Performances of the Assessment of SpondyloArthritis International Society Axial Spondyloarthritis Criteria for Diagnostic and Classification Purposes in Patients Visiting a Rheumatologist Because of Chronic Back Pain: Results From a Multicenter, Cross-Sectional Study
Universitat Autònoma de Barcelona, Barcelona, Spain
Hôpital Cochin, René Descartes University, Paris, France
Rheumatology B Service, Hôpital Cochin, 27 rue du Faubourg Saint-Jacques, 75014, Paris, France. E-mail: email@example.com
To evaluate the performances (sensitivity, specificity, and positive and negative predictive values) at diagnosis and study visit of the Assessment of SpondyloArthritis international Society (ASAS) criteria in axial spondyloarthritis in patients with chronic back pain (CBP). A secondary objective was to identify the most contributory item to diagnosis/classify spondyloarthritis.
We conducted a multicenter, cross-sectional study. Patients were ages <45 years with a history of CBP and seeing a rheumatologist in France. Data included items from the different sets of criteria, checking if present at diagnosis (“diagnosis”)/after diagnosis, but at study visit (“classification”), and the rheumatologist's diagnosis at study visit. Statistical analysis included descriptive characteristics and performances for diagnosis and classification. The diagnosis of the rheumatologist was considered the “gold standard.”
A total of 1,210 patients were eligible for our analysis. Sensitivity and specificity for ASAS axial criteria were 0.76 and 0.94, respectively, and 0.87 and 0.92 for diagnostic and classification purposes, respectively. The positive likelihood ratio of the ASAS axial criteria was 13.6 and 10.30 for diagnostic and classification purposes, respectively. The most contributory items to diagnosis and classification were radiographic sacroiliitis, followed by magnetic resonance imaging sacroiliitis for diagnosis and history of uveitis for classification.
We confirm the validity of the ASAS criteria for both diagnostic and classification purposes in a clinical setting of patients with CBP.
Spondyloarthritis (SpA) is a frequent disease in patients ages <45 years presenting with chronic (>3 months) back pain (CBP), but its real prevalence remains unknown (). Classically, axial involvement in SpA was diagnosed upon the modified New York (mNY) criteria () by assessing radiologic sacroiliitis, resulting in a diagnostic delay in the subgroup of patients without structural damage of the sacroiliac joints. In order to prevent this diagnostic delay and also to encompass the different clinical presentations of SpA (e.g., peripheral arthritis), other sets of classification criteria combining both clinical, biologic, and radiologic features, such as the Amor criteria () and the European Spondyloarthropathy Study Group (ESSG) criteria (), were defined and have been validated.
However, the performances of these sets of criteria in early diagnosis are rather poor (). In 2004, an international group of experts, the Assessment of SpondyloArthritis international Society (ASAS), decided to improve the classification criteria of SpA in order to permit an earlier diagnosis, including for the first time the investigation of the sacroiliac joint by magnetic resonance imaging (MRI). This approach led to the publication in 2009 of the ASAS classification criteria of SpA, especially for axial forms, and the proposition of modification of the Amor and ESSG criteria, by including the potential abnormalities in MRI ([6, 7]).
However, the performances of the ASAS criteria (sensitivity, specificity, and positive and negative predictive values) have been only scarcely tested in populations () other than the original validation population (). Moreover, only a systematic study of all consecutive patients presenting with present or past history of CBP occurring before age 45 years might allow an estimation of the performances of the axial ASAS criteria in daily practice.
The primary objective of our study was to evaluate the psychometric properties (sensitivity, specificity, and positive and negative predictive values) at diagnosis (diagnostic purpose) and at time of the study visit (classification purpose) of the ASAS criteria for axial SpA in patients visiting their rheumatologist for any reason, but with a history of CBP. Secondary objectives were 1) to identify the most contributory item of these criteria to the diagnosis of SpA, 2) to evaluate and compare the performances of the ASAS to that of the mNY, Amor, ESSG, modified Amor (mAmor), and modified ESSG (mESSG) criteria, 3) to evaluate the performances of the ASAS criteria in all SpA subgroups (ankylosing spondylitis, psoriatic arthritis, reactive arthritis, inflammatory bowel disease–related arthritis, and undifferentiated SpA), and 4) to describe the patients with regard to the ASAS criteria arm they are fulfilling, i.e., the “imaging arm” (either radiographic or MRI sacroiliitis plus 1 SpA feature) and the “clinical” arm (HLA–B27 positive and 2 SpA features).
