ORIGINAL ARTICLE
Septic Arthritis and Acute Rheumatic Fever in Children: The Diagnostic Value of Serological Inflammatory Markers Raakhi M. Mistry, MBChB,* Diana Lennon, MBChB, FRACP,w Matthew J. Boyle, MBChB, FRACS,w Karel Chivers, MBChB, FRACS,z Chris Frampton, BSc (Hons), PhD,y Ross Nicholson, MBChB, FRACP,z and Haemish Crawford, MBChB, FRACSw
Introduction: Joint pain and raised inflammatory markers are features of both acute rheumatic fever (ARF) and septic arthritis, often posing a diagnostic challenge to clinicians. Important differences in the presenting serological inflammatory marker profile may assist patient diagnosis, however, as clinical experience suggests that ARF is associated with a higher erythrocyte sedimentation rate (ESR), whereas other serological markers may be similarly elevated in these 2 conditions. Objective: The goal of this study was to determine the diagnostic value of serological inflammatory markers and white cell count (WCC) in children presenting with acute joint pain secondary to ARF or septic arthritis. Methods: Data were obtained from the Auckland regional rheumatic fever database and hospital computer records between 2005 and 2012. Records of all patients under the age of 16 years who were admitted with a new diagnosis of ARF or septic arthritis were analyzed. The diagnosis of ARF was defined on the basis of the New Zealand modification of the Jones Criteria, and the diagnosis of septic arthritis was defined on the basis of joint fluid cytology and culture. Baseline characteristics, serological inflammatory markers, and serum WCC were compared between the ARF and septic arthritis patient groups. Results: Children with ARF displayed significantly higher ESR, higher serum C-reactive protein, and lower serum WCC than children with septic arthritis on presentation to hospital. In children presenting with monoarthritis, an ESR > 64.5, serum WCC < 12.1 109/L, and age above 8.5 years were found to be significant independent predictors of ARF. Children with all 3 predictors had a 71% risk for ARF and a 29% risk for septic arthritis. A significant proportion (30%) of children with the final diagnosis of ARF initially presented with monoarthritis; 14% of these children (5/34) had received nonsteroidal antiinflammatory medication before hospital presentation, and 74% of these children (25/34) had abnormal echocardiograms on admission.
From the *Department of Pediatric Orthopedics, Starship Hospital, Park Road; wStarship Hospital; zKidz First Hospital, Auckland; zGisborne Hospital, Gisborne; and yDepartment of Biostatistics, University of Otago, Dunedin, New Zealand. The authors declare no conflicts of interest. Reprints: Raakhi M. Mistry, MBChB, Department of Pediatric Orthopedics, Starship Hospital, 2 Park Road, Auckland, New Zealand. E-mail: [email protected]. Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved.
318 | www.pedorthopaedics.com
Conclusions: ARF and septic arthritis are important diagnoses to consider in children presenting with acute joint pain in New Zealand. A significant proportion of patients with ARF initially present with acute monoarthritis. Serological inflammatory markers and WCC on presentation differ significantly between children with ARF and septic arthritis. Key Words: rheumatic fever, septic arthritis, monoarthritis (J Pediatr Orthop 2015;35:318–322)
J
oint pain and raised serological inflammatory markers are features of both acute rheumatic fever (ARF) and septic arthritis, often posing a diagnostic challenge to clinicians. Expeditiously distinguishing between these 2 diseases is of the utmost importance, as their clinical management is substantially different. Although ARF is managed symptomatically and does not typically require operative management, correct diagnosis is crucial to reduce the long-term cardiac morbidity. In contrast, septic arthritis is an orthopaedic emergency. The consequences of delayed diagnosis of septic arthritis include sepsis, cartilage damage, osteonecrosis, and growth arrest, making early operative drainage vital. Previous authors have reported on the value of serological inflammatory markers, in addition to other clinical features, in differentiating between transient synovitis of the hip and septic arthritis.1,2 In 1999, Kocher et al3 undertook a retrospective review of children with acute monoarthritis affecting the hip. They were able to develop a predictive algorithm on the basis of 4 variables: painful weight-bearing, history of fever, erythrocyte sedimentation rate (ESR) > 40, and white cell count (WCC) > 12. In 2006, Caird and colleagues similarly demonstrated that patients fulfilling Kocher’s criteria plus having a serological C-reactive protein (CRP) > 20 mg/L had a 98% chance of having septic arthritis. In addition, they found that a CRP > 20 mg/L was a strong independent risk factor for septic arthritis.1 The literature investigating the role of serological inflammatory markers in differentiating between ARF and septic arthritis is currently limited to small case series.4,5 One case series from Utah looked at 3 patients with ARF and 12 patients with septic arthritis presenting with monoarthritis.4 The serological inflammatory markers J Pediatr Orthop
Volume 35, Number 3, April/May 2015
J Pediatr Orthop
Volume 35, Number 3, April/May 2015
and WCC were elevated in both groups but there was no significant difference in values between the 2 groups. Although this study was limited by small sample size, it did demonstrate the important overlap in clinical and laboratory findings in ARF and septic arthritis. Another case series from Fiji looked at 3 patients with ARF who were initially misdiagnosed with septic arthritis despite having sterile joint fluid.5 The goal of our study was to determine the diagnostic value of serological inflammatory markers on presentation in differentiating between ARF and septic arthritis in children, in order to assist appropriate initial management.
METHODS Following institutional review board approval, clinical records of patients aged 16 years or younger with the final diagnosis of ARF or septic arthritis were identified. Patients with ARF were identified from the Auckland regional rheumatic fever database. ARF was defined on the basis of the New Zealand modification of the Jones Criteria.6,7 Each case of ARF was confirmed by an independent pediatrician. Patients with septic arthritis were identified from Starship Children’s Hospital and Middlemore Hospital’s computer records using the ICD10 code M00. Septic arthritis was defined as a positive synovial fluid on Gram staining, a positive synovial fluid culture, or a synovial fluid WCC > 50,000/mm3. Patients who fulfilled at least 1 of these criteria were defined as having definite septic arthritis. Patients who the laboratory reported as having a synovial WCC “too high to count” were defined as having probable septic arthritis. A power analysis indicated that 110 patients per group were required to detect a difference in ESR of 15% between the ARF group and the septic arthritis group as statistically significant (2-tailed a = 0.05) with 90% power. Consecutive hospital records from 2005 to 2012 were subsequently reviewed; hospital record review was concluded once >110 patients with confirmed diagnoses had been included in each patient group. The ARF group was compared with the septic arthritis group with respect to baseline characteristics [age, sex, ability to bear weight on affected limb, fever (presenting tympanic temperature >381C), use of nonsteroidal anti-inflammatory drug (NSAID) or antibiotic administration before presentation, and number of joints
Septic Arthritis and Acute Rheumatic Fever in Children
involved at presentation] and serum laboratory values on presentation (ESR, CRP, WCC). Blood culture and synovial aspirate results were also analyzed and compared between the 2 groups, where available.
