Abstract
Objective. To determine (1) the prevalence of central obesity in axial spondyloarthritis (axSpA) and its effect on disease-related outcomes and (2) how this differs between sexes.
Methods. Data were extracted from the Ankylosing Spondylitis Registry of Ireland. Patients with physical measurements for the calculation of anthropometric measures were included. BMI and waist-to-hip ratio (WHR) were used to compare classifications of obesity. Comparison analyses based on sex and central obesity were carried out. Multivariate analysis examined the effects of these factors on the following patient-reported outcomes: the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), the Bath Ankylosing Spondylitis Functional Index (BASFI), the Ankylosing Spondylitis Quality of Life (ASQoL) questionnaire, and the Health Assessment Questionnaire (HAQ).
Results. In total, 753 patients were included in the analysis. Of these patients, 29.6% (n = 223) were classified as obese based on their BMI, and 41.3% (n = 311) were classified as centrally obese according to the WHR. The prevalence of central obesity was significantly higher among women with axSpA compared to men (71.6% vs 29.9%, P < 0.01). Central obesity had a clear effect on patient outcomes, regardless of sex. Presence of central obesity was associated with significantly worse BASFI scores (P < 0.01), HAQ scores (P < 0.01), and ASQoL questionnaire scores (P = 0.01), with a nonsignificant trend toward worse BASDAI scores (P = 0.07).
Conclusion. There was a high prevalence of central obesity as assessed by the WHR in axSpA, most notably among women with axSpA. This modifiable comorbidity was significantly associated with worse quality of life, greater impairment of functional ability, and a trend toward worse disease activity. Regular use of the WHR to screen for central obesity as part of an axSpA assessment would provide an opportunity for prompt identification and intervention for at-risk patients.
Worldwide prevalence of obesity has been steadily increasing1,2 despite numerous public health campaigns to increase public awareness. Obesity is known to be associated with an increased risk of all-cause mortality3 and is the second most common preventable cause of death in the United States.4 As the prevalence of this health issue surges, so too do the frequencies of associated complications, such as type 2 diabetes and cardiovascular disease, which have a significant effect on patients’ overall quality of life (QOL) and place significant demands on the healthcare system.5
This rising prevalence is especially concerning in axial spondyloarthritis (axSpA), an inflammatory arthritis predominantly involving the axial spine.6 Obesity in axSpA has previously been associated with higher levels of disease activity7,8 and decreased response to treatment.9 The presence of obesity has even been linked to new syndesmophyte formation and enthesitis.10 This is related to adipose tissue secretion of proinflammatory cytokines, called adipocytokines, which in high concentrations can promote a persistent systemic inflammatory state.11 Serum levels of adipocytokines are directly related to body fat content, with elevated levels contributing to the development of metabolic disorders, including insulin resistance and accelerated atherosclerosis.12 Despite the established negative effect of obesity in axSpA, it remains a frequently encountered comorbidity with a reported prevalence of 14% to 27%.13,14
The presence of obesity is commonly determined by calculation of BMI. However, additional anthropometric scores have been developed to move beyond weight excess and focus on determining the prevalence of abdominal or central obesity. The waist-to-hip ratio (WHR) is a validated measure that is used to screen for central obesity in the general population.15 Central deposition of adipose tissue is associated with increased risk of cardiovascular disease, type 2 diabetes, and premature death.16,17,18,19 Typically, central obesity is more often found among males because of deposition of visceral fat, whereas females are more prone to gluteal-femoral subcutaneous fat deposition.20 During menopause, there is a shift toward more central fat deposition, which increases further in the postmenopausal period.21,22
The aim of this study was to capture the prevalence of central obesity in a large cohort of patients with axSpA and assess how this affects disease-related outcomes. This analysis also examined how this differs between men and women with axSpA.
METHODS
Patient population. Data for the analysis were extracted from the Ankylosing Spondylitis Registry of Ireland (ASRI), a rich source of epidemiological data on axSpA in Ireland. This large national registry recruited patients from 12 rheumatology centers, representing all major geographic regions of the country. Since being launched in 2013, 892 patients have been enrolled in the ASRI, with recruitment ongoing in multiple centers. This analysis included all participants with anthropometric data recorded who were enrolled in the ASRI since the time of its inception.
All participants were required to meet several inclusion criteria in order to be considered for enrollment: diagnosed with axSpA by a rheumatologist; met the Assessment of SpondyloArthritis international Society (ASAS) criteria for axSpA at the time of enrollment, which allowed for data collection on both radiographic axSpA (r-axSpA) and nonradiographic axSpA (nr-axSpA); aged > 18 years; had full capacity; and fluent in English. All eligibility criteria were required at the time of enrollment in order to be included in the ASRI.
