Differentation of Post-Streptococcal Reactive Arthritis from Acute

Rheumatic Fever
JUDITH BARASH, MD, ERAN MASHIACH, MD, PNINA NAVON-ELKAN, MD, YACKOV BERKUN, MD, LIORA HAREL, MD, TSIVIA TAUBER, MD,
SHAI PADEH, MD, PHILIP J. HASHKES, MD, MSC, AND YOSEF UZIEL, MD, FOR THE PEDIATRIC RHEUMATOLOGY STUDY GROUP OF ISRAEL

Objective To perform a retrospective study comparing clinical and laboratory aspects of patients with acute rheumatic fever
(ARF) and patients with post-streptococcal reactive arthritis (PSRA), to discern whether these are 2 separate entities or varying
clinical manifestations of the same disease.
Study design We located the records of 68 patients with ARF and 159 patients with PSRA, whose diseases were diagnosed
with standardized criteria and treated by 8 pediatric rheumatologists in 7 medical centers, using the Israeli internet-based
pediatric rheumatology registry. The medical records of these patients were reviewed for demographic, clinical, and laboratory
variables, and the data were compared and analyzed with univariate, multivariate, and discriminatory analysis.
Results Four variables were found to differ significantly between ARF and PSRA and serve also as predictors: sedimentation
rate, C-reactive protein, duration of joint symptoms after starting anti-inflammatory
treatment, and relapse of joint symptoms after cessation of treatment. A discriminative
equation was derived that enabled us to correctly classify >80% of the patients.
Conclusion On the basis of simple clinical and laboratory variables, we were able to From the Pediatric Day Care and Pediatric
differentiate ARF from PSRA and correctly classify >80% of the patients. It appears that Rheumatology service (J.B.), and Depart-
ARF and PSRA are distinct entities. (J Pediatr 2008;153:696-9) ment of Pediatrics (E.M.), Kaplan Medical
Center, Rehovot, affiliated to the Hadassah
Medical School, the Hebrew University
Jerusalem, Israel; Pediatric Day Hospital,
ost-streptococcal reactive arthritis (PSRA) is defined as arthritis of ⱖ1 joints,

P
Shaarei Zedek Medical Center, Jerusalem, af-
associated with a recent group A streptococcal infection in a patient who does not filiated to the Hadassah Medical School, the
Hebrew University, Jerusalem Israel (P.N.-E.);
fulfill the Jones criteria for the diagnosis of acute rheumatic fever (ARF).1,2 Department of Pediatrics, Sheba Medical
Although other definitions have been suggested, mainly for adult cases, this definition is Center, Tel Hashomer, affiliated to the
Sackler Medical School, Tel Aviv University,
the most accepted and used by most investigators. Some authors consider PSRA to be part Tel Aviv, Israel (Y.B., S.P.); Schneider Chil-
of the spectrum of ARF, whereas other authors consider it to be a different entity. dren Hospital, Petah Tikva, affiliated to the
Authors who consider PSRA to be part of the ARF spectrum refer mainly to a small Sackler Medical School, Tel Aviv University,
Tel Aviv, Israel (L.H.); Pediatric Day Care,
number of cases in which PSRA has been diagnosed and in which carditis suggestive of Assaf Harofeh Medical Center, Zrifin, affili-
ARF subsequently developed.3,4 ated to the Sackler Medical School, Tel Aviv
University, Tel Aviv, Israel (T.T.); Section of
In 1993, Deighton5 proposed the following as distinguishing features of PSRA: Pediatric Rheumatology, Cleveland Clinic,
1) onset within 10 days of group A streptococcal infection, as opposed to 2 to 3 weeks in affiliated to the Cleveland Clinic Lerner
School of Medicine of Case Western Uni-
ARF; 2) Prolonged or recurrent arthritis, in contrast to ARF, in which the arthritis lasts versity, Cleveland, OH (P.H.); and Depart-
a few days to 3 weeks; 3) Slow and partial response to aspirin, whereas in ARF the ment of Pediatrics, Meir Medical Center,
response to aspirin is usually dramatic. Kfar Saba, affiliated to the Sackler Medical
School Tel Aviv University, Tel Aviv, Israel
The following diagnostic criteria for PSRA have been proposed by Ayoub and (Y.U.).
Ahmed based on these clinical features6: 1) Arthritis of acute onset, symmetric or Drs Hashkes and Uziel have contributed
asymmetric, usually non-migratory, can affect any joint, persistent or recurrent. The equally to this study.
