Notably, no clinical variables were found to be independently associated with biofilm formation in our study. Unlike previous studies, we did not observe biofilm formation to be statistically associated with relapse/reinfection, or chronic infections [14]. These prior studies examine mainly chronic diabetic or vascular stasis ulcers, which were not the type of wounds in our population; many of our patients underwent aggressive serial debridement of wounds, which is a unique aspect of trauma care. This result also differs from our earlier publication on these isolates, in which biofilm formation was observed in isolates serially recovered from ongoing infections in several patients [8]. With the present study, clinical information recovered from chart review was able to differentiate colonizing from infecting organisms, distinguish between discrete infections, and assess for outcome. A case control study recently published from our institution, with a different patient population and bacterial isolates, demonstrated that biofilm formation was associated with persistence of skin and soft tissue wound infection beyond 14 days [3]. Although no differences were found between biofilm producing isolates versus non-producing isolates and cure, correlation to wound persistence or duration of infection was not examined in this study.
Interestingly, we did not find significant differences in previous antibiotic exposure or duration of antibiotic exposure prior to organism recovery among biofilm formers either on univariate or multivariate analysis. Some authors have found that antibiotic exposure can induce biofilm formation in vitro [15]. It is possible that this phenomenon is not replicated in vivo, or that the analysis is confounded by the relatively high rate of previous antibiotic exposure in this population in whom severe trauma is common.
More deaths occurred with initial infections by biofilm-producing isolates overall, but the attributable mortality (i.e. death from infection) remained low (1 out of 14). The reasons for this are not immediately clear as both groups were similar with respect to presence of burn, TBSA of burn, gender, age, and site of recovery; although low numbers in the comparator group make drawing conclusions difficult. Our literature search did not reveal any studies addressing this issue.
In univariate analysis, higher percentage TBSA was found to be associated with relapse/reinfection, or chronic infection. This is similar to findings indicating that the risk of burn wound infection increases in proportion to the amount of body surface area burned [16]. MSSA was more likely to be associated with relapse/reinfection, and chronic infection, but this analysis is limited by the small sample size (6 isolates). Study isolates recovered from wound infections were more likely to be cured without relapse or reinfection; perhaps this reflects the ability to obtain a surgical cure through tissue debridement. However, no clinical variables or microbiological characteristics were found in this study in multivariate analysis to be independent predictors of relapse/reinfection or chronic infection.
There are several limitations notable in this study. As a retrospective study, our data are subject to selection bias and misclassification bias, particularly since we were not able to control for these potential influences. Utilization of a randomly selected sample had consequences in microorganism representation as well as demographics. Species were unevenly represented in this sample (6 MSSA versus 53 A. calcoaceticus-baumannii complex). However, this distribution was similar to prospectively-acquired data from military casualties having persistent wound infections [3]. A majority of the isolates selected were biofilm producers. This may be due to the selection of isolates from a repository containing many multidrug resistant isolates. This composition may not accurately represent the frequency of biofilm formation among isolates in an unselected population, but did allow for the study of large numbers of biofilm and non-biofilm forming isolates.
Common medical comorbidities, such as coronary artery disease (10 of 144 patients) and diabetes (13 of 144 patients) were also underrepresented; this may have been the result of sampling bias from our burn center, the young age of patients in the cohort, or inaccurate documentation of patients’ past medical history. There are likely too few cases to reasonably draw conclusions regarding the presence of biofilm forming isolates and these medical comorbidities.
With regards to medical device-related infections, despite a large overall isolate database, we observed only 19 orthopedic device related infections, 5 urinary catheter related infections, and 6 central line related infections. The small numbers in these cohorts could explain why no association was found with catheter associated infections and also explain why the association with orthopaedic devices was not significant on multivariate analysis. Future studies could include a sample size determined by power analysis to detect differences for all potential associations.
Unlike previous studies [17], recovery of organism from a polymicrobial culture site was not correlated with biofilm producing isolates. Data examining the biofilm producing capability of the co-located organisms were not available due to the retrospective nature of this study; it is possible that a non-study isolate was a biofilm producer and co-located with the study isolate. With these in mind, future studies should investigate the characteristics of each isolate in a clinical polymicrobial culture to better understand the nature of interactions, and multiple organisms commonly isolated together from polymicrobial wound infections could be studied together to determine the influence of multiple species on biofilm formation.
Ultimately, the in vitro formation of biofilms may not be related to the in vivo phenotype, and the in vitro formation of biofilms may underrepresent the biofilm forming capability of an organism. For example, plasma enhances the in vitro biofilm formation of S. aureus. Current techniques assessing in vitro biofilm formation often exclude factors that impact expression of this phenotype in an in vivo infection, and laboratory protocols standardizing biofilm assessment are lacking. It is also possible that while isolates may have the ability to form biofilms in vitro and in vivo, this phenotype may not always be expressed in a particular infection. Furthermore, the link between the in vitro phenotype and clinical outcome is still unclear, and further studies are required to establish this link.