Written informed consent was obtained from each patient prior to enrollment. The consent process included discussion of the purpose of data collection, data protection, and publication of studies using data from the ASRI. Patients were informed of their right to withdraw their data at any time and encouraged to discuss questions with investigators prior to signing the consent declaration. Ethical approval was received from local hospital ethics boards at each site that contributed patients to the ASRI. For the main center, St James’s Hospital, the study was approved by the St James’s Hospital/Tallaght University Hospital Joint Research Ethics Committee (approval no. 2019-07[27]; Supplementary Table 1, available with the online version of this article).
Outcome definitions and assessment. Each participant attended a single clinical visit conducted by a trained investigator, allowing all data to be collected at a single timepoint. Medical records were reviewed for radiology results, patterns of disease, and medication usage. Data were collected on participant demographics; disease duration; delay to diagnosis; patterns of disease, including both articular and extraarticular manifestations; and presence of comorbidities.
Participants completed self-administered questionnaires to provide information on patient outcomes; responses were recorded on a numerical scale, with higher scores indicating greater severity. Specific outcomes that were investigated include those measured by the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), the Bath Ankylosing Spondylitis Functional Index (BASFI), Ankylosing Spondylitis Disease Activity Score (ASDAS), the Health Assessment Questionnaire (HAQ), and the Ankylosing Spondylitis Quality of Life (ASQoL) questionnaire. A focused clinical examination measured spinal mobility to allow for calculation of the Bath Ankylosing Spondylitis Metrology Index (BASMI). Physical measurements were also recorded: waist circumference measured halfway between the lowest rib and the iliac crest, and hip circumference measured at the widest part of the hips and buttocks region.23 All outcome measures in this analysis were validated for use in axSpA.
Participants were analyzed on the basis of sex and the presence of obesity. Obesity was defined by BMI, calculated as weight in kilograms divided by height in meters squared; a participant with a result > 30 was categorized as obese, as per the Centers for Disease Control and Prevention definition.24 Central obesity was assessed by the WHR and defined according to the World Health Organization guidelines25 (Table 1).
WHR definitions for classification of central obesity as per the WHO25.
Statistical analysis: IBM SPSS Statistics for Windows (version 26; IBM Corp) was used for statistical analysis and data processing. Comparison analyses determined variation in the prevalence of obesity and variation between sexes. Categorical variables were recorded as frequencies with percentages, with a chi-square test for independence or a Fisher exact test for statistical significance. Means and SDs were calculated for numerical variables within each group, with an independent t test or Mann-Whitney U test to assess the significance of differences between groups. A Shapiro-Wilk test was carried out on all included variables to determine normality of the distribution. To assess the relationship between the WHR and BMI, a Spearman rank-order correlation was run because variables were not normally distributed. Scatterplots were constructed to visualize the presence of a monotonic relationship for each analysis.
A 2-way multivariate ANOVA assessed the effects of sex (ie, male vs female) and central obesity, as measured by the WHR, on patient-reported outcomes (PROs; ie, BASDAI, BASFI, ASQoL, and HAQ). An α level of < 0.05 was deemed significant. A binary logistic regression model examined the effects of age, sex (ie, male vs female), axSpA classification (ie, radiographic vs nonradiographic), BASDAI score, BASFI score, and BASMI score on the likelihood of central obesity. A Bonferroni correction was applied to determine P values for levels of statistical significance of linearity. All necessary assumptions for each statistical test were met.
RESULTS
At data extraction, physical measurements were available for 753 participants enrolled in the ASRI; of these participants, 88.4% (n = 666) were White, 27.6% (n = 208) were female, and 71.4% (n = 538) were male (Table 2). The mean age of the participants was 45.6 (SD 12.6, range 18–85) years; the mean disease duration was 18.8 years (SD 12.3). Out of all the participants, 75.2% (n = 566) were classified as having r-axSpA, whereas 24.8% (n = 187) were classified as having nr-axSpA. Mean scores for the self-administered questionnaires were as follows: BASDAI, mean 4.11 (95% CI 3.91–4.26); BASFI, mean 3.68 (95% CI 3.46–3.84); BASMI, mean 4.0 (95% CI 3.37–4.45); HAQ, mean 0.527 (95% CI 0.49–0.57); ASQoL, mean 6.72 (95% CI 6.31–7.12); and ASDAS, mean 2.4 (95% CI 1.99–2.65). In terms of anthropometric measurements, the mean BMI was 28.2 (SD 6.3) and the mean WHR was 0.941 (SD 0.104).