The authors declare no potential conflict of
arthritis is at best only poorly responsive to salicylates or nonsteroidal anti-inflammatory interest.
drugs; 2) Evidence of antecedent group A streptococcal infection; and 3) Failure to fulfill Submitted for publication Dec 2, 2007; last
the modified Jones criteria for the diagnosis of ARF.2 revision received Mar 31, 2008; accepted
May 21, 2008.
The question whether PSRA is a distinct entity from ARF has not yet been
Reprint requests: Judith Barash, MD, Pediatric
answered. Our objective was to search for differences in these 2 entities in a well-defined, Day Care, Kaplan Medical Center, PO Box 1,
large cohort. We attempted to discern whether they are 2 separate entities or varying Rehovot 76100, Israel. E-mail: judith_b@
clalit.org.il; barashj@inter.net.il.
0022-3476/$ - see front matter
Copyright © 2008 Mosby Inc. All rights
reserved.
ARF Acute rheumatic fever PSRA Post-streptococcal reactive arthritis
10.1016/j.jpeds.2008.05.044

696
Table I. Clinical and laboratory data of patients
with acute rheumatic fever and post-streptococcal
reactive arthritis
ARF PSRA P
(n ⴝ 68) (n ⴝ 159) value
Fever °C (SD) 38.3 (0.4) 38.3 (0.5) NS
% with Fever ⬎38 66 16 .0004
Number of active joints 2.5 (1.2) 1.8 (1.3) .0004
(SD)
Migratory arthritis % 79 33 .004
Symetrical arthritis % 40 22 .05
Carditis % 60 0 ⬍.0001
ESR (SD) 92.2 (31.1) 57.1 (40.9) ⬍.0001
CRP mg/L (SD) 106.7 (83.5) 22.6 (44.1) ⬍.0001
ASO IU (SD) 1011 (1573) 889 (733) NS
Positive throat culture % 77% 76% NS
Interval from pharyngitis 15 (9.2) 14.6 (10.1) NS
to arthritis, days (SD)
Figure. The distribution in percent with active joints in patients with
Response to treatment, 2.2 (1.7) 6.9 (5.9) ⬍.0001 ARF and PSRA. Large LE, Large joint lower extremity; small LE, small
days (SD) joint lower extremity; large UE, large joint upper extremity; small UE,
Relapse % 7 21 0.013 small joint upper extremity. *P ⫽ .0002.
ESR, Erythrocyte sedimentation rate; CRP, C-reactive protein; ASO, anti-streptolysin
O; response to treatment, resolution of joint symptoms in response to ant-inflammatory
treatment; relapse, relapse of joint symptoms after cessation of anti-inflammatory treat-
ment; NS, not significant. years for ARF and 9.3 ⫾ 3.6 years for PSRA. A total of 63% of
patients with ARF and 54% of patients with PSRA were male,
and the number of persons in the household was 5.9 ⫾ 1.5 for
clinical manifestations of the same disease and, if possible, to ARF and 6.5 ⫾ 2.1 for the patients with PSRA. A family history
offer a simple clinical tool to differentiate between them. of ARF in first- or second-degree relatives was found in 7.2% of
the patients with ARF and 7.5% of the patients with PSRA.
METHODS The clinical and laboratory characteristics are summa-
We located patients, ⬍16 years old, who received a rized in Table I. Fever ⬎38°C was recorded in 46 patients
diagnosis and were treated by 8 pediatric rheumatologists with ARF (66%) and 25 patients with PSRA (16%). The
from 7 centers who participated in the Israeli internet-based erythrocyte sedimentation rate and the C-reactive protein
pediatric rheumatology registry.7 We searched for patients in level were significantly higher in the ARF group. The 2
whom ARF was diagnosed with joint involvement and pa- groups did not differ in the proportion with positive results of
tients with a diagnosis of PSRA between the years 1996 and throat cultures at onset of joint disease, and no significant
2005 (most after 2001). A total of 159 patients with PSRA difference was found in the levels of antistreptolysin O that
and 68 patients with ARF for whom there was sufficient were markedly increased in both groups.
analyzable data were found. The patients with ARF had significantly more active
ARF was diagnosed according to the revised Jones joints (P ⫽ .0004), with 79% of the patients having migratory
criteria,2 and PSRA was diagnosed in cases of arthritis in- arthritis, in contrast to 33% of the patients in the PSRA group
volving ⱖ1 joints associated with proven group A streptococ- (P ⫽ .004). The distribution of involved joints (Figure) did
cal infection in a patient not fulfilling the Jones criteria. The not differ in the groups, except large joints in the upper
medical charts of these patients were reviewed by a single extremities were more frequently involved in ARF (P ⫽
physician, and all demographic, clinical, and laboratory data .0002). Arthritis was symmetrical in 40% of patients with
were summarized. The characteristics of ARF and PSRA ARF and 22% of the patients with PSRA (P ⫽ .05). Axial
were compared with 2-sample t tests (for continuous vari- joint involvement was not documented in either group.
