The prevalence of sacroiliac (SI) joint abnormalities in a primary low back pain population remains unresolved. The aims of our study were to define the prevalence of SI joint disease in this cohort, and to identify clinical features that might accurately predict radiographic changes in the SI joint and spine.
Lumbar spine and anteroposterior pelvis radiographs taken over a 3-year period for the evaluation of back pain at a major chiropractic college were scored for the presence of inflammatory or degenerative features. Data were subsequently extracted by means of a predetermined template from the clinical notes. The outcomes were correlated using Spearman's correlation coefficients.
We identified 315 patients (173 men, 142 women), ages 18–60 years. Of these, 100 patients (31.7%) demonstrated SI joint abnormalities: 75 (23.8%) degenerative, 25 (7.9%) inflammatory. Sex was strongly associated with type of SI joint pathology; degenerative disease was predominantly found in women (68%), whereas inflammatory disease was predominantly found in men (63%). In women there was no correlation between degenerative SI joint abnormalities and degenerative changes in the lumbar spine. Of the clinical descriptors evaluated, none were associated with the radiographic findings with the exception of buttock pain, which was associated with inflammatory sacroiliitis. Neither being overweight nor pregnancy history was associated with degenerative changes in the SI joint.
In a primary back pain cohort, degenerative SI joint disease may be an under-recognized clinical entity. It is strongly influenced by sex but is unrelated to degenerative changes in the lumbar spine. Currently proposed clinical discriminators performed poorly in correlating with radiographic changes in the SI joint.
Low back pain affects a significant proportion of the population and has a major impact on health care costs. It has been estimated that back pain has a point prevalence of 15–30%, an annual prevalence of 20–60%, and a lifetime prevalence of 60–80% (1–3). Yet the contribution of ankylosing spondylitis (AS) and related forms of spondylarthritis (SpA) to chronic back pain has received little study in terms of systematic analysis. One study estimated that 0.3% of individuals with chronic low back pain may have an underlying SpA (4). On the other hand, a survey of 313 patients in a primary care setting completing a screening questionnaire estimated that 5% had SpA as the cause of their chronic low back pain (5).
The modified New York criteria, originally developed for classification purposes, have become the standard benchmark for the diagnosis of AS (6). Because these criteria require presence of sacroiliitis on an anteroposterior (AP) pelvis radiograph, this has become the defining feature for AS. Thus, characteristic radiographic changes in the sacroiliac (SI) joint have become pivotal for case identification in both genetic and therapeutic studies in AS. Yet the SI joint is a notoriously difficult joint to read accurately, and even training exercises make little impact on improving reading the SI joint for the presence of sacroiliitis (7). A significant complication in this area is the fact that definition of the spectrum of radiographic abnormalities in the SI joint has received very little attention in the literature. The overlap between degenerative and inflammatory features is not well defined, nor have the effect of obesity and the role of sex on SI joint pathology been examined.
Clinical symptoms characteristic of inflammatory back pain (IBP) were proposed in 1977 as the hallmark of AS (8). Since then, it has been common practice to classify back pain as inflammatory if 4 of the following 5 features are present: age at onset <40 years, duration of pain >3 months, insidious onset, morning stiffness, and improvement with exercise. When clinical symptoms of IBP are present in a patient with chronic low back pain, the post test probability for axial SpA is estimated at 14% (9). Recent refinement of these clinical features has identified a modified set of criteria for IBP (10). However, how these discriminating features perform in a nonspecific back pain population has yet to be examined.
In Canada, the majority of patients with back pain are first seen by chiropractors, family physicians, or a combination of the two. The most common complaint seen by chiropractors is chronic back pain, with studies estimating that low back pain accounts for >75% of all visits (11). Because only a small percentage of back pain cases are seen by a rheumatologist, it is possible that a significant number of patients with AS are going unrecognized.
The primary objective of this study was to determine the prevalence of sacroiliitis, the radiographic hallmark of AS, in a retrospective cohort of patients with low back pain in a primary care setting. Secondary objectives of the study were to document the spectrum of radiographic abnormalities found within the SI joint in a low back pain cohort, characterize the associated radiographic features in the lumbar spine in the cohort of patients with sacroiliitis, and compare and contrast clinical features associated with sacroiliitis and those associated with noninflammatory back pain.
