7)1.000SMI, kg/m26.3 (5.9, 6.7)6.4 (6.0, 6.7)6.3 (5.9, 6.7)0.503Grip power, kg14.8 (13.4, 16.1)14.7 (13.3, 15.9)14.8 (13.4, 16.2)0.834Walking speed, m/s1.6 (1.4, 1.9)1.7 (1.4, 1.8)1.6 (1.4, 1.9)0.869One-leg standing time, s31.1 (7.0, 99.1)23.1 (5.7, 79.7)33.5 (7.9, 100.2)0.341COP path length, cm2.1 (1.2, 79.8)59.7 (1.2, 85.7)1.4 (1.1, 73.6)0.266COP sway area, cm21.5 (1.0, 2.4)1.7 (0.9, 3.3)1.4 (1.0, 2.1)0.345IQR, interquartile range; ANCOVA, analysis of covariance; BMI, body mass index; RA, rheumatoid arthritis; MTX, methotrexate; GC, glucocorticoid; PSL, prednisolone; b/tsDMARD, biologic or targeted synthetic disease-modifying antirheumatic drug; ACPA, anti-citrullinated protein antibody; RF, rheumatoid factor; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; MMP-3, matrix metalloproteinase-3; DAS28, Disease Activity Score in 28 joints; mHAQ, modified Health Assessment Questionnaire; SMI, skeletal muscle mass index; PMI, psoas muscle mass index; COP, center of pressure.Values are shown as median (IQR) or n (%), as appropriate. *P < 0.05. P-values calculated using t-test, Mann–Whitney U test, Fisher's exact test, or chi-squared test, as appropriate.aACPA was classified as negative (<5.3 IU/mL), low-positive (5.3-15.9 IU/mL), and high-positive (>15.9 IU/mL).bRF was classified as negative (<15 U/mL), low-positive (15-45 U/mL), and high-positive (>45 U/mL). Cases with missing ACPA or RF data were excluded from the analysis.cPSL-equivalent dose: 0–5 mg/day. Baseline characteristics of patients with rheumatoid arthritis according to fall status. IQR, interquartile range; ANCOVA, analysis of covariance; BMI, body mass index; RA, rheumatoid arthritis; MTX, methotrexate; GC, glucocorticoid; PSL, prednisolone; b/tsDMARD, biologic or targeted synthetic disease-modifying antirheumatic drug; ACPA, anti-citrullinated protein antibody; RF, rheumatoid factor; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; MMP-3, matrix metalloproteinase-3; DAS28, Disease Activity Score in 28 joints; mHAQ, modified Health Assessment Questionnaire; SMI, skeletal muscle mass index; PMI, psoas muscle mass index; COP, center of pressure. Values are shown as median (IQR) or n (%), as appropriate. *P < 0.05. P-values calculated using t-test, Mann–Whitney U test, Fisher's exact test, or chi-squared test, as appropriate. ACPA was classified as negative (<5.3 IU/mL), low-positive (5.3-15.9 IU/mL), and high-positive (>15.9 IU/mL). RF was classified as negative (<15 U/mL), low-positive (15-45 U/mL), and high-positive (>45 U/mL). Cases with missing ACPA or RF data were excluded from the analysis. PSL-equivalent dose: 0–5 mg/day. Multivariable logistic regression analysis revealed that a history of falls (odds ratio [OR] = 2.76, 95% confidence interval [CI] 1.02–7.45, P = 0.045) and hypnotics or anxiolytics use (OR = 3.92, 95% CI 1.01–15.20, P = 0.048) were significantly associated with falls (Table 2). Table 2Multivariable logistic regression analysis of risk factors for falls.Table 2OR95% CIP-valueAge, years1.040.99–1.100.153Glucocorticoid use2.880.80–10.400.107DAS28-ESR0.860.49–1.530.623Sarcopenia0.480.12–2.000.312History of falls2.761.02–7.450.045*Hypnotics/anxiolytics use3.921.01–15.200.048*Antihypertensives/diuretics use1.900.76–4.760.173COP sway area, cm20.950.74–1.230.719OR, odds ratio; CI, confidence interval; DAS28-ESR, Disease Activity Score in 28 joints based on erythrocyte sedimentation rate; COP, center of pressure. *P < 0.05. Reference categories: no glucocorticoid use, no sarcopenia, no hypnotic/anxiolytic use, no antihypertensive/diuretic use, and no history of falls.Multivariable logistic regression analysis was performed using fall occurrence within 1 year as dependent variable. Covariates included age, glucocorticoid use, DAS28-ESR, sarcopenia, history of falls, hypnotics/anxiolytics use, antihypertensives/diuretics use, and COP sway area (eyes open). Multivariable logistic regression analysis of risk factors for falls. OR, odds ratio; CI, confidence interval; DAS28-ESR, Disease Activity Score in 28 joints based on erythrocyte sedimentation rate; COP, center of pressure. *P < 0.05. Reference categories: no glucocorticoid use, no sarcopenia, no hypnotic/anxiolytic use, no antihypertensive/diuretic use, and no history of