C

Epidemiologic approaches to in

fection and immunity:

the case of reactive arthritisSherry Rohekar and Janet Pope

Department of Rheumatology, St Joseph’s Healthcare,London, Ontario, Canada

Correspondence to Dr Sherry Rohekar, Department ofRheumatology, St Joseph’s Healthcare, 268 GrosvenorStreet, London, ON N6A 4V2, CanadaTel: +1 519 646 6242; fax: +1 519 646 6348;e-mail: sherry.rohekar@sjhc.london.on.ca

Current Opinion in Rheumatology 2009,21:386–390

Purpose of review

There is significant evidence that infection and arthritis are linked, but the nature of this

association is unclear. The goal of this review is to examine the case of reactive arthritis

(ReA), an inflammatory arthritis with a clear infectious trigger. We will first examine the

current state of knowledge of ReA epidemiology and follow it with a discussion of the

epidemiologic challenges that ReA studies face.

Recent findings

Recent studies have examined outbreaks of gastroenteritis to try and elucidate the

epidemiology of ReA. Some have found higher levels of self-reported arthritis than

previously thought, and others have implicated organisms such as Escherichia coli

O157:H7 that were not traditionally associated with ReA. There is also evidence that the

severity of initial infection may be associated with a higher relative risk of developing

ReA. New population-based studies have further clarified the natural history of infection

and subsequent ReA, demonstrating the power of community surveillance. Despite

these findings, several methodological issues complicate the study of ReA. Problems

include lack of standard diagnostic criteria, varying culture methods, selection bias and

difficulties in establishing a control population.

Summary

Recent studies have continued to increase our knowledge of the epidemiology of ReA.

Addressing the multiple challenges that face the study of infection and arthritis will be

very useful for future study.

Keywords

arthritis, epidemiology, infection, methodology, reactive arthritis

Curr Opin Rheumatol 21:386–390� 2009 Wolters Kluwer Health | Lippincott Williams & Wilkins1040-8711

IntroductionEpidemiologists strive to understand the factors and

determinants of health events in large populations, with

the goal of applying this knowledge toward the reduction

of disease burden. Many practical barriers prevent the

study of the epidemiology of infection as it relates to

arthritis, such as heterogeneity in populations and arthri-

tis types. Studies are often not directly comparable to

each other due to differences in methodology. Despite

these barriers, much can still be learned about the

relationship between infection, autoimmune response

and arthritis.

BackgroundThe complex path from infection to autoimmune arthritis

can be studied by examining reactive arthritis (ReA), an

inflammatory arthritis with a clear infectious trigger. ReA

is typically defined as a sterile synovitis that occurs after

an infection at a distant site, usually the gastrointestinal

opyright © Lippincott Williams & Wilkins. Unautho

1040-8711 � 2009 Wolters Kluwer Health | Lippincott Williams & Wilkins

or genitourinary tract. The most common triggers are

Yersinia, Salmonella, Shigella and Campylobacter [1]. ReA is

usually grouped with the seronegative spondyloarthro-

pathies (SpA) due to similarities in clinical presentation

[1–3]. Patients typically present with an asymmetric

monoarthritis or oligoarthritis of the lower extremities

[1–3]. They may also have sacroiliitis, back pain and

extraarticular features such as enthesitis, rash and ocular

inflammation [1–3]. Some patients will completely

resolve their arthritic symptoms; a proportion will pro-

gress to chronic disease [1–3]. Table 1 [4] summarizes the

common clinical features of ReA.

There is significant evidence that the inciting infection

and the arthritis are indeed linked. Microbial DNA and

RNA of the triggering organism can be retrieved from the

sites of inflammation, including synovial fluid [2,3,5,6].

There is also evidence that host factors, particularly

the presence of HLA-B27, play an important role in

the pathogenesis of ReA [2–4,6]. However, these simple

factors of antecedent infection in a genetically predis-

rized reproduction of this article is prohibited.

DOI:10.1097/BOR.0b013e32832aac66

C

Epidemiologic approaches to infection and immunity Rohekar and Pope 387

Table 1 Common clinical manifestations of reactive arthritis

System Manifestation

Musculoskeletal ArthralgiaArthritisEnthesitisDactylitisOsteitis/hyperostosis

Mucocutaneous Keratoderma blennorrhagicaPsoriasiformOral ulcersNail dystrophyErythema nodosum

Ocular ConjunctivitisUveitisEpiscleritisCorneal ulcers

Urogenital Circinate balanitisSterile urethritisProstatitisCystitisCervivitisSalpingo-oophoritis

Adapted from [1–4].

posed individual do not tell the full story of ReA as seen

in the rheumatology clinic. In this review, we will first

examine the current state of knowledge of epidemiology

of ReA and follow it with a discussion of the epidemio-

logic challenges that ReA studies face.

