To identify risk factors for functional limitations in patients with ankylosing spondylitis (AS) of at least 20 years' duration.
To identify risk factors for functional limitations in patients with ankylosing spondylitis (AS) of at least 20 years' duration.
Patients with AS for ≥20 years were enrolled in the cross-sectional component of the Prospective Study of Outcomes in AS. All patients had clinical evaluations and completed questionnaires on functional limitations and potential risk factors. Functional limitations were assessed using the Bath Ankylosing Spondylitis Functional Index (BASFI; score range 0–100, higher scores indicate more limitations) and the Health Assessment Questionnaire for the Spondylarthropathies (HAQS). Risk factors included demographic characteristics, duration of AS, smoking status, number of comorbid medical conditions, recalled level of recreational activity in teens and twenties, occupational physical activity throughout life (rated 1 = little, 2 = moderate, 3 = heavy, and weighted by the number of years in each job), and history of AS in a first-degree relative.
The 326 patients (74% men) had a mean ± SD age of 55.0 ± 10.7 years, a mean duration of AS symptoms of 31.7 ± 10.2 years, and a mean BASFI score of 40.7 ± 25.6. BASFI scores increased with higher lifetime occupational physical activity (r = 0.31; P < 0.0001), the number of comorbid conditions (r = 0.25; P < 0.0001), and the duration of AS (r = 0.12; P = 0.04). BASFI scores were higher among current smokers compared with former/nonsmokers (55.5 versus 38.9; P = 0.0002), and among nonwhites compared with whites (49.9 versus 39.3; P = 0.02). In multivariable analyses, lifetime occupational physical activity, current smoking, education level, number of comorbid conditions, and family history were significantly associated with BASFI scores. The same risk factors were associated with the HAQS.
Functional limitations in patients with AS for ≥20 years are greater among those with a history of more physically demanding jobs, more comorbid conditions, and among smokers, and are less severe among those with higher levels of education and a family history of AS.
Functional limitations and resultant disability are major consequences of ankylosing spondylitis (AS). The degree of functional limitation is an important determinant of health-related quality of life, influences the likelihood of work disability, and is the major predictor of medical costs in patients with AS (1–6). Identification of risk factors for functional limitations is important because it would enable better understanding of how these limitations develop, would help identify high-risk groups of patients, and may indicate interventions to prevent functional limitations and improve health outcomes.
Functional limitations in AS increase with age and with the duration of symptoms (1, 7–13). Additional risk factors noted in prior studies included the severity of pain and stiffness, peripheral arthritis, total hip arthroplasty, and smoking (2, 7, 8, 12–17). Results of studies differ on whether functional limitations are more severe among women or men, and on whether the age at onset of AS, familial AS, or unsupervised exercise are associated with functional limitations (6–10, 12–14, 18, 19). Few studies have examined possible associations with education level or comorbid medical conditions, even though these are important determinants of functional limitations in other rheumatic diseases (1, 8, 20–22). In addition, occupational characteristics have been largely overlooked as potential risk factors, despite strong associations between functional limitations and work disability, and between physically demanding jobs and work disability (3–8, 23, 24).
A potential reason for discrepant results among prior studies may be heterogeneity in the patients examined. In particular, risk factors for functional limitation may differ with the duration of AS. Functional limitations in early AS may be closely related to symptom severity and factors that modify symptoms, whereas functional limitations in long-standing AS may be more closely related to factors whose effects cumulate over time and that reflect spinal fusion and long-term structural damage. In this cross-sectional study, we examined demographic and clinical risk factors for functional limitations in a large group of patients with AS symptoms for ≥20 years.
Patients were participants in the cross-sectional component of the Prospective Study of Outcomes in Ankylosing Spondylitis, an observational study whose main aim is to investigate genetic markers of AS severity. Patients were recruited from the clinics of the investigators or local rheumatologists, from patient support and advocacy groups, and from the community by advertisement. A total of 62 participants were drawn from a study of patients with familial AS (25). Enrollment occurred in 2002–2004. Inclusion criteria were a diagnosis of AS by the modified New York criteria (26) and duration of AS for ≥20 years, dated from the onset of persistent musculoskeletal symptoms. All participants had a clinical evaluation by one of the study rheumatologists, had pelvic and spinal radiographs to confirm their diagnosis, and completed questionnaires about their personal and medical history and functional status.