falls. Multivariable logistic regression analysis was performed using fall occurrence within 1 year as dependent variable. Covariates included age, glucocorticoid use, DAS28-ESR, sarcopenia, history of falls, hypnotics/anxiolytics use, antihypertensives/diuretics use, and COP sway area (eyes open). Table S2 [Supplementary Materials] shows the distribution of hypnotic or anxiolytic use and fall rates according to medication category. Users of benzodiazepines (BZ) and non-BZ experienced higher fall rates, but these differences were not statistically significant compared to non-users. After classifying the cohort into 4 groups based on sarcopenia status and GC use, a significant difference in fall rates was observed (P = 0.035). Post hoc residual analysis showed that the group with both sarcopenia and GC use contributed significantly to the overall difference in fall rate (P = 0.035). Adjusted predicted probabilities for the 4 groups are shown in Table 3. Table 3Observed and adjusted fall risk according to sarcopenia and glucocorticoid use.Table 3Without GC and sarcopenia (n = 94)With GC and without sarcopenia (n = 10)Without GC and with sarcopenia (n = 14)With GC and sarcopenia (n = 6)P-valueNo falls in 1 year, n6961320.035*Falls in 1 year, n25414Residual analysis, P-value0.3730.2590.0780.035*Adjusted predicted probability, %21.524.24.442.5GC, glucocorticoid; DAS28-ESR, Disease Activity Score in 28 joints based on erythrocyte sedimentation rate; mHAQ, modified Health Assessment Questionnaire. *P < 0.05.Participants were categorized into 4 groups based on sarcopenia and GC use and fall rates compared using Fisher's exact test. When the overall comparison was significant, post hoc residual analysis was performed to identify subgroups contributing to the overall association. Adjusted predicted probabilities were estimated using the Firth penalized logistic regression model as described in Statistical Analysis section. Observed and adjusted fall risk according to sarcopenia and glucocorticoid use. GC, glucocorticoid; DAS28-ESR, Disease Activity Score in 28 joints based on erythrocyte sedimentation rate; mHAQ, modified Health Assessment Questionnaire. *P < 0.05. Participants were categorized into 4 groups based on sarcopenia and GC use and fall rates compared using Fisher's exact test. When the overall comparison was significant, post hoc residual analysis was performed to identify subgroups contributing to the overall association. Adjusted predicted probabilities were estimated using the Firth penalized logistic regression model as described in Statistical Analysis section. In the sensitivity analysis using Firth's penalized logistic regression adjusted for age, DAS28-ESR, mHAQ, history of falls, and hypnotic or anxiolytic use, the interaction between sarcopenia and GC use was no longer significant (OR = 13.73, 95% CI 0.80–412.62, P = 0.071; Table 4). Table 4Sensitivity analysis of fall risk using Firth penalized logistic regression.Table 4Adjusted OR95% CIP-valueAge, years1.030.98–1.090.234DAS28-ESR0.880.51–1.490.640mHAQ1.730.75–4.070.196Sarcopenia0.170.01–0.970.046GC use1.170.26–4.770.831Sarcopenia with GC use13.730.80–412.620.071History of falls2.450.93–6.440.068Hypnotic/anxiolytic use3.390.95–12.790.060OR, odds ratio; CI, confidence interval; GC, glucocorticoid; DAS28-ESR, Disease Activity Score in 28 joints based on erythrocyte sedimentation rate; mHAQ, modified Health Assessment Questionnaire.Sensitivity analysis performed using Firth's penalized logistic regression to account for the sparse data. Fall occurrence within 1 year was used as dependent variable. Covariates included age, DAS28-ESR, mHAQ, sarcopenia, GC use, interaction between sarcopenia and GC use, history of falls, and hypnotic/anxiolytic use. Sensitivity analysis of fall risk using Firth penalized logistic regression. OR, odds ratio; CI, confidence interval; GC, glucocorticoid; DAS28-ESR, Disease Activity Score in 28 joints based on erythrocyte sedimentation rate; mHAQ, modified Health Assessment Questionnaire. Sensitivity analysis performed using Firth's penalized logistic regression to account for the sparse data. Fall occurrence within 1 year was used as dependent variable. Covariates included age, DAS28-ESR, mHAQ, sarcopenia, GC use, interaction between