Acute haematogenous osteomyelitis (AHO) still remains a challenge, as it is associated with a significant worldwide morbidity and often requires prolonged treatment.1,2 In Europe, the incidence ranges from 2 to 13 cases per 100,000 inhabitants.2–4 The aetiology is variable, but Staphylococcus aureus is yet the most frequent causative pathogen, accounting for 36.5–70% of positive cases,1,2,5 followed by Kingella kingae in young children (14–22.2%).6,7

In Europe, empiric antibiotherapy for AHO is well established,3 but the indication for a surgical procedure remains controversial and there is a trend towards less aggressive and shorter treatments.1,8–11 However, in cases caused by S. aureus infection, especially those involving Panton-Valentine leucocidin-positive S. aureus (PVL-SA) or methicillin-resistant S. aureus (MRSA),3,12–14 treatment is longer and more aggressive when compared to cases of K. kingae infection.6,7 Furthermore, if AHO is concomitant with septic arthritis, abscesses or (pyo)myositis, prolonged treatment courses and multiple surgical interventions are often needed.1–3,13,15–17 However, few studies have identified predictors of complications, which are of outmost importance in heterogeneous series.16,18,19

In 2014, a revised protocol was introduced to manage AHO at our hospital. The objectives of our study were to identify children at higher risk for complicated AHO, analyse aetiological trends and management standard practices (to achieve 80% compliance with the current protocol) and to compare differences between S. aureus infections and infections by other pathogens.

We performed a longitudinal, observational study in children aged less than 18 years with AHO admitted to a tertiary care paediatric hospital between January 2008 and December 2018. We included patients retrospectively until 2014 and prospectively thereafter. A multidisciplinary team followed all patients from 2014. We collected data on demographic characteristics, clinical manifestations, microbiological and imaging findings, treatment and outcomes from medical charts. We excluded patients aged less than 3 months and patients with a history of surgery or open trauma at the site of infection.

The diagnosis of AHO and septic arthritis was based on classical clinical findings (inability to bear weight, limited range of movement [ROM], pain, local inflammatory signs, fever) for less than 14 days and compatible imaging findings, with or without microbial isolation.1,2 Good clinical response on follow-up further indicated septic arthritis. Complications included abscesses, (pyo)myositis, deep venous thrombosis (DVT), disseminated infection, pathologic fractures, joint dislocation, avascular necrosis or chronic osteomyelitis. Residual pain, limping, ROM limitation, stiffness, angular deformity or limb length discrepancy (LLD) caused by AHO and lasting for more than 6 months were considered sequelae.1,16

The recommended first-line antibiotics were flucloxacillin (150−200mg/kg/day given every 6h) or cefuroxime (150mg/kg/day given every 8h) for children aged 3 years or younger. A switch to oral therapy was considered after a minimum of 3 days if the patient was afebrile, improving clinically and had a 50% decrease in the C-reactive protein (CRP) level (or CRP≤20mg/L).3,10

Bacterial identification and antimicrobial susceptibility were performed at the hospital microbiology laboratory. Synovial fluid samples for culture were inoculated into blood culture bottles. After 2014, when available, joint fluid and bone samples were tested for Kingella kingae with specific real-time polymerase chain reaction (PCR), in children aged 4 years or younger.

Although our study was conducted in a single centre, it included one of the largest cohorts of patients with AHO to describe trends in the aetiology and management of AHO in Lisbon and Southern Portugal. Our findings show that (1) MSSA continues to be the most common causative agent of AHO, with no variation between years; (2) K. kingae is still not a frequent detectable cause of AHO in these patients, still molecular testing continues not to be performed routinely; (3) the management of AHO in our area continues to be traditional, with prolonged courses of antibiotherapy, longer lengths of stay and a higher frequency of surgical intervention, although this trend has been decreasing in recent years; (4) there is an unexpectedly high frequency of concomitant septic arthritis and complications, mostly myositis, which may explain some of these results. The increased accessibility and precision of MRI in recent years may also partly account for some of our findings. However, our study did not find an increase in the use of MRI over time.

