Objective To estimate the prevalence of spondyloarthritis and its subtypes.
Methods The Swedish healthcare organisation comprises a system where all inpatient and outpatient care is registered by a personal identifier. For the calendar years 2003–7, all residents aged ≥15 years in the southernmost county of Sweden (1.2 million inhabitants) diagnosed by a physician with spondyloarthritis (ankylosing spondylitis (AS), psoriatic arthritis (PsA), inflammatory arthritis associated with inflammatory bowel disease (Aa-IBD) or undifferentiated spondylarthritis (USpA)) were identified. To obtain valid point estimates of prevalence by the end of 2007, identification numbers were cross-referenced with the population register to exclude patients who had died or relocated.
Results The authors estimated the prevalence of spondyloarthritis (not including chronic reactive arthritis) as 0.45% (95% CI 0.44% to 0.47%). The mean (SD) age of patients with prevalent spondyloarthritis by the end of 2007 was 53 (15) years. Among the component subtypes, PsA accounted for 54% of cases, AS 21.4%, USpA 17.8% and Aa-IBD 2.3% with a prevalence of 0.25%, 0.12%, 0.10% and 0.015%, respectively. The remaining 6.4% had some form of combination of spondyloarthritis diagnoses. The prevalence of spondyloarthritis at large was about the same in men and women. However, the subtype PsA was more prevalent in women and AS was more prevalent in men.
Conclusion In Sweden the prevalence of spondyloarthritis leading to a doctor consultation is not much lower than rheumatoid arthritis. PsA was the most frequent subtype followed by AS and USpA, and the two most frequent subtypes PsA and AS also display some distinct sex patterns.
Statistics from Altmetric.com
Spondyloarthritis includes a variety of both chronic and non-chronic rheumatic conditions with a common theme of inflammatory back pain, asymmetrical peripheral oligoarthritis, enthesopathy and an association with HLA-B27 antigen. Spondyloarthritis can be classified into five component subtypes: ankylosing spondylitis (AS), psoriatic arthritis (PsA), inflammatory arthritis associated with inflammatory bowel disease (Aa-IBD), undifferentiated spondylarthritis (USpA) and reactive arthritis (ReA).1 In Europe the overall prevalence of spondyloarthritis has been reported to be in the range of 0.3–1.9%.2,–,5 Prevalence and incidence may vary owing to differences in genetics (HLA-B27) and environmental triggers, but estimates also vary as a result of differences in study design and the fact that widely used and well-accepted classification criteria are lacking.6,–,8 However, criteria for the classification of both axial and peripheral spondyloarthritis have recently been published by the Assessment for Spondyloarthritis International Society (ASAS).9 10
Spondyloarthritis often causes limitations of functional capacity, decreased ability to work and a negative impact on the overall quality of life.7 11 12 In studying the causes and impact of spondyloarthritis on society, it is important to have up-to-date estimates of disease occurrence. Better knowledge of the epidemiology of diseases and its burden improves the possibilities for better healthcare planning plan and resource utilisation. However, the scientific documentation for spondyloarthritis concerning prevalence and burden on society is limited in most countries. Furthermore, the studies performed to date have often been based only on AS and PsA. The use of a comprehensive population-based healthcare register with diagnoses by physicians offers a promising opportunity to estimate the prevalence of disease requiring management by healthcare providers.13 Thus, our objective was to estimate the prevalence of spondyloarthritis and the age and gender characteristics of the patients.
The Skåne Health Care Register
In Sweden, free healthcare is provided by both public and private healthcare providers, all with the same tax-based financing system. All healthcare providers are required to submit information for reimbursement purposes, which ensures high quality reporting.
In the southernmost county of Sweden all healthcare visits—public and private, inpatient and outpatient—are registered with the Skåne Health Care Register (SHCR) by the patient's personal identification number (PIN). The register also includes information on healthcare provider, date of visit and, for public healthcare, the International Classification of Diseases (ICD-10) diagnoses as given by any physician in primary or secondary care. The SHCR covers one-eighth of the Swedish population.13
The population register
All Swedish residents are registered in the legislative mandatory population register by their PIN. This register includes vital data of, for example, date of birth, sex, marriages, residential address and date of death. The population register is continually updated by the Swedish Tax Agency and is used for a variety of reasons including voting records, tax purposes, healthcare providers, school authority, social insurance agency and banks.
