Osteoarticular Infections (OAI) cause significant morbidity worldwide, and the incidence is increasing [1–4]. This likely reflects both a higher burden of comorbidity in the population, and an increased performance of surgical bone and joint procedures [1, 2].

Streptococci are consistently second only to staphylococci as causative agents of OAI, representing approximately 15–20 % of the cases [4–6]. However, knowledge on clinical and microbial characteristics specific for streptococcal OAI is scarce, and mostly available from small case series published decades ago [7]. The few retrospective cohort studies published revealed a predominance of β-haemolytic streptococci, with group A Streptococci (GAS) and group B streptococci (GBS) comprising the majority [8–10]. Furthermore, species-associated clinical features were suggested; GBS mainly affecting the multi-morbid and elderly, and GAS producing more severe clinical manifestations. However, the number of included cases in these studies was generally too small to permit statistical analyses or evaluate temporal trends.

Streptococcus dysgalactiae subspecies equisimilis (SDSE), a β-haemolytic streptococcus predominantly possessing Lancefield group C or G antigen (GCS and GGS), has recently emerged as an important pathogen, increasingly associated with invasive disease [11]. We wished to explore if this rising trend was reflected in the rates of osteoarticular GCS and GGS-infections, and to describe the host and pathogen characteristics of these infections in comparison to OAI caused by GAS and GBS.

The 15-year survey period yielded 24 GAS, 45 GBS and 42 GCGS acute Osteoarticular Infections, representing 8, 13 and 20 % of the total count of invasive episodes, respectively. Two patients with GCGS acute OAI declined participation in the study. They are represented in the cumulative incidence, but omitted from all further analyses. One GBS and three SDSE were identified based on partial sequencing of 16S rDNA gene from the four clinical samples that were culture negative. The proportions of the different subgroups of OAI are described in Table 1.

Annual incidence and seasonal variation

The cumulative incidence of GCGS OAI increased significantly from the first to the last 5-year period (IRR 5.7, p = 0.0003), with annual incidence peaking at 1.9/100 000 in 2013 (Fig. 1). Conversely, the cumulative incidence of GAS and GBS OAI did not display any significant trends, and remained relatively stable at 0.5/100 000 (IRR 1.27, p = 0.63) and 0.8/100 000 (IRR 1.77, p = 0.14), respectively.

All three groups exhibited seasonal variation, GAS peaking in the winter with 38 % of the cases occurring from December to February, and only 13 % during the summer-months of June through August. Differently, GBS and GCGS presented predominantly during summer (both 31 %), and less frequently in the winter (24 and 19 %, respectively).

Demographics and comorbidity

The demographics, risk factors and disease manifestations are shown in Table 1.

Both GBS and GCGS were significantly associated with the presence of comorbidity and risk factors; GBS clearly linked to systemic comorbidity, and GCGS predominantly associated with factors of local origin. Patients presenting with GBS or GCGS OAI were significantly older than the GAS OAI (corrected p-value = 0.007 and p = 0.049), and a male predominance was noted for GAS and GCGS. The age-difference is further delineated in Fig. 2, showing that the age-distribution for GBS and GCGS was skewed towards the elderly.

Molecular characteristics and antimicrobial susceptibility

Altogether, 21 GAS, 30 GBS and 29 GCGS isolates were available for typing, revealing a diverse microbial population (Fig. 3). The major GAS emm-types were emm1 and emm3, and stG485 dominated among GCGS, although neither a clonal outbreak nor a temporal trend could be inferred. Among GBS, capsule-type III and V were most frequently identified.

All isolates were sensitive to penicillin. Resistance to erythromycin was detected in 9/41 (22 %) of GBS and 4/37 (11 %) of GCGS, and resistance to clindamycin in 7/41 (17 %) of GBS but in none of the GCGS. GAS were uniformly susceptible to erythromycin and clindamycin.

