Functional limitations in ankylosing spondylitis (AS) may be due to peripheral joint or axial involvement. To determine if the Bath Ankylosing Spondylitis Functional Index (BASFI), an axial-focused measure, can detect limitations related to peripheral joint involvement equally as well as the Health Assessment Questionnaire modified for the spondyloarthropathies (HAQ-S), a peripheral arthritis–focused measure, and vice versa, we compared associations of each questionnaire with spinal and hip range of motion, peripheral arthritis, and enthesitis in patients with AS.
We examined patients every 4–6 months in this prospective longitudinal study. We used mixed linear models to analyze the associations between 10 physical examination measures and the BASFI and HAQ-S.
We studied 411 patients for a median of 1.5 years (median 3 visits). In multivariate analyses, cervical rotation, chest expansion, lateral thoracolumbar flexion, hip motion, tender joint count, and tender enthesis count were equally strongly associated with the BASFI and HAQ-S. Peripheral joint swelling was more strongly associated with the HAQ-S. Individual items of the BASFI were more likely than items of the HAQ-S to be associated with unrelated physical examination measures (e.g., the association between difficulty rising from a chair and cervical rotation), which may have diminished the axial/peripheral distinction for the BASFI.
The BASFI and HAQ-S had similar associations with impairments in axial measures, while the HAQ-S had stronger associations with the number of swollen peripheral joints. The HAQ-S should be considered for use in studies focused on spondyloarthritis with peripheral joint involvement.
Limitations in physical functioning are a major consequence of many chronic rheumatic diseases. These limitations are most commonly assessed by patient report, which provides a means to learn how patients appraise their ability to do typical daily activities (1). Patient-reported measures may be generic, assessing a broad range of common tasks, or may be tailored to focus on tasks more likely impacted by a particular disease. In ankylosing spondylitis (AS), the Bath Ankylosing Spondylitis Functional Index (BASFI) has become the recommended patient-reported measure of functional limitations, in part because its focus on functions related to the axial skeleton enhances its validity and sensitivity to change in patients with AS (2–5). Supporting its validity are studies that report the BASFI to be more highly correlated with spinal range of motion than other AS-specific measures of physical functioning (6–10).
Peripheral joints are affected in up to 50% of patients with AS (11, 12). It is unclear if the focus of the BASFI on primarily axial skeletal functions, such as bending, standing, and arising, affects how well it captures functional limitations in patients who have peripheral joint involvement. On one hand, there may be sufficient overlap between peripheral functions and axial functions, or such a high correlation between them, that the BASFI provides an accurate overall assessment of functional impairment in these patients (13). Alternatively, functional limitations associated with peripheral joint functions may not be adequately captured by the BASFI, resulting in an incomplete assessment of functional limitations in patients who have both peripheral joint and axial disease (14). Notably, most validation studies of the BASFI have examined its correlation with only measures of spinal impairment, and not with peripheral arthritis or enthesitis (6–10, 15–25). Studies of patients with psoriatic arthritis have used the Health Assessment Questionnaire (HAQ), which emphasizes peripheral joint functions, rather than the BASFI (26, 27).
The HAQ modified for the spondyloarthropathies (HAQ-S) was designed to enhance the validity of the HAQ for the assessment of functional limitations in patients with spondyloarthritis by adding 5 items related to axial functions (28). We considered whether the HAQ-S could be a measure particularly applicable to patients with axial spondyloarthritis and peripheral joint involvement (29). The purpose of this study was to compare the strength of association of the BASFI and HAQ-S with measures of spinal impairment and peripheral arthritis in patients with AS. To determine if the BASFI accurately captured limitations associated with peripheral joint involvement, we compared its associations with tender and swollen joint counts and enthesitis with those of the HAQ-S, both overall and in a subset of patients with peripheral arthritis. Similarly, to determine if the HAQ-S captured limitations in axial functions, we compared its associations with spinal impairments with those of the BASFI. Additionally, to determine the degree to which individual items on the BASFI and HAQ-S captured limitations in functioning irrespective of their axial or peripheral focus, we examined if the associations between questionnaire items and unrelated physical impairments were as strong as their associations with cognate impairments.
Significance & Innovations
This is the first study to examine if axial-focused measures of functional limitation adequately capture limitations due to peripheral joint involvement in patients with ankylosing spondylitis.
This study is the first to compare items of the Bath Ankylosing Spondylitis Functional Index and Health Assessment Questionnaire modified for the spondyloarthropathies with individual cognate and noncognate impairments.
This is the first study to demonstrate associations of the number of tender entheses with limitations in multiple functions.