Epidemiology of reactive arthritisThe annual incidence of ReA is usually estimated at 30–

40 cases per 100 000 people and the prevalence 1–7%

[7,8�,9]. ReA occurs 1–6 weeks after a preceding enteric

or urogenital infection [10]. The risk of developing ReA

after such an infection is 1–4%, but is substantially

increased to 20–25% in those who carry HLA-B27

[10]. HLA-B27 is an important factor in the presentation

of ReA, but it varies widely between ethnicities [10].

Recent research has indicated that there may be a dose–

response effect, with those suffering from more severe

initial infection more likely to develop arthritis [11��].

Table 2 demonstrates the increased relative risk of self-

reported arthritis in those with no, moderate and severe

symptoms of gastroenteritis [11��]. The accuracy of these

prevalence, incidence and risk estimates is questionable

due to several practical barriers to research, to be discussed

later.

opyright © Lippincott Williams & Wilkins. Unauth

Table 2 Results from a study of an outbreak of Campylobacter a

gastrointestinal symptoms were associated with an increased rela

Severity of gastroenteritisduring outbreak

Self-reported arthritis (%),age and sex-standardized

None 15.7Moderate 17.6Severe 21.6

CI, confidence interval; RR, relative risk. Modified with permission from [11

Most patients present with an asymmetrical oligoarthritis

of the lower extremities, but up to 20% may have an

inflammatory polyarthritis [10]. Inflammatory back pain

(40% of patients), enthesitis (20%) and ocular inflam-

mation (10%) are also common [10]. Approximately half

of those affected will be free of symptoms by 6 months,

but 20% may proceed to a chronic course [10]. One large

study [12] found objective physical findings on joint

examination of those with ReA 5 years after disease onset.

A wide variety of organisms have been implicated in ReA;

the most common triggers in the developed world include

Chlamydia, Salmonella, Shigella and Yersinia [2,4]. One

study [13] found that either Chlamydia trachomatis, Yersi-nia or Salmonella was identified as the triggering bacter-

ium in about 50% of patients with probable or possible

ReA. More recent data from outbreak sources indicate

that Campylobacter and possibly Escherichia coli O157:H7

are also significant triggers of ReA [11��,14��]. However, a

causative organism is detected in only 56% of those

diagnosed with ReA, even when sought in those with

typical symptoms of a gastrointestinal or genitourinary

infection [13]. Even fewer pathogens (47%) are identified

in those who present clinically with undifferentiated

oligoarthritis that is compatible with ReA but lacks an

infectious prodrome [13].

Several recent studies have examined outbreaks to try

and elucidate the epidemiology of ReA. A survey of the

victims of a large outbreak of Salmonella enteritidis in

Canada demonstrated a high rate of self-reported symp-

toms consistent with ReA (62.5% of respondents, 11.5%

of those infected) [15��]. In this study, the most com-

monly reported extraintestinal symptoms of the Salmo-nella outbreak included joint pain, swelling or stiffness

(46.2%), morning stiffness more than 30 min (35.6%) and

inflammatory eye symptoms (24.0%) [15��]. A few

(19.2%) patients reported visible joint swelling [15��].

Interestingly, the incidence of HLA-B27 was the same in

those with ReA compared with those without, though

HLA-B27 was present more frequently in those who were

infected than in healthy controls [odds ratio (OR) 3.0]

[15��]. This contrasts with previous outbreak studies,

which show a trend toward increased joint damage, a

greater number of affected joints and enthesitis in those

who were HLA-B27-positive [12]. A further analysis of

genetic samples obtained from the participants in the

orized reproduction of this article is prohibited.

nd Escherichia coli O157:H7 demonstrating that more severe

tive risk of self-reported arthritis

Crude RR (95% CI)Age and sex-adjusted

RR (95% CI)

1.00 (reference) 1.00 (reference)1.18 (0.98–1.43) 1.19 (0.99–1.43)1.34 (1.07–1.66) 1.33 (1.07–1.66)

��].

C

388 Infectious arthritis and immune dysfunction

most recent study [16�] revealed that genetic variants of

Toll-like receptor 2 (TLR-2) were significantly associ-

ated with acute ReA.

Another Canadian outbreak, this time of Campylobacterjejuni and E. coli O157:H7, was studied prospectively for

4.5 years for participants’ self-report of chronic arthritis

[11��]. This study examined both patients with reported

gastroenteritis and those who were asymptomatic [11��].