Statistical Analyses Baseline demographics, clinical features, and serum inflammatory markers were compared between the ARF and septic arthritis groups using independent t tests and w2 or Fisher exact tests according to variable types. Synovial fluid WCC levels and the percentage of polymorphonucleocytes were compared between the groups using Mann-Whitney U tests. Within the ARF group, the serum inflammatory markers were compared between patients with single joint involvement and patients with multiple joint involvement using Mann-Whitney U tests.
RESULTS We identified 114 children with ARF and 111 children with acute septic arthritis (103 definite septic arthritis, 8 probable septic arthritis). The cases of ARF were identified from hospital records covering 3 years and the cases of septic arthritis were identified from records covering 6 years, within the 7-year study period.
Baseline Information At the time of hospital presentation, the ARF patients were significantly older (mean age 10.4 vs. 5.2 y, P < 0.001) and were less often febrile (51.8% vs. 73.9%, P < 0.001) than the septic arthritis patients (Table 1). There was no significant difference in sex (59% vs. 64% of patients male, P = 0.399), or inability to bear weight on the affected limb (97.4% vs. 100% of patients, P = 0.85), between the 2 patient groups. Before hospital presentation, ARF patients received antibiotics (19.3% vs. 4.5%, P < 0.001) or NSAID medication (10.5% vs. 1%, P = 0.002) more often than septic arthritis patients. The majority of ARF patients presenting with monoarthritis did not receive NSAID medication before their admission (86%, 29/34).
Serum Inflammatory Markers At the time of hospital presentation, the ARF patients displayed a significantly higher ESR (median 95.0 vs. 50.0 mm/h, P < 0.001), a significantly higher CRP (median 92.0 vs. 52.5 mg/L, P = 0.003), and a significantly lower WCC (mean 10.8 109 vs. 13.1 109/L, P < 0.001) than the
TABLE 1. Features on Presentation Variable Mean age (y) Male sex [n (%)] Inability to bear weight [n (%)] Temperature > 381C [n (%)] Monoarthritis at presentation [n (%)] Polyarthritis at presentation [n (%)] History of NSAID use [n (%)] History of antibiotic use [n (%)]
Septic Arthritis (n = 111)
Rheumatic Fever (n = 119)
P
5.2 (range, 6 wk-14 y) 73 (64) 111 (100) 82 (73.9) 109 (98.2) 2 (1.8)w 1 (0.9) 5 (4.52)
10.4 (range, 4-16 y) 65 (59) 111 (97.4) 59 (51.8) 34 (29.8)* 80 (70.2) 12 (10.5) 22 (19.3)
< 0.001 0.399 0.85 < 0.001 < 0.001 < 0.001 0.002 < 0.001
*14% (5/34) history of prior NSAID use. wMicroorganisms cultured—methicillin-sensitive S. aureus; S. pyogenes.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.pedorthopaedics.com |
319
J Pediatr Orthop
Mistry et al
Volume 35, Number 3, April/May 2015
TABLE 2. Inflammatory Markers Variable
Septic Arthritis (n = 111)
ESR (mm/h) CRP (mg/L) WCC ( 109/L)
53.2 (range, 1-138) 87.4 (range, 0-412) 14.5 (range, 4.53-78)
septic arthritis patients (Table 2). Within the ARF group, there was no significant difference in ESR (P = 0.309), CRP (P = 0.151), or WCC (P = 0.872) between patients with single joint involvement and patients with multiple joint involvement on presentation.
Monoarthritis At the time of hospital presentation, there were more patients presenting with monoarthritis in the septic arthritis group (n = 109) than in the ARF group (n = 34). The ARF patients with monoarthritis were older than the septic arthritis patients with monoarthritis (mean 9.6 vs. 5.2 y, P < 0.0001). The ARF patients with monoarthritis had a higher ESR than the septic arthritis patients with monoarthritis (mean 84.9 vs. 52.4, P < 0.0001). There was no significant difference between the 2 groups with respect to CRP and WCC (Table 3). Twenty-five of 34 of patients with ARF had echo changes on admission.
Rheumatic Fever (n = 119)
P
90.2 (range, 10-140) 93.8 (range, 1-239) 12.2 (range, 4.7-136)
< 0.001 0.003 < 0.001
Independent Risk Factors for Final Diagnosis To determine which of the measures identified by univariate analyses as showing statistically significant associations independently contributed to the discrimination of ARF and septic arthritis, we performed forward and backward stepwise logistic regression analyses. These analyses identified age, serum ESR, and serum WCC as 3 independent factors. In an effort to create amenable scores to discriminate between ARF and septic arthritis, we further defined the 3 risk factors; patient age above 8.5 years, serum ESR > 64.5, and serum WCC < 12.1 109/L. If a patient displayed all 3 of these risk factors or presented with polyarthritis there was a 94% chance of ARF as the final diagnosis. If polyarthritic patients were excluded, a patient presenting with monoarthritis in addition to 1, 2, or 3 other risk factors had a 17%, 38%, or 71% likelihood of ARF (and an 83%, 62%, or 29% likelihood of septic arthritis) as the final diagnosis, respectively (Table 5).