PATIENTS AND METHODS
We examined patients enrolled in the Prospective Study of Outcomes in Ankylosing Spondylitis, a prospective observational study of clinical and genetic features of AS (30). We recruited patients from our clinics, by referral from local rheumatologists and patient advocacy groups, and by community advertisements. Eligible participants were at least age 18 years and had AS defined by the modified New York criteria (31). We performed pelvic and spinal radiographs to confirm the diagnosis in all participants. The study protocol was approved by the institutional review board at each site, and all participants provided written informed consent. Enrollment began in 2002 and is ongoing.
Initially, the study had 2 arms: a cross-sectional arm for patients whose duration of AS was ≥20 years and a longitudinal arm for patients whose duration of AS was <20 years (dated from the onset of symptoms). From 2002 to 2007, patients in the longitudinal arm were examined every 4 months. In 2007, we opened enrollment in the longitudinal arm to patients with AS of any duration, and changed the visit frequency to once every 6 months. In this analysis, we included all patients enrolled in the longitudinal study arm from 2002 to May 2011.
At each visit, patients completed written questionnaires about their health, including the BASFI and HAQ-S, and had a musculoskeletal examination by a study rheumatologist (MHW, LSG, JCD, MMW). The BASFI is a 10-item questionnaire on the degree of difficulty performing specific tasks, such as standing unsupported, reaching overhead, or looking behind oneself without turning the shoulders. Each item uses a visual analog scale, with anchors of 0 = easy and 100 = impossible (2). The index is calculated as the mean of the 10 responses. The HAQ-S has 25 items grouped in 10 functional areas (8 areas of the original HAQ and 2 areas related to back and neck functions) (28). Items ask about the degree of difficulty performing tasks, where possible responses range from 0 = no difficulty to 3 = unable to do. The total score is the mean of the highest score in each of the 10 areas (possible range 0–3). Both measures have good reliability (3).
The musculoskeletal examination was performed by the same rheumatologist at all visits. On isolated occasions when examinations were performed by a different examiner, these visits were excluded from the analysis. At the start of the study, the investigators established a written protocol for patient examinations and met to standardize measurement techniques. We included 10 measures of impairment in the musculoskeletal examination. We measured both cervical rotation and cervical lateral flexion as the sum of maximal right and left movements using a protractor. We measured occiput-to-wall distance to the nearest centimeter using a tape measure. We measured chest expansion as the difference between maximal inspiration and expiration and the 10-cm modified Schober test using a tape measure. We measured lateral thoracolumbar flexion as the maximum movement of the fingertips down a ruler while leaning to the right and left sides, with the patient standing against a wall. We measured hip range of motion as the sum of internal and external rotation using a goniometer. For lateral thoracolumbar flexion and hip range of motion, we used the poorer of the right and left movements in the analyses.
In addition, we examined 44 joints for the presence of tenderness and swelling (10 hand proximal interphalangeal joints, 10 metacarpophalangeal joints, 2 wrists, 2 elbows, 2 shoulders, 2 acromioclavicular joints, 2 sternoclavicular joints, 2 knees, 2 ankles, and 10 metatarsophalangeal joints) and examined 17 entheses for tenderness (San Francisco method) (32). For the analysis, we used the number of tender entheses.
Our primary interest was the association between measures of impairment on physical examination and functional limitations on the BASFI and HAQ-S. We used random-effects mixed linear models that can incorporate repeated observations to test these associations. In separate sets of models, we used either the BASFI or HAQ-S as the dependent variable. Because the distribution of both measures was positively skewed, we used their square root transformations, which provided normal distributions, in the analyses. We first tested associations with each of the 10 physical examination measures individually in models that adjusted for patient age, duration of AS, sex, ethnicity, education level, and time of observation in the study. Next, we tested all 10 physical examination measures simultaneously to determine the marginal association of each measure, adjusting for the influence of the other measures. In all models, we stratified by the examining physician to account for the nonindependent nature of examinations by the same physician. These models demonstrated the relative associations of spinal measures and peripheral joint measures with the BASFI and HAQ-S.
We then repeated this analysis in the subgroup of patients with peripheral arthritis. We defined patients with peripheral arthritis as those who had physician-recorded swelling of at least 1 peripheral joint at any study visit.
Next, we repeated these analyses using each of the 10 BASFI items and the 10 HAQ-S categories as dependent variables in separate models. This analysis allowed us to determine how the strength of association between given impairments and their cognate physical functions (e.g., the Schober test and picking objects up from the floor) compared to their associations with noncognate functions (e.g., the Schober test and reaching to a high shelf). We used clinical judgment to designate cognate pairs as the physical examination measure that would most impact performance of a particular function, emphasizing specificity rather than sensitivity. In these analyses, the BASFI responses were modeled as continuous variables and the HAQ-S responses were modeled as ordered categorical variables. Analyses were performed using SAS, version 9.2.