ables) and ␹2 tests (for categorical variables). Multivariate A total of 41 patients with ARF (60%) had carditis by
backwards stepwise logistic regression analysis was used to evaluation of a qualified pediatric cardiologist. None of the
model the probability of ARF versus PSRA. A prediction patients with PSRA had proven carditis. Two patients with
equation was calculated by using discriminant analysis.8 The PSRA had prolonged P-R intervals with electrocardiography,
study was approved by each center’s ethical committee. and 3 patients with PSRA had a trace of mitral regurgitation
on echocardiogram, which was not considered pathologic.
RESULTS The interval from onset of pharyngitis to the onset of arthritis
There was no significant difference in the 2 groups in was 15.1 ⫾ 9.2 days (documented in 76% of the patients) for
demographic characteristics. The age of onset was 10.2 ⫾ 3.0 the patients with ARF, versus 14.6 ⫾ 10.1 days for the

Differentiation of Post-Streptococcal Reactive Arthritis from Acute Rheumatic Fever 697

Table II. Significant predictors of acute rheumatic
fever by using stepwise logistic regression
Variable Odds ratio 95% CI Significance
ESR 1.015 1.000-1.031 0.043
CRP 1.016 1.004-1.028 0.007
Days to disappearance 0.565 0.389-0.820 0.003
of joint symptoms
Relapse after cessation 0.026 0.002-0.390 0.008
of treatment (yes/no)
patients with PSRA (documented in 70% of the patients; P ⫽
not significant).
The time required for resolution of joint symptoms in
response to anti-inflammatory treatment was significantly
shorter in patients with ARF as compared with patients with
PSRA (2.18 ⫾ 1.71 days versus 6.73 ⫾ 5.89, respectively;
P ⬍ .0001). Resolution of joint symptoms was defined at the
date stated by the attending physician that no arthritis or
arthralgia was noted on physical examination. Only 7% of
patients with ARF sustained a relapse after cessation of anti-
inflammatory therapy, compared with 21% of patients with
PSRA (P ⫽ .013) requiring repeated anti-inflammatory treat-
ment. The relapse of joint complaints suggests a longer nat-
ural course of arthritis.
Twenty-two percent of the patients with PSRA did not
receive anti-inflammatory treatment, and their joint disease
was treated only with rest, compared with only 1 patient in
the ARF group.
A total of 87% of patients with ARF received long-term
prophylactic antibiotic treatment, compared with 50% of pa-
tients with PSRA; 50% of the patients with ARF received
intramuscular penicillin versus 4% of the patients with PSRA.
Despite more intensive antibiotic prophylaxis in patients with
ARF, 12% had a recurrence, compared with 8% of patients in
the PSRA group.
The 2 most significant variables differentiating ARF
from PSRA on univariate analysis were the time required for
resolution of joint symptoms in response to anti-inflamma-
tory therapy and the number of patients who had a relapse of
joint disease after cessation of anti-inflammatory treatment.
By using multivariate, backward, stepwise logistic regres-
sion, we found 4 variables to be significant diagnostic discrimi-
nators between ARF and PSRA: erythrocyte sedimentation rate,
C-reactive protein, duration of joint symptoms in response to
treatment with anti-inflammatory medications, and relapse of
joint symptoms after cessation of anti-inflammatory treatment
(Table II). By using discriminant analysis, we calculated this
prediction equation: ⫺1.568 ⫹ 0.015 ⫻ sedimentation rate ⫹
0.02 ⫻ CRP – 0.162 ⫻ days to resolution of joint symptoms –
2.04 ⫻ return of joint symptom (yes ⫽ 1, no ⫽ 0).
If the value was ⬎0, the patient was classified as having
ARF; otherwise, the patient was classified as having PSRA.
This prediction gave 79% sensitivity rate (correct classification
as ARF) and 87.5% specificity rate (correct classification as
PSRA).
698 Barash et al
DISCUSSION
Two recent studies attempted to answer whether ARF
and PSRA are distinct entities and to validate the diagnostic
criteria proposed by Ayoub and Ahmed.6 Tutar et al9 com-
pared 24 children with PSRA with 20 patients with ARF.
The latency period from upper respiratory tract infection was
significantly shorter in patients with PSRA; however, 25% of
the patients with ARF also had a short latency period. Un-
responsiveness of articular symptoms to salicylate therapy was
more frequent in PSRA, but also 45% of the patients with
ARF with joint symptoms did not respond during the first 72
hours. The authors concluded that because there was consid-
erable overlap between PSRA and ARF, these 2 entities could
not easily be distinguished.