Two measures of functional limitation were used: the Bath Ankylosing Spondylitis Functional Index (BASFI) and the Health Assessment Questionnaire modified for the Spondylarthropathies (HAQS) (27, 28). The BASFI is a 10-item scale that asks respondents to rate the degree of difficulty they have performing tasks, using visual analog scales labeled from 0 (easy) to 100 (impossible). The BASFI score is calculated as the mean of the 10 responses. It has been demonstrated to have good reliability and construct validity (2, 11, 14, 27, 29–31). The HAQS is a 25-item scale that asks respondents to rate the degree of difficulty they have performing tasks in 10 functional areas (dressing, arising, eating, walking, hygiene, reach, grip, errands and chores, bending, and driving). Responses to each question can range from 0 (no difficulty) to 3 (unable to do), and the HAQS score is the average of the highest score in each of the 10 function categories. The HAQS has also been demonstrated to have good reliability and validity (11, 28, 32).
The risk factors considered for association with functional limitations were patient age, sex, ethnicity (white versus other), education level, smoking status (current, former, or nonsmoker), pack-years of smoking, and number of comorbid medical conditions (by patient self report). In addition, we examined associations with the duration of AS, age at onset of AS, history of AS in a first-degree relative (by patient report), and history of iritis (by patient report). We also asked participants to report their level of recreational physical activity in their teens and twenties, relative to their same-sex peers (1 = less than peers, 2 = same as peers, 3 = more than peers). Because previous studies demonstrated associations between functional limitations in AS and occupational physical activity (7), we also computed an occupational physical activity score for each patient. Each patient was asked to report each paid job they had in their lifetime, and to rate the level of physical activity in each job (1 = little, 2 = moderate, 3 = heavy) (5). The occupational physical activity score was then computed as the mean of these ratings, weighted by the number of years spent in each job (possible range 1–3). For example, the occupational physical activity score of a patient who reported working as a carpenter with heavy activity for 10 years, and then as a supervisor with moderate activity for 15 years would be calculated by the equation [(10 × 3) + (15 × 2)]/25 = 2.4. An alternative occupational physical activity score was also calculated, using the activity ratings for each job assigned by the US Department of Labor in the Dictionary of Occupational Titles (1 = sedentary, 2 = light, 3 = medium, 4 = heavy, 5 = very heavy) (33), and also weighted by the number of years in each job (possible range 1–5). The correlation between the 2 occupational physical activity scores was moderately high (r = 0.45; P < 0.0001).
Associations between potential risk factors and the BASFI were tested using t tests (for categorical variables) or Pearson's correlations (for continuous variables). Because values of the HAQS were not normally distributed, nonparametric rank sum tests and Spearman's correlations were used to test associations with this measure. Multivariable linear regression models were developed to examine the independent association of risk factors with the BASFI or HAQS. A square-root transformation of the HAQS was used as the dependent variable in the regression model for this measure, to make the distribution of this variable more normal (Shapiro-Wilk W = 0.985; skewness = −0.04). To determine if risk factors for functional limitations differed between men and women, sex-specific models were also developed. All hypotheses were 2-tailed, and P values less than 0.05 were considered significant. Analyses were performed using SAS version 8.2 programs (SAS Institute, Cary, NC).
The study included 326 patients, with a mean ± SD age of 55.0 ± 10.7 years and a mean duration of AS symptoms of 31.7 ± 10.2 years (Table 1). The patients were mainly white and well educated, and women comprised 26% of the group. The most common comorbid conditions were hypertension (35.6%), peptic ulcer disease (19.3%), depression (16%), osteoporosis (12.6%), and asthma (10.1%). BASFI scores were distributed over a broad range, with 75% of scores falling between 20 and 61 (Figure 1). HAQS scores were generally low, with a median of 0.7, and positively skewed.
