Sex differences in long-term functional deterioration and permanent disability in spondyloarthritis: a 17-year follow-up from the REGISPON-3 study

Abstract

Objective

The primary objective was to assess sex differences in long-term functional deterioration and permanent work disability in spondyloarthritis (SpA), and the secondary aim to evaluate whether systemic inflammation mediates the association between biologic disease-modifying antirheumatic drugs (bDMARDs) exposure and functional change in a sex-dependent manner.

Methods

A longitudinal observational study was conducted in 411 patients with SpA from the Spanish REGISPONSER registry (baseline 2004–2007), who were reassessed 17 years later in REGISPON-3 (2021–2024). Outcomes were Bath Ankylosing Spondylitis Functional Index (BASFI) at follow-up, ΔBASFI and permanent work disability. Sex-stratified logistic and multivariable linear regression models were used. Moderated mediation analysis evaluated C reactive protein (CRP) as a mediator of the association between bDMARD exposure and ΔBASFI, with sex as a moderator.

Results

A total of 271 male patients with SpA (65.9%) and 140 female patients with SpA (34.1%) were included. At follow-up, female patients showed significantly worse functional status and greater functional deterioration (ΔBASFI 1.47±2.81 vs 0.38±2.57, p<0.001), whereas permanent disability rates were similar (27.1% vs 32.1%, p=0.31). Higher Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and Bath Ankylosing Spondylitis Radiology Index (BASRI)-total were independently associated with disability in both sexes. Baseline BASFI and BASDAI predicted ΔBASFI in both sexes; additional predictors in male patients included BASRI-total and occiput-to-wall distance, whereas BASRI-column was relevant in female patients. The indirect effect of bDMARD exposure on ΔBASFI through CRP was significant in male patients but not in female patients with a significant sex interaction.

Conclusion

Female patients experienced greater long-term functional deterioration despite similar work disability rates at follow-up. The relationship between systemic inflammation and long-term functional outcomes may differ by sex.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Functional decline and work disability are important long-term outcomes in spondyloarthritis (SpA).

• Although inflammation and structural damage contribute to disease burden, it is unclear if these outcomes differ by sex.

WHAT THIS STUDY ADDS

• This study shows that worse functional outcomes in women do not translate into higher rates of permanent work disability at follow-up.

• It also suggests that long-term functional impairment is not explained by inflammation alone, particularly in women.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings support a sex-aware approach to long-term SpA outcomes and highlight the need to consider factors beyond objective inflammation when assessing functional impairment.

Introduction

Spondyloarthritis (SpA) represents a spectrum of chronic inflammatory rheumatic diseases characterised by axial and peripheral musculoskeletal involvement, often associated with extra-articular features such as uveitis, psoriasis and inflammatory bowel disease.^{1 2} Functional impairment is an important determinant of long-term outcomes in SpA, with substantial implications for quality of life. The Bath Ankylosing Spondylitis Functional Index (BASFI) is widely used to quantify patient-reported functional limitation and has been validated across different SpA populations.^{3 4} Long-term functional impairment in SpA reflects the cumulative effects of sustained disease activity, structural progression and reduced mobility,⁵,⁸ leading to a decreased quality of life and work participation.⁹,¹¹

Sex-related differences in SpA have become increasingly acknowledged in recent years. Male patients tend to present with earlier disease onset, more pronounced radiographic spinal involvement and a more favourable response and retention rate to tumour necrosis factor inhibitors (TNFi).¹²,¹⁴ In contrast, female patients more frequently exhibit peripheral manifestations and report higher levels of pain, fatigue, disease activity and poorer quality of life despite lower C reactive protein (CRP) levels and less radiographic structural damage.¹⁵,²¹ These discrepancies suggest a partial dissociation between patient-reported outcomes (PROs) and objective inflammatory or structural findings in SpA, particularly among female patients.¹⁷²⁰,²² Proposed contributors include non-inflammatory pain mechanisms, fatigue, mood disorders, fibromyalgia, psychosocial burden and differences in immune response.^{1819 21},²³ Recent registry data indicate socioeconomic vulnerability may also affect PROs, especially pain severity.²⁴ Although biologic disease-modifying antirheumatic drugs (bDMARDs) have substantially improved disease control, the mechanisms underlying long-term functional deterioration and disability progression remain incompletely understood, especially regarding sex differences. Most previous studies have focused predominantly on axial disease, short-term treatment response or radiographic progression,^{12 13 25 26} with limited data evaluating long-term real-world functional outcomes across the broader SpA spectrum.