Synovial Fluid Analysis The ARF patients underwent joint aspiration much less frequently than the septic arthritis patients (5.3% vs. 100% of patients, P < 0.001) on presentation to hospital. Of the patients who underwent aspiration, ARF patients displayed a significantly lower synovial fluid WCC (median 20.2 vs. 102.5/mm3, P = 0.004) than the septic arthritis patients; however, there was no significant difference in the percentage of polymorphonucleocytes (median 63.5% vs. 79.0% of synovial fluid WCC, P = 0.761) between the 2 groups. With respect to synovial fluid culture in the septic arthritis group, the most commonly cultured organisms were methicillin-sensitive Staphylococcus aureus (25.2% of patients), Streptococcus pneumoniae (7.2%), K. kingae (7.2%), S. pyogenes (6.3%), and methicillin-resistant S. aureus (2.7%) (Table 3). With respect to synovial fluid culture in the septic arthritis group in children aged below 2 years, the most commonly cultured organisms were K. kingae (18.1%), S. pyogenes (9.1%), S. pneumoniae (9.1%), methicillin-sensitive S. aureus (6.8%), Escherichia coli (4.5%), and S. agalactiae (4.5%); 31.8% of these patients had no growth from their synovial culture (Table 4).
DISCUSSION Accurate diagnosis of the child presenting with acute joint pain is essential. After radiographic abnormalities have been ruled out, the remaining differential diagnoses can pose a clinical challenge, with the most common diagnoses including transient synovitis, septic arthritis, peri-articular osteomyelitis, and inflammatory arthritis. Diagnosis and treatment of septic arthritis is time critical and therefore a high index of suspicion is important. Distinguishing between septic arthritis and transient synovitis in children has previously been well investigated,1,3 with the use of Kocher’s criteria well established in clinical practice. In the New Zealand context, ARF is an important diagnosis to consider in a child presenting with arthritis.
Polyarthritis Polyarthritis is one of the major Jones criteria for the diagnosis of ARF, with joint pain in ARF patients tending to be migratory and predominantly involving the large joints.7–9 In our study the majority of children with ARF presented with polyarthritis (70.2%, n = 80). Only a small proportion of children with septic arthritis presented with polyarthritis (1.8%, n = 2). With a sample
TABLE 3. Monoarthritis Variable
Septic Arthritis (n = 109)
ESR (mm/h) CRP (mg/L) WCC ( 109/L) Age (y)
320 | www.pedorthopaedics.com
52.4 82.9 14.4 5.2
(range, (range, (range, (range,
Rheumatic Fever (n = 34)
1-138) 0-412) 0-78) 0-14)
84.9 83.7 11.2 9.6
Copyright
r
(range, (range, (range, (range,
17-133) 1-239) 5.9-18.3) 6-13)
P < 0.001 0.968 0.026 < 0.001
2014 Wolters Kluwer Health, Inc. All rights reserved.
J Pediatr Orthop
Volume 35, Number 3, April/May 2015
TABLE 4. Most Common Organisms Cultured in the Septic Arthritis Group Microorganism
% (n = 111)
Methicillin-sensitive S. aureus S. pneumoniae K. kingae S. pyogenes Methicillin-resistant S. aureus
24.3 7.2 7.2 6.3 2.7
(n = 27) (n = 8) (n = 8) (n = 7) (n = 3)
size of 2 it is difficult to draw any significant comparative differences between the 2 groups; however, we did note that both children in the septic arthritis group presented with raised pulse rate, fever, elevated serum inflammatory markers and notably a CRP > 300, and positive Gram staining on joint aspirate.
Monoarthritis This study has confirmed that ARF can present early with monoarthritis. Earlier studies have identified monoarthritis as a potential presenting feature of ARF.2,4,10,11 One retrospective review from Brazil looked at 109 patients with ARF; they found that 3.3% of their 92 cases with arthritis presented with monoarthritis.12 Another retrospective review looked at 367 patients with ARF in Australia’s Northern Territory and found that 17% of the confirmed ARF cases presented with acute monoarthritis.3,8 Another study from New Zealand showed that the incidence of monoarthritis in ARF was 6% (5/79); however, the authors did note that all of the monoarthritic cases had received NSAIDs before their presentation.13 In our study, 30% of patients with ARF presented with monoarthritis. When these patients with monoarthritis were compared with the septic arthritis patients with monoarthritis, we were able to demonstrate significant differences in age and ESR between these 2 groups. It is interesting to note that 14% (5/34) of these patients received NSAIDs before their presentation. Although it is thought that pretreatment with NSAID therapy may halt the progression of monoarthritis to polyarthritis in ARF patients,5 it is more likely that the patients presenting with monoarthritis in our study may have been in the early phases of ARF disease development. Nevertheless, these findings highlight the importance of including ARF in the differential diagnosis of all children presenting with monoarthritis. TABLE 5. Predicted Probability of ARF Versus Septic Arthritis in Patients Presenting With Monoarthritis
No. Predictors* 1 2 3
Predicted Probability of ARF (%)
Predicted Probability of Septic Arthritis (%)
17 38 71
83 62 29
*Logistic regression analysis determined 3 independent predictors to discriminate between ARF and septic arthritis: patient age > 8.5 years, serum ESR > 64.5, and serum WCC < 12.1 109/L.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
Septic Arthritis and Acute Rheumatic Fever in Children
Laboratory Variables The literature investigating the diagnostic value of serological inflammatory markers and WCC in differentiating between ARF and septic arthritis is limited.4,5,9 Although previous studies have been limited by small sample size, they identify an important overlap in the clinical features of these 2 conditions. With larger patient numbers we were able to identify clear differences in laboratory results between these 2 patient groups, with a significantly higher ESR, higher CRP, and lower WCC seen in the ARF group compared with the septic arthritis group. Our findings are consistent with our hypothesis that septic arthritis has a more acute presentation, with a shorter period of inflammation, and thus may display lower serological inflammatory markers on presentation. In contrast ARF tends to have a more indolent presentation, with a longer duration of inflammation, and thus typically presents with higher serological inflammatory markers. Children with suspected ARF require prompt, appropriate investigation including streptococcal serology tests and echocardiography. In our study 5.3% of patients with a final diagnosis of ARF underwent joint washout for presumed septic arthritis. While delaying the diagnosis of septic arthritis is hugely detrimental, care must be taken not to significantly delay the diagnosis of ARF in affected patients. Delayed diagnosis of ARF can potentially result in progression of cardiac disease.14 One study showed that misdiagnosis on initial presentation resulted in progression of rheumatic heart disease.5 In this study, the authors reported a case series of 3 children from Fiji who presented with arthritis, fever, and elevated inflammatory markers. One child presented with monoarthritis and the other 2 presented with polyarthritis (2 joints involved). Joint aspirates were performed in all cases and all were negative. The children were commenced on antibiotics and subsequently discharged home. All 3 children returned 1 to 3 years later with established moderate to severe rheumatic heart disease. When we applied our predictive algorithm to the child with monoarthritis—an 8-year-old boy; ESR 75; and normal WCC—our predicted probability of ARF was 71%. Combined with a negative joint aspirate, this should alert the clinician to undertake further investigation.