PATIENTS AND METHODS
This study was conducted at the outpatient clinic of the Canadian Memorial Chiropractic College (CMCC) in Toronto, Ontario, Canada. There is significant emphasis placed on the role of radiography in assessing musculoskeletal complaints in the CMCC curriculum (12). No referral from a physician is required for a patient to be assessed at CMCC, and therefore back pain patients represent a cohort unselected by physicians or by predetermined guidelines.
The study population was a cohort of consecutive patients between the ages of 18 and 60 years who presented to CMCC outpatient clinics from 2003 to 2005, inclusive, and who had low back pain and received AP pelvis and/or lumbar radiographs as part of that assessment. The specific population of interest were those with back pain of ≥3 months' duration (Figure 1). The standard lumbar spine and pelvis radiographs included visualization of the SI joints.
The radiographs were read by 2 rheumatologists (FDO and RDI) and a musculoskeletal radiologist (DCS). There was no prior knowledge of medical history when the reading took place. The final radiographic interpretation was determined by means of a consensus. Each SI joint was assessed for 1) the presence of sacroiliitis; 2) the grade of sacroiliitis, using the standard modified New York grading scale for sacroiliitis; 3) the presence of degenerative features such as osteophytes, dense sclerosis, and joint space narrowing; and 4) osteitis condensans ilii (Figure 2).
The clinical charts of this cohort were subsequently retrieved. Each patient at CMCC is seen according to a standard protocol on their first visit, and a standard set of clinical variables are recorded. Clinical and demographic information was extracted using a standard chart extraction form. The form defined the presence of clinical features that may be associated with IBP. Specific variables recorded included the following: 1) current age, age at onset, and duration of symptoms; 2) description and character of the pain; 3) preexisting diagnosis of AS or SpA; 4) comorbidities (specifically psoriasis and inflammatory bowel disease); 5) medications, specifically nonsteroidal antiinflammatory drugs; 6) family history; 7) pregnancy details; 8) height, weight, and body mass index (BMI); and 9) pain intensity using a visual analog scale (VAS) and the Oswestry Low Back Pain Disability Questionnaire (13).
Following the consensus read, all radiographs were stratified by the presence or absence of abnormalities in the SI joint. The primary analysis consisted of calculating the prevalence (and 95% confidence interval [95% CI]) of sacroiliitis. The age-specific and sex-specific prevalence were also calculated. The spectrum of abnormalities (both inflammatory and degenerative) were documented and stratified by sex and age. Correlations between radiographic findings of the SI joint and the lumbar spine were calculated using Spearman's rho correlation. Descriptive statistics (mean, median, and percentage) were used to describe the clinical features of patients with or without sacroiliitis.
Specifically, the entire cohort was divided into 3 subgroups (normal, degenerative, or inflammatory) depending on the radiographic appearance of the SI joint. Bivariate analysis (Student's t-test and chi-square) was used to compare single clinical features. Multivariable logistic regression analysis with backwards stepping was used to compare patterns of clinical features between sacroiliitis and non-IBP. Significance level was set at P is equal to 0.10.
The appropriate ethical approval was granted from the Institutional Review Boards of CMCC and University Health Network.
Radiographic findings in the SI joint.
In the 3-year period 2003–2005, inclusive, 315 individuals (ages 18–60 years) had radiographs of the pelvis and/or lumbar spine performed for the investigation of back pain. This cohort consisted of 173 men and 142 women. Within this group, 100 patients (31.7%) had a radiographic abnormality of the SI joint (Table 1). Degenerative changes in the SI joint were seen in 23.8% of the cohort. Among individuals with degenerative changes in the SI joint, there were significantly more women (n = 50) than men (n = 23) (P < 0.05). Unilateral SI joint osteophytes were identified in 20 patients (11 men and 9 women) and bilateral SI joint osteophytes in 37 patients (5 men and 32 women). Unilateral osteosclerosis was identified in 10 patients (5 men and 5 women) and bilateral osteosclerosis in 7 patients (3 men and 4 women). Two female patients had osteitis condensans ilii.