Using data from the Spanish REGISPONSER registry²⁷ and its 17-year follow-up study REGISPON-3 (Spanish Registry of Spondyloarthritis - Phase 3),²⁸ we aimed to (1) assess sex differences in long-term functional outcomes and permanent work disability across patients with SpA and (2) examine whether the association between bDMARD exposure and functional change is mediated by inflammatory activity in a sex-dependent manner.

Patients and methods

Study design and population

This longitudinal observational study used data from the Spanish REGISPONSER cohort,²⁷ with its follow-up reassessment in REGISPON-3 after 17 years.²⁸ The cohort was multicentric, including patients recruited from several Spanish rheumatology centres participating in the REGISPONSER registry and its 17-year follow-up reassessment, REGISPON-3, as previously described in a recent publication by Puche-Larrubia et al.²⁸ Data were collected during single visits at each registry, enabling analysis of the same patient cohort over a 17-year period. Although no prospective intermediate study visits were performed, relevant clinical information between both assessments, including treatment exposure and major clinical events, was retrospectively collected through standardised medical record review and structured follow-up interviews, according to the REGISPON-3 study protocol.²⁸ Participants were included in this study if they received a diagnosis of SpA by a rheumatologist at their initial REGISPONSER baseline visit using the European Spondyloarthropathy Study Group (ESSG) classification criteria.²⁹

For this analysis, patients were eligible if they participated in both registry waves and completed baseline and follow-up functional assessments. A total of 411 patients (271 male patients and 140 female patients) met the inclusion criteria and were included in the study. The study was conducted following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.³⁰

Data collection

Clinical and demographic data were collected during face-to-face standardised registry visits at baseline and follow-up using harmonised case report forms.

At the baseline visit, demographic data included age at symptom onset, age at first signs/symptoms, age at diagnosis and sex. Clinical characteristics recorded included disease duration, human leucocyte antigen B27 status and the presence of extra-articular manifestations such as uveitis, psoriasis and inflammatory bowel disease, covering all well-known characteristics of SpA. Functional ability was evaluated using the BASFI score.³ Permanent work disability was defined as an officially recognised disability due to SpA, recorded in the registry using standardised case report forms with predefined categories (none, temporary or permanent disability).³¹ Disease activity was measured through the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI)³² and the Ankylosing Spondylitis Disease Activity Score (ASDAS).³³ BASDAI, BASFI, ASDAS and mobility assessments were collected according to the standardised registry protocol across the whole SpA cohort, including patients with peripheral-dominant phenotypes, to ensure consistency of longitudinal assessment across participating centres. Structural damage was assessed through plain radiographs of the sacroiliac joints, cervical and lumbar spine and hips and quantified using the Bath Ankylosing Spondylitis Radiology Index (BASRI).³⁴ Clinical and radiographic assessments were performed locally following a standardised protocol. Due to the multicentre, long-term nature of the study, assessments were conducted across participating centres instead of a central reading. Investigators were trained beforehand to standardise procedures. Socioeconomic status was assessed using the modified Graffar scale.³⁵ Treatment data included the use of non-steroidal anti-inflammatory drugs, corticosteroids, conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) and bDMARDs. PROs were gathered using standardised questionnaires assessing pain levels, fatigue and quality-of-life questionnaires, such as the Short Form-12 Health Survey.³⁶

At follow-up, updated clinical, laboratory, functional ability and treatment data were collected. The primary outcomes were permanent work disability at follow-up, defined as recognised disability due to SpA, recorded as no disability, temporary or permanent. Long-term functional deterioration was defined as the change in BASFI from baseline to follow-up (ΔBASFI).