Prevalence of spondyloarthritis
Using the SHCR, we identified all subjects who at any time during five calendar years (2003–7) had received a diagnosis of spondyloarthritis by searching for 12 separate ICD-10 codes for AS, PsA, Aa-IBD and USpA (table 1). All spondyloarthritis cases had to be aged ≥15 years. Our primary case definition required the diagnosis at least once by a rheumatologist or internist or at least on two separate occasions by any other physician in primary or secondary care. In clinical practice, over the years patients are given different diagnostic codes which are included in the spondyloarthritis group of disorders. A patient was only counted once for the overall estimate of spondyloarthritis prevalence, but once in each subset diagnosis (when fulfilling our case criteria for AS, PsA, Aa-IBD and USpA). Training in the diagnostic setting and use of the ICD system is an integral part of the medical training of physicians. We did not include diagnostic codes for ReA because of the typical non-chronic character of the disease.
In order to obtain valid point prevalence estimates of spondyloarthritis by 31 December 2007 (end of study period), we cross-referenced all cases identified with the population register to exclude subjects who had died or had relocated out of the county. We used the total population of Skåne at 31 December 2007 as the denominator, compensated by a 15% reduction to simulate loss of patients only seen by private practitioners whose diagnoses were not forwarded to the register (although the patients' PINs and dates of consultations are). This deduction was reached by the following reasoning. Approximately 30% of all outpatient clinic visits in the county are made to private practitioners. However, the majority of patients who visit private healthcare providers also consult public practising physicians in primary or secondary care where their diagnoses may get listed, or if admitted to a hospital (all hospitals are public). There were only four rheumatologists in private practice during this time frame (in total there were 46 specialists or physicians under specialty training in rheumatology practising in public healthcare by the end of 2007). We estimated that approximately half of those who seek private care also have a visit within public healthcare (eg, a public practising rheumatologist) where their spondyloarthritis diagnosis gets listed. We therefore reduced the denominator by only 15% so as not to yield inflated estimates of prevalence.
Validation of the ICD-10 diagnoses and sensitivity analyses
We performed one validation exercise of the diagnostic coding in the SHCR and three sensitivity analyses to evaluate the effect on our estimates of uncertainties in the diagnostic coding and the extent of missing data resulting from clinic visits without diagnoses forwarded to the SHCR. First, we validated the diagnostic coding by reviewing the medical records of patients who had visited the rheumatology clinic at Lund University Hospital (a major rheumatology clinic within the county). Second, we evaluated the differences in the prevalence by modifying our case criteria. We compared our primary more conservative case definition (the diagnosis given at least once by a rheumatologist/internist or at least two clinic visits with the diagnosis by any other physician in primary or secondary care) to a more liberal secondary case definition: the diagnosis at least once by any physician in primary or secondary care. Third, we varied the adjustment to the denominator (total population) by −5% to −25% to simulate the range of uncertainty of the proportion of cases with spondyloarthritis who were only diagnosed and seen by private healthcare providers. Fourth, we evaluated the prevalence of spondyloarthritis in a district of Skåne with no private rheumatologist and few private general practices without a reduction in the denominator.
We present both crude and age-standardised estimates (European standard population, direct method) to enable comparisons with datasets from other countries.14 Exact 95% CIs were calculated using binomial distribution (Stata/IC 10.0 for Windows software).
Prevalence of spondyloarthritis and its subtypes
Using our primary case definition of spondyloarthritis (including AS, PsA, Aa-IBD or USpA), we identified a total of 3852 patients (49.0% men) aged ≥15 years who were still alive and resident in the county (the number of cases identified who had died or relocated was 364). The population in the county aged ≥15 years was 849 253 (49.0% men). This resulted in an overall crude prevalence of spondyloarthritis of 0.45% (95% CI 0.44% to 0.47%) (figure 1). The standardised prevalence (European standard population) was also 0.45%. We also provide prevalence estimates for different age groups (15+, 20+ and 25+ years) and sexes to enable comparisons with other datasets (table 2). The mean (SD) age of patients with prevalent spondyloarthritis was 53 (15) years for both men and women. The peak prevalence was found in those aged 60–74 years but was also close to the peak prevalence in the group aged 45–59 years.