Table 1. Radiographic findings in the sacroiliac joint in a cohort of patients ages 18–60 years
Men (n = 173)
Women (n = 142)
Total (n = 315)
Definite inflammatory (at least bilateral grade 2 sacroiliitis, evidence of some erosions)
Possible inflammatory (less than bilateral grade 2 sacroiliitis, and sclerosis without erosions plus no other features of a degenerative process)
Possible degenerative (sclerosis without erosions plus other features of a degenerative process [e.g., osteophytes])
Definite degenerative (predominant feature is unilateral or bilateral osteophytes)
Osteitis condensans ilii
SI joint abnormalities consistent with sacroiliitis were seen in 7.9% of the cohort. In this group there was a trend toward more men affected than women (16 men versus 9 women; P value was not significant). These inflammatory changes were equally divided between early and established findings. Of the 315 SI joint radiographs reviewed, 12 (3.8%) demonstrated at least bilateral grade 2 changes or unilateral grade 3 changes, sufficient to meet the modified New York criteria for the classification of AS. Of the entire cohort, 13 patients (4.1%) had early inflammatory changes.
Radiographic findings in the lumbar spine.
Of the 315 patients with SI joint radiographs, 275 had concurrent views of the lumbar spine. The lumbar spine radiographs were scored for the presence of changes indicative of AS (syndesmophytes) or degenerative disc disease (disc space narrowing or osteophytosis). None of the lumbar spine radiographs showed changes indicative of AS. A total of 128 patients (47%) had evidence of ≥1 level of disc space narrowing, and 114 patients (41%) had evidence of ≥1 site of osteophytes (Table 2).
Table 2. Radiographic abnormalities in the lumbar spine
Total cohort (n = 275)
Disc space narrowing, no. (%)
Presence of osteophytes, no. (%)
The radiographic findings in the SI joints and lumbar spine were correlated using Spearman's correlation coefficients. There was poor correlation between the presence of degenerative changes in the SI joint and the lumbar spine, and no correlation between inflammatory changes in the SI joint and degeneration in the lumbar spine. When the men and women were analyzed separately, there was a modest correlation between disc space narrowing and SI joint degeneration (r = 0.27, P = 0.001) and between osteophyte formation and SI joint degeneration (r = 0.23, P = 0.004) for men. However, in women there was no correlation between degenerative changes in the SI joint and either disc space narrowing or spinal osteophyte formation (Tables 3 and 4).
Table 3. Correlation of sacroiliac (SI) joint findings and lumbar spine findings for men ages 18–60 years
Table 4. Correlation of sacroiliac (SI) joint findings and lumbar spine findings for women ages 18–60 years
Lumbar spine findings
Degenerative SI joint (n = 48)
Inflammatory SI joint (n = 7)
Disc space narrowing
≥1 level (n = 47)
≥2 levels (n = 23)
Presence of osteophytes
≥1 level (n = 41)
≥2 levels (n = 26)
Of the cohort of 315 patients, 274 (156 men and 18 women) had adequate clinical information for analysis. The remaining 41 patients had been referred from a peripheral clinic for radiographs for investigation of back pain, and these clinical records were not accessible. The clinical data were extracted by means of the predetermined chart extraction form. Only 1 patient (0.4%) of 274 had a preexisting diagnosis of any form of SpA. A positive family history for osteoarthritis (OA) was seen in 10%, for SpA in 0.4%, and for psoriasis in 0.7%.
The severity of the back pain was defined as moderate by a 10-point VAS (mean 5.1, median 5.0). The mean Oswestry Low Back Pain Disability score was 36/100, with a median of 32/100 suggesting moderate disability. Review of the height and weight allowed a calculation of the BMI. A total of 51% of the cohort were classified as normal weight (BMI 19–25 kg/m2), 35% were overweight (BMI >25–30 kg/m2), and 13% were obese (BMI >30). One percent were classified as underweight (BMI <19 kg/m2).