Statistical analysis

Continuous variables are presented as means with SD and categorical variables as frequencies and percentages. All analyses were stratified by sex. Between-sex comparisons were performed using independent samples t-tests or Mann-Whitney U tests for continuous variables and χ² or Fisher’s exact tests for categorical variables.

Univariable logistic regression analyses were conducted separately in male and female patients to identify baseline predictors of permanent disability at follow-up. All variables that were statistically significant (p<0.05) or considered clinically relevant based on previous literature and expert clinical judgement were subsequently entered into sex-stratified multivariable logistic regression models.

To examine long-term functional deterioration, sex-specific univariable linear regression analyses were performed with ΔBASFI as the outcome. Variables that met either statistical significance or clinical relevance, based on established associations with functional outcomes in SpA reported in previous studies, were incorporated into the final multivariable models. Multicollinearity was assessed using the variance inflation factor (VIF), with acceptable values below 5. The model’s performance was assessed through adjusted R² values.

To explore whether inflammatory burden mediated the association between biologic treatment exposure and long-term functional change, a moderated mediation analysis was performed using Model 8 of the PROCESS macro (V.4.2) for SPSS, estimating indirect effects with bias-corrected bootstrapped CIs.^{37 38} In this model, bDMARD exposure (ever vs never) was the independent variable, ΔBASFI the dependent variable, CRP the mediator and sex the moderator. CRP was analysed as a continuous variable; therefore, patients with normal CRP levels were retained in the moderated mediation analyses. Age and disease duration were included as covariates to capture different aspects of disease burden. Although possibly correlated, multicollinearity was assessed using VIF values, which remained below 5. The moderated mediation index was used to evaluate whether the indirect effect differs significantly between male and female patients.

No formal sample size calculation was performed, as all eligible patients participating in both registry waves were included in the analysis.

Missing data for the variables used in the analyses were minimal (<5%) and, importantly, there were no significant differences between complete and incomplete cases in terms of age, sex or functional ability-related variables. All statistical tests were two-tailed, with p<0.05 considered statistically significant. Moderated mediation analyses were performed using the PROCESS macro (V.4.2) for SPSS. All other analyses were conducted in Python (V.3.10).

Results

Baseline characteristics

A total of 411 patients with SpA were included, of whom 271 (65.9%) were male. Table 1 shows baseline and follow-up demographic and clinical data stratified by sex. All patients fulfilled the ESSG classification criteria for SpA at inclusion. At baseline, 270 patients met the modified New York criteria³⁹ for ankylosing spondylitis, corresponding to radiographic axial spondyloarthritis according to current terminology, of whom 196 (72.3%) were male. Additionally, 60 patients (28 females) were classified as having psoriatic arthritis, while 81 had other forms of SpA.

Table 1. Baseline and follow-up cohort characteristics.