Epidemiology Quality epidemiological data on the global incidence of ARF is lacking. Despite this there is a clear trend in developed countries showing a decline in incidence.15 In New Zealand, however, the incidence has remained high.13,14 A recent review of hospitalizations (not epidemiological survey) showed that the incidence in New Zealand was 17.2/100,000.16 It is well established that the burden of ARF is carried predominantly by minority groups within a given area who are of a lower socioeconomic status.13 In New Zealand this has been clearly demonstrated in indigenous Maori and New Zealand-born Pacific Island people. A study by Milne et al17 found that the incidence of ARF in school-age www.pedorthopaedics.com |
321
J Pediatr Orthop
Mistry et al
Pacific Island children in New Zealand is 80 to 100 per 100,000. In New Zealand where rheumatic fever is highly endemic, awareness of the clinical features of this debilitating disease is clearly essential for clinicians evaluating children with acute monoarthritis. This paper has shown that it is not uncommon for children with ARF to present to an orthopaedic service with acute monoarthritis. Raised awareness among clinicians in New Zealand has resulted in these children being referred for a pediatric medical opinion and/or an echocardiogram. Historically a number of severe cases of rheumatic heart disease have occurred in patients in whom the diagnosis was first missed.
Echocardiography In children suspected of ARF, echocardiogram is the investigation of choice. If the initial echocardiogram is normal, the standard protocol is to repeat it within 2 to 3 weeks.18 In our study when we reviewed the echocardiograms of the 34 children with ARF who presented with acute monoarthritis we found that 9/34 had normal echocardiograms on admission. Of those 9 normal echocardiogram results, 5 were repeated. Of these, 1 was normal and 4 had developed mild aortic or mitral regurgitation. Twenty-five of 34 children had abnormal echocardiograms on admission, 6 were severe, 13 were moderate, and 6 were mild. Thus, if heart disease was already established NSAID agents were used rather than waiting for polyarthritis to ensue. In our study, we found that a child above 8.5 years of age presenting with monoarthritis and a serological ESR > 64.5 and WCC < 12.1 had a 71% chance of having ARF. In our population correspondingly, we feel we have provided strong evidence to consider the diagnosis of ARF in all children presenting with monoarthritis, but particularly in older children with an elevated serological ESR and a low serum WCC.
CONCLUSIONS ARF and septic arthritis are important diagnoses to consider in children presenting with acute joint pain. Our results suggest that serological inflammatory markers and WCC on presentation differ significantly between children with ARF and septic arthritis. Contrary to traditional teaching, we have found that a significant proportion of
322 | www.pedorthopaedics.com
Volume 35, Number 3, April/May 2015
patients with ARF initially present with acute monoarthritis. REFERENCES 1. Caird MS, Flynn JM, Leung YL, et al. Factors distinguishing septic arthritis from transient synovitis of the hip in children. A prospective study. J Bone Joint Surg Am. 2006;88:1251–1257. 2. Carapetis JR, Currie BJ. Rheumatic fever in a high incidence population: the importance of monoarthritis and low grade fever. Arch Dis Child. 2001;85:223–227. 3. Kocher MS, Zurakowski D, Kasser JR. Differentiating between septic arthritis and transient synovitis of the hip in children: an evidence-based clinical prediction algorithm. J Bone Joint Surg Am. 1999;81:1662–1670. 4. Harlan GA, Tani LY, Byington CL. Rheumatic fever presenting as monoarticular arthritis. Pediatr Infect Dis J. 2006;25:743–746. 5. Mataika R, Carapetis JR, Kado J, et al. Acute rheumatic fever: an important differential diagnosis of septic arthritis. J Trop Pediatr. 2008;54:205–207. 6. New Zealand Heart Foundation. New Zealand Modification of the Jones Criteria. Auckland: New Zealand Heart Foundation; 2013. 7. Guidelines for the diagnosis of rheumatic fever. Jones Criteria, 1992 Update. Special Writing Group of the Committee on Rheumatic Fever, Endocarditis, and Kawasaki Disease of the Council on Cardiovascular Disease in the Young of the American Heart Association. JAMA. 1992;268:2069–73. [Erratum appears in JAMA, 1993, 27;269(4):476]. 8. Long S. Principles and Practice of Pediatric Infectious Diseases. Philadelphia, PA: Elsevier; 2012. 9. The National Heart Foundation of New Zealand. A Summary of the New Zealand Guidelines for Rheumatic Fever. Auckland, New Zealand: The National Heart Foundation of New Zealand; 2010. 10. Khriesat I, Najada AH. Acute rheumatic fever without early carditis: an atypical clinical presentation. Eur J Pediatr. 2003;162:868–871. 11. Wilson E, Wilson N, Voss L, et al. Monoarthritis in rheumatic fever? Pediatr Infect Dis J. 2007;26:369–370. 12. Pileggi GC. Atypical arthritis in children with rheumatic fever. J Pediatr (Rio J). 2000;76:49–54. 13. Harrison GJ, Cherry JD, Kaplan SL, et al. Feigin and Cherry’s Textbook of Pediatric Infectious Diseases. Vol. 7th Edition, Chapter 29 - Acute Rheumatic Fever. Philadelphia: Saunders; 2014. 14. Carapetis JR, McDonald M, Wilson NJ. Acute rheumatic fever. Lancet. 2005;366:155–168. 15. Miyake CY, Gauvreau K, Tani LY, et al. Characteristics of children discharged from hospitals in the United States in 2000 with the diagnosis of acute rheumatic fever. Pediatrics. 2007;120:503–508. 16. Milne RJ, Lennon DR, Stewart JM, et al. Incidence of acute rheumatic fever in New Zealand children and youth. J Paediatr Child Health. 2012;48:685–691. 17. Milne RJ, Lennon DR, Stewart JM, et al. Incidence of acute rheumatic fever in New Zealand children and youth. J Paediatr Child Health. 2011;48:685–691. 18. Abernethy M, Bass N, Sharpe N, et al. Doppler echocardiography and the early diagnosis of carditis in acute rheumatic fever. Aust N Z J Med. 1994;24:530–535.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
Septic Arthritis and Acute Rheumatic Fever in Children: The Diagnostic Value of Serological Inflammatory Markers Raakhi M. Mistry, MBChB,* Diana Lennon, MBChB, FRACP,w Matthew J. Boyle, MBChB, FRACS,w Karel Chivers, MBChB, FRACS,z Chris Frampton, BSc (Hons), PhD,y Ross Nicholson, MBChB, FRACP,z and Haemish Crawford, MBChB, FRACSw