On review of their clinical data, 61 patients had back pain of <3 months' duration; this resulted in a cohort of 213 patients (124 men, 89 women) with back pain of ≥3 months' duration, radiographs of the SI joint, and clinical data. The baseline clinical variables of the 3 subgroups as defined by the radiographic findings in the SI joint (normal, degenerative, and inflammatory) are outlined in Table 5. There was no significant difference between the 3 groups with respect to current age, age at onset of pain, intensity of pain, or BMI. As mentioned above, men were less likely to have degenerative changes in the SI joint than women (odds ratio [OR] 0.47, 95% CI 0.26–0.88). The clinical features, such as the character, location, and radiation of the pain, were compared between the 3 groups. The category of BMI and pregnancy data for the female patients were also analyzed under these 3 subgroups (Table 6).
Table 5. Distribution of clinical features among the chronic back pain cohort depending on radiographic finding in the SI joint*
Normal (n = 143)
Degenerative (n = 53)
Inflammatory (n = 17)
SI = sacroiliac; VAS = visual analog scale; BMI = body mass index.
Compared with normal, odds ratio (OR) 0.47, 95% confidence interval (95% CI) 0.25–0.88 (significant).
Compared with normal, OR 1.46, 95% CI 0.49–4.40 (no significant difference).
There was no significant difference in the characteristics of the pain, such as nocturnal pain, improvement with activity, improvement with rest, or insidious onset, between the radiographic subgroups. Location of the pain (SI, sacrum, buttock, lumbar, or thoracic spine) did not differ between the 3 groups. IBP cases were more likely to radiate to the buttock when compared with those patients with normal SI joints (OR 3.18, 95% CI 1.13–8.94).
The distribution of patients classified as overweight or obese (BMI >25 kg/m2) did not differ between the 3 cohorts, with 45% of the normal SI joint cohort, 58% of the degenerative SI joint, and 33% of the inflammatory SI joint groups having a BMI >25 kg/m2. Although there was a trend for a higher percentage of overweight patients with degenerative findings in the SI joint when compared with patients with normal weight, this did not reach statistical significance (OR 1.73, 95% CI 0.80–3.75). The proportion of female patients with a history of previous pregnancy appeared to be similar among the 3 groups: 48% in the normal SI joint group, 52% in the degenerative SI joint group, and 50% in the inflammatory SI joint group.
In this large primary back pain cohort, radiographic sacroiliitis consistent with AS was found in 3.8%. Radiographic changes consistent with early sacroiliitis were seen in another 4.1%. A significantly large proportion of the cohort (23.8%) had degenerative changes in the SI joint. Degenerative change in the SI joint has received little attention in prior investigations and is clinically under-recognized. The entity appears strongly influenced by sex, with a strong bias toward women. However, it appears unrelated to concurrent OA in the lumbar spine. Overall, 47% of this cohort had evidence of some degenerative features in the lumbar spine. There was only modest correlation between degenerative findings in the SI joint and lumbar spine for men, but no such correlation for women. Although there was a trend for patients with degenerative SI joints to be overweight, this did not reach statistical significance. Furthermore, a history of previous pregnancy was not seen more frequently in the female patients with degenerative SI joints. The current proposed discriminators of IBP performed poorly in correlating with radiographic changes in the SI joint; a history of nocturnal pain, improvement with activity, change with rest, or insidious onset did not differ between the patients with back pain defined by radiographic findings in the SI joint.
The modified New York classification criteria for AS rely on the combination of clinical symptoms of back pain or physical findings of restricted movement in the spine, plus radiographic sacroiliitis of at least grade 2 bilaterally or grade 3 unilaterally (14). Our study demonstrates a broader range of radiographic change in the SI joint, implicating a broader spectrum of pathology found in the SI joint. There is a significant prevalence of these changes in a primary back pain cohort, but only a minority of the radiographs have features consistent with an inflammatory process. Recent studies both on the genetic basis of AS and the therapeutic efficacy of biologic agents have used the modified New York criteria as the inclusion criterion. Both types of investigation make the assumption that AS is a homogeneous clinical entity. Sclerosis in the SI joint joint, particularly in the setting of back pain that is often chronic and shares some clinical features with IBP, might be a confounder that contaminates the homogeneity of the cohorts being analyzed for genetic susceptibility or responsiveness to treatment using tumor necrosis factor inhibitor therapy.
There is an unacceptably long delay between the onset of symptoms and the time of diagnosis for AS: an average interval of 7 years has been reported (15, 16). There are a number of factors contributing to this long delay. First, the established classification criteria for AS rely on the combination of clinical symptoms plus unequivocal radiographic sacroiliitis. Second, there is no pathognomonic clinical feature or laboratory test to make the diagnosis of AS.