	Total N=411	Male N=271	Female N=140	p-value
Baseline visit (2004–2007)
Diagnosis, % (n)
r-axSpA	65.7% (270)	72.3% (196)	52.9% (74)	0.001
PsA	14.6% (60)	11.8% (32)	20% (28)
uSpA	14.8% (61)	10.7% (29)	22.9% (32)
ReA	1.9% (8)	1.8% (5)	2.1% (3)
IBD arthritis	0.7% (3)	1.1% (3)	0
Juv-SpA	2.2% (9)	2.2% (6)	2.1% (3)
Age, mean (SD)	44±10.3	44.5±10.5	42.9±9.8	0.070
Age at disease onset, mean (SD)	26.2±10.5	25.8±10.3	26.8±10.8	0.390
Age at diagnosis, mean (SD)	33.2±10.2	32.9±11.2	33.7±9.8	0.360
Evolution time (years) from first sign/symptom, mean (SD)	17.6±10.8	18.6±11.1	15.8±10.8	0.020
Diagnostic delay (years), mean (SD)	6.8±8.5	7.05±8.8	6.6±8.02	0.620
Disease duration (years) from diagnosis, mean (SD)	10.7±8.4	11.5±8.9	9.1±7.1	0.004
ESR mm/hour, mean (SD)	18.1±14.8	16.5±15.4	21.3±13.2	<0.001
CRP mg/L, mean (SD)	9.9±12.7	11.6±14.7	6.5±6.1	0.007
Elevated CRP (>5 mg/L), % (n)	51.9% (205)	54.6% (143)	46.6% (62)	0.135
SpA family history, % (n)	17.4% (69)	15.8% (41)	20.4% (28)	0.260
HLA-B27 positivity, % (n)	81.3% (282)	83.3% (194)	77.2% (88)	0.220
Radiographic sacroiliitis, % (n)	83.2% (341)	87.8% (238)	74.1% (103)	<0.001
Asymmetric arthritis, % (n)	43.4% (177)	39.3% (106)	51.4% (71)	0.020
Enthesitis, % (n)	16.4% (67)	13.3% (36)	22.3% (31)	0.020
Dactylitis, % (n)	13% (53)	11.1% (30)	16.5% (23)	0.160
Uveitis, % (n)	17.6% (72)	15.5% (42)	21.7% (30)	0.150
Psoriasis, % (n)	21.8% (89)	18.6% (50)	27.9% (39)	0.040
IBD, % (n)	4.9% (20)	5.5% (15)	3.6% (5)	0.520
Clinical phenotype
Axial-predominant	84.7% (348)	88.2% (239)	77.9% (109)	0.006
Peripheral-predominant	15.3% (63)	11.8% (32)	22.1% (31)
Swollen joints (≥1 swollen joint)	15.1% (62)	12.2% (33)	20.9% (29)	0.020
bDMARDs, % (n)	13.1% (54)	12.9% (35)	13.6% (19)	0.870
csDMARDs ever, % (n)	29.5% (121)	25.8% (70)	36.7% (51)	0.020
Permanent disability, % (n)	18.1% (73)	21.8% (58)	10.9% (15)	0.007
BASDAI, mean (SD)	4.2±2.2	4.1±2.2	4.6±2.3	0.010
ASDAS, mean (SD)	2.7±1	2.7±1.1	2.7±0.9	0.600
BASFI, mean (SD)	3.3±2.5	3.2±2.6	3.3±2.5	0.640
SF-12 physical component, mean (SD)	37.2±7.7	38.2±7.5	35.3±7.9	<0.001
SF-12 mental component, mean (SD)	50.1±7.3	50±6.2	50.1±9.1	0.200
BASRI-total, mean (SD)	5±3.6	5.8±3.9	3.3±2.4	<0.001
BASRI-column, mean (SD)	4.5±3.1	5.2±3.3	3.1±2.1	<0.001
BMI kg/m2, mean (SD)	26.5±4.2	27.1±3.9	25.4±4.4	<0.001
Graffar scale
Low (lower+middle-lower)	51.5% (102)	54.5% (73)	45.4% (29)	0.700
Middle class	32.8% (65)	29.9% (40)	39.1% (25)
Upper (upper-middle+upper)	15.6% (31)	15.6% (21)	15.7% (10)
Follow-up visit (2021–2024)
BASFI, mean (SD)	4.0±2.6	3.6±2.6	4.7±2.5	<0.001
ΔBASFI, mean (SD)	0.7±2.7	0.3±2.5	1.4±2.8	<0.001
Permanent disability, % (n)	30.4% (125)	32.1% (87)	27.1% (38)	0.310
bDMARDs exposure, % (n)	51.6% (212)	51.7% (140)	51.4% (72)	0.960
FiRST, mean (SD)	2.1±1.9	1.6±1.7	2.9±1.9	<0.001
FiRST positive (≥5), % (n)	15.1% (62)	8.9% (24)	27.1% (38)	<0.001

Baseline and follow-up demographic and clinical characteristics according to sex. This table was adapted from our previous analysis of the same REGISPON-3 cohort,¹⁴ with modifications and additional variables relevant to the present study.

Bold p-values indicate statistically significant differences between males and females (p<0.05).

Percentages are based on available data; denominators vary due to missing data.

