Article Text

PDF

Aetiology of arthritis in hospitalised children: an observational study
  1. Camille Aupiais1,2,3,
  2. Brice Ilharreborde3,4,
  3. Catherine Doit5,
  4. Audrey Blachier6,
  5. Marie Desmarest7,
  6. Chantal Job-Deslandre8,9,
  7. Keyvan Mazda3,4,
  8. Albert Faye2,3,10,
  9. Stéphane Bonacorsi3,5,
  10. Corinne Alberti1,2,3,
  11. Mathie Lorrot2,3,10
  1. 1Unité d'Epidémiologie clinique, AP-HP, Hôpital Robert Debré, Paris, France
  2. 2Inserm, U1123, Paris, France
  3. 3Univ Denis Diderot Paris 7, Sorbonne Paris Cité, Paris, France
  4. 4Service d'orthopédie pédiatrique, AP-HP, Hôpital Robert Debré, Paris, France
  5. 5Service de Microbiologie, AP-HP, Hôpital Robert Debré, Paris, France
  6. 6Département informatique médicale, AP-HP, Hôpital Robert Debré, Paris, France
  7. 7Service d'Accueil des Urgences pédiatriques, AP-HP, Hôpital Robert Debré, Paris, France
  8. 8Service de rhumatologie, AP-HP, Hôpital Cochin, Paris, France
  9. 9Université René Descartes Paris 5, Paris, France
  10. 10Service de Pédiatrie générale, AP-HP, Hôpital Robert Debré, Paris, France
  1. Correspondence to Dr Camille Aupiais, Unité d'Epidémiologie Clinique, Hôpital Robert Debré (APHP), 48 boulevard Sérurier, Paris 75935, Cedex 19, France; camille.aupiais{at}rdb.aphp.fr

Abstract

Background and objective Arthritis in children has many causes and includes septic and viral arthritis, reactive arthritis and juvenile idiopathic arthritis (JIA). We aimed to describe the different types of arthritis among children hospitalised for a first episode of arthritis.

Design Retrospective, descriptive case series study.

Setting A French tertiary care centre.

Patients Children under 16 years of age hospitalised for an arthritis episode between 1 January 2008 and 31 December 2009.

Main outcome measures Demographic and clinical features were compared with χ2 or Fisher's exact tests and non-parametric tests.

Results 173 children were hospitalised for a first episode of arthritis during the study period, with a male/female ratio of 1.14. The most frequent cause of hospitalisation was septic arthritis (43.4% of cases, 69.3% of which were due to Kingella kingae and 10.7% to Staphylococcus aureus). JIA was responsible for 8.1% of cases and arthritis without any definitive diagnosis for 40.4%. Median age at diagnosis was 2.7 years (IQR 0.3–14.6) and was lower in the septic arthritis group (1.5 years; 1.1–3.4) than in the JIA group (4.7 years; 2.5–10.9) (p<0.01). Septic arthritis involved a single joint in 97.3% of cases, while JIA involved four joints in 14.3% of cases and two to four joints in 28.6% of cases (p<0.01).

Conclusions Septic arthritis was the most frequent cause of arthritis in hospitalised children. Despite the increasing application of microbiological molecular methods to synovial fluid analysis, further measures are required to improve the diagnosis of arthritis of unknown cause.

Statistics from Altmetric.com

What is already known on this topic

  • The epidemiological profile of arthritis has been described in two population-based studies; transient synovitis is presumed to be the commonest cause of arthritis during childhood.

  • Recent developments in molecular methods have considerably improved the diagnosis of septic arthritis in childhood.

  • Children with septic arthritis require early diagnosis and hospital treatment so that septicaemia, growth problems and joint damage can be avoided.

What this study adds

  • The epidemiological profile of arthritis among hospitalised children has not been previously described; we found that septic arthritis was the most frequent cause.

  • Despite the increasing use of microbiological molecular methods for synovial fluid analysis, further measures are required to improve the diagnosis of arthritis of unknown cause.