Compared with those who were asymptomatic, people

with moderate gastrointestinal symptoms had an adjusted

relative risk of arthritis of 1.19 [95% confidence interval

(CI) 0.99–1.43] [11��]. The relative risk was higher in

patients who reported severe diarrhea (1.33, 95% CI

1.07–1.66). Interestingly, rates of prescription medi-

cations for arthritis did not seem to associate with gastro-

enteritis, but those with dysentery may have a relative

contraindication to NSAIDs [11��].

Some studies rely on follow-up of community-reported

disease, rather than sporadic outbreaks. One study [17��]

retrospectively analyzed a cohort with gastroenteritis in

Sweden over a period of 7 years and standardized inci-

dence ratios (SIRs) were calculated by dividing the

observed number of cases of ReA by the expected

number of cases. Within 1 year, patients with Yersiniainfections had an elevated risk of ReA (SIR 47.0, 95% CI

21.5–89.2). Smaller but still substantial increases in

SIR for ReA were found in patients with Salmonella(SIR 18.2, 95% CI 12.0–26.5) and Campylobacter enteritis

(SIR 6.3, 95% CI 3.5–10.4) [17��]. Other musculoskeletal

complications of enteric infection included septic arthritis

and osteomyelitis [17��].

In another population-based study [18��], an American

telephone interview series was recently conducted on

those with culture-confirmed enteric infections (Campy-lobacter, E. coli O157:H7, Salmonella, Shigella and Yersi-nia). In this study, 13% of respondents developed symp-

toms compatible with ReA, with the most common

symptoms including new joint pain (56%), new low back

pain (44%) and new morning stiffness (32%) [18��].

Fifteen percent of those with rheumatic complaints

sought healthcare for their new arthritic symptoms; in

fact, 6% were formally diagnosed with ReA [18��]. The

incidence of ReA was highest after infection with Cam-pylobacter (2.1/100 000) and Salmonella (1.4/100 000)

[18��]. The investigators were able to actually take rheu-

matologic histories and perform physical examinations on

a subset of 82 patients [18��]. Of this subset, 66% were

confirmed to have ReA by a rheumatologist, with enthe-

sitis being the most common physical finding (88.8%)

[18��]. A smaller but still substantial proportion of

patients were found to have peripheral joint arthritis

(19%) [18��]. This large population-based study found

no association between ReA and HLA-B27, but did find

opyright © Lippincott Williams & Wilkins. Unautho

that arthritis correlated with increased severity of initial

infection [18��].

A recent study in Denmark [14��] also found a significant

association between the severity of the initial gastro-

enteritis and the development of ReA. In this study,

14% of individuals affected by gastroenteritis developed

symptoms of ReA by survey self-report [14��]. A subset of

respondents were clinically examined; 21% had objective

synovitis, 42% had findings compatible with ReA and

37% had symptoms not related to ReA [14��]. The OR for

ReA in an HLA-B27-positive individual was 2.62 (95% CI

1.67–3.93) [14��]. Thus, there appears to be controversy

on whether HLA-B27 is associated with community ReA

[12,14��].

Epidemiologic challenges in reactive arthritisresearchThe recent studies detailed above are thoughtful

approaches to elucidating the epidemiology of ReA.

However, ReA is not a clinical entity that is simple to

study and its epidemiologic characterization is fraught

with problems. In fact, even a standard definition of ReA

may be lacking. As one author eloquently wrote, ReA

occupies ‘the conceptual ground somewhere between

septic arthritis and the classic autoimmune rheumatic

diseases such as rheumatoid arthritis’ [6].

One of the greatest challenges facing researchers is that

there is no single defined set of criteria for the clinical

diagnosis of ReA. The American College of Rheumatol-

ogy (ACR) classification criteria for ReA are based on a

1981 study of the percentage sensitivity and specificity of

various criteria for ‘typical’ Reiter’s syndrome [19,20�]. In

this framework, a definition of an episode of arthritis of

more than 1 month with urethritis and/or cervicitis has a

specificity of 98.2% and sensitivity of 84.3% [19]. However,

a large proportion of those with gastrointestinal or genito-

urinary infection can remain completely asymptomatic,

with only subclinical disease. Some may not remember the

occurrence of mild extraarticular symptoms (recall bias). In

these individuals, ReA may present only as a chronic

undifferentiated oligoarthritis. Because of these conun-

drums in classification, studies vary widely in their

inclusion and exclusion criteria; furthermore, subsets

of those with ReA are likely not included due to lack

of classical symptoms. These issues predispose studies of

ReA to self-selection bias, wherein more severely

affected individuals are more likely to participate.