Box 1. Significance & Innovations
This study confirms the high frequency of spondyloarthritis (SpA) in patients visiting a rheumatologist because of chronic back pain, occurring before the age of 45 years, and also the relevant specificity and sensitivity of Assessment of SpondyloArthritis international Society (ASAS) criteria for classification and diagnosis of axial SpA. Moreover, the positive likelihood ratio (LR) of the ASAS criteria was >10 for both diagnostic and classification purposes.
This study provides relevant information regarding some items of the ASAS criteria that presented with a positive LR >10, suggesting they might be highly contributory in diagnosis or classification by themselves in this clinical setting. Some of them were expected, such as magnetic resonance imaging and radiographic sacroiliitis, or HLA-B27 positive, but others were not, such as raised C-reactive protein (CRP) level for diagnosis, and raised CRP level and anterior uveitis for classification.
This study confirms the rationale of having included the raised CRP level in the ASAS criteria, and the clinical impression that uveitis facilitates the classification of the disease.
PATIENTS AND METHODS
This was a multicenter and cross-sectional observational study. From a national comprehensive file of all 1,834 rheumatologists working in office-based and both office- and hospital-based practices in France (), 1,000 rheumatologists were randomly selected (cluster-sampling method) and invited to participate. A total of 384 rheumatologists signed the participation agreement. The study was conducted in agreement with good clinical practices, including an ethical committee agreement and a written informed consent from all the patients.
Patients had to be ages >18 years, with past or current history of CBP (≥3 months) occurring before age 45 years but starting after 1995, visiting a rheumatologist in France, and not participating in a clinical trial. Patients with onset of CBP before 1995 were excluded, as MRI was only widely used in the diagnosis of CBP in rheumatology daily practice in France after that date. Each rheumatologist had to include 4 consecutive patients responding to the inclusion criteria, but regardless of the reason for the visit and the definite diagnosis of CBP.
Questionnaires were completed by the rheumatologist from the available data at the time of visit, between January 2010 and May 2010 in 3 sections that included: 1) checking the inclusion and exclusion criteria, sociodemographic data (age, sex, height, and weight), and date and reason for visit; 2) defining the date of onset of CBP and whether etiologic diagnosis of CBP was definitively established before the visit, and if yes, to choose one of the following: vertebral fracture, spondylodiscitis, neoplastic disease, mechanical back disorders (e.g., discopathy or facet joint disease), SpA, or other diagnosis; the rheumatologist was not asked to answer whether or not the patient fulfilled any set of criteria; and 3) checking, in a random order, for the presence of the items of the ASAS, ESSG, Amor, and mNY criteria, and if present, whether the items were present before or at the time of diagnosis or only after the diagnosis was made. The informational items collected were: family history (SpA, psoriasis, reactive arthritis, inflammatory bowel disease, and uveitis), characteristics of CBP (description of CBP [age at first episode, continuous or chronic back pain], inflammatory back pain [yes/no; and checking for the characteristics (yes/no): insidious onset, morning stiffness, improved with rest, and awakening at night because of pain with improvement upon getting up], good response to nonsteroidal antiinflammatory drugs [NSAIDs] within 24–48 hours), other axial manifestations (cervical and thoracic manifestations, buttock pain [any or bilateral, alternating buttock pain]), peripheral clinical characteristics (arthritis, dactylitis, and enthesitis), description of extraarticular involvement (uveitis, psoriasis, nongonococcal urethritis/cervicitis 1 month prior to arthritis onset, diarrhea 1 month prior to arthritis onset, and inflammatory bowel disease), HLA–B27 status (positive/negative), C-reactive protein (CRP) level status (raised: yes/no), and imaging (plain pelvic radiograph [sacroiliitis: yes/no] and sacroiliac MRI [sacroiliitis (e.g., bone edema/osteitis suggesting an SpA-related sacroiliitis): yes/no]).
Description of population at baseline
All patients without a definite diagnosis were excluded. Population was divided into SpA and non-SpA groups according to the definite diagnosis of the rheumatologist at the time of the visit. For all measures, diagnosis of the rheumatologist was considered the “gold standard.” The non-SpA group was considered the control group. A descriptive analysis of the population was performed at baseline, comparing the demographic data in both groups by chi-square test and nonparametric Wilcoxon's test as appropriate. Prevalence of SpA in this population was calculated.