ASDAS, Ankylosing Spondylitis Disease Activity Score; BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; BASFI, Bath Ankylosing Spondylitis Functional Index; BASRI, Bath Ankylosing Spondylitis Radiology Index; bDMARDs, biologic disease-modifying antirheumatic drugs; BMI, body mass index; CRP, C reactive protein; csDMARDs, conventional synthetic disease-modifying antirheumatic drugs; ESR, erythrocyte sedimentation rate; FiRST, Fibromyalgia Rapid Screening Tool; HLA-B27, human leucocyte antigen B27; IBD, inflammatory bowel disease; Juv-SpA, juvenile-onset spondyloarthritis; PsA, psoriatic arthritis; r-axSpA, radiographic axial spondyloarthritis; ReA, reactive arthritis; SF-12, Short Form-12 Health Survey; uSpA, undifferentiated spondyloarthritis.

At baseline, male participants exhibited a longer duration of disease from diagnosis (11.5 vs 9.1 years, p=0.004) and greater radiographic damage, as evidenced by elevated BASRI-total scores (5.8 vs 3.3, p<0.001), along with a higher prevalence of sacroiliitis (87.8% vs 74.1%, p<0.001). Females showed higher disease activity (BASDAI 4.6 vs 4.1, p=0.01) and had more frequent arthritis (51.4% vs 39.3%, p=0.02) and more enthesitis (22.3% vs 13.3%, p=0.02), and they were also more frequently treated with csDMARDs (36.7% vs 25.8%, p=0.02).

Functional ability status and permanent disability

After 17 years, the rate of permanent work disability was similar between male and female patients (males: 32.1%, females: 27.1%; p=0.3). However, females experienced greater functional decline, indicated by higher BASFI scores (mean±SD: 4.7±2.5 vs 3.6±2.6; p<0.001) and higher ΔBASFI values (1.4±2.8 vs 0.3±2.5; p<0.001).

Simple logistic regression did not find a significant association between sex and permanent disability at the follow-up visit (OR 0.79, 95% CI 0.50 to 1.24; p=0.30). In univariable logistic regression analyses stratified by sex, higher disease activity (BASDAI, ASDAS), greater structural damage (BASRI), worse functional status (baseline BASFI), impaired mobility measures and higher BMI were associated with increased odds of permanent disability in male patients. In female patients, baseline BASFI, ASDAS and BASRI-column were significantly associated with permanent disability. Full univariable results are presented in online supplemental table S1.

In the multivariable logistic regression analysis presented in table 2, higher baseline BASDAI and BASRI-total scores independently predicted permanent disability across both sexes. Specifically, in male patients, BASDAI (OR 1.19, 95% CI 1.04 to 1.37, p=0.01) and BASRI-total (OR 1.13, 95% CI 1.05 to 1.21, p<0.001) remained statistically significant predictors. Similarly, in female patients, BASDAI (OR 1.28, 95% CI 1.03 to 1.60, p=0.02) and BASRI-total (OR 1.19, 95% CI 1.01 to 1.40, p=0.04) were independently associated with disability.

Table 2. Sex-stratified multivariable logistic regression for permanent disability at follow-up.

Predictor	Males OR (95% CI)	p-value	Females OR (95% CI)	p-value
BASDAI	1.19 (1.04 to 1.37)	0.01	1.28 (1.03 to 1.60)	0.02
BASRI-total	1.13 (1.05 to 1.21)	<0.001	1.19 (1.01 to 1.40)	0.04
SF-12 physical component	0.68 (0.44 to 1.03)	0.07	1.60 (0.75 to 3.41)	0.22

Additional variables included in the multivariable models (mobility measures, ESR, CRP, ASDAS, baseline BASFI, BMI and disease duration) were not independently associated with permanent disability.

OR, odds ratio; CI, confidence interval.

ASDAS, Ankylosing Spondylitis Disease Activity Score; BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; BASFI, Bath Ankylosing Spondylitis Functional Index; BASRI, Bath Ankylosing Spondylitis Radiology Index; BMI, body mass index; CRP, C reactive protein; ESR, erythrocyte sedimentation rate; SF-12, Short Form-12 Health Survey.