Characteristics of participants, including comparison by sex and presence of central obesity as measured by the WHR.
In total, 29.6% (n = 223) of the patients were considered obese based on BMI results, whereas 41.2% (n = 310) were considered centrally obese according to the WHR. Comparison analysis based on sex revealed no significant variation in mean BMI (27.9 vs 28.3, P = 0.44) or prevalence of obesity based on BMI (29.8% vs 29.9%, P = 0.21) between women and men (Table 3). Normal weight was the most common classification for women, based on BMI, whereas overweight was the most common classification for men.
Assessment of anthropometric measures by sex.
WHR analysis revealed higher mean WHRs among men compared to women (0.956 vs 0.897, P < 0.01). However, the prevalence of central obesity was significantly higher among women than men (71.6% vs 29.9%, P < 0.01; Figure 1). Obesity was the most common classification among women, affecting 71.6% (149/208) of them, whereas overweight was the most common classification among men, affecting 46.1% (248/538) of them. Further, only 12.0% (25/208) of women were classified as normal weight according to the WHR (Table 3).
Weight classification as determined using the waist-to-hip ratio (WHR) and BMI by sex.
Analysis of average PROs revealed significantly worse ASQoL scores (7.66 vs 6.05, P < 0.01), BASFI scores (4.23 vs 3.30, P < 0.01), HAQ scores (0.63 vs 0.45, P < 0.01), and BASDAI scores (4.37 vs 3.92, P = 0.01) among participants with central obesity compared to those without. As there was a higher proportion of women in the central obesity cohort, further analyses were carried out to determine if sex was a contributing factor in the differences observed.
A multivariate analysis was carried out to determine the effects of sex and central obesity on PROs, including the BASDAI, BASFI, ASQoL, and HAQ. A significant interaction effect was not detected between sex and central obesity (F = 0.5, Wilks λ = 0.997, P = 0.74). However, there was a significant main effect of both sex (F = 3.99, P < 0.01) and central obesity (F = 5.68, P < 0.01) on PROs when assessed individually.
A follow-up univariate 2-way ANOVA found a significant main effect of central obesity on several PROs. Specifically, effects were found on ASQoL scores (F = 7.87, P = 0.01), BASFI scores (F = 19.45, P < 0.01), and HAQ scores (F = 12.25, P < 0.01), with a nonsignificant trend detected for BASDAI scores (F = 3.63, P = 0.07). Follow-up analysis of PROs between centrally obese and normal-weight participants revealed the following average differences: ASQoL, 2.00 (95% CI 3.31–0.70, P < 0.01); BASFI, 1.40 (95% CI 2.02–0.80, P < 0.01); HAQ, 0.24 (95% CI 0.36–0.12, P < 0.01); and BASDAI, 0.65 (95% CI 1.21–0.08, P = 0.02). For ASQoL, HAQ, and BASFI, these differences remained significant, though lesser in magnitude, even when comparing those classified as overweight compared to centrally obese participants.
There was a statistically significant positive correlation between BMI and the WHR among patients with axSpA (rs = 0.456, P < 0.01). When assessed within each sex, the strength of this correlation was found to be stronger among men with axSpA (rs = 0.538, P < 0.01) but weaker among women with axSpA (rs = 0.349, P < 0.01), although both remained statistically significant.
A binomial regression analysis was performed to identify factors associated with central obesity as determined by the WHR. The logistic regression model was statistically significant (χ27 = 96.23, P < 0.01), with an area under the receiver-operating characteristic curve of 0.742 (95% CI 0.699–0.785), indicating an acceptable level of discrimination (Figure 2).
Receiver-operating characteristic (ROC) curve for factors associated with central obesity.
Of the 7 predictor variables assessed (ie, age, female sex, r-axSpA, BASMI, BASDAI, BASFI, and ASQoL), only 2 were statistically significant: age and sex (Table 4). This indicated that both increasing age (odds ratio [OR] 1.03, 95% CI 1.01–1.05) and female sex (OR 6.76, 95% CI 4.12–11.1) were associated with an increased likelihood of a patient with axSpA having central obesity.
Determining the likelihood of central obesity as measured by the WHR.
DISCUSSION
This study detailed the high prevalence and detrimental effects of central obesity in axSpA using data from a large, well-characterized national axSpA registry. Obesity, as classified by BMI, has previously been shown to be one of the most prevalent comorbidities in axSpA, with numerous negative implications regarding disease activity and disease progression. Although central obesity has known negative health consequences in the general population, there is scant literature on central obesity in the context of axSpA. This analysis is novel, not only in capturing the prevalence of central obesity in axSpA but also in demonstrating the significant negative effects of this modifiable risk factor on patient outcomes, including disease activity, functional ability, and QOL.
