What is the clinical course of transient synovitis in children: a systematic review of the literature

Background Transient synovitis of the hip (TS) is considered to be a self-limiting disease in childhood. However, because the etiology is unclear and some cases precede Legg-Perthes’ disease, data on follow-up are important. Our aim was to summarize the knowledge on the clinical course of TS in children. Methods The study design was a systematic review and a literature search was conducted in Medline and Embase. Studies describing short and/or long-term follow-up of TS in children were included. Case reports, reviews and studies describing traumatic hip pain were excluded. Study quality was scored and data extraction was performed. The main outcome measures were short-term and long-term clinical course, and recurrence of symptoms. Results A total of 25 studies were included of which 14 were of high quality. At two-week follow-up, almost all children with TS were symptom free. Those with symptoms persisting for over one month were more prone to develop other hip pathology, such as Legg-Perthes’ disease. The recurrence rate of TS ranged from 0–26.3%. At long-term follow-up, 0-10% of the children diagnosed with TS developed Legg-Perthes’ disease. Hip pain after intensive physical effort and limited range of motion of the hip at long-term follow-up was reported in 12-28% and in 0-18% of the children, respectively. Conclusions The majority of the studies indicate that children with TS recover within two weeks; recurrence was seen in 0-26% of the cases. Children with TS should be followed at least six months to increase the likelihood of not missing Legg-Perthes’ disease.


Background
Transient synovitis of the hip (TS) in children is a relatively common disorder with an average annual incidence of 0.2% in the general population [1]. There is a male-to-female ratio of slightly more than 2:1. The disease typically presents at a mean age of 6 (3)(4)(5)(6)(7)(8) year [1].
TS is characterized by pain and limited motion of the hip [1]. This hip pain and limited range of movement is caused by reactive effusion, leading to a position of the hip in flexion and external rotation [2]. TS is confirmed by excluding potentially more severe causes for hip symptoms, such as septic arthritis, osteomyelitis, Legg-Perthes' disease, juvenile idiopathic arthritis, fractures and tumors [1]. These conditions can be verified by imaging and laboratory investigations [3]. However, the mainstay of diagnosis for TS remains that of clinical findings [3]. For detection of hip effusion, ultrasound is considered to be the best non-invasive technique [2]. The etiology of TS remains largely unknown. Infections of the upper respiratory tract may be a predisposing factor, although identifiable causes or agents have not yet been found [4]. Treatment generally consists of rest and anti-inflammatory agents [5].
In many countries, the majority of children with nontraumatic hip symptoms will initially present to the general practitioner (GP). Krul et al. [6] reported an incidence rate for TS in general practice of 76.2 per 100,000 personyears [6]. According to Vijlbrief et al. [7], GPs prefer to adopt a wait-and-see approach to TS rather than referring patients for additional diagnostics [7]. However, TS has been associated with the risk of developing other diseases, such as Legg-Perthes' disease [1].
Increased understanding of the short and long-term course of this disease will help medical doctors to better determine follow-up policy and better inform patients about the prognosis. Early recognition of subsequent disorders may also lead to timely referral and more optimal treatment. Therefore, this study presents a systematic review of the literature on TS, focusing on its short-term symptomatic course, long-term clinical course, and recurrences.

Searches
A literature search using Medline and Embase was performed from Januari 1966 (Medline) and Januari 1947 (Embase) till September 2013. Medical subject headings (mesh) terms and corresponding entry terms for the diagnosis, age group and outcome were used (Additional file 1: Table S1).

Study inclusion and exclusion criteria
Studies describing recurrence or clinical follow-up (either on the short or long-term) of TS in children were included. Studies describing children with traumatic hip problems were excluded, as were reviews and case reports. Randomized controlled trials, and cohort and casecontrol studies were included. Relevant articles in English, German, French, Swedish, Norwegian, Danish and Dutch were included. The abstract of the article had to be available in order to select relevant studies. First, two reviewers (SA, RvR) selected relevant articles by title and abstract. All articles selected by either reviewer were ordered in full text. One reviewer (SA) made the final selection by reading the full text of the selected articles. In case of doubt, a second reviewer (SB) read the full text to ensure that no studies were excluded unjustly.

Study quality assessment
Three reviewers assessed the quality of the selected articles independently. One reviewer (SA) assessed the quality of all studies and two reviewers (SB, JB) each assessed the quality of half of the studies. Seven criteria were used, which were rated as positive, negative or inconclusive. These criteria were based on the quality assessment used by van Rijn et al. [8] and adjusted for the purpose of investigating TS. Disagreements were resolved by consensus. The initial agreement was measured by using the Kappa value (k). Agreement is considered poor if k ≤ 0.20; fair if 0.21≤ k ≤ 0.40; moderate if 0.41 ≤ k ≤ 0.60; substantial if 0.61 ≤ k ≤ 0.80; good if k > 0.80 [9]. Summing the positive criteria, a quality score was calculated ranging from 0 to 7 points. Studies with five or more positive criteria were considered to be of 'high quality'.