Introduction: Joint pain and raised inflammatory markers are features of both acute rheumatic fever (ARF) and septic arthritis, often posing a diagnostic challenge to clinicians. Important differences in the presenting serological inflammatory marker profile may assist patient diagnosis, however, as clinical experience suggests that ARF is associated with a higher erythrocyte sedimentation rate (ESR), whereas other serological markers may be similarly elevated in these 2 conditions. Objective: The goal of this study was to determine the diagnostic value of serological inflammatory markers and white cell count (WCC) in children presenting with acute joint pain secondary to ARF or septic arthritis. Methods: Data were obtained from the Auckland regional rheumatic fever database and hospital computer records between 2005 and 2012. Records of all patients under the age of 16 years who were admitted with a new diagnosis of ARF or septic arthritis were analyzed. The diagnosis of ARF was defined on the basis of the New Zealand modification of the Jones Criteria, and the diagnosis of septic arthritis was defined on the basis of joint fluid cytology and culture. Baseline characteristics, serological inflammatory markers, and serum WCC were compared between the ARF and septic arthritis patient groups. Results: Children with ARF displayed significantly higher ESR, higher serum C-reactive protein, and lower serum WCC than children with septic arthritis on presentation to hospital. In children presenting with monoarthritis, an ESR > 64.5, serum WCC < 12.1 109/L, and age above 8.5 years were found to be significant independent predictors of ARF. Children with all 3 predictors had a 71% risk for ARF and a 29% risk for septic arthritis. A significant proportion (30%) of children with the final diagnosis of ARF initially presented with monoarthritis; 14% of these children (5/34) had received nonsteroidal antiinflammatory medication before hospital presentation, and 74% of these children (25/34) had abnormal echocardiograms on admission.
From the *Department of Pediatric Orthopedics, Starship Hospital, Park Road; wStarship Hospital; zKidz First Hospital, Auckland; zGisborne Hospital, Gisborne; and yDepartment of Biostatistics, University of Otago, Dunedin, New Zealand. The authors declare no conflicts of interest. Reprints: Raakhi M. Mistry, MBChB, Department of Pediatric Orthopedics, Starship Hospital, 2 Park Road, Auckland, New Zealand. E-mail: [email protected]. Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved.
318 | www.pedorthopaedics.com
Conclusions: ARF and septic arthritis are important diagnoses to consider in children presenting with acute joint pain in New Zealand. A significant proportion of patients with ARF initially present with acute monoarthritis. Serological inflammatory markers and WCC on presentation differ significantly between children with ARF and septic arthritis. Key Words: rheumatic fever, septic arthritis, monoarthritis (J Pediatr Orthop 2015;35:318–322)
J
oint pain and raised serological inflammatory markers are features of both acute rheumatic fever (ARF) and septic arthritis, often posing a diagnostic challenge to clinicians. Expeditiously distinguishing between these 2 diseases is of the utmost importance, as their clinical management is substantially different. Although ARF is managed symptomatically and does not typically require operative management, correct diagnosis is crucial to reduce the long-term cardiac morbidity. In contrast, septic arthritis is an orthopaedic emergency. The consequences of delayed diagnosis of septic arthritis include sepsis, cartilage damage, osteonecrosis, and growth arrest, making early operative drainage vital. Previous authors have reported on the value of serological inflammatory markers, in addition to other clinical features, in differentiating between transient synovitis of the hip and septic arthritis.1,2 In 1999, Kocher et al3 undertook a retrospective review of children with acute monoarthritis affecting the hip. They were able to develop a predictive algorithm on the basis of 4 variables: painful weight-bearing, history of fever, erythrocyte sedimentation rate (ESR) > 40, and white cell count (WCC) > 12. In 2006, Caird and colleagues similarly demonstrated that patients fulfilling Kocher’s criteria plus having a serological C-reactive protein (CRP) > 20 mg/L had a 98% chance of having septic arthritis. In addition, they found that a CRP > 20 mg/L was a strong independent risk factor for septic arthritis.1 The literature investigating the role of serological inflammatory markers in differentiating between ARF and septic arthritis is currently limited to small case series.4,5 One case series from Utah looked at 3 patients with ARF and 12 patients with septic arthritis presenting with monoarthritis.4 The serological inflammatory markers J Pediatr Orthop
Volume 35, Number 3, April/May 2015
J Pediatr Orthop
Volume 35, Number 3, April/May 2015
and WCC were elevated in both groups but there was no significant difference in values between the 2 groups. Although this study was limited by small sample size, it did demonstrate the important overlap in clinical and laboratory findings in ARF and septic arthritis. Another case series from Fiji looked at 3 patients with ARF who were initially misdiagnosed with septic arthritis despite having sterile joint fluid.5 The goal of our study was to determine the diagnostic value of serological inflammatory markers on presentation in differentiating between ARF and septic arthritis in children, in order to assist appropriate initial management.
METHODS Following institutional review board approval, clinical records of patients aged 16 years or younger with the final diagnosis of ARF or septic arthritis were identified. Patients with ARF were identified from the Auckland regional rheumatic fever database. ARF was defined on the basis of the New Zealand modification of the Jones Criteria.6,7 Each case of ARF was confirmed by an independent pediatrician. Patients with septic arthritis were identified from Starship Children’s Hospital and Middlemore Hospital’s computer records using the ICD10 code M00. Septic arthritis was defined as a positive synovial fluid on Gram staining, a positive synovial fluid culture, or a synovial fluid WCC > 50,000/mm3. Patients who fulfilled at least 1 of these criteria were defined as having definite septic arthritis. Patients who the laboratory reported as having a synovial WCC “too high to count” were defined as having probable septic arthritis. A power analysis indicated that 110 patients per group were required to detect a difference in ESR of 15% between the ARF group and the septic arthritis group as statistically significant (2-tailed a = 0.05) with 90% power. Consecutive hospital records from 2005 to 2012 were subsequently reviewed; hospital record review was concluded once >110 patients with confirmed diagnoses had been included in each patient group. The ARF group was compared with the septic arthritis group with respect to baseline characteristics [age, sex, ability to bear weight on affected limb, fever (presenting tympanic temperature >381C), use of nonsteroidal anti-inflammatory drug (NSAID) or antibiotic administration before presentation, and number of joints
Septic Arthritis and Acute Rheumatic Fever in Children
involved at presentation] and serum laboratory values on presentation (ESR, CRP, WCC). Blood culture and synovial aspirate results were also analyzed and compared between the 2 groups, where available.