Introduction

Arthritis is defined as inflammation of the joint. Arthritis may be caused by transient synovitis, juvenile idiopathic arthritis (JIA), reactive arthritis, viral arthritis and other inflammatory diseases; septic arthritis also occurs as a complication of bacteraemia in children. The overall incidence of arthritis and its causes are not well documented in children, and have even been reported to be unknown.1 ,2 Only two population-based studies have described the incidence of different diagnoses of arthritis in childhood.3 ,4 Both of these studies were performed in Northern Europe and reported that transient synovitis is the most common cause.

The outcome and treatment of arthritis depend on its aetiology. Children with transient synovitis do not need any surgical procedures and are treated symptomatically at home. However, children with septic arthritis require early diagnosis and hospital treatment, including surgical joint drainage and intravenous antibiotics so that septicaemia and joint damage can be avoided.5–8

The epidemiological profile of arthritis among hospitalised children has not been previously described. Moreover, recent developments in molecular methods have greatly improved the diagnosis of septic arthritis in childhood.9–12 The aim of this retrospective study was to describe children hospitalised for a first episode of arthritis in a paediatric tertiary care centre over a 2-year period.

Methods

Study population and criteria for inclusion

Children under the age of 16 years, hospitalised for arthritis between 1 January 2008 and 31 December 2009 in a French tertiary referral centre for paediatric orthopaedics and rheumatology, were retrospectively included. The French national hospital discharge database (PMSI, Programme de médicalisation des systèmes d’information) was used to identify children discharged from hospital with a diagnosis of arthritis according to the International Classification of Diseases, Tenth Revision (ICD-10) (table 1).13 Moreover, to ensure that cases of arthritis were not missed due to coding errors, we also included all children who underwent joint aspiration by checking the records of synovial fluid stored by the microbiology laboratory. Data were retrospectively collected. All patient records were examined to confirm the diagnosis of arthritis, defined as joint pain and/or functional disability in association with clinical and/or radiological joint effusion. Patients with arthritis occurring after open wound or bone surgery were excluded from analysis. The study was approved by the Institutional Ethics Committee and the French national data protection agency (Commission nationale de l'informatique et des libertés).

Table 1

International Classification of Diseases, Tenth Revision (ICD-10): diagnosis codes for arthritis

Microbiological methods

In our centre, all children with a first episode of arthritis associated with fever, high C-reactive protein levels or high white blood cell counts, were suspected of having septic arthritis and underwent joint aspiration. A blood sample was taken before the intervention and stored in an aerobic blood culture vial, while aspirated joint fluid was immediately stored in aerobic blood culture bottles after the intervention. The blood culture vials were incubated in a continually monitored instrument (BacT/Alert 3D; BioMérieux) and no blind subcultures were performed. The remainder of the joint fluid sample was sent to the laboratory for Gram staining, cell counting, and immediate inoculation onto Columbia blood agar (incubated in anaerobic conditions), chocolate agar (incubated in CO2-enriched air) and brain-heart broth. Aliquots (100–200 mm3) were stored at 80°C for DNA extraction. Blood culture bottles and the other media were incubated for 5 and 10 days, respectively. When sufficient sample was available, 100–200 mm3 of plasma was separated from blood samples and stored at 80°C for DNA extraction. Kingella kingae was identified based on microbiological characteristics.14 K. kingae DNA was detected by a real-time PCR-based method for children under 6 years of age, as previously described.10

Diagnostic classification

After collecting all the data, we classified children according to clinical and biological information:

  • Septic arthritis was defined as arthritis associated with bacteria isolated from blood or synovial fluid, as described above. Serological testing for viruses or Lyme disease was not routinely carried out.

  • Patients were classified as having JIA according to the diagnosis of the rheumatologist paediatrician. This diagnosis was assumed to follow the International League of Associations for Rheumatology (ILAR) classification of JIA, that is, arthritis of unknown aetiology that begins before the 16th birthday and persists for at least 6 weeks.15

  • Arthritis that did not retrospectively fulfil criteria for septic arthritis, JIA or other well-established diagnoses (haemarthrosis, Kawasaki disease, malignant disease, Gaucher disease, systemic lupus erythematosus, villo-nodular synovitis, dermatomyositis and chronic recurrent multifocal osteitis) was classified as ‘arthritis without any definitive diagnosis’.