Prevalence of obesity, as classified by BMI, was similar among both sexes (29.8% vs 29.9%, P = 0.21). Obesity is one of the most prevalent comorbidities in axSpA, with a recent metaanalysis reporting a pooled prevalence of 14%.13 This is notably lower than the prevalence reported in the ASRI, which likely reflects the rising prevalence of obesity in the general population in Ireland.26,27 Although it is difficult to determine the true national prevalence of obesity, most recent data report an Irish obesity prevalence of 27%28 compared to 25.2% in the European Union.29 Obesity prevalence in Ireland is currently similar among men and women, although evidence suggests this will soon change.27 Longitudinal data indicate that the European population with the steepest rise in obesity is Irish women.27 Alarmingly, young women, aged 18–35 years, are driving this rapid increase. This emphasizes the need for early recognition and lifestyle modification from an early age.
The most surprising finding from this current analysis was the significantly higher prevalence of central obesity among women with axSpA compared to men (71.6% vs 29.9%, P < 0.01). Previously, central obesity in the general population was perceived as more common among males.20 However, an epidemiological study examining trends in central obesity in the US noted that prevalence was increasing steadily at an annual rate of 0.56% among males and 0.75% among females and that, by 2030, it would affect up to 55.6% of males and 80% of females.30 This finding is cause for alarm, as central obesity is known to be associated with insulin resistance,31 renal impairment,32 and higher risk of mortality, even among women with a normal BMI.16 For women of childbearing age, central obesity is associated with an increased risk of gestational diabetes, preeclampsia, and the need for cesarean sections.33,34 Unfortunately, there is scant literature examining this issue specifically in axSpA. Research in obesity has demonstrated that excess adipose tissue creates a persistent low-grade inflammatory state driven by proinflammatory adipocytokines.35 This is especially concerning in patients with inflammatory arthritis, such as axSpA, as persistent inflammation could drive worsening of disease activity. As the concentration of adipocytokines is directly related to body fat mass, reduction of excess adipose tissue has the potential to correct this systemic inflammatory reaction. Given the disproportionate percentage of females with central obesity and the trend for females with axSpA to average worse PROs compared to males, this is a key modifiable lifestyle factor to target in this population. Thus, central obesity in women, especially those with axSpA, is a critical issue that requires greater public awareness and recognition.
Identification of central obesity is the first step in improving recognition in axSpA. This analysis highlighted disparities in obesity detection using 2 anthropometric tools (ie, BMI and WHR) among women with axSpA. For men with axSpA, the prevalence of obesity was similar regardless of the tool used (BMI, 29.6% vs WHR, 29.9%). However, for women, there was a stark change in prevalence of obesity when comparing BMI (29.8%) and the WHR (71.6%). This difference in classification between sexes was quantified in a correlation analysis, which found the relationship between BMI and the WHR to be stronger among men than women with axSpA. This suggests that although BMI is a quick tool to screen for obesity, it underestimates obesity in women. This difference in the strength of association between sexes may reflect the fact that BMI was a tool developed and validated for use in males.36 Since its development in the 1830s by a Belgian statistician to describe the average man, BMI has become one of the most commonly used tools to quantify physical size and general health.37 However, studies in postmenopausal women indicate that BMI is not a reliable indicator of obesity status among women.38 In addition, universal BMI cut-offs for obesity have been found to perform poorly among women of differing ethnic backgrounds.39 This clearly demonstrates that it is time to reconsider the use of BMI to screen for obesity in females, especially those with axSpA.
The prevalence of central obesity was similar among those with r-axSpA and nr-axSpA (42.6% vs 37.4%, P = 0.21). However, the prevalence of HLA-B27 was higher in patients with central obesity compared to those with no central obesity (90.4% vs 87.6%, P < 0.01; Table 2). There is limited knowledge of central obesity in axSpA; however, an ankylosing spondylitis study did note that central obesity was associated with radiographic progression over time.40 The ASAS COMOrbidities in SPondyloArthritis (COMOSPA) study also demonstrated that mean waist circumference was significantly lower in patients with axSpA who were positive for HLA-B27 compared to those who were negative for HLA-B27.41 Additional longitudinal data are required to investigate the relationship between radiographic disease, presence of HLA-B27, and central obesity in axSpA.