Statistical Analyses Baseline demographics, clinical features, and serum inflammatory markers were compared between the ARF and septic arthritis groups using independent t tests and w2 or Fisher exact tests according to variable types. Synovial fluid WCC levels and the percentage of polymorphonucleocytes were compared between the groups using Mann-Whitney U tests. Within the ARF group, the serum inflammatory markers were compared between patients with single joint involvement and patients with multiple joint involvement using Mann-Whitney U tests.
RESULTS We identified 114 children with ARF and 111 children with acute septic arthritis (103 definite septic arthritis, 8 probable septic arthritis). The cases of ARF were identified from hospital records covering 3 years and the cases of septic arthritis were identified from records covering 6 years, within the 7-year study period.
Baseline Information At the time of hospital presentation, the ARF patients were significantly older (mean age 10.4 vs. 5.2 y, P < 0.001) and were less often febrile (51.8% vs. 73.9%, P < 0.001) than the septic arthritis patients (Table 1). There was no significant difference in sex (59% vs. 64% of patients male, P = 0.399), or inability to bear weight on the affected limb (97.4% vs. 100% of patients, P = 0.85), between the 2 patient groups. Before hospital presentation, ARF patients received antibiotics (19.3% vs. 4.5%, P < 0.001) or NSAID medication (10.5% vs. 1%, P = 0.002) more often than septic arthritis patients. The majority of ARF patients presenting with monoarthritis did not receive NSAID medication before their admission (86%, 29/34).
Serum Inflammatory Markers At the time of hospital presentation, the ARF patients displayed a significantly higher ESR (median 95.0 vs. 50.0 mm/h, P < 0.001), a significantly higher CRP (median 92.0 vs. 52.5 mg/L, P = 0.003), and a significantly lower WCC (mean 10.8 109 vs. 13.1 109/L, P < 0.001) than the
TABLE 1. Features on Presentation Variable Mean age (y) Male sex [n (%)] Inability to bear weight [n (%)] Temperature > 381C [n (%)] Monoarthritis at presentation [n (%)] Polyarthritis at presentation [n (%)] History of NSAID use [n (%)] History of antibiotic use [n (%)]
Septic Arthritis (n = 111)
Rheumatic Fever (n = 119)
P
5.2 (range, 6 wk-14 y) 73 (64) 111 (100) 82 (73.9) 109 (98.2) 2 (1.8)w 1 (0.9) 5 (4.52)
10.4 (range, 4-16 y) 65 (59) 111 (97.4) 59 (51.8) 34 (29.8)* 80 (70.2) 12 (10.5) 22 (19.3)
< 0.001 0.399 0.85 < 0.001 < 0.001 < 0.001 0.002 < 0.001
*14% (5/34) history of prior NSAID use. wMicroorganisms cultured—methicillin-sensitive S. aureus; S. pyogenes.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.pedorthopaedics.com |
319
J Pediatr Orthop
Mistry et al
Volume 35, Number 3, April/May 2015
TABLE 2. Inflammatory Markers Variable
Septic Arthritis (n = 111)
ESR (mm/h) CRP (mg/L) WCC ( 109/L)
53.2 (range, 1-138) 87.4 (range, 0-412) 14.5 (range, 4.53-78)
septic arthritis patients (Table 2). Within the ARF group, there was no significant difference in ESR (P = 0.309), CRP (P = 0.151), or WCC (P = 0.872) between patients with single joint involvement and patients with multiple joint involvement on presentation.
Monoarthritis At the time of hospital presentation, there were more patients presenting with monoarthritis in the septic arthritis group (n = 109) than in the ARF group (n = 34). The ARF patients with monoarthritis were older than the septic arthritis patients with monoarthritis (mean 9.6 vs. 5.2 y, P < 0.0001). The ARF patients with monoarthritis had a higher ESR than the septic arthritis patients with monoarthritis (mean 84.9 vs. 52.4, P < 0.0001). There was no significant difference between the 2 groups with respect to CRP and WCC (Table 3). Twenty-five of 34 of patients with ARF had echo changes on admission.
Rheumatic Fever (n = 119)
P
90.2 (range, 10-140) 93.8 (range, 1-239) 12.2 (range, 4.7-136)
< 0.001 0.003 < 0.001
Independent Risk Factors for Final Diagnosis To determine which of the measures identified by univariate analyses as showing statistically significant associations independently contributed to the discrimination of ARF and septic arthritis, we performed forward and backward stepwise logistic regression analyses. These analyses identified age, serum ESR, and serum WCC as 3 independent factors. In an effort to create amenable scores to discriminate between ARF and septic arthritis, we further defined the 3 risk factors; patient age above 8.5 years, serum ESR > 64.5, and serum WCC < 12.1 109/L. If a patient displayed all 3 of these risk factors or presented with polyarthritis there was a 94% chance of ARF as the final diagnosis. If polyarthritic patients were excluded, a patient presenting with monoarthritis in addition to 1, 2, or 3 other risk factors had a 17%, 38%, or 71% likelihood of ARF (and an 83%, 62%, or 29% likelihood of septic arthritis) as the final diagnosis, respectively (Table 5).