The most frequent subtype of spondyloarthritis was PsA with a prevalence of 0.25% (95% CI 0.24% to 0.26%) followed by AS with a prevalence of 0.12% (95% CI 0.11% to 0.124%) and USpA with a prevalence of 0.10% (95% CI 0.098% to 0.112%). Aa-IBD was the least frequent subtype with a prevalence of 0.015 % (95% CI 0.011% to 0.018%) (figure 2, table 2 and table in online supplement). If we merged the hallmark subtype of axial spondyloarthritis (M45) with psoriatic spondylitis (M07.2), an additional 36 individuals were identified in the group with axial SpA. However, this small group did not affect the overall prevalence of AS (still 0.12%). Most cases of spondyloarthritis (n=3258, 84.6%) had been diagnosed by a rheumatologist or internist.
Overlap between spondyloarthritis diagnoses
Of the total number of prevalent spondyloarthritis cases, 93.6% did not have a combination of diagnoses. PsA occurred in 51.9%, AS in 21.5%, USpA in 17.6% and Aa-IBD in 2.3%. The remaining 248 patients (6.4%) had some form of combination of spondyloarthritis diagnoses. The most frequent combinations were AS and USpA (n=133) and PsA and USpA (n=52) (table 3).
Validation of diagnoses in the SHCR
In a validation exercise of the accuracy of the diagnostic coding, we evaluated 347 patients identified in the SHCR with a spondyloarthritis diagnosis who were also included in a consecutive series of patients seen at an outpatient rheumatology clinic. The medical records were reviewed with regard to the items included in the classification criteria for AS (New York),15 PsA (CASPAR),16 17 USpA as well as a diagnosis of IBD (ESSG).1 A valid spondyloarthritis diagnosis was found in 340 (98%) of the cases. In the remaining seven patients (2%), ReA was found in two, otherwise the cause was still an unclear rheumatic condition.
Of the 347 patients identified in the SHCR, 197 (57%) fulfilled the ASAS classification criteria for peripheral spondyloarthritis. Of the 172 patients who had information available on HLA-B27 status and/or radiography of the sacroiliac joint (ASAS main criteria for axial spondyloarthritis), 157 (91%) fullfilled the ASAS classification criteria for axial spondyloarthritis and 70 (45%) of these also fulfilled the ASAS criteria for peripheral spondyloarthritis.
When we used a more liberal case definition where the spondyloarthritis diagnosis was only required on one occasion by any physician in primary or secondary care, we identified a total of 4618 patients resulting in a crude prevalence of 0.54% (95% CI 0.53% to 0.56%), an increase in prevalence of 0.09% from our primary estimate.
In the second sensitivity analysis we adjusted the denominator of the equation to simulate the loss of cases that were diagnosed and only seen in private healthcare (whose diagnoses were not forwarded to the SHCR). The error bars in figures 1 and 2 illustrate the effect on the prevalence estimates.
In a third and final sensitivity analysis we evaluated the prevalence of spondyloarthritis in a healthcare district with no private rheumatologist and few private practices without any adjustment to the denominator. The prevalence of spondyloarthritis standardised to the age and sex structure of the whole county was 0.51% (95% CI 0.46% to 0.56%) using our primary case definition (ie, marginally higher than 0.45%, our estimate from the county at large).
We estimated the prevalence of spondyloarthritis leading to a consultation with a doctor to be 0.45% in persons aged ≥15 years in southern Sweden. When comparing the prevalence of spondyloarthritis in persons aged ≥20 years (0.51%) with the estimate obtained for rheumatoid arthritis (0.66%) in the same county, this was only moderately lower.13
Our estimate is in line with findings from other population-based studies in Europe which also suggested a prevalence of about 0.5%.2 4 18 In the USA the National Arthritis Data Workgroup estimated the prevalence to be in the range of 0.35–1.3%, partly based on studies performed in Europe.19 The presence of HLA-B27 in the population varies between countries and ethnic groups and is higher in Scandinavia than in many other parts of the world.20 21 The varying estimates of the prevalence of spondyloarthritis should therefore be considered in relation to several methodological factors and study design limitations, making direct comparisons between studies difficult.