Assessment of baseline demographics of patients with central obesity and those without revealed that those with central obesity were older (mean 48.6 vs 43.7 years, P < 0.01) and had a later age of onset (mean 28.7 vs 25.7 years, P < 0.01; Table 2). Risk of central obesity in the general population is recognized to increase with age,42 with men and postmenopausal women having up to twice as much visceral adipose tissue (VAT) compared to premenopausal females.43 In women, these changes are mediated by increased androgen secretion and a decline in estrogen concentration associated with reduction in ovarian function during menopause.21 While women from the general population report an increase in myalgia and arthralgia during the onset of menopause,44 there are limited and conflicting data on the effects of changes in estrogen concentration on disease activity among women with axSpA.45
To examine this further, a model was constructed to determine the factors that are associated with the presence of central obesity in axSpA. Despite the numerous predictors assessed, only age and female sex were found to be significantly associated with central obesity in axSpA. The association was especially strong among women with axSpA, who averaged 6.76 times higher odds of central obesity compared to men with axSpA, highlighting the dramatically higher prevalence of central obesity among women with axSpA.
Women with axSpA have previously demonstrated worse PROs compared to men, despite a lower prevalence of radiographic disease and better spinal mobility.45 This disconnect between radiographic burden and outcome scores has perplexed axSpA researchers for some time. Obesity in axSpA is known to contribute to worse patient outcomes regardless of sex.7,46 Given the unexpectedly high prevalence of central obesity detected among women in this study, a follow-up analysis examined the interaction between female sex and central obesity to determine if there was a synergistic effect on PROs. Although the interaction did not reach statistical significance, the effects of both female sex and central obesity remained significant when considered individually. Central obesity was found to be significantly associated with all PROs, with the exception of the BASDAI outcome, which demonstrated a nonsignificant trend toward worse scores in central obesity (P = 0.07). Differences in BASDAI scores between the 2 cohorts were more subtle than for the other PROs, which is likely the reason that the association between central obesity and this outcome did not reach significance. However, if tested on a larger scale, it is suspected that the effect of central obesity on BASDAI scores would be significant. This suggests that central obesity could play a key role in explaining the sex differences observed in PROs in axSpA. Further studies are needed to corroborate these findings in other axSpA cohorts; development of longitudinal data is also needed to determine if there is a causal relationship between central obesity and worse PROs.
The health risks associated with central obesity are attributable to the presence of VAT, which can be directly measured by various imaging modalities, including dual-energy x-ray absorptiometry (DXA) and magnetic resonance imaging.47 Studies using DXA suggest that persistent disease activity in axSpA is associated with increased free fat mass in women,48 with higher body fat content associated with a worse response to anti–tumor necrosis factor-α therapies.49 In clinical practice, it is not practical or economical to repeatedly image patients to screen for central obesity. During an axSpA consultation, physical measurements are routinely recorded for calculation of the BASMI. In this setting, calculation of the WHR would be a simple yet effective modality to effectively assess a patient’s body habitus. Improving recognition of central obesity is important, as this is a potentially modifiable state. Prompt identification creates a window of opportunity for multidisciplinary intervention, including education on lifestyle modification, dietary management with dieticians, and exercise therapy with physiotherapists. Obesity research has proven that early intervention and comprehensive structured programs are the most effective strategies for weight reduction and long-term maintenance.50 Thus, incorporation of the WHR is essential for improving identification of central obesity in axSpA.
Our study has some limitations. Due to the cross-sectional nature of the ASRI, relationships between variables could only be assessed for association and not causation. Plans are in place to collect longitudinal data within the ASRI, which would allow for further analysis of these relationships. Not all patients within the ASRI had physical measurements recorded; as such, they were excluded from the analysis where such measurements were not available.
In conclusion, there is a high prevalence of central obesity in axSpA, as classified by the WHR, most notably among women with axSpA. This modifiable comorbidity was significantly associated with worse functional ability, poorer QOL, and a trend toward worse disease activity. Presence of central obesity was also strongly associated with increasing age and female sex. These results are highly concerning and would support the use of the WHR when screening for obesity in this population. Central obesity is a potentially modifiable state that, if identified, may be a key target for intervention among patients with axSpA. Further public health campaigns are needed to raise awareness and improve recognition of this issue in axSpA.
Footnotes
The Ankylosing Spondylitis Registry of Ireland is supported by unrestricted funding from AbbVie, Pfizer, and UCB.
SM is the recipient of the Gilead inflammation fellowship and has received grants for educational events from AbbVie, Pfizer, Lilly, UCB, and Janssen. The remaining authors declare no conflicts of interest relevant to this article.
- Accepted for publication February 4, 2022.
- Copyright © 2022 by the Journal of Rheumatology