The incidence of AHO has increased in recent years.2 Gafur et al. report a 2.8-fold increase in the last 20 years,4 while Samara et al. describe a 79% increase in the mean annual incidence,7 although only in young children. Similarly, we found a 1.8-fold increase in the past 5 years, but without changes in age distribution. A growing awareness of AHO, prospective enrolment after 2014 and improvements in imaging combined with an actual incidence rise may explain this trend. The use of molecular detection methods, which was minimal in our study, cannot account for this increase.

In our study, a pathogen was identified in only 36.6% of patients, which was consistent with the findings of Filleron et al. in France (40.7%)20 or Calvo et al in Spain (33%).16 Yet, it is much lower than the 62.7% reported by Samara et al in Geneva,7 where they routinely used molecular methods for detection of K. kingae, or the 76% reported by McNeil et al in the United States.21 In addition, the proportion of positive blood cultures in our study was only 21%, lower compared to the previous literature,1,2,10 although it was similar to the proportion reported by Samara et al.7 In our sample, infections without microbiological identification were more common in children aged less than 4 years, with CRP levels of less than 20mg/L and fewer complications. It is very likely that K. kingae caused a fair proportion of these infections in young children. In our study, only 2 of the 5 children tested for K. kingae by PCR had positive results, with detection in a synovial fluid sample in one and in a throat swab sample in the other. Synovial fluid samples were collected in 7 other cases, but the small amount of fluid precluded the use of this assay. Although diagnosis based on a positive throat swab is questionable given the prevalence of oropharyngeal colonization by K. kingae in young children (8%–12%), the throat swab seems to offer a good sensitivity (90.5%) for diagnosis of K. kingae in osteoarticular infections.7,22 The routine use of molecular techniques would increase the reported incidence of K. kingae infection,7,23 especially since the incidence of septic arthritis is increasing. In any case, bone aspiration was less frequent after 2014, which was probably associated with a decrease in microbiological identification.

Staphylococcus aureus was the bacterium isolated most frequently (57.7%), as expected based on its involvement in 36.5% to 63% of cases described in the literature.1,10,13,14 In our study, most cases involved MSSA (93.3%) and MRSA was still uncommon (6.6%), which was similar to the findings in other European countries (7.8%)12 and much lower compared to the United States (18.5%–70%).13,17,24 In a previous study conducted in our hospital between 2005 and 2008, community-acquired S. aureus infections were PVL-positive in 23% of cases, with isolation of MRSA in 4.3%.25 Although MRSA infections are associated with complications and poorer outcomes, MSSA can be equally aggressive, independently of PVL production or the USA300 genotype.12–14,17 Most studies report a higher frequency of fever, a higher mean CRP value, more complications, longer antibiotic courses and longer lengths of stay in patients with S. aureus infections.13,14,17,24 Concomitant septic arthritis, subperiosteal abscess, pyomyositis and DVT are common complications of these infections.13,14,17 In our series, although we only tested for LPV in 1 case with a negative result, S. aureus infections were also associated with higher elevation of inflammatory markers, longer duration of treatment and length of stay and a higher incidence of complications (60%).

The incidence of S. pneumoniae (8%) and GAS infections remained stable in our series (19%). However, in Europe, especially in the United Kingdom, there is evidence of an increased incidence of infection by invasive GAS, specially emm1 strains.26 In our study, the only GAS strain available for typing was an emm4, which is not among the most prevalent emm types involved in invasive infections in Portugal (emm1, emm89 and emm3).27 However, this was consistent with the findings of a study conducted in Spain, in which the most frequent emm types involved in osteoarticular infections in children were emm1, emm12 and emm4.28 In our cohort, although the incidence of GAS infections did not increase, they were severe and required not only prolonged IV treatment longer hospital stays, but also multiple surgical interventions.