Diagnostic versus classification definition
In case of a positive answer for any of the items (e.g., psoriasis), the rheumatologist was asked whether this information was available at the time of diagnosis (e.g., the psoriasis had appeared 5 years before the diagnosis of CBP) or only after diagnosis but before the study visit (e.g., psoriasis appearing 5 years after the diagnosis of CBP). As explained above, we did not ask the rheumatologists to confirm whether the patient was fulfilling any of the SpA sets of criteria, but by collecting the presence/absence of all the items composing the criteria, we were able to calculate the percentage of patients fulfilling them. If one set of criteria (e.g., Amor) was fulfilled because of the presence of the items at diagnosis (e.g., presence at diagnosis of inflammatory back pain, defined according to the criteria by Calin et al , alternate buttock pain, family history of SpA, and good NSAID response), then this set of criteria was considered as diagnostic criteria for diagnostic purposes. When the fulfillment of one set of criteria (e.g., Amor) was possible because of the presence of the items both at diagnosis or after diagnosis, but prior to study visit (e.g., presence at diagnosis of inflammatory back pain, alternate buttock pain, and good NSAID response, with asymmetrical oligoarthritis appearing only 4 years after diagnosis, but 2 years before study visit), the set of criteria was considered as classification criteria, as the diagnosis was already made and the fulfillment of this set of criteria would only help to classify the patient.
Evaluation of the performances of the items of the ASAS criteria
For this purpose, frequencies of every item were calculated both at the time the diagnosis was made and at the time of the study visit in SpA and non-SpA groups, and compared by chi-square test as appropriate.
We also calculated the positive likelihood ratio (LR; positive LR = sensitivity/1 − specificity, or the probability of a person who has the disease testing positive divided by the probability of a person who does not have the disease testing positive), the negative LR (negative LR = 1 − sensitivity/specificity, or the probability of a person who has the disease testing negative divided by the probability of a person who does not have the disease testing negative), and their 95% confidence interval (95% CI) for every item using the diagnosis of the rheumatologist as the “gold standard.” Positive LR and negative LR capture both sensitivity and specificity of a given test parameter in a single figure and are indicators of the diagnostic value of the respective tests; the higher the positive LR and the lower the negative LR, the better the diagnostic value of the test. We also calculated the positive posttest probability (posttest probability = pretest probability × positive LR), which is the subjective probability of the presence of a condition after a diagnostic test, pretest probability being the prevalence of the disease, in this case SpA ([12, 13]).
Evaluation of performances of ASAS criteria
Psychometric performances for the axial ASAS criteria were calculated using diagnosis of the rheumatologist as the “gold standard” as follows: sensitivity = number of true positive diagnoses/(number of true positives + number of false negatives); specificity = number of true negatives/(number of true negatives + number of false positives); positive predictive value = number of true positive/(number of true positives + number of false positives); negative predictive value = number of true negatives/(number of true negatives + number of false negatives); and positive LR, negative LR, and the posttest probability as defined above. All performances were calculated for diagnostic and classification purposes.
Comparison of the performances of all criteria sets
For this purpose we calculated the performances (as defined above) for each set of criteria: mNY, Amor, mAmor, ESSG, and mESSG. For the mAmor and mESSG criteria, sacroiliitis was defined either by radiographs or MRI, both for diagnosis and classification.
Comparison of the performances of all criteria sets in the different SpA subgroups
For this purpose, prevalence of every type of SpA subgroup was calculated. SpA subgroups were ankylosing spondylitis, psoriatic arthritis, reactive arthritis, inflammatory bowel disease–related arthritis, and undifferentiated SpA. The performances for each subgroup were calculated both for diagnosis and classification.
Description of the patients with regard to the ASAS criteria arm they are fulfilling
For this purpose, we considered as a first step 2 different arms in the ASAS criteria, i.e., a “clinical” arm of the ASAS criteria (HLA–B27 plus 2 other clinical SpA features) and an “imaging” arm of the ASAS criteria (radiograph or MRI sacroiliitis plus 1 other SpA clinical feature). As a second step, we considered 3 subgroups within the imaging arm as follows: 1) MRI positive and radiograph positive arm (MRI sacroiliitis and mNY sacroiliitis plus 1 other SpA clinical feature), 2) MRI positive and radiograph negative arm (MRI sacroiliitis [without mNY sacroiliitis] plus 1 other SpA clinical feature), and 3) radiograph positive and MRI negative arm (mNY sacroiliitis [without MRI sacroiliitis] plus 1 other SpA clinical feature). We described the demographics of each of the ASAS subgroups, comparing the data by chi-square test and nonparametric Wilcoxon's test as appropriate for both diagnostic and classification purposes. Statistical analyses were performed with SAS software, version 9. P values less than 0.05 were considered significant.