|Age, mean ± SD years||55.0 ± 10.7|
|African American||12 (3.7)|
|Asian/Pacific Islander||7 (2.1)|
|Native American||4 (1.2)|
|Professional occupation||212 (68.0)|
|Education level, mean ± SD years||15.9 ± 3.0|
|Lifetime occupational physical activity score, mean ± SD (range 1–3)||1.8 ± 0.7|
|Recreational activity in teens and twenties, mean ± SD (range 1–3)||2.1 ± 0.6|
|Former smoker||146 (44.8)|
|Current smoker||35 (10.7)|
|Number of comorbid conditions|
|Inflammatory bowel disease||10 (3.1)|
|History of iritis||134 (41.1)|
|Duration of AS, mean ± SD years||31.7 ± 10.2|
|Age at onset, mean ± SD years||23.2 ± 7.5|
|History of AS in a first-degree relative||101 (31.0)|
|BASFI score, mean ± SD||40.7 ± 25.6|
|HAQS score, mean ± SD||0.8 ± 0.6|
In univariable analyses, scores for both the BASFI and HAQS increased with older age, longer duration of AS, and the number of comorbid conditions (Table 2). The BASFI and HAQS scores were lower among patients with higher levels of education, but scores were not associated with either the age at onset of AS or with recalled levels of recreational physical activity in young adulthood. However, both the BASFI and HAQS were strongly associated with lifetime occupational physical activity scores. Patients who had more physically demanding jobs throughout their lifetimes had more functional limitations. This association was present when measured both by the patients' subjective ratings of job activity and by the standard ratings of the Dictionary of Occupational Titles.
|Duration of AS||0.12||0.04||0.06||0.23|
|Age at onset||−0.01||0.96||−0.01||0.97|
|Education level||−0.24||< 0.0001||−0.29||< 0.0001|
|Number of comorbid conditions||0.25||< 0.0001||0.21||0.0001|
|Recreational activity in teens and twenties||−0.03||0.61||−0.01||0.99|
|Lifetime occupational physical activity score||0.31||< 0.0001||0.32||< 0.0001|
|Lifetime DOT occupational physical activity score||0.19||0.0007||0.16||0.004|
Scores for the BASFI and HAQS were similar in men and women, in patients with and those without a family history of AS, and in those with and those without a history of iritis (Table 3). However, among men, patients with a family history of AS tended to have lower BASFI scores than those without a family history (30.3 versus 38.2; P = 0.11). Nonwhites had more functional limitations than whites, and current smokers had much higher scores on the BASFI and HAQS than nonsmokers or former smokers. Scores were comparable among nonsmokers and former smokers (mean BASFI 36.9 and 40.8, respectively; median HAQS 0.6 and 0.6, respectively). Among current and former smokers, there was no association between BASFI or HAQS scores and the number of pack-years of smoking (BASFI r = 0.12, P = 0.12; HAQS r = 0.09; P = 0.24).
|Mean ± SD||P||Median (IQR)||P|
|Men||40.5 ± 25.5||0.89||0.6 (0.3–1.1)||0.19|
|Women||41.0 ± 25.8||0.7 (0.4–1.4)|
|White||39.3 ± 24.6||0.02||0.6 (0.3–1.2)||0.01|
|Nonwhite||49.9 ± 30.0||0.9 (0.5–1.5)|
|Family history of AS||38.0 ± 24.6||0.22||0.6 (0.3–1.1)||0.25|
|No family history of AS||41.8 ± 26.0||0.7 (0.1–1.2)|
|History of iritis||38.0 ± 24.2||0.13||0.6 (0.3–1.1)||0.15|
|No history of iritis||42.5 ± 26.4||0.7 (0.3–1.25)|
|Current smoker||55.5 ± 27.4||0.0002||1.1 (0.7–1.4)||< 0.0001|
|Nonsmoker or former smoker||38.9 ± 24.8||0.6 (0.3–1.1)|
Based on the results of the univariable analyses, age, ethnicity, education level, number of comorbid conditions, smoking status, lifetime occupational physical activity score, and family history of AS were considered as potential risk factors in the multivariable models. Sex was also included because it was a potential confounder of the association between functional limitations and age and occupational physical activity. Age and duration of AS were highly correlated (r = 0.70) and could not be included simultaneously in the models because of collinearity. Age was used in the primary analysis rather than duration of AS because age was more highly correlated with education level (r = 0.12, P = 0.03) than was duration of AS (r = 0.07, P = 0.21), and thus provided better adjustment for the association of education level with functional limitation scores.