sarcopenia and GC use, history of falls, and hypnotic/anxiolytic use. Discussion This study is a follow-up to our previously published cohort, using the same methodology at the same institution. In our 2006–2007 cohort, 50.0% of patients experienced ≥1 fall over a 1-year period, whereas the fall rate in the contemporary cohort, despite being older, was 27.4% [20]. The contemporary cohort also exhibited lower GC use and disease activity, as well as a lower prevalence of sarcopenia. These findings match improvements in RA management and clinical status. However, causality cannot be inferred from this observational study. The factors associated with falls also differed from those in our previous study. In the 2006–2007 cohort, swollen joint count, antihypertensive or diuretic use, shorter one-leg standing time, and greater postural sway were independently associated with falls [20]. Conversely, in the 2022–2023 cohort, history of falls and hypnotic or anxiolytic use were independently associated with falls, whereas disease activity and postural sway were not. The current fall rate was comparable to that reported in the general older adult population [21]. Sarcopenia is common in patients with RA and associated with increased risk of falls and fractures [5]. Further, it has been recently linked to moderate-to-severe HV in patients with RA [27]. Although aging contributes to the development of sarcopenia [28,29], chronic inflammation may also contribute in patients with RA [5,30]. The influence of RA treatment remains controversial [5,31]. Compared to our previous findings, GC use was significantly lower in the current cohort. This coincides with major changes in RA treatment strategies over the intervening years. Nevertheless, GCs are still used in a substantial proportion of patients with RA, although they can promote muscle protein catabolism, muscle wasting, and metabolic abnormalities [32]. Previous studies have associated GC exposure with the development of sarcopenia in patients with RA [33]. In the present study, both the prevalence of GC use and the PSL-equivalent dose among GC users were higher in the fall group, and combined sarcopenia and GC use were showed the highest observed fall rate among the 4 subgroups. However, after adjustment for disease activity, functional disability, history of falls, and hypnotic or anxiolytic use, sensitivity analyses using Firth's penalized logistic regression showed a weaker interaction. Given the small number of patients with both sarcopenia and GC use and the lack of robustness in sensitivity analyses, this interaction is an exploratory finding requiring confirmation in larger, adequately powered prospective studies. Collectively, these findings support a comprehensive approach to fall prevention in patients with RA extending beyond disease activity control alone. This approach should involve assessing previous falls, reviewing hypnotic or anxiolytic use, and preserving muscle function. This study has several limitations. First, we only included women with RA aged ≥50 years. Although this population is at high risk of falls, caution is warranted when generalizing the findings to the broader RA population, including men and younger individuals. Second, this was a single-institution study, and regional characteristics and clinical practice may have influenced the results. Despite using the same methodology as our previous prospective study, which allowed comparisons across 2 treatment eras, the observational design does not allow us to determine whether changes in RA treatment or disease control caused the reduction in falls. Third, despite assessing a range of clinical and functional variables, potential confounders such as visual impairment, physical activity, living environment, and psychological factors were not fully captured. These factors may have influenced fall risk and should be considered in future studies. In addition, detailed information regarding GC treatment duration, cumulative dose, and dose changes during follow-up was unavailable. Therefore, dose–response relationships could not be evaluated, and residual confounding by indication cannot be excluded, as GC use might reflect more severe or difficult-to-control RA rather than being a direct causal determinant