Study variables

Demographic and clinical characteristics were retrospectively recorded on a standardised form. The number of joints initially involved was classified as follows: monoarthritis, involving one joint; oligoarthritis, involving between two and four joints; and polyarthritis, involving five or more joints.

Statistical analysis

Characteristics were described as frequencies or the median. Categorical variables were compared between groups with the χ2 test or Fisher's exact test, and continuous variables were compared with non-parametric tests (Mann–Whitney–Wilcoxon for comparisons involving two groups and Kruskal–Wallis for comparisons involving three groups). The chosen level of significance was p<0.05. Statistical analyses were performed with SAS statistical software (V.9.3; SAS Institute).

Results

Children with arthritis, aetiology and patient characteristics

Over the 2-year study period, 26 493 patients under 16 years of age were admitted to our centre. Of these patients, 188 (0.7%) were hospitalised for arthritis, including 173 first episodes (92.0%) and 15 recurrences.

The aetiology of the 173 first episodes of arthritis is presented table 2. The sex ratio was 1.14. Nine of the 173 children had underlying disease: four children had sickle cell disease (two were diagnosed with Salmonella infection and two had no definitive diagnosis), three children had arthritis without any definitive diagnosis after bone marrow transplantation, one had septic arthritis due to Staphylococcus aureus in the context of chronic renal failure, and one had septic arthritis due to Streptococcus pneumoniae in the context of hypoxic ischaemic encephalopathy.

Table 2

Patients hospitalised for a first episode of arthritis in 2008–2009: aetiology and patient characteristics

In children under 2 years of age, the most common diagnosis was septic arthritis (64.8%), while in children above 2 years of age, it was arthritis without any definitive diagnosis (53.5% of arthritis cases in children between 2 and 6 years of age and 44.2% in children above 6 years of age) (figure 1).

Figure 1

Distribution of diagnoses according to age in children hospitalised for a first episode of arthritis in 2008–2009. *Arthritis without any definitive diagnosis indicates arthritis with no established association with infection and that did not fulfil the classification criteria for JIA or other well-established diagnoses (haemarthrosis, Kawasaki disease, malignant disease, Gaucher disease, systemic lupus erythematosus, villo-nodular synovitis, dermatomyositis and chronic recurrent multifocal osteitis). JIA, juvenile idiopathic arthritis.

Children with septic arthritis

The median age was significantly lower in the group with septic arthritis compared to the JIA group (p<0.01; table 2) and 61.3% of children with septic arthritis were under 2 years of age.

K. kingae was the most common pathogen found in septic arthritis, and accounted for 69.3% of cases (n=52), followed by S. aureus in 10.7% of cases (n=8), and group A Streptococcus in 9.3% of cases (n=7). Other agents were S. pneumoniae (n=2), Escherichia coli (n=2), Haemophilus spp, Salmonella, serogroup B Meningococcus and serogroup B Streptococcus (one case of each).

Arthritis due to K. kingae occurred throughout the year, with a peak in October and a nadir in July and August.

Children with JIA

The sex ratio was different among the various subgroups of arthritis (p=0.04; table 2); in particular, two thirds of JIA patients were female. JIA exclusively occurred in children above 18 months of age.

Joint involvement in the subgroups of arthritis

The most frequently affected joints were the knee and the hip (including six cases with bilateral involvement) (table 3). Only one case of JIA initially affected the hip (figure 2). The two cases of septic arthritis involving more than one joint occurred during S. aureus septicaemia in a healthy child and during Salmonella septicaemia in a child with sickle cell disease. The joints involved also differed according to the causative organism: the knee was affected in 65.4% of cases of arthritis due to K. kingae, in comparison to other organisms which affected the hip in 39.1% and the ankle in 26.1% of cases (p<0.01).