In the multivariate analysis, BASFI scores were significantly associated with education level, the number of comorbid medical conditions, smoking status, lifetime occupational physical activity, and family history of AS (Table 4). On average, the BASFI score decreased by 1.1 points (out of 100) with each additional year of education, increased by 3.1 points with each additional comorbid condition, was 11.8 points higher among current smokers than among nonsmokers or former smokers, and was 6.7 points lower in patients with a family history of AS. Also, the BASFI score increased by 8.9 points with each additional point in the lifetime occupational physical activity score, meaning that patients whose past jobs all involved moderate physical activity had BASFI scores that were on average 8.9 points higher than those whose past jobs all involved little physical activity. Similarly, patients whose past jobs all involved heavy physical activity had BASFI scores that were on average 8.9 points higher than those whose past jobs all involved moderate activity. When the job activity ratings of the Dictionary of Occupational Titles (33) were used in place of patients' own ratings, BASFI scores were also significantly associated with occupational physical activity, increasing on average by 3.5 points with each grade in the 5-grade scale (P = 0.05). There was no evidence of a threshold effect for occupational physical activity or the number of comorbid medical conditions. There were no associations with age, sex, or ethnicity in the multivariate analysis.
|Education level, years||−1.1||−2.42||0.02||−0.02||−3.29||0.002|
|Number of comorbid conditions||3.1||4.45||< 0.0001||0.04||3.82||0.0002|
|Lifetime occupational physical activity score||8.9||4.42||< 0.0001||0.12||3.93||0.0001|
|Family history of AS||−6.7||−2.38||0.02||−0.08||−1.95||0.06|
Results for the HAQS were similar to those for the BASFI, with significant associations with education level, the number of comorbid conditions, smoking status, family history, and occupational physical activity (Table 4). Results were also similar when duration of AS was used instead of age as the measure of time (Table 5).
|Duration of AS, years||0.2||1.23||0.22||0.001||0.61||0.55|
|Education level, years||−1.1||−2.36||0.02||−0.02||−3.24||0.002|
|Number of comorbid conditions||3.0||4.47||< 0.0001||0.04||4.12||< 0.0001|
|Lifetime occupational physical activity score||9.0||4.46||< 0.0001||0.12||3.97||< 0.0001|
|Family history of AS||−6.9||−2.44||0.02||−0.08||−2.02||0.05|
Because risk factors for functional limitations may differ between men and women, we also performed analyses stratified by sex (Table 6). BASFI scores were associated with smoking status, the number of comorbid conditions, and lifetime occupational physical activity in both men and women. P values for these associations were higher among women, and were likely due to the smaller number of women in the study. The association with education level was also similar in both sexes. However, a family history of AS was more strongly associated with lower BASFI scores in men than in women. White ethnicity was marginally associated with lower BASFI scores only among men. Sex-specific analyses of the HAQS demonstrated similar results (data not shown).
|Education level, years||−0.9||−1.70||0.09||−1.7||−1.65||0.11|
|Number of comorbid conditions||2.9||3.51||0.0005||3.8||2.81||0.007|
|Lifetime occupational physical activity score||9.0||3.79||0.0002||8.6||2.19||0.04|
|Family history of AS||−8.5||−2.48||0.02||−3.6||−0.7||0.50|
To explore further the association of BASFI scores with a family history of AS among men, we categorized patients by the type of relative affected. In adjusted analyses, BASFI scores were lower among men with a child affected with AS (10.7 points lower versus men without a family history, P = 0.20) than among those with a sibling with AS (5.5 points lower versus men without a family history, P = 0.17) or a parent with AS (2.9 points lower versus men without a family history, P = 0.62).
In this study, occupational physical activity, smoking status, comorbidity, level of formal education, and family history of AS were important predictors of the degree of functional limitation in patients with long-standing AS. Because the sample was limited to patients with AS duration of at least 20 years, these factors may be more closely associated with functional limitations due to spinal fusion or permanent structural damage, and less associated with limitations due to inflammatory symptoms. The restricted nature of the sample was also likely to be the reason that the association of age and duration of AS with functional limitations was less pronounced in this study compared with other studies, which included patients with AS of any duration (1, 8–10). In this sample, we did not find the degree of functional limitation to be associated with age at onset of AS, sex, ethnicity, history of iritis, or recreational activity in young adulthood.
Occupational physical activity, assessed over the patient's working life, was strongly associated with functional limitations. In this sample, scores on the BASFI were estimated to be, on average, ∼18 points higher in patients who had always worked in physically demanding jobs, compared with those who worked in sedentary jobs. When computed using the standard job activity ratings of the Dictionary of Occupational Titles, BASFI scores were estimated to be 14 points higher in patients who consistently worked in jobs involving heavy labor, compared with those with sedentary jobs. Although the nature of work differs between men and women, similar associations were found in both sexes. These associations were present in analyses that adjusted for the level of formal education, suggesting that the association between occupational activity and functional limitation was not likely due to confounding by socioeconomic status. However, we cannot be certain that residual confounding by socioeconomic status was not present. Patients with severe functional limitations may have overestimated the physical demands of their past jobs, resulting in information bias, but the results were similar when standard ratings of occupational activity from the Dictionary of Occupational Titles were used. It is important to note that the association here was between a cumulative measure of occupational activity and functional limitations, which by the restricted nature of the sample would also be more likely to represent cumulative effects. In early AS, this association would not be expected, because physically demanding jobs would more likely be held by individuals with few or no functional limitations. Studies of groups of patients that are heterogeneous in the duration or stage of AS may mask such associations.