Long-term functional deterioration (ΔBASFI) analysis

ΔBASFI was calculated as follow-up BASFI minus baseline BASFI; therefore, positive values indicate functional deterioration, whereas negative values indicate functional improvement. Female patients experienced significantly greater long-term functional decline than male patients (ΔBASFI: 1.4±2.8 vs 0.3±2.5, p<0.001).

In univariable linear regression analyses, baseline disease activity scores, baseline BASFI, PROs and certain mobility parameters were associated with greater ΔBASFI in both sexes. Structural damage measures were associated with ΔBASFI, particularly in male patients. Detailed results are available in online supplemental table S2.

Sex-stratified multivariable linear regression models are presented in table 3. In both sexes, higher baseline BASDAI was independently associated with a greater increase in BASFI, whereas higher baseline BASFI was independently associated with a smaller increase in BASFI. Among males, increased structural damage (BASRI-total; B=0.17, 95% CI 0.06 to 0.27, p=0.001) and reduced cervical mobility (occiput-wall distance; B=0.08, 95% CI 0.02 to 0.14, p=0.007) remained independently associated with ΔBASFI. Among females, the BASRI-column (B=0.27, 95% CI 0.05 to 0.50, p=0.01) was independently associated with functional deterioration. The models accounted for a similar proportion of variance in both males and females (adjusted R²=0.31 and 0.30, respectively).

Table 3. Sex-stratified multivariable linear regression for ΔBASFI.

Predictor	Males, B (95% CI)	p-value	Females, B (95% CI)	p-value
BASDAI	0.30 (0.13 to 0.48)	0.001	0.25 (0.02 to 0.47)	0.02
Baseline BASFI	−0.75 (−0.91 to −0.58)	<0.001	−0.73 (−0.95 to −0.52)	<0.001
Occiput-to-wall distance, cm	0.08 (0.02 to 0.14)	0.007	–	–
BASRI-total	0.17 (0.06 to 0.27)	0.001	–	–
BASRI-column	–	–	0.27 (0.05 to 0.50)	0.01

B, unstandardized regression coefficient; CI, confidence interval.

ΔBASFI, follow-up BASFI minus baseline BASFI.

Both models were adjusted for disease duration and bDMARDs exposure (ever). Additional variables included in the multivariable models (ESR, HLA-B27 status, lumbar back pain, ASDAS, SF-12 physical component, diagnostic delay, BMI and lumbar lateral flexion) were not independently associated with ΔBASFI.

ASDAS, Ankylosing Spondylitis Disease Activity Score; BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; BASFI, Bath Ankylosing Spondylitis Functional Index; BASRI, Bath Ankylosing Spondylitis Radiology Index; bDMARDs, biologic disease-modifying antirheumatic drugs; BMI, body mass index; ESR, erythrocyte sedimentation rate; HLA-B27, human leucocyte antigen B27; SF-12, Short Form-12 Health Survey.

Moderated mediation analysis

To explore whether systemic inflammation (measured by CRP) influenced long-term functional deterioration (ΔBASFI), we conducted a moderated mediation analysis (figure 1; table 4). In male patients, part of the beneficial effect of bDMARD exposure on functional change (ΔBASFI) appeared to be related to reductions in CRP levels, as reflected by a significant indirect effect (β=−0.21, 95% CI −0.43 to −0.04). In contrast, this pattern was not observed in female patients, for whom the indirect effect was not statistically significant (β=−0.04, 95% CI −0.23 to 0.11). The moderated mediation index confirmed that this pathway differed significantly between sexes (β=0.17, 95% CI 0.03 to 0.33). Analyses were adjusted for age and disease duration.

Figure 1. Moderated mediation model of the association between biologic disease-modifying antirheumatic drug (bDMARD) exposure and functional change (change in Bath Ankylosing Spondylitis Functional Index (ΔBASFI)), with C reactive protein (CRP) as mediator and sex as moderator. The diagram illustrates the moderated mediation model used to assess whether the association between bDMARD exposure and functional deterioration (ΔBASFI) is mediated by CRP and moderated by sex. The indirect effect of bDMARD exposure on ΔBASFI through CRP was significant in male patients but not in female patients. The moderated mediation index indicates that this indirect effect differed significantly between sexes. The values shown correspond to the estimates reported in table 4. NS, not statistically significant.