of falls. The small number of patients with both sarcopenia and GC limited the precision and robustness of the interaction analysis. Consequently, this finding should be considered exploratory. Conclusions This study demonstrated a lower fall rate among women with RA than reported in our previous cohort. These findings suggest that the factors associated with falls in contemporary patients with RA may differ from those previously identified. While the coexistence of sarcopenia and GC use appeared to identify a subgroup with a high observed risk of falls, this finding should be considered exploratory. CRediT author statement Naoki Nakagawa: Conceptualization, Data curation, Formal analysis, Writing – original draft. Keita Nagira: Conceptualization, Data curation, Investigation, Resources, Writing – original draft. Masako Hayashibara: Data curation, Investigation, Resources. Hiroshi Hagino: Conceptualization, Resources. Hideki Nagashima: Conceptualization. Takashi Wada: Data curation, Investigation. Yuki Kitsuda: Data curation, Investigation. Chikako Takeda: Data curation, Investigation. Mari Osaki: Data curation, Investigation. Yuta Takasu: Resources. Data availability statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Conflicts of interest Our institution has received research support from Simizu Hospital, Motomachi Hospital, Medical Corporation Jujikai, Healthcare Corporation Yowakai, Misasa Onsen Hospital, Tsuyama Daiichi Hospital, Hino Hospital, Chugai Pharmaceutical Co., Ltd., Asahi Kasei Pharma Co., Kaken Pharmaceutical Co., Ltd., Nippon Zoki Pharmaceutical Co., Ltd., Teijin Pharma Ltd., Otsuka Pharmaceutical Co., Ltd., Japan Medical Dynamic Marketing, INC., The Japanese Society for Spine Surgery and Related Research, Sakaguchi Limited Liability Co., Hamamoto Clinic, Yonago Brain & Nerve Clinic and The Western Japanese Research Society for Spine. Hideki Nagashima has received honoraria from Astellas Pharma, Inc., AO Spine, Hisamitsu Pharmaceutical Co., Inc.,Daiichi Sankyo Co., Ltd., Taisho Pharmaceutical Co., Ltd., Eisai Co., Ltd., Teijin Pharma Ltd., Asahi Kasei Pharma Co., Johnson & Johnson K.K., Medtronic, Amgen Inc., Kyowa Kirin Co., Ltd., Nipro Co., Smith & Nephew, Healthcare Corporation Yowakai and Takeda Pharmaceutical Co., Ltd. He also holds leadership role in Tottori Medical Association (paid) and unpaid roles in Japanese Society for Spine Surgery and Related Research, Japan Medical Society of Spinal Cord Lesion, Japan Spinal Instrumentation Society, The Japanese Society of Lumbar Spine Disorders, Japanese Society for Study of Bone and Joint Infections and The Chugoku-Shikoku Orthopaedic Association. Hiroshi Hagino has received honoraria from Amgen Inc., Astellas Pharma Inc., Eisai Co., Ltd., Kyowa Kirin Co., Ltd., Daiichi Sankyo Co. Ltd., Teijin Healthcare Ltd., Towa Pharmaceutical Co., Ltd., UCB Japan Co., Ltd., Asahi Kasei Pharma Co., Ltd., and Taisho Pharmaceutical Co., Ltd. Keita Nagira reports lecture honoraria from Ono Pharmaceutical Co., Ltd., Asahi Kasei Pharma Corporation, Mitsubishi Tanabe Pharma Corporation, AYUMI Pharmaceutical Corporation, Nippon Kayaku Co., Ltd., Taisho Pharmaceutical Co., Ltd., Eisai Co., Ltd., and Eli Lilly Japan K.K., and research support from Taisho Pharmaceutical Co., Ltd., not related to this work. Masako Hayashibara has received honoraria from Eisai Co., Ltd., Asahi Kasei Pharma Corporation, UCB Japan Co. Ltd. and Taisho Pharma Co., Ltd. Naoki Nakagawa has received honoraria from Teijin Pharma Ltd. All other authors declare that they have no conflicts of interest relevant to this study. Acknowledgments We sincerely thank the nurses, physical therapists, and clinical staff of the Department of Orthopedic Surgery at Tottori University Hospital for their dedicated support. ORCID Naoki Nakagawa: 0009-0007-4649-3823. Keita Nagira: 0000-0003-0973-6295. Masako Hayashibara: 0000-0001-9694-1726. Takashi Wada: 0000-0002-2902-9427. Yuki Kitsuda: 0000-0001-7598-206X. Yuta Takasu: 0009-0003-6995-8866. Chikako Takeda: 0000-0002-0895-9940. Mari Osaki: 0000-0002-0315-0616. Hiroshi Hagino: 0000-0001-8053-1095. Hideki Nagashima: 0000-0002-2896-8406. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.afos.2026.09.003. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.