PsA and AS were found to be the two most prevalent subtypes of spondyloarthritis, making up about 80% of the cases which is in line with previous studies.5 22 The prevalence of the hallmark subtype of spondyloarthritis, AS, was previously estimated to be in the range 0.06–0.9% based on studies from the western world.2 3 5 19 23,–,26 A US report suggests a prevalence of AS of 0.13% in Caucasian populations,25 and a lower prevalence in African-Americans than in white subjects has been reported.19 According to our expectations, we found a higher prevalence of AS (0.12%) in Sweden than in parts of the world where the distribution of HLA-B27 is lower.27
We found the overall prevalence of spondyloarthritis to be as common in women as in men, which is in line with a community-wide study of Alaskan Eskimos and some European studies5 18 31 32 but contradictory to other European studies.2 3 The dominant male prevalence in AS is well accepted,33 and the studies with contradictory results either included a larger proportion of individuals from inhospital care or the subjects were mostly HLA-B27 positive blood donors with AS. Our study is based on an entire population. In line with previous studies, we found AS to be more common in men with a sex ratio of 3:1.23,–,25 34 The results of studies of PsA are conflicting, with some reporting that PsA is equally common in men and women,35 36 some that it is more frequent in men29 37 and others that it occurs more frequently in women.28
Consistent with the findings of previous studies, spondyloarthritis was most prevalent in the age group 60–74 years, decreasing thereafter.2 5 38 The prevalence in those aged 45–60 years was almost as high. Importantly, disease onset often occurs at a relatively young age. Thus, disease-related disability and associated health economic consequences may well be of the same magnitude as for rheumatoid arthritis. There is some evidence that the disease is associated with increased mortality,39 40 which may be one possible explanation for the declining prevalence at a higher age. It is also plausible that, when older people seek healthcare, a diagnosis of spondyloarthritis may not be in focus compared with other common diagnoses in older people. Interestingly, the prevalence curve for USpA in our study showed a shift towards the left (younger age) compared with the other subtypes, suggesting that this non-specific subtype of spondyloarthritis may possibly evolve into one of the other more distinct subtypes with time.
We would like to acknowledge important limitations of this study. First, we did not capture spondyloarthritis which has not led to a consultation with a doctor. However, importantly, we measured the proportion of the population who place a burden on the healthcare system. Second, there are no formal guidelines for the diagnosis of spondyloarthritis. The data are solely based on patient-driven clinic visits and the judgement of the individual physician. However, validation of the diagnoses via review of medical records at one of the major rheumatological clinics of the county indicated a high level of accuracy of the clinical diagnoses. Third, the diagnosis of patients seen only in private practice would not be listed in the register. However, the sensitivity analyses revealed only a marginal effect on the estimates in the range of this uncertainty.
To conclude, the occurrence of spondyloarthritis is much less studied than many other rheumatic diseases and up-to-date estimates of disease occurrence are highly valuable in order to understand the basic epidemiology and burden of disease on society (figure 2). In southern Sweden we found prevalent spondyloarthritis leading to a consultation with a doctor in 0.45% of the population aged ≥15 years.
The authors would like to thank the personnel at TietoEnator and Dennis Noreén and Thor Lithman for their assistance in the data extraction procedure from SHRC. The authors would also like to thank the EpiHealth network at Lund University, Sweden.
Competing interests None.
Funding This study was supported by grants from Region Skåne, the Swedish Social Insurance Agency, the Swedish Research Council, Faculty of Medicine, Lund University, the Österlund and Kock Foundations, King Gustav V 80 Year Fund and the Swedish Rheumatism Association.
Ethics approval This study was conducted with the approval of the ethics committee in Lund, Sweden.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.