The 411 patients were predominantly men and middle aged, with a mean duration of AS of 17.6 years (Table 1). At study entry, the mean ± SD BASFI score was 30.7 ± 24.5 and the mean ± SD HAQ-S score was 0.57 ± 0.55, indicating generally mild functional limitations. Impairments on physical examination varied widely among the patients (Table 1). Although only 12% of patients had peripheral arthritis at the baseline visit, 100 patients (24.3%) had peripheral arthritis at ≥1 study visits. Patients were examined for up to 13 visits (median 3 visits) over 7.6 years (median 1.5 years).
Table 1. Patient characteristics at the baseline visit*
Values are the median (25th, 75th percentiles) unless otherwise indicated. AS = ankylosing spondylitis; BASFI = Bath Ankylosing Spondylitis Functional Index; HAQ-S = Health Assessment Questionnaire modified for the spondyloarthropathies.
Age, mean ± SD years
41.9 ± 14.3
42.0 ± 14.3
Men, no. (%)
White, no. (%)
Education level, mean ± SD years
15.8 ± 2.8
15.4 ± 2.8
Duration of AS, mean ± SD years
17.8 ± 13.5
18.5 ± 14.2
BASFI score (range 0–100)
24 (10.2, 48.1)
39.8 (17.1, 60.0)
HAQ-S score (range 0–3)
0.4 (0.1, 0.9)
0.8 (0.2, 1.2)
Cervical rotation, °
100 (75, 120)
92.5 (70, 115)
Cervical lateral flexion, °
58 (37, 75)
50 (37.3, 74.2)
Occiput-to-wall distance, cm
0 (0, 6.0)
0 (0, 6.0)
Chest expansion, cm
4.0 (2.5, 5.5)
3.5 (2.4, 5.0)
Schober test, cm
3.5 (2.2, 4.5)
3.5 (2.1, 4.5)
Lateral thoracolumbar flexion, cm
12.0 (7.0, 17.0)
10.5 (7.5, 17.0)
Hip rotation, °
60 (50, 71)
60 (46, 68)
Tender enthesis count (range 0–17)
0 (0, 3)
2 (0, 5)
Tender joint count (range 0–44)
0 (0, 1)
2 (0, 4)
Swollen joint count (range 0–44)
0 (0, 0)
0 (0, 1)
Associations with impairments in all patients.
Higher BASFI scores were associated with greater impairments in each of the 10 physical examination measures when examined individually (Table 2). Considering all physical examination measures together in a multivariate model, higher BASFI scores were associated with more limited cervical rotation, more limited chest expansion, less lateral thoracolumbar flexion, more limited hip rotation, and more tender entheses (Table 2). In addition, those with any tender peripheral joints had higher BASFI scores than those with no tender peripheral joints. Those with ≥3 swollen peripheral joints also had higher BASFI scores than those with no swollen peripheral joints, whereas BASFI scores for those with either 1 or 2 swollen joints were not different from those with no swollen joints.
Table 2. Associations between physical examination measures and the BASFI and HAQ-S*
Estimates were based on random-effects mixed linear models that used square root transformations of the Bath Ankylosing Spondylitis Functional Index (BASFI) and the Health Assessment Questionnaire modified for the spondyloarthropathies (HAQ-S) as the dependent variables, and that were adjusted for patient age at study entry, sex, ethnicity, education level, duration of ankylosing spondylitis, time of observation in the study, and the examining physician.
Cervical lateral flexion
Lateral thoracolumbar flexion
Tender enthesis count
1 tender joint (versus 0)
2 tender joints (versus 0)
≥3 tender joints (versus 0)
1 swollen joint (versus 0)
2 swollen joints (versus 0)
≥3 swollen joints (versus 0)
Associations with the HAQ-S were very similar, with greater impairments in cervical rotation, chest expansion, lateral thoracolumbar flexion, and hip rotation, and more tender entheses and tender joints were each associated with higher HAQ-S scores in the multivariate analysis (Table 2). Comparison of the T statistics between models for the HAQ-S and BASFI indicated that the strengths of association for both axial and peripheral joint impairments were similar for the 2 questionnaires. The major difference between the HAQ-S and BASFI was in their association with the swollen joint count. The HAQ-S was significantly higher not only among those with ≥3 swollen joints, but also among those with 1 or 2 swollen joints, whereas having 1 or 2 swollen joints was not associated with limitations as measured by the BASFI.
Associations with impairments in patients with peripheral arthritis.