Among the 1,379 patients recruited, 1,364 fulfilled the inclusion criteria, but only 1,210 patients had a definite diagnosis (SpA or non-SpA) at the time of the study visit and were therefore included in the subsequent analyses. Prevalence of SpA was 35.1% (n = 425 patients), and among the non-SpA patients (n = 785 patients) diagnoses were mechanical back disorders in 760 (62.8%) patients, vertebral fractures in 13 (1.1%) patients, infectious spondylodiscitis in 1 (0.1%) patient, neoplastic disease in 1 (0.1%) patient, and other diagnoses for 10 (0.8%) patients. Diagnostic delay (i.e., time from back pain onset to diagnosis) was longer among SpA patients (median delay 1.08 years [interquartile range (IQR) −0.16, 3.90 years] versus 0.28 years [IQR 0.0, 2.51 years]; P < 0.001). SpA patients were younger (median age 38 years [IQR 32, 45 years] versus 44 years [IQR 36, 51 years]; P < 0.001) and slimmer (median body mass index [BMI] 23.7 [IQR 21.7, 26.3 kg/m2] versus 24.2 [IQR 22, 27 kg/m2]; P < 0.015) with male predominance in both groups (56% men in SpA group versus 52.2% in non-SpA group, nonsignificant).
The most frequent missing data were HLA–B27 typing and imaging of the sacroiliac joints. HLA–B27 typing, plain pelvic radiography, and MRI of the sacroiliac joints were available in 393 (92.5%) versus 219 (28.9%) patients, 327 (76.9%) versus 381 (48.5%) patients, and 131 (30.8%) versus 62 patients (7.9%) in the SpA versus non-SpA groups, respectively. All missing data were considered as negative (or normal) for the analysis (e.g., in case MRI of the sacroiliac joints was not available, MRI was considered as negative in the analysis) in order to reflect the application of these criteria in real life, i.e., if an item (e.g., HLA–B27) would be missing, the rheumatologist would consider it negative and still try to apply the criteria.
Performances of ASAS criteria items for diagnostic and classification purposes
Table 1 represents the frequencies for every item of the composite ASAS criteria, both for diagnostic and classification purposes. All items were significantly more frequent in the SpA group, except for balanitis and family history of reactive arthritis.
Table 1. Comparison of the frequency of the different features in the SpA vs. the control group and the corresponding positive and negative LRs*
SpA, no. (%)
Controls, no. (%)
Positive LR (95% CI)
Negative LR (95% CI)
SpA, no. (%)
Controls, no. (%)
Positive LR (95% CI)
Negative LR (95% CI)
SpA = spondyloarthritis; LRs = likelihood ratios; 95% CI = 95% confidence interval; IBP = inflammatory back pain; NSAID = nonsteroidal antiinflammatory drug; IBD = inflammatory bowel disease; AS = ankylosing spondylitis; ReA = reactive arthritis; CRP = C-reactive protein; MRI = magnetic resonance imaging.
Table 1 presents also the positive LR and negative LR of every item of the ASAS criteria, both for diagnostic and classification purposes; Figure 1 also presents the positive LR of every set of criteria. Radiographic sacroiliitis, followed by MRI sacroiliitis, HLA–B27 positivity, and raised CRP level had the higher positive LR and posttest probabilities (data not shown) for diagnosis, with lower negative LR. Interestingly enough, radiographic sacroiliitis had the higher positive LR and posttest probabilities (data not shown) for classification, but was followed by a history of anterior uveitis, MRI sacroiliitis, and raised CRP level.
Positive LRs of all items, excluding missing data, were also calculated with, as expected, lower values (HLA–B27 positive LR 3.39 [95% CI 2.42–4.76] and 3.36 [95% CI 2.60–4.34] and MRI sacroiliitis positive LR 5.50 [95% CI 2.74–9.32] and 3.80 [2.51–5.75] for diagnosis and classification, respectively).
Performances of ASAS criteria for diagnostic and classification purposes
At diagnosis, 324 (76.2%) patients among the SpA group fulfilled the ASAS criteria versus 368 (88.6%) patients at study visit. The positive LR of ASAS criteria both for diagnosis and classification are summarized in Figure 1. The performances of the ASAS criteria for diagnosis and classification are presented in Figure 2, with high specificity and acceptable sensitivity for both diagnosis and classification.