Beta-haemolytic streptococci are an important cause of Osteoarticular infections (OAI), but their temporal trends have previously not been explored in detail. We found a highly significant increase in the GCGS OAI incidence during the 15-year survey period, the annual incidence peaking at 1.9/100 000 in 2013. To our knowledge, this has not been previously reported, although it is consistent with several epidemiological studies on invasive GCGS disease in general [11, 18, 19]. Conversely, the rates of GAS and GBS OAI remained relatively stable at 0.5/100 000 and 0.8/100 000, respectively. This contrasts the findings from a recent study, where a 3-fold increase in the incidence of GBS bone-infections in the period 1995–2012 was documented [20]. However, that study also included chronic osteomyelitis, and 76 % of the patients in their bone-infection cohort were diabetics. In line with our observations, though, they reported a stable incidence of 0.8/100 000 for GBS joint infections.

The increasing GCGS OAI incidence in our material could partly be due to increased microbial sampling or improved diagnostic procedures. However, one would expect this to be reflected also in the rates of GAS and GBS OAI, but these remained stationary. The ageing and increasingly comorbid population is probably of importance, but surprisingly did not affect the rates of GBS OAI, although GBS and GCGS both appear to preferentially target the elderly and the vulnerable patients. Acquisition of novel virulence factors enhancing invasive potential or conferring specific osteoarticular tissue tropism could be a contributing factor, and horizontal genetic transfer between streptococci has been inferred [21]. Experimental arthritis studies indicate a pivotal role for bacterial adhesins in the pathogenesis, but the knowledge on adhesins in GCGS is limited at present [22–24]. Osteoarticular infections are one of the major disease manifestations of GCGS, constituting approximately 20 % of invasive disease, and further exploration of the molecular basis for their arthritogenicity is warranted [11].

We observed several species-associated clinical characteristics. GAS OAI tended to present more acutely, to produce a more severe clinical picture, and displayed the highest propensity to cause native joint infections. GBS and GCGS affected primarily the elderly, and were significantly associated with the presence of comorbidity and risk factors; GBS were clearly linked to systemic comorbidity, whereas GCGS were predominantly associated with local risk factors. Furthermore, GBS OAI were frequently polyarticular, and comprised the majority of the recurring infections. Our findings are consistent with those from previous publications [7–10].

The antibiotic treatment was concordant with established national guidelines regarding choice of antimicrobial agent and length of treatment. The surgical intervention, however, occasionally diverged from current standards [13]. Several PJIs were treated with curative intention with antibiotics alone, particularly in the earliest time-period. This probably had an impact on the rates of recurring infections and sequelae, although, GBS have been reported to be associated with high relapse rates in prosthetic joint infections compared to other pathogens [25].

Previously, emm4, emm12 and emm28 have been associated with infective arthritis [26–28]. We found a substantial molecular diversity among our isolates, and although these three emm-types were represented, they did not constitute the majority (Fig. 3). On the contrary, the identified emm-types for both GAS and GCGS conform closely to the prevailing emm-types encountered in invasive disease in general during the study period [11]. Similarly, it is tempting to speculate that the OAI GBS serotypes likely reflects the most prevalent capsule types in the population, as previously documented in Norway [29].

The present study is limited by its retrospective design, increasing the risk of diagnostic misclassification. However, the microbe-specific emphasis of the study, along with a review of the local microbiological records, likely led to the inclusion of all cases of β-haemolytic streptococcal infection during the study period. Such an approach probably conforms more closely to the true disease burden than studies based on selected isolates submitted to National surveillance laboratories.

The total incidence and characteristics of OAI in our health region were unfortunately not available. Hence, a more comprehensive comparison of the clinical features of streptococcal OAI and OAI caused by other bacteria was not feasible.

Sepsis has recently been redefined, and the new classification should ideally have been applied to our sample material [30]. However, the use of the former sepsis definition based on SIRS-criteria permit a better comparison with existing literature on OAI and streptococcal disease in general. Furthermore, the retrospective design of the study hampered the categorization of the patients according to the new sepsis definition based on Sequential Organ Failure Assessment score results.

We found a significantly increasing incidence of GCGS Osteoarticular infections, whereas GAS and GBS OAI did not display any temporal trends. GCGS OAI were clearly associated with the presence of comorbidity and risk factors, predominantly of focal origin. With an increasing application of therapeutic and diagnostic bone and joint procedures, the rising trend of OAI caused by GCGS is likely to continue. Sustained epidemiological attentiveness to GCGS seems warranted.