In 2004, Mackie and Keat10 addressed the same issue by
conducting a systematic search on Medline using strict inclu-
sion criteria. They identified 188 cases of PSRA published in
the literature between 1982 and 2002. One hundred of these
cases were in adults ⬎18 years old. The authors concluded
that PSRA is a heterogeneous group of clinical entities, some
of which share features of ARF and others with HLA B27-
related spondyloarthropathies and that the assumed causal
role of streptococcal infection is far from proven.
There are several studies addressing the association of
ARF and PSRA with class II HLA-DR antigens. Ahmed
et al11 found an increased frequency of HLA DRB1*01 in
patients with PSRA compared with healthy control subjects
and to patients with ARF. In patients with ARF, there was an
increased frequency of HLA DRB1*16 allele compared with
control subjects. This association may suggest that the patho-
genesis of PSRA, like that of ARF, may be related to the
inheritance of certain class II HLA alleles. In 2004, Simo-
nini12 published a different observation on genotyping of the
HLA-DRB1 locus in 25 patients with ARF, 34 patients with
PSRA, and healthy control subjects. The frequency of
DRB1*01 and DRB1*16 in patients with PSRA and ARF did
not show any statistical significance when compared with each
other or with that of the control subjects. The conclusion in
this study was that the 2 diseases are not genetically different
in these alleles.
The presence of a non-HLA B cell alloantigen D8/17
in a significantly greater percentage of patients with ARF and
their first-degree relatives compared with control subjects has
been shown by Khanna.13 In a study of Israeli patients,
Harel14 has found a significantly higher percentage of B cells
expressing the D8/17 antigen in patients with a history of
ARF than in control subjects. Later the same author investi-
gated the presence of D8/17 alloantigen on B cells from
patients with PSRA compared with control subjects.15 There
was a small difference between the expression of the antigen
in patients with PSRA and subjects in the control group, but
with significant overlap in the 2 groups. Moreover, there was
a weak negative correlation between the percentage of D8/17
positive cells and the time elapsed from diagnosis. It was
suggested previously that this expression is regulated by the
The Journal of Pediatrics • November 2008
infection, so it is questionable whether it is a true genetic
marker.
In this study, we have shown that we were able to
differentiate between ARF and PSRA in ⬎80% of the cases
on the basis of 4 criteria: the sedimentation rate and CRP at
onset, the number of days before resolution of joint symptoms
after starting anti-inflammatory therapy, and the presence or
absence of a recurrence of arthritis after discontinuation of
anti-inflammatory therapy. Other variables proposed as dif-
ferentiating factors in the 2 entities, including the latency
period between development of pharyngitis and joint symp-
toms which was approximately 3 to 10 days for PSRA in
earlier studies,9,11,12 did not differ between ARF and PSRA
in our population. This observation in our cohort is not easily
explained; 1 possibility is that there was a parental recall bias
and that possibly the parents, in response to this question,
referred to an earlier upper respiratory illness and not neces-
sarily to the one that caused the arthritis.
Although not significant in multivariate analysis, more
patients with ARF had involvement of ⬎1 joint, symmetrical
and migratory arthritis. There are some similarities and a few
differences in the characteristics of the arthritis in our patients as
compared with patients in earlier studies. There were more joints
involved and significantly more patients with migratory arthritis
in our study and in the earlier studies. Contrary to the other
studies, none of our patients had axial involvement, and there
were more patients with symmetrical arthritis in the ARF group.
Because our analysis revealed that more patients with
ARF had fever and markedly elevated erythrocyte sedimen-
tation rate, we reexamined the charts of all patients classified
as having PSRA who had fever, sedimentation rate of ⬎60
mm/hour, and at least 2 active joints. Ten of the 11 patients
with PSRA we found with these features had positive results
on a throat culture. We assume that the high erythrocyte
sedimentation rate and fever in these patients represented the
acute streptococcal illness in some or most of these cases.
On the basis of the results of this study, ARF and PSRA
appear to be 2 distinct entities. Although ARF has a more acute
presentation, with fever, acute phase response, a greater number
of joints, and frequency of cardiac involvement, the response to
treatment is much quicker and the course of arthritis is shorter
than in PSRA. If PSRA was a milder form of the spectrum of
ARF, we would not expect a slower response to treatment or a
longer course of the arthritis.
We suggest that our equation might be used by clini-
cians to correctly classify their patients, at least until there are
more accurate and easier to use guidelines. However, this
equation will need to be validated in a prospective study.