Table 3

Children hospitalised for a first episode of arthritis in 2008–2009: joint involvement at first hospitalisation

Figure 2

Distribution of diagnoses according to the joint involved in children hospitalised for a first episode of arthritis in 2008–2009. *Arthritis without any definitive diagnosis indicates arthritis with no established association with infection and that did not fulfil the classification criteria for JIA or other well-established diagnoses (haemarthrosis, Kawasaki disease, malignant disease, Gaucher disease, systemic lupus erythematosus, villo-nodular synovitis, dermatomyositis and chronic recurrent multifocal osteitis). **Other joints: finger or foot joints (six cases), shoulder (four cases), elbow (three cases), wrist (one case), association of many of the above-mentioned joints (13 cases). JIA, juvenile idiopathic arthritis.

Children without any definitive diagnosis

Of the 70 children classified as ‘arthritis without any definitive diagnosis’, 85.7% underwent joint aspiration (n=60). K. kingae-specific real-time PCR was negative in 71.7% (n=43) of these 60 synovial fluid samples, and was not carried out for the remaining 17. Finally, at time of discharge, 53 (75.7%) patients were assumed to have septic arthritis (although the bacteria could not be identified), nine (12.9%) viral arthritis, and eight (11.4%) could not be diagnosed.

Length and service of care

The length of hospital stay is shown in table 2. Of the 173 children, 139 (80.3%) were managed under orthopaedic services, 25 (14.5%) under rheumatology services, and nine under other services (intensive care unit, neonatal care unit or another specialised paediatric unit).

Discussion

The annual incidence rate of septic arthritis in France was evaluated as 7.1 per 100 000 children under 16 years of age in 2008–2009,16 while that of JIA ranges between 1.6 and 3.2 per 100 000 children.17 This is the first study to describe all children hospitalised for a first episode of arthritis regardless of the cause of arthritis.

We enrolled 173 children during a 2-year period, corresponding to 0.7% of all patients hospitalised at our centre. Septic arthritis was the most frequent cause of hospitalisation, accounting for 43% of all cases, and 65% of cases in children under 2 years of age. Three population-based prospective studies have described the annual incidence of arthritis in children in Northern Europe.3 ,4 ,18 One of these studies focussed on JIA and did not describe septic arthritis.18 The two other studies found transient synovitis to be the most common diagnosis (47.8%3 and 60.5%4). Septic arthritis and transient synovitis can be differentiated using clinical and biological criteria19–21 and are managed differently due to the risk of general or local dissemination and joint sequelae in septic arthritis.1 ,22 Septic arthritis requires emergency hospitalisation for joint drainage and antibiotics,5–7 while transient synovitis can be treated with rest and painkillers.23 Thus, children with transient synovitis were not included in our hospital-based study, because they do not need hospitalisation and can be managed by secondary care clinicians. Therefore, we found a higher proportion of children with septic arthritis than reported in previous studies (43.3% vs 6.2%3 and 7.0%4). This reveals the burden of septic arthritis in children because these individuals are routinely hospitalised. We found that the median duration of stay was 7 days for children with septic arthritis, which is consistent with the results of a French study which found that the mean length of hospital stay for paediatric bone and joint infections was 8.4 days, costing approximately €5200 per stay.24

S. aureus has been historically described as the most common agent of osteoarticular infection in children.1 ,16 ,25–27 However, following the development of molecular techniques, more recent studies have described K. kingae, a Gram-negative coccobacillus, as the primary pathogen of arthritis, especially in children under 4 years of age.9 ,10 ,12 ,28 In our study, the most common pathogen among 75 children with proven septic arthritis was K. kingae (69% of cases). This is higher than reported in two previous French studies that found K. kingae in 45% of osteoarticular infections9 and 52% of cases of septic arthritis.10 This latter study was carried out in our centre between 2006 and 2008. Differences between the findings of this study and the current study may be explained by the more recent systematic use of molecular techniques; nonetheless, these differences show that changes in the epidemiological profile of arthritis in children should be followed carefully. Consistent with reports in the literature, we found that septic arthritis due to K. kingae occurred in children under 4 years of age,9 ,29–31 showed seasonal variability14 ,30 ,32 and mostly affected the knee.9 ,10 ,14 K. kingae has become the most common microbial agent of osteoarticular infections in young children, followed by S. aureus. Thus, the empiric choice of initial antibiotics in septic arthritis in children must take into consideration the prevalence of S. aureus and K. kingae, their respective antibiotic sensitivity profiles, their pharmacokinetic and pharmacodynamic properties, and the results of clinical studies. K. kingae is highly susceptible to β-lactam antibiotics. K. kingae arthritis diagnosed by PCR may be treated with a β-lactam drug such as amoxicillin, which allows intravenous antibiotic therapy to be changed to oral antibiotic therapy.10 JIA frequently involved two or more joints, whereas all septic arthritis cases affected only one joint, with the exception of two multifocal arthritis cases secondary to septicaemia.