This is the diagnostic challenge that commonly faces clinicians dealing with patients with back pain (17). More recently, magnetic resonance imaging (MRI) has added an important diagnostic dimension in this context (18, 19). However, the sensitivity and specificity of MRI-defined lesions have not been defined in an unselected back pain cohort, and in many regions, access to MRI is limited by cost or availability. Therefore, clinicians are forced to continue to rely on specific features of the history to screen for underlying inflammation of the SI joint, and on plain radiography as the confirmatory test.
It is a significant challenge to accurately identify the relatively small number of patients (3.8% in our cohort) with chronic low back pain who have AS. In this regard, AS presents a distinct diagnostic problem because it occurs in the context of a highly prevalent condition (low back pain) in which it represents a small subset. Studies of the prevalence of AS in the population have generally used the modified New York criteria, which include both clinical features and the radiographic changes in the SI joint. Yet the sensitivity and specificity of the SI joint radiographic changes for AS have not been defined in unselected back pain populations. The higher prevalence of degenerative joint disease in the SI joint is an important factor for interpreting prevalence studies, and for case definition in genetic studies or therapeutic trials.
Clinical features that might serve as surrogate markers for IBP are a high priority in improving early detection of AS. Our findings indicate that in a back pain cohort characteristic of that seen by primary care clinicians, the currently proposed clinical discriminators performed poorly in correlating with radiographic changes in the SI joint.
The high prevalence of back pain in the population has been acknowledged by a large number of investigators (20–22). Under-recognition of AS within this large patient population has contributed materially to the long delay in diagnosis of AS. This has become a higher priority with the advent of effective new biologic treatment for AS. Yet recent guidelines on the diagnosis and management of back pain have given little attention to IBP (4, 23).
One of the limitations of generalizing from prior studies of IBP is that there is usually a preselection process by virtue of referral patterns from primary care physicians to specialists. It has proved a challenge to develop an accurate portrait of the contribution of AS to a primary, unselected back pain population. Our retrospective study allowed us to address this issue. In our cohort, the indication for radiographs was based solely on clinical grounds using established guidelines for the investigation of low back pain. In this sense, it is a portrait of back pain in a real-world setting, without bias introduced by referral patterns and practices.
Another potential confounder in this type of research relates to variability in radiographic technique between centers. This pitfall was minimized by utilizing a single radiology unit with preset specifications for radiologic studies of the lumbosacral spine and SI joints. Variability in clinical record keeping has also been problematic in primary back cohort studies. In our study, the patients were evaluated using a standard case report form and all clinical records were retrievable for the study.
Erosive changes in the SI joint are interpreted as reflecting current or prior inflammation. This is clearly a surrogate marker, and sacroiliitis would ideally be confirmed by either biopsy or at least MRI. Yet neither is feasible in a large population study, and the New York criteria, which serve to anchor the diagnosis of AS, use the radiographic changes in the SI joint as reflecting inflammatory disease. Future studies might include MRI analysis in a large primary back pain cohort, but this was beyond the scope of the present study, and for the time being, it is plain radiography, not MRI, that is the gold standard for the diagnosis of AS.
Degenerative change of the SI joint is an under-recognized clinical entity that contributed substantially to the radiographic findings of back pain in an outpatient setting. Whether this entity was the primary cause of the back pain itself cannot be resolved in the present study and might be addressed in the future with local treatments directed at this site, SI joint injection with a steroid having demonstrated a reasonable success rate in previous studies (24, 25). Degeneration of the SI joint appears to be strongly influenced by sex but is unrelated to degenerative OA in the lumbar spine. The pathogenesis of this localized site of degeneration in the SI joint is unresolved and warrants further investigation.
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 submitted for publication. Dr. Inman 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. O'Shea, Peterson, Inman.
Acquisition of data. O'Shea, Salonen, Ammendolia, Hsu, Inman.
Analysis and interpretation of data. O'Shea, Boyle, Salonen, Ammendolia, Inman.
We thank Brendan Carney Killian for his assistance in data extraction and data entry.