In our study, 39.4% of patients had concomitant septic arthritis, with an increased frequency (54.3%) since 2015. This is comparable to the 35.2% reported by McNeil et al.21 or the 39.8% reported by Schallert et al.29 but higher compared to the 12% reported in the Spanish cohort.16 Despite the high sensitivity of MRI for both AHO and septic arthritis, specific virulent factors may account for this difference.3,13,14,29 In our series, there was a clear increase in AHO with concomitant septic arthritis after 2015, which was more frequent in children with a positive culture, fever at admission and greater elevation of inflammatory markers and was associated with longer treatment duration, as observed in other studies.17,18,29

The incidence of myositis is increasing, and it has been suggested that this is related to the increase in complicated MRSA and PVL-positive infections. In 2006, Bocchini et al. reported than PVL-positive S. aureus isolates were significantly associated with a higher probability of concomitant myositis and more severe disease.13 Furthermore, Dohin et al. considered myositis as a musculoskeletal complication.14 In our study, myositis was associated with greater elevation of inflammatory markers, S. aureus infection, septic arthritis and prolonged IV treatment, suggesting an intense local inflammatory response that would contribute to disease severity.

We found adherence with the hospital treatment protocol in 80% of cases, although the duration of antibiotherapy was too long: although it had decreased modestly in recent years, it was still far from the short course of IV antibiotherapy proposed by Peltola and Pääkkönen.10 On the other hand, in their case series only 3.8% of patients had hip involvement, compared to 12.7% in our sample, and the authors did not report on complications, but in our sample 45.1% of patients experienced complications, more frequently after 2015 (54.2%). These differences may partially explain the longer duration of treatment, along with the more traditional approach used in our hospital.3,17,19 Indeed, patients with complicated high-risk osteoarticular infections or with slow clinical improvement may require longer courses of IV and oral antibiotherapy.3,19

According to the European guidelines, surgery should be avoided in uncomplicated AHO.3 In agreement with this, Peltola and Pääkkönen mention that up to 90% of patients with early AHO can be treated conservatively.10 However, in a case series previously published by these authors, only 24% of children with AHO did not require surgery.31 Calvo et al. reported surgical intervention in only 20% of patients with AHO, with most surgeries (46%) performed on account of complications, which only happened in 13.4% of patients.16 In our study, 40.8% of patients underwent invasive procedures, which were more frequent in patients with complicated AHO (62.5%), which was consistent with previous studies.11,21,24 Indeed, abscess drainage should be considered in cases in which fever, symptoms, marked elevation of CRP or bacteraemia persist.3 Our study found a less aggressive approach to the management of AHO in recent years, with a decreased frequency of bone aspiration and abscess drainage despite a similar frequency of abscesses.

Sequelae occur in 1.1%–4.5% of cases and are associated with strain virulence (more frequent in MRSA infections, 4.7%–8.6%),17,24 disease severity, delayed treatment and concomitant hip or shoulder infection.3,21,30 In our study, 3.6% of the patients had poor long-term outcomes, all minor and associated with previous complications (OR, 4.4). It is likely that strain differences and host susceptibility explain small variations in outcomes.

Identifying children at risk of complications is essential.16,19,31 In our study, the main predictors of complicated AHO were age 3 years or greater, time elapsed from onset to diagnosis of 5 or more days, CRP levels of 20mg/L or greater and pathogen isolation. Furthermore, microbiological isolation was the only factor that remained significant in the logistic regression analysis. The presence of sequelae at 6 months of follow-up was associated with complications. However, we were unable to identify clinical and laboratory predictors of sequelae of AHO due to the small sample size. Other limitations of the study include the partial retrospective design and it being conducted in a single centre.

AHO still is a major health problem. The most frequent aetiological agent continues to be MSSA, which causes more severe infections. Microbiological diagnosis in our area still needs to improve, mainly through the use of molecular testing. Duration of treatment should be reduced through the use of validated scores to identify patients with a lower risk of complications.