A further challenge to the study of ReA is the lack of a

single diagnostic test for the inciting infectious organism.

Stool cultures vary between centers and are often nega-

tive by the time patients present with their arthritic

symptoms. The measure of antibody levels to pathogenic

rized reproduction of this article is prohibited.

C

Epidemiologic approaches to infection and immunity Rohekar and Pope 389

organisms is complicated by high rates of positivity in

control populations [21]. Crossreactivity with other

species may also occur; some tests for Chlamydia do

not differentiate between C. trachomatis and the common

C. pneumoniae [3]. Even the results of PCR techniques to

identify organisms in the genitourinary tract or directly

from synovial fluid vary greatly between centers [21].

Diagnostic difficulties are amplified when host properties

are considered, such as the presence or absence of HLA-

B27 and its subgroups [21]. Populations vary significantly

in their baseline prevalence of HLA-B27, from 4 to 53%

[1]. Additionally, we now have the ability to examine

increasingly more specific patient properties, such as the

use of genotyping to detect variants in TLRs [16�]. Some

of these issues in diagnosis have been carefully explored

in a review article by Sieper et al. [21], who calculated a

series of posttest probabilities for the diagnosis of ReA

following a variety of diagnostic tests.

Another issue complicating ReA research is that of set-

ting. The prevalence of pathogens varies greatly from

country to country, and between communities, and fre-

quently changes with time. For example, gastroenteritis

due to Yersinia infection is common in parts of Europe,

but rare in North America [21]. This makes direct com-

parison between studies difficult. Studies conducted in

hospitals are prone to selection or referral bias, wherein

included patients are more likely to have severe or

prolonged disease.

Outbreaks afford a unique opportunity to investigators

interested in studying ReA. By prospectively studying

those affected by point-source infections, investigators

can minimize ascertainment bias, in which patients

associate prior noninfectious gastrointestinal or genito-

urinary symptoms with their arthritis [15��]. Unfortu-

nately, outbreak studies are often difficult to conduct

practically. Due to cost and practicality, it is difficult to

have the arthritis assessed by a physician; most studies rely

on patients’ self-report. Again, patients with more severe

disease may be more likely to respond. In the case of some

recent outbreaks, the possibility of class-action lawsuits

against companies with contaminated food products may

affect study participation. It is difficult to ascertain the true

number of exposed individuals in any outbreak, thus

making it difficult to establish a true sampling frame of

those affected and controls. Additional problems include

accessing patient records in outbreaks (often protected by

privacy legislation) and obtaining timely ethics approval to

conduct the study [15��].

Finally, it is possible that ‘ReA’ encompasses a constella-

tion of different diseases rather than a unified entity.

Arthritis triggered by urinary infections may be clinically

significantly different from those with gastrointestinal

opyright © Lippincott Williams & Wilkins. Unauth

origins. Several studies [3,6,21] have suggested that anti-

biotics may be effective in genitourinary Chlamydia-

induced ReA, but not after a gastroenteritis.

ConclusionReA typifies the complexities involved in studying dis-

ease processes in which the interplay between genetics

and environment is critical. ReA, by definition, involves

an infectious trigger and an inflammatory musculoskele-

tal outcome, but the intermediate steps of autoimmune

reaction remain elusive.

Recent studies have continued to increase our knowledge

of the epidemiology of ReA. These studies include out-

break research and large population-based cohorts. Many

suggest [9,11��,12,15��,17��,18��] that ReA may be a fairly

common consequence of genitourinary or gastrointestinal

infection. Organisms previously overlooked as triggers,

such as E. coli H7:O157, are emerging as potential

arthritic pathogens [11��]. With recent trends to treat

‘undifferentiated’ arthritis early, ReA is an important

clinical consideration for any rheumatologist.

There are great challenges that face the study of ReA.

The establishment of a clear set of diagnostic criteria

would be a great benefit and should be the focus of

further discussion. Establishing widely accepted diagnos-

tic criteria would allow researchers studying diverse

populations to compare their results more directly with

those of other centers. This consistency would simplify

the goal of fully understanding which determinants of

health may be used to reduce the burden of autoimmune

inflammatory arthritis.

References and recommended readingPapers of particular interest, published within the annual period of review, havebeen highlighted as:� of special interest�� of outstanding interest

Additional references related to this topic can also be found in the CurrentWorld Literature section in this issue (p. 430).

1 Hill Gaston JS, Lillicrap MS. Arthritis associated with enteric infection. BestPract Res Clin Rheumatol 2003; 17:219–239.