Table 4. Moderated mediation analysis results for ΔBASFI with CRP as mediator and sex as moderator.

Mediator	Males indirect effect (95% CI)	Females indirect effect (95% CI)	Moderated mediation index (95% CI)
CRP	−0.21 (−0.43 to −0.04)	−0.04 (−0.23 to 0.11)	0.17 (0.03 to 0.33)

CI, confidence interval.

Indirect effects represent the mediation of bDMARD exposure on ΔBASFI through the mediator (CRP), stratified by sex. The moderated mediation index indicates whether the mediation effect differs significantly between sexes.

bDMARD, biologic disease-modifying antirheumatic drug; CRP, C reactive protein; ΔBASFI, change in Bath Ankylosing Spondylitis Functional Index.

Discussion

In this longitudinal SpA cohort with a 17-year follow-up, we identified sex-specific differences in long-term functional outcomes and in the mechanisms associated with functional deterioration. Female patients showed greater functional decline, as reflected by ΔBASFI, despite similar rates of permanent work disability compared with male patients. Baseline disease activity and baseline functional impairment were independently associated with subsequent deterioration in BASFI scores in both sexes, whereas structural damage and reduced spinal mobility appeared more strongly associated with long-term functional loss in male patients. Importantly, the association between bDMARD exposure, systemic inflammation and functional change differed by sex, with a significant CRP-mediated pathway observed only in male patients. These findings suggest that long-term functional impairment in male patients is more closely associated with inflammatory burden, whereas in female patients, additional factors beyond systemic inflammation may play a more significant role.

The observed sex disparity in long-term functional outcomes aligns with previous research indicating that female patients with axial SpA (axSpA) often report worse PROs, including increased pain, higher levels of fatigue and diminished health-related quality of life compared with male patients, despite exhibiting less radiographic damage and experiencing slower structural progression.^{12 13 36} In males, functional impairment seems more closely associated with inflammatory burden and structural damage, whereas in females, additional factors such as pain perception, fatigue and comorbidities may play a more significant role in functional limitations.^{9 19} While some studies report no significant sex differences in BASFI scores among patients with radiographic axSpA,²⁰ others have shown that female patients report worse functional outcomes than male patients among patients treated with TNFi, indicating persistent sex differences in patient-reported disease burden.²¹ Additionally, delays in diagnosis, which are more common in female patients, have been associated with worse BASFI scores, reduced spinal mobility and greater radiographic progression.²²

In our study, the rates of permanent work disability due to SpA were similar between male and female patients, despite the greater functional ability deterioration observed among female patients. This suggests that a decline in functional status, as captured by ΔBASFI, does not necessarily translate directly into permanent work disability. Higher baseline disease activity and greater radiographic damage were associated with permanent work disability at follow-up in our analysis. Previous registry studies have similarly shown that disease severity and structural progression are important determinants of work disability and reduced work participation in axSpA.⁴⁰,⁴²

A notable finding of our study was the sex-specific mediation pattern observed in the analysis. Inflammation (quantified by CRP) mediated the association between bDMARD exposure and ΔBASFI in male patients but not in female patients. In male patients, the indirect effect was statistically significant, suggesting that part of the functional benefit associated with bDMARD therapy may be explained by the reduction of systemic inflammation. In contrast, this pathway was not observed in female patients, suggesting that the relationship between bDMARD exposure and long-term functional outcomes may be less strongly related to CRP levels. Previous real-world studies have also reported sex differences in response to TNFi and other biologics, with female patients often demonstrating smaller improvements in disease activity and PROs compared with male patients.^{12 43} Biological differences in immune regulation may contribute to these observations.^{44 45} This sex-specific pattern may also reflect the known discordance between PROs and inflammatory markers, such as CRP, in female patients with SpA. Previous studies have shown that female patients often report higher disease activity, pain and fatigue, while male patients tend to show more radiographic damage and higher objective inflammatory burden.^{16 21 43} Therefore, the absence of a significant CRP-mediated pathway in female patients should not necessarily be interpreted as absence of clinically relevant disease impact. Instead, the decline in function observed in female patients may be influenced by a broader range of factors, including fatigue, comorbid pain syndromes, fibromyalgia, central sensitisation and psychosocial burden, which may affect BASFI independently of CRP levels.^{18 19 23 43} In line with this hypothesis, female patients in our cohort showed higher Fibromyalgia Rapid Screening Tool (FiRST) scores at follow-up, suggesting a greater burden of fibromyalgia-related symptoms. This may partly explain why greater functional deterioration in female patients did not translate into higher permanent work disability rates. Work disability depends on disease activity and on occupational demands, physical function, spinal mobility, socioeconomic conditions, social support and coping strategies.^{40 46 47}