Children in the JIA group were older than children with septic arthritis, and were all at least 18 months old at the time of the first episode of arthritis, with a median age of 4.7 years. This is consistent with previous studies showing that the incidence of JIA peaks between 2 and 4 years of age, and occurs earlier in girls than in boys.3 ,4 ,17 ,33–35 We found that 71% of children with JIA were female, consistent with previous data about particular types of JIA.4 ,17 ,18 ,36 ,37

The second largest arthritis group consisted of children without any definitive diagnosis, defined as those who had no signs of septic arthritis and who did not fulfil the classification criteria for JIA or other well-established diagnoses. This classification was applied at the time of the study; therefore, in some cases, this a posteriori diagnosis differed from the diagnosis given at time of discharge. At discharge, many children were considered to have septic arthritis without any biological evidence of bacterial infection. This has been described in the literature as culture-negative septic arthritis.38 However, these children could also have had transient synovitis, reactive arthritis, or viral or other bacterial infections. Children with arthritis without any definitive cause were older than those with septic arthritis, and the condition was more common in boys (63%) and preferentially affected the hip (47%). This suggests that transient synovitis or septic arthritis may be involved because these conditions are believed to be more prevalent in boys.4 A large proportion of children without any definitive diagnosis underwent joint aspiration because it was suspected that they might have septic arthritis. Unfortunately, K. kingae-specific real-time PCR was carried out for only 68.3% of synovial fluid samples from children with arthritis of unknown cause. The diagnosis of arthritis in such cases may be improved by the systematic use of molecular biology techniques to detect bacteria in cultures for septic arthritis, and by the taking of detailed medical records and clinical descriptions to detect signs of reactive arthritis, in addition to oriented serological testing and the long-term follow-up of children.

Our study has several limitations. First, we only analysed children who were hospitalised in a single centre. Thus, our methodology did not enable us to estimate the incidence of the different subgroups of arthritis, and the general applicability of the results is limited. Furthermore, this was a retrospective study, and data from the PMSI may not be sufficiently reliable. A recent French study estimated that the positive predictive value of the osteoarticular infections recorded in the PMSI was 81%.24 Therefore, we verified that all children who underwent joint aspiration were included in our study by verifying records from the microbiology laboratory and we re-examined all files to confirm the diagnosis of arthritis. Nevertheless, these steps limit but do not prevent misdiagnosis. Some bias in diagnosis classification may also have existed, because not all children underwent the same exploratory procedures, such as joint aspiration. Moreover, this is a retrospective study and some information was not available, such as previous exposure to antibiotics.

Conclusions

Septic arthritis was the most frequent cause of arthritis in hospitalised children. Despite the development of molecular techniques, 40% of arthritis cases had no definitive diagnosis. Improvements are needed in the diagnosis of arthritis and comparative studies are required to describe the clinical and biological characteristics of the various causes of this condition.

References

View Abstract

Footnotes

  • Contributors Study concept and design: CAu, AF and ML; acquisition of data: CAu, AB, MD and SB; analysis and interpretation of data: CAu, BI, CJ-D, KM, AF, SB, CAl and ML; drafting of the manuscript: CAu, CD and ML; revision of the manuscript: CAu, BI, CD, AB, MD, CJ-D, KM, AF, SB, CAl and ML; statistical analysis: CAu and CAl.

  • Competing interests None

  • Ethics approval The study was approved by the Institutional Review Board of Paris-North Hospitals, Paris 7 University, AP-HP (IRB no. 2013-84, 17 September 2013) and the French national data protection agency (Commission nationale de l'informatique et des libertés).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.