Synovial Fluid Analysis The ARF patients underwent joint aspiration much less frequently than the septic arthritis patients (5.3% vs. 100% of patients, P < 0.001) on presentation to hospital. Of the patients who underwent aspiration, ARF patients displayed a significantly lower synovial fluid WCC (median 20.2 vs. 102.5/mm3, P = 0.004) than the septic arthritis patients; however, there was no significant difference in the percentage of polymorphonucleocytes (median 63.5% vs. 79.0% of synovial fluid WCC, P = 0.761) between the 2 groups. With respect to synovial fluid culture in the septic arthritis group, the most commonly cultured organisms were methicillin-sensitive Staphylococcus aureus (25.2% of patients), Streptococcus pneumoniae (7.2%), K. kingae (7.2%), S. pyogenes (6.3%), and methicillin-resistant S. aureus (2.7%) (Table 3). With respect to synovial fluid culture in the septic arthritis group in children aged below 2 years, the most commonly cultured organisms were K. kingae (18.1%), S. pyogenes (9.1%), S. pneumoniae (9.1%), methicillin-sensitive S. aureus (6.8%), Escherichia coli (4.5%), and S. agalactiae (4.5%); 31.8% of these patients had no growth from their synovial culture (Table 4).
DISCUSSION Accurate diagnosis of the child presenting with acute joint pain is essential. After radiographic abnormalities have been ruled out, the remaining differential diagnoses can pose a clinical challenge, with the most common diagnoses including transient synovitis, septic arthritis, peri-articular osteomyelitis, and inflammatory arthritis. Diagnosis and treatment of septic arthritis is time critical and therefore a high index of suspicion is important. Distinguishing between septic arthritis and transient synovitis in children has previously been well investigated,1,3 with the use of Kocher’s criteria well established in clinical practice. In the New Zealand context, ARF is an important diagnosis to consider in a child presenting with arthritis.
Polyarthritis Polyarthritis is one of the major Jones criteria for the diagnosis of ARF, with joint pain in ARF patients tending to be migratory and predominantly involving the large joints.7–9 In our study the majority of children with ARF presented with polyarthritis (70.2%, n = 80). Only a small proportion of children with septic arthritis presented with polyarthritis (1.8%, n = 2). With a sample
TABLE 3. Monoarthritis Variable
Septic Arthritis (n = 109)
ESR (mm/h) CRP (mg/L) WCC ( 109/L) Age (y)
320 | www.pedorthopaedics.com
52.4 82.9 14.4 5.2
(range, (range, (range, (range,
Rheumatic Fever (n = 34)
1-138) 0-412) 0-78) 0-14)
84.9 83.7 11.2 9.6
Copyright
r
(range, (range, (range, (range,
17-133) 1-239) 5.9-18.3) 6-13)
P < 0.001 0.968 0.026 < 0.001
2014 Wolters Kluwer Health, Inc. All rights reserved.
J Pediatr Orthop
Volume 35, Number 3, April/May 2015
TABLE 4. Most Common Organisms Cultured in the Septic Arthritis Group Microorganism
% (n = 111)
Methicillin-sensitive S. aureus S. pneumoniae K. kingae S. pyogenes Methicillin-resistant S. aureus
24.3 7.2 7.2 6.3 2.7
(n = 27) (n = 8) (n = 8) (n = 7) (n = 3)
size of 2 it is difficult to draw any significant comparative differences between the 2 groups; however, we did note that both children in the septic arthritis group presented with raised pulse rate, fever, elevated serum inflammatory markers and notably a CRP > 300, and positive Gram staining on joint aspirate.
Monoarthritis This study has confirmed that ARF can present early with monoarthritis. Earlier studies have identified monoarthritis as a potential presenting feature of ARF.2,4,10,11 One retrospective review from Brazil looked at 109 patients with ARF; they found that 3.3% of their 92 cases with arthritis presented with monoarthritis.12 Another retrospective review looked at 367 patients with ARF in Australia’s Northern Territory and found that 17% of the confirmed ARF cases presented with acute monoarthritis.3,8 Another study from New Zealand showed that the incidence of monoarthritis in ARF was 6% (5/79); however, the authors did note that all of the monoarthritic cases had received NSAIDs before their presentation.13 In our study, 30% of patients with ARF presented with monoarthritis. When these patients with monoarthritis were compared with the septic arthritis patients with monoarthritis, we were able to demonstrate significant differences in age and ESR between these 2 groups. It is interesting to note that 14% (5/34) of these patients received NSAIDs before their presentation. Although it is thought that pretreatment with NSAID therapy may halt the progression of monoarthritis to polyarthritis in ARF patients,5 it is more likely that the patients presenting with monoarthritis in our study may have been in the early phases of ARF disease development. Nevertheless, these findings highlight the importance of including ARF in the differential diagnosis of all children presenting with monoarthritis. TABLE 5. Predicted Probability of ARF Versus Septic Arthritis in Patients Presenting With Monoarthritis
No. Predictors* 1 2 3
Predicted Probability of ARF (%)
Predicted Probability of Septic Arthritis (%)
17 38 71
83 62 29
*Logistic regression analysis determined 3 independent predictors to discriminate between ARF and septic arthritis: patient age > 8.5 years, serum ESR > 64.5, and serum WCC < 12.1 109/L.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
Septic Arthritis and Acute Rheumatic Fever in Children
Laboratory Variables The literature investigating the diagnostic value of serological inflammatory markers and WCC in differentiating between ARF and septic arthritis is limited.4,5,9 Although previous studies have been limited by small sample size, they identify an important overlap in the clinical features of these 2 conditions. With larger patient numbers we were able to identify clear differences in laboratory results between these 2 patient groups, with a significantly higher ESR, higher CRP, and lower WCC seen in the ARF group compared with the septic arthritis group. Our findings are consistent with our hypothesis that septic arthritis has a more acute presentation, with a shorter period of inflammation, and thus may display lower serological inflammatory markers on presentation. In contrast ARF tends to have a more indolent presentation, with a longer duration of inflammation, and thus typically presents with higher serological inflammatory markers. Children with suspected ARF require prompt, appropriate investigation including streptococcal serology tests and echocardiography. In our study 5.3% of patients with a final diagnosis of ARF underwent joint washout for presumed septic arthritis. While delaying the diagnosis of septic arthritis is hugely detrimental, care must be taken not to significantly delay the diagnosis of ARF in affected patients. Delayed diagnosis of ARF can potentially result in progression of cardiac disease.14 One study showed that misdiagnosis on initial presentation resulted in progression of rheumatic heart disease.5 In this study, the authors reported a case series of 3 children from Fiji who presented with arthritis, fever, and elevated inflammatory markers. One child presented with monoarthritis and the other 2 presented with polyarthritis (2 joints involved). Joint aspirates were performed in all cases and all were negative. The children were commenced on antibiotics and subsequently discharged home. All 3 children returned 1 to 3 years later with established moderate to severe rheumatic heart disease. When we applied our predictive algorithm to the child with monoarthritis—an 8-year-old boy; ESR 75; and normal WCC—our predicted probability of ARF was 71%. Combined with a negative joint aspirate, this should alert the clinician to undertake further investigation.