Performances of all criteria sets for diagnostic and classification purposes
Three hundred sixty-five (83.8%) patients, 345 (81.2%), 342 (80.5%), 327 (76.9%), and 190 (44.7%) patients fulfilled the mAmor, mESSG, Amor, ESSG, and mNY criteria at diagnosis, respectively. Three hundred ninety-five (92.9%) patients, 395 (92.9%), 388 (91.3%), 381 (90.0%), and 229 (53.9%) patients fulfilled the mAmor, mESSG, Amor, ESSG and mNY criteria for classification, respectively.
The performances of the different sets of criteria are presented in Figure 2. The mAmor criteria showed the highest sensitivity for both diagnostic and classification purposes, followed by mESSG, ESSG, and ASAS criteria sets. The mNY criteria presented the highest specificity, both for diagnosis and classification, followed by the ASAS criteria.
Figure 1 also summarizes the positive LR for each set of criteria; the mNY criteria had the highest positive LR, followed by ASAS criteria both for diagnostic and classification purposes, although ASAS criteria had lower negative LRs compared to the mNY criteria. Interestingly, only ASAS and mNY criteria sets showed positive LRs >10 for both purposes.
Performances of all sets of criteria in the different SpA subgroups
Performances of all criteria sets in all SpA subgroups are described in Figure 3. The ASAS criteria had the highest sensitivity in ankylosing spondylitis both for diagnosis and classification, but a rather low sensitivity for the other forms of SpA. As expected, the mNY criteria were the most specific for ankylosing spondylitis diagnosis and classification, but also in all other subgroups of SpA, followed by the ASAS criteria also in all subgroups.
The ESSG and mESSG criteria had the highest sensitivity for psoriatic arthritis both for diagnosis and classification, and the Amor and mAmor criteria were the most sensitive for both diagnosis and classification of reactive arthritis and inflammatory bowel disease–related arthritis.
Description of the patients with regard to the ASAS criteria arm they are fulfilling
For diagnosis, 107 (29.1% of the patients fulfilling ASAS criteria for diagnostic purposes) and 261 (70–79%) patients fulfilled the clinical and imaging arm of the ASAS criteria, respectively. Within the imaging arm of the ASAS criteria, 49 patients (13.3%), 56 patients (15.2%), and 156 patients (42.4%) fulfilled the MRI and radiographic positive arm, the MRI positive and radiographic negative, and the MRI negative and radiographic positive arms, respectively. Patients fulfilling the clinical arm of the ASAS criteria at diagnosis were older (median age 41 years [range 36–51 years]) compared to the imaging arm (median age 38.0 years [range 32–44 years]) (P < 0.0001), but younger than controls (median age 43 years [range 36–51 years]). No differences were assessed for sex or BMI. Within the imaging arm, patients from the MRI and radiographic positive group, MRI positive and radiographic negative, and radiographic positive and MRI negative had a median age of 37 years (range 32–44 years), 35 years (range 28–40 years), and 38 years (range 32–46 years), respectively.
For classification purposes, 111 (25.6% of the patients fulfilling ASAS criteria for classification purposes) and 323 (74.4%) patients fulfilled the clinical and imaging arms of the ASAS criteria, respectively. Within the imaging arm of the ASAS criteria, 85 (19.5%), 74 (17.0%), and 164 (37.8%) patients fulfilled the MRI and radiographic positive arm, the MRI positive and radiographic negative, and the MRI negative and radiographic positive arms, respectively. Identically, patients fulfilling the clinical arm of the ASAS criteria were older (median age 41 years [range 33–49 years]) compared to the imaging arm (median age 38 years [range 32–45 years]) but younger than controls (median age 43 years [range 36–51 years]) (P < 0.0001) and no differences were assessed for sex or BMI. Within the imaging arm, patients from the MRI and radiographic positive group, MRI positive and radiographic negative, and radiographic positive and MRI negative arms had a median age of 39 years (range 33–45 years), 37 years (range 29–42 years), and 38 years (range 32–46 years), respectively.
This study confirms that ASAS criteria are highly specific and have acceptable sensitivity both for diagnosing and classifying SpA patients, with similar psychometric properties as described in the original validation study (). Tests with very high positive LR (>10) are considered of diagnostic value () and in this study, the positive LR of the ASAS criteria were >10 for both diagnosis and classification.