The subgroup of patients with peripheral arthritis generally had more severe impairments and higher BASFI and HAQ-S scores than the patient group overall (Table 1). In this subgroup, the strengths of association of the BASFI and HAQ-S with cervical rotation and chest expansion were similar, while neither questionnaire was associated with cervical flexion, occiput-to-wall distance, the Schober test, lateral thoracolumbar flexion, or hip rotation in multivariate analyses (Figure 1). Both questionnaires were equally strongly associated with the tender enthesis count. The HAQ-S was more strongly associated with the tender joint count than was the BASFI, and was also higher in those with only 1 swollen joint compared to those with no swollen joints. The strength of association with ≥3 swollen joints was similar for both questionnaires.
Associations with individual questionnaire items.
To determine if responses on the BASFI reflected only axial skeleton impairments or if the BASFI items also captured impairments due to peripheral arthritis, we investigated the association of each BASFI item with each of the 10 physical examination measures and examined the specificity of associations using cognate pairs of BASFI items and their corresponding physical examination measure. The cognate pairs demonstrated expected associations. For example, limitations in cervical rotation and cervical flexion were strongly associated with difficulty in looking behind oneself, occiput-to-wall distance was associated with difficulty reaching a high shelf, and the Schober test and hip rotation were associated with limitations in donning socks (Figure 2A). However, many physical measurements were associated with difficulty in noncognate functions. For example, limitations in cervical rotation were associated with seemingly unrelated tasks, such as rising from a chair, ascending stairs, and donning socks, while hip rotation was associated with looking behind oneself. Lateral thoracolumbar flexion was associated with all 10 BASFI items and chest expansion was associated with 7 BASFI items; neither had direct cognate functions. Tender joint counts and to a lesser extent swollen joint counts were associated not only with limitations in the BASFI items mediated largely by peripheral joints, but also with axial functions, such as looking behind oneself and standing. These findings indicate that peripheral joint tenderness and swelling contribute to differences in how patients rate limitations in functions not principally related to peripheral joints. These associations, present in multivariate analyses that included all physical measures simultaneously, indicate that responses on the BASFI were relatively nonspecific with respect to the function being assessed. Of note, the number of tender entheses had uniformly strong associations with all BASFI items.
In a similar analysis of the HAQ-S (Figure 2B), tender and swollen joint counts were associated with all functions mediated primarily by peripheral joints, with one exception: there was a lack of an association between walking/ascending stairs and the swollen joint count. However, the swollen joint count was not associated with the 2 primarily axial skeleton categories (desk work and driving), and only the tender joint count was associated with desk work. Cognate pairs demonstrated strong associations between cervical rotation and cervical flexion and the 2 axial items of the HAQ-S, but cervical rotation and cervical flexion were not associated with the noncognate functions of rising, reach, or ascending stairs. Hip rotation demonstrated strong cognate associations with dressing, hygiene, rising, and walking, but also with reaching. Lateral thoracolumbar flexion was associated with 7 categories of the HAQ-S, but chest expansion, which has no specific cognate functions, was not associated with any HAQ-S categories. Only 3 noncognate items of the HAQ-S were associated with cervical rotation. These associations indicate that HAQ-S responses were fairly specific, with few noncognate associations. However, the number of tender entheses had strong associations with all HAQ-S categories.
In this prospective longitudinal study, the BASFI and HAQ-S had similarly strong associations with measures of impairment on physical examination in a large cohort of patients with AS. Impairments in cervical rotation, chest expansion, lateral thoracolumbar flexion, and hip range of motion, along with more tender entheses, tender joints, and swollen joints, were associated with higher scores on both the BASFI and HAQ-S. Despite the focus of the BASFI on axial skeletal functions and the focus of the HAQ-S on peripheral joint functions, scores on both measures were influenced to a similar degree by variations in the same subset of physical examination measures that included both axial and peripheral joint measures.
Several factors may explain this apparent contradiction. First, functional limitations in this cohort were more strongly associated with peripheral joint manifestations, including tender entheses and tender and swollen joint counts, than with axial manifestations. Several axial measures, including cervical lateral flexion, occiput-to-wall distance, and the Schober test, were not independently associated with functional limitations by either the BASFI or HAQ-S. Second, several items of the BASFI involved peripheral joint functions, including reaching, ascending stairs, rising from the floor, and standing up from a chair. These 2 factors may diminish the influence of axial-specific items on the overall BASFI score. Third, the items of the HAQ-S that focused on neck and upper back functions were sufficiently represented and sufficiently strongly associated with impairments in neck motion to balance results for the overall HAQ-S so that it had similar associations to the overall BASFI.