Clinical implications

Our findings suggest that long-term functional outcomes in SpA should be interpreted with attention to potential sex differences. In male patients, the association between disease activity, structural damage and disability reinforces the importance of sustained inflammatory control and prevention of structural progression. In female patients, functional impairment may be less closely related to inflammatory markers alone, and clinicians should also consider non-inflammatory contributors such as comorbid pain syndromes, fatigue or psychosocial burden, which are recognised in axSpA.^{18 43} Future studies on sex-specific molecular and biological profiles may help to better understand these differences. Functional assessment remains essential; however, BASFI results should be interpreted within the wider clinical context and, where appropriate, supplemented with comprehensive measures such as the ASAS Health Index and its derived utility measures, which capture broader physical and psychosocial health aspects in SpA.^{48 49}

Strengths and limitations

The main strength of this study lies in its long-term follow-up, with reassessment of the same cohort after 17 years, a design rarely achieved in SpA research. Using validated outcome measures and sex-stratified analyses enhances the robustness and clinical relevance of the findings.

The inclusion of different SpA subtypes may have introduced clinical heterogeneity, potentially limiting subtype-specific interpretation of long-term functional trajectories and disability mechanisms. Future studies specifically designed to evaluate axial and psoriatic phenotypes separately may provide further insight into sex-specific disease patterns. In addition, although relevant clinical information between the two visits was retrospectively reconstructed using standardised medical record review and structured interviews, prospective intermediate assessments were not available; therefore, temporal fluctuations in disease activity, treatment response and functional progression during follow-up could not be fully characterised.

Although general health-related quality of life data were collected, fibromyalgia, depressive symptoms and other pain-related or psychosocial comorbidities were not systematically assessed at the baseline visit and therefore could not be formally controlled for in the longitudinal analyses. At follow-up, fibromyalgia-related symptoms were assessed using the FiRST questionnaire,⁵⁰ showing a higher burden among women; however, these data were not available at baseline and should therefore be interpreted descriptively. These factors may have influenced PROs such as BASDAI and BASFI, particularly among female patients, and may partly contribute to sex-related differences in functional deterioration. Furthermore, some disease activity and functional instruments included in the registry, such as BASDAI, BASFI and ASDAS, were developed for axial disease and may have more limited applicability in peripheral-dominant SpA phenotypes. However, the cohort was predominantly axial, and REGISPON-3 documented diagnostic transitions over 17 years, supporting the use of standardised longitudinal assessments across the whole SpA cohort.²⁸ Additionally, several sociodemographic variables were excluded from the analysis due to missing data, potentially limiting the capacity to adjust for possible socioeconomic confounders. Also, bDMARD exposure was analysed as a dichotomous variable (ever vs never exposed). Cumulative exposure, treatment duration and persistence may also influence long-term functional outcomes, but were beyond the scope of the present analysis and have been evaluated separately in the same REGISPON-3 cohort.¹⁴ Finally, the prolonged 17-year follow-up may have introduced attrition or survivor bias, as patients lost to follow-up or who died may differ from those reassessed.

Conclusion

In this long-term SpA cohort, female patients experienced greater functional deterioration despite similar rates of permanent work disability. While disease activity and structural damage were associated with functional outcomes in both sexes, the association between bDMARD exposure, systemic inflammation and functional change appeared to differ by sex. These findings suggest that long-term functional outcomes in female patients are less related to inflammation, supporting a sex-aware, personalised approach.

Data availability statement

Data are available on reasonable request.