2 Toivanen A, Toivanen P. Reactive arthritis. Best Pract Res Clin Rheumatol2004; 18:689–703.

3 Colmegna I, Cuchacovich R, Espinoza LR. HLA-B27-associated reactivearthritis: pathogenetic and clinical considerations. Clin Microbiol Rev2004; 17:348–369.

4 Toivanen A, Toivanen P. Reactive arthritis. Curr Opin Rheumatol 2000;12:300–305.

5 Kim TH, Uhm WS, Inman RD. Pathogenesis of ankylosing spondylitis andreactive arthritis. Curr Opin Rheumatol 2005; 17:400–405.

6 Inman RD. Mechanisms of disease: infection and spondyloarthritis. Nat ClinPract Rheumatol 2006; 2:163–169.

7 Hochberg MC, Silman AJ, Smolen JS, et al., editors. Rheumatology. 3rd ed.London: Mosby; 2003.

8

�Pope JE, Krizova A, Garg AX, et al. Campylobacter reactive arthritis: asystematic review. Semin Arthritis Rheum 2007; 37:48–55.

An excellent review of the literature on the epidemiology of Campylobacter-associated ReA.

orized reproduction of this article is prohibited.

C

390 Infectious arthritis and immune dysfunction

9 Soderlin MK, Borjesson O, Kautiainen H, et al. Annual incidence of inflam-matory joint diseases in a population based study in southern Sweden. AnnRheum Dis 2002; 61:911–915.

10 Sieper J, Rudwaleit M, Khan MA, Braun J. Concepts and epidemiologyof spondyloarthritis. Best Pract Res Clin Rheumatol 2006; 20:401–417.

11

��Garg AX, Pope JE, Thiessen-Philbrook H, et al., Walkerton Health StudyInvestigators. Arthritis risk after acute bacterial gastroenteritis. Rheumatology(Oxford) 2008; 47:200–204.

A study of a large community outbreak of E. coli O157:H7 and Campylobacter with4.5 years of follow-up. Demonstrated that those with more severe diarrhealsymptoms had an increased relative risk of arthritis.

12 Thomson GT, DeRubeis DA, Hodge MA, et al. Post-Salmonella reactivearthritis: late clinical sequelae in a point source cohort. Am J Med 1995;98:13–21.

13 Fendler C, Laitko S, Sorensen H, et al. Frequency of triggering bacteria inpatients with reactive arthritis and undifferentiated oligoarthritis and therelative importance of the tests used for diagnosis. Ann Rheum Dis 2001;60:337–343.

14

��Schiellerup P, Krogfelt KA, Locht H. A comparison of self-reported jointsymptoms following infection with different enteric pathogens: effect ofHLA-B27. J Rheumatol 2008; 35:480–487.

Another study demonstrating severity of ReA was correlated with the severity ofgastrointestinal symptoms, as well as an association with HLA-B27.

opyright © Lippincott Williams & Wilkins. Unautho

15

��Rohekar S, Tsui FW, Tsui HW, et al. Symptomatic acute reactive arthritis afteran outbreak of Salmonella. J Rheumatol 2008; 35:1599–1602.

Outbreak study demonstrating high rates of self-reported ReA after Salmonellainfection.

16

�Tsui FW, Xi N, Rohekar S, et al. Toll-like receptor 2 variants are associatedwith acute reactive arthritis. Arthritis Rheum 2008; 58:3436–3438.

This study showed that genetic variants of TLR-2, but not TLR-4, were associatedwith acute ReA after infection with Salmonella.

17

��Ternhag A, Torner A, Svensson A, et al. Short- and long-term effects ofbacterial gastrointestinal infections. Emerg Infect Dis 2008; 14:143–148.

Population-based study that calculated SIRs for ReA with Yersinia, Salmonella andCampylobacter infections.

18

��Townes JM, Deodhar AA, Laine ES, et al. Reactive arthritis following culture-confirmed infections with bacterial enteric pathogens in Minnesota andOregon: a population-based study. Ann Rheum Dis 2008; 67:1689–1696.

Large population-based study of ReA after Salmonella, Shigella and Yersiniainfections in the United States.

19 Willkens RF, Arnett FC, Bitter T, et al. Reiter’s syndrome: evaluation ofpreliminary criteria for definite disease. Arthritis Rheum 1981; 24:844–849.

20

�American College of Rheumatology. 2009. www.rheumatology.org.

Includes the current ACR classification criteria for ReA.

21 Sieper J, Rudwaleit M, Braun J, van der Heijde D. Diagnosing reactive arthritis:role of clinical setting in the value of serologic and microbiologic assays.Arthritis Rheum 2002; 46:319–327.

rized reproduction of this article is prohibited.