Data extraction strategy
One reviewer (SA) performed data extraction for all the studies, which was subsequently checked by two other reviewers (RvR and MK) each checking half of the studies. Data extraction included number of patients with TS; characteristics of patients with TS (age, male/female ratio, primary or secondary care, duration of symptoms prior to admission); study design; tests done before diagnosing TS; treatment for TS; length of follow-up period; and outcomes (short-term symptomatic course, recurrence, longterm course). Disagreements were resolved by consensus.

Data synthesis and presentation
Included articles were classified in three groups based on their outcome: short-term symptomatic course, recurrence and/or long-term course. Short-term symptomatic course implied clinical follow-up < 3 months. Studies classified as recurrence reported a recurrence rate and/or clinical course of patients with recurrent TS. Long-term course was defined as clinical course of TS > 3 months; this also included the development of other diseases subsequent to TS.
All studies included patients treated for TS, of which three studies [17,29,30] had a specific target group. One study [30] focused on patients with recurrences of TS, defined as at least two separate episodes of TS at least two months apart. One study [17] included patients with recurrent or prolonged TS, defined as recurrence of symptoms within one month after initial resolution, or symptoms that lasted more than one month. Another study [29] compared patients with no recurrence of TS with patients who had one or more recurrence(s). Table 2 presents the methodological quality assessment of the included studies. Of the 25 studies, 15 were considered to be of high quality (i.e. 5 or more criteria were rated positive). The initial agreement of the total quality assessment was 84% (147 of 175 items). The kappa score calculated on the total quality assessment was 0.65, which is considered to be a substantial agreement. Critical items were 'same duration of follow-up' and 'follow-up data available from at least 80% of the study population' , which were present in 48% and 52% of the studies, respectively. All initial disagreements were resolved by discussion.
An overview of six studies [3,18,21,22,24,28], which exclusively reported the percentage of patients with symptoms on short-term follow-up, is shown in Figure 2. Of these, three are of high quality and three of low quality.
In the high quality studies, the length of symptoms (hip pain, limp, limited range of motion and/or refusal to bear weight), ranged from 2.5 days to 3 months [3,12,18,22]. In three of these studies [12,18,22], between 3-8% of patients had symptoms persisting for over one month.
In one high quality study [12], all nine patients with persisting symptoms for over one month developed Legg-Perthes' disease.
According to one high quality study by Terjesen et al. [22], 75% of the recurrences occur in the first six months after the initial episode of TS. Also, in 25% of the cases in this study the opposite hip was involved at recurrence.

Studies excluded -No clinical follow-up TS (n=5) -Did not meet language criterion (n=2) -Not available in Dutch or British library (n=1)
Studies included for systematic review (n=25) Figure 1 Flow chart of the selection process.     [ [24][25][26]29,30] showed that in 20-50% of the cases, the opposite hip is involved at recurrence.
Three high quality studies reported clinical follow-up of TS on the long-term. Limited range of motion was found in 7% to 8% of the patients with TS [15,16].
Ten high quality studies [4,10,[13][14][15][16][17]19,20,23] reported development of Legg-Perthes' disease subsequent to TS, ranging from 0% [15,16,19] to 38% [17]. In a study by by Kallio et al. [15] two patients with TS in one hip had recent symptom-free Legg-Perthes' disease in the opposite hip. Also, in this study all patients initially diagnosed with TS who developed Legg-Perthes' disease on longterm follow-up had early signs of Legg-Perthes' disease on the original radiographs. Early signs of Legg-Perthes' disease on radiographs were defined as signs of avascular necrosis and/or increased joint space.
Two high quality studies mentioned the long-term prevalence of coxa magna subsequent to TS ranging from 0% 23 to 32% [16].
Five low quality studies reported clinical follow-up of TS on the long-term. Hip pain after intensive physical effort ranged from 12% to 28% [21,26,30]. Limited range of motion of the hip was found in 0% to 18% of the patients with TS [21,28].
Two low quality studies reported that patients initially diagnosed with TS who developed Legg-Perthes' disease on long-term follow-up, had early signs of Legg-Perthes' disease on the initial radiographs in 67% [2] and 80% [1] of the cases.