When there is doubt about the diagnosis, we suggest to favor
the diagnosis of ARF and administer the usual antibiotic
prophylactic treatment.
Differentiation of Post-Streptococcal Reactive Arthritis from Acute Rheumatic Fever
There is a growing need for guidelines about the ne-
cessity for and duration of anti-streptococcal prophylaxis in
patients in whom PSRA is diagnosed. This study was not
designed to answer that question. Shulman and Ayoub,16 the
American Heart Association, and the Red Book of the
AAP17 suggest that antibiotic prophylaxis be given for 1 year,
and if no carditis is observed, then prophylaxis should be
discontinued.
Our retrospective data are subject to parental recall and
physician diagnostic biases, but the large size of this series
may help to diminish these concerns.
We conclude that ARF and PSRA are different enti-
ties, although both are consequences of group A streptococcal
infection and involve the joints. Different approaches to an-
tibiotic prophylaxis in these 2 entities may be justified.
REFERENCES
1. Goldsmith DP, Long SS. Post streptococcal disease of childhood: a changing
syndrome. Arthritis Rheum 1982;25(Suppl 4):S18.
2. Special writing group of the committee on rheumatic fever, endocarditis and
Kawasaki disease of the council on cardiovascular disease in the young, American Heart
Association Guidelines for the diagnosis of rheumatic fever: Jones criteria, updated.
JAMA 1992;68:2069-73.
3. De Cunto CM, Giannini EH, Fink CW, Brewer EJ, Person DA. Prognosis of
children with poststreptococcal reactive arthritis. Pediatr Inf Dis J 1988;7:683-6.
4. Schaffer EM, Agarwal R, Helm J, Gingell RL, Roland JM, O’Neil KM. Post-
streptococcal reactive arthritis and silent carditis: a case report and review of the
literature. Pediatrics 1994;93:837-9.
5. Deighton C. Beta hemolytic streptococci and reactive arthritis in adults. Ann
Rheum Dis 1993;52:475-82.
6. Ayoub EM, Ahmed S. Update on the complications of group A streptococcal
infections. Curr Probl Pediatr 1997;27:90-101.
7. Hashkes PJ, Uziel Y, Rubenstein M, Navon P, Padeh S, Mukamel M, et al for the
Pediatric Rheumatology Study Group of Israel. The Israeli Internet-based registry:
novel concept of multicenter data collection in pediatric rheumatology. Ped Rheum
Online J 2004;2:37-50.
8. Dillon WR, Goldstein M. Multivariate analysis: methods and applications. New
York, New York: John Wiley and Sons; 1984.
9. Tutar E, Atalay S, Yilmaz E, Ucar T, Kocak G, Imamoglu A. Poststreptococcal
reactive arthritis in children. Rheumatology Int 2002;22:80-3.
10. Mackie SL, Keat A. Poststreptococcal reactive arthritis: what is it and how do we
know? Rheumatology 2004;43:949-54.
11. Ahmed S, Ayoub EM, Scornik JC, Wang CY, She JX. Poststreptococcal reactive
arthritis: clinical characteristics and association with HLA-DR alleles. Arthritis Rheum
1998;41:1096-102.
12. Simonini G, Porfirio B, Cimaz R, Calabri GB, Giani T, Falcini F. Lack of
association between the HLA-DRB1 locus and post-streptococcal reactive arthritis and
acute rheumatic fever in Italian children. Semin Arthritis Rheum 2004;34:553-8.
13. Khanna AK, Buskirk DR, Williams RC Jr, Gibofsky A, Ceow MK, Menon A, et al.
Presence of non-HLA B cell antigen in RF patients and their families as defined by a
monoclonal antibody. J Clin Invest 1989;8:1710-6.
14. Harel L, Zeharia A, Kodman Y, Straussberg R, Zabriskie JB, Amir J. Presence of
D8/17 B-cell marker in children with rheumatic fever in Israel. Clin Genet
2002;61:293-8.
15. Harel L, Mukamel M, Zeharia A, Kodman Y, Prais D, Uziel Y, et al. Presence of
D8/17 B-cell marker in patients with poststreptococcal reactive arthritis. Rheumatol Int
2007;27:695-8.
16. Shulman S, Ayoub EM. Poststreptococcal reactive arthritis. Curr Opin Rheumatol
2002;14:562-5.
17. Hall CB, Kimberlin DW. Group A streptococcal infection. In: Red Book, Report of
the Committee on Infectious Diseases, American Academy of Pediatrics 27th ed. Elk Grove
Village, Illinois, 2006. p. 610-20.
699