When testing the performances of the other SpA criteria, lower sensitivity and higher specificity for all criteria were found in the diagnostic setting, and higher sensitivity with similar specificity in the classification setting compared to the original cohorts that tested those criteria ([2-4]). It is worth noting that with the addition of the MRI to the old Amor and ESSG criteria, sensitivity was increased with a similar specificity; however, specificity of both mAmor and mESSG criteria remained lower than ASAS criteria specificity. The mNY criteria showed the highest positive LRs for both diagnosis and classification, followed by the ASAS criteria, and those 2 criteria sets were the only sets with positive LR values >10.
This study has some weaknesses, but also some strengths. One could consider that circularity, as the gold standard was defined by the opinion of the rheumatologist, is the main weakness of this study. However, this methodology has been used in the elaboration of the main sets of criteria used in the field of SpA (e.g., mNY, Amor, ESSG, and axial ASAS criteria). Moreover, this study was not aimed at the elaboration of criteria but at its evaluation, and therefore we admit that a potential circularity exists, but in a lower magnitude than in other studies aiming at elaborating criteria. Furthermore, the case report form was carefully prepared for the participating rheumatologists to first report their opinion on the diagnosis of CBP, and thereafter and only then to check (in a random order) for the presence or not of the different items of the different sets of criteria.
Another weakness is the retrospective design of the study for the diagnostic purpose; on the other hand, in a single study, data to assess the performances of the criteria both for diagnostic and classification purposes were collected in the same group of patients. The potential bias in the selection of patients, due to the study design, might also be seen as a weakness, i.e., the participating rheumatologists being aware that our center was involved in the design of the study might be more prone to recruit SpA patients, despite that it was well specified in the protocol that patients had to be included consecutively regardless of the reason for the visit and regardless of the definite diagnosis of CBP. This might explain the high prevalence of SpA in this study (35%); however, such findings are in line with other recently reported studies on the prevalence of SpA in patients presenting with CBP occurring before age 45 years (). Another possible bias might be due to the patient selection (less established disease than in other studies); by including patients diagnosed only after 1995, we limited the disease duration to 15 years. This might explain the differences in the performances of the different sets of criteria, particularly the low sensitivity rate for diagnosis compared to other studies. However, this reflects the clinical practice regarding MRI prescription in France for the diagnosis of CBP.
Furthermore, we considered in our regression analyses all missing data for HLA–B27, radiographic, and MRI as negative results, and this might lead to higher specificity rates, as control patients did not necessarily undergo those examinations (which we considered negative). However, once again, this study reflects our clinical daily practice, and HLA–B27 and MRI have been well reported to be frequently missing data in other observational studies in SpA. Furthermore, by considering a missing item as negative, we tried to reproduce the application of these criteria in real life, i.e., if an item (for example, HLA–B27) would be missing, the rheumatologist would consider it negative and still try to apply the criteria. Finally, regarding imaging modality, no central reading was performed, but here again, this reflects the clinical daily practice.
The study also has some strengths, as by recruiting consecutive patients, a representative sample of the population is ensured. Other remarkable findings from the study are that some individual items of the ASAS criteria had positive LRs >10 with posttest probabilities at approximately 90%; some of these were expected, such as MRI and radiographic sacroiliitis, or HLA–B27 positive, but others were not, such as raised CRP level for diagnostic purpose (posttest probability of 89%) and anterior uveitis and raised CRP level (posttest probability 89% and 86%, respectively) for classification purpose. This confirms the rationale of having added the raised CRP level in the ASAS criteria, as well as the clinical feeling that uveitis facilitates the recognition of the disease. In contrast, in our study, inflammatory back pain presented a very low positive LR, but this might be explained by the fact that all our patients already had CBP before they were age 45 years (inclusion criteria).
With this study, we confirm the validity of the ASAS criteria both for diagnosis and classification, in a clinical rheumatologic setting of young patients with CBP, for both its arms, suggesting that those criteria might help not only classifying, but also guiding the rheumatologist in diagnosing these patients. Other studies are required in other sets of patients to confirm (or not) our findings.
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. Moltó 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. Thibout, Claudepierre, van der Heijde, Dougados.
Acquisition of data. Thibout, Dougados.
Analysis and interpretation of data. Moltó, Paternotte, Comet, Thibout, Rudwaleit, van der Heijde, Dougados.
ROLE OF THE STUDY SPONSOR
Abbott had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was contingent upon approval by Abbott.
Author Comet is an employee of Axonal France, who was contracted for statistical analysis and article review.