Working at physically demanding jobs may affect long-term functional ability by stressing joints in the axial skeleton, thereby perpetuating inflammation. Alternatively, persons with physically demanding jobs may be less likely to perform therapeutic or recreational exercises, which then could lead to greater functional limitations in the future (34). Both of these factors may contribute to functional limitations. In previous studies, occupational activity was characterized globally as manual or sedentary, and it was not considered that physical activity can vary greatly within these categories, or that individuals change jobs, often moving into less physically demanding positions with time (7, 14). The more detailed characterization of occupational physical activity in this study may have permitted better detection of associations between work activity and functional limitations. The association of occupational physical activity with functional limitations provides a mechanistic link between work demands, functional limitations, and work disability (3–8, 23, 24, 35).
Current smokers also had much more functional limitation than nonsmokers or former smokers. Smoking status has rarely been examined as a risk factor for functional limitation in AS, but smoking has been associated with more rapid progression of functional disability and with higher levels of functional limitations in prior studies of this risk factor (13, 14, 17). In the study by Doran and colleagues, current and former smokers were examined together, and it was not clear if most or all of the association was due to current smokers (14). In our study, former smokers had levels of functional limitation similar to that of nonsmokers, and only current smokers had increased functional limitation. This finding, along with the absence of an association between pack-years of smoking and functional limitations, suggests that the primary association may not be a direct effect of smoking on functional ability. Rather, smoking may be a marker of poor health behaviors, including perhaps lack of exercise, that contribute to functional limitations.
Two previous studies reported that patients with familial AS have similar degrees of functional impairment as patients with sporadic AS, but a third study reported less severe functional limitations in patients with familial AS (12, 14, 19). Less severe disease in those with familial AS may be a consequence of genetic effects, but may also result from earlier diagnosis and better treatment, more knowledge, or more support regarding AS in affected families. Alternatively, mildly affected individuals in families with AS may be more likely to be diagnosed than mildly affected individuals without a family history of AS, in whom low back pain and stiffness may be attributed to other conditions. Similar selection factors may also influence which patients elect to participate in research studies, and it may not be possible to know if familial AS influences severity without a population-based study. However, our finding that the protective association of family history was more pronounced among fathers of children with AS than among brothers or sons of persons with AS suggests that earlier diagnosis, education, or social support may not be the mechanism underlying this association.
The strengths of this study include the large, well-characterized sample; examination of a number of risk factors; and replication of the findings using 2 different measures of functional limitation. The focus on patients with long-standing AS may have permitted better identification of risk factors associated with irreversible changes and structural damage. However, the study was cross-sectional, and causal inferences cannot be drawn. Despite the cross-sectional design, all risk factors were either invariant (e.g., ethnicity) or historical (e.g., lifetime occupational physical activity, education level, comorbid conditions), which preserved the correct temporal relationship between the risk factors and the outcome. Information on familial AS was collected by patient report, and diagnoses in family members were not verified for all patients. We did not collect information on symptoms or peripheral arthritis, and we cannot determine how much of the variation in functional limitations in these patients may have been associated with differences in AS activity. We also did not examine associations with treatment or past medication use, but few patients had extensive treatment with anti–tumor necrosis factor α medications. The measures of functional limitations we used have good reliability and validity, but the HAQS has previously been demonstrated to have a floor effect (36). This may limit detection of mild functional limitations but would not be expected to limit detection of more severe impairment. Lastly, the patients represented a volunteer sample, and findings may have been different had a community-based or population-based sample been examined.
Our findings indicate that studies of long-term functional outcomes in AS should include assessment of the patient's level of education, comorbid conditions, smoking history, family history, and occupational history. Identification of the specific work activities and the specific behaviors related to smoking that are associated with functional limitations in AS will provide guidance for recommendations that may improve long-term functional outcomes for patients.
The authors would like to thank Cheryl Kallmann, Lori Guthrie, Erin Skrok, Stephanie Morgan, and Laura Diekman for their assistance.