Despite the generally similar results of the 2 questionnaires, the HAQ-S was more sensitive than the BASFI to limitations associated with 1 or 2 swollen joints. Scores on the BASFI were significantly higher only among patients with at least 3 swollen peripheral joints. In the subset of patients with peripheral arthritis, the tender and swollen joint counts were more strongly associated with the HAQ-S than with the BASFI, while both measures had similar associations with the number of tender entheses, hip rotation, and the spinal examination measures. These results indicate that the HAQ-S would more accurately detect functional limitations in patients with AS and peripheral joint involvement, with no loss of information regarding functional limitations due to axial manifestations.
The analysis of individual questionnaire items provides additional information on how impairments were related to self-reported functional limitations, and how these associations differed between the BASFI and HAQ-S. The individual BASFI items showed not only associations with cognate impairments, but also commonly with noncognate impairments. For example, difficulty looking behind oneself was associated not only with neck movements, but also with chest expansion, the Schober test, lateral thoracolumbar flexion, hip rotation, tender entheses, and tender and swollen joint counts. Conversely, many impairments were associated with BASFI items that were seemingly related. These results suggest a lack of specificity in the association of the physical examination measures with BASFI items. Poor correspondence between self-reported functional difficulties and measured impairments has been reported for other self-reported measures of physical function (33–36). These findings suggest that these measures capture general appraisals of frailty more so than limitations in specific functions (13). The associations seen with the BASFI are consistent with the pattern expected if many patients answered the items based on a gestalt of their overall level of functioning.
Similar patterns were not seen for the HAQ-S because it had more specific associations between individual function categories and their cognate physical examination measures. This difference between the BASFI and HAQ-S may be due to the differences in the specific functions assessed, but also may be related to the differences in their formats. The 10 consecutive visual analog scales of the BASFI may be susceptible to response bias if patients tend to carry responses from one item to the next, either because it is easier to do this than to think about each item individually, or to present a picture of consistency (37). Carrying responses from one item to the next would lead to nonspecific associations between individual items and the physical impairments but would maintain the association between overall BASFI scores and overall level of physical impairment. The HAQ-S asks respondents to make ratings in discrete categories of difficulty, which may elicit more focused responses and therefore may retain associations with physical impairments better than the visual analog scale format.
The tender enthesis count was strongly associated with worse functioning on both the BASFI and HAQ-S and impacted all questionnaire items uniformly. This lack of specificity may reflect that those with multiple tender entheses had tenderness in many anatomic regions. However, multiple tender entheses may also reflect a widespread pain syndrome, with associated limitations in many functions, rather than being a specific measure of spondyloarthritis activity. Because tender entheses are an important correlate of self-reported functional limitations, it is important to assess tender entheses to understand interpatient differences in functioning, but it is also important to consider that tenderness at these sites may not necessarily indicate inflammation. The association between tender entheses and self-reported functional limitations warrants further investigation.
The strengths of this study include the large sample size, prospective design, and repeated measurement of 10 physical examination measures, which provided a comprehensive assessment of impairments. However, we did not have cognate physical examination measures for each questionnaire item, and therefore could not assess the specificity of associations for each item. Cognate pairs were based on clinical judgment and represent our opinion about the most specific associations. We did not include some measures, such as intermalleolar distance or lumbar extension, that may have provided additional information. Additionally, we did not include measures of peripheral joint damage, such as deformity, limited range of motion, or fusion, but these manifestations are rare in AS (38, 39). We did not study the Bath Ankylosing Spondylitis Metrology Index or other composite measures because we were interested in associations with individual impairments.
In conclusion, our results indicate that despite the axial focus of most items of the BASFI, this measure captured functional limitations due to impairments in both the peripheral joints and axial skeleton. However, the HAQ-S had stronger associations with the swollen joint count, particularly when only 1 or 2 joints were affected, and yet was as strongly associated with axial impairments as the BASFI. In studies focused specifically on patients with AS or axial spondyloarthritis and peripheral joint involvement, the HAQ-S may be preferable, given its ability to detect associations with even a small number of involved joints and yet comparable ability to detect associations with axial impairments.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Ward had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Bethi, Weisman, Learch, Davis, Reveille, Ward.
Acquisition of data. Weisman, Learch, Gensler, Davis, Reveille, Ward.
Analysis and interpretation of data. Bethi, Dasgupta, Weisman, Learch, Gensler, Davis, Reveille, Ward.
Dr. Davis is an employee of Genentech.
We thank Lori Guthrie, Felice Lin, Stephanie Brown, Stephanie Morgan, Vera Wirawan, and Laura Diekman for their assistance.