Discussion
As no prognostic review of TS has yet been published, this systematic review is the first to provide an overview of the short and long-term clinical course of children with TS. Both high and low quality studies show that most children with TS experience symptoms for less than two weeks. Recurrence ranged from 0% to 18.3% in the high quality studies and from 4.1% to 26.3% in the low quality studies. In 20% to 50% of the children with a second episode of TS the opposite hip is affected. On long-term follow-up of high quality studies, limited range of motion was found in 7% to 8% of children with TS. Legg-Perthes' disease subsequent to TS, which was reported most frequently on long-term follow-up, ranged from 0% to 38% in the high-quality studies and from 0% to 4% in the low-quality studies.
Although all the studies followed patients with TS over time their objective(s) sometimes differed, leading to heterogeneity and adversely affecting comparability. With 40% of the included studies having a retrospective design, selection bias may have interfered with the overall outcomes. All included studies described children with TS who were hospitalised or treated as outpatients, none of the patients were seen by a GP only. This might imply that potentially useful information is lost, since many children with TS will be followed only in primary care. Also noteworthy is that the follow-up of children with TS began from the moment of inclusion in the study. This means that the numbers of symptomatic days prior to inclusion are not included in the follow-up period, which might have led to underestimation of the duration of symptoms on short-term follow-up.
The scoring system we used for methodological quality assessment was adopted from an earlier review on assessing clinical course [8], but has not yet been validated. Furthermore, a distinction into high and low quality studies may be too black-and-white a representation of the quality of the included studies. Nevertheless, according to our rating, the methodological quality was high for more than 50% of the included studies. Critical items were criteria 5 and 6 ('same length of follow-up' and 'follow-up available from at least 80% of the study population'). Both these criteria induced loss of patient information during follow-up and might have influenced the outcomes of the present review. It showed that even within high quality studies large differences in outcomes were seen. The percentage of children with persistent symptoms after one to three weeks of follow-up in Figure 2 ranged from 10% to 40%. The studies , that showed the extreme values of 33% [24] and of 40% [3] did neither use the same test for all patients nor did they have a follow-up larger than 80%. The other studies showed a more homogeneous percentage of 10-17% for persistent symptoms after one to three weeks of follow-up.
For recurrence rate the largest differences was 0% after 9.3 months [20] and 19% after one year [16]. In this last study an extensive systematic re-examination in all patients was used, while the other study re-examined only some of the patients at three months based on initial findings and just reported that there was no recurrence between three months and three years of follow-up. A low quality study also showed a large recurrence rate after seven months of follow-up, but this study was retrospective and had more than 20% loss to follow-up [2].
The relatively high recurrence rate of TS at the opposite hip supports the hypothesis that TS is due to a systemic Source Figure 2 Data on short-term symptoms reported in six studies arranged by high (black) and low (white) methodological quality of the studies. factor rather than being a local problem. A review by Do [5] includes various studies suggesting that a viral agent (causing respiratory tract infection or other flulike syndromes) may also lead to the development of TS in children; elevated viral titers have been found in children with TS, although an infectious agent has not yet been identified. The prevalence of Legg-Perthes' disease subsequent to TS (reported in nine high-quality studies) [4,10,[13][14][15][16]19,20,23] ranged from 0% to 10%. One high-quality study by Keenan et al. [17] reported that 38% of the cases initially diagnosed as TS subsequently developed Legg-Perthes' disease. The aim of this latter study differs from the rest since only 13 patients with either prolonged or recurrent TS were studied. Five of them had a delay of more than two years in bone age and all five developed changes of Legg-Perthes' disease at three-months follow-up. These outcomes are consistent with the earlier mentioned hypothesis of a higher risk of Legg-Perthes' disease in children with prolonged symptoms of TS [1]. However, due to the low number of included patients and the preselected group of children with TS, the prevalence of 38% should not be generalized to the risk of developing Legg-Perthes' disease subsequent to TS.
Gaughan et al. describe an estimated prevalence rate of Legg-Perthes' disease (reported in the general pediatric population in the USA) of 0.023% [31]. The high quality studies in our review report a prevalence of Legg-Perthes' disease subsequent to TS of 0-10%. The question remains whether Legg-Perthes' disease is a direct consequence of TS in children. Symptoms of TS could also represent early signs of Legg-Perthes' disease. Children initially diagnosed with TS and (after ongoing symptoms) subsequently diagnosed with Legg-Perthes' disease fit this hypothesis. Early signs of Legg-Perthes' disease found at retrospective analysis of initial radiographs of patients with TS [15], also support this theory. Moreover, subtle radiologic changes of Legg-Perthes' disease are easily overlooked. In the literature, Gopakumar et al. found an absence of early radiologic signs in about 20% of the patients at first presentation, and in about 50% of the patients diagnosed with Legg-Perthes' disease [2].
Regrettably, some studies excluded children with Legg-Perthes' disease whereas information about the medical history of these patients might have revealed a misdiagnosis in TS. The reported follow-up periods after the initial diagnosis of TS in which subsequent Legg-Perthes' was diagnosed ranged between 1 and 6 months [1,10,11,17,20,24,28]. This might indicate that some studies did not follow their patients long enough to detect this.