In the present study 14.3% of the infected nonunions could be diagnosed as polymicrobial infection. Infected nonunions which were defined by FRI criteria and cessation of bone healing after at least six months can be regarded as chronic FRIs. Thus, infected nonunions can be compared to posttraumatic osteomyelitis being a chronic bone infection as well. Jorge et al. reported 37.8% polymicrobial infections using standard culturing techniques in posttraumatic osteomyelitis [
12]. Wimmer and co-workers reported 46.6% polymicrobial infections in PJI [
8], while Bozhkova et al. found 28.5% PJIs to be polymicrobial [
13]. Both studies used standard culturing techniques and did not differ between acute and chronic PJI as well. In addition, comparing different entities such as PJI with infected nonunion makes comparison of data difficult. Reasons for the low rate of polymicrobial infections (Table
1) could be previous antibiotic treatment as part of fracture care. Preceding antibiotic prophylaxis in fracture treatment might reduce numbers of different germs at the fracture site and thus later polymicrobial infection in infected nonunion. Results, however, did not show any statistically significant difference of monomicrobial, polymicrobial infections as well as germ-changes in course of treatment in regard to type and duration of antibiotic treatment (Table
3), which might be due to the low volume in case numbers. Type of accident was considered an indicator of associated soft tissue damage. Data, however did not show any difference concerning type of infection. Open fractures are expected to be exposed to several different pathogens compared to closed fracture. Further, patients with open fractures usually obtain longer perioperative antibiotic treatment than patients with closed fractures in which antibiotic treatment is not recommended beyond 24 h [
14]. Intriguinly, we found same numbers of open and closed fractures for polymicrobial infections in infected nonunion (
n = 3 each). Since almost equal numbers of infected nonunions for closed (57.1%) and open fractures (42.9%) could be figured out, longer antibiotic treatment could be reason for similar, but not higher numbers of polymicrobial infections in open fractures compared to closed ones. Nevertheless, data suffices not to proof this theory. Another aspect which might explain low numbers of polymicrobial infection is tissue culture itself. It is traditionally the gold standard for diagnosing infection, although low sensitivity is reported for culturing intraoperatively taken tissue samples [
15]. Hospital-based culture techniques apply heavy selection pressure. Different germs in originally polymicrobial infections could be simply overgrown in conventional plating. In addition, artificial growth conditions favor bacteria capable growing under those conditions [
16]. Furthermore, biofilm bacteria which are in their sessile phase are generally difficult to culture [
17]. Another aspect is the possibility of presence of viable but non-culturable bacteria (VBNC). Many planktonic as well as biofilm bacteria are able to enter a starvation survival state, which results from external stress (e.g. antibiotic treatment). Those bacteria can not be grown on laboratory media and are therefore difficult to detect with standard microbiological assays [
18,
19]. Furthermore, additional PCR was not performed routinely in patients with positive culture results. Thus, the authors could not determine a significantly higher rate of polymicrobial infections as reported by Palmer and co-workers who compared PCR combined with time-of-flight mass spectrometer and Fluorescence In Situ Hybridization (FISH) with microbiological culture [
15].
In literature, coagulase-negative
Staphylococcus spp. and
Staphylococcus aureus are the most common infection causing germs in infected nonunions and PJI [
20‐
22]. We found
Staphylococcus spp. being the most frequently evidenced pathogen in infected nonunion patients, followed by
Enterococcus spp.. Interestingly, gram-negative bacteria and
Streptococcus spp. were only found in polymicrobial infections and cases with germ-change evidenced after follow-up revision surgeries (Table
4). An explanation for detecting gram-negative bacteria in those cases might be empirical broad-spectrum antibiotic therapy with amoxicillin/clavulanic acid and generation I and II cephalosporines after the initial revision surgery. Those antibiotics target predominantly gram-positive pathogens. Since cefazolin was most often used for empirical treatment in infected nonunion treatment, resistance of
Enterococcus spp. to cephalosporins may explain the more frequent detection of
Enterococcus spp. after revision surgeries [
23]. Despite the shift of detected germs during multiple revision surgeries, the present data shows that all germ-changes in course of multiple revision surgeries occurred in patients with initial monomicrobial infection (21/21 patients, 100%). This can either be attributed to superinfection during surgical treatment itself or to failure in microbiological detection of all pathogens responsible for index infection. Moreover, germ-changes may also be regarded as laboratory contaminations, especially towards the identification of coagulase-negative staphylococci. Even a standardiszd collection of relevant samples from the suspected osseous defect cannot exclude subsequent contamination within the framework of microbiological diagnostics [
24].
Limitations of our study are its retrospective design and the low volume in numbers of patients. Since infected nonunion is a relative rare diagnosed sequela, not more than 42 patients were involved according to our inclusion criteria within a long observation period. The low number in cases with a small group of polymicrobial infections prevents statistically significant conclusion of the results. Inclusion of infected long bone nonunions of the lower and upper extremities leads to further heterogenity of the study group. Although different microbiological patterns in musculoskeletal infections of the upper and lower extremities are well known [
25], the authors considered reporting those cases useful due to lack of reported microbial findings in infected nounions of the upper extremities. The retrospective design of the study accounts for a high rate of unknown previous antibiotic treatment. Patients suffering from infected nonunion usually undergo long-term suppressive antibiotic therapy. This previous treatment may alter the diagnostic yield significantly as well as microbiological patterns detected by different diagnostic methods. Retrospectively reviewing medical records over a ten-year period did not allow to discriminate between diagnostic yield of different diagnostic methods. This can be regarded as a major drawback of our study, although standardization of tissue sampling and diagnostic work-up was established for the included patients. The fact, that instead of five relevant tissue samples, at least three relevant tissue samples were sent to microbiological diagnostics is a further downside. The recent understanding, that five relevant tissue samples increase microbial sensitivity in PJI and FRI, was not known for the observation period and thus retrospective analysis not possible [
26] . The same applies to the laboratory culture work-up. Recent findings of higher sensitivity when culturing tissue samples in liquid media was not established as microbiological culture standard in our laboratory over the whole study period. Nevertheless, the standard diagnostics in our study should be similar to most of the microbiological diagnostic set-ups performed worldwide which prompted us to report the present data. The additional use of sonication did only result in three additional evidenced pathogens in cases otherwise being treated as culture negative infected nonunion. These limited case numbers did not allow any conclusions for diagnostic relevance of sonication which needs to be clarified in future studies. In addition, an exact classification of soft tissue damage according to Gustilo-Anderson in open fractures was not possible reviewing patient records. Lastly, the question if adaption of antibiotic treatment to germ-changes evidenced after follow-up revision surgeries is appropriate for treatment cannot be answered by this descriptive analysis.