Epidemiology Quality epidemiological data on the global incidence of ARF is lacking. Despite this there is a clear trend in developed countries showing a decline in incidence.15 In New Zealand, however, the incidence has remained high.13,14 A recent review of hospitalizations (not epidemiological survey) showed that the incidence in New Zealand was 17.2/100,000.16 It is well established that the burden of ARF is carried predominantly by minority groups within a given area who are of a lower socioeconomic status.13 In New Zealand this has been clearly demonstrated in indigenous Maori and New Zealand-born Pacific Island people. A study by Milne et al17 found that the incidence of ARF in school-age www.pedorthopaedics.com |
321
J Pediatr Orthop
Mistry et al
Pacific Island children in New Zealand is 80 to 100 per 100,000. In New Zealand where rheumatic fever is highly endemic, awareness of the clinical features of this debilitating disease is clearly essential for clinicians evaluating children with acute monoarthritis. This paper has shown that it is not uncommon for children with ARF to present to an orthopaedic service with acute monoarthritis. Raised awareness among clinicians in New Zealand has resulted in these children being referred for a pediatric medical opinion and/or an echocardiogram. Historically a number of severe cases of rheumatic heart disease have occurred in patients in whom the diagnosis was first missed.
Echocardiography In children suspected of ARF, echocardiogram is the investigation of choice. If the initial echocardiogram is normal, the standard protocol is to repeat it within 2 to 3 weeks.18 In our study when we reviewed the echocardiograms of the 34 children with ARF who presented with acute monoarthritis we found that 9/34 had normal echocardiograms on admission. Of those 9 normal echocardiogram results, 5 were repeated. Of these, 1 was normal and 4 had developed mild aortic or mitral regurgitation. Twenty-five of 34 children had abnormal echocardiograms on admission, 6 were severe, 13 were moderate, and 6 were mild. Thus, if heart disease was already established NSAID agents were used rather than waiting for polyarthritis to ensue. In our study, we found that a child above 8.5 years of age presenting with monoarthritis and a serological ESR > 64.5 and WCC < 12.1 had a 71% chance of having ARF. In our population correspondingly, we feel we have provided strong evidence to consider the diagnosis of ARF in all children presenting with monoarthritis, but particularly in older children with an elevated serological ESR and a low serum WCC.
CONCLUSIONS ARF and septic arthritis are important diagnoses to consider in children presenting with acute joint pain. Our results suggest that serological inflammatory markers and WCC on presentation differ significantly between children with ARF and septic arthritis. Contrary to traditional teaching, we have found that a significant proportion of
322 | www.pedorthopaedics.com
Volume 35, Number 3, April/May 2015
patients with ARF initially present with acute monoarthritis. REFERENCES 1. Caird MS, Flynn JM, Leung YL, et al. Factors distinguishing septic arthritis from transient synovitis of the hip in children. A prospective study. J Bone Joint Surg Am. 2006;88:1251–1257. 2. Carapetis JR, Currie BJ. Rheumatic fever in a high incidence population: the importance of monoarthritis and low grade fever. Arch Dis Child. 2001;85:223–227. 3. Kocher MS, Zurakowski D, Kasser JR. Differentiating between septic arthritis and transient synovitis of the hip in children: an evidence-based clinical prediction algorithm. J Bone Joint Surg Am. 1999;81:1662–1670. 4. Harlan GA, Tani LY, Byington CL. Rheumatic fever presenting as monoarticular arthritis. Pediatr Infect Dis J. 2006;25:743–746. 5. Mataika R, Carapetis JR, Kado J, et al. Acute rheumatic fever: an important differential diagnosis of septic arthritis. J Trop Pediatr. 2008;54:205–207. 6. New Zealand Heart Foundation. New Zealand Modification of the Jones Criteria. Auckland: New Zealand Heart Foundation; 2013. 7. Guidelines for the diagnosis of rheumatic fever. Jones Criteria, 1992 Update. Special Writing Group of the Committee on Rheumatic Fever, Endocarditis, and Kawasaki Disease of the Council on Cardiovascular Disease in the Young of the American Heart Association. JAMA. 1992;268:2069–73. [Erratum appears in JAMA, 1993, 27;269(4):476]. 8. Long S. Principles and Practice of Pediatric Infectious Diseases. Philadelphia, PA: Elsevier; 2012. 9. The National Heart Foundation of New Zealand. A Summary of the New Zealand Guidelines for Rheumatic Fever. Auckland, New Zealand: The National Heart Foundation of New Zealand; 2010. 10. Khriesat I, Najada AH. Acute rheumatic fever without early carditis: an atypical clinical presentation. Eur J Pediatr. 2003;162:868–871. 11. Wilson E, Wilson N, Voss L, et al. Monoarthritis in rheumatic fever? Pediatr Infect Dis J. 2007;26:369–370. 12. Pileggi GC. Atypical arthritis in children with rheumatic fever. J Pediatr (Rio J). 2000;76:49–54. 13. Harrison GJ, Cherry JD, Kaplan SL, et al. Feigin and Cherry’s Textbook of Pediatric Infectious Diseases. Vol. 7th Edition, Chapter 29 - Acute Rheumatic Fever. Philadelphia: Saunders; 2014. 14. Carapetis JR, McDonald M, Wilson NJ. Acute rheumatic fever. Lancet. 2005;366:155–168. 15. Miyake CY, Gauvreau K, Tani LY, et al. Characteristics of children discharged from hospitals in the United States in 2000 with the diagnosis of acute rheumatic fever. Pediatrics. 2007;120:503–508. 16. Milne RJ, Lennon DR, Stewart JM, et al. Incidence of acute rheumatic fever in New Zealand children and youth. J Paediatr Child Health. 2012;48:685–691. 17. Milne RJ, Lennon DR, Stewart JM, et al. Incidence of acute rheumatic fever in New Zealand children and youth. J Paediatr Child Health. 2011;48:685–691. 18. Abernethy M, Bass N, Sharpe N, et al. Doppler echocardiography and the early diagnosis of carditis in acute rheumatic fever. Aust N Z J Med. 1994;24:530–535.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
